Genel Introduction to Ore Forming Processes.p ITOA01 09/03/2009 14:30 Page iiINTRODUCTION TO ORE-FORMING PROCESSES ITOA01 09/03/2009 14:30 Page iITOA01 09/03/2009 14:30 Page iiIntroduction to Ore-Forming Processes LAURENCE ROBB ITOA01 09/03/2009 14:30 Page iii© 2005 by Blackwell Science Ltd a Blackwell Publishing company 350 Main Street, Malden, MA 02148-50120 USA 108 Cowley Road, Oxford OX4 1JF, UK 550 Swanston Street, Carlton, Victoria 3053, Australia The right of Laurence Robb to be identi?ed as the Author of this Work has been asserted in accordance with the UK Copyright, Designs, and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs, and Patents Act 1988, without the prior permission of the publisher. First published 2005 by Blackwell Publishing Library of Congress Cataloging-in-Publication Data Robb, L.J. Introduction to ore-forming processes / Laurence Robb. p. cm. Includes bibliographical references and index. ISBN 0-632-06378-5 (pbk. : alk. paper) 1. Ores. I. Title. QE390.R32 2004 553'.1—dc22 2003014049 A catalogue record for this title is available from the British Library. Set in 9/11 1 – 2 pt Trump Mediaeval by Graphicraft Limited, Hong Kong Printed and bound in the United Kingdom by TJ International, Padstow, Cornwall For further information on Blackwell Publishing, visit our website: ITOA01 09/03/2009 14:30 Page ivPreface vii INTRODUCTION: MINERAL RESOURCES 1 Introduction and aims 1 A classi?cation scheme for ore deposits 2 What makes a viable mineral deposit? 4 Some useful de?nitions and compilations 6 Natural resources, sustainability, and environmental responsibility 11 Summary and further reading 15 PART 1 IGNEOUS PROCESSES 1I GNEOUS ORE-FORMING PROCESSES 19 1.1 Introduction 19 1.2 Magmas and metallogeny 20 1.3 Why are some magmas more fertile than others? The “inheritance factor” 28 1.4 Partial melting and crystal fractionation as ore-forming processes 37 1.5 Liquid immiscibility as an ore-forming process 54 1.6 A more detailed consideration of mineralization processes in ma?c magmas 57 1.7 A model for mineralization in layered ma?c intrusions 71 Summary and further reading 74 2M AGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES 75 2.1 Introduction 75 2.2 Some physical and chemical properties of water 76 2.3 Formation of a magmatic aqueous phase 79 2.4 The composition and characteristics of magmatic- hydrothermal solutions 85 2.5 A note on pegmatites and their signi?cance to granite-related ore-forming processes 93 2.6 Fluid–melt trace element partitioning 96 2.7 Water content and depth of emplacement of granites – relationships to ore-forming processes 101 2.8 Models for the formation of porphyry-type Cu, Mo, and W deposits 106 2.9 Fluid ?ow in and around granite plutons 108 2.10 Skarn deposits 113 2.11 Near-surface magmatic- hydrothermal processes – the “epithermal” family of Au–Ag–(Cu) deposits 117 2.12 The role of hydrothermal ?uids in mineralized ma?c rocks 122 Summary and further reading 125 PART 2 HYDROTHERMAL PROCESSES 3H YDROTHERMAL ORE-FORMING PROCESSES 129 3.1 Introduction 129 3.2 Other ?uids in the Earth’s crust and their origins 130 Contents ITOA01 09/03/2009 14:30 Page v3.3 The movement of hydrothermal ?uids in the Earth’s crust 138 3.4 Further factors affecting metal solubility 147 3.5 Precipitation mechanisms for metals in solution 153 3.6 More on ?uid/rock interaction – an introduction to hydrothermal alteration 166 3.7 Metal zoning and paragenetic sequence 174 3.8 Modern analogues of ore-forming processes – the VMS–SEDEX continuum 177 3.9 Mineral deposits associated with aqueo-carbonic metamorphic ?uids 189 3.10 Ore deposits associated with connate ?uids 197 3.11 Ore deposits associated with near surface meteoric ?uids (groundwater) 209 Summary and further reading 214 PART 3 SEDIMENTARY/ SURFICIAL PROCESSES 4S URFICIAL AND SUPERGENE ORE-FORMING PROCESSES 219 4.1 Introduction 219 4.2 Principles of chemical weathering 220 4.3 Lateritic deposits 223 4.4 Clay deposits 233 4.5 Calcrete-hosted deposits 235 4.6 Supergene enrichment of Cu and other metals in near surface deposits 238 Summary and further reading 245 5S EDIMENTARY ORE-FORMING PROCESSES 246 5.1 Introduction 246 5.2 Clastic sedimentation and heavy mineral concentration – placer deposits 247 5.3 Chemical sedimentation – banded iron-formations, phosphorites, and evaporites 266 5.4 Fossil fuels – oil/gas formation and coali?cation 287 Summary and further reading 307 PART 4 GLOBAL TECTONICS AND METALLOGENY 6O RE DEPOSITS IN A GLOBAL TECTONIC CONTEXT 311 6.1 Introduction 311 6.2 Patterns in the distribution of mineral deposits 312 6.3 Continental growth rates 312 6.4 Crustal evolution and metallogenesis 315 6.5 Metallogeny through time 319 6.6 Plate tectonics and ore deposits – a summary 339 Summary and further reading 343 References 345 Index 368 vi CONTENTS ITOA01 09/03/2009 14:30 Page viThere are many excellent texts, available at both introductory and advanced levels, that describe the Earth’s mineral deposits. Several describe the deposits themselves and others do so in com- bination with explanations that provide an under- standing of how such mineral occurrences form. Few are dedicated entirely to the multitude of processes that give rise to the ore deposits of the world. The main purpose of this book is to provide a better understanding of the processes, as well as the nature and origin, of mineral occurrences and how they ?t into the Earth system. It is intended for use at a senior undergraduate level (third and fourth year levels), or graduate level (North America), and assumes a basic knowledge in a wide range of core earth science disciplines, as well as in chemistry and physics. Although meant to be introductory, it is reasonably comprehen- sive in its treatment of topics, and it is hoped that practicing geologists in the minerals and related industries will also ?nd the book useful as a sum- mary and update of ore-forming processes. To this end the text is punctuated by a number of boxed case studies in which actual ore deposits, selected as classic examples from around the world, are brie?y described to give context and relevance to processes being discussed in the main text. Metallogeny, or the study of the genesis of ore deposits in relation to the global tectonic paradigm, is a topic that traditionally has been, and should remain, a core component of the university earth science curriculum. It is also the discipline that underpins the training of professional earth scien- tists working in the minerals and related industries of the world. A tendency in the past has been to treat economic geology as a vocational topic and to provide instruction only to those individuals who wished to specialize in the discipline or to follow a career in the minerals industries. In more recent years, changes in earth science curricula have resulted in a trend, at least in a good many parts of the world, in which economic geology has been sidelined. A more holistic, process-orientated approach (earth systems science) has led to a wider appreciation of the Earth as a complex interrelated system. Another aim of this book, therefore, is to emphasize the range of processes responsible for the formation of the enormously diverse ore deposit types found on Earth and to integrate these into a description of Earth evolution and global tectonics. In so doing it is hoped that metallogenic studies will increasingly be reinte- grated into the university earth science curricula. Teaching the processes involved in the formation of the world’s diminishing resource inventory is necessary, not only because of its practical relevance to the real world, but also because such processes form an integral and informative part of the Earth system. This book was written mainly while on a protracted sabbatical in the Department of Earth Sciences at the University of Oxford. I am very grateful to John Woodhouse and the departmental staff who accommodated me and helped to provide the combination of academic rigor and quietitude that made writing this book such a pleasure. In particular Jenny Colls, Earth Science Librarian, was a tower of support in locating reference material. The “tea club” at the Banbury Road annexe pro- vided both stimulation and the requisite libations Preface ITOA01 09/03/2009 14:30 Page viito break the monotony. The staff at Blackwell managed to combine being really nice people with a truly professional attitude, and Ian Francis, Delia Sandford, Rosie Hayden, and Cee Pike were all a pleasure to work with. Dave Coles drafted all the diagrams and I am extremely grateful for his forebearance in dealing amiably with a list of ?gures that seemingly did not end. Several people took time to read through the manuscript for me and in so doing greatly improved both the style and content. They include John Taylor (copy- editing), Judith Kinnaird and Dave Waters (Intro- duction), Grant Cawthorn (Chapter 1), Philip Candela (Chapter 2), Franco Pirajno (Chapter 3), Michael Meyer (Chapter 4), John Parnell and Harold Reading (Chapter 5), and Mark Barley, Kevin Burke, and John Dewey (Chapter 6). The de?ciencies that remain, though, are entirely my own. A particularly debt of gratitude is owed to David Rickard, who undertook the onerous task of reviewing the entire manuscript; his lucid comments helped to eliminate a number of ?aws and omissions. Financial support for this project came from BHP Billiton in London and the Geo- logical Society of South Africa Trust. My col- leagues at Wits were extremely supportive during my long absences, and I am very grateful to Spike McCarthy, Paul Dirks, Carl Anhauesser, Johan Kruger, and Judith Kinnaird for their input in so many ways. Finally, my family, Vicki, Nicole, and Brendan, were subjected to a life-style that involved making personal sacri?ces for the fruition of this project – there is no way of saying thank you and it is to them that I dedicate this book. Laurence Robb Johannesburg viii PREFACE ITOA01 09/03/2009 14:30 Page viiiINTRODUCTION AND AIMS Given the unprecedented growth of human population over the past century, as well as the related increase in demand for and production of natural resources, it is evident that under- standing the nature, origin and distribution of the world’s mineral deposits remains a strategic topic. The discipline of “economic geology,” which covers all aspects pertaining to the description and understanding of mineral resources, is, there- fore, one which traditionally has been, and should remain, a core component of the university earth science curriculum. It is also the discipline that underpins the training of professional earth scient- ists working in the minerals and related indus- tries of the world. A tendency in the past has been to treat economic geology as a vocational topic and to provide instruction only to those individ- uals who wished to specialize in the discipline or to follow a career in the minerals industry. In more recent years, changes in earth science curricula have resulted in a trend, at least in a good many parts of the world, in which economic geology has been sidelined. The conceptual development of earth systems science, also a feature of the latter years of the twentieth century, has led to dramatic shifts in the way in which the earth sciences are taught. A more holistic, process-orientated approach has led to a much wider appreciation of the Earth as a complex interrelated system. The under- standing of feedback mechanisms has brought an appreciation that the solid Earth, its oceans and atmosphere, and the organic life forms that occupy niches above, at and below its surface, are intimately connected and can only be under- stood properly in terms of an interplay of pro- cesses. Examples include the links between global tectonics and climate patterns, and also between the evolution of unicellular organisms and the formation of certain types of ore deposits. In this context the teaching of many of the traditional geological disciplines assumes new relevance and the challenge to successfully teaching earth sys- tem science is how best to integrate the wide range of topics into a curriculum that provides under- standing of the entity. Teaching the processes involved in the formation of the enormously Introduction: mineral resources GENERAL INTRODUCTION AND AIMS OF THE BOOK A SIMPLE CLASSIFICATION SCHEME FOR MINERAL DEPOSITS SOME IMPORTANT DEFINITIONS metallogeny, syngenetic, epigenetic, mesothermal, epithermal, supergene, hypogene, etc. SOME RELEVANT COMPILATIONS periodic table of the elements tables of the main ore and gangue minerals geological time scale FACTORS THAT MAKE A VIABLE MINERAL DEPOSIT enrichment factors required to make ore deposits how are mineral resources and ore reserves de?ned? NATURAL RESOURCES AND THEIR FUTURE EXPLOITATION sustainability environmental responsibility ITOA02 09/03/2009 14:31 Page 1diverse ore deposit types found on Earth is neces- sary, not only because of its practical relevance to the real world, but also because such processes form an integral and informative part of the Earth’s evolution. The purpose of this process-orientated book is to provide a better understanding of the nature and origin of mineral occurrences and how they ?t into the Earth system. It is intended for use at a senior undergraduate level (third and fourth year levels), or a graduate level, and assumes a basic knowledge in a wide range of core earth science disciplines, as well as in chemistry and physics. It is also hoped that practicing geologists in the minerals and related industries will ?nd the book useful as a summary and update of ore-forming processes. To this end the text is punctuated by a number of boxed case studies in which actual ore deposits, selected as classic examples from around the world, are brie?y described to give context and relevance to processes being discussed in the main text. A CLASSIFICATION SCHEME FOR ORE DEPOSITS There are many different ways of categorizing ore deposits. Most people who have written about and described ore deposits have either unwit- tingly or deliberately been involved in their classi?cation. This is especially true of textbooks where the task of providing order and structure to a set of descriptions invariably involves some form of classi?cation. The best classi?cation schemes are probably those that remain as independent of genetic linkages as possible, thereby minimizing the scope for mistakes and controversy. Never- theless, genetic classi?cation schemes are ulti- mately desirable, as there is considerable advantage to having processes of ore formation re?ected in a set of descriptive categories. Guilbert and Park (1986) discuss the problem of ore deposit classi- ?cation at some length in Chapters 1 and 9 of their seminal book on the geology of ore deposits. They show how classi?cation schemes re?ect the development of theory and techniques, as well as the level of understanding, in the discipline. Given the dramatic improvements in the level of understanding in economic geology over the past few years, the Guilbert and Park (1986) classi- ?cation scheme, modi?ed after Lindgren’s (1933) scheme, is both detailed and complex, and be?ts the comprehensive coverage of the subject matter provided by their book. In a more recent, but equally comprehensive, coverage of ore deposits, Misra (2000) has opted for a categorization based essentially on genetic type and rock association, similar to a scheme by Meyer (1981). It is the asso- ciation between ore deposits and host rock that is particularly appealing for its simplicity, and that has been selected as the framework within which the processes described in this book are placed. Rocks are classi?ed universally in terms of a threefold subdivision, namely igneous, sediment- ary and metamorphic, that re?ects the funda- mental processes active in the Earth’s crust (Figure 1a). The scheme is universal because rocks are (generally!) recognizably either igneous or sedimentary, or, in the case of both precursors, have been substantially modi?ed to form a meta- morphic rock. Likewise, ores are rocks and can often be relatively easily attributed to an igneous or sedimentary/sur?cial origin, a feature that rep- resents a good basis for classi?cation. Such a clas- si?cation also re?ects the genetic process involved in ore formation, since igneous and sedimentary deposits are often syngenetic and formed at the same time as the host rock itself. Although many ores are metamorphosed, and whereas pressure and temperature increases can substantially mod- ify the original nature of ore deposits, it is evident that metamorphism does not itself represent a fundamental process whereby ore deposits are created. Hydrothermalism, however, is a viable analogue in ore-forming processes for metamorph- ism and also involves modi?cation of either igneous or sedimentary rocks, as well as heat (and mass) transfer and pressure ?uctuation. A very simple classi?cation of ores is, therefore, achieved on the basis of igneous, sedimentary/ sur?cial and hydrothermal categories (Figure 1b), and this forms the basis for the structure and lay- out of this book. This subdivision is very similar to one used by Einaudi (2000), who stated that all mineral deposits can be classi?ed into three types based on process, namely magmatic deposits, hydro- thermal deposits and sur?cial deposits formed by 2 INTRODUCTION: MINERAL RESOURCES ITOA02 09/03/2009 14:31 Page 2INTRODUCTION: MINERAL RESOURCES 3 (a) Rocks Igneous Metamorphic Sedimentary (b) Ore deposits Igneous Hydrothermal Sedimentary- surficial Part 1 Part 3 Part 2 Figure 1 Classi?cation of the principal rock types (a) and an analogous, but much simpli?ed, classi?cation of ore deposit types (b). Photographs show the interplay between ore forming processes. (c) Igneous ore type: the PGE- bearing Merensky Reef, Bushveld Complex, South Africa. This unit and the ores within it can be altered and redistributed by hydrothermal solutions. (d) Sedimentary ore type: Au- and U-bearing conglomerate from the Witwatersrand Basin, South Africa. Quartz veins cutting this unit attest to the action of later hydrothermal ?uids in the sequence. (e) Hydrothermal ore type: quartz-carbonate vein network in an Archean orogenic or lode-gold deposit from the Abitibi greenstone belt in Canada. The deposit is associated with igneous (lamprophyre) intrusions that may be implicated in the mineralization process. (c) (d) (e) ITOA02 09/03/2009 14:31 Page 3surface and groundwaters. One drawback of this type of classi?cation, however, is that ore-forming processes are complex and episodic. Ore formation also involves processes that evolve, sometimes over signi?cant periods of geologic time. For example, igneous processes become magmatic- hydrothermal as the intrusion cools and crys- tallizes, and sediments undergo diagenesis and metamorphism as they are progressively buried, with accompanying ?uid ?ow and alteration. In addition, deformation of the Earth’s crust intro- duces new conduits that also facilitate ?uid ?ow and promote the potential for mineralization in virtually any rock type. Ore-forming processes can, therefore, span more than one of the three categories, and there is considerable overlap be- tween igneous and hydrothermal and between sedimentary and hydrothermal, as illustrated diagrammatically in Figure 1b, and also in the accompanying photographs of the three major categories of ore types. The main part of this book is subdivided into three sections termed Igneous (Part 1), Hydrothermal (Part 2), and Sedimentary/sur?cial (Part 3). Part 1 comprises Chapters 1 and 2, which deal with igneous and magmatic-hydrothermal ore-forming processes respectively. Part 2 con- tains Chapter 3 and covers the large and diverse range of hydrothermal processes not covered in Part 1. Part 3 comprises Chapter 4 on sur?cial and supergene processes, as well as Chapter 5, which covers sedimentary ore deposits, including a short section on the fossil fuels. The ?nal chapter of the book, Chapter 6, is effectively an addendum to this threefold subdivision and is an attempt to describe the distribution of ore deposits, both spa- tially in the context of global tectonics and tem- porally in terms of crustal evolution through Earth history. This chapter is considered relevant in this day and age because the plate tectonic paradigm, which has so pervasively in?uenced geological thought since the early 1970s, provides another conceptual basis within which to classify ore deposits. In fact, modern economic geology, and the scienti?c exploration of mineral deposits, is now ?rmly conceptualized in terms of global tectonics and crustal evolution. Although there is still a great deal to be learnt, the links between plate tectonics and ore genesis are now suf?ci- ently well established that studies of ore deposits are starting to contribute to a better understand- ing of the Earth system. WHAT MAKES A VIABLE MINERAL DEPOSIT? Ore deposits form when a useful commodity is suf?ciently concentrated in an accessible part of the Earth’s crust so that it can be pro?tably extracted. The processes by which this concentra- tion occurs are the topic of this book. As an intro- duction it is pertinent to consider the range of concentration factors that characterize the forma- tion of different ore deposit types. Some of the strategically important metals, such as Fe, Al, Mg, Ti, and Mn, are abundantly distributed in the Earth’s crust (i.e. between about 0.5 and 10%) and only require a relatively small degree of enrich- ment in order to make a viable deposit. Table 1 shows that Fe and Al, for example, need to be concentrated by factors of 9 and 4 respectively, relative to average crustal abundances, in order to form potentially viable deposits. By contrast, base metals such as Cu, Zn, and Ni are much more sparsely distributed and average crustal abundances are only in the range 55–75 parts per million (ppm). The economics of mining dictate that these metals need to be concentrated by factors in the hundreds in order to form poten- 4 INTRODUCTION: MINERAL RESOURCES Table 1 Average crustal abundances for selected metals and typical concentration factors that need to be achieved in order to produce a viable ore deposit Average Typical Approximate crustal exploitable concentration abundance grade factor Al 8.2% 30% ×4 Fe 5.6% 50% ×9 Cu 55 ppm 1% ×180 Ni 75 ppm 1% ×130 Zn 70 ppm 5% ×700 Sn 2 ppm 0.5% ×2500 Au 4 ppb 5 g t -1 ×1250 Pt 5 ppb 5 g t -1 ×1000 Note: 1 ppm is the same as 1 g t -1 . ITOA02 09/03/2009 14:31 Page 4tially viable deposits, degrees of enrichment that are an order of magnitude higher than those applic- able to the more abundant metals. The degree of concentration required for the precious metals is even more demanding, where the required enrich- ment factors are in the thousands. Table 1 shows that average crustal abundances for Au and Pt are in the range 4–5 parts per billion (ppb) and even though ore deposits routinely extract these metals at grades of around 5g t -1 , the enrichment factors involved are between 1000 and 1250 times. Another useful way to distinguish between the geochemically abundant and scarce metals is to plot average crustal abundances against produc- tion estimates. This type of analysis was ?rst car- ried out by Skinner (1976), who used a plot like that in Figure 2 to con?rm that crustal abundance is a reasonable measure of the availability of a given resource. It is by design and of necessity that we use more of the geochemically abundant metals than we do the scarce ones. The nature of our technologies and the materials we use to man- ufacture mechanical items depend in large meas- ure on the availability of raw materials. As an example, the technologies (geological and metal- lurgical) that resulted in the dramatic increase in global aluminum production over the latter part of the twentieth century allowed iron to be replaced by aluminum in many products such as motor vehicles. More importantly, though, Figure 2 allows estimates to be made of the relative rates of depletion of certain metals relative to others. These trends are discussed again below. Mineral resources and ore reserves In the course of this book reference is made to the term “ore deposit” with little or no consideration of whether such occurrences might be economic- ally viable. Although such considerations might seem irrelevant in the present context it is neces- sary to emphasize that professional institutions now insist on the correct de?nition and usage of terminology pertaining to exploration results, mineral resources, and ore reserves. Such termi- nology should be widely used and applied, as it would help in reducing the irresponsible, and some- times even fraudulent, usage of terminology, espe- cially with respect to the investor public. Correct terminology can also assist in the description and identi?cation of genuine ore deposits from zones of marginal economic interest or simply anomal- ous concentrations of a given commodity. Although the legislation that governs the pub- lic reporting of mineral occurrences obviously varies from one country to the next, there is INTRODUCTION: MINERAL RESOURCES 5 0 Tons produced (1992) 10 9 10 0 Crustal abundance (wt%) 10 7 10 6 10 5 10 4 10 3 10 2 10 1 10 8 10 –1 10 –2 10 –3 10 –4 10 –5 10 –6 Au Bi Hg PGE Ag Se Sb Cd Mo W Sn As Pb Ca Cr Zn Ni Co S Mn Ti Mg Al Fe 10 1 Figure 2 Plot of crustal abundances against global production for an number of metal commodities (after Einaudi, 2000). The line through Fe can be regarded as a datum against which the rates of production of the other metals can be compared in the context of crustal abundances. ITOA02 09/03/2009 14:31 Page 5now reasonable agreement as to the de?nition of terms. In general it is agreed that different terms should apply to mineral occurrences depending on the level of knowledge and degree of con?d- ence that is associated with the measurement of its quantity. Figure 3 is a matrix that re?ects the terminology associated with an increased level of geological knowledge and con?dence, and modifying factors such as those related to mining techniques, metallurgical extraction, mar- keting, and environmental reclamation. Explora- tion results can be translated into a mineral resource once it is clear that an occurrence of intrinsic economic interest exists in such form and quantity that there are reasonable prospects for its eventual exploitation. Such a resource can only be referred to as an ore reserve if it is a part of an economically extractable measured or indicated mineral resource. One problem with this terminology is that what is economically extract- able in a Third World artisinal operation may not of course be viable in a technically developed First World economy, and vice versa. The term “ore deposit” has no signi?cance in the professional description of a mineral occurrence and is best used as a simply descriptive or generic term. SOME USEFUL DEFINITIONS AND COMPILATIONS Some general de?nitions This section is not intended to provide a com- prehensive glossary of terms used in this book. There are, however, several terms that are used throughout the text where a de?nition is either useful or necessary in order to avoid ambiguity. The following de?nitions are consistent with those provided in the Glossary of Geology (Bates and Jackson, 1987) and The Encyclopedia of the Solid Earth Sciences (Kearey, 1993). • Ore: any naturally occurring material from which a mineral or aggregate of value can be extracted at a pro?t. In this book the concept extends to coal (a combustible rock comprising more than 50% by weight carbonaceous material) and petroleum (naturally occurring hydrocarbon in gaseous, liquid, or solid state). • Syngenetic: refers to ore deposits that form at the same time as their host rocks. In this book this includes deposits that form during the early stages of sediment diagenesis. • Epigenetic: refers to ore deposits that form after their host rocks. • Hypogene: refers to mineralization caused by ascending hydrothermal solutions. • Supergene: refers to mineralization caused by descending solutions. Generally refers to the enrichment processes accompanying the weath- ering and oxidation of sul?de and oxide ores at or near the surface. • Metallogeny: the study of the genesis of min- eral deposits, with emphasis on their relation- ships in space and time to geological features of the Earth’s crust. • Metallotect: any geological, tectonic, litholo- gical, or geochemical feature that has played a role 6 INTRODUCTION: MINERAL RESOURCES Increasing level of knowledge and confidence Exploration results Mineral resources (Reported as in situ mineralization estimates) Inferred Indicated Measured Mineral reserves (Reported as mineable production estimates) Probable Proved Mining, metallurgical, economic, marketing, legal, environmental, social, and governmental factors (Factors that contribute to the successful exploitation of a deposit) Figure 3 Simpli?ed scheme illustrating the conceptual difference between mineral resources and ore reserves as applied to mineral occurrences. The scheme forms the basis for the professional description of ore deposits as de?ned by the Australian and South African Institutes of Mining and Metallurgy. ITOA02 09/03/2009 14:31 Page 6in the concentration of one or more elements in the Earth’s crust. • Metallogenic Epoch: a unit of geologic time favorable for the deposition of ores or character- ized by a particular assemblage of deposit types. • Metallogenic Province: a region characterized by a particular assemblage of mineral deposit types. • Epithermal: hydrothermal ore deposits formed at shallow depths (less than 1500 meters) and fairly low temperatures (50–200 °C). • Mesothermal: hydrothermal ore deposits formed at intermediate depths (1500–4500 meters) and temperatures (200–400 °C). • Hypothermal: hydrothermal ore deposits formed at substantial depths (greater than 4500 meters) and elevated temperatures (400–600 °C). Periodic table of the elements The question of the number of elements present on Earth is a dif?cult one to answer. Most of the element compilations relevant to the earth sci- ences show that there are 92 elements, the major- ity of which occur in readily detectable amounts in the Earth’s crust. Figure 4 shows a periodic table in which these elements are presented in ascending atomic number and also categorized into groupings that are relevant to metallogenesis. There are in fact as many as 118 elements known to man, but those with atomic numbers greater than 92 (U: uranium) either occur in vanishingly small amounts as unstable isotopes that are the products of various natural radioactive decay reactions or are synthetically created in nuclear reactors. The heaviest known element, ununoc- tium (Uuo, atomic number 118), has been only transiently detected in a nuclear reactor and its actual existence is still conjectural. Some of the heavy, unstable elements are, however, manu- factured synthetically and serve a variety of uses. Plutonium (Pu, atomic number 94), for example, is manufactured in fast breeder reactors and is INTRODUCTION: MINERAL RESOURCES 7 5 B 33 As 34 Se 52 Te 83 Bi 50 Sn 48 Cd 47 Ag 30 Zn 29 Cu 28 Ni 46 Pd 80 Hg 79 Au 78 Pt 27 Co 45 Rh 26 Fe 25 Mn 44 Ru 76 Os 74 W 42 Mo 41 Nb Atmophile 1 H 3 Li 11 Na 19 K 37 Rb 55 Cs 87 Fr 4 Be 12 Mg 20 Ca 38 Sr 56 Ba 88 Ra 21 Sc 39 Y 57 La 89 Ac 22 Ti 40 Zr 72 Hf 23 V 73 Ta 58 Ce 24 Cr 59 Pr 43 Tc 75 Re 60 Nd 61 Pm 77 Ir 62 Sm 63 Eu 64 Gd 65 Tb 13 AI 31 Ga 49 In 81 TI 66 Dy 6 C 14 Si 32 Ge 82 Pb 67 Ho 7 N 15 P 51 Sb 68 Er 8 O 16 S 84 Po 69 Tm 9 F 17 Cl 35 Br 53 I 85 At 70 Yb 10 Ne 18 Ar 36 Kr 54 Xe 86 Rn 71 Lu 90 Th 91 Pa 92 U 2 He Metals Metalloids Non-metals Oxide Oxide or sul?de Sulfide Native metal/alloy Lithophile Rb- Chalcophile Cu- Siderophile Au- Ne- Generally decreasing electronegativity Generally decreasing electronegativity No known use Ore mineral generally as:- Figure 4 Periodic table showing the 92 geologically relevant elements classi?ed on the basis of their rock and mineral associations. ITOA02 09/03/2009 14:31 Page 7used as a nuclear fuel and in weapons manufacture. Americium (Am, atomic number 95) is also man- ufactured in reactors and is widely used as the active agent in smoke detectors. Of the 92 elements shown in Figure 4, almost all have some use in our modern technologically driven societies. Some of the elements (iron and aluminum) are required in copious quantities as raw materials for the manufacture of vehicles and in construction, whereas others (the rare earths, for example) are needed in very much smaller amounts for use in the alloys and electronics industries. Only three elements appear at this stage to have little or no use at all (Figure 4). These are astatine (At, atomic number 85), francium (Fr, atomic number 87), and protactinium (Pa, atomic number 91). Francium is radioactive and so short- lived that only some 20–30 g exists in the entire Earth’s crust at any one time! Astatine, likewise, is very unstable and exists in vanishingly small amounts in the crust, or is manufactured synthet- ically. Radon (Rn, atomic number 86) is an inert or noble gas that is formed as a radioactive decay product of radium. It has limited use in medical applications, but, conversely, if allowed to accu- mulate can represent a serious health hazard in certain environments. The useful elements can be broadly subdivided in a number of different ways. Most of the ele- ments can be classi?ed as metals (Figure 4), with a smaller fraction being non-metals. The elements B, Si, As, Se, Te, and At have intermediate proper- ties and are referred to as metalloids. Another classi?cation of elements, attributed to the pion- eering geochemist Goldschmidt, is based on their rock associations and forms the basis for distinguishing between lithophile (associated with silicates and concentrated in the crust), chalco- phile (associated with sul?des), siderophile (occur as the native metal and concentrated in the core), and atmophile (occur as gases in the atmosphere) elements. It is also useful to consider elements in terms of their ore mineral associations, with some preferentially occurring as sul?des and others as oxides (see Figure 4). Some elements have proper- ties that enable them to be classi?ed in more than one way and iron is a good example, in that it occurs readily as both an oxide and sul?de. Common ore and gangue minerals It is estimated that there are about 3800 known minerals that have been identi?ed and classi?ed (Battey and Pring, 1997). Only a very small propor- tion of these make up the bulk of the rocks of the Earth’s crust, as the common rock forming min- erals. Likewise, a relatively small number of min- erals make up most of the economically viable ore deposits of the world. The following compilation is a breakdown of the more common ore minerals in terms of chemical classes based essentially on the anionic part of the mineral formula. Also included are some of the more common “gangue,” which are those minerals that form part of the ore body, but do not contribute to the economically extractable part of the deposit. Most of these are alteration assemblages formed during hydrother- mal processes. The compilation, including ideal chemical formulae, is subdivided into six sec- tions, namely native elements, halides, sul?des and sulfo-salts, oxides and hydroxides, oxy-salts (such as carbonates, phosphates, tungstates, sul- fates), and silicates. More detailed descriptions of both ore and gangue minerals can be found in a variety of mineralogical texts, such as Deer et al. (1982), Berry et al. (1983), and Battey and Pring (1997). More information on ore mineral textures and occurrences can be found in Craig and Vaughan (1994) and Ixer (1990). 1 Native elements Both metals and non-metals exist in nature in the native form, where essentially only one element exists in the structure. Copper, silver, gold, and platinum are all characterized by cubic close packing of atoms, have high densities, and are malleable and soft. The carbon atoms in diamond are linked in tetrahedral groups forming well cleaved, very hard, translucent crystals. Sulfur oc- curs as rings of eight atoms and forms bipyramids or is amorphous. Metals Gold – Au Silver – Ag Platinum – Pt 8 INTRODUCTION: MINERAL RESOURCES ITOA02 09/03/2009 14:31 Page 8Palladium – Pd Copper – Cu Non-metals Sulfur – S Diamond – C Graphite – C 2 Halides The halide mineral group comprises compounds made up by ionic bonding. Minerals such as halite and sylvite are cubic, have simple chemical for- mulae, and are highly soluble in water. Halides sometimes form as ore minerals, such as chlorar- gyrite and atacamite. Halite – NaCl Sylvite – KCl Chlorargyrite – AgCl Fluorite – CaF 2 Atacamite – Cu 2 Cl(OH) 3 3 Sul?des and sulfo-salts This is a large and complex group of minerals in which bonding is both ionic and covalent in char- acter. The sul?de group has the general formula A M X P , where X, the larger atom, is typically S but can be As, Sb, Te, Bi, or Se, and A is one or more metals. The sulfo-salts, which are much rarer than sul?des, have the general formula A M B N X P , where A is commonly Ag, Cu, or Pb, B is com- monly As, Sb, or Bi, and X is S. The sul?de and sulfo-salt minerals are generally opaque, heavy and have a metallic to sub-metallic lustre. Sul?des Chalcocite – Cu 2 S Bornite – Cu 5 FeS 4 Galena – PbS Sphalerite – ZnS Chalcopyrite – CuFeS 2 Pyrrhotite – Fe 1–x S Pentlandite – (Fe,Ni) 9 S 8 Millerite – NiS Covellite – CuS Cinnabar – HgS Skutterudite – (Co,Ni)As 3 Sperrylite – PtAs 2 Braggite/cooperite – (Pt,Pd,Ni)S Moncheite – (Pt,Pd)(Te,Bi) 2 Cobaltite – CoAsS Gersdorf?te – NiAsS Loellingite – FeAs 2 Molybdenite – MoS 2 Realgar – AsS Orpiment – As 2 S 3 Stibnite – Sb 2 S 3 Bismuthinite – Bi 2 S 3 Argentite – Ag 2 S Calaverite – AuTe 2 Pyrite – FeS 2 Laurite – RuS 2 Sulfo-salts Tetrahedrite – (Cu,Ag) 12 Sb 4 S 13 Tennantite – (Cu,Ag) 12 As 4 S 13 Enargite – Cu 3 AsS 4 4 Oxides and hydroxides This group of minerals is variable in its properties, but is characterized by one or more metal in combination with oxygen or a hydroxyl group. The oxides and hydroxides typically exhibit ionic bonding. The oxide minerals can be hard, dense, and refractory in nature (magnetite, cassiterite) but can also be softer and less dense, forming as products of hydrothermal alteration and weather- ing (hematite, anatase, pyrolucite). Hydroxides, such as goethite and gibbsite, are typically the products of extreme weathering and alteration. Oxides Cuprite – Cu 2 O Hematite – Fe 2 O 3 Ilmenite – FeTiO 3 Hercynite – FeAl 2 O 4 Gahnite – ZnAl 2 O 4 Magnetite – Fe 3 O 4 Chromite – FeCr 2 O 4 Rutile – TiO 2 Anatase – TiO 2 Pyrolucite – MnO 2 Cassiterite – SnO 2 Uraninite – UO 2 INTRODUCTION: MINERAL RESOURCES 9 ITOA02 09/03/2009 14:31 Page 9Thorianite – ThO 2 Columbite-tantalite – (Fe,Mn)(Nb,Ta) 2 O 6 Hydroxides (or oxyhydroxides) Goethite – FeO(OH) Gibbsite – Al(OH) 3 Boehmite – AlO(OH) Manganite – MnO(OH) 5 Oxy-salts The carbonate group of minerals form when anionic carbonate groups (CO 3 2- ) are linked by intermediate cations such as Ca, Mg, and Fe. Hydroxyl bearing and hydrated carbonates can also form, usually as a result of weathering and alteration. The other oxy-salts, such as the tung- states, sulfates, phosphates, and vanadates, are analogous to the carbonates, but are built around an anionic group of the form XO 4 n- . Carbonates Calcite – CaCO 3 Dolomite – CaMg(CO 3 ) 2 Ankerite – CaFe(CO 3 ) 2 Siderite – FeCO 3 Rhodochrosite – MnCO 3 Smithsonite – ZnCO 3 Cerussite – PbCO 3 Azurite – Cu 3 (OH) 2 (CO 3 ) 2 Malachite – Cu 2 (OH) 2 CO 3 Tungstates Scheelite – CaWO 4 Wolframite – (Fe,Mn)WO 4 Sulfates Baryte(s) – BaSO 4 Anhydrite – CaSO 4 Alunite – KAl 3 (OH) 6 (SO 4 ) 2 Gypsum – CaSO 4 .2H 2 O Epsomite – MgSO 4 .7H 2 O Phosphates Xenotime – YPO 4 Monazite – (Ce,La,Th)PO 4 Apatite – Ca 5 (PO 4 ) 3 (F,Cl,OH) Vanadates Carnotite – K 2 (UO 2 )(VO 4 ) 2 .3H 2 O 6 Silicates The bulk of the Earth’s crust and mantle is made up of silicate minerals that can be subdivided into several mineral series based on the structure and coordination of the tetrahedral SiO 4 4- anionic group. Silicate minerals are generally hard, refract- ory and translucent. Most of them cannot be regarded as ore minerals in that they do not repres- ent the extractable part of an ore body, and the list provided below shows only some of the silic- ates more commonly associated with mineral occurrences as gangue or alteration products. Some silicate minerals, such as zircon and spo- dumene, are ore minerals and represent important sources of metals such as zirconium and lithium, respectively. Others, such as kaolinite, are mined for their intrinsic properties (i.e. as a clay for the ceramics industry). Tekto (framework) Quartz – SiO 2 Orthoclase – (K,Na)AlSi 3 O 8 Albite – (Na,Ca)AlSi 3 O 8 Scapolite – (Na,Ca) 4 [(Al,Si) 4 O 8 )] 3 (Cl, CO 3 ) Zeolite (analcime) – NaAlSi 2 O 6 .H 2 O Neso (ortho) Zircon – Zr(SiO 4 ) Garnet (almandine) – Fe 3 Al 2 (SiO 4 ) 3 Garnet (grossular) – Ca 3 Al 2 (SiO 4 ) 3 Sillimanite – Al 2 SiO 5 Topaz – Al 2 SiO 4 (F,OH) 2 Chloritoid – (Fe,Mg,Mn) 2 (Al,Fe)Al 3 O 2 (SiO 4 ) 2 (OH) 4 Cyclo (ring) Beryl – Be 3 Al 2 Si 6 O 18 Tourmaline – (Na,Ca)(Mg,Fe,Mn,Al) 3 (Al,Mg,Fe) 6 Si 6 O 18 (BO 3 ) 3 (OH,F) 4 Soro (di) Lawsonite – CaAl 2 Si 2 O 7 (OH) 2 .H 2 O Epidote – Ca 2 (Al,Fe) 3 Si 3 O 12 (OH) 10 INTRODUCTION: MINERAL RESOURCES ITOA02 09/03/2009 14:31 Page 10Phyllo (sheet) Kaolinite – Al 4 Si 4 O 10 (OH) 8 Montmorillonite – (Na,Ca) 0.3 (Al,Mg) 2 Si 4 O 10 (OH) 2 .nH 2 O Illite – KAl 2 (Si,Al) 4 O 10 (H 2 O)(OH) 2 Pyrophyllite – Al 2 Si 4 O 10 (OH) 2 Talc – Mg 3 Si 4 O 10 (OH) 2 Muscovite – KAl 2 (AlSi 3 O 10 )(OH) 2 Biotite – K(Fe,Mg) 3 (Al,Fe)Si 3 O 10 (OH,F) 2 Lepidolite – K(Li,Al) 3 (Si,Al) 4 O 10 (OH,F) 2 Chlorite – (Fe,Mg,Al) 5–6 (Si,Al) 4 O 10 (OH) 8 Ino (chain) Tremolite-actinolite – Ca 2 (Fe,Mg) 5 Si 8 O 22 (OH) 2 Spodumene – LiAlSi 2 O 6 Wollastonite – CaSiO 3 Unknown structure Chrysocolla – (Cu,Al) 2 H 2 Si 2 O 5 (OH) 4 .nH 2 O Geological time scale The development of a geological time scale has been the subject of a considerable amount of thought and research over the past few decades and continues to occupy the minds and activities of a large number of earth scientists around the world. The de?nition of a framework within which to describe the secular evolution of rocks, and hence the Earth, has been, and continues to be, a contentious exercise. The International Commission on Stratigraphy (ICS is a working group of the International Union of Geological Sciences: IUGS) has been given the task of formal- izing the geological time scale and this work is ongoing ( In this book reference is often made to the timing of various events and processes and the provision of a time scale to which the reader can refer is, there- fore, useful. Figure 5 is a time scale based on the 2000 edition of the International Stratigraphic Chart published and sanctioned by the ICS and IUGS. In this diagram global chronostratigraphic terms are presented in terms of eons, eras, periods, and epochs, and de?ned by absolute ages in mil- lions of years before present (Ma). Also shown are the positions on the time scale of many of the ore deposits or metallogenic provinces referred to in the text. NATURAL RESOURCES, SUSTAINABILITY, AND ENVIRONMENTAL RESPONSIBILITY One of the major issues that characterized so- cial and economic development toward the end of the twentieth century revolved around the widespread acceptance that the Earth’s natural resources are ?nite, and that their exploitation should be carried out in a manner that will not detrimentally affect future generations. The con- cept of “sustainable development” in terms of the exploitation of mineral occurrences implies that future social and economic practice should endeavor not to deplete natural resources to the point where the needs of the future cannot be met. This would seem to be an impossible goal given the unprecedented population growth over the past century and the fact that many commod- ities will become depleted within the next 100 years. The challenge for commodity supply over the next century is a multifaceted one and will require a better understanding of the earth sys- tem, improved incentives to promote more ef?cient recycling of existing resources, and the means to ?nd alternative sources for commodit- ies that are in danger of serious depletion. There has been a dramatic rise in global popula- tion over the past 150 years. The number of humans on Earth has risen from 1 billion in 1830 to 6 billion at the end of the twentieth century. Most predictions suggest that the populations of most countries will start to level off over the next 30 years and that global numbers will stabilize at around 11 billion people by the end of the twenty- ?rst century. Societies in the next 100 years are, nevertheless, facing a scenario in which the demand for, and utilization of, natural resources continues to increase and certain commodities might well become depleted in this interval. Production trends for commodities such as oil, bauxite, copper, and gold (Figure 6) con?rm that demand for resources mirrors population growth and is likely to continue to do so over the next few INTRODUCTION: MINERAL RESOURCES 11 ITOA02 09/03/2009 14:31 Page 110.01 Ma 1.81 EON ERA PERIOD EPOCH QUAT. Holocene Pliocene 5.32 Pleistocene Miocene 23.8 NEOGENE Oligocene 33.7 Eocene 55.0 Paleocene 65.5 Late PALEOGENE Early CRETACEOUS Late Middle 142 PHANEROZOIC CENOZOIC MESOZOIC JURASSIC Early 205 251 Early Middle Late TRIASSIC Hishikari El Laco Kasuga Orange River diamonds Supergene/exotics El Salvador La Escondida/ Chuquicamata Carlin Skaergaard Los Pijiguaos MacTung/Troodos Orapa Andean magmatism (west east progression) WEST 251 Ma EON ERA PERIOD EPOCH Lopingian Guadalup- ian Cisuralian 292 Pennsylv- anian Mississipp- ian 354 Late PERMIAN Early DEVONIAN Wenlock Llandovery 417 PHANEROZOIC PALEOZOIC SILURIAN Early 495 543 Early Middle Late CAMBRIAN Coal measures Cornubian batholith Red Dog Lachlan S-and I-type granites CARBONIFEROUS Middle Pridoli 440 Middle Viburnum Trend Ludlow Rossing .. 543 Ma EON ERA NED- PROTEROZOIC MESO- PROTEROZOIC 1000 1600 2500 NEO- ARCHEAN MESO- ARCHEAN 2800 PRE CAMBRIAN ARCHEAN Witwatersrand Bikita PROTEROZOIC PALEO- PROTEROZOIC PALEO- ARCHEAN EO- ARCHEAN 3200 Kambalda Golden Mile Great Dyke Hamersley BIF Bushveld/ Phalaborwa Sudbury Kiruna 3600 (Not to scale) 4560 Ma (origin of solar system) Olympic Dam Argyle Zambian Copperbelt Late EAST Main Arabian Gulf reservoirs ORDOVICIAN Figure 5 Geological time scale after the ICS (2000). Also shown are the ages of the various deposits and metallogenic provinces mentioned in the book. ITOA02 09/03/2009 14:31 Page 12decades. World oil production increased precip- itously until the late 1970s, but since then a variety of political and economic factors have contributed to tempering production (Figure 6a), thereby ensuring a longer-term reserve base. A similar levelling of production is evident for baux- ite (Figure 6b) but such a trend is not yet evident for the precious metals such as gold or platinum. For some commodities, such as copper (Figure 6c), the world reserve base is also levelling off, a fea- ture that in part also re?ects fewer new and large discoveries. Critical shortages of most natural commodities are not likely to present a problem during the early part of the twenty-?rst century (Einaudi, 2000), but this situation will deteriorate unless strategies for sustainability are put into place immediately. The depletion of commodities in the Earth’s crust is particularly serious for those metals that are already scarce in terms of crustal abundances and for which high degrees of enrichment are required in order to make viable ore deposits. Figure 2 illustrates the point by referring to the production of iron as a baseline measure against INTRODUCTION: MINERAL RESOURCES 13 Figure 6 Global production trends for oil (a), bauxite (b), copper (c), and gold (d) over the twentieth century (after compilations in Craig et al., 1996). 30 25 20 15 10 5 Billion barrels per year Oil (a) 1900 1920 1940 1960 1980 2000 USA World 120 110 100 60 40 20 Million tons Bauxite (b) 1940 90 80 70 50 30 10 1950 1960 1970 1980 1990 600 500 400 300 200 100 Million tons Copper (c) 1950 1960 1970 1980 1990 World reserve base 2000 Tons Gold (d) 1500 1000 500 1860 1980 World cumulative production 1880 1900 1920 1940 1960 2000 Year Year Year Year ITOA02 09/03/2009 14:31 Page 13which extraction of other metals can be compared (Skinner, 1976). Those elements which fall above the Fe production line (notably Au, Ag, Bi, Sb, Sn, Cu, Pb, and Zn) are being extracted or depleted at faster rates, relative to their crustal abundances, than Fe. It is these metals that are in most danger of depletion in the next 50 years or so unless production is ameliorated or the reserve base is replaced. Conversely, those metals that plot beneath the Fe production line (such as Ti, Mg, and Al) are being extracted at slower rates than Fe and are in less danger of serious depletion during this century. One of the ways in which metallogeny can assist in the creation of a sustainable pattern of resource utilization is to better understand the processes by which ores are concentrated in the Earth’s crust. The replacement of the global com- modity reserve base is obviously dependent on exploration success and the ability to ?nd new ore deposits that can replace those that are being depleted. It is, of course, increasingly dif?cult to ?nd new and large deposits of conventional ores, since most of the accessible parts of the globe have been extensively surveyed and assessed for their mineral potential. The search for deeper deposits is an option but this is dependent to a large extent on the availability of technologies that will enable mining to take place safely and pro?tably at depths in excess of 4000 meters (currently the deepest level of mining in South African gold mines). Another option is to extract material from inaccessible parts of the globe, such as the ocean ?oor, a proposal that has received serious consid- eration with respect to metals such as Mn and Cu. Again, there are technological barriers to such processes at present, but these can be overcome, as demonstrated by the now widespread explo- ration for, and extraction of, oil and gas from the sea ?oor. A third option to improve the sustain- ability of resource exploitation is to extract useful commodities from rocks that traditionally have not been thought of as viable ores. Such a develop- ment can only be achieved if the so-called “miner- alogical barrier” (Skinner, 1976) is overcome. This concept can be described in terms of the amount of energy (or cost) required to extract a commod- ity from its ore. It is, for example, considerably cheaper to extract Fe from a banded iron- formation than it is from olivine or orthopyrox- ene in an igneous rock, even though both rock types might contain signi?cant amounts of the metal. The economics of mining and the wide- spread availability of banded iron-formation dict- ate that extraction of Fe from silicate minerals is essentially not feasible. The same is not true of nickel. Although it is cheaper and easier to extract Ni from sul?de ore minerals (such as pentlandite) there is now widespread extraction of the metal from nickeliferous silicate minerals (garnierite) that form during the lateritic weathering of ultrama?c rocks. Even though Ni is more dif?cult and expensive to extract from laterite than from sul?de ores, the high tonnages and grades, as well as the widespread development and ease of access of the former, mean that they represent viable mining propositions despite the extractive dif?- culties. Ultimately, it may also become desirable to consider mining iron laterites, but this would only happen if conventional banded iron-formation hosted deposits were depleted, or if the economics of the whole operation favored laterites over iron- formations. This is not likely to happen in the short term, but, if planned for, the scenario does offer hope for sustainability in the long term. In short, sustainable production of mineral resources requires a thorough understanding of ore-forming processes and the means to apply these to the discovery of new mineral occurrences. It also requires the timely development of technologies, both in the earth sciences and in related ?elds of mining and extractive metallurgy, that will enable alternative supplies of mineral resources to be economically exploited in the future. Mining and environmental responsibility A global population of 11 billion by the end of the century presents a major problem in terms of the supply of most of the world’s mineral resources. What is even more serious, though, is the enorm- ous strain it will place on the Earth’s fragile envir- onment arising from the justi?able expectation that future technologically advanced societies will provide an adequate standard of living, in terms of food, water, housing, recreation, and 14 INTRODUCTION: MINERAL RESOURCES ITOA02 09/03/2009 14:31 Page 14material bene?ts, to all their peoples. In addition to commodity supply problems, the twenty-?rst century will be also be characterized by unpre- cedented depletion of even more critical resources in the form of soil, water, and clean air (Fyfe, 2000). Legislation that is aimed at dealing with issues such as atmospheric pollution and green- house gas emissions, factory waste and acid drainage, the burning and destruction of forests, the protection of endangered species, overgrazing, and erosion is highly desirable but far from glob- ally applicable because it is perceived as a luxury that only the developed world can afford. The study of ore-forming processes is occa- sionally viewed as an undesirable topic that ultimately contributes to the exploitation of the world’s precious natural resources. Nothing could be further from the truth. An understanding of the processes by which metals are concentrated in the Earth’s crust is essential knowledge for anyone The discipline of “economic geology” and in par- ticular the ?eld of metallogeny (the study of the genesis of ore deposits) remains critical to the teach- ing of earth systems science. A holistic approach involving the integration of knowledge relevant to the atmosphere, biosphere, and lithosphere is now regarded as essential to understanding the complexities of the earth system. The develop- ment of environmentally responsible and sustain- able policies with respect to the future supply of all natural resources will demand a thorough Blunden, J. (1983) Mineral Resources and Their Man- agement. Harlow: Longman, 302 pp. Craig, J.R., Vaughan, D.J. and Skinner, B.J. (1996) Re- sources of the Earth – Origin, Use and Environmental Impact. Englewood Cliffs, NJ: Prentice Hall, 472 pp. Ernst, W.G. (2000) Earth Systems – Processes and Issues. Cambridge: Cambridge University Press, 559 pp. Kesler, S.E. (1994) Mineral Resources, Economics and the Environment. London: Macmillan, 400 pp. knowledge of the nature and workings of the earth system. Central to this is an understanding of metallogeny and the nature and origin of the entire spectrum of mineral resources, including the fossil fuels. The classi?cation of mineral deposits in terms of process can be simply and effectively achieved in terms of rock associations, namely igneous, hydrothermal, and sedimentary. This breakdown forms the basis for the layout of this book. concerned with the preservation and remediation of the environment. The principles that underpin the natural concentration of ores in the crust are the same as those that can be utilized in issues such as the control of acid mine drainage, and soil and erosion management. Mining operations around the world are increasingly having to assume responsibility for reclamation of the land- scape once the resource has been depleted. The industry now encompasses a range of activities extending from geological exploration and evalu- ation, through mining and bene?ciation, and eventually to environmental reclamation. This is the mining cycle and its effective management in the future will be a multidisciplinary exercise carried out by highly skilled scientists and engin- eers. Earth systems science, and in particular the geological processes that gave rise to the natural concentration of the ore in the ?rst place, will be central to this entire operation. INTRODUCTION: MINERAL RESOURCES 15 ITOA02 09/03/2009 14:31 Page 15ITOA02 09/03/2009 14:31 Page 16Igneous Processes ITOC01 09/03/2009 14:37 Page 17ITOC01 09/03/2009 14:37 Page 181.1 INTRODUCTION Igneous rocks host a large number of different ore deposit types. Both ma?c and felsic rocks are linked to mineral deposits, examples of which range from the chromite ores resulting from crystal fractionation of ma?c magmas to tin deposits associated with certain types of granites. The processes described in this chapter relate to properties that are intrinsic to the magma itself and can be linked genetically to its cooling and solidi?cation. Discussion of related pro- cesses, whereby an aqueous ?uid phase forms or “exsolves” from the magma as it crystallizes, is placed in Chapter 2. The topics discussed under the banners of igneous and magmatic– hydrothermal ore-forming processes are intimately linked and form Part 1 of this book. A measure of the economic importance of ore deposits hosted in igneous rocks can be obtained from a compilation of mineral production data as a function of host rock type. A country like South Africa, for example, is underlain dominantly by sedimentary rocks and these undoubtedly host many of the valuable mineral resources (especially Igneous ore-forming processes Box 1.1 Diamondiferous kimberlites and lamproites: the Orapa (Botswana) and the Argyle (Western Australia) diamond mines Box 1.2 Partial melting and concentration of incompatible elements: the Rössing uranium deposit Box 1.3 Boundary layer differentiation in granites and incompatible element concentration: the Zaaiplaats tin deposit, Bushveld Complex Box 1.4 Crystal fractionation and formation of monomineralic chromitite layers: the UG1 chromitite seam, Bushveld Complex Box 1.5 Silicate–sul?de immiscibility: the komatiite hosted Ni–Cu deposits at Kambalda, Western Australia Box 1.6 New magma injection and magma mixing: the Merensky Reef, Bushveld Complex Box 1.7 Magma contamination and sul?de immiscibility: the Sudbury Ni–Cu deposits METALLOGENY OF OCEANIC AND CONTINENTAL CRUST FUNDAMENTAL MAGMA TYPES AND THEIR METAL ENDOWMENT THE RELATIVE FERTILITY OF MAGMAS AND THE “INHERITANCE FACTOR” “late-veneer” hypothesis diamonds and kimberlite/lamproite metal concentrations in metasomatized mantle S- and I-type granites PARTIAL MELTING AND CRYSTAL FRACTIONATION AS ORE- FORMING PROCESSES TRACE ELEMENT DISTRIBUTION DURING PARTIAL MELTING TRACE ELEMENT DISTRIBUTION DURING FRACTIONAL CRYSTALLIZATION MONOMINERALIC CHROMITITE LAYERS LIQUID IMMISCIBILITY AS AN ORE-FORMING PROCESS SPECIAL EMPHASIS ON MINERALIZATION PROCESSES IN LAYERED MAFIC INTRUSIONS sul?de solubility sul?de–silicate partition coef?cients the R factor PGE clusters and hiatus models ITOC01 09/03/2009 14:37 Page 1920 PART 1 IGNEOUS PROCESSES if the fossil fuels are taken into consideration). Nevertheless, the value of ores hosted in igneous rocks per unit area of outcrop can be comparable with that for sedimentary rocks, as indicated in Table 1.1. Although South Africa is character- ized by a rather special endowment of mineral wealth related to the huge Bushveld Complex, the importance of igneous-hosted ore deposits is nevertheless apparent. 1.2 MAGMAS AND METALLOGENY It is well known that different igneous rocks host ore deposits with different metal associ- ations, and that this must be related somehow to the environments in which magmas are generated and the resulting compositional characteristics they inherit from their various settings. It is widely recognized, for example, that many of the chalcophile and siderophile elements (such as Ni, Co, Pt, Pd, and Au) are more likely to be asso- ciated with ma?c rock types, whereas concentra- tions of many lithophile elements (such as Li, Sn, Zr, U, and W) are typically found in association with felsic or alkaline rock types. This has impli- cations for understanding ore genesis and, conse- quently, some of the factors related to these differences are discussed below. 1.2.1 Crustal architecture and mineral wealth Although the greatest concentrations of sidero- phile and chalcophile elements almost certainly reside in the mantle and core of the Earth, these are generally inaccessible due to their very great depths. In fact, most of the world’s economically exploitable mineral wealth effectively lies on the surface or just below the surface of the Earth. The world’s deepest mine, the Western Deep Levels gold mine near Johannesburg, South Africa, extends to just over 4000 m depth and this places an effec- tive limit on ore body exploitation, at least in terms of present technologies. Nevertheless, many mineral commodities are formed much deeper in the crust than 4 km, with some even being derived from the mantle. Diamonds, for example, are hosted in kimberlite magmas that have been brought to exploitable depths by a variety of igneous or tectonic mechanisms. Understanding ore genesis processes, therefore, requires a know- ledge of lithospheric (i.e. crust and upper mantle) architecture, and also of the origin and nature of the igneous rocks in this section of the Earth. The oceanic crust, which covers some two- thirds of the Earth surface, is thin (less than 10 km) and, compared to the continents, has a composi- tion and structure that is relatively simple and consistent over its entire extent. The upper layer, on average only 0.4 km thick (Kearey and Vine, 1996), comprises a combination of terrigenous and pelagic sediments that are distributed mainly by turbidity currents. They are often highly reduced and metal charged. This is underlain by a layer, typically 1–2.5 km thick, that is both extrusive and intrusive in character and dominantly basaltic in composition. The basalts are, in turn, underlain by the main body of oceanic crust that is plutonic in character and formed by crystallization and fractionation of basaltic magma. This cumulate assemblage comprises mainly gabbro, pyroxenite, and peridotite. Sections of tectonized and meta- Table 1.1 A comparison of the value of mineral production from igneous and sedimentary rocks in South Africa Mineralization hosted in Area (km 2 ) Value of sales, 1971 % of total area % of total value Unit value (US$/km 2 ) (10 6 US$) Granites 163 100 1973 13.3 3.4 12 000 Ma?c layered complexes 36 400 7288 3.0 12.5 200 200 Total (igneous) 199 500 9261 16.3 15.9 46 400 Sedimentary rocks 1 023 900 49 137 83.7 84.1 47 900 Source: after Pretorius (1976). ITOC01 09/03/2009 14:37 Page 20morphosed oceanic lithosphere can be observed in ophiolite complexes which represent segments of the ocean crust (usually back-arc basins) that have been thrust or obducted onto continental margins during continent–ocean collision. The types of ore deposits that one might ex- pect to ?nd associated with ophiolitic rocks are shown in Figure 1.1. They include the category of podiform chromite deposits that are related to crystal fractionation of mid-ocean ridge basalt (MORB), and also have potential for Ni and Pt group element (PGE) mineralization. Accumula- tions of manganese in nodules on the sea ?oor, metal-rich concentrations in pelagic muds, and exhalative volcanogenic massive sul?de (VMS) Cu–Zn deposits also occur in this tectonic set- ting, but are not directly related to igneous pro- cesses and are discussed elsewhere (Chapters 3 and 5). The continental crust differs markedly from its oceanic counterpart. It is typically 35–40 km thick, but thins to around 20 km under rift zones and thickens to 80 km or more beneath young mountain belts. Historically, the continental crust was thought to comprise an upper zone made up largely of granite (and its sedimentary derivatives) and a lower, more ma?c zone, with the two layers separated by the Conrad discontinuity (which marks a change in seismic velocities, and, there- fore crustal density). More recent geophysical and geological studies clearly indicate that crustal architecture is more complex and re?ects a long-lived tectonic and magmatic history, extend- ing back in some cases over 3800 million years (Figure 1.2). The continents have been progressively con- structed throughout geological time by a variety of magmatic, sedimentary, and orogenic processes taking place along active plate margins and, to a lesser extent, within the continents themselves. In addition, continental land masses have repeatedly broken apart and reamalgamated throughout geo- logical history. These episodes, known as Wilson cycles, have rearranged the con?guration of con- tinental fragments several times in the geological past. In the early Proterozoic, for example, it is conceivable that segments of southern Africa and western Australia might have been part of the same continent. The signi?cance of these cycles, and the pattern of crustal evolution with time, to global metallogeny is discussed in more detail in Chapter 6. IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 21 dykes Oceanic lithosphere Cr VMS Cu, Co, Zn Sheeted Partial Ocean Mid- ocean ridge melting Plagioclase Pyroxene Olivine Cumulates Tectonized dunite and harzburgite Lavas (pillowed) Mn, Co, Ni Pelagic sediment Plagioclase lherzolite Spinel lherzolite Cr, Ni, Pt Figure 1.1 Oceanic crustal architecture showing the main types of ore deposits characteristic of this environment. Only chromite and related deposits (Cr–Ni–Pt) are related to igneous ore-forming processes; VMS (Cu, Co, Zn) and sediment-hosted deposits (Mn, Co, Ni) are discussed in Chapters 3 and 5 respectively. ITOC01 09/03/2009 14:37 Page 2122 PART 1 IGNEOUS PROCESSES The upper crust, which in some continental sections is de?ned as extending to the Conrad discontinuity at some 6 km depth, is made up of felsic to intermediate compositions (granite to diorite) together with the sedimentary detritus derived from the weathering and erosion of this material. Archean continental fragments (greater than 2500 Myr old) also contain a signi?cant com- ponent of greenstone belt material, representing preserved fragments of ancient oceanic crust. The lower crust, between the Conrad and Mohorovicic discontinuities, is variable in composition but is typically made up of hotter, and usually more dense, material. This is because temperatures and pressures in the crust increase with depth at average rates of some 25 °C km -1 and 30 MPa km -1 respectively (Kearey and Vine, 1996). The lower crust is not necessarily compositionally differ- ent from the upper crust, but exists at higher metamorphic grades. It is also likely to be more anhydrous and residual, in the sense that magma now present at higher levels was extracted from the lower crust, leaving a residue of modi?ed material. Some of the lower crust may be more ma?c in com- position, comprising material such as amphibolite, gabbro, and anorthosite. Most of the world’s known ore deposits are, of course, hosted in rocks of the continental crust, and the full range is not shown in Figure 1.2. Some of the more important igneous rock-related deposit types are shown and these include dia- mondiferous kimberlites, anorthosite-hosted Ti deposits, the Cr–V–Pt–Cu–Ni assemblage of ores in continental layered ma?c suites, and the Sn–W–F–Nb–REE–P–U family of lithophile ores related to granites and alkaline intrusions. 1.2.2 Magma types and metal contents Although their rheological properties are different, the outer two layers of the Earth, the more rigid lithosphere and the ductile asthenosphere, are largely solid. Zones within these layers that are anomalous in terms of pressure or temperature do, however, form and can cause localized melt- ing of the rocks present. The nature of the rock undergoing melting and the extent to which it is melted are the main factors that control the com- position of the magma that is formed. The magma composition, in turn, dictates the nature of metal concentrations that are likely to form in the rocks that solidify from that magma. Although it is theoretically possible to form an almost in?nite range of magma compositions (from ultrama?c to highly alkaline), for ease of discussion this section is subdivided into four parts, each representing what is considered to be a fundamental magma type – these are basalt, andesite, rhyolite, and alkaline magmas, the latter including kimberlite. Cu, Mo, Pb, Zn Sn, W Diamond Cu, REE, P Cr, Cu, Ni, PGE, V Sn, W, Cu, Au Rift U, Th Tectonically thickened Continental crust Lithosphere Asthenosphere Oceanic crust Volcanic arc Continental crust I-type S-type Kimberlite Figure 1.2 Continental crustal architecture showing the main types of igneous-related ore deposits characteristic of this environment. ITOC01 09/03/2009 14:37 Page 22Basalt Basalts form in almost every tectonic environment, but the majority of basaltic magma production takes place along the mid-ocean ridges, and in response to hot-spot related plumes, to form oceanic crust. In addition, basalts are formed together with a variety of more felsic magmas, along island arcs and orogenic continental margins. Basaltic magma may also intrude or extrude continental crust, either along well de?ned fractures or rifts (such as continental ?ood basalt provinces, or the Great Dyke of Zimbabwe) or in response to intra- plate hot-spot activity (which might have been responsible for the formation of the Bushveld Complex of South Africa). Basalt forms by partial melting of mantle material, much of which can generally be described as peridotitic in composition. Certain mantle rocks, such as lherzolite (a peridotite which contains clinopyroxene and either garnet or spinel), have been shown experimentally to produce basaltic liquids on melting, whereas others, like alpine- type peridotite (comprising mainly olivine and orthopyroxene), are too refractory to yield basaltic liquids and may indeed represent the residues left behind after basaltic magma has already been extracted from the mantle. Likewise, oceanic crust made up of hydrated (serpentinized) basalt and drawn down into a subduction zone is also a potential source rock for island arc and contin- ental margin type magmatism. Komatiites, which are ultrama?c basalt magmas (with >18% MgO) mainly restricted to Archean greenstone belts, have a controversial origin but are generally believed to represent high degrees of partial melting of mantle during the high heat-?ow conditions that prevailed in the early stages of crust formation prior to 2500 Ma. Ore deposits associated with ma?c igneous rocks typically comprise a distinctive (mainly siderophile and chalcophile) metal assemblage of, among others, Ni, Co, Cr, V, Cu, Pt, and Au. Examination of Table 1.2 shows that this list corresponds to those elements that are intrinsically enriched in basaltic magmas. Figure 1.3 illustrates the relative abundances of these metals in three fundamental magma types and the signi?cantly higher con- centrations in basalt by comparison with andesite and rhyolite. The enhanced concentration of these metals in each case is related to the fact that the source materials from which the basalt formed must likewise have been enriched in those con- stituents. In addition, enhanced abundances also re?ect the chemical af?nity that these metals have for the major elements that characterize a basaltic magma (Mg and Fe) and dictate its mineral com- position (olivine and the pyroxenes). The chemical af?nity that one element has for another is related to their atomic properties as re?ected by their relative positions in the periodic table (see Figure 4, Introduction). The alkali earth elements (i.e. K, Na, Rb, Cs, etc.), for example, are all very similar to one another, but have properties that are quite different to the transition metals (such as Fe, Co, Ni, Pt, Pd). In addition, minor or trace elements, which occur in such low abundances in magmas that they cannot form a discrete mineral phase, are present by virtue of their ability either to substitute for another chemically similar element in a mineral lattice or to occupy a defect site in a crystal lattice. This behavior is referred to as diadochy or substitu- tion and explains much, but not all, of the trace element behavior in rocks. Substitution of a trace element for a major element in a crystal takes place if their ionic radii and charges are similar. Typically radii should be within 15% of one another and charges should differ by no more than one unit provided the charge difference can be compensated by another substitution. Bond strength and type also effects diadochy and it preferentially occurs in crystals where ionic bond- ing dominates. A good example of diadochic behavior is the substitution of Ni 2+ for Mg 2+ in olivine, or V 3+ for Fe 3+ in magnetite. Analytical data for the Ni content of basalts shows an excellent correla- tion between Ni and MgO contents (Figure 1.4), con?rming the notion that the minor metal sub- stitutes readily for Mg. The higher intrinsic Ni content of ultrama?c basalts and komatiites would suggest that the latter rocks are perhaps better suited to hosting viable magmatic nickel deposits, an observation borne out by the presence of world class nickel deposits hosted in the Archean IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 23 ITOC01 09/03/2009 14:37 Page 2324 PART 1 IGNEOUS PROCESSES Table 1.2 Average abundances of selected elements in the major magma types Basalt Andesite Rhyolite Alkaline magma Kimberlite Clarke* Li 10 12 50 – – 20 Be 0.7 1.5 4.1 4–24 – 2.8 F 380 210 480 640 – 625 P 3200 2800 1200 1800 0.6–0.9% 1050 V 266 148 72 235 – 135 Cr 307 55 4 – – 10 Co 48 24 4.4 – – 25 Ni 134 18 6 – 1050 75 Cu 65 60 6 – 103 55 Zn 94 87 38 108 – 70 Zr 87 205 136 1800 2200 165 Mo 0.9–2.7 0.8–1.2 1 15 – 1.5 Sn 0.9 1.5 3.6 – – 2 Nb 5 4–11 28 140 240 20 Sb 0.1–1.4 0.2 0.1–0.6 – – 0.2 Ta 0.9 – 2.3 10 – 2 W 1.2 1.1 2.4 16 – 1.5 Pb 6.4 5 21 15 – 13 Bi 0.02 0.12 0.12 – – 0.17 U 0.1–0.6 0.8 5 10 – 2.7 Th 0.2 1.9 26 35 – 7.2 Ag † 100 80 37 – – 70 Au † 3.6 – 1.5 – – 4 Pt † 17–30 – 3–12 – 19 10 S 782 423 284 598 2100 260 Ge 1.1 1.2 1.0–1.3 1.3–2.1 0.5 1.5 As 0.8 1.8 3.5 – – 1.8 Cd 0.02 0.02 0.2–0.5 0.04 – 0.2 If no average is available, a range of values is provided. *Clarke is a term that refers to the average crustal abundance. † Values as ppb, all other values as ppm. Source: data from Taylor (1964), Wedepohl (1969), Krauskopf and Bird (1995). Pt × 10 ppb 400 0 Ni Abundance (ppm) 300 200 100 Basalt Andesite Rhyolite Co Cr V Cu Zn Au × 100 ppb Figure 1.3 Relative abundances of selected metals in basalt, andesite, and rhyolite (data from Table 1.2). ITOC01 09/03/2009 14:37 Page 24komatiites of the Kambalda mining district in Western Australia (see Box 1.5) and elsewhere in the world. Andesite Andesites are rocks that crystallize from magmas of composition intermediate between basalt and rhyolite (typically with SiO 2 contents between 53 and 63 wt%). Their petrogenesis remains contentious, although it is well known that they tend to occur dominantly in orogenic zones, either along island arcs or on continental margins beneath which subduction of oceanic crust is taking place (Hall, 1996). Discussion about the origin of andesite revolves around whether it represents a primary magma composition derived directly by an appropriate degree of melting of a suitable source rock, or an evolved melt formed by differentiation of a more ma?c magma such as basalt. Geological observations support the notion that andesite can be formed both as a primary magma composition and by in situ fractionation. The observation that andesitic volcanoes occur directly above aseismic sections of a Benioff zone (i.e. the subducted slab that produces earth- quakes due to movement and fracturing of rock) suggests that melt production (and damping of seismic waves) has occurred in these areas. This would support the notion that andesitic magma is produced by direct melting of hydrous oceanic crust or, more likely, the mantle wedge over- lying the subduction zone as it is permeated by ?uids expelled from the subjacent oceanic crust. Alternatively, andesitic magma can be produced by fractionation of phases such as hornblende and magnetite from relatively water-rich parent magmas (Osborn, 1979), or by contamination of an originally more ma?c melt by felsic material or melt. Irrespective of the mode of formation of andesite it is apparent that as a magma type it does not exhibit a primary association with any particu- lar suite of metals or ore deposits. It appears instead that ore deposits tend to be associated with magmas representing the ends of the com- positional spectrum, and that intermediate melt compositions are simply characterized by inter- mediate trace element abundances. Examination of Table 1.2 shows that andesites appear to have little or no metal speci?city and are characterized by trace element abundances that are intermedi- ate between those of basalt on the one hand and either granite or alkaline rocks on the other. Rhyolite Felsic magmas can also form in a variety of geo- logical environments. They crystallize at depth to form a spectrum of rock compositions ranging from Na-rich tonalite to K-rich alkali granite, or extrude on surface to form dacitic to rhyolitic volcanic rocks. Very little granite magma forms in oceanic crust or along island arcs that have formed between two oceanic plates. Where oceanic granite does occur it is typically the result of differenti- ation of a more ma?c magma type originally formed by mantle melting. Along the mid-Atlantic ridge in Iceland, for example, eruptions of the volcano Hekla are initiated by a pulse of felsic ash pro- duction which is rapidly followed by eruption of more typical basaltic andesite. This suggests that the intervening period between eruptions was characterized by differentiation of the magma and that the accompanying build-up of volatiles may have been responsible for the subsequent eruption (Baldridge et al., 1973). These observations, among many others, clearly indicate that granitic melts IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 25 0 Ni (ppm) 2000 1000 0 500 1500 5 10 15 35 20 25 30 Wt% MgO Figure 1.4 The relationship between Ni and MgO contents of basalts within which base metal mineralization does not occur (data from Naldrett, 1989a). ITOC01 09/03/2009 14:37 Page 2526 PART 1 IGNEOUS PROCESSES can be the products of differentiation of more ma?c magmas in oceanic settings. Most felsic magmas, however, are derived from the partial melting of predominantly crustal mater- ial along ocean–continent island arcs and orogenic continental margins. Although Andean-type sub- duction zones might facilitate partial melting of the downgoing slab itself, the much higher pro- portion of felsic magma formed in this environ- ment compared to oceanic settings points to a signi?cant role for continental crust as a source. There is now general agreement that Andean- type subduction-related magmatism receives melt contributions from both the mantle lithosphere and the continental crust, with the wide-ranging compositions of so-called “calc–alkaline” igneous suites being attributed to a combination of both magma mingling and fractional crystallization (Best, 2003). Signi?cant quantities of felsic magma are pro- duced in the latter stages of continent–continent collision and also in anorogenic continental set- tings where rifting and crustal thinning has taken place. Himalayan-type continent collision, for example, is usually accompanied by crustal thick- ening associated with intense thrusting, tectonic duplication and reverse metamorphic gradients. These processes cause dewatering of crustal material, which, in turn, promotes partial melting to form high-level leucogranite magmas derived from source rocks that often contain signi?cant proportions of sedimentary material (Le Fort, 1975). Anorogenic continental magmatism, on the other hand, is usually related to crustal thinning (accompanying plume or hot-spot activity?) and is typi?ed by the production of magmas with bimodal compositions (i.e. basalt plus rhyolite). A good example is the 2060 Myr old Bushveld Complex in South Africa, where early ma?c mag- mas intruded to form the world’s largest layered igneous complex, followed by emplacement of a voluminous suite of granites. Ore deposits associated with felsic igneous rocks often comprise concentrations of the lithophile elements such as Li, Be, F, Sn, W, U, and Th. Table 1.2 shows that this list corresponds to those elements that are intrinsically enriched in rhy- olitic magmas and Figure 1.5 illustrates, in bar graph form, the relative abundances of these ele- ments and, in particular, the higher abundances in rhyolite by comparison with andesite and basalt. The relative enrichment of certain lithophile elements in rhyolitic magmas is partially related to their geochemically incompatible nature. An incompatible element is one whose ionic charge and radius make it dif?cult to substitute for any of the stoichiometric elements in rock-forming minerals. Thus, incompatible elements tend to be excluded from the products of crystallization and concentrated into residual or differentiated magmas (such as the granitic magmas that might form by crystal fractionation of ma?c magmas in oceanic settings). Alternatively, incompatible elements also tend to be concentrated in crustal 60 0 Li Abundance (ppm) 50 30 10 20 40 Be × 10 F/10 Sn × 10 W × 10 U × 10 Th Basalt Andesite Rhyolite Figure 1.5 Relative abundances of selected “granitophile” elements in basalt, andesite, and rhyolite (data from Table 1.2). ITOC01 09/03/2009 14:37 Page 26melts derived from low degrees of partial melting of source rocks that may themselves have been endowed in the lithophile elements. These con- cepts are discussed in more detail in section 1.4 below. A well known and interesting feature of ore deposits that are genetically associated with granite intrusions is that the origin and composi- tion of the magma generally controls the nature of the metal assemblage in the deposit (Chappell and White, 1974; Ishihara, 1978, 1981). This con- trol is almost certainly related in part to the metal endowment inherited by the magma from the rocks that were melted to produce it. Where a felsic magma is derived from melting of a sedimentary or supracrustal protolith (termed S-type granites), associated ore deposits are characterized by con- centrations of metals such as Sn, W, U, and Th. Where it is derived from melting of older igneous protoliths in the crust (I-type granite) the ore asso- ciation is typi?ed by metals such as Cu, Mo, Pb, Zn, and Au. This association is metallogenically very signi?cant and is discussed in more detail in section 1.3.4 below and again in Chapter 2. Alkaline magmas and kimberlite Although most magma compositions can be rep- resented by the basalt–andesite–rhyolite spectrum, some deviate from this trend and are composi- tionally unusual. For example, magmas that are depleted in SiO 2 but highly enriched in the alkali elements (Na, K, and Ca) are relatively rare, but may be economically important as they frequently contain impressive concentrations of a wide range of ore-forming metals (such as Cu, Fe, P, Zr, Nb, REE, F, U, and Th). In addition, kimberlitic and related magma types (such as lamproites) are the main primary source of diamonds. The most common alkaline ma?c magma is nephelinite, which crystallizes to give a range of rock types (the ijolite suite; Hall, 1996) com- prising rather unusual minerals, such as felspa- thoid, calcic-pyroxene, and carbonate assemblages. Nephelinite lavas are observed in oceanic settings such as the Cape Verde and Hawaiian islands, but are best seen in young (Paleocene to recent) con- tinental volcanic settings such as the East African rift valley, central Europe and southeast Australia. Old alkaline igneous complexes are rare, one of the best preserved being the 2050Myr old Phalaborwa Complex in South Africa, which is mined for copper and phosphate as well as a host of minor by-products. Nephelinite, as well as the associated, but rare, carbonatite melts (i.e. magmas comprising essen- tially CaCO 3 and lesser Na 2 CO 3 ), are undoubtedly primary magma types derived from the mantle by very low degrees of partial melting under condi- tions of high P total and P CO 2 (Hall, 1996). The rela- tionship between nephelinitic and carbonatitic magmas is generally attributed to liquid immis- cibility, whereby an original alkali-rich silicate magma rich in a carbonate component exsolves into two liquid fractions, one a silicate and the other a carbonate (Ferguson and Currie, 1971; Le Bas, 1987). Low degrees (2%) of partial melting of a garnet lherzolitic source in the mantle will typic- ally yield olivine nephelinite compositions and these magmas may be spatially and temporally associated with basaltic volcanism (Le Bas, 1987). Nephelinite magma associated with carbonatite, on the other hand, is only considered possible if the source material also contained a carbonate phase (such as dolomite) and a soda-amphibole. This type of mantle source rock is likely to be the result of extensive metasomatism, a process that involves ?uid ingress and enrichment of volatile and other incompatible elements. Melting of a fertile mantle source rock is probably the main reason why alkaline magmas are so enriched in the variety of ore constituents mentioned above. The extent of metal enrichment relative to aver- age basalt is illustrated in Figure 1.6. Kimberlitic and related ultrama?c magmas crystallize to form very rare and unusual rocks, containing among other minerals both mica and olivine. Kimberlites are rich in potassium (K 2 O typically 1–3 wt%) and, although derived from deep in the mantle, are also hydrated and carbonated. They usually occur in small (<1k m diameter) pipe-like bodies, or dykes and sills, and commonly extrude in highly explosive, gas-charged eruptions. The deep-seated origin of kimberlite is evident from the fact that it commonly transports garnet lherzolitic and eclogitic xenoliths to the IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 27 ITOC01 09/03/2009 14:37 Page 2728 PART 1 IGNEOUS PROCESSES surface, rock types made up of very high pressure mineral assemblages that could only have come from the mantle. In addition, a small proportion of kimberlites also contain diamond xenocrysts. Diamond is the stable carbon polymorph under very reducing conditions, and at depths in excess of about 100 km and temperatures greater than 900 °C (Haggerty, 1999). The origin of kimberlitic magmas is not too dif- ferent from that of the alkaline rocks described above, and high pressure partial melting of a garnet peridotite source rock containing additional phlogopite or K-amphibole (richterite), as well as a carbonate phase, is viewed as a likely scenario (Hall, 1996). The enrichment of incompatible constituents (such as K, Rb, H 2 O, and CO 2 ) in kimberlite, as with alkaline magmas in general, again indicates that metasomatism of the mantle has played an important role in the provision of a deep-seated environment capable of producing highly enriched, or fertile, magmas. These aspects are discussed in more detail in section 1.3 below. 1.3 WHY ARE SOME MAGMAS MORE FERTILE THAN OTHERS? THE “INHERITANCE FACTOR” Geochemical inheritance is clearly an important factor in understanding the nature of ore-forming processes in igneous rocks. Magma may inherit a surplus of potential ore-forming trace elements because the source material from which it was derived was itself enriched in these components. Further concentration of incompatible trace ele- ments into residual magma, or of compatible trace elements into crystallizing phases, will take place during cooling and solidi?cation of the magma, and these processes are discussed in more detail in section 1.4 below. A pertinent question that relates to the issue of geochemical inheritance is why certain portions of the Earth’s crust appear to be so much better endowed in mineral deposits than others. The spectacular concentrations of, for example, gold and platinum deposits in South Africa, related to the Witwatersrand Basin and Bushveld Com- plex respectively, perhaps point to some form of (mantle-related?) enrichment of these metals that is speci?c to this region and not to other parts of the globe. Why is it that Sudbury is so rich in Ni and the Andes are so well endowed with large Cu deposits, and is there an explanation for the tendency that diamondiferous kimberlites only occur in ancient, cratonic areas? These questions are addressed below in relation to some rather novel ideas suggesting that metals may have been preferentially concentrated in certain parts of the mantle and then subsequently transferred into the crust through a variety of mineralizing agencies. 1.3.1 The “late veneer” hypothesis of siderophile metal concentration – an extraterrestrial origin for Au and Pt? In the very early stages of Earth evolution, prevailing theory suggests that the originally 700 0 F Abundance (ppm) 300 200 100 Basalt Alkaline magma/kimberlite 400 500 600 P/10 Cu Nb Zr/10 U × 10 Th × 10 Figure 1.6 Relative abundances of selected metals in alkaline magmas (and kimberlite in the case of Cu and P) relative to average basalt (data from Table 1.2). ITOC01 09/03/2009 14:37 Page 28homogeneous, molten planet differentiated into a metallic core, comprising essentially Fe and FeO with lesser Ni, and a mantle which had a silicate composition. As this differentiation took place the siderophile metals (i.e. those with a strong af?nity for Fe such as Au and the platinum group elements or PGE) were comprehensively parti- tioned into the core. Experiments by Holzheid et al. (2000) indicate that the average concentra- tions of elements such as Au, Pt and Pd in the Earth’s mantle should be at least 10 -4 times lower than average chondritic abundances. Such con- centrations are, in fact, so low that they virtually preclude the possibility of forming ore deposits in rocks extracted from the mantle (i.e. the crust). Yet the actual concentration of these precious metals in the mantle, although depleted, is only about 150 times lower than average chondritic abundances. This depletion might be accommod- ated by the fact that there have been numerous ore deposits formed over geological time that have extracted these metals from the mantle. Another explanation is that the ef?ciency with which siderophile metals are partitioned between metal core and silicate mantle decreases with increasing pressure (depth) and that this could explain why the mantle is not as depleted as theory predicts. Although the latter notion probably applies to nickel, recent experiments suggest that it is not applicable to the precious metals and that some other explanation must be sought for the higher than expected concentrations of the latter in the mantle. A clue as to why the mantle might be relatively enriched in siderophile metals lies in the fact that their abundance ratios (i.e. the abundance of one element relative to another, such as Au/Pt or Pt/Pd) are generally similar to chondritic abundance ratios as determined from analyses of meteorites that have fallen to Earth. The only way to explain this is by having a substantial proportion of the preci- ous metals in the mantle derived from meteorites that impacted the proto-crust during the early stages of Earth evolution, but after the differentia- tion of core and mantle (Figure 1.7). This idea, known as the “late-veneer” hypothesis (Kimura et al., 1974), suggests that much, if not all, the Au and Pt that is mined from ore deposits on the IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 29 t o + 500 Myr intense meteorite flux Au Pt Au Pd Pt t o – core/mantle segregation "Late veneer" of siderophile element enrichment (Au, Pt, Pd, etc.) Fe, Ni, Au, Pt Figure 1.7 Schematic representation of the “late veneer” hypothesis for the siderophile (precious) metal enrichment of the Earth’s lithosphere. During initial segregation of the Earth (t o ) siderophile metals were comprehensively partitioned into the Fe–Ni core. Approximately 500 million years later (t o + 500 Myr) intense meteorite bombardment of the Earth added to the siderophile metal budget of the Earth’s lithosphere. ITOC01 09/03/2009 14:37 Page 29Earth’s surface today ultimately had an extrater- restrial origin and that the planet’s own inventory of these metals is presently locked away in the core. Since both the meteorite ?ux itself and the subsequent distribution of this material through the mantle are likely to have been irregular, this hypothesis is also consistent with the hetero- geneous distribution of precious metals over the Earth’s surface. As a footnote it is intriguing to note that pre- cious metals may not have been the only valuable commodity introduced to Earth by meteorites. The enigmatic “carbonado” diamonds found only in 1500 Myr old metasediments of Bahia State in Brazil and the Central African Republic have mineralogical and isotopic characteristics unlike any diamond of terrestrial origin. Haggerty (1999) has suggested that carbonado diamonds are derived from the fall-back of a fragmented carbon-type asteroid that impacted the Earth’s crust at a time when the relevant parts of Brazil and Africa formed a single continental entity. Although con- tentious, the notion of an extraterrestrial origin for certain constituents of the Earth’s surface (and of life itself?) is one that is likely to continue attracting attention in the future. 1.3.2 Diamonds and the story they tell The Earth’s mantle, between about 35 and 2900 km depth, is the ultimate source of material that, over geological time, has contributed, either directly or by recycling, to formation of the crust. Considering that the mantle is essentially inaccessible it has, nevertheless, been the sub- ject of numerous studies by both geologists and geophysicists and its structure and composition are now reasonably well known. One reason for studying the mantle is to understand the origin of diamond. This remarkable mineral, together with the magmas that bring it to the surface, has provided a great deal of information about the deep Earth, much of which is very relevant to understanding those properties of the mantle that also relate to the source of metals in other igneous ore deposits. Most diamonds are brought to the Earth’s sur- face by kimberlitic magmas (see section 1.2) or a compositionally similar melt known as lamproite. Most kimberlites and lamproites are barren, and diamondiferous magmas only intrude into ancient, stable continental crust that is typically older than 2500 Myr, but sometimes as young as 1500 Myr. The kimberlite magmas that transport diamonds to the surface, however, are typically much younger than the rocks they intrude, forming in discrete episodes in the Mesozoic and Cenozoic eras. Older intrusive episodes have also been observed in the Devonian, as well as at around 500 Ma and again at 1000Ma (Haggerty, 1999). Diamondiferous kimberlites must also have been emplaced during the Archean, since the Witwatersrand conglomer- ates in South Africa, for example, are known to contain green detrital diamonds. To further com- plicate the story, the diamonds themselves tend to be much older than their kimberlitic host rocks and range in age from 1500 to 3000Ma, indicating that they have resided in the mantle for consider- able periods of time prior to their eruption onto the Earth’s surface. Diamonds did not, therefore, crystallize from the kimberlite but were intro- duced to the Earth’s surface as xenocrysts within the magma (Richardson et al., 1984). Diamond xenocrysts occur either as isolated single crystals in the kimberlitic matrix or as minerals within discrete xenoliths of either peridotite (P-type diamonds – the more common) or eclogite (E-type diamonds). The high pressure phase relations that characterize mineral assemblages in these mantle xenoliths indicate that diamonds are derived from zones of thickened, sub-cratonic lithosphere, at least 200 km thick, that extend beneath stable Archean–Proterozoic shield areas (Figure 1.8). These lithospheric keels comprise old, depleted, peri- dotite (i.e. from which mantle melts have already been extracted) as well as primitive, but younger, eclogite that has generally not had a melt fraction extracted and is, therefore, more fertile with respect to crust-forming elements (Haggerty, 1999). It is now generally accepted that diamonds were generated from deep in the mantle, in the layer known as the Transition Zone between the lower and upper mantle at around 400 to 650 km depth. Because the upper mantle is relatively depleted in carbon (100 ppm compared with 1000–3700 ppm in the lower mantle; Wood et al., 1996) it is 30 PART 1 IGNEOUS PROCESSES ITOC01 09/03/2009 14:37 Page 30unlikely to be a viable source for the primordial carbon that makes up diamond. The more fertile lower mantle is more likely to be the source of the carbon, and this is supported by the presence of very high pressure minerals occurring as tiny inclusions in many diamonds. However, the upper mantle is more reduced than the lower mantle, which, in addition to its high carbon contents, also contains substantially more water (500–1900 ppm compared to only 200 ppm in the upper mantle). The upper mantle is, therefore, more likely to pre- serve diamond because the mineral’s long-term stability depends on the existence of a reducing environment. Carbon in the relatively oxidized, ?uid-rich lower mantle would, despite the higher pressures, not occur as diamond at all, but as CO 2 , CCO, or MgCO 3 (Wood et al., 1996). The model for diamond formation (Figure 1.8), therefore, sug- gests that plumes transfer melt and volatiles from the lower mantle, and precipitate diamond at higher levels either in the reduced environment represented by the Transition Zone or in the keels extending below thick, cratonic lithosphere. Thus, the more common P-type diamonds form when the relatively oxidized carbonic ?uids dis- solved in ascending plumes interact with reduced mantle at higher levels and precipitate elemental carbon. This mass transfer process is referred to as metasomatism and involves the movement of ?uids and volatiles from deep in the Earth’s IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 31 Spinel Lherzolite Spinel Lherzolite 100 200 300 400 500 600 km GRAPHITE DIAMOND UPPER MANTLE LOWER MANTLE TRANSITION ZONE CRUST OROGENIC BELT Moho Harzburgite CO 2 H 2 O LILE Dunite Eclogite LILE DEPLETED CO 2 H 2 O LILE PLUME CO 2 H 2 O LILE PLUME LILE ENRICHED LILE CO 2 MgCO 3 1500°C 1200°C 900°C LAMPROITE Metasomatized mantle KIMBERLITES Eclogite SUB-CRATONIC LITHOSPHERIC KEEL (ARCHEAN?) H 2 O: 200 ppm C: 100 ppm H 2 O H 2 O Carbon as CO 2 , MgCO 3 , CCO fO 2 >FMQ fO 2 10 –2 to 10 –1 FMQ H 2 O: 520–1900 ppm C: 1000–3700 ppm Figure 1.8 Schematic diagram illustrating features pertinent to the formation of diamond and the fertilization of the Earth’s mantle by plume-related magmas and their associated aqueo-carbonic ?uids (after Haggerty, 1999). LILE refers to the large ion lithophile elements; FMQ refers to the fayalite–magnetite–quartz oxygen buffer. ITOC01 09/03/2009 14:37 Page 3132 PART 1 IGNEOUS PROCESSES mantle to higher levels. This process is turning out to be very relevant to the concepts of mantle fertilization and geochemical inheritance. The more rare E-type diamonds, by contrast, are con- sidered to have crystallized directly from a magma intruded into or ponded below the keels (Haggerty, 1999). Formation of the kimberlitic magma that trans- ports diamond to the Earth’s surface has also been attributed to plume activity and the metasomatic transfer of volatile constituents from a fertile lower mantle into depleted upper mantle. Evidence for this process comes from the observation that many of the major episodes of kimberlite intrusion men- tioned above correlate with “superchron” events that are de?ned as geologically long time periods of unidirectional polarity in the Earth’s magnetic ?eld. Superchrons are caused by core–mantle boundary disruptions which increase the rate of liquid core convection, causing a damping of the Diamond mining around the world was worth some US$7 billion in 2001, a substantial proportion of which was derived from exploitation of primary kimberlitic and lamproitic deposits. The biggest single deposit is at Argyle in Western Australia, which is hosted in lamproite and produces some 26 million carats per year. Most of the diamonds produced here, however, are of low value. The Orapa and Jwaneng deposits of Botswana, by contrast, produce less than half the number of carats per year, but their stones are much more valuable. Orapa and Jwaneng together are the richest diamond deposits in the world. Kimberlites are by far the most important primary source of diamonds (see section 1.3.2) and there are over 5000 occurrences known world wide (Nixon, 1995). By contrast there are only some 24 known occurrences of lamproite. Both kimberlites and lamproites are emplaced into the Earth’s crust as “diatreme–maar” volcanoes which are the product of highly overpressured, volatile-rich magma. Kimberlitic or lamproitic magma is injected into the crust, along zones of structural weakness, to within 2–3 km of the surface. At this point volatiles (H 2 O and CO 2 ) either exsolve from the magma itself (see Chapter 2), or the magma interacts with groundwater, with the resulting vapor phase Diamondiferous kimberlites and lamproites: The Orapa diamond mine, Botswana and the Argyle diamond mine, Western Australia causing violent phreatomagmatic eruption of magma and disruption of country rock. Figure 1 shows the anatomy of a diatreme–maar system that has applicability to the nature and geometry of both kimberlites and lamproites. Kimberlites are richer in CO 2 than lamproites and since CO 2 has a lower solubility than water in silicate melts, kimberlite magmas will usually exsolve a volatile fraction at lower depths than lamproites (Nixon, 1995). Lamproite venting is quite often a function of magma interaction with groundwater, the availability and depth of which dictates the geometry of the crater. These factors account for the carrot-like shape of kimberlites compared to a broader, champagne glass shape for many lamproites. The distribution of diamonds in any one kimberlite may be highly erratic and there seems to be little or no relationship between grade and depth (Nixon, 1995). Some hypabyssal dykes are very rich in diamonds, such as the Marsfontein mine in South Africa which has an average grade of 200–300 carats per 100 tons of ore mined. Diatreme facies kimberlites are often characterized by multiple injections of magma, some of which are barren and others economically viable. Kimberlitic pyro- clastic sediments in crater facies are also often richer than geomagnetic ?eld intensity but promoting plume activity and mantle metasomatism. Intrusion of diamondiferous kimberlites has also been linked in time to major geological events, such as con- tinental break-up and ?ood basaltic magmatism (Haggerty, 1994). England and Houseman (1984) suggested that enhanced kimberlite intrusion could be related to periods of low plate velocity when uninterrupted mantle convection gave rise to partial melting and volatile production in the lithosphere, and the subsequent formation of plume activity. Accompanying epeirogenic uplift created the fractures that allowed kimberlite magma to intrude rapidly upwards and, in many cases, to extrude violently onto the Earth’s surface. This explanation is certainly consistent with the geodynamic setting of kimberlites, such as their predominance in Africa during the Mesozoic still-stand (see Chapter 6). The relative rarity of kimberlite formation and penetration to the surface ITOC01 09/03/2009 14:37 Page 32IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 33 crater facies made up of both epiclastic and pyroclastic kimberlite debris. All these phases are diamondiferous. At Orapa the northern diatreme is believed to have been emplaced ?rst, followed by residual volatile build-up and explosive volcanic activity. This was shortly followed by a similar sequence of events to form the southern diatreme, with the subsequent merging of its crater facies into a single maar (Field et al., 1997). Dyke 5 Depth (km) Fallback material 4 3 2 1 0 Sill Maar Tuffaceous kimberlite Diatreme Kimberlite breccia . . . . . . . . . . . . . Crater facies Diatreme facies Hypabyssal fecies Figure 1 Idealized geometry of a diatreme–maar type volcano, showing the nature of the hypabyssal facies of magma emplacement along crustal weaknesses (after Smith et al., 1979). Kimberlites around the world are seen as hypabyssal, diatreme, or crater facies, depending on the extent of preservation (or depth of erosion) of the bodies. Figure 2 Large boulder of country rock in tuf?sitic and lacustrine maar sediments, Orapa Mine, Botswana. associated diatreme rocks, possibly due to enrichment of heavy minerals, including diamond, by wind (during eruption) or water (in the crater lake). Argyle The Argyle diamond mine occurs in a 1200 million year old lamproite diatreme that has intruded older Proterozoic sediments in the Kimberley region of Western Australia (Boxer et al., 1989). It forms an elongate 2km long body that varies from 100 to 500 m in width and is the result of at least two coalesced vents along a fault line. It is mined in the southern section where it has grades in excess of 5 carats per ton. Most of the body is made up of pyroclastic or tuffaceous material, with marginal breccia and occa- sional lamproitic dykes. The diatreme was formed when lamproite magma encountered groundwater in largely unconsolidated sediments, resulting in multiple phreato- magmatic eruptions and venting of lamproite at the surface. Orapa The Orapa diamond mine occurs in a mid-Cretaceous kimberlite in the north of Botswana, and is well known for the excellent preservation of its crater facies. The kimberlite was emplaced in two pulses that merge at about 200 m depth into a single maar. The diatremes, com- prising tuf?sitic kimberlite breccia, grade progressively into ITOC01 09/03/2009 14:37 Page 3334 PART 1 IGNEOUS PROCESSES is also consistent with this model, since magma formation requires that a number of coincident- ally optimal conditions apply. 1.3.3 Metal concentrations in metasomatized mantle and their transfer into the crust Although it has been evident for many years that there is an association between mantle meta- somatism and diamond formation, a link to mantle fertilization with respect to other constituents, such as the base and precious metals, has only recently been suggested. A comparison of Re–Os isotopic ratios of epithermal Cu–Au ores from the giant Ladolam mine on Lihir Island, near Papua New Guinea, with peridotite xenoliths transported to the ocean surface by volcanic activity from the underlying mantle wedge, shows that the ore con- stituents are derived from the mantle. Although this is not unexpected, closer study of the peri- dotitic material involved reveals that some of this material has been extensively metasomatized to form a high temperature hydrothermal mineral assemblage comprising olivine, pyroxene, phlo- gopite, magnetite, and Fe–Ni sul?des (McInnes et al., 1999). Metasomatism is considered to be the result of dehydration of the oceanic slab as it moves down the subduction zone, yielding ?uids that migrate upwards into the overlying mantle wedge. Metasomatized peridotite contains pre- cious and base metal concentrations that are up to two orders of magnitude enriched relative to unaltered mantle (Figure 1.9). Subduction of oceanic crust beneath an island arc, such as that of which Lihir Island is part, has resulted in the formation of alkaline basalt which builds up the arc and ultimately forms the host rocks to the Ladolam Cu–Au deposit. The deposit itself occurs at a high level in the crust by the circulation of metal-charged hydrothermal ?uids, a process that is not relevant to the present dis- cussion except for the fact that these ?uids have dissolved the ore constituents that they carry from the immediate country rocks (i.e. the alkaline basalts). An interesting feature of the study by McInnes et al. (1999) is that basaltic and syenitic country rocks to the deposit (refer to b and s in Figure 1.9) were analyzed well away from the CO 2 CO 2 Cu Au Pt Pd 9 0.04 2.60 0.09 Unaltered harzburgite Magma production H 2 O H 2 O Metasomatized mantle H 2 O Lihir island arc Ladolam Cu-Au-deposit Cu Au Pt Pd 54 1.73 90.2 41.7 Alkali basalt Ontong-Java plateau Cu Au Pt Pd 61 0.79 1.75 3.00 145 0.3 15.50 21.3 Country rocks b s Dehydration of subducted slab Figure 1.9 Schematic diagram illustrating the concept of mantle metasomatism and metal enrichment associated with subduction, and the subsequent inheritance of an enhanced metal budget by magmas derived from melting of metasomatized mantle. Metal abundances of the relevant rock types (in ppm) are from McInnes et al. (1999); the two analyses showing metal abundances for the magmatic products of subduction refer to basalt (b) and syenite (s). ITOC01 09/03/2009 14:37 Page 34mineralization itself and also exhibit signi?cant enrichments in their base and precious metal con- tents. The inference is, therefore, that although the ore metals resided originally in the mantle, they were redistributed, and their concentrations signi?cantly upgraded, by the high temperature metasomatic processes observed in the mantle peridotite. Because it is the metasomatized peri- dotite that is likely to have been preferentially melted above the subduction zone (by virtue of its hydrous nature and lower solidus temperature), the resulting alkali basalt magma will have inherited a signi?cant metal endowment, thus enhancing the chances of creating a substantial ore deposit during subsequent (hydrothermal) stages of ore formation. The Ladolam study illustrates three important features: ?rst, that metasomatism and ?uid ?ow are very important processes in redistri- buting and concentrating incompatible elements in the mantle; second, that partial melting is the main process by which matter is transferred from the mantle to the crust; and, third, that inherit- ance is critical to the nature and formation of igneous hosted ore deposits, as well as to whether subsequent ?uid circulation is likely to form viable hydrothermal deposits. 1.3.4 I- and S-type granite magmas and metal speci?city As mentioned previously, the different types of granite, and more speci?cally the origins of felsic magma, can be linked to distinct metal associ- ations. Of the many classi?cation schemes that exist for granitic rocks one of the most relevant, with respect to studies of ore deposits, is the I- and S-type scheme, originally devised for the Lachlan Fold Belt in southeast Australia (Figure 1.10a) by Chappell and White (1974). In its simplest form the scheme implies that orogenic granites can be subdivided on the basis of whether their parental magmas were derived by partial melting of predominantly igneous (I-type) or sedimentary (S-type) source rock. In general, I-type granites tend to be metaluminous and typi?ed by tonalitic (or quartz-dioritic) to granodioritic compositions, whereas S-types are often peraluminous and have adamellitic (or quartz-monzonitic) to granitic com- IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 35 positions. Also very important from a metallogenic viewpoint is the fact that I-type granites tend to be more oxidized (i.e. they have a higher magmatic fO 2 ) than S-type granites, whose magmas were originally fairly reduced because of the presence of graphite in their source rocks. An approximate indication of the oxidation state of granitic magmas can be obtained from their whole rock Fe 2 O 3 /FeO ratio (which effectively records the ferric/ferrous ratio). Blevin and Chappell (1992) have shown that an Fe 2 O 3 /FeO ratio of about 0.3 provides a useful discriminant between I- (with Fe 2 O 3 /FeO > 0.3) and S-type (with Fe 2 O 3 /FeO < 0.3) granites, at least for the Australian case (Figure 1.10 b and c). A classi?cation of granites according to oxida- tion state was, in fact, made relatively early on by Ishihara (1977), who distinguished between reduced granite magmas (forming ilmenite-series granitoids) and more oxidized equivalents (forming magnetite-series granitoids). The metallogenic sig- ni?cance of this type of granite classi?cation was also recognized by Ishihara (1981), who indicated that Sn–W deposits were preferentially associated with reduced ilmenite-series granitoids, whereas Cu–Mo–Au ores could be linked genetically to oxidized magnetite-series granitoids. Magnetite- series granitoids are equivalent to most I-types, whereas ilmenite-series granitoids encompass all S-types as well as the more reduced I-types. Although now regarded as somewhat over- simpli?ed, at least with respect to more recent ideas regarding granite petrogenesis, the S- and I-type classi?cation scheme is nevertheless appeal- ing because it has tectonic implications and can be used to infer positioning relative to subduc- tion along Andean-type continental margins (see Figure 1.2). The scheme also has metallogenic signi?cance because of the empirical observation that porphyry Cu–Mo mineralization (with asso- ciated Pb–Zn–Au–Ag ores) is typically associated with I-type granites, whereas Sn–W mineralization (together with concentrations of U and Th) is more generally hosted by S-type granites. Although this relationship is broadly applicable it, too, is oversimpli?ed. Some granites, notably those that are post-tectonic or anorogenic, do not accord with the scheme, such as the alkaline granites of the Bushveld Complex, which are polymetallic and ITOC01 09/03/2009 14:37 Page 3536 PART 1 IGNEOUS PROCESSES contain both Sn–W and base metal mineraliza- tion (Robb et al., 2000). A more accurate appraisal of the relationship between magma composi- tion (including oxidation state) and metallogenic association is given by Barton (1996). Figure 1.11 shows Barton’s scheme, which regards granites as a continuum of compositional types and their metal associations in terms of different intrusion- related ore deposit types. Adjacent to Andean type subduction zones a clearly de?ned spatial pattern exists with respect to the distribution of I- and S-type granite intru- sions, as well as associated metallogenic zonation (Sillitoe, 1976; Clark et al., 1990). The leading edge (i.e. the oceanic side) of the subduction zone tends to correlate with the production of I-type granite magmas and is associated with the formation of porphyry Cu styles of mineralization. By contrast, the continental side of the subduction zone con- tains more differentiated granite types that are often S-type in character and with which Sn–W styles of mineralization are associated. Other examples of this type of regional zonation are seen in the Cordilleran granites of the western United States and in the Lachlan Fold Belt of southeast Australia, where I- and S-type granites were origin- ally de?ned (Figure 1.10a). Again, although excep- tions and complications do exist, the recognition and delineation of these patterns is clearly import- ant with respect to understanding the spatial dis- tribution of different types of ore deposits hosted in granitoid rocks. The Lachlan Fold Belt provides an excellent example of the relationships between magma type and metallogenic association and, as predicted, granite related mineralization is dominated by Sn–W in the mainly S-type granites to the west of the I–S line, whereas largely Mo with lesser Cu– Au ores are found in the I-type terrane to its east (Figure 1.10a). Another very important feature of the metal content of granites is, however, also apparent in the Lachlan Fold Belt. When the Fe 2 O 3 / FeO ratio (an indication of magmatic oxidation state) is plotted against Rb content (an indicator of degree of fractionation) for granites that are mineralized in terms of either Sn–W or Cu–Mo–Au km 0 100 200 I-type granite S-type granite Canberra Melbourne Mainly Sn-W I-S line Mainly Mo and Cu-Au (a) Fe 2 O 3 /FeO 1 0.1 10 Rb ppm 100 Fe 2 O 3 /FeO 200 0 1 0.1 10 0.01 300 400 500 600 Oxidation Oxidation Cu W I S I S Sn Mo Au Fractionation (b) (c) Figure 1.10 (a) Simpli?ed map showing the distribution of S- and I-type granites, and associated metallogenic trends, in the Lachlan Fold Belt of southeastern Australia (after Chappell and White, 1974). (b and c) Plots of Fe 2 O 3 /FeO versus Rb for granites of the Lachlan Fold Belt that are mineralized with respect to Sn–W and Cu–Mo–Au (after Blevin and Chappell, 1992). ITOC01 09/03/2009 14:37 Page 36(Figure 1.10b and c) it is clear that metal content is not simply a function of magma type alone. It is apparent that Cu–Mo–Au related intrusions typic- ally have higher Fe 2 O 3 /FeO ratios than those associ- ated with Sn–W mineralization, and are, therefore, preferentially associated with I-type granites. What is perhaps more clearly apparent, however, is that Sn and W mineralization is associated with intrusions that are more highly fractionated than those containing Cu–Mo–Au (Blevin and Chappell, 1992). Metal contents are, therefore, also a function of processes that happen as the magma cools and fractionates and these processes are discussed in more detail in section 1.4 below. In fact, mineralization in granites also involves hydrothermal processes that are quite distinct from either geochemical inheritance or fractiona- tion, and these are examined in Chapter 2. 1.4 PARTIAL MELTING AND CRYSTAL FRACTIONATION AS ORE-FORMING PROCESSES The previous section discussed the various magma- forming processes and some of the reasons why they are variably endowed with respect to their trace element and mineral contents. This section examines how trace elements and minerals behave in the magma, both as it is forming and then sub- sequently during cooling and solidi?cation. Trace element abundances can be very good indicators of petrogenetic processes, during both IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 37 Monzodiorite Monzonite Qtz-monzonite/adamellite Alkali granite F, Li, Be Zn, F, Ag, Be Mo, F (W) 'PORPHYRY' Zn, W (Pb, Ag) Cu (Mo) Cu, Zn,(Au, Ag, Fe) Cu (Au, Mo) ALKALIC Au Cu, (Au, Fe) Zn, F, Mo (Ag, Be) Sn, W LITHOPHILE Paralkaline/ qtz undersaturated Metaluminous Peraluminous Oxidized (high magmatic fO 2 ) Reduced (low magmatic fO 2 ) Mafic Felsic Diorite Granodiorite Granite Qtz-diorite/tonalite Figure 1.11 Generalized scheme that links granite compositions and magmatic oxidation state to metal associations and intrusion-related ore deposit types (modi?ed after Barton, 1996). Ore deposit types referred to are “alkalic,” “porphyry,” and “lithophile” and are discussed in more detail in Chapter 2. Metals shown in bold re?ect the more important associations. ITOC01 09/03/2009 14:37 Page 37partial melting and crystal fractionation. Since many magmatic ore deposits arise out of con- centrations of metals that were originally present in very small abundances, trace element behavior during igneous processes is also very useful in understanding ore formation. A trace element is de?ned as an element that is present in a rock at concentrations lower than 0.1 wt% (or 1000 ppm), although this limit places a rather arti?cial constraint on the de?nition. In general trace elements substitute for major ele- ments in the rock forming minerals, but in certain cases they can and do form the stoichiometric components of accessory mineral phases (Rollinson, 1993). Many of the ores associated with igneous rocks are formed from elements (Cu, Ni, Cr, Ti, P, Sn, W, U, etc.) that originally existed at trace concentrations in a magma or rock and were sub- sequently enriched to ore grades by processes dis- cussed in this and later sections. When rocks undergo partial melting trace ele- ments partition themselves between the melt phase and solid residue. Those that prefer the solid are referred to as compatible (i.e. they have an af?nity with elements making up the crystal lattice of an existing mineral), whereas those whose preference is the melt are termed incompatible. Likewise, during cooling and solidi?cation of magma, com- patible elements are preferentially taken up in the crystals, whereas incompatible elements are enriched in the residual melt. Enrichment of trace elements and potential ore formation can, there- fore, be linked to the concentration of incompatible elements in the early melt phase of a rock under- going anatexis (see Box 1.2), or the residual magma during progressive crystallization (see Box 1.3). Compatible trace elements tend to be “locked up” in early formed rock-forming minerals and are typically not concentrated ef?ciently enough to form viable ore grade material. An exception to this is provided, for example, during the formation of chromitite layers, as discussed in section 1.4.3 below. A brief description of how one can quantify trace element distribution during igneous pro- cesses is presented below, with an indication of how these processes can be applied to the under- standing of selected ore-forming processes. 1.4.1 The conditions of melting Despite the fact that temperatures in the upper mantle reach 1500°C and more, melting (or anatexis) is not as widespread as might be expected because of the positive correlation that exists between pressure and the beginning of melting of a rock (i.e. the solidus temperature). The astheno- sphere, which is de?ned as that zone in the mantle where rocks are closest to their solidus and where deformation occurs in a ductile fashion (which explains the lower strength of the asthenosphere relative to the elastically deformable lithosphere), is the “engine-room” where a considerable amount of magma is formed. Major magma-generating episodes, however, do not occur randomly and without cause, but are catalyzed by processes such as a decrease of pressure (caused, for example, by crustal thinning in an extensional regime such as the mid-ocean ridge) or addition of volatiles to lower the solidus temperature (such as during subduction and metasomatism). An increase in local heat supply is generally not important in the promotion of partial melting, although it is possible that mantle plumes might be implicated in this role. Partial melting, so called because source rocks very seldom melt to completion, invariably leaves behind a solid residue. It is a complex process that is affected by a number of variables, the most important being the nature of the mineral assemblage making up the protolith, as well as local pressure, temperature and water/volatile con- tent. Early experimental work on the progressive fusion of peridotite with increasing temperature (Mysen and Kushiro, 1976) helps to explain the process of partial melting in the asthenosphere, even though the description outlined below is probably not a particularly good indication of the very complicated processes actually involved. In Figure 1.12 melting of peridotite starts just above 1400°C with the formation of a small melt fraction in equilibrium with pyroxenes and olivine. Melting can continue without signi?cant addition of heat until between 30 and 40% of the rock is molten and the clinopyroxene is totally consumed. Once the clinopyroxene is gone, further melting can only continue with addition of heat (the 38 PART 1 IGNEOUS PROCESSES ITOC01 09/03/2009 14:37 Page 38?rst in?ection point on the curves in Figure 1.12), which, if present, would then promote the melt- ing of orthopyroxene. Once orthopyroxene has been consumed, after some 50–60% of the rock has melted, a further input of heat is required (the second in?ection point) if olivine is to be wholly included into the melt product. A small degree of partial melting of peridotite will, therefore, yield magma with a composition that is dominated by the melt products of clinopyroxene (an alkali basalt), with the residue re?ecting the bulk com- position of orthopyroxene +olivine. In reality melting processes are more complex and usually involve the fusion of more than just one mineral at a time. It is also pertinent to note that the presence of even a small amount of water in the system will catalyze the extent of melting and also allow anatexis to occur at lower temperatures. Anatexis in the crust is better explained by con- sidering the partial melting of a metasedimentary protolith, to form, for example, an S-type granite. If the sediment consisted only of quartz then the temperature would have to exceed 1170°C (the melting temperature of SiO 2 at P H 2 O = 1 kbar) in order for it to start melting. If the sediment were an arkose, however, made up of a recrystallized assemblage of quartz + orthoclase + albite (as shown in Figure 1.13), then melting would commence at much lower temperatures because of the depres- sion of melting points that occurs for binary or ternary eutectic mixtures. Thus, where quartz + albite are in contact, melting would commence at IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 39 1400 Temperature (°C) 1800 1600 100 1500 1700 50 10 80 09 0 20 30 Melt (%) 40 60 70 20 kbar Ol + Opx + Cpx + Liq Ol + Opx + Liq Ol + Liq Anhydrous With 1.9 wt% H 2 O (a) (b) Figure 1.12 The sequential melting behavior of peridotite in the mantle at 20 kbar pressure, (a) without water and (b) in the presence of some 2% water (after Mysen and Kushiro, 1976). Ab Or Qz Or Or Qz Qz Qz Ab Ab Ab Or Qz Qz Or No melting Partial melt Figure 1.13 The pattern of grain boundary melting of an arkosic protolith subjected to temperatures of 700–800 °C at P H 2 O = 1 kbar (after Hall, 1996). ITOC01 09/03/2009 14:37 Page 39temperatures around 790 °C and, where quartz + albite + orthoclase meet at a triple junction, melt- ing could start as low as 720°C. Disaggregation of the arkose protolith would, therefore, take place by small increments of partial melt forming along selected grain boundaries within the rock. The residue left behind during such a process is likely to be made up of fragmented minerals that cannot melt on their own at a given temperature. Melt and residue are also likely to have different compositions. The extraction of a partial melt from its residue, whether it be from an igneous or sedimentary protolith, is a process which segregates chemical components and is referred to as fractionation. Partial melts can be considerably enriched in certain elements, but depleted in others, relative to the source rock. The mechanisms whereby melt is removed from its residue may differ, and this is discussed in more detail below. More detailed, quantitative descriptions of partial melt modelling are provided in Cox et al. (1979), Rollinson (1993), and Albarede (1996). Trace element distribution during partial melting In theory there are two limiting extremes by which partial melting can occur. The ?rst envisages forma- tion of a single melt increment that remains in equilibrium with its solid residue until physical removal and emplacement as a magma. This pro- cess may be applicable to the formation of high viscosity granitic melts (Rollinson, 1993) and is known as “batch melting.” It is quanti?ed by the following equation, whose derivation (together with others discussed below) is provided in Wood and Fraser (1976): C liq /C o = 1/[D res + F(1 - D res )] [1.1] where: C liq is the concentration of a trace element in the liquid (melt); C o is the concentration of trace element in parental (unmelted) solid; D res is the bulk partition coef?cient of the residual solid (after the melt is extracted); F is the weight fraction of melt produced. A diagrammatic illustration of the extent of enrichment (or depletion) of an incompatible (or compatible) element in a batch partial melt is presented in Figure 1.14a, where the ratio C liq /C o is plotted as a function of the degree of melt pro- duced (F) for a variety of values of D res . Marked enrichments of highly incompatible elements (i.e. those with very small values of D res ) can occur in small melt fractions, with the maximum enrichment factor being 1/D res as F approaches 0. The second partial melt process is referred to as “fractional melting” and is the process whereby small increments of melt are instantaneously removed from their solid residue, aggregating else- where to form a magma body. This process may be more applicable to low viscosity basaltic magmas where small melt fractions can be removed from their source regions. The distribution of trace ele- ments during fractional melting is quanti?ed in terms of the following equation: C liq /C o = 1/D o (1 - F) (1/D o - 1) [1.2] where: D o is the bulk partition coef?cient of the original solid (prior to melting); and the other symbols are the same as for equation [1.1]. The extent of trace element enrichment and depletion in a fractional partial melt is shown in Figure 1.14b. For very small degrees of melting the changes in trace element concentrations relat- ive to the source material are extreme and vary from a maximum value of enrichment (1/D o as F approaches 0) to depletions in the magma as melt- ing progresses. Unlike the batch melt situation, compatible element enrichment can also occur during fractional melting, but this is only likely to happen in the unlikely event of more than 70% partial melting. An example of incompatible ele- ment concentration in a granitic partial melt is provided in Box 1.2. 1.4.2 Crystallization of magmas The melt sequence for peridotite described in Figure 1.12 can be used in reverse to illustrate the crystallization of an ultrama?c magma as it cools (bearing in mind that in reality a magma derived from complete melting of a peridotitic protolith at around 1800 °C is unlikely ever to have existed). 40 PART 1 IGNEOUS PROCESSES ITOC01 09/03/2009 14:37 Page 40IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 41 The Rössing mine in Namibia is one of the largest uranium deposits in the world, and certainly the largest associated with an igneous rock. This deposit is associated with leucogranite (locally termed “alaskite”). The ore occurs mainly in the form of disseminated uraninite (UO 2 ) dis- tributed unevenly throughout the leucogranite, together with secondary uranium silicate and oxide minerals such as beta-uranophane and beta?te. Rössing occurs in the central zone of the c.500Ma Damaran orogen and the leucogranites are thought to be a product of partial melting of older basement comprising granite and supra- crustal (metasedimentary) sequences. Upper amphibolite to granulite grades of metamorphism apply regionally and melts are considered to have been derived from depths not signi?cantly greater than the actual level of emplace- ment, as indicated by the profusion of leucogranitic dykes in the region. The Rössing mine is located where several of the larger leucogranitic dykes have coalesced to form a signi?cant mass of mineable material. The deposit com- prises several hundred million tons of low grade uranium ore running on average at about 0.031% (or 310 ppm) U 3 O 8 . Although the deposit is low grade, a concentration of 310 ppm uranium in the leucogranite nevertheless represents a signi?cant enrichment factor relative to the Clarke value (i.e. about 2.7 ppm for U; see Table 1.2). This degree of enrichment is consistent with what might be expected for a magma formed by a low degree of partial melting of an already reasonably enriched protolith. This can be tested very simply in terms of the partial (batch) melt equation, given as equation [1.1] in section 1.4.1 of this chapter. If the protolith contained 10 ppm U and its bulk partition coef?cient relative to the melt fraction was signi?cantly less than 1 (i.e. D res between 0.01 and 0.1) then a 5% batch melt would yield a magma that is enriched by factors of between 7 (for D res = 0.1) and 17 (for D res = 0.01), as shown in Figure 2. This process would have resulted in uranium concentrations in the leucogranite of between 70 and 170 ppm, which is at best only one-half of the observed average concentration. In order to achieve concentrations of around 300 ppm U by batch melting either the degree of melting would have to have been very low (i.e. <5%), or the protolith would itself have to have been signi?cantly enriched, and have contained substantially more than 10 ppm U. Partial melting and concentration of incompatible elements: the Rössing uranium deposit Figure 1 Two phases of leucogranite cutting through high-grade metasediments in the vicinity of the Rössing uranium mine, Namibia (photograph by Paul Nex). 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F D res = 0.1 D res = 0.01 Batch melting = C liq C O D res + F (1 – D res ) 1 C liq C O Progressive melting 17× 7× 5% partial melt Figure 2 Simpli?ed batch melt model showing the degree of enrichment expected for an incompatible element such as uranium (with D res of either 0.1 or 0.01) after 5% melting (further details provided in section 1.4.1). ITOC01 09/03/2009 14:37 Page 4142 PART 1 IGNEOUS PROCESSES The crystallization sequence would have com- menced with olivine, followed by orthopyroxene and clinopyroxene. Together with plagioclase, this mineral assemblage and the crystallization sequence is typical for a basaltic magma. The sequential crystallization of a ma?c magma is, of course, another way of fractionating chemical components, since minerals of one composition are often physically separated from the composi- tionally different magma from which they form. Once minerals are removed from the magma (by a process such as crystal settling) there is little or no further chemical communication (or equilibration) between the solid and liquid components of the chamber. This process is referred to as fractional crystallization or Rayleigh fractionation. The nature of fractional crystallization, and the geo- metry of cumulate rocks that form during these processes, are both very relevant to an under- standing of ore deposits hosted in igneous rocks. It is, therefore, necessary to consider these processes in more detail and, in particular, their relationship to metal concentration and ore formation. It should also be emphasized at this stage that emplace- ment and crystallization of magma is generally accompanied by varying degrees of assimilation of country rock by the magma. Consequently, chem- ical trends re?ecting fractionation processes are Research by Nex et al. (2001) has shown that partial melting in the central zone of the Damaran orogeny was an episodic affair and that only one of several generations of leucogranite is signi?cantly enriched in uranium. Since all are likely to have been generated by approximately the same degrees of partial melting, this would suggest that the single enriched episode was derived from a particularly fertile protolith. It is apparent that deposits representing enrichment of incompatible trace elements in rocks formed by small degrees of melting, like the Rössing leucogranites, are not very common in the Earth’s crust. Another example might be the Johan Beetz uranium deposit in Quebec, Canada. = 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F C O 0.1 1 2 5 D = 10 0.01 0.0001 (a) Batch melting – liquids 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F C liq C O 0.1 10 2 5 D = 1 0.01 (b) Fractional melting C liq 1 D res + F (1 – D res ) = C liq C O C liq C O 1 D O (1 – F ) (1/D – 1) O Figure 1.14 The enrichment/depletion of a trace element in a partial melt relative to its concentration in the source rock (C liq /C o ) as a function of increasing degrees of melting (F). (a) Batch melting using equation [1.1], and (b) fractional melting using equation [1.2] (after Rollinson, 1993). ITOC01 09/03/2009 14:37 Page 42accompanied by the effects of magma contamina- tion and this, too, has important implications for igneous ore-forming processes (see sections 1.4.3 and 1.6). The form and internal zonation of igneous bodies The emplacement mechanisms of magma into the Earth’s crust and the resulting shapes of igneous intrusions is a complex topic and is reviewed in Hall (1996), Pitcher (1997), and Best (2003). Basaltic intrusions are typically quite different in shape and form from granitic batholiths. In addition, the mechanisms of crystal fractionation within a ma?c magma chamber are also differ- ent to those applicable inside granite intrusions. These differences are relevant to ore formation in crystallizing magmas, as exempli?ed by the stratiform nature of chromitite seams in ma?c intrusions such as the Bushveld Complex (see Box 1.4), compared to tin deposits in granites of the same Complex which tend to be concentrated in disseminated form toward the center of such intrusions (see Box 1.3). Ma?c intrusions The viscosity of ma?c magma is relatively low, as illustrated by the ease with which basaltic magmas ?ow on eruption. The densities of ma?c minerals are typically greater than 3gcm -3 , whereas ma?c magma has a density of around 2.6 g cm -3 . The low viscosity of ma?c magma and the high densities of minerals crystallizing from it imply that minerals such as olivine and the pyroxenes will typically sink in a magma at velocities of anywhere between 40 and 1000 m yr -1 depending on their composition and size (Hall, 1996). By contrast, less dense minerals, such as the felspathoids, might ?oat in an alkaline magma as they have densities less than 2.5 g cm -3 . Plagioclase would ?oat in a basaltic magma at pressures greater than about 5 kbar, but would sink in the same magma emplaced at high crustal levels (Kushiro, 1980). The sequence of layered rocks that result from gravitationally induced crystal settling are referred to as cumulates and their compositions differ from that of the starting magma. Crystal settling is, therefore, a form of fractional crystallization and this process could explain the segregation of chemical constituents and their possible con- centration into either of the solid or liquid phases of the chamber. Trace elements that are readily incorporated (by substitution) into cumulus min- erals are referred to as compatible elements. By contrast, trace elements which are excluded from the cumulate assemblage (because they cannot easily substitute into the crystal lattice sites of the major rock-forming minerals) are called incom- patible elements and they will, naturally, become progressively enriched in the residual magma as the cumulate assemblage is formed. Gravitationally induced settling of minerals in ma?c intrusions would seem to be the most logical way of explaining the well layered internal structure so often observed in these bodies. Early models for explaining the crystallization and layering in ma?c intrusions were derived from the classic work of Wager and Brown (1968) on the Skaergaard intrusion of Greenland. The latter is particularly well suited to petrogenetic study because it is considered to represent a single pulse of magma that crystallized to completion in a closed chamber without substantial addition or removal of magma during solidi?cation. The sub- horizontal layers that form in ma?c intrusions arise from the initial accumulation of early formed minerals (such as olivine and orthopyroxene), fol- lowed, in a stratigraphic sense, by accumulations of later formed minerals such as the pyroxenes and plagioclase. Minor oxide minerals such as chromite and magnetite are also observed to accumulate among the major silicate mineral phases. The Skaergaard intrusion is made up of a chilled mar- ginal phase and a relatively simple internal zone made up of sub-horizontal rhythmic and cryptic (i.e. where the layering is not visually obvious but is re?ected only in chemical variations) layers, ranging in composition upwards from gabbro to ferrodiorite and granophyre (Figure 1.15). Although it is tempting to attribute the relat- ively ordered, sub-horizontal layering that is so characteristic of intrusions such as Skaergaard to simple gravitationally induced crystal settling, solidi?cation of magma chambers, including Skaergaard, is a much more complex process (McBirney and Noyes, 1979). The temperature gradients and associated variations in magma density that form in cooling magmatic bodies IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 43 ITOC01 09/03/2009 14:37 Page 4344 PART 1 IGNEOUS PROCESSES have been shown to result in pronounced density strati?cation, with the formation of liquid layers through which elements diffuse in response to both chemical and temperature gradients (Huppert and Sparks, 1980; Turner, 1980; Irvine et al., 1983; McBirney, 1985). Density variations over time in a magma chamber are also a product of crystal fractionation itself. In Figure 1.16a it is apparent that during early stages of olivine crystallization the residual magma density decreases because the chemical components extracted by olivine have higher densities than that of the starting liquid. The trend changes, however, with the appearance on the liquidus of a mineral such as plagioclase, 3 2 1 0 km N S UZ MZ LZ MBS Platinova "reefs" Au, Pd ore Skaergaard intrusion HZ Country rocks Figure 1.15 Schematic illustration showing the nature of layering in the Skaergaard intrusion, east Greenland (after Anderson et al., 1998). The relative stratigraphic position of the Au- and Pd-bearing Platinova Reefs is shown in relation to the Marginal Border Series (MBS) and the Layered Series, which is subdivided into the Hidden Zone (HZ), Lower Zone (LZ), Middle Zone (MZ), and Upper Zone (UZ). 2.61 2.57 1400 Density (g cm –3 ) 2.60 2.59 2.58 1350 1300 1250 1200 1150 Temperature (°C) 0 5 10 Olivine Opx 15 20 30 40 50% (a) (b) Height Layer 2 Layer 1 “Fountain” Layer 2 Layer 1 “Plume” (40:60) Opx + plag New magma Figure 1.16 (a) Density variations in a fractionating magma similar in starting composition to that of the Bushveld Complex. (b) Contrasting behavior of a new magma injected into a density strati?ed magma chamber within which crystal fractionation has already occurred (after Campbell et al., 1983; Naldrett and von Gruenewaldt, 1989). ITOC01 09/03/2009 14:37 Page 44which has a density lower than that of the magma. In such a situation it is possible, after an interval of crystallization, for a residual magma to become more dense than when it started solidifying. This pattern of density variation has major implications for the behavior of crystals settling in a magma chamber, particularly when a new injection of magma takes place into an already evolved chamber. In Figure 1.16b two situations are considered. If the new magma has a density that is greater than the liquid residue in the chamber (as might be the case if the new magma was injected fairly early after crystallization had commenced), then a rather muted fountain-like feature will form and mixing between the new and evolved liquids will be limited to a layer along the base of the chamber. By contrast, if the new magma is injected late in the crystallization sequence, there is a possibility that its density will be less than that of the residual liquid and a more impressive plume-like feature would form (see Figure 1.16b) and the new magma would rise to its own density level, or to the very top of the chamber. In the latter scenario turbulent mixing between the new and residual liquids is likely to be more complete. Injection of a plume that reached and interacted with the roof of the magma chamber has been suggested for the Bushveld Complex (Schoenberg et al., 1999). This plume has also been strongly implicated in the formation of both the chromite and PGE mineralization in the Bushveld (Kinnaird et al. 2002). At a ?rst glance, layered ma?c intrusions such as Skaergaard may seem to be the products of relatively straightforward, gravitationally induced, crystal settling of minerals as they appear sequen- tially in the growth sequence. In detail, the nature of the layers may be quite complex and the normal, usually rather thick, cumulate assemblages are likely to be affected by complex processes invol- ving the relative rates of thermal and chemical diffusion. In addition, convective currents in the magma chamber, as well as turbidity currents in which a crystal-laden ?ow of magma surges down a slope on the margins or ?oor of the chamber, add to the complexities of crystallization. Finally, the composition of the minerals crystallizing at any one time will be sensitive to perturbations in the chemistry of the magma and rapid changes caused either by injection of new magma or con- tamination of existing magma by wall rock. The processes resulting in crystal fractionation, density variations and changes in magma composition are all critical to the formation of ore grade concen- trations of chromite, magnetite, platinum group elements, base metal sul?des and even gold in layered ma?c complexes. As the processes are better understood, such intrusions are increas- ingly attracting the attention of explorationists worldwide and even well studied bodies, such as Skaergaard, are now known to be mineralized. Although the presence of sul?de minerals at Skaergaard was recognized early on (Wager et al., 1957), the recognition of a potentially signi?cant Au and Pd resource only came much later (Bird et al., 1991; Andersen et al., 1998). The interaction between magmatic processes and the role that they play in mineralization are discussed in more detail below, and also illustrated in case studies. Felsic intrusions Granite intrusions do not exhibit the well de?ned sub-horizontal layering that typi?es large ma?c intrusions. This is largely due to the fact that felsic magmas are several orders of magnitude more viscous than ma?c ones (i.e. up to 10 6 Pa s for rhyolite compared to 10 2 Pa s for basalt at the same temperature; Hall, 1996). In addition, the density contrasts that exist in crystallizing ma?c magma chambers are not as marked as they are with respect to quartz and felspars forming in felsic chambers. Crystal set- tling is inconsequential in all but a few granite plutons, such as the relatively hot, hydrous alka- line granites whose parental magmas were emplaced at shallow depths in the crust. It is nevertheless evident that substantial fractionation does occur in felsic intrusions and there are many recorded examples of internal zonation in granite plutons (Pitcher, 1997). Most plutons record a con- centric zonation, with the outer zones preserving more ma?c compositions (i.e. diorite, tonalite, granodiorite) and rock types becoming progres- sively more fractionated toward the center. Since many granite plutons intrude at shallow crustal levels and most of their heat is, therefore, lost to the sidewalls, this type of zonation is logically attributed to crystallization that commenced from the sides and roof of the magma chamber IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 45 ITOC01 09/03/2009 14:37 Page 4546 PART 1 IGNEOUS PROCESSES and progressed inwards. Although crystals are not being removed from the magma by the settling process applicable in ma?c intrusions, they are effectively isolated from the residual melt by the crystallization front which advances in towards the center of the chamber. This process, referred to as sidewall boundary layer differentiation, can also, therefore, be regarded as a form of crystal fractionation and is characterized by concentra- tion of incompatible elements in the center of the intrusion where the ?nal increments of differ- entiated granite melt accumulate. Trace element distribution during fractional crystallization There are several ways by which crystallization can be modeled but the most appropriate would appear to be where crystals are removed from the site of formation with only limited equilibration between solid and liquid phases. In such a case trace element distributions can be described in terms of Rayleigh fractionation, in which: C liq /C o = F (D - 1) [1.3] for trace element concentration in the residual magma (liquid), and C sol /C o = DF (D - 1) [1.4] for trace element concentration in the crystalliz- ing assemblage (solid), where: C o is the original concentration of trace element in the parental liquid; D is the bulk partition coef?cient of the fractionating assemblage; F is the weight fraction of melt remaining. The extent of trace element enrichment and depletion during fractional crystallization of a magma is shown in Figure 1.17. The enrichment of an incompatible trace element in the residual melt relative to the original concentration in the parental magma is demonstrated in Figure 1.17a, where abundances can be seen to increase expon- entially as crystallization proceeds (i.e. as the fraction of melt remaining, F, decreases). Compat- ible elements will, of course, continuously decrease in the residual melt as they are extracted into the solid phases. Figure 1.17b shows, however, that the relative concentration of a compatible element in the crystallizing assemblage will start off as enriched only for rocks formed in the early stages of fractionation. Relative concentrations will decrease as crystallization progresses because the magma rapidly depletes itself of the com- patible elements as they are withdrawn into the crystallizing assemblage. Rayleigh fractionation equations apply reason- ably well to low viscosity basaltic magmas where the effects of crystal fractionation are very evident. As mentioned previously, though, granitic magmas do not exhibit well de?ned igneous strati?cation and they tend to solidify by inward nucleation of a sidewall boundary layer. This in situ style of crystal fractionation, where the crystallizing front is spatially distinct from the magma residue, is akin to Rayleigh fractionation, but the equations that govern trace element distribution have to be modi?ed, as shown below, to accommodate the melt fraction that is returned to the magma chamber as the solidi?cation zone moves pro- gressively into the chamber (Langmuir, 1989): C liq /C o = (M liq /M o ) [f(D - 1)/(D(1 - f) + f)] [1.5] where: M liq is the mass of liquid remaining in the magma chamber; M o is the initial mass of the magma chamber; f is the fraction of liquid in the solidi?cation zone returned to the magma chamber; and the other symbols are as in equa- tion [1.4]. The distribution trends of compatible and incompatible trace elements during this style of progressive inward solidi?cation are essentially similar to Rayleigh fractionation (Figure 1.17) although the degrees of concentration, especially at low values of f, are not as extreme (Rollinson, 1993). This accords with observations that the effects of crystal fractionation in granites tend to be less evident than in ma?c intrusions. An example of trace element concentration during inward nucleation of a sidewall boundary layer in granite is provided by the Zaaiplaats Sn deposit in the granitic phase of the Bushveld Complex in South Africa (see Box 1.3). There ITOC01 09/03/2009 14:37 Page 46IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 47 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F C liq C O 0.1 1 2 5 D = 10 0.01 (a) Rayleigh fractionation – liquids = F (D-1) C liq C O = D.F (D-1 ) 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F C sol C O 0.1 1 2 5 D = 10 (b) Rayleigh fractionation – residue, instantaneous solid C sol C O Figure 1.17 Trace element distribution during fractional crystallization. (a) The enrichment/depletion of a trace element in the residual magma relative to its concentration in the original melt (C liq /C o ) as a function of the fraction of the remaining magma (F) using equation [1.3]. (b) The enrichment/depletion of a trace element in a crystallized assemblage (immediately removed from its melt) relative to its concentration in the original melt (C sol /C o ) as a function of the fraction of the remaining magma (F) using equation [1.4] (after Rollinson, 1993). The granites of the Bushveld Complex, which overlie the better known layered ma?c intrusion, occur as large sill-like intrusions and are also known to be strongly fractionated (McCarthy and Hasty, 1976; Groves and McCarthy, 1978). The granites of the Bushveld Complex are enriched in Sn and several deposits occur in the more highly fractionated parts of the suite. One of the better studied examples is the Zaaiplaats tin mine, which obtains the bulk of its tonnage from a zone of low-grade, disseminated cassiterite (SnO 2 ) mineralization that occurs within the central portion of the granite (Figure 1). A cross section through the ore body, from diamond drill core, de?nes the low grade zone of disseminated cassiterite toward the center of the granite sheet. This is consistent with the suggestion that solidi?cation occurred from the margins inwards and that Sn was fractionated by processes akin to boundary layer differentiation. This type of crystal fractionation and its implications for ore-forming processes can be evaluated in terms of trace element distribution patterns. If boundary layer differentiation is modeled in terms of Rayleigh Fractionation (equations [1.3] and [1.4] in section 1.4.2), then the degree of crystallization required to concentrate Sn in the residual magma to the levels observed in the disseminated zone at Zaaiplaats can be calculated. Assuming that Sn is incompatible in the crystallizing granitic magma (i.e. D = 0.1) then some 96% crystallization is required in order to achieve an enrichment factor of 20 × (Figure 2). Coetzee and Twist (1989) have Boundary layer differentiation in granites and incompatible element concentration: the Zaaiplaats tin deposit, Bushveld Complex ITOC01 09/03/2009 14:37 Page 4748 PART 1 IGNEOUS PROCESSES shown that unmineralized granite of the type that hosts tin mineralization in the Bushveld Complex contains between 8 and 14 ppm Sn, whereas mineralized portions of the suite, like the disseminated zone at Zaaiplaats, average around 270ppmSn. The disseminated mineralization at Zaaiplaats is, therefore, consistent with Sn concentra- tion by crystal fractionation. After an advanced degree of solidi?cation, it is suggested that the Sn content of the residual magma was suf?ciently enriched to promote cassiterite crystallization. Other factors are also necessary in order to stabilize cassiterite in granites (Taylor and Wall, 1992), and these include fO 2 and magma composi- tion (speci?cally the Na/K ratio). These factors, together with the degree of enrichment required, account for the fact that cassiterite is seldom seen as an accessory mineral in granites. Chilled marginal granite Granophyre Meters 100 0 200 300 Zone of Sn enrichment (>35 ppm Sn) Pegmatite f S N Surface Granite (2–5 ppm Sn) 100 0.1 0 10 1.0 0.2 0.4 0.8 1.0 0.6 F D = 0.1 Rayleigh fractionation C liq C O 20× 96% fractional crystallization = C liq C O F (D–1) Fractional crystallization Figure 1 Simpli?ed cross section through the Zaaiplaats tin mine in the northern portion of the Bushveld Complex, South Africa (after Coetzee and Twist, 1989). Figure 2 (left) Rayleigh fractionation model showing the degree of enrichment expected for an incompatible element (D = 0.1) such as Sn after 96% crystallization (further details provided in section 1.4.2). ITOC01 09/03/2009 14:37 Page 48IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 49 are many examples of trace element concentra- tions in layered ma?c intrusions and subsequent sections discuss some of the complex processes involved in accumulation of ore grade Pt, Cu, and Ni deposits such as those of the Bushveld Complex, Kambalda, and Sudbury (see Boxes 1.4, 1.5, 1.6, and 1.7). A case where crystal fractiona- tion alone, without the added complication of features such as magma replenishment or con- tamination, has resulted in signi?cant enrichment of Au, Pd, and S is provided by the Platinova Reefs of the Skaergaard Complex. Figure 1.15 shows that this zone of potentially ore grade precious element enrichment occurs toward the top of the Middle Zone and formed after a substantial proportion of the magma chamber had crystallized. The incom- patible nature of Au, Pd, and S with respect to the early formed crystals (dominantly plagioclase, olivine, and pyroxene with lesser magnetite and apatite) resulted in a progressive enrichment of these metals in the residual magma (Anderson et al., 1998). Once the concentration of sulfur had reached saturation levels (believed to be between 0.16 and 0.3 wt% in the Skaergaard magma) the metals precipitated out as a Au–Pd alloy to be further concentrated as inclusions within the sul?de minerals that now occur distributed interstitially among the normal cumulus mineral assemblage. The concentrations of Au and Pd that resulted from the progressive in situ crystal fractionation of the Skaergaard magma amounts to about 2 ppm for each metal and it has been estimated that some 90 million tons of potential ore grade material is available in the Platinova Reefs (Anderson et al., 1998). 1.4.3 Fractional crystallization and the formation of monomineralic chromitite layers Crystal settling, convective ?uid ?ow, and diffusion-related chemical segregation across density strati?ed layers are the processes which give rise to the characteristically sub-horizontal, well ordered layering evident in most layered ma?c intrusions. These processes do not, how- ever, account for the occasional development of monomineralic layers of chromite or magnet- ite so spectacularly developed in many layered ma?c intrusions around the world. Nor do they explain the formation of the massive pods of chromitite that occur in the lithospheric portions of ophiolite complexes. This section provides some insight into the processes by which layers and pods of massive chromitite can form in ma?c cumulate rocks. There are many ma?c intrusions that con- tain layers of near monomineralic chromite or (vanadium-rich) magnetite that are typically 0.5– 1 meter in thickness and extend laterally for tens of kilometers, representing enormous reserves of Cr and Fe–V ore (see Box 1.4). Notable among these are the Bushveld Complex in South Africa and the Great Dyke in Zimbabwe. The forma- tion of such layers, which might comprise up to 90% of a single mineral (chromite or magnetite), would appear to require that normal crystalliza- tion of silicate minerals (dominated by olivine, the pyroxenes, and plagioclase) be “switched off” and replaced by a brief interlude where only the single oxide phase is on the liquidus. A simple, but elegant, explanation of this process was provided by Irvine (1977) with respect to the formation of chromitite seams. The Irvine model The Irvine model refers to part of the phase diagram for a basaltic system in which only the olivine–chromite–silica end-members are por- trayed in a ternary plot (Figure 1.18a). The normal crystallization sequence in a basaltic magma with starting composition at A (Figure 1.18b) would commence with olivine as the only mineral on the liquidus, settling of which would result in the formation of a dunitic cumulate rock. Extraction of olivine from the magma composition at A would result in evolution of the magma composition away from the olivine end-member composition and toward the cotectic phase boundary at B. At B a small amount of chromite (around 1%) would also start to crystallize together with olivine, and the magma composition would then evolve along the cotectic towards C. At C the SiO 2 con- tent of the magma has increased to a level where olivine and chromite can no longer be the stable liquidus assemblage and orthopyroxene starts to ITOC01 09/03/2009 14:37 Page 4950 PART 1 IGNEOUS PROCESSES crystallize to form a bronzitite cumulate rock. From this stage magma composition evolves toward D. Continued fractional crystallization will eventu- ally lead to the appearance of plagioclase together with orthopyroxene on the liquidus, but felspar compositions are not re?ected on the simpli?ed phase projection shown here. This crystalliza- tion sequence cannot lead to the formation of a chromite seam and the latter mineral would only occur as an accessory phase in the early formed cumulates. In order to make an ore deposit something unusual, or different from the norm, needs to take place. One way of disturbing the normal crystal- lization sequence is to introduce, at point D, a new magma with a composition at E (i.e. not as primitive as the original starting liquid), that is injected into the chamber and allowed to mix with the evolved liquid at F (Figure 1.18c). Mingling of the two liquids represented by D and E would result in a mixture whose composition must lie somewhere along the mixing line DE. Exactly 10 0.5 Wt% Chromite Mixing line between magmas at D and E 1.0 1.5 20 30 40 50 Wt% SiO 2 F G E B A D (c) 10 0.5 Wt% Chromite 1.0 1.5 20 30 40 50 Wt% SiO 2 H G E B A D (d) 2.0 SiO 2 Contamination trajectory C 10 0.5 Wt% Chromite Chromite 1.0 1.5 20 30 40 Wt% SiO 2 B A D (b) 2.0 C Olivine (a) Quartz Olivine Chromite Area covered by b, c, d Orthopyroxene Olivine Olivine Olivine Figure 1.18 A portion of the ternary system quartz–olivine–chromite (a) showing the nature of crystallization in a ma?c magma (b). Scenarios in which magma mixing (c) and magma contamination (d) occur as mechanisms for promoting the transient crystallization of only chromite are also shown (after Irvine, 1977). ITOC01 09/03/2009 14:37 Page 50where along DE this mixture would lie depends on the relative proportions of D and E that are mixed together. For most mixtures (at point F, for example, in Figure 1.18c) the magma com- position would lie within the stability ?eld of chromite and for a brief interval of crystallization (from F to G) only chromite would crystallize from the mixture. Being relatively dense, chromite would settle fairly ef?ciently and a single, near monomineralic layer of chromite would form. In a very large magma chamber, such as the one from which the Bushveld Complex must have developed, the crystallization of such a layer could lead to the formation of an ore body with IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 51 Crystal fractionation and formation of monomineralic chromitite layers: the UG1 chromitite seam, Bushveld Complex Figure 1 Simpli?ed geological map and section through the Bushveld Complex. The distribution of chromitite layers or seams in the Critical Zone is shown in relation to the PGE-rich Merensky Reef and the vanadium-rich magnetite seams at the base of the Upper Zone. Also shown is the distribution of associated granite and rhyolite that overlies the layered ma?c intrusion. The Bushveld Complex is the world’s largest layered ma?c intrusion (Figure 1), covering an area of some 67 000 km 2 . It also contains a substantial proportion (more than 75%) of the world’s chromite reserves. Several major chromitite seams (at least 14 in number) occur within the Critical Zone of the Bushveld Complex and these are subdivided into three groups termed the Lower Group (LG1 to LG7), the Middle Group (MG1 to MG4), and the Upper Group (UG1 to UG3). The LG6 chromitite seam is the most important in terms of production and reserves and can be traced for over 160 km in both the western and eastern portions of the complex. In section 1.4.3 of this chapter the mechanisms by which monomineralic layers of chromitite could form in a layered ma?c intrusion were discussed with speci?c reference to the Irvine model. The UG1 chromitite layer is Pretoria Rustenburg Johannesburg Zaaiplaats Sn mine Bushveld Complex Dwars River N 05 0 km Magnetite layers Merensky Reef Chromitite layers Granite and rhyolite Upper zone 3000 m Main zone Critical zone UG MG LG Lower zone ITOC01 09/03/2009 14:37 Page 5152 PART 1 IGNEOUS PROCESSES potentially vast Cr reserves. Experimental con- ?rmation of these processes has been provided by Murck and Campbell (1986). Once the magma composition reaches point G (after extraction of only chromite) on the cotectic, crystallization will again be dominated by olivine and the rocks that form in the hanging wall of the chromite seam will again contain only accessory amounts of chromite. It should also be mentioned that another way of forcing the magma com- position into the chromite ?eld is shown in Figure 1.18d, where the magma at point E (or any- where along the cotectic for that matter) becomes contaminated with siliceous material (perhaps by assimilation of crustal material forming either the ?oor or the roof of the magma chamber). The contaminated magma would have a composition that lies somewhere along the mixing line join- ing E to the SiO 2 apex of the ternary diagram. This composition would also lie transiently in the chromite ?eld and result in the formation of a monomineralic cumulate layer of chromite (between H and G). This is an indication of how important contamination of magma can be to igneous ore-forming processes. Other mechanisms for the formation of chromitite layers or pods Although the Irvine model very neatly explains many of the characteristics of chromitite layers it is unlikely to apply to all situations and there are several other mechanisms that might pertain to the accumulation of monomineralic layers or pods. Two of the most likely, since they have been con?rmed experimentally, include changes in oxygen fugacity (fO 2 ) and total pressure (P Tot ) of the crystallizing magma. An increase in fO 2 will promote the stability of chromite and possibly allow the mineral to crystallize alone for a period of time (Ulmer, 1969). Increasing fO 2 in the magma could be achieved by a devolatilization reaction such as [1.6] below (after Lipin, 1993). 4FeCO 3 - 2Fe 2 O 3 + 4CO + O 2 [1.6] where CO 3 2- in solution breaks down to form carbon monoxide and free oxygen. However, because CO 2 and not CO is likely to be the domin- ant carbon species in basalt, it seems unlikely of particular interest because it is hosted essentially within anorthositic rocks (rather than olivine and orthopyroxene cumulates), and is also spectacularly well exposed in a gorge of the Dwars River in the eastern portion of the Bushveld Complex. The UG1 shows some intriguing features that require special explanation. One of the inter- esting aspects of the UG1 chromitite layer is the way in which the seam bifurcates. In some sections the seam splits into several thinner layers, whose cumulative thickness is similar to that of the single seam. Chromitite splits tend to occur in areas where anorthosite layers thicken into domal features, with the bifurcations opening out towards the core of the dome. Nex (2002) has explained this by suggest- ing that normal “sedimentation” of chromite (according to Irvine’s model) was interrupted by liquefaction of the footwall crystalline assemblage to form a slurry of plagio- clase feldspar and melt that erupted at the magma–cumu- late interface. The bottom-up accumulation of the slurry is responsible for the formation of the domal features in the anorthosite and also serves to dilute the top-down set- tling of chromite grains. There is, therefore, a correlation between liquefaction of the footwall cumulates, doming Figure 2 Bifurcating chromitite seams exposed at Dwars River in the eastern Bushveld Complex. of the anorthosites, and splitting of chromitite layers. This type of feature, together with many other intriguing textures and relationships, indicates that crystallization processes in magma chambers can be very complex. ITOC01 09/03/2009 14:37 Page 52that oxidation of a magma will be easily achieved and the process is generally not called upon as an explanation for chromite accumulation. By contrast, small increases in P Tot of the magma have now been shown to occur readily in basaltic chambers. Observations in the Kilauea volcano (Hawaii) have shown that a pressure increase (of up to 0.25kbar) can occur in the roof of the magma chamber as a result of CO 2 exsolution and expansion of the gas bubbles as they stream upwards in the magma (Bottinga and Javoy, 1990; Lipin, 1993). An increase in total pressure in the magma chamber will have the effect of shifting the phase boundary between olivine and chromite such that the ?eld of the latter phase expands. This would have the same effect as that pre- dicted in the Irvine model, in that chromite would crystallize alone until such time as the ambient pressure is restored (by egress of magma, volatiles or both). This mechanism has been proposed for the formation of chromitite layers in the Stillwater Complex (Montana), where it is suggested that CO 2 streaming and associated pressure increase accompanied ingress of a new magma pulse into the chamber (Lipin, 1993). Exsolution of volatiles and ?uids from a magma is discussed in more detail in Chapter 2. A note on podiform chromitites Although not as large as the deposits in layered ma?c intrusions, podiform chromitite ores in ophiolite complexes represent important resources in many parts of the world. The chromitite ores are typically found as irregular, stratiform to discordant, pods within dunitic and harzburgitic host rocks which themselves are often intensely deformed. In detail the ore textures, characterized by nodular and orbicular associations of chromite and olivine, suggest that the mingling of two magmas has given rise to crystallization of the chromitite ores. Ballhaus (1998) has suggested that the sites of chromite mineralization repres- ent zones in the oceanic lithosphere where low viscosity, olivine-normative melt mingled with a more siliceous, higher viscosity magma. The two melt fractions remain segregated, at least for the time it takes to accumulate chromitite ore. For thermodynamic reasons chromite nucleates pre- ferentially in the ultrama?c melt globules, with crystals forming initially along the metastable liquid interface (where mixing takes place at a small scale and a situation akin to the Irvine model (Figure 1.18) pertains) and then progressively throughout the globule. Diffusion of chromium from the siliceous magma, where no chromite nucleation has occurred, across the liquid inter- face into the ultrama?c melt globules, also takes place, with the result that the latter might ultim- ately be entirely replaced by chromitite. Because the siliceous magma acts as a chromium reservoir, the richest ores are considered to occur where the volume of the latter is high relative to that of the ultrama?c globules within which accumulation of chromitite is occurring. 1.4.4 Filter pressing as a process of crystal fractionation The separation of crystal phases from residual melt during the solidi?cation of magma is gener- ally attributed to gravitational segregation where crystals of either higher or lower density than the magma settle or ?oat to form horizontally layered cumulate rocks. As mentioned previ- ously, this is an oversimpli?cation of the actual processes, which often involve complex density and chemical diffusive controls. Another mech- anism by which crystal melt segregation can occur is the process known as ?lter pressing. The residual magma within a network of accumulat- ing crystals in a partially solidi?ed chamber can be pressed out into regions of lower pressure such as overlying non-crystalline magma or fractures in the country rock. The process is considered to apply even in more viscous granitic magma chambers where evolved, water-saturated melts are ?lter pressed into adjacent fractures created during hydrofracturing. Anorthosite hosted Ti–Fe deposits Large massif-anorthosite intrusions of Mesopro- terozoic age, located in the Paleohelikian and Grenvillian orogenies extending from North America into the Sveconorwegian province, are IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 53 ITOC01 09/03/2009 14:37 Page 53–carbonate or silicate–sul?de immiscibility). The phenomenon of immiscibility is best observed in extrusive rocks where rapid quenching prevents the segregated products from being rehomogen- ized. Philpotts (1982) noted two compositionally distinct glasses interstitial to cumulus minerals in a tholeiitic basalt. Their compositions were essentially granitic, on the one hand, and an unusual ma?c assemblage, comprising pyroxene, magnetite–ilmenite, and apatite, on the other. The August 1963 eruption of the Kilauea volcano in Hawaii provided evidence of another form of liquid immiscibility in which a directly observ- able sul?de melt separated from a cooling, basaltic magma (Skinner and Peck, 1969). The Duluth Complex in Minnesota contains several small occurrences of massive Cu–Ni sul?de mineraliza- tion, as well as rare discordant bodies of ilmenite– magnetite–apatite (in oxide to apatite proportions of about 2:1) referred to as nelsonites. These oc- currences are considered to provide evidence that both sul?de and Fe–Ti–P immiscible fractions separated from the Duluth magma and that these represent a compositional continuum of immis- cible products (Ripley et al., 1998). Although its occurrence is dif?cult to prove, and despite the fact that it is not a major process during magma evolution, immiscibility is very important as an ore-forming process in ma?c magmas and can lead to the formation of large and important deposits such as the PGE sul?de deposits of the Merensky Reef in the Bushveld Complex, South Africa (Box 1.6), the Ni–Cu sul?de ores at Kambalda in Western Australia (Box 1.5), and at Sudbury in Ontario (Box 1.7). Various types of immiscibility are discussed below, with special emphasis on silicate–sul?de immiscibility. 1.5.1 Silicate–oxide immiscibility It is well known that unusual, discordant bodies of magnetite–apatite or ilmenite/rutile–apatite (nelsonite) are preferentially associated with some alkaline rocks, as well as with anorthosite complexes. Early experimental work showed that it is possible to create two immiscible liquids, one quenching to form a mixture of magnetite and apatite in proportions of about 2:1, and the other a 54 PART 1 IGNEOUS PROCESSES the hosts to very important Ti and Fe deposits of magmatic origin (Force, 1991a; Gross et al., 1997). Well known examples of such deposits include Sanford Lake in the Adirondacks of New York state, USA, the Lac Tio deposit in the Allard Lake region of Quebec, Canada, and the Tellnes deposit in southern Norway. The deposits are typically thought to be related to large differentiated intru- sive complexes made up mainly of anorthosite, gabbro, norite, and monzonite rocks emplaced in the late tectonic to extensional stages of the orogenic cycle. Although mineralogically vari- able, the more important category economically is the andesine anorthosite type (or Adirondack type), which contains ilmenite–hematite as the principal ore minerals. The Ti–Fe oxide ore accumulations occur as stratiform layers and disseminations within the intrusive complexes themselves, or as more massive, higher-grade, cross-cutting or dyke-like bodies. These deposits are clearly a product of in situ crystal fractionation. Early extraction of a plagio- clase-dominated crystal assemblage results in con- centration of Fe and Ti in the residual magmas, which crystallize to form ferrogabbro or ferrodior- ite. Titaniferous magnetite or hemo-ilmenite (depending on the magma composition) also crystallize with disseminated layers formed by crystal settling and accumulation on the chamber ?oor. The more massive discordant bodies are considered to be a product of the pressing out of an Fe–Ti oxide mineral slurry – the slurry concen- trates to form an intrusive body often along the margins of the largely consolidated anorthosite complex, or into fractures and breccia in the host rocks. The Lac Tio orebody is an irregular, tabular intrusive mass some 1100 m long and 1000 m wide, whereas the Tellnes ores form part of a 14 km long dyke (Gross et al., 1997). 1.5 LIQUID IMMISCIBILITY AS AN ORE-FORMING PROCESS Liquid immiscibility is the segregation of two coexisting liquid fractions from an originally homogeneous magma. The two fractions may be mineralogically similar (silicate–silicate immis- cibility) or be very different (silicate–oxide, silicate ITOC01 09/03/2009 14:37 Page 54rock that is dioritic in composition (Philpotts, 1967). Further experimental studies showed that for a fairly broad range of rock compositions under conditions of high fO 2 , an immiscible FeO melt will separate from a magma of felsic composition (Naslund, 1976). A large immiscibiliity ?eld exists, for example, in the system KAlSi 3 O 8 -SiO 2 -FeO (Figure 1.19) at approximately atmospheric fO 2 conditions, but the gap is greatly diminished in a more reducing environment. The existence of immiscibility is also enhanced in magmas with high concentrations of P, Ti, and Fe, but the ?eld diminishes with increasing Ca and Mg (Naslund, 1983). The observation that oxidized iron-rich mag- mas can segregate into two immiscible liquids, one which is Fe-rich and the other a more normal silicate composition, has relevance to the forma- tion of Fe- and Ti-rich magmatic segregations in nature. In slowly cooled plutonic rocks resorption reactions are largely responsible for the rehomo- genization of segregated liquids, such that immis- cibility is not often seen. In extrusive magmas, however, quenching can occur, or a more dense oxide liquid could separate from its silicate coun- terpart to be forcibly injected into a different part of the magma chamber, or into fractured country rock. Such an immiscible Fe- or Ti-rich ?uid could also form a discrete magma or lava ?ow, a possibility that provides a theoretical basis for understanding the very important magnetite– hematite–apatite ores in, for example, the Kiruna district of northern Sweden (Freitsch, 1978). Another example of what appears to be a ?ow of immiscible magnetite lava that separated from andesitic lavas has been documented at the El Laco volcano in northern Chile (Naslund et al., 2002). A word of caution though – in the cases of both Kiruna and El Laco there is considerable debate about the origin of the iron ores, and in the latter example in particular, a strong case has been made suggesting a hydrothermal origin for the Fe mineralization (Rhodes and Oreskes, 1999; Sillitoe and Burrows, 2002). The question of sili- cate–oxide immiscibility as a viable ore-forming process is still very contentious, despite the fact that the process has been proven experimentally. 1.5.2 Silicate–sul?de immiscibility By contrast with the process of silicate–oxide im- miscibility, where there is some controversy over its extent in nature, the existence of silicate– sul?de immiscibility in ma?c magmas is widely accepted as a common feature of magma crystal- lization. Experimental data in the system SiO 2 - FeO-FeS (MacLean, 1969) con?rms that silicate liquid can coexist with sul?de liquid over a large volume of the system (Figure 1.20; two-liquid ?eld). Magma at A in Figure 1.20, crystallizing Fe-rich olivine (fayalite), would evolve along A–A' and eventually intersect the two-liquid phase IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 55 (b) (a) fO 2 = 10 –9 bars fO 2 = 10 –1 bars FeO KAISi 3 O 8 SiO 2 FeO KAISi 3 O 8 SiO 2 Immiscibility gap Figure 1.19 The nature of the “immiscibility gap” (shaded) that exists in felsic magmas under conditions of variable oxygen fugacity. The tie-lines indicate the coexisting liquid compositions (after Naslund, 1976). ITOC01 09/03/2009 14:37 Page 5556 PART 1 IGNEOUS PROCESSES boundary where the residual melt would com- prise conjugate silicate and sul?de melts. Sulfur as sul?de is dissolved in magmas by displacing oxygen bonded to ferrous iron. Sul?de solubility is, therefore, a function of FeO activity in the magma, but is also controlled by oxygen fugacity (fO 2 ), decreasing as fO 2 increases (MacLean, 1969). Sul?de solubility, or the amount of sul?de dissolved in the magma at saturation, will vary as a magma progressively crystallizes and, at any point when saturation is reached, small immis- cible globules of sul?de melt will form. Sul?de sat- uration can be achieved as solidi?cation proceeds and magma temperature falls, or by an increase in fO 2 , or by a decrease in the amount of ferrous iron in the magma (such as might occur during extraction of an Fe-rich phase; see Figure 1.20). As will be shown later, other factors, such as addition of externally derived sulfur, or ingress of new magma, can also promote saturation and the for- mation of an immiscible sul?de phase. The immiscible sul?de melt that segregated from basaltic magma on the Kilauea volcano con- tained approximately 61% Fe, 31% S, 4% Cu, and 4% O. It subsequently solidi?ed to form minerals such as pyrrhotite, chalcopyrite, and magnetite (Skinner and Peck, 1969). Segregated sul?de melts forming in this fashion clearly have enormous potential to host concentrations of metals with both chalcophile and siderophile tendencies, such as base (Cu, Ni, Co) and precious (Au, Pt) metal ores. There are many large and important ore deposits associated with the development of an immiscible sul?de fraction in ma?c and ultra- ma?c magmas (see Boxes 1.5, 1.6, and 1.7). Central to the formation of all these deposits are three fundamental steps: • the appearance of a substantial fraction of immiscible sul?de melt; • creation of conditions whereby the sul?de globules can effectively equilibrate with a large volume of silicate magma; • effective accumulation of the sul?de globules into a single cohesive layer or spatial entity. The processes whereby ore deposits are created during silicate–sul?de immiscibility are, there- fore, complex and multifaceted, and are discussed in more detail in the following section. FeO FeS Troilite Wustite Fayalite 2-liquids A' Tr A (fayalite) Cr 3-liquids 2-liquids 2-liquids SiO 2 Fe 2 SiO 4 Figure 1.20 Phase equilibria established experimentally in the system SiO 2 –FeO–FeS (after MacLean, 1969). The ?eld of two coexisting silicate and sul?de liquids is shown by the stipple accentuated line. Oxidation of the magma would shift the phase boundary in the direction shown by the double arrow, and expand the ?eld of two-liquids. A magma represented by composition A and crystallizing fayalite would evolve along the line A–A'. Silica phases are represented by tridymite (Tr) and cristobalite (Cr). ITOC01 09/03/2009 14:37 Page 561.6 A MORE DETAILED CONSIDERATION OF MINERALIZATION PROCESSES IN MAFIC MAGMAS 1.6.1 A closer look at sul?de solubility The question of whether an immiscible fraction will develop in a magma or not is related to the amount of sulfur required to achieve sul?de sat- uration. This concept is referred to simply as sul- ?de solubility. As mentioned previously, sul?de solubility decreases with increasing oxygen con- tent in a magma because solution of sulfur appears to be controlled by the following equilibrium reaction (after Naldrett, 1989a): FeO (melt) + 1 /2S 2 ? FeS (melt) + 1 /2O 2 [1.7] Sul?de solubility increases as a function of increasing temperature and FeO content of the magma, but decreases with increasing pressure and SiO 2 content. The data for sul?de solubility (reviewed in Naldrett, 1989a) have been used to construct a generalized solubility curve for a fractionating Bushveld type ma?c magma and are shown in Figure 1.21. The sulfur content at sul?de saturation is seen to decrease from a maximum of about 0.4 wt% sul?de at the start of crystallization to below 0.1 wt% as olivine and orthopyroxene crystallize, but then levels off after plagioclase forms. A magma with an original sul?de content of 0.3 wt% would initially be undersaturated and its position would plot below the curve in Figure 1.21. Extraction of olivine and pyroxene as cumulus minerals would cause the sul?de content to increase in the residual magma until the satura- tion limit was attained (i.e. after about 10% crys- tallization). At this point immiscible globules of homogeneous sul?de melt would form and the magma would remain saturated with respect to sul?de. The amount of sul?de remaining in the magma will, however, decrease as dictated by the solubility curve. If the high density sul?de globules are extracted from the magma by gravitational IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 57 0.4 0 0 Wt% sul?de 0.2 0.1 10 20 40 30 50 60 70 0.3 Crystallization (%) C (1180°C) AD A B (1300°C) D Olivine Orthopyroxene AC Plagioclase plus orthopyroxene (+clinopyroxene) Sul?de solubility curve Figure 1.21 Variation in sul?de solubility as a function of progressive crystallization in a ma?c magma such as that from which the Bushveld Complex formed (after Naldrett and Von Grünewaldt, 1989). ITOC01 09/03/2009 14:37 Page 5758 PART 1 IGNEOUS PROCESSES settling, then the distribution of trace elements be- tween sul?de and silicate liquids can be modeled in terms of crystal fractionation mechanisms. After solidi?cation the cumulate rocks that had formed during this stage of the crystallization sequence would contain minor disseminations of sul?de minerals (such as pyrrhotite, FeS, and chalcopyrite, CuFeS 2 ) among the olivine and orthopyroxene crystals. Alternatively, since the sul?de globules have a high density, they might also accumulate as a separate sul?de layer toward the base of the magma chamber. The decrease in sul?de solubility with progressive crystallization of a ma?c magma, as well as the shape of the solub- ility curve shown in Figure 1.21, have important implications for the understanding of ore-forming processes in these rocks and this diagram is referred to again below. 1.6.2 Sul?de–silicate partition coef?cients The ability of an immiscible sul?de fraction to concentrate base and precious metals will obvi- ously depend on the extent to which metals parti- tion themselves between sul?de and silicate melts (i.e. the magnitudes of the relevant sul?de-silicate partition coef?cients). All metals that have chal- cophile tendencies are likely to partition strongly into the sul?de phase rather than remain in the silicate melt, and the data presented in Table 1.3 con?rms that both base and precious metals are markedly compatible with respect to sul?de minerals. The partition coef?cients indicate that although Cu, Ni and Co partition strongly into the sul?de phase, scavenging of platinum group elements (PGE)* by sul?de melt is even more ef?cient. Values of partition coef?cients up to 100 000 for the PGE (Table 1.3) indicate that the presence of an immiscible sul?de fraction is poten- tially a very powerful concentrating mechanism. In reality, however, even though the sul?de– silicate partition coef?cients for the PGE are so high, one seldom ?nds economically viable con- centrations of these elements in layered ma?c complexes. One reason for this is that the original concentrations of PGE in magmatic reservoirs are very low. Another reason is that, even in magmas where immiscible globules of sul?de do form, the concentration mechanism may be diminished because sul?des are not able to communicate (chemically) with the entire magma reservoir that contains the metals. In the case of the Skaergaard intrusion, for example, sul?de disseminations occur mainly in the upper portions of the layered body and sul?de immiscibility is believed to have occurred relatively late in the crystallization sequence (Anderson et al., 1998). The sul?de minerals at Skaergaard are depleted in Ni because a signi?cant proportion of the latter metal was extracted by early formed olivine and orthopyrox- ene and was no longer available to be scavenged by the sul?de phase. On the other hand, the Skaergaard sul?de zone is cupriferous, and is also enriched in Au and Pd, as both these metals were concentrated into the residual magma as incom- patible elements during crystal fractionation. The Skaergaard sul?des have only, therefore, equilib- rated with the more differentiated parts of the magma chamber. * Six elements make up the platinum group – they are platinum (Pt), palladium (Pd), and rhodium (Rh), collectively referred to as the Pd subgroup, and osmium (Os), iridium (Ir), and ruthenium (Ru), referred to as the iridium subgroup. Both the Pd and Ir subgroups have strong af?nities with Fe–Cu–Ni sul?des, whereas the iridium subgroup is also often associated with chromite and olivine cumulates. Table 1.3 Estimates of sul?de–silicate partition coef?cients for base and precious metals in ultrama?c and ma?c magmas Ni Cu Co Pt Pd Komatiite 27% MgO 100 250–3000 40 10 4 –10 5 10 4 –10 5 19% MgO 175 250–2000 60 10 4 –10 5 10 4 –10 5 Basalt 275 250–2000 80 10 4 –10 5 10 4 –10 5 Source: from Naldrett (1989a), Barnes and Francis (1995), Tredoux et al. (1995). ITOC01 09/03/2009 14:38 Page 58The effects of early removal of trace elements are demonstrated in Figure 1.17, where it is seen that partitioning of a highly compatible trace ele- ment (e.g. D = 10 in Figure 1.17b) leads to enrich- ments in the solid cumulates only for the ?rst 20–30% of crystallization (i.e. for values of F from 1.0 to about 0.7) and, thereafter, the solids are increasingly depleted in that element. This is because a high-D trace element will comprehens- ively enter the early formed solids, leaving the residual magma strongly depleted. Later solids will have to equilibrate with a depleted reservoir and will likewise be depleted, irrespective of the magnitude of the partition coef?cient. An immis- cible sul?de melt within a silicate magma can also be regarded as a cumulus mineral phase and the same situation would, therefore, apply. In fact, the problem would be particularly acute for the PGE with respect to an immiscible sul?de fraction because their partition coef?cients are so high, and the original abundances in the parental magma so low (generally only a few ppb), that all but the very ?rst formed sul?de fraction will be depleted in these metals. One does, therefore, have to ask the very pertinent question as to how large and economically viable concentrations of low abundance metals actually form. Or, put another way, how do the enormous concentra- tions of PGE-bearing sul?de ores such as those of the Bushveld Complex form when early depletion of compatible metals apparently limits the extent to which viable concentrations of metals can be achieved? 1.6.3 The R factor and concentration of low abundance trace elements If a globule of sul?de melt equilibrates with an in?nite reservoir of magma then the concentra- tion of a trace metal in that, or any other similar, globule is given simply by the product of the parti- tion coef?cient and the initial concentration of the metal in the coexisting silicate melt. In reality of course the sul?de globule is only likely to inter- act with a small, restricted mass of silicate melt, which is why the problem alluded to at the end of the previous section is very relevant. Clearly, the greater the proportion of silicate relative to sul?de melt, the higher and more persistent the concentration of compatible trace metals in that globule is likely to be. Campbell and Naldrett (1979) quanti?ed the problem by proposing that the concentrations of low-abundance trace elements, such as the PGE, into an immiscible sul?de phase should be calculated in terms not only of the original abundance in the magma and the relevant partition coef?cients, but also of the silicate/sul?de liquid mass ratio, termed the “R” factor. An expression which incorporates the R factor into the normal distribution law is provided in equation [1.8] (after Campbell and Naldrett, 1979): C sul = C o D(R + 1)/(R + D) [1.8] where: C sul is the concentration of a trace element in the sul?de fraction; C o is the original trace element concentration in the host magma; D is the sul?de–silicate partition coef?cient; and R is the mass ratio of silicate magma to sul?de melt. Although the R factor is simply a mass fraction, it is a parameter that also effectively records the extent to which the immiscible sul?de fraction interacts with the silicate magma from which it is derived (and which also contains the metals that need to be concentrated if an ore deposit is to be formed). If a sul?de globule sinks through a lengthy column of magma (or is caught up in a turbulent plume) it is effectively interacting with a large volume of silicate liquid, and this can be equated to a high R factor (Campbell et al., 1983; Barnes and Maier, 2002). Even though a sul?de globule may rapidly deplete the surrounding magma at any one instant of time by virtue of a very high sul?de–silicate partition coef?cient, it may be provided with the opportunity to scavenge compatible elements by continuously moving through undepleted magma. A low R factor is analogous to a situation where a sul?de globule is static, or is removed early on from the magma, so that it is not able to ef?ciently scavenge compatible elements even though the partition coef?cients may be very high. The way in which the concentrations of com- patible trace metals in sul?de melts vary as a function of both partition coef?cient (D) and R is IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 59 ITOC01 09/03/2009 14:38 Page 5960 PART 1 IGNEOUS PROCESSES shown in Figure 1.22a (after Barnes and Francis, 1995). For cases where D is much larger then R, the enrichment factor (C sul /C o ) approximates the value of the R factor (i.e. the sloped portions of the curves in Figure 1.22a). Conversely, when the R factor is much larger than D, then the enrich- ment factor is approximately the value of D (Figure 1.22a – the ?at parts of the curves). Another way of emphasizing the importance of the R factor is to examine its in?uence on the concentrations in sul?de melt of two compatible trace elements, one with a moderately high parti- tion coef?cient (for example, Ni with D = 275) and the other with a very high partition coef?cient (Pt with D = 100 000). Figure 1.22b shows that where the R factor is low (say 10 3 ) Ni concentrations in sul?de fractions are high (i.e. typical of most Ni-sul?de ores) because of the combination of high D and high initial Ni content (350 ppm) of the parental magma. By contrast, the Pt concen- trations in the same sul?de fraction will be low because, despite the very large D, they have not had the opportunity of scavenging a substantial mass of the element from a magma that initially had a very small Pt abundance (5 ppb). By con- trast, in situations where the R factor is high (say 10 6 ) the Ni concentrations of the sul?de fraction will not be signi?cantly higher than the lower R factor case, but the Pt contents will have increased substantially because of more extensive interac- tion between the immiscible sul?de fraction and the silicate magma. Variations in the R factor have real implica- tions for the grades of sul?de phases within indi- vidual ma?c intrusions. In the Muskox intrusion of northern Canada, for example, disseminated Cu–Ni sul?des on the margins of the body con- tain only moderate concentrations of PGE, whereas the more voluminous interior of the intrusion contains sul?des with higher PGE grades (Barnes and Francis, 1995). This is attributed to a lower mass ratio of silicate to sul?de melt, and therefore a lower R factor, on the margins of the intrusion. Considerations such as this might be of consider- able relevance to an exploration strategy for this type of deposit. 1.6.4 Factors that promote sul?de immiscibility It is conceivable that situations in which sul?de saturation is attained relatively late in the crystal- lization history of a magma could be disadvant- ageous to the formation of, for example, Ni- or PGE-sul?de ores if the latter metals had already been extracted from the residual magma. It is also possible that late sul?de saturation could equate 8 Ni (wt%) 6 4 2 654 1 Log R 200 100 300 Pt (ppm) D = 10 5 X i = 5 ppb Pt Ni 500 3 2 400 D = 275 X i = 350 ppm Ni (R = 10 6 ) Ni (R = 10 3 ) Pt (R = 10 6 ) (b) Pt (R = 10 3 ) 10 1 C sul /C o 10 4 R factor (a) 10 2 10 3 10 4 10 5 10 3 10 2 10 1 D = 100 000 D = 10 000 D = 1000 D = 100 Figure 1.22 (a) Plot showing the relationships between partition coef?cients (D), the R factor (i.e. the mass ratio of silicate melt to sul?de melt), and the degree of enrichment (C sul /C o ) of compatible trace elements in the sul?de phase (after Barnes and Francis, 1995). (b) Diagram showing the effect of variations in the R factor on the concentration of Ni and Pt in an immiscible sul?de fraction that is in equilibrium with a basaltic magma (after Naldrett, 1989b). ITOC01 09/03/2009 14:38 Page 60with a low R factor simply because the volume of residual silicate magma is low. A scenario whereby magmas become sul?de saturated relat- ively early in the crystallization history of an intrusion is considered to be advantageous for the development of a wide range of magmatic sul?de deposits. There are several processes by which sul?de saturation can be promoted and these are discussed below. Addition of externally derived sulfur Perhaps the simplest way of promoting saturation in a magma whose composition is sulfur under- saturated is to increase the global amount of sulfur in the melt by addition from an external source. Some of the komatiite-hosted Ni–Cu deposits, such as Kambalda in Western Australia (Box 1.5) and several examples in Zimbabwe and Canada, are hosted in lava ?ows that have extruded onto sul?de-bearing footwall sediments such as chert, banded iron-formation, or shale. The komatiitic host lava commonly cuts down into its footwall, a feature attributed to a combina- tion of thermal erosion and structural dislocation. Good geological and geochemical evidence exists to suggest that the komatiitic magma arrived on surface undersaturated with respect to sulfur, and that saturation was achieved by a combination of crystal fractionation and assimilation of crustally derived sulfur derived from footwall sediments (Lesher, 1989). This concept is consistent with factors such as the localization of Ni–Cu ore within the footwall embayments created by ex- truded komatiitic lavas and is discussed in more detail in Box 1.5. Injection of a new magma and magma mixing As mentioned in section 1.3.2 above, large magma chambers are seldom the result of a single injec- tion of melt and most witness one or more replen- ishments during the crystallization sequence. The Bushveld Complex and the Stillwater Com- plex in Montana contain important chromite deposits as well as PGE-enriched base metal ores associated with a sul?de-rich layer (the Merensky reef in the Bushveld and the J-M reef at Stillwater). These two, economically important, layered ma?c intrusions were both characterized by crystall- ization that was interrupted by periodic injections of new, less differentiated magma. The magma replenishment events broadly coincide, in both cases, with the development of PGE-enriched sul?de horizons. In the Bushveld Complex very detailed Rb–Sr and Re–Os isotopic measurements indicate that the new magma was more radio- genic (perhaps due to crustal contamination) but less differentiated than the magma remaining in the chamber at the time of injection (Kruger and Marsh, 1982; Lee and Butcher, 1990; Schoenberg et al., 1999). Further discussion on magma replen- ishment and its role in mineralization in other ma?c intrusions is presented in reviews by Naldrett (1989a, b, 1997, 1999). The sul?de solubility diagram in Figure 1.21 can be used to demonstrate how the mixing of new and residual magmas can promote sul?de satura- tion. Although the proportions of cumulus min- erals being extracted from the magma, as well as the sulfur solubility curve itself, may not be uni- versally applicable, the principles could be modi- ?ed for application elsewhere. An initial magma composition represented by A in Figure 1.21, con- taining 0.3 wt% sul?de, would be undersaturated relative to sul?de (i.e. sul?de solubility at this stage is 0.4%). Extraction of olivine from the magma (A–B in Figure 1.21) would decrease its FeO content and result in sul?de saturation with formation of an immiscible sul?de fraction. With continued crystallization and segregation of sul?des the magma would remain saturated, follow- ing the solubility curve as shown in Figure 1.21. If after some 20% crystallization (at point C just prior to the appearance of plagioclase as a cumulus phase) the chamber is replenished with the injec- tion of a new magma (similar to A) and mixing occurs, then the composition of the mixture would lie somewhere along the mixing line between A and C. If the mixture is at AC it would clearly be undersaturated again and any sul?de globules present would be resorbed back into the magma. If, however, the chamber is replenished with the injection of a new magma at point D (after 35% crystallization, and after the introduction of plagio- clase as a cumulus phase), then a mixed magma IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 61 ITOC01 09/03/2009 14:38 Page 6162 PART 1 IGNEOUS PROCESSES An important category of magmatic Ni–Cu sul?de deposits is that related to ma?c and ultrama?c volcanic rocks (komatiites) in Archean greenstone belts. Deposits of this type occur in Canada (the Thompson Ni belt of Manitoba) and Zimbabwe (Trojan and Shangani mines), but the largest and richest occurrences occur in the Kambalda region of Western Australia. Although a great deal is now known about these deposits (Lesher, 1989) this case study focuses speci?cally on the source of sulfur in the immiscible globules of sul?de within the komatiitic lava ?ows that host the mineralization. Komatiites are ultrama?c extrusive rocks that were ?rst described in the Barberton greenstone belt, South Africa, by Viljoen and Viljoen (1969). They comprise mainly olivine + clinopyroxene and typically contain MgO > 18 wt% and low alkalis. As indicated in section 1.2.2 of this chapter they are characterized by high Ni contents. Komatiites are extruded onto the Earth’s surface as high temperature, low viscosity lava ?ows characterized by a variety of volcanic forms of which pillowed lava and bladed spinifex textures are the most diagnostic. Most komatiites are not mineralized and their Ni is resident mainly in cumulus olivine, suggesting that on extrusion these lavas were under- saturated with respect to sul?de. Evidence from the Kambalda region indicates that voluminous eruption of hot, low viscosity komatiitic lava ?ows formed extensive sheet ?ows and gradient-controlled lava rivers or channels (showing many of the features seen in parts of present day Hawaii). Hot lava rivers are believed to have thermally eroded discrete channels into the previously consolidated footwall and some of the Ni–Cu ores at Kambalda exhibit sul?de concentrations at the base of well de?ned, linear channelways (Figure 1). Another feature of this type of ore is that the mineralized channelways are devoid of inter?ow sediments even though such sediments occur laterally away from the ore zones at that level. These sediments comprise a variety of carbonaceous and sul?dic shales as well as sul?dic chert and banded iron-formation (Bavinton, 1981). The nature of these sediments has led to suggestions that they represented a source of sulfur and this is supported by S isotope studies showing similarities in the isotope ratios of ores and inter?ow sediments (Lesher, 1989). One implication of this idea is that assimilation of sediment by thermally eroding komatiitic lava channels enhanced the sul?de content within the magma and promoted local sul?de saturation and immiscibility. Sulfur saturation and immiscibility of a sul?de fraction early in the magma crystallization history is the funda- mental process responsible for mineralization in komatiitic lava ?ows. Chalcophile metals, in particular Ni and lesser Cu, were scavenged from the turbulent, ?owing komatiite magma by the immiscible sul?de globules, which eventu- ally accumulated as massive sul?de ore along the bottom of the channelways (Figure 1). The disseminated ore that overlies the massive sul?de ore re?ects the static buoy- ancy contrast that existed between komatiitic crystal mush and massive sul?de. External derivation of sulfur has also been suggested for the promotion of sul?de saturation in the very large Noril’sk–Talnakh Cu–Ni–PGE deposits in the Russian Federation (Grinenko, 1985) and also for the Duluth Complex in Minnesota (Ripley and Al-Jassar, 1987). Silicate–sul?de immiscibility: the komatiite-hosted Ni–Cu deposits at Kambalda, Western Australia Figure 1 The characteristics of komatiite-hosted Ni–Ci deposits in the Kambalda region, Western Australia (after Solomon et al., 2000). Spinifex textured komatiite Disseminated ore Lava channel (thermally eroded footwall) Massive Ni-Cu ore Massive basalt Sul?dic interflow sediment Pillowed komatiite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ITOC01 09/03/2009 14:38 Page 62IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 63 In addition to its huge chromite reserves, the Bushveld Complex also hosts the world’s largest reserves of platinum group elements (PGE). About 80% of the world’s PGE reserves are located in the complex, from three speci?c horizons, namely the Merensky Reef (where exploitation commenced and which itself contains about 22% of the world’s total PGE reserves; Misra, 2000), the UG2 chromitite layer, and the Plat Reef. A simpli?ed outline and section of the Bushveld Complex, showing the extent of the Merensky Reef, is presented in Box 1.4. This case study focuses speci?cally on the Merensky Reef as it was formed at the time when a major injection of new magma occurred into the chamber, and it also marks a regional mineralogical hiatus separating the Critical Zone from the Main Zone (Kruger and Marsh, 1985). It is typically repres- ented by a 1 meter thick, coarse-grained (or pegmatoidal), feldspathic pyroxenite that extends along strike for about 250 km. The origin of the Merensky Reef is a contentious issue (Naldrett, 1989a) and arguments range from purely magmatic models to those involving interaction with magmatic-hydrothermal ?uids (Cawthorn, 1999). There is ample evidence that the Merensky Reef and its contained PGE mineralization have been involved with a hydro- thermal ?uid, and this topic is discussed in more detail in Chapter 2. For the purposes of this discussion, however, it is accepted that the Merensky unit (i.e. the mineralized reef and the rocks immediately overlying it, called the Bastard unit) owes its origin to the turbulence and magma mixing that accompanied a magmatic replenishment event (Schoenberg et al., 1999). Mineralization in the Merensky Reef is evident in the presence of disseminated base metal sul?des, mainly chalcopyrite and pyrrhotite–pentlandite, with which minor PGE sul?des (such as braggite, cooperite, laurite, and New magma injection and magma mixing: the Merensky Reef, Bushveld Complex Lower Zone Initial 87 Sr/ 86 Sr ratio 0.704 0.705 0.706 0.707 0.708 0.709 0.710 Ferro-gabbronorite Harzburgite Critical Zone Main Zone Lower Lower Upper Upper Zone Upper Pyroxenite marker Merensky and Bastard units Gabbronorite Norite and Anorthosite Orthopyroxenite Figure 1 Sr isotope variations (in terms of the initial 87 Sr/ 86 Sr ratio or R o ) with stratigraphic height in the Bushveld Complex (after F.J. Kruger, personal communication). ITOC01 09/03/2009 14:38 Page 6364 PART 1 IGNEOUS PROCESSES Migration of residual liquid Movement Recrystallized/ metasomatized zone Flatter down-dip wall Steep up-dip wall Merensky unit magma Normal Merensky Reef Movement (b) Continued down-dip extension Recrystallized/ metasomatized zone New Merensky magma (a) Initial rupture Footwall cumulate assemblage Pothole reef Figure 3 Schematic cross sections illustrating the progressive development of potholes in the Merensky Reef by syn-magmatic extension in the footwall cumulate rocks just prior to and during injection of a new magma (after Carr et al., 1999). moncheite) and PGE metal alloys are associated. As a general statement, magma replenishment and mixing are believed to have been responsible for the formation of an immiscible sul?de fraction, and the concentration of PGE into these sul?des, at the time of Merensky Reef formation (Naldrett, 1989a). It must be emphasized, however, that mineralization in the Merensky Reef, as well as in the UG2 chromitite and Plat Reef, is a very complex process and this is discussed in somewhat more detail in section 1.6 and also again in Chapter 2. Evidence for a new magma pulse at the Merensky unit is most obvious in terms of mineral–chemical trends. The most convincing evidence for the input of a new magma, however, comes from a dramatic shift in the initial 87 Sr/ 86 Sr ratio (R o ) of rocks below and above the Merensky unit (Figure 1). Such a change cannot be achieved by crystal fractionation and must record the input of new magma. In this case the new magma had a higher initial 87 Sr/ 86 Sr ratio Figure 2 The edge of a small pothole showing anorthosite in contact with pyroxenite. The Merensky Reef should occur at the contact between the two rock types. ITOC01 09/03/2009 14:38 Page 64IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 65 composition at, for example, AD would be over- saturated with respect to sul?de. This would result in the segregation of an immiscible sul?de frac- tion, the mass of which exceeds the expected cotectic proportion. A period of enhanced sul?de production could be expected, therefore, to accom- pany the injection of a new magma at point D. Magma mixing is, therefore, a process that can lead to the production of an immiscible sul?de fraction. The process seems to be linked with the formation of some very important PGE-sul?de bearing horizons such as the J-M Reef in the Stillwater Complex and the Merensky Reef in the Bushveld Complex. It is considered again in the context of a more holistic model in section 1.7 below. Magma contamination In the discussion on the formation of monominer- alic chromitite seams (section 1.4.3 and Box 1.4) it was mentioned that one way of forcing a magma composition off its cotectic crystallization path into the stability ?eld of chromite was to contam- inate the magma with foreign material rich in SiO 2 . The same process will also result in a lower- ing of the sulfur content required for sul?de saturation, thereby promoting the formation of an immiscible sul?de fraction. Figure 1.23 shows a 1200 °C isothermal section of the SiO 2 -FeO-FeS system in which the silicate–sul?de immiscibil- ity ?eld is identi?ed. The composition of a homo- geneous magma undersaturated in sul?de at point A can be forced into the ?eld of two liquids by the addition of SiO 2 (toward B). At B the two liquids in equilibrium will have compositions at Y (silicate rich) and X (sul?de rich). This process could have applied to the formation of the Ni–Cu ores at Sudbury (see Box 1.7) where contamination of than the original magma pulse, a trait that is attributed to crustal contamination of the new magma increment. Magma mixing, sul?de segregation, and mineralization in the same interval are also, therefore, likely to be related to this event. Recognition of replenishment events should be regarded as a useful criterion for the exploration of PGE mineralization in layered ma?c intrusions. Another intriguing feature of the Merensky Reef linked to injection of new magma is the presence of enigmatic potholes (Figure 2) that are well revealed in underground workings. Potholes are elliptical areas, 100–200 m in diameter, where the Merensky Reef cuts down into its footwall, often by as much as 20–30 m. They are of con- siderable inconvenience to mining operations and also re?ect changes in the mineralogy of the PGE, in that metal alloys tend to dominate over PGE minerals (i.e. sul?des, tellurides, etc.). Their origin is highly debated (see Buntin et al., 1985) and has been attributed to either downward erosional processes (such as convective scouring or thermal erosion) or upward ?uid/melt migration (involv- ing focused vertical migration of ?lter-pressed interstitial magma or a hydrothermal ?uid). A model that accom- modates the Sr isotopic characteristics of the Merensky unit, however, envisages syn-magmatic extension of the footwall cumulus assemblage just prior to injection of the new magma (Carr et al., 1999). The pull-apart rupture grows in a down-dip direction (perhaps due to subsidence and slumping) and is then ?lled by the new magma to form the Merensky unit, with the mineralized reef at its base (Figure 3). Potholes are another feature that illustrate just how complex ore-forming processes actually are when deposits are examined in detail. Y SiO 2 FeO FeS Wustite + liquid X Fayalite + liquid B Tridymite + liquids Immiscibility field- 2 liquids (sul?de and silicate) 1200 °C A Figure 1.23 The ternary system FeO–SiO 2 –FeS at 1200 °C showing how the addition of silica to a homogeneous, sul?de-undersaturated magma will force its composition below the solvus into the ?eld of two liquids, one at Y (silicate-rich) and the other at X (sul?de-rich) (after Naldrett and MacDonald, 1980). ITOC01 09/03/2009 14:38 Page 6566 PART 1 IGNEOUS PROCESSES Ni–Cu mineralization associated with the Sudbury Com- plex in Ontario, Canada represents a fascinating and unique ore occurrence. Although small (only 1100 km 2 ) in com- parison to the Bushveld Complex, the Sudbury layered ma?c intrusion has for a long time been the world’s leading producer of nickel, and contains over 1.5 billion tons of ore at an average grade of some 1.2 wt% Ni, together with substantial Cu and PGE credits (Misra, 2000). It is even more fascinating for the widely held view that it is the product of a large meteorite impact that struck Earth about 1850 million years ago (Krogh et al., 1984). The meteorite impacted into a composite crust containing Archean granite gneisses and the Paleoproterozoic volcano-sedimentary Huronian Supergroup. The main mass of the Sudbury Complex is made up of a differentiated suite of norite, quartz gabbro, and granophyre. The Ni– Cu ores, however, are found in an enigmatic ma?c unit at the base of the succession termed the sublayer (Figure 1). The interior of the structure is underlain by a chaotic sequence of brecciated debris and volcaniclastic material interpreted as the fall-back from the meteorite impact. The granitic and sedimentary ?oor to the structure also contains dramatic evidence for a violent meteorite impact in the form of breccia, pseudotachylite (frictional melt rock formed by very high strain rates), and shatter cones. A widely accepted model suggests that Sudbury magmatism was triggered by the meteorite impact event, or more speci?cally by partial melting of crustal source rocks, during the rebound and pressure release that immediately followed impact (Naldrett, 1989a). Ni–Cu sul?des (mainly chalcopyrite and pyrrhotite– pentlandite) are found along the basal contacts of the sublayer as massive ore which grades upwards into more Magma contamination and sul?de immiscibility: the Sudbury Ni–Cu deposits Whistle Sudbury Frood, Stobie Worthington Copper Cliff Foy Fraser N km 0 10 Granite- Gneiss Levack McCreedy W. Whitewater Group Grean Creighton Huronian Supergroup Sudbury Complex Contact sublayer/ Offset dyke Fault Ministic Figure 1 Simpli?ed geological map of the Sudbury Complex showing the relationship between the main mass of the layered ma?c intrusion and the Ni–Cu mineralized sublayer unit and offset dykes (after Prevec et al., 2000). Figure 2 Typical appearance of the Ni–Cu sul?de- bearing “sublayer” at Sudbury showing its heterogeneous and contaminated nature. ITOC01 09/03/2009 14:38 Page 66IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 67 disseminated mineralization. Signi?cant sul?de mineraliza- tion also occurs in the brecciated ?oor of the structure where sul?de melt percolated downwards into embayments and fractures. The offset dykes also contain steeply plunging pods of sul?de mineralization. An enigmatic feature of the Sudbury ores, however, is their high Ni contents relative to Cu. High Ni magmatic sul?de ores are generally associated with ultrama?c rocks (see, for example, Box 1.5 for a de- scription of the Kambalda deposits), with more different- iated rocks being characterized by higher Cu/Ni ratios. An important, and relevant, feature of the Sudbury ores in the sublayer is their intimate association with exotic ultrama?c inclusions. These inclusions are now known to be the same age as the main Sudbury Complex itself and display no mantle isotopic or geochemical signatures (Prevec et al., 2000). Their presence has important implications for the source of metals in the Sudbury deposits. An intriguing aspect of the relationship between mete- orite impact and Ni–Cu sul?de mineralization at Sudbury is the notion that the catastrophic impact was responsible MgO (wt%) 5 500 Sr (ppm) 10 15 20 25 30 35 40 0 400 300 200 100 North range South range Basalt Mafic-ultramafic inclusions Olivine cumulate Granite- gneiss Norite Granite- gneiss Breccia Massive sul?de mineralization Sublayer Mafic/ultramafic inclusions Basalt slurry (magma, inclusions, sul?de) Sublayer injection along footwall contact Sudbury complex (main mass) Figure 3 (a) Plot of Sr versus MgO showing the compositional ?elds of the mineralized sublayer from the north and south ranges of the Sudbury structure, and that of the ma?c–ultrama?c inclusions within the sublayer. The sublayer is a mixture of melts derived by near total melting of granite–gneiss and tholeiitic basalt of the Huronian Supergroup. The inclusions formed by olivine and pyroxene fractionation of a high-Mg melt at and differentiated toward lower MgO contents (after Prevec et al., 2000). (b) Schematic and simpli?ed cross section through a sublayer- hosted Ni–Cu sul?de orebody in the Sudbury Complex showing some of the features of the Prevec et al. (2000) model for the origin of the sublayer. (a) (b) ITOC01 09/03/2009 14:38 Page 6768 PART 1 IGNEOUS PROCESSES the host magma with siliceous country rock is believed to have facilitated the formation of an immiscible sul?de fraction which scavenged Ni and Cu from the magma to form the ore deposits. 1.6.5 Other magmatic models for mineralization in layered ma?c intrusions The preceding sections described processes such as magma replenishment and sul?de immiscibility that have come to be regarded as providing good “?rst-order” explanations for mineralization in layered ma?c intrusions. In detail, however, it is evident that additional mechanisms are required to explain the many differences that exist from one deposit type to the next, and also the enigmatic patterns of metal distribution evident from very detailed study and more rigorous exploration (Cawthorn, 1999; Barnes and Maier, 2002). Sug- gestions have been made, for example, that even in rocks as well mineralized as the Critical Zone of the Bushveld Complex, sul?de saturation might not have been achieved and the appearance of a sul?de melt was not the most important feature of the mineralization process. Instead, it is argued that, under conditions of low sulfur fugacity (fS 2 ), PGE might crystallize directly from the silicate magma in the form of various platinum group minerals (PGM, such as braggite, laurite, malanite, moncheite, cooperite) or alloys. Although it is dif?cult to conceive how this might happen given the very low abundances of the PGE in the magma, a novel mechanism involving clustering of PGE ions has lent credibility to this as an alter- native process to sul?de segregation (Tredoux et al., 1995; Ballhaus and Sylvester, 2000). PGE clusters Detailed studies of PGE mineralization have shown that these elements are heterogeneously distributed in rocks and minerals, even at a sub-microscopic level. The application of the relatively new ?eld of cluster chemistry has indicated that the heavy transition metals in a magma will tend to coalesce by metal–metal bonding into clusters of 10 to 100 atoms (Schmid, 1994). Tredoux et al. (1995) have suggested that this mechanism might also apply to mineraliza- tion in natural PGE ores such as the Merensky Reef of the Bushveld Complex and the J-M Reef at Stillwater. Theory predicts that heavy transition metals form more stable clusters than light ones and that the PGE will therefore tend to cluster preferentially relative to Cu and Ni. Likewise, the heavy PGE (i.e. higher atomic numbers – Os, Ir, for the extensive and widespread contamination of magma as it intruded into the dust and debris of the impact scar (Naldrett et al., 1986). The rocks of the Sudbury Complex are highly contaminated by crustal material and this is readily apparent in their high silica and potassium con- tents relative to normal continental basalts. One of the features of the Sudbury ores is that this type of contamina- tion might have been responsible for the promotion of sul?de immiscibility and mineralization, as discussed in section 1.6.4 of this chapter. In detail, however, the picture is much more complex and the sublayer is now thought to have an origin that is different to the remainder of the intrusive complex (Prevec et al., 2000; Lightfoot et al., 2001). The main mass of intrusive magma is thought to have been derived by wholesale melting of granite– gneiss and volcano-sedimentary target rocks immediately after meteorite impact and elastic rebound. These crustally contaminated melts crystallized a dominantly plagioclase + pyroxene assemblage to form the bulk of the Sudbury complex, from which the offset dykes were also tapped. The sublayer, however, is considered to have been derived by melting of a much higher proportion of Huronian basaltic target material such that its parental magma was more ma?c than that of the main mass. It crystallized an assemblage dominated initially by olivine, forming ultra- ma?c cumulate rocks at depth in the structure (Prevec et al., 2000). This particular magma was also sulfur saturated and segregated signi?cant volumes of immiscible Ni–Cu sul?de melt, as well as accumulating sul?des from above. It was then emplaced along the brecciated footwall con- tact as a basaltic slurry, comprising magma, ultrama?c inclusions, and sul?de melt, at a relatively late stage in the evolution of the complex. Although convoluted, a meteorite impact model would appear to best explain the complex geology of the Sudbury Complex and to accommodate the intricate processes of anatexis, crustal contamination, and sul?de segregation required to explain its contained ores. ITOC01 09/03/2009 14:38 Page 68and Pt) will form more stable clusters than the light PGE (Ru, Rh, and Pd). Clustering may, there- fore, provide a mechanism for fractionating met- als in a magma chamber. Clustering of atoms and molecules in magmas is a complex process and is controlled by variables such as temperature and composition. Figure 1.24 schematically illustrates the nature of PGE clus- tering in magma. Initial coalescence of metals occurs by metal–metal bonding, but the clusters are then stabilized by an envelope of ligands (see Chapter 3) such as sulfur or aluminosilicate species. Although there is no direct experimental evidence for PGE clusters in magmas (Mathez, 1999), circumstantial evidence comes from fea- tures such as light/heavy PGE fractionation and microbeam detection of minute PGE metal alloys in natural materials. It has also been noted that cumulus olivine- and chromite-rich rocks, such as the harzburgites and chilled margins of the Bushveld Complex, which show no association with sul?de segregations, are notably enriched in PGE relative to less ma?c rocks (Davies and Tredoux, 1985; Ballhaus and Sylvester, 2000). Such an enrichment could be attributed to inclu- sion of PGE clusters into early formed cumulus phases such as olivine and chromite. Stability of PGE clusters and their precipitation from the magma as discrete platinum group minerals or metal alloys is likely to be promoted by a low fS 2 environment. The existence of PGE clusters in ma?c magmas has important implications for understanding the mineralization process. If PGE clusters do form then crystal-chemical considerations and parti- tioning behavior might be less important than mechanical concentration mechanisms involving trapping of clusters as micro-inclusions in any suitable cumulus phase, be it silicate, oxide, or sul?de. Figure 1.24 illustrates two options, one where PGE clusters are incorporated into sul?de globules and the other where they are trapped as micro-inclusions in either olivine or chromite (Tredoux et al., 1995; Ballhaus and Sylvester, 2000). Detailed analyses of PGE in minerals of the Merensky Reef indicate that cluster theory might well explain some of the metal distribution pat- terns not previously explained by conventional sul?de partitioning behaviour. Micro-inclusions of Os–Ir–Pt alloy are found in pyrrhotite and pent- landite, as are other discrete aggregates of Pt with lesser Pd and Au. In olivine and chromite, micro- inclusions are dominantly the Pt–Pd–Au type (Ballhaus and Sylvester, 2000). The latter category is particularly signi?cant as it suggests that the Bushveld magma contained abundant Pt– (Pd–Au) clusters early on in its crystallization history and that concentration of these metals (by trapping them as inclusions in early cumulus phases) can be achieved without sul?de saturation having occurred at all. Admittedly, the concentrations of PGE in silicate phases have not as yet proved to be of economic interest, although the process should remain of interest to platinum explora- tionists in the future. Chromitite layers can, by contrast, contain very signi?cant concentrations of PGE, as in the case of the huge reserves asso- ciated with the UG2 chromitite seam in the Bushveld Complex. This association is discussed in more detail below. The role of chromite in PGE concentration The very signi?cant concentration of PGE in chromitite layers such as the UG2 in the Bushveld Complex suggests that mineralization processes other than sul?de scavenging must have played a role. The UG2 also contains a differ- ent assemblage of the PGE than the Merensky Reef, for example, and is relatively enriched in Ru, Os, and Ir, but with similar Pt contents (von Grünewaldt et al., 1986). At a ?rst glance this would seem to suggest that heavy PGE clusters had been preferentially concentrated as metal- alloy inclusions in cumulus chromite grains (sim- ilar to the process shown in Figure 2.24), a model ?rst suggested by Hiemstra (1979). More detailed work, however, reveals the following character- istics of PGE mineralization in chromitite layers: 1 The chromitite layers do, in fact, contain minor sul?de minerals, both interstitially to the chromite grains and as inclusions (of which laur- ite RuS 2 is predominant; Merkle, 1992) within them. 2 Pyrrhotite is virtually absent from the chromit- ites probably because of reaction between sul?des IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 69 ITOC01 09/03/2009 14:38 Page 6970 PART 1 IGNEOUS PROCESSES and chromite which partitions additional Fe into chromite and liberates sulfur (as shown in equa- tion [1.9], after Naldrett and Lehmann, 1988): 4Fe 2 O 3 (chr) + FeS ? 3Fe 3 O 4 (chr) + 0.5S 2 [1.9] 3 The assemblage Pt–(Pd–Au) occasionally occurs as discrete metal-alloy inclusions in chromite (Ballhaus and Sylvester, 2000), indicating that the other PGE are associated with sul?de minerals, either as inclusions within, or interstitially to, chromite. These characteristics led Barnes and Maier (2002) to suggest that an association between PGE con- centration and chromitite layers could be related to both sul?de accumulation and metal cluster- ing. In their model sul?de segregation under high R factor conditions leads to the accumulation of PGE-enriched sul?de cumulates together with the normal silicate minerals and chromite. The sul?de phase reacts with chromite consuming Fe and liberating S, resulting in a Cu–Ni rich residual sul?de assemblage that crystallizes minor chal- copyrite and pentlandite. The localized lowering of fS 2 that results from sul?de–chromite inter- action promotes the direct crystallization of PGM or metal alloys from clusters in the magma so that PGE concentration in the immiscible sul?de Coalescence (metal–metal bonding) (PGE cluster 10–100 atoms) Stabilization (ligand envelope) (S-poor pathway) PGE cluster captured by sul?de melt Key: PGE Fe S PGE inclusions in olivine and chromite (S-rich pathway) Figure 1.24 Schematic diagram illustrating the formation of PGE clusters in a magmatic system and their eventual inclusion either in an immiscible sul?de fraction (S-rich pathway) or in a silicate or oxide cumulus phase such as olivine or chromite (S-poor pathway) (after Tredoux et al., 1995; Ballhaus and Sylvester, 2000). ITOC01 09/03/2009 14:38 Page 70fraction is dictated by clustering principles rather than strict partitioning behavior. The unusual PGE content of chromite-rich cumulate assem- blages arises from the fact that Ru and, to a lesser extent, Rh partition strongly into the oxide phase whereas other PGE (Pd and Pt) are speci?cally excluded from the chromite lattice and remain in solution. Since Ru and Rh are fractionated into the chromite, the remainder of the PGE cluster (i.e. Os, Ir, and Pt) destabilizes and is likely to pre- cipitate as PGM or metal alloys. The latter are scavenged by any interstitial sul?des present in and around the chromitite layer. Application of composite models, as illustrated above, seems to provide an adequate explanation for several of the intriguing characteristics of PGE mineralization in both sul?de-rich and sul?de- poor environments. Detailed observations, in par- ticular intricacies such as the location of PGE in different mineralogical sites, clearly provide the sort of constraints that are needed to improve our understanding of the relevant ore-forming processes. Hiatus models One of the tendencies of traditional models for the formation of magmatic deposits is to consider the processes in terms of accumulation of cumul- us phases, be they silicate or oxide minerals, or sul?de melt. An intriguing thought by Cawthorn (1999) has suggested that the opposite, namely the lack of accumulation of cumulus minerals, might be the essential ingredient in the formation of PGE and base metal sul?de mineralization styles in certain types of deposits. The Merensky Reef, for example, represents a layer in the Bushveld magma chamber where a major injection of new magma occurred and where there is a pronounced mineralogical hiatus in the magmatic “stratigra- phy” (see Box 1.6). Magma replenishment could create a zone of mixing where the liquidus tem- perature is substantially above that relevant to local crystallization, resulting in a transient stage when no silicate minerals formed. This would also result in a hiatus in the solidi?cation process, equating to a period, at least in terms of simpli?ed gravitative settling, with little or no accumula- tion of minerals on the chamber ?oor. Sul?de ac- cumulation and/or metal clustering could, never- theless, have continued during this interval so that enrichment and mineralization was assisted by the absence of phases that might normally contri- bute to dilution of the PGE. The view of Ballhaus and Sylvester (2000) with respect to Merensky Reef formation is somewhat analogous. They suggest that PGE concentration occurred by metal clus- tering in the magma to form an in situ stratiform PGE anomaly. The anomaly was effectively “frozen in” at the hiatus created by magma replenish- ment and local sul?de saturation, as the sul?de globules scavenged PGE clusters from the magma and then settled into the underlying cumulate mush. Fluid-related in?ltration of PGE There is abundant mineralogical evidence to suggest that many layered ma?c intrusions, as well as their contained mineralization, have been effected by alteration related to hydrothermal ?uids derived either from the evolving magma itself or from an external source. There is little doubt that such ?uids have played a role in the remobilization of existing magmatic sul?de min- eralization. A more pertinent question, however, relates to whether such hydrothermal processes may not in fact have been dominant in ore forma- tion (Schiffries and Rye, 1990; Boudreau and Meurer, 1999). Discussion of this topic falls out- side the con?nes of this chapter but it is revisited in Chapter 2. 1.7 A MODEL FOR MINERALIZATION IN LAYERED MAFIC INTRUSIONS The Bushveld Complex in South Africa, together with the Great Dyke of Zimbabwe and the Stillwater Complex of Montana, represent the prime examples of chromite and PGE-sul?de ore formation in layered ma?c intrusions. These ex- amples have many features in common and the detailed work of researchers such as Irvine (1977), Naldrett (1989a, b), Naldrett and von Grünewaldt (1986), Naldrett and Wilson (1991), and Campbell IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 71 ITOC01 09/03/2009 14:38 Page 7172 PART 1 IGNEOUS PROCESSES et al. (1983), as well as many others, facilitates the compilation of a relatively simple model to explain the occurrence and formation of these very important deposits. The nature and origin of im- portant Cu–Ni sul?de deposits such as Sudbury (Ontario), Noril’sk-Talnakh (Siberia), and Duluth (Minnesota) are somewhat different, but their formation can nevertheless be explained by some of the processes described below. The model described in this section represents a useful sum- mary, but for ease of explanation is simpli?ed, both conceptually and with respect to the actual nature of the ore deposits that it represents. Readers should be cautious in applying this model too rigorously to the real situation. Additional detail can be obtained from the original reference articles, in particular Naldrett (1989b, 1997). Previous sections of this chapter have emphas- ized the following features as being important with respect to the formation of ores in igneous intrusions: 1 Crystal fractionation and gravity-induced crys- tal settling. 2 Density strati?cation of magma chambers and the ability of magmas to undergo density changes as crystallization proceeds. 3 Repeated recharge of chambers by injection of fresh magma. 4 The ability of fresh magma to ?nd its own dens- ity level and to turbulently mix with the residual magma. 5 The existence of transient periods of crystalliza- tion when only a single phase such as chromite is on the liquidus. 6 The formation of immiscible globules of sul?de liquid once the magma becomes saturated or supersaturated with respect to sul?de. 7 The very high partition coef?cients of siderophilic elements for the immiscible sul?de fraction. These features are all used to construct the model in Figure 1.25, which shows a schematized layered ma?c intrusion containing elements of mineralization representing well known deposits such as the PGE-sul?de bearing Merensky, J-M (Stillwater), and LSZ (Lower Sul?de Zone, Great Dyke) reefs, and the chromitiferous UG and LG (upper and lower groups respectively) seams of the Bushveld Complex. The UG2 chromitite is particularly important because it contains vast reserves of Cr as well as PGE occurring together in the same, very extensive, horizon. One empirical observation which has not been previously mentioned, but which turns out to be of considerable importance to understanding both the position and tenor of these ores, is the “strati- graphic” level in the layered intrusion where plagioclase ?rst appears as a cumulus phase. This seemingly innocuous point in the evolution of an igneous intrusion (clearly marked in Figure 1.25) divides ore horizons which tend to be less PGE- enriched below it from those above it, which tend to be both more enriched and more substantial (at least in the Bushveld and Stillwater). In Fig- ure 1.16a it is clear that an evolving magma will become more dense than when it started once signi?cant plagioclase is extracted. This has im- portant implications for the behavior of newly injected melt into the magma chamber and affects the formation of fountains and plumes (Figure 1.16b) as well as the magnitude of the R factor, which in turn controls metal concentration into the immiscible sul?de fraction (Figure 1.22). Thus, with reference to Figure 1.25, any new magma which is introduced into the chamber early in its crystallization sequence (i.e. prior to the appear- ance of plagioclase on the liquidus) will tend to behave as a fountain. The limited degree of mixing that does occur (point AC in the “Irvine model,” Figure 1.25) will force the magma composition into the chromite ?eld and give rise to a chromitite layer associated with ultrama?c cumulates such as those, for example, of the Great Dyke. The magma composition at this stage will be undersaturated with respect to sul?de (point AC in the “Naldrett and von Grünewaldt model,” Figure 1.25) and no sul?de ores will, therefore, form. As crystallization continues, however (A to B in the “Naldrett and von Grünewaldt model,” Figure 1.25), saturation will be achieved and immiscible sul?des will segregate to form disseminations of sul?de ore in an orthopyroxene cumulate rock. The very high partition coef?cients of Cu, Ni, and the PGE into the immiscible fraction will mean that even with limited buoyancy and mixing the earliest sul?des will be enriched in these elements. As further sul?des exsolve, however, they will rapidly become depleted in low abundance metals because the residual magma in equilibrium with ITOC01 09/03/2009 14:38 Page 72these sul?des, which itself originally only had very low abundances of the PGE, becomes depleted (refer to Figure 1.22 for a quanti?cation of this effect). This feature characterizes the Pt–Pd ores of the Lower Sul?de Zone of the Great Dyke, for example, where despite the existence of Cu–F sul?des over a 2–3 meter interval, the ores con- tain viable grades of PGE over only a few centime- ters at the base of the mineralized zone. By contrast, if a new injection of magma takes place after the ?rst appearance of plagioclase then it will likely have a lower density than the residual magma and emplace itself in a turbulent, buoyant fashion at a relatively high level in the chamber. The more ef?cient mixing that occurs will again promote chromite precipitation (point AD in the “Irvine model,” Figure 1.25). These chromite seams will be associated with either gabbroic or ultrama?c cumulate rocks, depending on the relative proportions of liquids A and D that are mixed together, and it is also possible that they will be better developed. Mixing of primitive magma A with an evolved liquid at D will also, in this case, result in a mixture that is sul?de over- saturated (point AD in the “Naldrett and von Grünewaldt model,” Figure 1.25), resulting in the formation of a sul?de immiscible fraction. The plume effect that is created by the injection of the lower density primitive magma will result in buoyancy and a high degree of convective circula- tion of these sul?des, which translates into a high R factor. PGE are again strongly partitioned into the sul?de fraction but the grades will be higher and more evenly distributed because the sul?des have the opportunity of equilibrating with a much larger mass of the magma. This situation is applicable to the formation of the Merensky and J-M reefs, both of which are formed well above the ?rst appearance of plagioclase cumulate rocks. It also applies to the formation of the UG2 seam where PGE-sul?des are intimately associated with a major chromitite layer. IGNEOUS ORE-FORMING PROCESSES CHAPTER 1 73 ‘Plume’ High R factor Gabbroic cumulates Ultramafic cumulates (bronzitites, harzburgites) ‘Fountain’ Low R factor AD AC A Chromite Olivine C D Quartz D AD C AC A B Merensky and J-M UG-2 LSZ 02 0 7 0 0 0.4 Sul?de (wt%) Crystallization (%) LG “IRVINE MODEL” “Naldrett and von Grünewaldt model” Figure 1.25 Generalized model showing the nature of igneous processes that give rise to some of the important styles of chromite and PGE–base metal sul?de deposits associated with layered ma?c intrusions. LG and UG refer to the Lower and Upper Group (UG2 speci?cally) chromite seams of the Bushveld Complex; LSZ is the Lower Sul?de Zone of PGE mineralization in the Great Dyke; Merensky refers to the Merensky Reef of the Bushveld Complex and J-M is the J-M reef of the Stillwater Complex, both of which contain PGE–sul?de mineralization. The difference between ultrama?c and gabbroic cumulates in this model is marked by the ?rst appearance of cumulus plagioclase in the latter (modi?ed after Naldrett, 1997). ITOC01 09/03/2009 14:38 Page 73There are many different types of ore deposits associated with igneous rocks; there is also a wide range of igneous rock compositions with which ore deposits are linked. Magmas tend to inherit their metal endowment from the source area from which they are partially melted. Fertile source areas, such as metasomatized mantle or sediment- ary rock, are usually themselves a product of some sort of metal concentration process. Felsic magmas crystallize to form granites, or their extrusive equivalents, and are associated with concentrations of elements such as Sn, W, U, Th, Li, Be, and Cs, as well as Cu, Mo, Pb, Zn, and Au. Incompatible elements in felsic magmas are con- centrated into the products of very small degrees of partial melting or into the residual magma at an advanced stage of crystallization. Such processes do not, however, very often result in econom- ically viable ore deposits. Crystal fractionation in ma?c magmas, on the other hand, results in important concentrations of elements such as Cr, Ti, Fe, and V, while associated sul?de immiscib- ility in these rocks results in accumulations of PGE, Cu, Ni, and Au. Layered ma?c intrusions are very important exploration targets for this suite of metals worldwide. Primary diamond deposits represent the very unusual situation where deep-seated ma?c magma vents to the sur- face as explosive diatreme-maar type volcanoes, bringing with them older, xenocrystic diamond from fertilized mantle. In both ma?c and felsic magmas the latter stages of crystallization are accompanied by the exsolution of a dominantly aqueous and carbonic ?uid phase that ultimately plays a very important role in ore formation. It is this process that is the subject of Chapter 2. For those readers wishing to delve further into mag- matic ore-forming processes, the following references to books and journal special issues will help: Economic Geology, volume 80 (1985) Special Issue on the Bushveld Complex. Kirkham, R.V. et al. (eds) (1997) Mineral Deposit Model- ing. Geological Association of Canada, Special Paper 40. Misra, K.C. (2000) Understanding Mineral Deposits. Dordrecht: Kluwer Academic Publishers, 845 pp. Naldrett, A.J. (1989) Magmatic Sul?de Deposits. Oxford Monographs on Geology and Geophysics. Oxford: Clarendon Press, 186 pp. Naldrett, A.J. (1999) World-class Ni–Cu–PGE deposits: key factors in their genesis. Mineralium Deposita, 34, 227–40. Taylor, R.P. and Strong, D.F. (eds) (1988) Recent advances in the geology of granite-related mineral deposits. Canadian Institute of Mining and Metallurgy, special volume 39, 445 pp. Whitney, J.A. and Naldrett, A.J. (1989) Ore deposition associated with magmas. Reviews in Economic Geology, volume 4. El Paso, TX: Society of Economic Geologists, 250 pp. 74 PART 1 IGNEOUS PROCESSES ITOC01 09/03/2009 14:38 Page 742.1 INTRODUCTION In Chapter 1 emphasis was placed on the concen- tration of metals during the igneous processes associated speci?cally with magma formation and its subsequent cooling and crystallization. Little mention was made in the previous chapter of aqueous solutions (or hydrothermal ?uids), even though it is well known that such ?uids have often played a role in either the formation or modi?cation of various magmatic deposits. The interaction between igneous and hydrothermal processes is an extremely important one for the formation of a wide variety of ore deposit types, especially in near surface environments where magmas and ?uids are spatially and genetically linked. This chapter introduces the concept of “magmatic-hydrothermal” ?uids and concentrates on those ?uids speci?cally derived from within the body of magma itself. Particular emphasis is given to the ?uids derived from granitic magmas crystallizing in the Earth’s crust. There is little doubt that the majority of ore deposits around the world either are a direct prod- uct of concentration processes arising from the circulation of hot, aqueous solutions through the Earth’s crust, or have been signi?cantly modi?ed by such ?uids. A wide variety of ore-forming pro- cesses are associated with hydrothermal ?uids and these can be applied to both igneous and sedimentary environments, and at pressures and temperatures that range from those applicable at shallow crustal levels to those deep in the Magmatic-hydrothermal ore-forming processes Box 2.1 Magmatic-hydrothermal ?uids associated with granite intrusions. 1 La Escondida porphyry Cu deposit, Chile Box 2.2 Fluid ?ow in and around granite plutons – the Cornish granites, SW England Box 2.3 Magmatic-hydrothermal ?uids associated with granite intrusions. 2 The MacTung W skarn deposit, Yukon, Canada Box 2.4 Magmatic-hydrothermal ?uids in volcanic environments – the high- and low-sul?dation epithermal gold deposits of Kyushu, Japan SOME PHYSICAL AND CHEMICAL PROPERTIES OF WATER MAGMATIC-HYDROTHERMAL FLUIDS water solubility in magmas ?rst boiling, second boiling granite-related magmatic-hydrothermal ore deposits COMPOSITION AND CHARACTERISTICS OF MAGMATIC- HYDROTHERMAL FLUIDS PEGMATITES AND THEIR SIGNIFICANCE METAL TRANSPORT IN MAGMATIC-HYDROTHERMAL FLUIDS ?uid-melt partitioning of trace elements WATER CONTENT AND DEPTH OF EMPLACEMENT OF GRANITIC MAGMAS FLUID FLOW IN AND AROUND GRANITE INTRUSIONS ORIGIN OF PORPHYRY CU, MO, AND W DEPOSITS POLYMETALLIC SKARN DEPOSITS EPITHERMAL AU–AG–(CU) DEPOSITS ROLE OF FLUIDS IN MINERALIZED MAFIC ROCKS ITOC02 09/03/2009 14:37 Page 75lithosphere. Many different types of ?uids are involved in hydrothermal ore-forming processes. The most primitive or “juvenile” of these are the magmatic-hydrothermal ?uids that originate from magmas as they cool and crystallize at various levels in the Earth’s crust, and a number of important ore deposit types are related to the concentration of metals that arise from circula- tion of such solutions. This chapter has particular relevance to ore deposits such as the large family of porphyry Cu and Mo deposits, as well as the “high-sul?dation” epithermal Au–Ag deposits that often represent the surface or volcanic mani- festations of porphyry deposits. In addition, the formation of greisen-related Sn–W ores, poly- metallic skarn mineralization, and pegmatite- related deposits are also explained in this chapter. Discussion of processes involved in the formation of hydrothermal ores that are not directly linked to magmatic activity follows in Part 2 of the book (Chapter 3). 2.2 SOME PHYSICAL AND CHEMICAL PROPERTIES OF WATER Water (H 2 O) is a liquid at room or standard tem- perature and pressure (STP). By contrast, other light molecules (CH 4 , NH 3 , H 2 S, etc.) at STP exist as vapor and form gaseous hydrides. H 2 O boils at a much higher temperature than the hydrides formed by other light elements. The main reason for this is that the hydrogen of a water molecule has a very strong af?nity for oxygen such that temporary hydrogen bonds are formed between it and the oxygen atoms of the neighboring water molecule (Figure 2.1). Water is, therefore, a liquid polymer at STP, and this is responsible for its many anomalous properties. The physical properties of water that make it so important for chemical, biological, and geological processes include: 1 A high heat capacity, which means that it can conduct heat more readily than other liquids. 2 High surface tension implying that it can easily “wet” mineral surfaces. 3 A density maximum at temperatures just above the freezing point, which means that solid-H 2 O (ice) will ?oat on liquid-H 2 O. 4 A high dielectric constant and, hence, an ability to dissolve more ionic substances, and in greater quantities, than any other natural liquid. The last property is particularly relevant to the formation of ore deposits, since water is largely responsible for the dissolution, transport, and, hence, concentration of a wide range of elements and compounds, including metals, in the Earth’s crust. The chemical properties of water are dominated by its high dipole moment, a feature related to the fact that the water molecule is not symmetrical but comprises two hydrogen atoms that are not directly opposite one another but offset relative to the diametric axis of the single, much larger, oxygen atom (Figure 2.1). The H 2 O molecule is, therefore, an electric dipole in which the center of positive charge does not coincide with the center of negative charge. Consequently, water 76 PART 1 IGNEOUS PROCESSES Lone pairs of electrons Oxygen atom Hydrogen atoms (a) (b) Hydrogen bond Lone pair of electrons O HH O HH –ve –ve +ve +ve +ve –ve Figure 2.1 Idealized illustration of the water molecule and the nature of hydrogen bonding. ITOC02 09/03/2009 14:37 Page 76MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 77 molecules can align themselves in an electrical ?eld or orientate themselves around other charged ionic species. The clustering of water molecules around a dissolved ion is referred to as hydration, and this promotes the stability of ions in aqueous solution. The dielectric constant is a measure of a ?uid’s ability to separate ions or other dipoles and is a function of the number of dipoles per unit volume. The high dielectric constant of water also has the effect of enhancing dissolution and increasing solubility. In nature water exists as three different phases, solid (or ice), vapor (or steam), and liquid (Figure 2.2a). H 2 O-ice typically exists below about 0 °C and forms when water molecules are arranged in a hexagonal crystalline structure similar to many rock forming silicate minerals. When H 2 O-ice melts only a small proportion of the hydrogen bonds are broken and H 2 O-liquid retains a large measure of tetrahedral coordina- tion. At temperatures close to the freezing point, this structure is more tightly knit than ice, which explains why the liquid form of water is more dense than its solid, and ice ?oats. The density of H 2 O-liquid varies mainly as a function of tem- perature, but also pressure, as illustrated in Fig- ure 2.2b. Density de?nes the difference between the liquid phase (typically around 1 g cm -3 at room temperatures) and its coexisting vapor (where densities are one to two orders of magnitude lower). Lines of equal density (or volume) in P–T space are referred to as isochores (Figure 2.2c). The phase boundary along which liquid and vapor are in equilibrium (i.e. between the triple point and the critical point in Figure 2.2a) de?nes the equilibrium or saturation vapor pressure, also known simply as the boiling point curve. On the Earth’s surface, water effectively “boils” when equilibrium vapor pressure exceeds the prevailing atmospheric pressure and vapor can bubble off from the liquid. As temperature rises and the density of H 2 O- liquid decreases, the latter must coexist in equi- librium with a vapor whose partial vapor pressure (and density) is increasing. The boiling point tem- perature of pure water increases progressively with pressure until a maximum limit is reached, called the critical point, at 374 °C and 221 bar (Figure 2.2a). The critical point is where it is no longer possible to increase the boiling point by increase of pressure and is effectively de?ned as the stage where there is no longer a physical distinction (i.e. a contrast in density) between liquid and vapor. The densities of liquid and vapor at the critical point have merged to a value of around 0.3 g cm -3 (Figure 2.2c). Since the terms “liquid” and “vapor” no longer have any meaning at the critical point, the term “supercritical ?uid” (see section 2.3.3 below) is used to describe the homogeneous single phase that exists at pressures and temperatures above the critical point. For an ideal gas which obeys the Gas Law, the relation between pressure (P), temperature (T), and volume (V) or density is expressed by the well known equation: PV = RT [2.1] where R is the gas constant. Water only behaves as an ideal gas at very high temperatures and low pressures and equation [2.1] cannot, therefore, be used to predict aqueous phase relations under natural conditions in the Earth’s crust. Consequently, many attempts have been made to introduce correction factors and modi?cations to the ideal gas equation such that it will more accurately re?ect non-ideal behavior. The modi?ed forms of the equation, of which there are many that have been derived by both empirical and theoretical means, are referred to as equations of state. The simplest, general form of an equation of state, which incorporates two corrective terms, is an expression known as van der Waals’s equation: P = [RT/(V - b)] - a(V) [2.2] where a is the corrective term that accounts for the attractive potential between molecules, and b is a term that accounts for the volume occupied by the molecules. A widely utilized equation of state for water at higher pressures and temperatures is the modi?ed Redlich–Kwong equation, a more detailed dis- cussion of which, together with its applicability to the chemical and thermodynamic properties ITOC02 09/03/2009 14:37 Page 7778 PART 1 IGNEOUS PROCESSES Pressure (bars) 300 200 100 600 500 400 100 150 200 250 300 350 400 Temperature (°C) 0.01 0.03 0.05 0.08 0.15 Isochores 0.20 0.30 0.40 0.50 0.60 0.70 0.75 0.80 0.85 0.90 0.95 1.00 cp (c) Pressure (bars) 221 0.06 0.008 100 374 Temperature (°C) Liquid water (a) Critical point Ice (solid) (liquid) Water vapor (gas) Triple point Vapor Liquid H 2 O Supercritical fluid 2000 Depth (meters) 500 1000 1500 100 200 300 400 500 600 (b) 900 600 300 Pressure (atmospheres) Temperature (°C) cp Figure 2.2 (a) Pressure– temperature phase diagram (not to scale) for pure H 2 O showing the occurrences of the three phases (solid/ice, liquid water, water vapor/gas). The Triple Point, where solid, liquid, and vapor coexist, is at 0.008 °C and 0.06 bar. The Critical Point, beyond which there is no longer a physical distinction between liquid and vapor, is at 374 °C and 221 bar.The critical density is 0.322 g cm -3 . (b) Schematic visual representation of H 2 O ?uid densities in relation to the boiling point curve and critical point, cp (after Helgeson, 1964). (c) Part of the H 2 O phase diagram showing actual variations of liquid and vapor equilibrium densities in pressure–temperature space (isochoral densities in g cm -3 ). ITOC02 09/03/2009 14:37 Page 78MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 79 of magmatic ?uids, can be found in Holloway (1987). 2.3 FORMATION OF A MAGMATIC AQUEOUS PHASE 2.3.1 Magmatic water – where does it come from? In the very early stages of the Earth’s development there was probably very little water either in the atmosphere or on the surface of the planet. It seems likely that the incipient oceans developed as soon as the early crust had stabilized and rain water ponded on its surface. Substantial bodies of water (oceans or seas) existed by at least 3800 Myr ago as evident from the preservation of early Archean sediments and subaqueously deposited volcanic rocks. Some of this early water was derived from the degassing of volcanic magmas as they extruded onto the early crust, and is referred to as juvenile water. As plate tectonic processes progressively dominated Earth processes, water has been subjected to extensive recycling and it is likely that much of the ?uid introduced by magmatic activity at the surface in more recent geological times is no longer juvenile, although it is still referred to as magmatic. Subduction of oceanic crust that has been highly altered and hydrated by percolating sea water is a scenario that can be used to explain the H 2 O contents of most arc-related andesitic and basaltic magmas. Arc-related magmas probably, therefore, contain dissolved water derived from mixing of primitive, mantle-derived ?uids and sea water. A minor component from meteoric ?uid (i.e. derived from the hydrosphere) is also possi- ble. It is interesting to note that hydrous basaltic magmas that pond at the base of the crust, and whose heat is responsible for initiating anatexis of the lower crust to form granitic melts, may contribute their water to these melts. Whitney (1989) has suggested that underplating of a felsic magma chamber with a more dense ma?c magma will result in diffusive transfer of elements and volatile species across the boundary layer. Most of the water present in granitic magmas is, however, derived from the dehydration of min- erals in the crust that were themselves melted to form the magma. The concept of “?uid-absent” or dehydration melting (i.e. where melting occurs without the presence of free water in the rock) is regarded as a realistic model for the generation of granites in the Earth’s crust. The process can best be explained by considering the phase equilibria of essentially three minerals, namely muscovite, biotite, and hornblende. Although the detailed reactions are complex (see Clemens and Vielzeuf, 1987; Whitney, 1989; Burnham, 1997), the approximate conditions for dehydration melt- ing of muscovite, biotite, and hornblende in relation to an average geothermal gradient are shown in Figure 2.3. The amount of H 2 O con- tained within these three minerals decreases from around 8–10% in muscovite, to 3–5% in biotite and 2–3% in hornblende. Accordingly, the water activity in melts formed at breakdown of these hydrous minerals will vary considerably and a magma derived from anatexis of a muscovite- bearing precursor is likely to contain more dis- solved water than one derived by melting of an amphibolite. Source material comprising domin- antly muscovite and progressively buried along the 25 °C km -1 geotherm in Figure 2.3 will start to melt at point A. The water content of that ?rst- formed magma will be 7.4 wt% (Burnham, 1997). If the source material contained mainly biotite as the hydrous phase, melting would begin at a signi?cantly higher temperature and pressure, at point B, and in this case the ?rst-formed melt would contain only 3.3 wt% H 2 O. Likewise, dehydration melting of an amphibolitic source rock would only produce a melt at even higher P–T (point C, Figure 2.3), and this would contain 2.7 wt% H 2 O. It is apparent, therefore, that gran- ites of variable composition derived from differ- ent levels in the crust will initially contain very different H 2 O contents. It is clear that substantial volumes of magmatic water will be added to the Earth’s crust as granite magmas progressively build the continents. It is relevant to point out that melting of a muscovite- or a muscovite +biotite-bearing source rock (represented in nature by rocks such as metasedi- ments) is likely to yield peraluminous, S-type granite compositions (see section 1.3.4). The types of granites with which most Sn–W–U ore ITOC02 09/03/2009 14:37 Page 79assemblages are typically associated will, there- fore, be relatively “wet,” and contain high initial H 2 O contents. S-type granites generally correlate with the relatively reduced “ilmenite-series” granites of Ishihara (1977), which inherit their low fO 2 character by melting of graphite-bearing metasedimentary material. By contrast, melting of biotite- or biotite + hornblende-bearing source material (represented by meta-igneous rocks) will yield metaluminous, I-type granite compositions. The porphyry Cu–Mo suite of ore deposits are associated globally with the latter and are typic- ally relatively “dry” compared to S-type granites. Ishihara’s “magnetite-series” granites are often, but not always, correlatable with I-type granites and these are characterized by higher fO 2 magmas (Ishihara, 1981). 2.3.2 H 2 O solubility in silicate magmas It is evident from the above discussion that magmas derived in different tectonic settings are likely to have had variable initial H 2 O contents, and that this is partly a function of the amount of water supplied by the source material during melting. There is, however, a maximum limit to the amount of H 2 O that a given magma can dis- solve, and this is de?ned by the solubility of water in any given magma. The solubility of H 2 O in silicate magmas is determined mainly by pressure and to a lesser extent temperature. Figure 2.4a shows the results of experimental determinations of water content in melts of basaltic, andesitic, and granitic, or pegmatitic, compositions. The experiments suggest that water content is strongly dependent on pressure, with magmas at the base of the crust (circa 10 kbars) able to dissolve between 10 and 15% H 2 O. It would also appear from Figure 2.4a that for any given pressure, felsic melts are able to dissolve more water than ma?c ones. When water dissolves in a magma it exists essentially as hydroxyl (OH) groups, although it is likely that at higher pressure and water con- tents discrete molecular water (H 2 O) also exists (Stolper, 1982). The solubility of water in silicate 80 PART 1 IGNEOUS PROCESSES 0 Pressure (kb) 12 8 4 16 20 10 20 30 40 50 60 70 600 800 1000 Temperature (°C) Depth (km) 25° km –1 Granodiorite (solidus) Basalt (solidus) Muscovite (7.4% H 2 O) Biotite (3.3% H 2 O) Amphibolite (2.7% H 2 O) C B A Figure 2.3 Pressure–temperature plot showing the approximate conditions under which dehydration melting of muscovite-, biotite-, and hornblende-bearing assemblages would take place in relation to a 25 °C km -1 geothermal gradient. Melts formed at A, B, and C respectively are likely to contain different initial water contents. The solidus curves for granite and basalt are for water-saturated conditions (redrawn after Burnham, 1997). ITOC02 09/03/2009 14:37 Page 80MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 81 magmas is thought to be governed by the follow- ing equilibrium reaction: H 2 O (molecular) + O o - 2OH [2.3] where O o refers to the oxygen that bridges or polymerizes the silicate structure of the magma. Low viscosity basaltic magmas comprise a smaller proportion of bridging O o than more highly polymerized granitic melts. Basaltic melts may, therefore, accommodate fewer OH groups in O o substituted sites, which could explain their inability to dissolve as much water as granite. At high pressures, however, water solubilities are essentially independent of magma composition. As a generalization, the mole fraction (squared) of water dissolved in any magma is proportional to the water fugacity (f H 2 O or the effective concentra- tion in a non-ideal solution) of the magma and this is strongly pressure dependent. An indication of the amount of water dissolved in a hydrous melt as a function of pressure and water fugacity is provided in Figure 2.4b. At low pressures, where total solubilities are low, OH groups are the domi- nant species of dissolved water in silicate mag- mas, whereas at higher pressures water solubility is dominated by molecular H 2 O (Stolper, 1982). 2.3.3 The Burnham model The importance of processes whereby zones of H 2 O-saturated magma are formed and localized toward the roof of a granite intrusion, and their signi?cance with respect to granitoid related ore deposits, was emphasized by the pioneering work of C. W. Burnham (1967, 1979, 1997). The concept has stimulated much fruitful research and continues to receive experimental and theor- etical re?nement through the work of Whitney (1975, 1989), Candela (1989a, b, 1991, 1992, 1997), Shinohara (1994), and many others. When a granitic magma crystallizes the liquidus assemblage is dominated by anhydrous minerals and the concentration of dissolved incompatible constituents, including H 2 O and other volatile species, increases by processes akin to Rayleigh H 2 O (wt%) 10 5 15 20 1 P (kb) 9 357 10 5 0 1 52 02 53 03 5 Basalt (1100°C) Andesite (1100°C) Depth (km) (a) 11 40 20 Granite pegmatite (820–660°C) f H 2 O (kb) 2 0.1 (X H 2 O ) 2 (b) 0 4 6 8 10 12 1100°C 10 kb 0 0.2 0.3 0.4 0.5 8 kb 6 kb 4 kb 2 kb H 2 O saturation H 2 O dominant OH dominant Figure 2.4 (a) Experimentally determined solubilities (in wt%) of H 2 O in silicate melts as function of pressure. 1, basalt (at 1100 °C); 2, andesite (at 1100 °C ); 3, granitic pegmatite (at 660–820 °C) (after Burnham, 1979). (b) Calculated estimates of the mole fraction of water (X H 2 O ) (squared) in hydrous magmas at 1100 °C as a function of the water fugacity (f H 2 O ) and pressure (after Stolper, 1982). H 2 O-saturation is de?ned by the encompassing curve. ITOC02 09/03/2009 14:37 Page 81fractionation (see Chapter 1). At some stage, either early or late in the crystallization sequence, granitic magma will become water-saturated, resulting in the exsolution of an aqueous ?uid to form a chemically distinct phase in the silicate melt. This process is called H 2 O-saturation, but it is also often referred to as either “boiling” or “vapor-saturation.” These terms often lead to semantic confusion and the footnote* should provide some clarity on the issue. Because the aqueous ?uid has a density that is considerably lower (usually less than or around 1 g cm -3 , see Figure 2.2c) than that of the granitic magma (which is typically around 2.5g cm -3 ) it will tend to rise and concentrate in the roof, or cara- pace, of the magma chamber. Although some of the original OH - in the magma may be utilized to form hydrous rock-forming minerals (such as biotite and hornblende), the amount of magmatic- hydrothermal water formed in this way can be very substantial. The concept of the formation of a zone of H 2 O-saturated magma in a high-level granite intru- sion (2 km depth) which initially contained some 2.7 wt% H 2 O is schematically illustrated in Fig- ure 2.5. At these shallow depths H 2 O-saturation is achieved after only about 10% crystallization, when the water content of the residual magma 82 PART 1 IGNEOUS PROCESSES saturated carapace 8 Depth (km) 4 5 7 0 21 2 Distance (km) 01 1 2 3 6 2.0 1.5 1.0 0.5 Pressure (kb) H 2 O 1000°C 1000°C Solidified portions Country rock Residual melt S Figure 2.5 Section through a high-level granodioritic intrusion undergoing progressive crystallization and showing the hypothetical position in space of the H 2 O-saturated granite solidus (S), as well as the zone (in gray) where aqueous ?uid saturation occurs in the residual magma (after Burnham, 1979). * Unless the pressure, temperature, and composition of a magmatic aqueous solution are speci?ed, one cannot say whether it exists as liquid or vapor, or a homogeneous supercritical phase (see Figure 2.2). In such a case it should be referred to by the generic term “H 2 O ?uid.” Since the magmatic aqueous phase is so much less dense than the silicate melt from which it was derived, and because it may contain other low solubility volatile species such as CO 2 or SO 2 , it is often referred to as the “vapor” or “volatile” phase. In addition, a homogeneous supercritical ?uid is one that would effectively ?ll its container, and in this sense should be regarded as a gas or vapor, even though its density might be much higher than a gas as we might envision one at the Earth’s surface. Accordingly, in this book the terms H 2 O-saturation, boiling, degassing, and vapor- saturation are used interchangeably. The concepts are discussed again in section 2.4.4. ITOC02 09/03/2009 14:37 Page 82MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 83 reaches 3.3 wt%. At low pressures such as those pertaining in Figure 2.5, the ?uid does in fact boil, since the equilibrium vapor pressure equals that of the load pressure on the magmatic system and bubbles of gas (i.e. water vapor plus other volatiles such as CO 2 ) vesiculate. The process whereby vaporsaturation is achieved by virtue of decreas- ing pressure (i.e. because of upward emplacement of magma or mechanical failure of the chamber) is called “?rst boiling” and is particularly applicable to high level systems. As mentioned earlier, it is also possible to achieve saturation with respect to an aqueous ?uid by progressive crystalliza- tion of dominantly anhydrous minerals under isobaric conditions, and this process is referred to speci?cally as “second boiling.” Second boiling generally occurs in more deep-seated magmatic systems and occurs only after a relatively ad- vanced stage of crystallization. As shown later, the differences between ?rst and second boiling, and more speci?cally the timing of H 2 O ?uid sat- uration relative to the progress of solidi?cation of the magma, is very important in understanding how different granite related ore deposits form (see sections 2.6 and 2.7). In addition to the strong dependence of boiling on pressure, it is obvious that ?uid saturation will also be a function of the original water content of the initial melt. Melts that are more enriched in water and volatiles will achieve saturation earl- ier (relative to the progression of crystallization) than those that are poorer in these components. Experimental studies have con?rmed the effects of pressure and initial water content on ?uid saturation. Figure 2.6 compares the attainment of H 2 O-saturation in a typical granite melt at high and low pressures, and in terms of crystallization sequences and initial H 2 O content. In a situation deep in the Earth’s crust (i.e. 8 kbar), where the original granite melt contained 2 wt% H 2 O, crystallization would have commenced with nucleation of plagioclase at temperatures around 1100 °C, followed by the appearance of K-feldspar and quartz on the liquidus at lower temperatures (A–A'–A¨, Figure 2.6a). H 2 O-saturation is achieved at temperatures that are just a few degrees above the solidus (at A') and only after over 80% of the melt had crystallized. The solidus is intersected at 650 °C, at which point the granite has totally solidi?ed. In the unlikely event that this granite was initially H 2 O saturated at 8 kbar it would have contained at least 12 wt% H 2 O and, as illustrated by the crystallization path B–B' (Figure 2.6a), would not have started to crystallize plagioclase until the melt temperature had cooled to around 750°C. In this case solidi?cation would have progressed quite rapidly between 750 and 650 °C and entirely in the presence of H 2 O ?uid. At shallower levels in the crust (2 kbar) the sit- uation is quite different (Figure 2.6b). The same granite magma composition would be saturated in water if it originally contained only 6–7 wt% H 2 O and in this situation (path D–D') crystalliza- tion in the presence of H 2 O ?uid would take place over a wider temperature interval than at greater depth. The same crystallization path at 8 kbar would have existed over much of its temperature range in the H 2 O undersaturated ?eld. A granitic melt at 2 kbar with low initial water content (2 wt%; path C–C'–C¨ in Figure 2.7b) would also crystallize over a signi?cant temperature interval in the undersaturated ?eld, as with deeper in the crust, but in this case H 2 O-saturation would be achieved at a higher temperature (around 700 °C at C') and after only some 60–70% crystallization. These experimental data reinforce the concept that an aqueous ?uid will exsolve from a granitic melt as a normal consequence of its crystalliza- tion. The Burnham model has great relevance to the formation of a wide range of ore deposit types. The porphyry Cu–Mo suite of deposits, epither- mal precious metal ores, and polymetallic skarn type deposits are all examples of deposits whose origins are related to the processes conceptualized in this model. More detailed discussion of these deposit types is presented in later sections. A note on the mechanical effects of boiling An aqueous ?uid constrained to the roof zone of a granite magma chamber will have limited effect on the concentration of metals unless it has the opportunity to circulate ef?ciently in and around the intrusive complex from which it is derived. The appearance of an exsolved H 2 O ?uid within a magma is, however, also accompanied by the ITOC02 09/03/2009 14:37 Page 8384 PART 1 IGNEOUS PROCESSES 1000 T (°C) 600 800 1200 24681 01 2 Wt% H 2 O (b) CD D’ C" 1000 T (°C) 600 800 1200 2 4 6 8 10 12 Wt% H 2 O (a) A" Liquidus H 2 O – saturation Solidus 14 B A Solidus L + V H 2 O – saturation Liquidus B’ P = 8 kb P = 2 kb L PI+L PI+L+V PI+Af+L+V PI+Af+Q+V PI+Af+Q+L+V L+V PI+Af+Q+L PI+Af+L+V PI+L+V PI+L L PI+Af+Q+L PI+Af+L PI+Af+L PI+Af+Q+V PI+Af+Q+L+V A' C' Figure 2.6 Plots of temperature versus H 2 O content showing the crystallization sequences for granitic melts cooling and solidifying at (a) deep crustal levels (8 kbar) and (b) shallower crustal levels (2 kbar). The bold lines in both cases refer to the H 2 O- saturation curve, and also the liquidus and solidus (after Whitney, 1989). Pl, plagioclase; Q, quartz; Af, alkali feldspar; L, melt; V, H 2 O ?uid. Crystallization paths refer to hypothetical situations where the granite was either initially oversaturated in H 2 O (B–B' and D–D') or markedly undersaturated in H 2 O (A–A'–A¨ and C–C'–C¨). ITOC02 09/03/2009 14:37 Page 84MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 85 release of mechanical energy, since the volume per unit mass of silicate melt plus low density H 2 O ?uid is greater than the equivalent mass of H 2 O saturated magma (Burnham, 1979). At shallow levels in the crust the volume change accompanying H 2 O ?uid production may be as much as 30% (at P total of 1 kbar). This results in overpressuring of the chamber interior and can cause brittle failure of the surrounding rocks. The hydrofracturing that results from this type of failure usually forms fractures with a steep dip, because expansion of the rock mass takes place in the direction of least principal stress, which is usually in the horizontal plane. Hydrofractures tend to emanate from zones of H 2 O ?uid produc- tion in the apical portions of the granite body and may propogate into the country rock and even reach surface. This concept is schematized in Figure 2.7. Experimental work has con?rmed that high level granite emplacement enhances the possibil- ity of brittle failure, both in the intrusion itself and in the surrounding country rocks (Dingwell et al., 1997), thereby providing excellent ground preparation for the ef?cient circulation of ore- bearing ?uids. The factors that help to promote brittle failure in high level granite-related ore- forming systems include volatile saturation, which increases magma viscosity because of dehydration, bubble vesiculation, and rapid cooling. 2.4 THE COMPOSITION AND CHARACTERISTICS OF MAGMATIC-HYDROTHERMAL SOLUTIONS 2.4.1 Quartz veins – what do they tell us about ?uid compositions? Quartz veins are the products of precipitation of silica from hot aqueous solutions percolating through fractures in the Earth’s crust. As pressure and temperature increases, water becomes an increasingly powerful solvent and can dissolve signi?cant quantities of most rock-forming min- erals. The solubility of quartz in water increases to about 8 wt% at temperatures of 900 °C and pressures up to 7 kbar (Anderson and Burnham, 1965). When dissolved in water, silica exists in 8 Depth (km) 4 5 7 0 21 2 Distance (km) 01 1 2 3 6 2.0 1.5 1.0 0.5 0.0 Pressure (kb) saturated carapace H 2 O 1000°C S S S S 1000°C Breccia pipe Quartz vein Figure 2.7 Section through a high-level granodioritic intrusion showing the nature of hydrofracturing and breccia pipe formation that could form around the apical portion of a granite body (after Burnham, 1979). ITOC02 09/03/2009 14:37 Page 8586 PART 1 IGNEOUS PROCESSES 0.5 Solubility (molal) 0.4 0.3 0.2 0.1 6 5 4 3 2 0.5 700 Solubility (molal) 0.4 0.3 0.2 0.1 600 500 400 300 Temperature (°C) Pressure (kb) (a) 0.10 0 Solubility (molal) 0.08 0.06 0.04 0.02 10 9.6 9.2 8.8 8.4 0.10 Solubility (molal) 0.08 0.06 0.04 0.02 300 200 100 Temperature (°C) pH (b) Figure 2.8 The variation of quartz solubility in aqueous solution as a function of (a) pressure and temperature and (b) temperature and pH (after Rimstidt, 1979). ITOC02 09/03/2009 14:37 Page 86MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 87 the form H 4 SiO 4 . This is a typical constituent of hydrothermal solutions, and explains the com- mon occurence of quartz in veins. The variation of quartz solubility in aqueous solution is, how- ever, quite complex (Rimstidt, 1997) and other parameters in addition to P and T, such as pH and salinity, play an important role. The combined effects of P, T, and pH on quartz solubility in aqueous solutions is shown in Figure 2.8. 2.4.2 Major elements in magmatic aqueous solutions In addition to silica, water can dissolve signi?c- ant amounts of other major elements, such as the alkali metals. In a now classic experiment, Burnham (1967) reacted granite with pure water under a variety of conditions and showed that at high pressures and temperatures (10 kbar and 650 °C) the total solute content of the solution was about 9 wt% and comprised Si, Na and K in proportions that are approximately the same as the granite eutectic (or minimum melt) composition (Figure 2.9). This indicated that material precipit- ating from an aqueous solution at high P–T could have the same composition and mineralogy as granite that crystallized from a silicate melt (i.e. quartz + plagioclase + K feldspar in approximately equal proportions). At progressively lower pres- sures and temperatures, however, the total solute content of the solution decreased (to a minimum of about 0.7 wt% at 2 kbar), with the alkali metals (i.e. Na +K) also decreasing relative to silica (Figure 2.9). Close to the surface, therefore, the products precipitating from an aqueous hydro- thermal solution comprise mainly silica. In addition to their ability to dissolve cationic species (i.e. electron acceptors) such as Na + , K + , and Si 4+ , magmatic aqueous solutions can also transport signi?cant amounts of Ca 2+ , Mg 2+ , and Fe 2+ , as well as a variety of anionic substances, in particular Cl - . The anions are referred to as lig- ands (electron donors) and there are several others which are also commonly found in aqueous solu- tions, including HS - , HCO 3 - , and SO 4 2- . These additional components are important in ore-form- ing processes and are discussed in more detail below. Further discussion of solution chemistry, and of the ability of hydrothermal ?uids to trans- port different metals, is presented in Chapter 3. 2.4.3 Other important components of magmatic aqueous solutions Rocks are mainly made up of some ten major element oxides that are used to form the more abundant rock-forming minerals. Any given rock sample also comprises most other known ele- ments, albeit for many of them in vanishingly small quantities. Their detection, for the most part, depends on the barriers imposed by analyt- ical technology. To a certain extent, the same is true of the solute content of aqueous solutions in the Earth’s crust, which, although dominated by a few highly soluble species, contain a wide range of other constituents, some of which are easily detectable, with the remainder occurring in only minute quantities. The trace constituents of an aqueous solution cannot simply be dismissed, especially in an ore-forming context, as it is these ingredients that distinguish an ore-forming ?uid from one that is likely to be barren. The typical composition of a magmatic aque- ous solution in the Earth’s crust is probably best obtained by direct analysis of waters produced adjacent to active volcanoes or geothermal springs. Although it is now known that such ?uids are not entirely magmatic in origin (most have signi?cant contributions from meteoric waters), their com- position gives a reasonable indication of the types and quantities of dissolved components in aque- ous solutions. Table 2.1 shows the chemical compositions of volcanic and geothermal ?uids from a variety of places around the world. The data show that these natural ?uids are variable in composition and this re?ects the different rocks through which they have been circulating, and also, possibly, contamination by other types of ?uid. Typically the solute content of magmatic ?uids is dominated by the alkali and alkali-earth metal cations and by chlorine as the major ligand, although exceptions do occur (e.g. Rotorua; Table 2.1). A few additional comments are relevant. There is often a signi?cant amount of carbon dioxide associated with magmatic ?uids and this is ITOC02 09/03/2009 14:37 Page 87discussed again below. The amount of sulfur in magmatic ?uids is generally low, but this may re?ect the fact that at high crustal levels SO 2 partitions into the vapor phase on boiling. The oxidized and reduced forms of sulfur are essen- tially mutually exclusive and exist either as the SO 4 2- complex (with S 6+ ), or as the reduced HS - complex (with S 2- ). Relatively oxidized I-type magmatic ?uids tend to comprise sulfur as SO 4 2- which fractionates into the aqueous liquid phase. Porphyry Cu and Mo type deposits, therefore, are associated with abundant sul?de minerals in the form of pyrite and chalcopyrite. S-type magmas, which are more reducing because of equilibration with graphite-bearing metasediments, exsolve aqueous ?uids that contain mainly H 2 S or HS - and have lower total sulfur contents. The presence of reduced sulfur species promotes the stability of 88 PART 1 IGNEOUS PROCESSES 70 90 80 70 60 50 40 30 Wt% 20 10 90 80 70 60 50 40 30 20 10 90 80 60 50 40 30 20 10 10 8 6 4 2 2.2 2.0 3.0 2.0 2.0 4.0 4.0 4.0 4.0 6.0 6.0 6.0 6.0 5.7 8.3 7.6 7.7 8.4 7.8 9.8 10.0 Spruce Pine pegmatite SiO 2 Qtz Ab NaAISi 3 O 8 KAISi 3 O 8 Or Total solute (wt%) (Temperature 900–600°C) Compositional trend with decreasing pressure 10.0 = pressure (kb) Figure 2.9 Normalized compositions of aqueous ?uid in equilibrium with a granitic pegmatite at pressures varying from 2 to 10 kbar (pressure shown next to each point on the diagram) and temperatures between 600 and 900 °C. The auxiliary diagram on the left shows the approximate variation in total solute content of the aqueous solutions as a function of pressure (after Burnham, 1967). ITOC02 09/03/2009 14:37 Page 88MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 89 sul?de minerals down to lower temperatures, such that the Sn and W oxide minerals (cassiterite and scheelite) associated with S-type granites tend to form early in the mineralizing sequence of these systems (Burnham, 1997). It is important to point out that Table 2.1 illustrates the major dissolved components of natural aqueous solutions. The concentrations of ore metals, which should be regarded as trace constituents in hydrothermal ?uids, are usually considerably lower than those shown in the table. Recent analytical developments have allowed measurements of magmatic ore-forming ?uids directly from the tiny volumes of ?uid trapped in individual ?uid inclusions and this has provided some important new insights into the composi- tions of both the aqueous liquid and vapor phases in granite ore-forming systems. Magmatic ?uid compositions from ?uid inclusion analysis Ore-forming ?uids can perhaps best and most directly be studied by examining the ?uid inclu- sions (typically 5–30 µm in diameter) that exist particularly in quartz, but also in many other rock- and ore-forming minerals. Studies of por- phyry Cu and Mo systems have shown that the early generation of ?uid trapped in primary ?uid inclusions is characterized by high temperatures and high salinities, and also provides evidence of liquid–vapor phase separation, or boiling (Roedder, 1984). The high salinities of these ore-forming ?uids are con?rmed by the fact that their asso- ciated ?uid inclusions often contain daughter crystals (i.e. tiny minerals that precipitated from the ?uid on cooling after it was trapped within the inclusion; see inset microphotograph in Figure 2.10a). The majority of these daughter crystals are identi?ed as halite (NaCl) and, to a lesser extent, sylvite (KCl), and con?rm the fact that magmatic ?uids contain signi?cant amounts of K, Na, and Cl. A plot of ?uid compositions derived from microthermometric homogeniza- tion experiments on individual ?uid inclusions for a wide range of porphyry Cu deposits is shown in Figure 2.10a. A more detailed description of these data and their signi?cance can be found in Roedder (1984). A much more accurate and complete assessment of porphyry Cu-related ore ?uid composition, however, is obtained from the quantitative ana- lysis of individual ?uid inclusions by laser-ablation inductively coupled plasma mass-spectrometry (LA-ICP-MS). This remarkable technique allows recognition and analysis of a wide range of cationic species in the tiny volume of inclusion ?uid, and down to levels of about 1 ppm. Figure 2.10b shows the results of LA-ICP-MS analysis of two coexisting ?uid inclusions in quartz from the Bajo de la Alumbrera porphyry Cu–Au deposit (Ulrich et al., 2001). The two coexisting ?uid inclusions, one a vapor-rich inclusion and the other a hyper- saline liquid-rich inclusion containing several daughter crystals, indicate that the ore ?uid was boiling as it was trapped. Their analysis re?ects the extent to which elements were partitioned between liquid and vapor, as well as the metal abundances in each. The plot shows that the majority of ele- ments analyzed partition preferentially (by a factor Table 2.1 Typical solute concentrations (in mg kg -1 or ppm) of the main components that comprise natural aqueous solutions from a variety of geothermal and volcanic ?uids Location pH Na + K + Ca 2+ Mg 2+ Fe 2+ Cl - HCO 3 - - SO 4 2- - White Island 1.1 8630 960 2010 3200 6100 7300 <1 6600 Mahagnao 5.8 20 340 4840 2900 95 <0.1 46 235 20 138 Rotorua 6.8 147 20 8 1 <0.1 13 560 <5 Sea water – 10 560 380 400 1270 – 18 900 140 2710 Sea water is shown for comparison. Source: after Giggenbach (1997). ITOC02 09/03/2009 14:37 Page 8990 PART 1 IGNEOUS PROCESSES NaCl H 2 O NaCl Ice Porphyry Cu fluid inclusions KCl KCl Hydrohalite (a) Concentration (vapor-rich fluid inclusion) (ppm) 1 1 10 100 Concentration (hypersaline liquid-rich fluid inclusion) (ppm) 1000 10 000 100 000 Hypersaline, liquid-rich fluid inclusion Zn Pb Rb Sr Ba Bi Ag Cs Mn Ce 10 100 10 000 100 000 1000 Cu Vapor-rich fluid inclusion Na K Fe 1:1 Th = 720°C P = 680 bars (b) Figure 2.10 (a) Plot of the system H 2 O–NaCl–KCl showing the compositional range of ?uids in ?uid inclusions from a variety of different porphyry Cu deposits (after Roedder, 1984). Inset microphotographs show (top) a hypersaline (containing a vapor bubble, saturated brine, and a daughter crystal) ?uid inclusion, and (bottom) a trail of vapor-rich ?uid inclusions. (b) Plot of element concentrations (obtained by laser- ablation ICP-MS analysis) in a vapor-rich ?uid inclusion versus element concentrations in a related hypersaline liquid-rich ?uid inclusion. Schematic illustrations of the nature of the two ?uid inclusions analyzed in order to create the plot are shown next to the respective axes. The data pertain to the Bajo de la Alumbrera porphyry Cu–Au deposit in Argentina. The plot is redrawn after Ulrich et al. (2001). ITOC02 09/03/2009 14:37 Page 90MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 91 of about 10 to 20 ×) into the liquid phase rather than the vapor (Figure 2.10b). As expected, the Na and K contents of the ore ?uid are high, but so is the Fe content. The liquid also contains signi?cant quantities of Mn, Cu, Pb, and Zn in solution. Interestingly, Cu behaves differently to the other metals in that it is preferentially frac- tionated into the vapor phase, a feature also previ- ously noted by Lowenstern et al. (1991) and Heinrich et al. (1999). 2.4.4 Carbon dioxide in magmatic ?uids After water, CO 2 is the most common gas eman- ating from volcanic eruptions and its presence is important in a variety of ore-forming processes, albeit in a more obscure way than for H 2 O. Many different studies have con?rmed that the solubil- ity of CO 2 in magmas is typically an order of magnitude lower than that of H 2 O. It dissolves as molecular CO 2 in normal felsic or ma?c melts, but in alkaline magmas it also exists as carbonate ionic complexes in solution (Lowenstern, 2001). As an indication, rhyolite and basalt melts at 3 kbar will normally contain between 1000 and 2000 ppm dissolved CO 2 compared to values at least 4–5 times higher in a leucitite melt. CO 2 solubilities, therefore, increase as a function of pressure and magma alkalinity. Because CO 2 solubilities are much lower than those of H 2 O, it follows that CO 2 will exsolve early such that vapors generated at an early stage of solidi?cation will tend to be more CO 2 -rich than those forming later. Carbon dioxide will partition preferentially into the vapor phase and it is probable that many normal magmas, both felsic and ma?c, will be vapor-saturated at sig- ni?cant crustal depths, even to the mid-crust (Lowenstern, 2001). Other low solubility gases are also likely to be associated with the vapor phase created by early CO 2 saturation or effervescence. Volatiles that are generated in systems that have undergone protracted differentiation, or reach shallow crustal levels, will tend to be H 2 O- dominated since the relatively insoluble gases like CO 2 , as well as N 2 and others, will already have bubbled away. It is, therefore, possible for mag- mas to become volatile-saturated before, and inde- pendently of, H 2 O ?uid saturation. The presence of CO 2 in an evolving aqueous ?uid within a cry- stallizing granite will promote immiscibility between vapor and saline liquid phases of the solution. Such processes can be very important during ore-forming processes since they pro- mote the precipitation of metals from solution (Lowenstern, 2001). Effervescence of CO 2 from the ?uid will also promote certain types of altera- tion in the host rocks and increase pH in the remaining ?uid, further in?uencing ore-forming processes (see Chapter 3 for more discussion of these and related topics). Thus, although CO 2 does not appear to play a direct role in the trans- port and concentration of metals in hydrothermal solutions, it nevertheless has a role to play in the distribution and precipitation of metals, espe- cially via the volatile or vapor phase. 2.4.5 Other important features of magmatic ?uids An important characteristic of magmatic aqueous solutions at low pressures is that they tend to segregate into two phases of differing densities and composition. This feature of hydrothermal solutions is often overlooked, and it also further complicates the description of these ?uids. It is evident that the so-called boiling-point curve (Figure 2.2) is actually part of a solvus which separ- ates low density H 2 O from high density H 2 O. As emphasized above, this distinction becomes meaningless in P–T space above the critical point (or the solvus) where densities of liquid and vapor have merged and only a single “?uid” phase exists. The critical point of pure water, at 374 °C, 220 bar, and 0.322 g cm -3 , migrates towards higher temperatures and pressures as dissolved salts (usually in the form of alkali chloride complexes) are added to water. A representation of the critical points for hydrothermal solutions of differing salinities and the effect on ?uid densities (iso- chores) is shown in Figure 2.11a. The locus of critical points, as a function of increasing salinity, represents a solvus boundary which governs the existence or otherwise of immiscible behavior in aqueous solutions. The phase relationships of magmatic aqueous solutions (brines) are also, therefore, dependent ITOC02 09/03/2009 14:37 Page 91on this solvus which, depending on the pressure, temperature, and composition of the ?uid, dict- ates whether a single homogeneous phase exists above the critical point, or two immiscible phases below it. Figure 2.11b shows a projection of the H 2 O–NaCl system onto the T–X plane in which the solvus curves for various pressures are shown. The phase relationships show that, for a granite at 600bar in which an exsolved H 2 O ?uid has formed with a composition at point A, the ?uid exists below the relevant solvus. Such a ?uid must segregate into two phases, a small propor- tion of brine containing 78 wt% NaCl (A¨) and a more voluminous low density ?uid (A') contain- ing only 1 wt% NaCl (Burnham, 1997). The same granite at 1 kbar (point B, Figure 2.11b) would also 92 PART 1 IGNEOUS PROCESSES CP 700 600 400 200 200 Temperature (°C) 1400 1000 Pressure (bars) (a) 300 400 500 600 800 1200 0.92 0.99 1.11 Liquid Vapor CP CP 0.58 0.64 0.79 Pure water 10 wt% NaCl 25 wt% NaCl 1000 400 0 Temperature (°C) 500 800 700 600 900 50 100 NaCl (wt%) 500 bars 600 750 1000 1300 1500 B A C 2000 bars 1750 Critical point curve Solvus curves A' A" B' B" (b) Figure 2.11 (a) Boiling point (liquid– vapor) curves for pure water and for two aqueous solutions containing 10 and 25 wt% NaCl, showing the shift of critical point to higher temperatures and pressures with increasing salinity. Selected isochores, with relevant densities (in g cm -3 ), are also shown for each of the three cases. The diagram is redrawn after Roedder and Bodnar (1980). (b) A projection of the system H 2 O–NaCl showing the solvus curves, for a variety of pressures between 500 and 1750 bars, that de?ne whether a homogeneous aqueous solution will segregate into two immiscible phases or not (after Bodnar et al., 1985). ITOC02 09/03/2009 14:37 Page 92MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 93 exsolve two aqueous solutions, but with different compositions. The brine would in this case con- tain 53 wt% NaCl (B¨), whereas the remaining ?uid would have 2 wt% NaCl (B'). The granite at 2 kbar (point C), however, would exist above the solvus (which falls off the diagram in Figure 2.11b) and would exsolve a single, homogeneous ?uid phase containing about 12.5 wt% NaCl. It should be noted that the presence of CO 2 in the ?uid will promote immiscibility to much higher pressures and, as mentioned above, this is one of the import- ant roles that this common gas has to play in ore- forming processes. These features of an H 2 O ?uid exsolving from a granite magma are extremely important to the understanding of ore-forming processes because metals partition themselves differently between the high solute content brine and the more pure H 2 O phase. This topic is dis- cussed in more detail in section 2.6 below. The discussion of magmatic-hydrothermal ?uids and ore-forming processes in the preceding sections has given rise to some confusing termi- nology. As a summary, and before we proceed further, the following de?nitions are presented in order to clarify some of the concepts referred to thus far. • H 2 O-saturation. At some stage, either early or late in the crystallization sequence, granitic magma will become water saturated, resulting in the formation of a chemically distinct, aqueous phase in the silicate melt. The magmatic aqueous phase can exist as a liquid, vapor, or homogeneous supercritical ?uid. In the last case it should also be regarded as gas since it is a substance that would ?ll its container. For this reason the magmatic aqueous ?uid is often sim- ply referred to as the “vapor phase.” • Vapor-saturation, also referred to as boiling, occurs when the equilibrium vapor pressure of the magma equals that of the load pressure on the system and bubbles of gas (i.e. steam or water vapor, CO 2 , N 2 , SO 2 , H 2 S, etc.) nucleate in the magma. Vapor-saturation of low solubility volatiles such as CO 2 may be unrelated to, and can precede, H 2 O-saturation. The process of H 2 O-saturation can be achieved in two ways, either by progressive crystallization of magma, or by decreasing the pressure of the system: • First boiling refers to the case where vapor- saturation is achieved by virtue of decreasing pressure (i.e. because of upward emplacement of magma or mechanical failure of the chamber) and is particularly applicable to high level sys- tems. It re?ects the fact that solubilities of volatile phases in a melt increase as a function of pressure. • Second boiling refers to the achievement of H 2 O ?uid saturation by progressive crystalliza- tion of dominantly anhydrous minerals under isobaric conditions. It pertains to more deep- seated magmatic systems and occurs only after a relatively advanced stage of crystallization. • Immiscibility refers to the tendency for brine solutions at low pressures, or in the presence of CO 2 , to segregate into two phases, one a dense, more saline brine and the other a lower density, low salinity aqueous solution. 2.5 A NOTE ON PEGMATITES AND THEIR SIGNIFICANCE TO GRANITE-RELATED ORE-FORMING PROCESSES Pegmatites are commonly regarded as rocks derived from magma that may have crystallized in the presence of a magmatic aqueous ?uid. They are de?ned as very coarse-grained rocks, typically associated with granites and comprising the major granite rock-forming minerals. The large crystals of quartz, feldspar, and muscovite that make up the bulk of most pegmatites are often extracted for industrial purposes. In addition, they can comprise a wide variety of minor minerals of more exotic and semi-precious character, such as tourmaline, topaz, and beryl. Pegmatites also con- tain concentrations of the large ion lithophile and high ?eld strength elements, such as Sn, W, U, Th, Li, Be, B, Ta, Nb, Cs, Ce, and Zr. Their origin is a topic that provides a fruitful area of research because of the clues they provide for ore-forming processes and igneous petrogenesis in general. Although there are several classi?cation schemes for pegmatites, the model by Cerny (1991) is particularly convenient because it separ- ates them into characteristic metal assemblages that have an implied genetic connotation. Two families of pegmatites are recognized, the Nb–Y–F ITOC02 09/03/2009 14:37 Page 93suite associated with sub-alkaline to metalum- inous (largely I-type) granites and the Li–Cs–Ta suite, also enriched in boron, and typically asso- ciated with peraluminous (dominantly S-type) granites. Some pegmatites clearly form as the result of small degrees of partial melting and form minor dykes and segregations in high grade meta- morphic terranes. Other pegmatites are spatially associated with the cupola zones of large granite intrusions and may be genetically linked to the most highly differentiated, water-saturated portions of such bodies. Many pegmatites, on the other hand, are almost certainly not the products of water-saturated granite crystallization. The origin of pegmatites is one of the more fascinating problems in igneous petrology and, even though the processes are now well understood, they remain complex and somewhat contentious. Both the early and later developments in the under- standing of these rocks are instructive to ore- forming processes, and are discussed below. 2.5.1 Early models of pegmatite genesis Prior to the classic paper by Jahns and Burnham (1969), pegmatites were viewed as the products of extreme crystal fractionation of mainly granitic magmas. Experimental work, however, suggested that pegmatites were distinguished from “norm- ally textured” granite by crystallization in the presence of an exsolved H 2 O ?uid phase. More speci?cally, Jahns and Burnham (1969) suggested that the transition from granite to pegmatite marked the point at which H 2 O ?uid saturation occurred in the crystallization sequence and that pegmatites, therefore, formed in the presence of an immiscible H 2 O + volatile fraction. The exist- ence of H 2 O ?uid in the magma was used to explain the large crystal size of pegmatites, the latter growing, because of extended crystalliza- tion due to either volatile-related depression of the granite solidus, or more ef?cient diffusion of major elements into low viscosity H 2 O ?uid- saturated magma. Mineral zonation in pegmatites was attributed largely to the separation of relat- ively dense silicate melt from the low density H 2 O ?uid phase, and the attendant segregation of alkalis such that Na remained in the silicate melt and K partitioned in the aqueous phase. In addition, it was envisaged that precipitation of dissolved constituents from the H 2 O ?uid phase could explain the pockets of exotic minerals (including gemstone-quality crystals formed in an open space or vug from ?uids enriched in Li, B, Cs, Be, etc.) so often characteristic of pegmatites. The Jahns and Burnham model was particularly attract- ive because it was supported by experimental evidence demonstrating that an aqueous ?uid at high temperatures and pressures is capable of dissolving very signi?cant proportions of solute (see Figure 2.9). 2.5.2 More recent ideas on the origin of pegmatites The Jahns–Burnham model represented a major advance in the understanding of pegmatite genesis, as well as the crystallization and mineralization of granites generally. More recent work, however, has shown that there are other factors, in addition to H 2 O-saturation, which need to be considered in order to fully understand the formation of these rocks and their associated ore deposits. The research of D. London and co-workers (1990, 1992, 1996) in particular has shown that it is possible to generate pegmatites from H 2 O under- saturated granitic melts by undercooling the magma below its normal liquidus temperature. They have demonstrated experimentally that kinetic delays in the initiation of crystallization in felsic magmas that are rapidly undercooled means that melts persist in a metastable condi- tion and result in non-equilibrium crystal growth. Many of the features of pegmatites, including mineral zonation, large grain size, variable tex- tures, and highly fractionated chemistry, can be replicated experimentally in terms of the slow crystallization response to undercooling of granitic melts. This work has enabled pegmatites to again be explained by magmatic crystal fractionation process (as had been suggested prior to the Jahns– Burnham model) rather than to silicate melt–H 2 O ?uid exsolution and alkali element segregation. The London model has removed a fundamental dif?culty always faced by the Jahns–Burnham model, namely how to explain the symmetric 94 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 94MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 95 mineral zonation typical of so many pegmatites in terms of segregation mechanisms that, being gravitationally induced, should have been intrin- sically asymmetric. It also provided a rather neat explanation for why pegmatitic textures are uncommon in more ma?c rocks. Ma?c melts nucleate crystals much more readily than their felsic equivalents, resulting in a tendency toward equilibrium crystallization conditions and the formation of more uniform textures. It should be emphasized, however, that more recent work on pegmatites has not, by any means, precluded the important role of water in the formation of these rocks. A ?uid inclusion study of the Tanco Li–Cs–Ta pegmatite in Manitoba, Canada, showed that it crystallized from a mix- ture of alumina–silicate melt plus H 2 O (with minor CO 2 ) ?uid over a temperature interval that extended from about 700 °C down to below 300 °C (Thomas et al., 1988). These observations were, in fact, interpreted as support for the Jahns–Burnham model. The processes are further complicated when one takes into account the major in?uence of other volatile species, such as B, F, and P, on pegmatite formation (London, 1996). These three elements, individually and collectively, lower the granite solidus temper- ature to below 500 °C and increase the range of temperatures over which magmatic crystalliza- tion occurs. They also dramatically increase the solubility of H 2 O in the melt and promote the crystallization of quartz. In a study of the Ehrenfriedersdorf Sn–W–B–F–P-rich pegmatites of the Erzgebirge in southeast Germany, it has been shown that ?uid inclusions are made up of coex- isting silicate-rich H 2 O-poor and silicate-poor H 2 O-rich melts (Thomas et al., 2000). On progress- ive reheating from 500 to 700 °C at 1 kbar pres- sure, the latter inclusions become poorer in H 2 O, whereas the former undergo a steady increase in water content (Figure 2.12). At just above 700 °C the two inclusion populations are effectively identical, with both containing about 20 wt% H 2 O. This indicates that the silicate-rich H 2 O- poor and silicate-poor H 2 O-rich melts were in fact derived from a single, originally homogeneous, melt. Even at low pressures (the Ehrenfriedersdorf pegmatites crystallized at about 1 kbar) these B–F–P-rich pegmatitic magmas exhibit complete miscibility between silicate melt and aqueous ?uid such that water can be regarded as in?nitely soluble in this melt. On cooling below 700 °C, however, immiscibility occurs to form two coex- isting melts. One of these is a silicate melt with low H 2 O content, while the other is a silicate melt with very high H 2 O content. The latter, however, has a density, viscosity, and diffusivity that is more akin to an aqueous solution than to a silicate melt. Although its properties are water- like, it is by de?nition different from the exsolved H 2 O ?uid envisaged to have formed in the Jahns– Burnham model. It is apparent, therefore, that a magmatic aque- ous phase has to be represented by a ?uid that is more complex than simply exsolved H 2 O. At high pressures and temperatures water can contain signi?cant quantities of dissolved silica and alka- lis which, on cooling, will yield precipitates with bulk compositions not dissimiliar to a granite. At lower pressures H 2 O ?uid will exsolve from a normal granitic magma, such that crystallization 500 Temperature (°C) 600 550 750 01 0 4 05 0 H 2 O (wt%) 650 700 20 30 Silicate melt (A) Single phase melt (supercritical) Fluid-rich melt (B) Melt A + melt B Figure 2.12 Plot of H 2 O content versus temperature for rehomogenized (at 1 kbar) silicate-rich H 2 O-poor (A) and silicate-poor H 2 O-rich (B) melt inclusions from the Ehrenfriedersdorf pegmatite. The curve de?nes a solvus with a critical point (above which there is no longer a distinction between the two melts and they are miscible) at about 700 °C and 20 wt% H 2 O content (after Thomas et al., 2000). ITOC02 09/03/2009 14:37 Page 95will occur in the presence of an aqueous phase. At low pressures and in the presence of signi?cant concentrations of elements such as B, F, and P, granite melt solidus temperatures are signi?c- antly depressed and H 2 O solubility increases to such an extent that H 2 O-saturation might not occur at all. In this case two immiscible melts form, one of which is H 2 O-rich and has physical properties akin to an aqueous solution with high solute content. Pegmatites could conceivably form in all these situations. 2.6 FLUID–MELT TRACE ELEMENT PARTITIONING The rise of magma to shallow levels of the Earth’s crust will inevitably result in saturation with respect to an aqueous ?uid or vapor phase. Even though the diffusivity of H 2 O is relatively low in most felsic magmas, the buoyant, low density H 2 O-?uid, together with entrained melt and crystals, will move to the apical portions of the magma chamber (Candela, 1991). In an equilib- rium situation, trace components of the magma must then partition themselves between melt, crystals, and H 2 O-?uid. The extent to which trace components, such as metals, distribute them- selves between these phases is a process that is potentially quanti?able in terms of partition coef?cients. It is possible to derive equations that express the partitioning of trace constituents between melt and H 2 O-?uid in exactly the same way as was discussed for the partitioning of trace elements between melt and crystals in section 1.3.4 of Chapter 1. However, considerations of melt–?uid partitioning are complicated by the fact that it is necessary to consider partition- ing behavior prior to water saturation, and also because the sequestration of incompatible con- stituents by the H 2 O-?uid depends on solubilities of both the ?uid and the magma from which it is derived. A detailed account of the quanti?cation of melt–?uid partitioning behavior is provided by the work of Candela and co-workers, summarized in Candela and Holland (1984, 1986), Candela (1989a, b, 1992), and Candela and Piccoli (1995). Before the results of this work are discussed in more detail, brief mention will ?rst be made of some of the classic experimental work that pro- vided initial insights into why hydrothermal ?uids are such ef?cient scavengers of metals and, therefore, so important in ore-forming processes. Again, more details on this topic are provided in Chapter 3. 2.6.1 The classic experiments of H. D. Holland and others In a now classic set of experiments, Holland (1972) showed that the solubilities of many metals in a magmatic H 2 O-?uid are strongly dependent on its Cl - concentration, in addition to other parameters such as temperature and pH. When experiment- ally reacting an aqueous solution with a granitic melt it was observed that metals such as Zn, Mn, Fe, and Pb are strongly partitioned into the H 2 O- ?uid and that the ratio of metal concentration between melt and ?uid is exponentially propor- tional to the Cl - concentration (Figure 2.13a and b). These experiments demonstrated quite clearly that certain metals will only readily dissolve in an aqueous solution if the latter contains appreciable amounts of the chloride anion (Cl - ). Furthermore, metals will exhibit different magnitudes of parti- tioning into H 2 O-?uid according to their variable abilities to bond, or form complexes, with Cl - . Figure 2.13b also shows that the total Fe content of the H 2 O-?uid varies as a function of temperature, and is higher for any value of Cl - at 750 °C than it is at 650 °C. This con?rms the intuitive notion that solubility normally increases as a function of temperature. The diagram also shows that the Al content of the ?uid remains constant (because the total Fe content changes in the same way as does the total Fe/Al ratio) and that it is not affected by the Cl - concentration of the aqueous ?uid. This indicates that not all metals will have their solubilities controlled by Cl - concentrations in the ?uid (although they may still be affected by the presence of other lig- ands) and that some metals may not even choose to partition into the aqueous phase at all. The dependence of metal solubility on Cl - con- centrations presupposes that the ligand itself will partition effectively into the H 2 O-?uid phase. Kilinc and Burnham (1972) and Shinohara et al. 96 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 96MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 97 (1989) have con?rmed experimentally that chlor- ide partitions strongly into the magmatic aqueous phase, according to the reaction: Cl - (melt) + OH - (melt) ? HCl (?uid) + O 2- (melt) [2.4] Reaction between granitic melt and a chloride- bearing aqueous ?uid showed that the Cl - ion par- titions strongly into the ?uid, but that the value of the partition coef?cient varies in a complex fashion (Figure 2.14). At low pressures where the H 2 O-?uid phase segregates into two immiscible fractions (one vapor-rich and the other liquid- rich), the partitioning of Cl - into the H 2 O-?uid phase is ?xed by the saturation of Cl - in the coex- isting vapor-rich phase. At higher pressures the partitioning of Cl - into a homogeneous H 2 O-?uid phase (i.e. above the critical point) is controlled by the concentration (or activity) of Cl - in the melt, but increases with increasing pressure. The early experiments of Holland and others showed how important the composition, and more speci?cally the nature and content of the ligand 80 0 1.0 60 40 20 2.0 3.0 4.0 5.0 6.0 Cl – concentration in H 2 O fluid (mol kg –1 ) Zinc [Zn] f [Zn] m (a) K Zn ? [Cl ] f/m –2 0.2 6 4 2 0.4 0.6 0.8 1.0 1.2 Cl – concentration in H 2 O fluid (moles) ?Fe = a + b . [Cl – ] 3 Molal (?Fe/Al) v 1 7 5 3 8 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 Moles ?Fe v 750°C 650°C 650°C 750°C ?Fe/Al ?Fe/Al ?Fe ?Fe Iron (b) Figure 2.13 (a) The ratio of the concentration of Zn in H 2 O-?uid [Zn] f to that in a granitic melt [Zn] m as a function of the Cl - concentration; experiments carried out at various temperatures between 770 and 880 °C and pressures between 1.4 and 2.4 kbar (after Holland, 1972). (b) The total Fe content (left ordinate) and the total Fe to Al ratio (right ordinate) in H 2 O-?uid that equilibrated with a granitic melt as a function of the Cl - concentration; experiments carried out at temperatures of 650 and 750 °C and a pressure of 2.0 kbar (after Burnham, 1967). ITOC02 09/03/2009 14:37 Page 97species, of the aqueous ?uid phase is to metal solubility in aqueous solutions. Metal contents of aqueous solutions are not, of course, only dependent on the Cl - concentration and many other factors affect the ability of hydrothermal ?uids to transport metals. A more complete dis- cussion of this topic is presented in section 3.4 of Chapter 3. In the section below, a more de- tailed consideration is given to the partitioning of metals between silicate melt and an exsolved magmatic H 2 O-?uid in the light of recent theoret- ical and experimental work. 2.6.2 Complications in the ?uid–melt partitioning of metals The partitioning of metals between a silicate melt and an exsolved H 2 O-?uid is a complex matter. The extent to which metals are partitioned between melt and ?uid is variable and is largely controlled by the Cl - (and other ligand) con- centration of the ?uid which, in an evolving system, itself changes continuously. In addition, however, factors such as temperature, pressure, the amount of water exsolved relative to the amount of water remaining in the silicate melt (i.e. when H 2 O-?uid saturation is achieved during crystallization), and the oxygen fugacity (fO 2 ) of the silicate–?uid system also in?uence metal partitioning. This means that a partition coef?- cient cannot be regarded as a constant because it pertains only to a very speci?c, transient, set of conditions. Melt–?uid partition coef?cients, therefore, show wide variations as the magmato- hydrothermal system evolves. Candela (1989) pointed out that the metal con- tent of an exsolved H 2 O-?uid will depend to a large extent on when in the crystallization sequ- ence water saturation is achieved. This is because crystal–melt partitioning can be as ef?cient a means of distributing and concentrating trace metals in a magma as can H 2 O ?uid–melt parti- tioning. It is particularly important to distinguish between “?rst boiling” (when vapor-saturation is caused by pressure decrease) and “second boiling” (when crystallization of a dominantly anhydrous assemblage causes the H 2 O content of the re- sidual melt to rise to the saturation water con- tent for a given level of emplacement). First boiling processes are simpler to model math- ematically and pertain to ore-forming processes associated with high level granite intrusions such as those associated with porphyry Cu and Mo deposits. Second boiling requires a more com- plicated mathematical model since the metals have to partition among melt, crystal, and H 2 O ?uid phases. Fluid–melt partitioning during “?rst boiling” Candela (1989) has provided equations that quantify the distribution of a trace component between H 2 O ?uid and silicate melt for a ?rst boiling situation. These particular equations are applicable to those components whose partition- ing does not change as a function of the ligand concentration of the ?uid. They are somewhat similar in form to the exponential equations 98 PART 1 IGNEOUS PROCESSES 0.06 0 2 Concentration Cl – in melt (moles) Concentration Cl – in aqueous phase (moles) 0.04 0.02 4 6 8 1.2 kb 2.2 kb 4.2 kb 6.0 kb 0.6 kb 3.5 kb Figure 2.14 Variations in the concentration of Cl - in a silicate melt with that in an exsolved H 2 O-?uid for a variety of different pressures. Cl - is seen to be partitioned strongly into the aqueous phase relative to the melt. At low pressures, where liquid–vapor immiscibility occurs (0.6 and 1.2 kbar), the concentration of Cl - is ?xed by the saturation of Cl - in the coexisting vapor-rich phase; at higher pressures, for any given concentration of Cl - in the melt a homogeneous H 2 O-?uid phase will contain more Cl - at higher pressures (after Shinohara et al., 1989). ITOC02 09/03/2009 14:37 Page 98MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 99 applicable to fractional melting and crystalliza- tion presented in Chapter 1, and are expressed as: C i ?uid = D i ?uid/melt C i O (1 - [(1 - F)C O water ]) D ?uid/melt-1 [2.5] where: C i ?uid is the concentration of a component (i) in the H 2 O ?uid at any instant in the evolu- tion of the aqueous phase; D i ?uid/melt is the H 2 O ?uid/silicate melt partition coef?cient for the component (i); C i O is the initial concentration of the component (i) in the silicate melt (de?ned at the instant of water saturation); C O water is the initial concentration of water in the silicate melt (de?ned at the instant of water saturation); and F is the ratio of the mass of water in the silicate melt (at any given instant after water saturation) to the initial mass of water in the silicate melt. The concentration of the component (i) in the associated silicate melt (C i melt ) is simply given by: C i melt = C i ?uid /D i ?uid/melt [2.6] where C i ?uid is the value obtained from equation [2.5]. Knowing the magnitude of the partition coef?- cient (D i ?uid/melt ), which changes as a function of pressure and, hence, saturation water content, the concentrations of a component such as chloride in an evolving aqueous phase, as well as in the associated silicate melt, can be calculated. Dia- grammatic representations of Cl concentration trends in an evolving silicate melt–H 2 O-?uid system are shown in Figure 2.15. The appearance of a magmatic aqueous ?uid will typically result in depletion of the volatile component in the melt (Figure 2.15b) and this will be accompanied by a concomitant decrease of the same component in the H 2 O-?uid (Figure 2.15a), as the system evolves toward complete solidi?cation. The ef?ciency with which the aqueous phase extracts components, such as chlorine or other chloride- complexing metals, from the melt is effectively measured by the ratio of the total or integrated amount of the component in the H 2 O-?uid to the amount initially present in the melt at the time of water-saturation (Candela, 1989). The ef?ciency factor will be dependent not only on obvious criteria such as D i ?uid/melt , but also on parameters such as F (the ratio of the mass of water in the silic- ate melt at any given instant after water satura- tion to the initial mass of water in the silicate melt). Note that not all volatile components will act in the same way, as is the case for ?uorine, whose concentration in the melt is predicted to remain relatively constant during ?rst boiling (Figure 2.15b). 2000 1.0 Cl in melt (ppm) 3000 4000 0.8 0.4 0.2 Proportion H 2 O remaining in melt 0.6 Chlorine Fluorine (b) CI in vapor (mol kg –1 ) 0.8 0.4 0.2 Proportion H 2 O remaining in melt 0.6 (a) 1.0 2.0 3.0 4.0 5.0 Chlorine Figure 2.15 Calculated concentrations of chlorine in (a) an aqueous ?uid exsolved from a granitic magma, and (b) the residual magma itself, both as a function of the proportion of water remaining in the melt. The calculations were carried out for a situation where water saturation is achieved in a “?rst boiling” scenario with a magma rising adiabatically upwards from 5.4 km depth (where water content at saturation is 5 wt%) to the surface (after which 1 wt% water remains in the rock). C O Cl is assumed to be 4000 ppm and D Cl ?uid/melt is calculated to vary from 39 at the point of water saturation to 1 as the magma reaches the surface (after Candela, 1989b). ITOC02 09/03/2009 14:37 Page 99Calculation of ?uid–melt partitioning is more complex for components such as Cu or Zn, whose behavior not only varies as a function of pressure, but is also strongly controlled by the Cl - concen- tration of the ?uid phase. Likewise, the partition- ing of metals into the aqueous ?uid phase in a situation where “second boiling” has taken place also involves many additional assumptions and a more complex mathematical derivation. The detailed discussion of both of these situations is beyond the scope of this book, and the interested reader is referred to Candela (1989) for more detail. It is pertinent to note, however, that any metal which bonds strongly with Cl - in an aque- ous solution (such as Cu) will tend to exhibit par- titioning behavior that is similar to that shown in Figure 2.15. In the case of metals such as copper, therefore, the maximum concentration into the aqueous ?uid occurs immediately after water- saturation has been achieved. By contrast, metals such as molybdenum, which do not bond strongly with Cl - , and are not, therefore, controlled by concentrations of this ligand, partition differently and may be most ef?ciently concentrated in the very last portions of the evolving H 2 O-?uid phase. The contrasting behavior of Cu and Mo in the magmato-hydrothermal environment, sug- gested by these theoretical calculations, has been con?rmed by direct experimental evidence and this is discussed in the following section. 2.6.3 Experimental con?rmation of Cu, Mo, and W partitioning behavior In a series of experiments run at 750 °C and 1.4 kbar, Candela and Holland (1984) were able to measure the extent to which Cu and Mo partition between granitic melt and a coexisting aqueous ?uid containing both Cl - and F - . The results of these experiments are shown in Figure 2.18 and clearly demonstrate that Cu is a metal whose partitioning behavior is strongly controlled by the Cl - concentration of the ?uid phase, whereas Mo remains unaffected by the latter. The value of D Cu ?uid/melt varies from about 1, for low salinity ?uids, up to about 50 at high salinities (Figure 2.16a) and is determined by the relationship D Cu ?uid/melt = 9.1[Cl - ]. By contrast, although Mo is also preferentially partitioned into the H 2 O-?uid (D Mo ?uid/melt = 2.5), its partition coef?cient is ?xed and remains constant irrespective of the Cl - concentration (Figure 2.16b). It is interesting to note that the partitioning behavior of both Cu and Mo is unaffected by the presence of F - in the aqueous solution. This is because ?uorine tends to partition more strongly into the silicate melt (D F ?uid/melt = 0.2–0.3) than into the aqueous ?uid and is, therefore, generally not present as a complexing agent for metals in the ?uid phase. The ?uid–melt partitioning behavior of W is not as well constrained as that for Cu and Mo, although experiments do suggest that its proper- ties are similar to those for Mo, but that its partition coef?cient is even lower (D W ?uid/melt ? 1: Manning and Henderson, 1984; Keppler and Wyllie, 1991). However, it is also known that W behaves as an incompatible element in terms of its crystal–melt partitioning behavior in relat- ively reduced (i.e. S-type) granitic magmas solidifying at some depth in the Earth’s crust. By contrast, Mo behaves in a more compatible manner in the same magma types such that fractionation will tend to result in increasing W/Mo ratios (Candela, 1992). The ef?ciency of metal extraction by an aqueous phase exsolving late in the crystallization sequence of such a granite will, therefore, favor concentration of W over Mo, despite the lower D ?uid/melt values. The crystal–melt partitioning behavior of W also contrasts with that of copper, which tends to act as a compatible element in virtually any magma composition (Candela and Holland, 1986). A value of D Cu crystal/melt =3 was calculated for a crystallizing basaltic magma composition and is likely to be >1 for a granite too, since Cu is able to substitute very ef?ciently into accessory sul?de phases and, to a lesser extent, into biotite and magnetite. It is apparent that both ?uid–melt and crystal– melt partition coef?cients play an important role in the distribution of metals in and around crystallizing granite plutons. These parameters, when coupled with considerations of granite type, depth of emplacement, and the timing of water-saturation relative to the crystallization sequence, can be used to explain the nature and 100 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 100MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 101 origin of many different magmatic-hydrothermal ore deposit types. A model that summarizes some of the features discussed in this chapter is presented in section 2.8 below as an explanation for the characteristics of granite-related Cu, Mo, and W deposits. 2.7 WATER CONTENT AND DEPTH OF EMPLACEMENT OF GRANITES – RELATIONSHIPS TO ORE-FORMING PROCESSES Many of the ore deposits associated with granites, such as porphyry Cu and epithermal Au–Ag ores, are related to magma emplacement at high levels 10 3 2 1 6 5 4 9 8 7 10 20 30 40 50 60 [Cl – ] (mol kg –1 ) 10 20 30 40 50 60 [Cl – ] (mol kg –1 ) (a) D Cu f/m D Cu = 9.1 [Cl – ] f/m D Mo = 2.5 f/m (b) 60 30 20 10 50 40 D Mo f/m Figure 2.16 (a) The relationship between H 2 O ?uid/silicate melt partition coef?cient for Cu (D Cu f/m ) and the Cl - concentration of the aqueous ?uid. (b) The relationship between the H 2 O ?uid/silicate melt partition coef?cient for Mo (D Mo f/m ) and the Cl - concentration of the aqueous ?uid (after Candela and Holland, 1984). ITOC02 09/03/2009 14:37 Page 101of the crust where H 2 O-?uid production and hydrofracturing can take place. These types of deposits are commonly located in the volcanic or subvolcanic environment and have formed as much from the action of surface-derived (or mete- oric) water as they have from the circulation of magmatic waters. Other deposit types, such as porphyry Mo and granitoid hosted Sn–W deposits, are generally associated with magmas emplaced at deeper levels in the crust. The depth of emplacement of a granite magma, together with related parameters such as magma composition and initial water content, plays a very important role in determining the nature and origin of ore deposits associated with felsic igneous rocks, and these topics are discussed below. Section 1.3.1 described how granitic magmas derived from the melting of different source materials, at different pressures and temperatures in the crust, would contain variable initial amounts of water. A melt derived by anatexis of a rock comprising mainly muscovite, for example, would contain in the region of 7–8 wt% H 2 O. By contrast, dehydration melting of an amphibolitic source rock could only produce a melt at higher pressure and temperature (i.e. deeper in the crust) and this would contain only 2–3 wt% H 2 O (Figure 2.3). It was suggested on this basis that S-type granite magmas would initially contain more water than I-type granite magmas. In terms of this model the drier I-type granite magmas would be derived from the deep crust (possibly with contributions from the upper mantle), whereas S-type granites come from material melted in the mid- to lower-crust. Several workers have used these concepts to develop models that link gran- ite emplacement depths with their metallogenic characteristics (Hyndman, 1981; Strong, 1981). Figure 2.17a shows the same P–T diagram as that in Figure 2.3 but inverted to re?ect the sur- face (i.e. low pressures) at the top of the diagram. Hypothetical zones of melting are shown for each of three cases where the water required to initiate melting is supplied by the breakdown of mus- covite, biotite, and amphibole. If suf?cient melt is allowed to accumulate and then to rise upwards in the crust along an adiabatic cooling path, it is apparent, at least theoretically, that each of these magmas would crystallize at different levels in the crust. Adiabatic upward movement of magma (i.e. where conductive heat loss to the wall rocks is ignored) would involve cooling at a rate of about 1.5 °C kbar -1 and in P–T space would approxim- ately follow the steep curves that de?ne min- eral phase boundaries. Conceptually, therefore, magmas could rise upwards in the crust until they intersect the water-saturated granite solidus, by which time they would have become completely solid and could not intrude any further. In reality crystallization is likely to have occurred prior to this level because of heat loss to the wall rocks and the water-saturated solidus effectively repres- ents the depth above which a magma is unlikely to be emplaced. These considerations suggest that S-type granite magmas would be emplaced at mid-crustal depths (4–5 kbar), as dictated by intersection of the muscovite breakdown curve with the water-saturated granite solidus. By con- trast, intersection of the biotite or amphibole breakdown curves with the water-saturated gran- ite solidus indicates that an I-type magma could move to much shallower crustal levels (1 kbar or less) before completely solidifying (Figure 2.17a). Figure 2.17b is a schematic crustal pro?le showing the relationships between depth of emplacement and the metallogenic character of various granite-related deposit types. I-type mag- mas generated deep in the lithosphere usually form adjacent to subduction zones and commonly receive a contribution from mantle-derived ma?c melts. Forming at high temperatures (1000°C or more) and being relatively dry (H 2 O contents < 3–4 wt%) they will rise to shallow levels of the 102 PART 1 IGNEOUS PROCESSES Figure 2.17 (Opposite) Strong’s model showing the relationship between level of granite emplacement and metallogenic character. (a) Pressure–temperature plot showing the approximate conditions where ?uid-absent melting of amphibolite- (I), biotite- (II), and muscovite- bearing (III) protoliths would occur, and the expected levels in the Earth’s crust to which melt fractions would rise adiabatically, as a function of the water- saturated solidus. (b) Schematic diagram illustrating the emplacement style and metallogenic character of granites formed under each of the conditions portrayed in part (a) (after Strong, 1988). ITOC02 09/03/2009 14:37 Page 1020 Pressure (kb) 3 2 1 900 800 700 600 500 400 Temperature (°C) 4 5 6 7 8 9 Water-saturated 0 12 8 4 16 20 24 28 32 36 granite solidus Muscovite solidus Biotite solidus A C D Amphibolite solidus (a) III Granite melt (8.4% H 2 O) II Diorite melt (2.7% H 2 O) I Granodiorite melt (3.3% H 2 O) Depth (km) (b) 0 Pressure (kb) 3 2 1 4 5 6 7 8 9 12 8 4 16 20 24 28 32 36 Depth (km) PORPHYRY/EPITHERMAL DEPOSITS (Cu, Mo, Au) Amphibole (–biotite) granitoid Biotite granitoid Muscovite (–biotite) granitoid (Ground water) A Connate water Metamorphic water Veins Pegmatites (Barren) Partial melts III II I Second boiling First boiling B Volatile depressed solidus GRANOPHILE DEPOSITS (Sn, W, U) B D C ITOC02 09/03/2009 14:37 Page 103crust and may even extrude to form substantial volcanic structures. Such magmas will typically exsolve a magmatic vapor phase by ?rst boiling, an event that will also promote hydrofracturing, brecciation, and the widespread circulation of hydrothermal solutions in and around the sites of magmatic activity. These are the environments in which porphyry Cu, as well as epithermal Au–Ag deposit types, occur. By contrast, S-type magmas are generated in the mid- to lower-crust by partial melting of a source rock that comprises a substantial proportion of metasedimentary material. These melts will form at relatively low temperatures (around 700 °C) and will initially comprise signi?cant H 2 O dis- solved in the magma. Such a magma type will crystallize in the mid-crust, not too far from its site of generation, and will typically be barren. If substantial crystal fractionation occurs, however, incompatible trace elements will become concen- trated in residual melts, and in those rocks repres- enting the crystallized products of differentiated magma (see Box 1.3, Chapter 1). H 2 O-?uid satura- tion will also eventually occur in the residual magma by second boiling, to form pegmatites and related deposits. The concentration of volatiles, 104 PART 1 IGNEOUS PROCESSES Chile is the world’s leading copper producer, with an output in 2000 of about 4.6 million tons of ?ne copper, or 36% of global production (Camus and Dilles, 2001). Northern Chile in particular contains several giant por- phyry copper deposits, including world-renowned mines such as Chuquicamata, El Teniente, and El Salvador. The porphyry deposits of this huge metallogenic province formed during ?ve discrete magmatic episodes extending from the Cretaceous to the Pliocene. The most proli?c of these is the late Eocene–Oligocene period of magmatism, during which time at least 10 major porphyry copper deposits were formed, including the two biggest copper mines in the world, La Escondida and Chuquicamata. La Escondida was discovered as recently as 1981 and in 1999 produced over 800 000 tons of ?ne copper from a reserve base of more than 2 billion tons of ore at 1.15 wt% copper (Padilla Garza et al., 2001). La Escondida, together with other giant porphyry systems such as Potrerillos, El Salvador, Chuquicamata, El Abra, and Collahuasi to its north and south, lies on the Domeyko fault system (Figure 1). This faulting was dominantly trans- current during Eocene–Oligocene magmatism. Locally developed areas of transtension and dilation are con- sidered to have created an environment conducive to high level ascent of magma and accompanying magmatic- hydrothermal mineralization (Richards et al., 2001). The Escondida district actually comprises six separate deposits (of which La Escondida is the largest) contained within a system of left-lateral strike-slip faults. Mineralization is associated with quartz monzonite (adamellite) and granodiorite intrusions within which alteration evolved from early potassic to sericite–chlorite and quartz–sericite. These assemblages have been overprinted by a younger Magmatic-hydrothermal ?uids associated with granite intrusions: 1 La Escondida porphyry copper deposit, Chile (34–31 Ma) Collahuasi Domeyko Fault South America El Abra Chuquicamata (39–37 Ma) (38–34 Ma) La Escondida (41–39 Ma) El Salvador (37 Ma) Potrerillos 70° W 200 0 km 26° S 22° S Antofagasta N Figure 1 Simpli?ed map showing the position, in northern Chile, of the Domeyko fault system and the six giant porphyry copper deposits located on it. ITOC02 09/03/2009 14:37 Page 104advanced argillic alteration assemblage, and both hypo- gene sul?de and supergene sulphide + oxide ores occur (Padilla Garza et al., 2001). Cross sections through La Escondida (Figure 2) show that the system is related to a multiphase, 38 Myr old, intrusion, the Escondida stock, that cuts through Paleocene andesites. This was capped, 35 million years ago, by a rhyolite dome. Three phases of sequential alteration are recorded in the deposit. The earliest stage involves potassic alteration (K-feldspar + biotite) with associated silici?cation and pro- pylitic (chlorite–sericite) alteration, and is linked to mag- netite, chalcopyrite, and bornite. This was followed by phyllic alteration (sericite–chlorite–quartz) with which chalcopyrite, pyrite, and molybdenite are associated. The ?nal hydrothermal alteration stage involves acid-sulfate or advanced argillic alteration (quartz–pyrophyllite–alunite). Precipitation of the hypogene sul?de assemblages, mainly during the ?rst two stages of alteration, is attributed to a drop in temperature as the magmatic system cooled and was uplifted, as well as to progressive dilution of magmatic ?uids by meteoric solutions (Padilla Garza et al., 2001). An important aspect of the ore-forming system at La Escondida is the development of supergene mineralization (see Chapter 4 and Box 4.2) in areas where the three stages of hydrothermal alteration overlap (i.e. where max- imum sul?des were concentrated). Supergene processes were active between 18 and 15 Ma and were responsible for 65% of the total copper resources of the mine (Figure 2). Kfel-Chl-Ser Chl-Ser (Bt) Qz-Ser Advanced argillic Chl-Ser (Bt) Epid-Chl f (c) (b) (a) Leached capping Primary Cu-Fe sul?des (hypogene) Chalcocite blanket (supergene) Mineralization Alteration Lithology Andesite Rhyolitic dome f f f f f Pit outline 1997 Escondida stock Figure 2 North–south sections through La Escondida showing the lithologies, alteration zonation, and mineralization characteristics (after Padilla Garza et al., 2001). The upper portions of the supergene zone comprise a leached cap characterized by low Cu and Mo contents and limonite, hematite, and goethite. This is underlain by a supergene-enriched sul?de blanket containing mainly chalcocite, with lesser covellite and digenite. The best grades (up to 3.5 wt% Cu) occur in the thickest parts of the supergene zone, and also coincide with zones of intense vein stockwork and overlapping hydrothermal alteration. Copper “oxide” minerals such as chrysocolla, brochantite, and atacamite also occur, mainly in late fractures. Figure 3 View of the Escondida open pit showing the nature of altered host rocks (photo courtesy of Rio Tinto) ITOC02 09/03/2009 14:37 Page 105as well as elements such as boron and phosphorus, will, together with exsolved water in the residual magma, depress the temperature of the water- saturated granite solidus to lower temperatures (500–600 °C) such that the magma could con- tinue its upward migration to higher levels in the crust (Figure 2.17b). Fluid pressures would prob- ably not be suf?cient to fracture rock at these depths, but other structural controls would never- theless promote ?uid circulation and it is these environments in which Sn greisen deposits, por- phyry Mo deposits, polymetallic skarn ores, and mesothermal veins might form. 2.8 MODELS FOR THE FORMATION OF PORPHYRY-TYPE CU, MO, AND W DEPOSITS Considerable progress has been made in under- standing the processes that occur in a crystalliz- ing granite intrusion in order to form genetically associated base metal deposits (Candela and Holland, 1984, 1986; Candela, 1989, 1991, 1994; Candela and Piccoli, 1995). As a generalization, porphyry Cu–Mo deposits are associated with arc- related “calc–alkaline” or I-type magmas gener- ated adjacent to Andean type subduction zones. Sn–W deposits are more often associated with S- type granites that are derived by partial melting of continental crust that would normally include a signi?cant proportion of metasedimentary material. In certain settings the latter deposits are also regarded as “porphyry-type” even though they may differ signi?cantly in their geological characteristics from the Cu–Mo types. Each of these deposit types is considered in turn below. 2.8.1 The origin of porphyry Cu–(Mo) and porphyry Mo–(Cu) type deposits The family of deposit types known as porphyry Cu–Mo ores can be subdivided into two groups, one in which Cu is the dominant exploitable metal (together with minor Mo and occasionally Au), and the other where Mo is the dominant exploitable metal (with minor Cu and sometimes W). The two groups are referred to below as the Cu–(Mo) and the Mo–(Cu) porphyry types, respect- ively. Porphyry deposits are the world’s most important source of both Cu and Mo and, espe- cially in the circum-Paci?c region, there are many world-class deposits that exploit these metals. A description of one of the world’s great porphyry Cu–(Mo) deposits, La Escondida in Chile, is pro- vided in Box 2.1. As illustrated in Figure 2.18, both types are associated with the generation of oxidized I-type granite magmas associated with melting processes adjacent to subducted oceanic crust. Porphyry Cu–(Mo) can be explained in terms of a body of magma with a relatively low initial H 2 O content (inherited from the ?uid absent melting of an amphibolitic protolith) rising to high levels in the crust before signi?cant crystallization takes place. It is considered likely that some melt fractions from this high level magma chamber will be tapped off and extrude on the surface. These fractions will crystallize to form volcanic and subvolcanic (porphyry) suites of rocks whose compositions will not be highly differentiated (i.e. granodioritic or rhyodacitic) because of the low degree of fractionation that has taken place prior to extrusion. Because the magma is emplaced at low load pressures the saturation water content will be relatively low and probably not signi?c- antly different from the initial water content. Vapor-saturation will, therefore, occur early in the crystallization sequence, essentially due to “?rst boiling.” Even though Cu is a compatible element in a crystallizing granitic melt (seques- tration of Cu into accessory sul?de phases and biotite result in D Cu crystal/melt > 1), the lack of crys- tallization means that very little of the metal will have been removed from the melt by the time water-saturation occurs. The vapor phase, by contrast, is characterized by high Cl - concentra- tions and it will, therefore, ef?ciently scavenge Cu from the silicate melt. In this setting, however, the situation with respect to Mo is different. Mo is an incom- patible element in a crystallizing granitic melt (D Mo crystal/melt < 1) but, because of the low degree of crystallization, its concentration in the residual magma will not have increased signi?cantly prior to water saturation. On saturation, Mo will partition into the H 2 O-?uid phase but, because its 106 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 106MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 107 partition coef?cient is relatively small and unaf- fected by the Cl - concentration (D Mo ?uid/melt = 2.5) it will never attain very signi?cant concentrations in the ?uid phase. In this setting, therefore, a high level granodioritic I-type magma will exsolve an aqueous ?uid phase that is highly enriched in Cu, but only moderately so in Mo, and form a typical porphyry Cu–(Mo) deposit. A different scenario can be constructed for the formation of Mo–(Cu) porphyry deposits, where it is envisaged that the parental I-type magma might originally have contained a slightly higher initial water content (perhaps due to the ?uid absent melting of a biotite-bearing protolith) than the Cu-dominant situation. As shown in Figure 2.18, this magma would not normally rise to the same shallow crustal levels as its drier equivalent. The saturation water content of the magma is also signi?cantly higher in this situation and, consequently, a greater degree of crystallization needs to take place before water-saturation is achieved (in this case vapor-saturation might occur by second boiling). As the magma cry- stallizes so will Cu be extracted from the melt and end up distributed somewhat evenly through- out the rocks representing the marginal zones of the pluton. This Cu will not, of course, be available for partitioning into the aqueous ?uid once saturation is achieved. Mo, on the other hand, is incompatible and its concentration will continue to increase in the residual melt. When saturation does occur, Mo will be concentrated D crystal/melt = 2 Cu 1C u -(Mo) Cu Mo Mo OH – OH – Mo Cu Cu Mo Surface 2M o -(Cu) 3W -(Mo) Exsolved H 2 O D crystal/melt = 0.05 w D fluid/melt = 1 w Melt Melt Mo Mo OH – Mo Mo Mo Mo Mo W W W W OH – Cu Cu Cu Cu Cu Crystals D crystal/melt = 0.02 Mo D fluid/melt = 2.5 Mo D fluid/melt = 9.1[Cl – ] Cu (2% H 2 O) (3% H 2 O) Subducted oceanic crust Mo I-type magmas (oxidized) S-type magmas (reduced) (8% H 2 O) Figure 2.18 Schematic model for the origin and formation of porphyry-type Cu, Mo, and W deposits. Detailed descriptions are provided in the text (modi?ed after the models of Candela and Holland, 1984; 1986; Strong, 1988; Candela, 1992). ITOC02 09/03/2009 14:37 Page 107even further into the H 2 O-?uid phase because of its favorable partition coef?cient. Despite the fact that Cu will be strongly partitioned into the highly saline H 2 O-?uid, its abundance in the melt is now signi?cantly depleted due to earlier crystal–melt sequestration and the vapor phase will not be signi?cantly concentrated in Cu. The extraction of a melt fraction from a more deep-seated magma chamber will give rise to a pluton that is less likely to reach the surface and whose composition is more highly differentiated due to the greater degree of crystal fractionation prior to extraction. Porphyry-style mineraliza- tion associated with the circulation of a mag- matic hydrothermal ?uid in and around such an intrusion may well have Mo concentrations in excess of Cu, giving rise to Mo–(Cu) porphyry- type mineralization. 2.8.2 The origin of porphyry W-type deposits The above model can also be used to explain some of the characteristics of W-dominated, porphyry- type ore deposits that are associated with many magmatic arc settings. Candela (1992) has empha- sized the relationship that exists between W-rich porphyry-type deposits and the more reduced, ilmenite-bearing, S-type granites that crystallize at relatively deep levels of the Earth’s crust and whose setting is illustrated under (3) in Figure 2.18. Under oxidizing conditions, such as those applic- able to the I-type, magnetite-bearing granites discussed above, the ef?ciency of extracting W from a magmatic-hydrothermal system into an ore body is relatively low and this metal is not a normal component of porphyry Cu–Mo deposits. Under reducing conditions, however, W behaves as an incompatible element in terms of its cry- stal–melt partitioning behavior and its concentra- tion will increase during crystal fractionation. When a magma forms from the partial melting of a metasedimentary precursor, the melt formed will be relatively hydrous and will tend to cry- stallize relatively deep in the crust. Such a melt is also likely to be peraluminous and relatively reduced since it will have equilibrated with meta- sedimentary material that might have contained organic carbon. The saturation water content under these conditions will be high and a signi?cant degree of crystallization will occur before satura- tion is attained (by second boiling). Because W is incompatible under these conditions its concen- tration will rise in the residual melt, whereas Mo concentrations will tend to decrease because it behaves in a more compatible way under reduc- ing conditions. When the H 2 O-?uid phase does exsolve it will interact with a highly different- iated melt that is signi?cantly enriched in W. The ?uid phase will scavenge some W but the rela- tively low partition coef?cient (D W ?uid/melt ? 1) will ensure that most of the concentration is achieved prior to water saturation. The deposits that form in this setting will be associated with deep-level, highly differentiated S-type granites with metal concentrations dominated by W and only minor Mo. 2.9 FLUID FLOW IN AND AROUND GRANITE PLUTONS Discussion has focused in this chapter thus far on the mechanisms by which a magmatic aqueous ?uid phase is formed and its role in the transport of metals derived essentially from the melt. Of equal importance with respect to the formation of magmatic-hydrothermal ore deposits are con- siderations of the ?ow patterns in and around the intrusion that forms the ?uid source, and also the length of time that magmatic-hydrothermal ?uids can remain active subsequent to magma emplacement. Fossilized ?uid ?ow pathways are revealed by recognizing the effects of alteration and mineralization that a hydrothermal solution imposes on the rocks through which it circulates (see Chapter 3). The prediction of where such ?uids have circulated, however, is best achieved by modeling the thermal effects of an evolving ?uid phase in terms of both conductive and convective heat loss in and around a theoretical intrusion. There have been many attempts to model ?uid ?ow around granite plutons and these have proven useful in predicting where optimal ?uid ?ux, and hence mineralization, should be located in relation to the intrusion. Cathles (1981) used a thermal modeling approach to demonstrate that intrusions cool very rapidly, 108 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 108MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 109 at least in terms of a geological time scale. In Figure 2.19a model curves are plotted which relate the time for an intrusion to cool to 25% of its initial temperature, and the size and geometry of the intrusion. The models can be erected for cases where heat is lost either by conduction alone, or by a combination of conduction and convective ?uid ?ow. For the case where a small (1–2 km wide) intrusion cools by conduction of heat to the wall-rocks, solidi?cation will be complete and magmatic-hydrothermal ?uid circulation effect- ively over in about 10 5 –10 6 years. If cooling is ac- companied by convective heat loss caused by the circulation of a hydrothermal ?uid through a per- meable network of fractures (such as would be the case if hydrofracturing had occurred; see section 2.3.3 and Figure 2.7) then such a pluton could be reduced to 25% of its initial temperature within as little as 10 4 years. Thus, in a highly permeable system where ?uid ?ow is maximized by ?uid exsolution and hydrofracturing, heat loss will be very rapid and a single intrusion will not be able to sustain a geologically long-lived period of hydro- thermal circulation. This feature could work 10 7 T 25% (years) 10 1 a (km) 10 6 10 5 10 4 Intrusive geometry 0.0 md 0.10 md 0.25 md 1.0 md Convective cooling Conductive cooling (a) Surface T = 0°C T + 100 000 years ? = 0.5 (c) T + 20 000 years (b) ? = 1 100°C 500°C 2 3 4 3.75 km 2.75 km 2.75 km 200°C 100°C 300°C 600°C 2a 3a Figure 2.19 Models illustrating the thermal and ?uid ?ow characteristics in and around a cooling igneous intrusive body. (a) Plot of the time taken for the temperature of an intrusion to fall to 25% of its initial temperature (T 25% ) as a function of the intrusion dimensions (a in km and de?ned in the inset diagram). Two situations, one where heat is lost by conduction alone (steep slopes) and another where a combination of conduction and convective heat loss occurs (shallow slopes), are shown. Permeabilities (in millidarcy, md) are shown for the two cases (after Cathles, 1981). (b) Thermal model showing ?uid ?ow and temperatures in and around an intrusion some 20 000 years after emplacement. For clarity, streamlines (quanti?ed in terms of the stream function ?) are shown on the left and isotherms on the right hand side, although the model is symmetric about the intrusion. The intrusion was initially at 750 °C with a permeability of 0.07 md and the country rocks at 0 °C with a permeability of 0.13 md. (c) Similar model for the situation 100 000 years after emplacement. Parts b and c are after Norton and Cathles (1979). ITOC02 09/03/2009 14:37 Page 109against the development of a viable ore body, if it is considered that only magmatic ?uids are respons- ible for ore formation. It should, however, be noted that mineralization in and around magmatic intru- sions can also be the result of externally derived ?uids and that these can circulate in response to other factors besides thermal gradients. Norton and Cathles (1979) have been able to model the evolution of hydrothermal ?uid ?ow with time in and around a granite intrusion. Figure 2.19b and c illustrate the isotherms and ?uid ?ow streamlines that are likely to exist around a granite intrusion at two discrete periods (i.e. 2 × 10 4 and 10 5 years after magma emplace- ment and ?uid saturation) for situations where heat is lost by both conduction and convection. Soon after intrusion of even a small granite body (Figure 2.19b) a substantial thermal anomaly is created for up to 2 km around the intrusion and an active system of circulatory ?uid ?ow is set up from both the sides and top of the body. Aqueous solutions establish a convective cell where ?ow is upwards from the intrusion, and circulates back downwards at some distance away. These solutions could incorporate waters from the surrounding country rocks as well as the mag- matic ?uid derived from the cooling pluton itself. An indication of the ?uid ?ux around the intrusion is obtained in the diagrams from the gra- dients of the stream function which de?nes the streamlines (i.e. the closer together the stream- lines, the greater the ?uid ?ux). The demise of the hydrothermal ?uid cell is demonstrated in Figure 2.19c, where it is clear that the thermal anomaly providing the energy for ?uid circulation, as well as the ?uid convection itself, has diminished to insigni?cant proportions within 10 5 years of magma emplacement and ?uid saturation. This again reinforces the view that ore-forming events are likely to be short-lived in and around small granite intrusions representing just a single pulse of magma. Likewise, the ore bodies that form under these sorts of conditions are also likely to be small and perhaps sub-economic. Larger, multi-episodic intrusions are required in order to create a system where a substantial magmatic- hydrothermal ore body will form. A good example of where the nature of ?uid ?ow in and around a granite intrusion contributes greatly to understanding the distribution of mag- matic-hydrothermal mineralization is provided by the polymetallic (Sn–W–Cu–Pb–Zn) deposits of the Cornubian batholith in Cornwall and Devon, southwest England. In this classic mining district mineralized veins occur along the mar- gins of individual granite plutons, but also extend out into the surrounding metasedimentary coun- try rocks. Mineralization is also characterized by a pronounced regional zonation in the distribu- tion of metals. These patterns can be explained in terms of the nature of ?uid ?ow in and around individual granite plutons, as well as the shape of the intrusions and the extent to which they have been exhumed. A description of mineralization and metal zoning in the Cornubian batholith is provided in Box 2.2. 110 PART 1 IGNEOUS PROCESSES The Cornubian granite batholith, emplaced at around 280– 290 Ma during the Carboniferous to Permian Variscan Orogen, is associated with one of the classic polymetallic mining districts of the world. The batholith outcrops as a series of ?ve major plutons, in addition to a host of minor granite bodies and related dykes, in the counties of Cornwall and Devon, southwest England (Figure 1). The batholith is believed to have been emplaced in response to subduction caused by convergence of Laurasia and Africa (the assembly of Pangea), with similar granites and styles of mineralization also evident elsewhere along the orogenic belt, notably in Spain, Portugal, France, and the Czech Republic in Europe, and New Brunswick– Newfoundland in North America. In southwest England mining can be traced back to the Bronze Age, but the peak of activities occurred in the nineteenth century. Historically, the district has been a predominantly Sn (2.5 million tons of cumulative metal production) and Fluid ?ow in and around granite intrusions: polymetallic mineralization associated with the granites of Cornwall, southwest England ITOC02 09/03/2009 14:37 Page 110MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 111 30 0 km 10 20 N St Agnes- Cligga Land's End Cornwall St Austell Carnmenellis Copper Tin Lead-zinc Bodmin Dartmoor Devon Exposed granite Mineralized veins Figure 1 Distribution of the major exposed plutons in Cornwall and Devon making up part of the Cornubian granite batholith, and the distribution of Sn, Cu, and Pb–Zn mineralization in and around these plutons (after Atkinson and Baker, 1986). Cu (2 million tons of cumulative metal production) pro- ducer, although other metals such as Fe, As, Pb, Zn, W, U, and Ag have also been extracted (Alderton, 1993). Presently the district is still a major producer of kaolinite (China clay) for the ceramics industry (see section 4.4.1 of Chapter 4). The Cornish granite, as with many similarly mineralized bodies elsewhere along the Variscan belt, is peralumin- ous in composition, enriched in Rb, Li, F, B, and Be, and can be classi?ed as an S-type granite (Willis-Richards and Jackson, 1989). Mineralization in the granite is related to the early exsolution of a voluminous aqueous phase from the magma which is responsible for widespread alteration of pluton margins and much of the mineraliza- tion (Jackson et al., 1989). Alteration is typically zoned and characterized by the sequence tourmaline – potassic alteration (K-feldspar and biotite) – sericite – chlorite as one progresses away from mineralization. Magmatic- hydrothermal ?uids are also responsible for the formation of mineralized quartz veins that are both endo- and exogranitic. Particularly characteristic of the Cornubian mineralization is the presence of sheeted, greisen-bordered vein sets which generally represent the sites of preferential Sn–W mineralization. These veins consist essentially of quartz and tourmaline with variable cassiterite and wol- framite, and a greisen alteration halo comprising quartz, muscovite, Li-mica, and topaz. These vein systems often extend for several kilometers out into the surrounding country rock and record a well de?ned metallogenic zon- ing or paragenetic sequence. The ore mineral zonation progresses from an early oxide-dominated assemblage (cassiterite and wolframite together with tourmaline) to a later sul?de assemblage comprising chalcopyrite, spha- lerite, and galena, together with ?uorite and chlorite. The zonation pattern is evident both vertically within a single vein system and laterally away from a granite body as shown in Figures 1 and 3. Although metal zonation is complex (Willis-Richards and Jackson, 1989), the progression from Sn–W to Cu to Pb–Zn re?ects an ore ?uid that evolved, in terms of both cooling and chemical changes due to ?uid–?uid and ?uid–rock interactions. Computerized numerical simulations of the ?uid ?ow in and around the Cornubian granites have provided use- ful insights into the distribution of mineralization, as well as the controls on metal zonation that is such of feature of the district. One of the characteristics of the Cornubian ITOC02 09/03/2009 14:37 Page 111batholith is that mineralization is preferentially distributed along one margin of the exposed granite pluton, such as the northern margin of the Carnmenellis pluton and the southern margins of the St Austell and Bodmin plutons (Figure 1). Two of the factors that in?uence the pattern of magmatic-hydrothermal ?uid ?ow in and around granite plutons are the shape of the body and its depth of emplacement (Sams and Thomas-Betts, 1988). Maximum ?uid ?ow tends to be concentrated on the more gently dipping margins of a granite body, which accounts for the asymmetry in distribution of mineralization. In addition, progressive unroo?ng of a granite by erosion will cause the locus of convective ?uid ?ow to migrate outwards into the surrounding sediments. A model for the southeastern margin of the St Agnes–Cligga Head granite (Sams and Thomas-Betts, 1988), much of which is under the sea, illustrates how progressive unroo?ng of the body causes the locus of vertical ?uid ?ow to move progressively away from the pluton margins. Thermal and chemical evolution of the mineralizing solutions over the time interval repre- sented by unroo?ng could account for the successive increments of metal deposition and the regional zonation of metals observed. 20 A –20 40 Fluid flow (x 10 –8 kg m –2 s –1 ) 1 2 3 4 5 6 10 20 30 Distance (km) B Granite Depth (km) B (c) Present surface Coastline Zones Sn Cu Pb-Zn (b) 3 0 km 2 1 Granite (a) Porphyry dyke Mineralized vein (showing dip direction) N St Agnes Head Cligga Head Tin zone Copper zone Lead-zinc zone B Line of section A A Exhumation Original surface Figure 3 The regional pattern of metal zonation around the southeast margin of the St Agnes–Cligga Head granite (a). Numerical simulation of ?uid ?ow in and around the granite as a function of progressive unroo?ng of the granite with time (b and c; both diagrams after Sams and Thomas-Betts, 1988). Figure 2 (left) Sheeted Sn–W-bearing, greisen- bordered veins exposed at Cligga Head, Cornwall. ITOC02 09/03/2009 14:37 Page 112MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 113 2.10 SKARN DEPOSITS The word “skarn” is an old Swedish term that originally referred to the very hard rocks com- posed dominantly of calc–silicate minerals (i.e. Ca-rich garnet, pyroxene, amphibole, and epidote) that identify the rather unusual alteration assem- blages associated with magnetite and chalcopy- rite deposits in that country. It is now widely used to refer to the metasomatic replacement of carbonate rocks (limestone and dolomite) by calc– silicate mineral assemblages during either con- tact or regional metamorphic processes. Mineral deposits associated with skarn assemblages are referred to as skarn deposits, and are typically the product of contact metamorphism and metaso- matism associated with intrusion of granite into carbonate rocks. A wide variety of deposit types and metal associations are grouped into the categ- ory of skarn deposits, and these include W, Sn, Mo, Cu, Fe, Pb–Zn, and Au ores. The different metals found in skarn deposits are a product of the differing compositions, oxidation state, and metallogenic af?nities of the igneous intrusion, as described in Chapter 1 (Einaudi et al., 1981; Misra, 2000). A simple diagram relating granitoid composition to skarn deposit type is shown in Figure 2.20. As a general rule Fe and Au skarn deposits tend to be associated with intrusions of more ma?c to intermediate compositions. Cu, Pb, Zn, and W are linked to calc–alkaline, magnetite- bearing, oxidized (I-type) granitic intrusions, and Mo and Sn with more differentiated granites that might be reduced (S-type) and ilmenite-bearing. It should be noted that there are exceptions to this general trend. Skarn deposits can be classi?ed into calcic or magnesian types, depending largely on whether the host rock is a limestone or dolomite. They are also described as either endo- or exo-skarns, depending on whether the metasomatic assem- blage is internal or external to the intruding plu- ton. Most of the large, economically viable skarn deposits are associated with calcic exoskarns. Tungsten skarns produce the bulk of the world’s W production and are typically associated with intrusion of calc–alkaline intrusions, emplaced relatively deep in the crust. Examples include the King Island mine of Tasmania and the MacTung deposit in the Yukon territory of Canada (see Box 2.3). Copper skarns, by contrast, are often associated with high level porphyry-style intrusions 8 [FeO + Fe 2 O 3 + CaO + Na 2 O]/K 2 O 75 70 60 SiO 2 (wt%) 6 2 65 4 Fe Au Cu Zn W Mo Sn Skarn deposit types Figure 2.20 Plot of SiO 2 versus [FeO + Fe 2 O 3 + CaO + Na 2 O]/K 2 O showing the relationship between the composition of igneous intrusions and the dominant metal in various skarn deposits (after Meinert, 1992). ITOC02 09/03/2009 14:37 Page 113114 PART 1 IGNEOUS PROCESSES The northern Cordillera of Canada, extending from British Columbia into the Yukon Territory, is a highly prospective zone that contains several major porphyry, SEDEX, and VMS base metal mines, as well as gold deposits. In addi- tion, several large W–Cu–(Zn–Mo) skarn deposits occur in the region, including the world’s largest known deposit in the Macmillan Pass area. The deposit, known as MacTung, has yet to be mined but contains ore reserves of some 63 million tons at 0.95 wt% WO 3 , with minor copper (Misra, 2000). The MacTung deposit occurs in Cambrian–Ordovician clastic and carbonate sediments deposited on a con- tinental margin in a platformal shelf setting (Atkinson and Baker, 1986). The sequence is characterized by alternat- ing siltstones, carbonaceous shales, and limestones. It is intruded by the 90 Myr old Cirque Lake stock, a biotite- and garnet-bearing quartz monzonite of probable S-type af?nity. This granite is implicated in the mineralization process, although the actual source of the magmatic- hydrothermal ?uids associated with alteration and metal deposition is believed to be a hidden intrusion at depth to the south of the deposit (Figure 1). Two distinct zones of skarn mineralization are evident at MacTung. The lower zone is hosted within a folded limestone breccia, whereas the upper zone straddles three separate lithological units, namely a lower limestone breccia, an intermediate pelitic unit, and an upper unit comprising alternating shale and limestone. Prograde hydrothermal alteration varies as a function of the host sediment composition and reactivity with ?uids. Non-reactive and less porous shales are char- acterized mainly by quartz veining with narrow bleached halos and low concentrations of scheelite. Highly reactive and porous limestones are the main hosts to skarn mineral- ization, and are pervasively altered. A distinct zonation is evident and alteration progresses from marginal limestone cut by occasional garnet–pyroxene veins, to a pervasively altered intermediate zone comprising extensive lime- stone replacement by garnet–pyroxene skarn, and ?nally a core of pyroxene and pyroxene–pyrrhotite skarn. Ore grades improve progressively from the margins to the core, where WO 3 and Cu concentrations are better than 1.5 and 0.2 wt% respectively (Atkinson and Baker, 1986). Magmatic-hydrothermal ?uids associated with granite intrusions: 2 The MacTung tungsten skarn deposit, Yukon, Canada Limestone/ shale Limestone breccia Hidden granite stock (?) Mineralized envelope Shale Mt Allen Upper ore zone Lower ore zone Limestone Fluid path through pervious, reactive limestone (pervasive skarn) Fluid flow through impervious, non-reactive lithologies (quartz veins) f f Figure 1 Generalized geology of the MacTung deposit and a cross section of the ore zone showing mineralized lithologies and the possible location of a hidden intrusion suggested to be the source of the ore-bearing magmatic-hydrothermal ?uids (after Atkinson and Baker, 1986). ITOC02 09/03/2009 14:37 Page 114MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 115 and many porphyry copper systems that intrude carbonate host rocks have copper skarns associ- ated with them. A classic example is the Bingham district of Utah, USA, which contains not only a huge porphyry Cu deposit, but also the world’s largest Cu skarn deposit. In addition, the non- porphyry ores at Bingham contain economically viable Pb–Zn–Ag ores in limestones that are distal to the copper mineralization (Einaudi, 1982). It is interesting to note that the world’s largest gold mine, at Grasberg in the Ertsberg district of West Papua, also exhibits a porphyry–skarn association and actually produces gold as a by-product of the copper mining process (Meinert, 2000). Gold skarns associated with porphyry Cu mineral- ization are associated with emplacement of high level, oxidized, magnetite-bearing granitoids. Other gold-speci?c skarn deposits, where Au oc- curs in association with a Bi–Te–As metal assem- blage, are linked to more reducing, ilmenite bearing granitoid intrusions (Meinert, 2000). Fe skarns, which occasionally form large, economically viable magnetite deposits, such as at Sverdlovsk and Sarbai in Russia, are associated with more ma?c gabbroic to granodioritic intrusions, and are typi?ed by endoskarn alteration and sodium metasomatism (Einaudi et al., 1981). Tin skarns are generally associated with highly differentiated S-type (or ilmenite-bearing) granitoids, a good example of which is the Renison Bell mine in Tasmania. Even though there are so many different metal associations in skarn deposits, the processes by which they form are similar, namely granitoid emplacement and magmatic-hydrothermal act- ivity, albeit at different levels in the crust. An association with granite intrusion cannot always be demonstrated, but is usually inferred. Skarn deposits typically form as a result of three sequen- tial processes (Einaudi et al., 1981; Meinert, 1992). These are isochemical contact metamor- phism during early stages of pluton emplacement and crystallization, followed by open system metasomatism and alteration during magmatic ?uid saturation, and, ?nally, draw-down and mix- ing with meteoric ?uids (for a de?nition see Chapter 3) during cooling of the pluton. Prograde – isochemical contact metamorphism (Figure 2.21a) As the granite pluton intrudes the country rock sediments are subjected to contact metamor- phism and the formation of a variety of hornfelsic textures. The mineral assemblages that form at this stage re?ect the composition of the litho- types within which they form. Contact metamor- phism is largely a thermal effect although ?uids Retrograde alteration is minimal at MacTung and is charac- terized by late quartz and calcite veins with coarse grained scheelite, and minor amphibole formation representing hydration of the pyroxene skarn. The pattern of alteration zonation, which is discordant to the Cirque Lake stock contact, as well as the distribu- tion of quartz veins, suggest that ?uids migrated up-dip and toward the stock. For these reasons the source of magmatic-hydrothermal ?uids involved in skarn forma- tion is believed to be a hidden intrusion at depth to the south of the mineralization, as suggested in Figure 1. It is not known whether such an intrusion is merely part of the Cirque Lake stock or an unrelated event. Figure 2 The Cirque Lake stock intruding Cambrian–Ordovician clastic and carbonate sediments in the vicinity of the MacTung skarn deposit. ITOC02 09/03/2009 14:37 Page 115are likely to circulate during this process and are largely a product of prograde metamorphic reactions and, therefore, comprise mainly H 2 O and CO 2 (see Chapter 3 for a description of meta- morphic ?uids). In dolomitic units metamorphic mineral zonation approximates the sequence garnet–clinopyroxene–tremolite–talc/phlogopite, re?ecting increased distance and progressively more hydrous assemblages away from the intru- sion. In limestone units the mineral zonation is garnet–vesuvianite +wollastinite–marble. There is no mineralization associated with this stage, although the process of dehydration close to the pluton margins may be important for increasing porosity of the source rocks and facilitating ?uid ?ow during later episodes of mineralization. Prograde – metasomatism and replacement (Figure 2.21a) The second stage in the formation of skarn deposits involves H 2 O-?uid and vapor-saturation of the intruding magma (as a function of either ?rst or second boiling or both) and the egress of the ?uid phase into the surrounding contact metamorphic halo. At deeper crustal levels ?uid 116 PART 1 IGNEOUS PROCESSES (a) Carbonate Exoskarn (metasomatic) Stage 2 Magmatic fluids PROGRADE: Stage 1 Isochemical (contact metamorphism) Stage 2 Metasomatic (magmatic fluid exsolution) (b) RETROGRADE: Stage 3 Meteoric fluid influx, metal precipitation Isochemical alteration Crystallizing granite intrusion Stage 1 Stage 3 Descending meteoric waters Figure 2.21 The evolution of intrusion-related skarn deposits showing the three sequential stages of formation. (a) Prograde stages, and (b) retrograde stage (modi?ed after Corbett and Leach, 1998). ITOC02 09/03/2009 14:37 Page 116MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 117 ?ow is likely to be concentrated along discrete structural or bedding parallel conduits, whereas at higher levels ?uid ?ow will be more pervasively distributed, perhaps due to hydrofracturing, in a broad halo in and around the granite cupola. Metasomatic mineral assemblages will be similar to those formed during contact metamorphism but alteration will be more pervasive and coarser- grained, and will replace earlier formed assem- blages. Si, Al, and Fe, as well as other components, will be introduced into the calcareous sediments by the aqueous magmatic ?uid, while Ca, Mg, and CO 2 are locally derived and also introduced into the metasomatic system. Sul?de mineralization does not form at this stage, although magnetite and scheelite (in W skarns) do precipitate in the waning stages of prograde metasomatism. Retrograde – meteoric ?uid in?ux and main metal precipitation (Figure 2.21b) All magmatic-hydrothermal systems undergo progressive cooling and decay of the high temper- ature magmatic ?uid system. As ?uids become progressively dominated by shallow meteoric waters, a series of complex retrograde reactions takes place, as well as the precipitation of the main stages of base and precious metal, sul?de- related mineralization (Einaudi et al., 1981). The retrograde alteration assemblages are superim- posed onto earlier metamorphic and metasomatic minerals and, typically, this process is recognized by paragenetically late formation of epidote, biotite, chlorite, plagioclase, calcite, quartz (all after various garnet types), tremolite–actinolite, and talc (after pyroxenes) and serpentine (after olivine). Sul?de ore minerals, as well as mag- netite and hematite, occur as disseminations, or veins, that cut across prograde assemblages. Assemblages such as pyrite–chalcopyrite–mag- netite characterize proximal settings, whereas bornite and sphalerite–galena are typically more distal in occurrence. The paragenetically late pre- cipitation of most skarn-related ores suggests that metal precipitation is related to decreasing tem- perature of the ore ?uids (and a resulting drop in solubilities), ?uid mixing, or neutralization of the ore ?uid by reaction with carbonate lithologies. Mixing of the magmatic ore ?uid with the late meteoric component, and related redox reactions in the ?uid, may be additional controls on the ore formation process. A more detailed description of precipitation mechanisms in hydrothermal solu- tions is provided in Chapter 3. 2.11 NEAR-SURFACE MAGMATIC-HYDROTHERMAL PROCESSES – THE “EPITHERMAL” FAMILY OF AU–AG–(CU) DEPOSITS Exploration for gold in the circum-Paci?c region, especially since the 1970s, has led to the discov- ery of a large number of world-class gold deposits associated with either active or geologically re- cent volcanic environments. These deposits are now regarded as an important and very prospect- ive category of gold deposit type, termed epither- mal deposits. The term “epithermal” is derived from Lindgren’s (1933) classi?cation of ore de- posits and refers to those that formed at shallow crustal levels (i.e. the epizone). Many studies of this ore-forming environment, and in particular comparisons with active, modern analogues such as the Taupo volcanic zone on the north island of New Zealand, have shown that epithermal deposits typically form at temperatures between 160 and 270°C and pressures equivalent to depths of between 50 and 1000 m (Cooke and Simmons, 2000; Hedenquist et al., 2000). There are two contrasting styles of mineraliza- tion that are now recognized in epithermal depos- its, and these are referred to as high-sul?dation and low-sul?dation types. These terms refer spe- ci?cally to the oxidation state of sulfur in the ore ?uid, the chemistry and pH of which also relates to the nature of alteration associated with each type. Unfortunately, the geological literature con- tains a confusing plethora of synonyms for these two types. Table 2.2 attempts to summarize these terms and also illustrates the main character- istics of each. Note that the two terms do not relate to the abundance of sulfur, as this is highly variable in each deposit type, but in some pub- lications “high sulfur” and “low sulfur” have been equated with high-sul?dation and low-sul?dation respectively. ITOC02 09/03/2009 14:37 Page 117High- and low-sul?dation epithermal deposits can be viewed as end-members of processes re- lated to ?uid evolution and circulation in and around volcanoes. High-sul?dation deposits occur in proximal settings and are commonly found within or close to the volcanic vent itself. The ?uids involved with mineralization are derived directly from the magma as a product of vapor- and ?uid-saturation and are usually boiling in the ore-forming environment. The ?uids are very acidic (pH of 1–3) and oxidized, carrying the oxid- ized S 4+ or S 6+ species as SO 2 , SO 4 2- , or HSO 4 - in solution. As this ?uid boils and SO 2 and CO 2 are partitioned into the vapor phase, the remain- ing liquid carries a surplus of H + which makes it very acidic (pH = 1; Hedenquist et al., 2000). This acidic ?uid is also capable of leaching most of the major elements from the host volcanic or volcano-sedimentary rocks through which it circulates, resulting in vuggy textures and an advanced argillic style of alteration (see Table 2.2 and also the more detailed account of hydrother- mal alteration in Chapter 3). By contrast, low- sul?dation deposits are associated with ?uids that are similar to those involved with hotsprings and other geothermal manifestations in areas of enhanced heat ?ow. These ?uids have equilib- rated with their host rocks and generally com- prise a dominantly meteoric component, although it is likely that this will have been mixed with an evolved magmatic ?uid if active volcanism is located nearby. Consequently, low-sul?dation deposits may form within the volcanic edi?ce, especially during the waning stages of magmatic activity when draw-down of meteoric ?uids is perhaps more likely. More typically they may form at locations that are somewhat removed from the focus of volcanism. The ?uid involved is near- neutral and has low salinities, but as with high- sul?dation environments, it is also likely to have boiled in and around the zone of ore formation. The characteristics of high- and low-sul?dation deposits are shown schematically in Figure 2.22. This diagram suggests that a spatial and genetic link exists between the two types, but it should be emphasized that this may not be the case. Many gold districts contain either low-sul?dation epi- thermal deposits (such as the major deposits in Nevada, USA, including Round Mountain, Com- stock Lode, Midas, and Sleeper) or high-sul?dation deposits (such as many of the Andean deposits, in- cluding Yanacocha, Pierina, and El Indio–Tambo). There is, however, an increasing number of cases where spatial and genetic links are evident. One area where both high- and low-sul?dation epither- mal deposits occur in fairly close proximity to one another is in Kyushu, the southernmost island of Japan (Box 2.4). It is evident in Figure 2.22 that there should also be a genetic link between por- phyry Cu–(Mo–Au) deposits formed in the sub- 118 PART 1 IGNEOUS PROCESSES Table 2.2 Characteristics of high- and low-sul?dation epithermal deposits High-sul?dation Oxidized sulfur species (SO 2 , SO 4 2- , HSO 4 - ) in ore ?uid/vapor Also referred to as Gold–alunite, acid–sulfate, alunite–kaolinite Fluids Acidic pH, probably saline initially, dominantly magmatic Alteration assemblage Advanced argillic (zonation: quartz–alunite–kaolinite– illite–montmorillonite–chlorite) Metal associations Au–Cu (lesser Ag, Bi, Te) Low-sul?dation Reduced sulfur species (HS - , H 2 S) in ore ?uid/vapor Adularia–sericite, hotspring-related Near-neutral pH, low salinity, gas-rich (CO 2 , H 2 S), dominantly meteoric Adularia–sericite (zonation: quartz/chalcedony–calcite– adularia–sericite–chlorite) Au–Ag (lesser As, Sb, Se, Hg) ITOC02 09/03/2009 14:37 Page 118MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 119 volcanic environment and high-sul?dation Au–Cu deposits at surface. This link can be demonstrated in the case of the Lepanto epithermal Au–Cu and the Far Southeast porphyry Cu–Au deposits on Luzon in the Phillipines. These two deposits are adjacent to one another in late Pliocene rocks, with the mineralization in both occurring over a brief 300 000 year time period at around 1.3 Ma (Arribas et al., 1995). It is also pertin-ent to note that low- sul?dation Au–Ag mineraliza-tion is located with- in a few kilometers of the Lepanto–Far Southeast system, indicating that spatial and genetic links between high- and low-sul?dation epithermal and porphyry styles of min-eralization are likely. The fact that their coexistence is only rarely docu- mented is, at least partially, therefore, a function of erosion and lack of preservation. 2.11.1 Gold precipitation mechanisms in epithermal deposits The detailed mechanisms of gold precipitation in high- and low-sul?dation deposits are complex and vary depending on the geological setting and nature of ?uids involved (Cooke and Simmons, 2000). The following section brie?y considers gold precipitation mechanisms as applicable to epithermal deposits, although the principles dis- cussed here also apply to other gold deposit types. The reader is encouraged to read the relevant sections on metal solubility and precipitation mechanisms in Chapter 3 (sections 3.4 and 3.5) before continuing with this section. Anionic species, or ligands, of chloride (Cl - ) and sul?de (speci?cally HS - ) are regarded as being Boiling Meteoric water CO 2 , H 2 S Acid springs solfataras Magmatic-hydrothermal system Volcano CO 2 , SO 2 , HCI Crater lake 200° Low sulfidation (Au, Ag) –Neutral fluids –Adularia, sericite alteration Fluid neutralization Saline magmatic fluid Liquid flow Vapor flow 300° 300° High sulfidation (Au, Cu) –acidic fluids –advanced argillic alteration Hotsprings, mud pools fumaroles Geothermal system 600° Porphyry Cu, (Mo, Au) 600° Figure 2.22 The geological setting and characteristics of high-sul?dation and low-sul?dation epithermal deposits. A genetic link between high-sul?dation epithermal Au–Cu and sub-volcanic porphyry type Cu–Au deposits is also suggested (after Hedenquist et al., 2000). ITOC02 09/03/2009 14:37 Page 119120 PART 1 IGNEOUS PROCESSES Kyushu, the southernmost major island of the Japanese arc, is the country’s principal gold producing region. Late Cenozoic volcanic activity in this region has given rise to several epithermal Au–Ag deposits, all related to either extinct or waning magmatic-hydrothermal systems aris- ing from this volcanism. Both styles of epithermal gold mineralization are preserved on Kyushu, namely high- sul?dation (or acid-sulfate) mineralization represented by the Nansatsu type ores of which Kasuga is an example, and low-sul?dation (or adularia–sericite) mineralization represented by the very rich Hishikari deposit. Kasuga The Nansatsu district in the southern portion of Kyushu has been the site of calc–alkaline volcanism for the past 10 million years. Several deposits occur in the district, including Akeshi, Iwato, and Kasuga. Kasuga is a small deposit that produces about 120 000 tons of ore annually at an average Au grade of about 3 g t -1 . The ore body is associated with a residual high-silica zone in andesites of the Nansatsu Group, which overlies basement metasedi- ments of the Cretaceous Shimanto Supergroup (Hedenquist et al., 1994). Alteration in the open pit is characterized by a central quartz-rich zone in which the volcanic host rock has been almost entirely leached of all elements except Si (Figure 1a). The quartz body is surrounded by a zone of advanced argillic alteration (comprising alunite, dickite, and kaolinite), which in turn grades out into a propylitic zone (chlorite + illite). Ore minerals consist mainly of pyrite Magmatic-hydrothermal processes in volcanic environments: the Kasuga and Hishikari epithermal Au–Ag deposits, Kyushu, Japan Gravels Se-Sm Pr Kasuga open pit Al Dt-Se Py Quaternary pyroclastic flows Pr Altered Nansatsu Group volcanics Shimanto Supergroup (basement) Quartz body Alunite Sericite-smectite Dickite-sericite-pyrite Sericite-chlorite Propylitic Al Se-Sm Dt-Se-Py Se-Ch Pr 100 0 m Dt-Se-Py Se-Ch Alteration: Se-Ch 400 200 0 –200 (meters) Recent pyroclastic flows Hishikari andesites (0.98–1.62 Ma) Yamada Zone (0.60–1.15 Ma) Shishimano dacites/mudstones etc. Honko-Sajin Zone (0.66–1.10 Ma) 500 0 m Shimanto Supergroup (65–70 Ma) Figure 1 (a) Cross section through the Kasuga deposit showing the nature and geometry of the alteration zoning (after Hedenquist et al., 1994). (b) Cross section through the Hishikari deposits illustrating the concentration of mineralization at one particular elevation (after Izawa et al., 2001). (b) (a) ITOC02 09/03/2009 14:37 Page 120MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 121 essential in order to solubilize gold in aqueous solutions (Seward and Barnes, 1997). In reduced, near-neutral pH, aqueous solutions Au is likely to be transported as the Au(HS) 2 - complex and this is also likely to be the preferred medium for the movement of gold in low-sul?dation environments. By contrast, at higher temperatures (>300 °C) and for solutions that are both more acidic and saline, gold is preferentially transported as the Au(Cl) 2 - complex, and this mode of transport probably applies to high-sul?dation environments. Since the mechanism whereby gold is transported could be fundamentally different in the two epithermal deposit types, it follows that the chemical and physical controls that precipitate the gold from the hydrothermal ?uids are also likely to differ. For low-sul?dation deposits gold precipitation is relatively straightforward and is linked to one, or both, of the two processes that characterize ?uid evolution in this environment, namely boil- ing and ?uid mixing. Boiling of an ore ?uid in this case will result in loss of H 2 S to the vapor phase, which causes destabilization of the Au(HS) - 2 complex and precipitation of Au, as described in equation [2.7] (after Cooke and Simmons, 2000): Au(HS) - 2 + H + + 0.5H 2 - Au + 2H 2 S [2.7] Mixing of an oxidized meteoric water with the same ore ?uid would also have the effect of pre- cipitating gold, as shown in equation [2.8]: Figure 2 Gold and silver rich Hosen vein photographed underground at the Hishikari mine, Kyushu. the zonation in the alteration halo and the late oxidation of pre-existing assemblages (Hedenquist et al., 1994). Hishikari The Hishikari deposit, discovered in 1981, is a very rich low-sul?dation epithermal deposit containing about 260 tons of mineable gold (Izawa et al., 2001). It is located some 20km to the northwest of the recently active Kirishima volcano in south-central Kyushu. The deposit occurs as a series of vertical, en-echelon veins largely within the Cretaceous Shimanto Supergroup basement, but also extending short distances into overlying Pleistocene andesites. The veins consist mainly of quartz and adularia, with lesser smectite clay. Vein formation and mineraliza- tion is believed to be linked to volcanism that started about 1.6 Ma and terminated at about 0.7 Ma. Vein related adularia, however, has been dated at between 1.15 and 0.60Ma, suggesting that the geothermal system and mineralization commenced about 0.5 Myr after volcanism (Izawa et al., 2001). The spectacularly rich Hosen No. 1 vein (which averages over 3000 g t -1 Au and 2000 g t -1 Ag!) records evidence of three phases of vein deposition. Each consists mainly of quartz, adularia, and electrum. Minor pyrite and chalcopyrite also occur but only during the early phases of deposition. The high grade ores at Hishikari are restricted to a speci?c elevation, extending over about 200 m of vertical extent (Figure 1b), and veins either ter- minate or become sub-economic above and below this range. This, together with ?uid inclusion evidence, suggests that boiling of a dominantly meteoric ?uid was responsible for the precipitation of gold and silver from solution. and enargite with later native sulfur, covellite, and goethite. The Au–Ag (electrum) is largely contained within the quartz body in the center of the alteration halo, although considerable remobilization into late oxidized phases also occurs. Mineralization is linked to the exsolution of a metal charged ?uid phase from the host andesite, with subsequent segregation into vapor and liquid phases. Highly acidic vapors are implicated in the leaching of volcanic rocks to form the porous quartz body. The metal charged ?uid phase subsequently percolated through the porous rock and precipitated Cu, Ag, and Au. Progressive mixing of this ?uid with meteoric waters was probably the main precipitation mechanism, and was also responsible for ITOC02 09/03/2009 14:37 Page 121Au(HS) 2 - + 8H 2 O - Au + 2SO 4 2- + 3H + + 7.5H 2 [2.8] Evidence for boiling-induced Au precipitation is provided from modern geothermal systems that exploit steam to drive electricity-generating tur- bines, such as at Broadlands in New Zealand (Cooke and Simmons, 2000). It is well known in these power stations that the siliceous scale that plates the inside of pipes and accompanies the ?ashing of water to steam is often enriched in both Au and Ag. In actual low-sul?dation deposits the narrow vertical interval over which vein-hosted mineralization occurs (such as at Hishikari; Box 2.4) is another indication that boil- ing acts as a fundamental control on ore precipita- tion. Although it is clear that ?uid mixing also plays a role, in deposits such as Creede in Colorado, the evidence for a widespread role for this mechan- ism is less clear (Cooke and Simmons, 2000). For high-sul?dation deposits the Au deposition mechanisms are less well understood and more complex, since it is feasible to transport gold, not only as a Au(Cl) 2 - complex, as suggested above, but also as a bisul?de complex (Au(HS)) in ?uids with a high sulfur activity. It is also considered possible that Au is transported together with Cu in the vapor phase (see section 2.4.2 above). If gold is trans- ported as a bisul?de complex in high-sul?dation environments then the precipitation mechanisms are also likely to be related to boiling and ?uid mixing, as for low-sul?dation systems. If, on the other hand, gold is transported as a chloride com- plex, then boiling and extraction of oxidized sulfur species (SO 2 or SO 4 2- ) into the vapor phase will have little effect on its stability. Fluid mixing, between a hot, acidic and saline Au(Cl) 2 - bearing ore ?uid and a cooler, neutral meteoric solution, could, however, be an important precipitation mechanism, since it will have the effect of in- creasing pH and decreasing salinity (by dilution) of the ore ?uid. Studies at Lepanto, for example, indicate that ?uid mixing did occur with varia- tions in the ratio of magmatic to meteoric ?uids ranging from 9:1 to 1:1 over the ore deposit (Hedenquist et al., 1998). Arribas et al. (1995) have provided a model for the formation of high-sul?dation epithermal systems that re?ects the complexities related to possible variations of gold speciation in this envir- onment. The model envisages two stages of ore formation. The ?rst involves degassing of hot, acid rich magmatic vapors that are responsible for intense leaching of the host rock to form the porous vuggy quartz zone in the fumarolic conduit and the advanced argillic alteration halo around it (Figure 2.23a). The vapor phases could also mix with meteoric waters to form an acid sulfate ?uid that has a low Au solubility, but is also implicated in the alteration process. The secondary porosity and permeability created during this alteration process is considered to be a necessary preparatory stage for the in?ux of later metal-bearing ?uids. Subsequent ore deposition can occur in one of two ways. A hot, acidic, and saline ore ?uid, carrying gold as a Au(Cl) 2 - complex, could be derived directly from the subjacent magma and move directly up into the alteration zone. Ore precipitation would occur as a result of mixing and dilution of this ?uid by cooler meteoric waters (Figure 2.23b). Alternatively, it is suggested that Cu and Au are initially removed from the magma in the vapor phase and that these metal charged gases mix with heated ground waters circulating around the intrusion to form a low salinity ?uid in which gold is transported as Au(HS). This ?uid would then precipitate metals by boiling in the near surface environment, or mixing with meteoric waters, or both (Figure 2.23b). 2.12 THE ROLE OF HYDROTHERMAL FLUIDS IN MINERALIZED MAFIC ROCKS Much of this chapter has concentrated on the relationships between felsic magmas and the ?uids that exsolve from them. As mentioned at the beginning of this chapter, however, it is also feasible for a ma?c melt to exsolve a mag- matic ?uid phase, although the resultant mass fraction of exsolved water may be lower than in felsic melts. In ma?c rocks the role of a magmatic ?uid in ore-forming processes is commonly over- looked, even though it is now apparent that such ?uids may be important in the concentra- tion of metals such as Cu, Ni, and the platinum 122 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 122MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 123 group elements (PGE) in layered ma?c intrusions (Mathez, 1989a, b). The solubility of H 2 O in a melt is higher than that of CO 2 and it follows, therefore, that the latter will exsolve before the aqueous species. The ?rst ?uids to form from a crystallizing ma?c melt are rich in CO 2 (as well as CO in more reduced environments), but they will evolve to more H 2 O- rich compositions as crystallization progresses. In addition the ?uid will also contain appreciable chlorine (as HCl or FeCl 2 ) and sulfur (as HS - or SO 4 2- depending on fO 2 ). The change in ?uid com- position from CO 2 to H 2 O-dominated occurs quite rapidly in ?uids derived from ma?c melts and is marked by the precipitation of C (graphite) from the ?uid. These ?uids are, therefore, unlikely, at least initially, to be able to dissolve much metal, but this would change as the proportion of water in the solutions increases. Considerations of metal concentration in ma?c rocks, such as those discussed at some length in Chapter 1, assume that distribution patterns of Cu, Ni, and PGE re?ect crystal–melt or melt– melt partitioning during progressive crystallization or silicate–sul?de immiscibility. These assump- tions presuppose that the ore deposits observed have not re-equilibrated with a vapor phase and that the metal distribution patterns are genuinely magmatic and not overprinted by hydrothermal processes. This is not always the case and min- eralogical observations in serpentinites from eastern Australia, for example, show that certain of the PGE (in particular Pd, Pt, and Rh) tend to be readily redistributed by hydrothermal solutions, whereas others (speci?cally Os, Ir, and Ru) remain unaffected by such processes (Yang and Seccombe, 1994). The rare cases where nuggets, overgrowths and dendrites of PGE are found in placer de- posits, and in laterite enrichments, also indicate that under certain conditions these metals are labile and can be put into solution even at low temperatures (see Chapter 4). In the case of the PGE it is apparent, with the possible exception of palladium (Pd), that these metals tend to exhibit low solubilities in mag- matic-hydrothermal ?uids at high temperatures and under fairly reduced (low fO 2 ) conditions. In the case of Pd it appears that this metal is fairly soluble in saline solutions at elevated temper- atures (Sassani and Shock, 1990). In general, how- ever, the presence of ligands such as Cl - enhances PGE solubilities only slightly and signi?cant Magmatic vapor (SO 2 , HCI) 0 (a) 1 2 km Alteration Cool meteoric water Magmatic brine 300°C 400°C 500°C Vuggy quartz Alunite Kaolinite Sericite Potassic Fumaroles Alteration (b1) (b2) Ore deposition Ore deposition by mixing with shallow meteoric water Acid magmatic brine transports gold as AuCl 2 – Mixing produces reduced, acid, low salinity, water carrying Au as AuHS Ores Alteration envelope Gas phase metal transport (as AuS, CuS) Hot ground water Mixing Figure 2.23 Two stage model for the formation of high-sul?dation epithermal deposits (after Arribas et al., 1995). (a) Initial stage where a dominantly magmatic vapor phase is responsible for leaching of the country rock and development of an advanced argillic alteration halo around the main fumarolic conduit. (b1) Ore deposition stage, in this case where gold is transported as a chloride complex; and (b2) ore deposition stage where gold is transported as a bisul?de complex. ITOC02 09/03/2009 14:37 Page 123dissolution of these metals is only feasible under highly oxidized, acidic conditions (Wood et al., 1992). Hydroxide (OH - ) complexation is also regarded as unlikely to contribute toward PGE solubility in hydrothermal solutions, although it may be important in sur?cial waters and PGE– OH - complexes may be implicated in placer and laterite deposits. Likewise, bisul?de (HS - ) com- plexes result in very low (ppb) quantities of Pt and Pd being transported in most geologically pertinent solutions. These observations tend to support the conventional view that metal concen- trations in ma?c rocks have not been markedly modi?ed by hydrothermal processes. However, under conditions of very low fS 2 (where the metal and not the metal sul?de is stable) it appears that PGE solubilities in a highly saline brine might be high (10 2 –10 3 ppb), mainly because PGE complex- ation with Cl - is not detrimentally affected by the presence of sul?de ligands. Under such condi- tions, a ?uid could play an important role in the redistribution of these metals. 2.12.1 The effects of a magmatic ?uid on PGE mineralization in the Bushveld Complex In the Bushveld Complex the grades of platinum group element (PGE) mineralization and the relative proportions of these elements remain uniform over tens of kilometers of strike, in both the Merensky Reef and the UG2 chromitite seam (Cawthorn et al., 2002). By contrast the mineral- ogy of the PGE varies markedly. The normal strat- iform reef is generally dominated by PGE sul?de minerals such as cooperite, braggite, and laurite. Atypical situations, represented by mineralized potholes (see Box 1.6) and unusual discordant Pt-rich dunite pipes, are characterized by PGE–Fe alloys and tellurides. The general consensus is that the sul?de-dominated mineralogy of normal reefs is the product of magmatic processes, as dis- cussed in Chapter 1, but that the sulfur-de?cient PGE mineralogy in potholes and pipes re?ects areas where a volatile-rich ?uid phase has inter- acted with the magmatic ores, resulting in local- ized low fS 2 conditions. The Bushveld Complex contains several un- usual rock types such as iron-rich ultrama?c pipes, as well as iron-rich pegmatites (with concentra- tions of Pb, As, Sb, and Bi) and plagioclase– amphibole–phlogopite veins, that have been used as evidence for vapor- or ?uid-saturation in the late stages of crystallization of the Bushveld mag- mas. Schiffries (1982) regarded the Fe- and Pt-rich dunite pipes as metasomatic in origin and implic- ated an aqueous magmatic brine that reacted with the country rocks at around 600 °C and 3.5 kbar. More recent work has indicated that these bod- ies are probably discrete magmatic intrusions, but that they did act as channelways for later, lower temperature hydrothermal ?uids (Cawthorn et al., 2000). The sul?de-poor, Pt-dominated min- eralogy of these bodies is considered to represent re-equilibration of an original magmatic sul?de assemblage by later hydrothermal ?uids. One of the still contentious issues concerning the Bushveld Complex, however, is the extent to which the major mineralized horizons, such as the Merensky Reef and the UG2, owe their PGE and base metal mineralization to hydrothermal rather than magmatic processes. Detailed minera- logical studies have shown that in certain environ- ments magmatic ?uids must have played a role in redistribution of sulfur and recrystallization of the PGE. Where the UG2 chromitite horizon is cut by a dunite pipe, for example, it contains the same distinctive Pt–Fe alloy and Pt–arsenide (sperrylite) phases as the pipe itself. By contrast, the UG2 horizon well removed from such bodies comprises the normal, magmatic PGE–sul?de min- erals such as braggite and cooperite (Peyerl, 1982). In addition, the Merensky Reef itself often exhibits the same sort of variations in PGE mineralogy as does the UG2, but in this case the occurrence of sul?de-de?cient PGE minerals is usually related to the formation of “potholes” and not necessarily to cross-cutting dunite re- placement pipes (Figure 2.24). The Merensky Reef potholes were described in Chapter 1 (see Box 1.6) and are attributed to syn-magmatic faulting of the footwall cumulates just prior to injection of a new magma pulse, with accompanying metasom- atism of rocks within the pothole structure by hydrothermal ?uid. The occurrence of desul?d- ized PGE mineral assemblages in the potholes is consistent both with this process and with the 124 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 124MAGMATIC-HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 2 125 fact that potholes were probably also the sites of ?uid-saturation and enhanced circulation. The presence of saline, but relatively oxidized, ?uids is considered to have been responsible for modify- ing the primary magmatic ore mineralogy, desta- bilizing the dominant base metal and PGE sul?de phases, with the resultant loss of sulfur and reac- tion between PGE and Fe to form alloys or tel- luride and arsenide minerals (Kinloch, 1982). It is important to note that, during the hydrothermal overprint, PGE and base metals do not appear to be redistributed and the metal budget, therefore, remains constant (Cawthorn et al., 2002). There seems little evidence for a solely hydrothermal origin for the base and precious metal ores of the All magmas contain the constituents that, on crystallization, combine to exsolve discrete ?uid and vapor phases. Most magmas will exsolve sub- stantial quantities (up to several wt%) of water, as well as an order of magnitude or so less of carbon dioxide, and these are the two dominant mag- matic-hydrothermal ?uids. Water in particular has the ability to dissolve signi?cant quantities of Merensky Reef, or any other major mineralized part of the Bushveld Complex, although it is clear that magmatic ?uids have reacted extensively with the rocks of this layered ma?c intrusion. In summary, it would appear that in contrast to granitoid-related systems, magmatic ?uids in ma?c melts have played a relatively small role in the mineralization processes that accompany the crystallization of these rock types. Vapor- saturation does occur but its in?uence as a miner- alizing agent would generally appear to be limited to re-equilibration of existing magmatic PGE and base metal sul?des, in rocks where metasomat- ism is clearly demonstratable, such as potholes, Fe-rich dunite pipes, pegmatites, and veins. Pothole depression Pothole reef Pt-Fe alloys + base metal sul?des 89.0 Pt-Pd sul?des Laurite Ru(Os,Ir)S 2 Tellurides and Cu–Au alloy 0.1 0.1 10.8 11.5 0 87.0 1.5 0 92.6 0.4 7.0 Normal Contact Pothole Normal reef Contact reef Zone of fluid–rock interaction Figure 2.24 Schematic illustration showing the nature of “potholes” in the Merensky Reef of the Bushveld Complex and the distinctive mineral assemblages of these features relative to normal reef (after Kinloch, 1982; Mathez, 1989b). ITOC02 09/03/2009 14:37 Page 125anionic substances, in particular Cl - , which in turn promotes the dissolution of other alkali and transition metal cations. The magmatic aqueous phase can exist as a liquid, vapor, or homogeneous supercritical ?uid. The process of H 2 O-saturation can be achieved in two ways, either by decreasing the pressure of the system (called ?rst boiling) or by progressive crystallization of magma (second boiling). H 2 O-saturation is particularly relevant to ore- forming processes during the emplacement and crystallization of granitic magmas at moderate to shallow crustal levels. This environment gives rise to the formation of a wide variety of import- ant ore deposit types including porphyry Cu and Mo deposits, polymetallic skarn ores, granite- related Sn–W deposits, and the family of volcanic- related epithermal Au–Ag–(Cu) deposit types. Many metals will partition strongly into the liq- uid or vapor that forms on H 2 O-saturation and, in such cases, mineralization accompanies the altera- tion of host rocks, both within and external to the intrusion. The formation of either Cu-dominant or Mo-dominant porphyry deposits re?ects a For those readers wishing to read more about magmatic- hydrothermal ore-forming processes, the following references to books and journal special issues will be useful. Corbett, G.J. and Leach, T.M. (1998) Southwest Paci?c Rim Gold–Copper Systems: Structure, Alteration and Mineralization. Special Publication 6. El Paso, TX: Society of Economic Geologists, 237 pp. Geochemistry of Hydrothermal Ore Deposits, ed. H.L. Barnes. 1st edn, 1967. Holt, Rinehart and Winston Inc., pp. 34–76. 2nd edn 1979. John Wiley and Sons, pp. 71–136. 3rd edn, 1997. John Wiley and Sons, pp. 63–124 and 737–96. Misra, K.C. (2000) Understanding Mineral Deposits. Dordrecht: Kluwer Academic Publishers, 845 pp. Pirajno, F. (1992) Hydrothermal Mineral Deposits. New York: Springer-Verlag, 709 pp. Reviews in Economic Geology, Volume 2: Berger, B.R. and Bethke, B.M. (eds) (1985) Geology and Geochem- subtle interplay between the depth of intrusion of a granitic body (itself a function of the original water content), the timing of H 2 O-saturation rel- ative to the progress of crystallization, and the behavior of metals during melt–?uid partitioning. Egress of ?uids and vapor from the magma and their subsequent circulation are dependent on the permeability of the surrounds, and may be modi- ?ed by boiling-related hydrofracturing. Polymetal- lic skarn deposits re?ect the interaction between the exsolved magmatic ?uids from different types of granite and calcareous sediments. H 2 O- saturation and boiling in volcanic environments, producing signi?cant volumes of volatile-rich vapor, is conducive to the formation of epithermal deposits. High- and low-sul?dation epithermal deposits re?ect end-members in a continuum of magmatic-hydrothermal processes that progress- ively incorporate more non-magmatic waters as the volcanic system wanes, or as one moves away from the volcanic center. Many ore deposit types are the product of ?uids that are unrelated to a magmatic source and these are the subject of Chapter 3. istry of Epithermal Systems. El Paso, TX: Society of Economic Geologists, 298 pp. Reviews in Economic Geology, Volume 4: Whitney, J.A. and Naldrett, A.J. (eds) (1989) Ore Deposition Associated with Magmas. El Paso, TX: Society of Economic Geologists, 250 pp. Reviews in Economic Geology, Volume 13: Hagemann, S.G. and Brown, P.E. (eds) (2000) Gold in 2000. El Paso, TX: Society of Economic Geologists, 559 pp. Seltmann, R., Lehmann, B., Lowenstern, J.B. and Candela, P.A. (eds) (1997) High-level silicic magmat- ism and related hydrothermal systems. Special issue, Journal of Petrology, 38(12), 1617–807. Taylor, R.P. and Strong, D.F. (eds) (1988) Recent Advances in the Geology of Granite-related Mineral Deposits. Canadian Institute of Mining and Metallurgy, Special Volume 39, 445 pp. Thompson, J.F.H. (ed.) (1995) Magmas, Fluids and Ore Deposits. Mineralogical Association of Canada, Short Course Handbook, 23, 525 pp. 126 PART 1 IGNEOUS PROCESSES ITOC02 09/03/2009 14:37 Page 126Hydrothermal Processes ITOC03 09/03/2009 14:36 Page 127ITOC03 09/03/2009 14:36 Page 1283.1 INTRODUCTION This chapter extends the concept of hydrothermal mineralization to deposits related to ?uids derived from sources other than magmatic solutions. Such ?uids include those formed from metamor- phic dehydration reactions, from the expulsion of pore ?uids during compaction of sediment, and from meteoric waters. It also considers sea water as a hydrothermal ?uid with speci?c reference to the formation of base metal deposits on the ocean ?oor. Unlike the previous chapter, which was mainly concerned with granite-related ore deposits, the present chapter discusses a much broader range of ore-forming processes and environments. Hydrothermal ore-forming processes are ubi- quitous and there is scarcely an ore deposit any- where on Earth that has not been formed directly from hot aqueous solutions ?owing through the crust, or modi?ed to varying degrees by such ?uids. This view is supported by the example in section 2.12 of Chapter 2, where a compelling case for hydrothermal precipitation of PGE min- eralization in the potholes of the Merensky Reef, Hydrothermal ore-forming processes Box 3.1 Fluid mixing and metal precipitation: the Olympic Dam iron oxide–copper–gold deposit, South Australia Box 3.2 Alteration and metal precipitation: the Golden Mile, Kalgoorlie, Western Australia – an Archean orogenic gold deposit Box 3.3 Exhalative venting and “black smokers” on the sea ?oor: the Cyprus-type VMS deposits Box 3.4 Sedimentary exhalative processes: the Red Dog deposit, Alaska, USA Box 3.5 Circulation of orogeny driven aqueo-carbonic ?uids: Twin Creeks – a Carlin type gold deposit, Nevada, USA Box 3.6 Circulation of sediment-hosted connate ?uids. 1 The Central African Copperbelt Box 3.7 Circulation of sediment-hosted connate ?uids. 2 The Viburnum Trend, Missouri, USA ORIGIN OF FLUIDS IN THE EARTH’S CRUST DEFORMATION, PRESSURE GRADIENTS, AND HYDROTHERMAL FLUID FLOW METAL SOLUBILITIES IN AQUEOUS SOLUTIONS the nature of metal–ligand complexes and Pearson’s Principle FLUID–ROCK INTERACTIONS AND ALTERATION PRECIPITATION MECHANISMS physico-chemical processes adsorption biologically mediated processes METAL ZONING AND PARAGENETIC SEQUENCES MODERN ANALOGUES OF HYDROTHERMAL ORE-FORMING PROCESSES the VMS–SEDEX continuum ORE DEPOSITS ASSOCIATED WITH AQUEO-CARBONIC HYDROTHERMAL FLUIDS orogenic, Carlin-type, and quartz pebble conglomerate hosted gold deposits ORE DEPOSITS ASSOCIATED WITH CONNATE FLUIDS stratiform sediment hosted copper (SSC) and Mississippi valley type (MVT) lead–zinc deposits ORE DEPOSITS ASSOCIATED WITH METEORIC FLUIDS sandstone-hosted uranium deposits ITOC03 09/03/2009 14:36 Page 129Bushveld Complex, traditionally regarded as igneous in origin, is presented. Likewise, the Au–U ores of the Witwatersrand Basin can no longer be regarded simply as paleoplacer deposits and hydrothermal processes have clearly played a signi?cant role in their formation (see section 3.9 below). Similarly, oil and gas deposits (see Chap- ter 5) have migrated to their present locations in the presence of hot water, during processes akin to those discussed below for hydrothermal solu- tions. Many of the giant ore deposits of the world owe their origins to the ?ow of hydrothermal ?uids in the Earth’s crust and the ability of aque- ous solutions to effectively scavenge, transport, and concentrate a wide range of economically important components. Over the past few decades in particular, a great deal of research has been directed toward bet- ter understanding the complexity of hydrother- mal processes. Concepts such as the source of hydrothermal solutions, their passage through the Earth’s crust, and the precipitation mechan- isms involved in the formation of ore bodies are now relatively well understood. The three edi- tions of Barnes’s Geochemistry of Hydrothermal Ore Deposits (1967, 1979a, 1997) provide an account of the progress of this research over sev- eral decades. There are, of course, some features of hydrothermal ores about which we understand very little, and these include the ages and dura- tion of ore-forming processes, the detailed recog- nition of ancient ?uid pathways, and the depths of ?uid ?ow in the crust, as well as the relation- ship between global tectonics and metallogeny (Skinner, 1997). The last topic is a particularly important one (see Chapter 6) that applies not only to hydrothermal ores, but to the entire range of mineral deposit types. Finally, the role of microorganisms in the formation of ore deposits is a topic that is beginning to attract attention and may turn out to be much more important than previously thought. It will be evident that the separation of “magmatic-hydrothermal” and “hydrothermal” processes into two sections (Chapters 2 and 3 respectively) is not so much a conceptual neces- sity as a requirement of the organization and structure of the book. The two chapters should be seen as covering a spectrum of processes rang- ing from magmatic ?uid ?ow to shallow level meteoric in?ltrations. In order to emphasize the continuum, this chapter is terminated with a summary diagram that attempts to relate the sources of hydrothermal solutions to ore deposit types, and applies to both Chapters 2 and 3. 3.2 OTHER FLUIDS IN THE EARTH’S CRUST AND THEIR ORIGINS Figure 3.1a shows that, in addition to magmatic ?uids, there are four other major water types on or near the Earth’s surface. Although they may all have had similar origins, each of these ?uid reser- voirs is different in terms of its composition and temperature and will, therefore, play different roles in the formation of ore deposits. The major water types are de?ned as sea water, meteoric water, connate water and metamorphic water, listed typically in order of increasing depth (and temperature) in the crust. A ?fth ?uid reservoir, where waters are derived from a mixing of two or more other water types, is also described below, speci?cally because mixed ?uids can be very important in certain ore-forming environments. In their present settings each of these ?uids can be identi?ed because the environment from which the ?uid came is known. In the geological past, however, where only indirect manifesta- tions of ?uid ?ow are apparent, it is much more dif?cult to determine the origin of a particular water type. Fortunately, the hydrogen and oxygen isotope characteristics of water are reasonably diagnostic of its source and can be used to deduce the origins of an ancient reservoir. Figure 3.1b is a plot of ?D versus ? 18 O that shows trends and ?elds which serve to ?ngerprint the major ?uid reservoirs on or near the Earth’s surface. The origins of the various water types and their stable isotopic characteristics are brie?y described below. 3.2.1 Sea water The oceans collectively represent the largest ?uid reservoir on the Earth’s surface (Figure 3.2), cover- ing some 70% of the surface and containing about 130 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 130HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 131 98% of its free water. As mentioned in Chapter 2, the Earth’s surface has been covered by substan- tial volumes of water since very early in Earth his- tory. Sea water is relatively well mixed at a global scale, and is weakly saline because of reaction with both continental and oceanic erosion pro- ducts over time. The principal dissolved con- stituents in sea water are the cations Na + , K + , Ca 2+ , and Mg 2+ and anions Cl - , HCO 3 - , and SO 4 2- , which typically occur at a total concentration (or salinity) of around 35 g of solids per kg of sea water (3.5 wt%). Sea water is extensively circulated through the oceanic crust and is responsible for widespread alteration and metal redistribution in this portion of the Earth’s crust. The drawdown of sea water into major faults associated with the mid-ocean ridges and its subsequent emergence from exhal- ative vents or “black smokers” is a major oceano- graphic discovery that has revolutionized the understanding of volcanogenic massive sul?de (VMS) deposits (see section 3.8 below). VMS deposits occur in many different parts of the world, and in rocks of all ages, con?rming the 0 –120 –20 –40 –80 –10 0 10 20 ?D (% ) ? 18 O (% ) Primary magmatic water Metamorphic water (300–600°C) B Meteoric water line (?D = 8 × ? 18 O + 10) Latitudinal shifts High Low Michigan basin Alberta basin A Gulf Coast SMOW A B Steamboat Springs, Nevada Salton Sea (b) (a) Connate Meteoric Ocean Sea water Metamorphosed rocks Metamorphic Sedimentary basin Igneous intrusion Magmatic Chapter 2 Connate fluids ? Figure 3.1 (a) The major types of liquid water that exist at or near the Earth’s surface; magmatic water was speci?cally discussed in Chapter 2, while the other ?uids are discussed in this chapter. (b) Plot of hydrogen (?D permil) and oxygen (? 18 O permil) isotopic ratios for various water types (after Taylor, 1997). Standard Mean Ocean Water (SMOW) is de?ned to be zero for both ?D and ? 18 O. Some ore-forming environments are clearly related to mixed ?uid reservoirs; in the cases shown (see A and B) mixing of meteoric and connate ?uids has taken place. ITOC03 09/03/2009 14:36 Page 131132 PART 2 HYDROTHERMAL PROCESSES importance of sea water as a hydrothermal ?uid source. Table 3.1 illustrates the concentrations of major ionic species in sea water and compares this to ?uids exhaled from black smoker vents, where signi?cant concentrations of metals have taken place due to interaction between hot sea water and oceanic crust. Sea water is also compared to rain water and meteoric derived groundwaters, which again have higher dissolved solute con- tents at higher temperatures. 3.2.2 Meteoric water Meteoric water has its immediate origin within the hydrological cycle and has, therefore, been in contact with the atmosphere. In a geological con- text it refers to groundwater that has in?ltrated into the upper crust, through either rainfall or seepage from standing or ?owing surface water. In this sense sea water in?ltrating into the crust should also be regarded as a source of meteoric groundwater. Groundwater is the second largest reservoir of liquid water (Figure 3.2) and generally exists close to the surface in the interstitial pore spaces of rocks and soil. It does not refer to the water that forms part of the crystal structure of hydrous minerals, nor does it refer to the micro- meter-sized ?uid inclusions that occur within many of the rock-forming minerals of both the crust and mantle. Meteoric ?uid can nevertheless Ice (43.4) Precipitation 0.107 Lakes/rivers (0.13) Groundwater (15.3) Precipitation 0.398 f f f Mid-ocean ridge (black smoker) Ocean (1400) f Figure 3.2 Simpli?ed diagram illustrating the water budget on or close to the surface of the Earth (after Berner and Berner, 1996). Precipitation ?uxes are in 10 6 km 3 yr -1 and reservoir volumes in 10 6 km 3 . Table 3.1 Representative concentrations (in ppm) of major ionic species in natural meteoric waters Species (ppm) Sea water Black smoker sea water Rain water Meteoric spring Meteoric geothermal (at 100–300 °C) water (at 6 °C) water (at 300 °C) Na + 10000 6000–14 000 1.0 23 187 Cl - 20 000 15 000–25 000 1.1 3.1 21 SO 4 2- 2700 0 1.5 11 103 Mg 2+ 1300 0 0.2 2.4 0 Ca 2+ 410 36 0.4 5.1 0.5 K + 400 26 0.5 1.0 27 Si 4+ 0.5–10 20 1.2 17 780 Metals (Fe, Cu, Zn, Mn, etc.) depleted enriched depleted depleted enriched Source: after data in Krauskopf and Bird (1995), Scott (1997). ITOC03 09/03/2009 14:36 Page 132HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 133 penetrate along fractures to deep levels in the crust and is, therefore, involved in widespread circulation throughout the crustal regime. It is responsible for the formation of many different types of hydrothermal ore deposits, but especially those characterized by relatively low temperature transport and precipitation, such as sandstone- hosted and sur?cial uranium ores (see section 3.11 and Chapter 4). The oxygen and hydrogen isotope compositions of meteoric waters, both on the surface and as groundwater, vary systematically over the entire globe as a function of latitude and elevation. The linear relationship between ?D and ? 18 O values (the meteoric water line in Figure 3.1b) exists because deuterium and hydrogen, as well as 18 O and 16 O, are systematically fractionated between liquid water and water vapor during the evapora- tion–condensation processes of the hydrological cycle, and this fractionation increases as a func- tion of temperature (Craig, 1961). Thus, most meteoric waters have ?D and ? 18 O values that vary along a straight line (?D = 8 ×? 18 O + 10; Figure 3.1b) and which help to broadly identify the climatic conditions or latitude from where the water came. 3.2.3 Connate water Water that is included within the interstitial pore spaces of sediment as it is deposited is referred to as connate or formational water. Originally this water is either meteoric or sea water, but it under- goes substantial modi?cation as the sediment is buried, compacted, and lithi?ed. The various stages of diagenesis that result in the transforma- tion from uncompacted particles of sediment to lithi?ed sedimentary rock produce aqueous solu- tions that evolve with time and depth. Such ?uids invariably move through the sequence and are often involved in the formation of ore deposits. The progressive burial of sediment to depths of around 300 meters results in a rapid reduction of porosity and the initial production of a sub- stantial volume of water. Shales are initially very porous and the early stages of burial can result in the production of up to 3500 litres of water for each cubic meter of sediment deposited (Hanor, 1979). Over 75% of the interstitial pore ?uid is likely to be expelled from a shale by the time it is buried to 300 meters depth and its porosity will, accordingly, decrease rapidly at this early stage. By contrast, uncompacted sandstones are initially less porous than shales and will release only 700 litres of water for each cubic meter of sediment deposited (Hanor, 1979). Calculated average rates of water release for shales and sands are shown as a function of depth of burial in Figure 3.3a. An additional source of formation water is the “structural” or “bound” water that occurs either as loosely bound H 2 O or OH - molecules within clay mineral particles in argillaceous sediments. Such water is able to leave the host mineral once temperatures of 50–100°C have been reached. This is shown as the episodic peak-like increases in the rate of release of water from the sediment in the depth pro?les of Figure 3.3a. The nature and volume of bound water release will obviously vary from one situation to the next and depend on factors such as the local geothermal gradient and the type and proportions of clay host minerals. The two main stages of connate ?uid production (i.e. pore ?uid and bound water) are also schematic- ally illustrated in Figure 3.4, where they are com- pared to the production of metamorphic ?uids that occurs at somewhat higher temperatures and greater burial depths. The temperature of connate ?uids increases with depth in the sedimentary sequence, with the exact rate of increase being a function of the local geothermal gradient. The latter varies typically between 15 and 40°C km -1 . Fluid pressures will also increase with depth, although the nature of pressure variation with progressive burial is likely to be complex and depend on the amount of pore ?uid present in any given sedimentary unit and the interplay between hydrostatic and lithostatic pressure gradients. A more detailed description of the relationships between hydrostatic and lithostatic pressure is presented in section 3.3.2 below, where its signi?cance with respect to the movement of hydrothermal ?uids in sedimentary sequences is discussed. Figure 3.3b illustrates the way ?uid pressure varies with depth in the Gulf Coast sediments of the southern USA and points ITOC03 09/03/2009 14:36 Page 133134 PART 2 HYDROTHERMAL PROCESSES Depth (km) 0 5 0.01 1 2 0.1 1 10 Depth (km) Rate of water release (l m –1 of burial) (a) 3 4 0 5 0 1 2 500 1000 Depth (km) Pressure (bars) (b) 3 4 0 5 1 2 1.1 1.2 1.3 Density (g cm –3 ) (c) 3 4 0 5 1 2 100 200 500 Depth (km) (d) 3 4 Sands Shales Episodic release of “bound” water from clays Release of interstitial water on compaction Gulf Basin Lithostatic pressure gradient Hydrostatic pressure gradient Gulf Basin Michigan Alberta 400 300 Michigan Alberta Sands Shales Gulf Basin Salinity (% ) ITOC03 09/03/2009 14:36 Page 134HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 135 to signi?cant deviations above the hydrostatic gradient (i.e “overpressuring”) caused by low per- meability horizons within the sequence which impede the expulsion of ?uids so that pore water supports the weight of the overlying strata and high ?uid pressures result (Hanor, 1979). Zones of overpressured ?uids dictate the nature of ?uid ?ow and mass/heat transfer, which has important implications for the migration of oil brines (see Chapter 5) and the formation of a variety of sediment-hosted hydrothermal ore deposits. Connate ?uids also undergo increases in dens- ity and salinity with depth, as illustrated in Fig- ure 3.3c and d. The density increases are related to increases in pressure and salinity, although there is a limit to such a trend since temperature is also augmented and this has an inverse effect on dens- ity. Deep basinal waters are commonly saline to the extent that they are generally unpotable. The increase in salinity is sometimes related to inter- action of connate waters with evaporitic horizons which contain easily dissolvable minerals such as halite, sylvite, gypsum, and anhydrite. Salinity increases are also, however, apparent in sedimen- tary sequences which do not contain evaporites and in such cases the process of “membrane-” or “salt-?ltration” probably applies. In argillaceous layers, where clay particles are strongly com- pacted, the interaction of the net negative charge (caused by atomic substitutions within the crys- tal lattice) around each particle creates what is termed a “Gouy layer” (Berner, 1971). The latter acts as a semipermeable membrane or ?lter which serves to reject anions in solutions passing through the shale layer. Since anions are excluded from the ?ltrate then so too must be cations, in order to maintain electroneutrality. The result is that any solution passing through the compacted clay will emerge with a lower salinity. Upwardly migrating brines will, therefore, become less sa- line, whereas the dissolved salt content of deeper residual waters will increase. Although doubts have been expressed about the ef?cacy of forcing solutions upwards in a sedimentary sequence across semipermeable shale layers, the process of salt ?ltration is generally regarded as a feasible explanation for the widespread existence of sub- surface connate brines. The topic is of consider- able importance to ore genesis, as the availability of ligands in hydrothermal ?uids, especially Cl - , is crucial to increasing the solubility and trans- port ef?ciency of metals in solution (see Chapter 2 and section 3.4 below). The salinity characteristics of connate waters will vary from one sedimentary basin to another, as seen in Figure 3.3d. Likewise, the ?D and ? 18 O characteristics of such ?uids will also vary depending on the climatic characteristics and lati- tude of the host basin. Since the original trapped pore ?uids in most sedimentary sequences will be either meteoric or sea water, their hydrogen and oxygen isotope compositions will initially re?ect either the appropriate point along the meteoric water line or SMOW (Figure 3.1b). As the ?uids evolve, through both temperature increases and interaction with the host rocks, their stable iso- tope characteristics will also change and de?ne trends that typically deviate from a position on the meteoric water line along paths of shallower slope. Fluids will tend to re?ect higher ? 18 O values as their temperatures and salinities in- crease in sedimentary sequences (Taylor, 1997). In Figure 3.1b, for example, the low latitude Gulf Coast formational waters de?ne a trend that is quite distinct from the ?uids in the high latitude Alberta Basin of Canada. 3.2.4 Metamorphic water As rocks are progressively buried and temperatures exceed about 200°C, the process of diagenesis evolves to one of metamorphism. Metamorphism is a multifaceted process but essentially involves the transformation of one mineral or mineral assemblage to another whose stability is more in equilibrium with the prevailing conditions at Figure 3.3 (Opposite) The formation and characteristics of connate waters. (a) Depth pro?le illustrating the rate of water release due to initial compaction and porosity reduction followed by dehydration of clay minerals such as montmorillonite. (b, c, and d) Depth pro?les showing typical pressures, ?uid densities, and ?uid salinities, respectively, in a compacting sediment pile. Different pro?les refer to the Michigan, Alberta, and Gulf basins in the USA (after Hanor, 1979). ITOC03 09/03/2009 14:36 Page 135136 PART 2 HYDROTHERMAL PROCESSES higher pressure and temperature. Of importance in this discussion is the transformation of hydrous silicate and carbonate minerals to newly formed assemblages that contain lesser volatiles than before. Dehydration and decarbonation reactions during prograde metamorphism are, therefore, very important processes that produce substantial volumes of metamorphic water in the mid- and lower-crust. An example from the Witwatersrand Basin in South Africa (Figure 3.4) shows that at around 300°C a metamorphic reaction involving the breakdown of kaolinite to pyrophyllite will result in the production of meta- morphic water. The latter arises from the fact that kaolinite contains more water than does pyrophyllite so that the excess water liberated by the phase transformation has to be expelled into the sedimentary sequence. This reaction is fol- lowed at around 400 °C by the transformation of muscovite and chlorite to biotite, when more ?uid will be produced. Both H 2 O and CO 2 (the latter speci?cally when the breakdown reactions involve carbonate minerals), as well as volatiles such as CH 4 and sulfur species, are produced as metamorphic ?uids during prograde reactions of this type. Fluids produced in low to medium grade meta- morphic terranes are dominated compositionally by H 2 O, CO 2 , and CH 4 in approximately that order of abundance. High grade rocks tend to be dominated by dense CO 2 with lesser amounts of associated water and methane. Most metamor- phic ?uids have low salinities and low concentra- tions of reduced sulfur. Fluids of this nature are globally implicated in the formation of orogenic gold deposits, which represent a very important category of gold deposit types (Phillips et al., 1994; see Box 3.2). Although it has not been mentioned before, it is clear that CO 2 may be an important component of hydrothermal solu- tions, not only in those originating through meta- morphic devolatilization but also in magmatic waters. Carbon dioxide is, after water, the most abund- ant component in hydrothermal solutions and its phase equilibria relative to those of water are brie?y considered below. It is a larger molecule than water and is non-polar (see Chapter 2), which accounts for its lower melting and critical points. It is of little importance as a solvent other than for non-anionic species such as the hydrocarbons, especially CH 4 . Muscovite + chlorite biotite 10 0 0 Proportion of water released (wt%) 3 2 1 100 4 7 6 5 9 8 200 300 400 500 600 T (°C) Connate fluid Metamorphic fluid Pore fluid produced during compaction “Bound” water produced during breakdown of clays Kaolinite + quartz pyrophyllite } Biotite + quartz garnet } Figure 3.4 The relationship between connate ?uid production during diagenesis and metamorphic ?uid production due to systematic dehydration reactions involving the breakdown of one mineral assemblage to another containing less water. The conditions pertaining here were calculated speci?cally for the Witwatersrand Basin (after Stevens et al., 1997). ITOC03 09/03/2009 14:36 Page 136HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 137 A part of the phase diagram for pure CO 2 is shown in Figure 3.5a and when compared to a similar diagram for H 2 O (Figure 2.2c) is seen to have remarkably similar densities and phase tran- sition characteristics to water, except that it has lower melting (-56.6°C) and critical (31.1°C) points. At high temperatures H 2 O and CO 2 are completely miscible and form a single ?uid phase where the one compound is dissolved in the other. At lower temperatures H 2 O and CO 2 become 400 0 –60 300 100 200 (a) CO 2 –40 –20 0 20 40 60 Temperature (°C) Liquid Triple point 1.15 1.10 1.05 1.00 0.95 Isochore 0.85 Critical point 0.75 0.65 0.55 0.45 0.15 0.02 Pressure (bars) Vapor 0.10 Pure H 2 O–CO 2 at 2kb H 2 O–CO 2 with 2.6% NaCl H 2 O–CO 2 with 6% NaCl at 1.5kb H 2 O–CO 2 with 6% NaCl at 0.5kb A 0 20 40 100 Mole % CO 2 60 80 600 0 Temperature (°C) 300 100 200 (b) 400 500 H 2 O–CO 2 Two phase(immiscible) One phase(miscible) Figure 3.5 (a) Part of the phase diagram for CO 2 showing the variations in ?uid density (i.e. isochores in g cm -3 ) as a function of pressure and temperature. The diagram is similar to that for H 2 O (see Figure 2.2c) except for the lower melting and critical points. (b) Various solvus curves for the system H 2 O–CO 2 ; the solvus (i.e. the curve which de?nes where in P–T–X space H 2 O and CO 2 unmix) moves to higher temperatures with decreasing pressure and increasing salinity of the mixture (compilations after Brown, 1998). ITOC03 09/03/2009 14:36 Page 137meteoric and connate waters, but which may overlap to a certain extent with magmatic waters (Figure 3.1b). The reasons for this are complex and related to the variable nature of metamorphic protoliths and the ubiquitous effects of ?uid– rock interaction during regional metamorphism (Taylor, 1997). 3.2.5 Waters of mixed origin Although the previous discussion has emphasized the nature and characteristics of different indi- vidual ?uid reservoirs in the Earth’s crust, it is unreasonable to expect that in nature such waters would necessarily retain their integrity with time. In the upper crust especially, it is expected that ?uids will undergo mixing and effectively become hybrids with more than just a single source or origin. It is apparent that the mixing of ore-bearing hydrothermal ?uids may be an import- ant consideration in the precipitation of metals from such solutions. Magmatic and meteoric waters, for example, often undergo mixing at the volcanic to subvolcanic crustal levels associated with porphyry copper and epithermal Au–Ag ore- forming environments. Mixing of connate and meteoric waters is also a process known to char- acterize sedimentary basins forming in rapidly ?lled continental rift environments. The major effect of ?uid mixing is to promote the precipita- tion of metals from ore-forming solutions and, con- sequently, this topic is discussed in more detail in section 3.5.1 below. 3.3 THE MOVEMENT OF HYDROTHERMAL FLUIDS IN THE EARTH’S CRUST In order to be effective as a mineralizing agent, hydrothermal ?uids need to circulate through the Earth’s crust. The main reason for this is that they need to interact with large volumes of rock in order to dissolve and transport the metals required to form hydrothermal ore deposits. The ?ow of an ore ?uid should preferably be focused so that the dissolved constituents can be concen- trated into an accessible portion of the Earth’s crust that has dimensions consistent with those 138 PART 2 HYDROTHERMAL PROCESSES immiscible (like oil and water) and exist as two separate phases (usually as H 2 O liquid and CO 2 supercritical ?uid). This relationship is shown in Figure 3.5b, where the solvus (i.e. the curve which de?nes where unmixing of the single phase mix- ture occurs) for pure H 2 O–CO 2 at 2 kbar identi?es a plateau at around 250–300 °C, above which the two compounds occur as a single miscible phase. As the temperature falls the single phase ?uid progressively unmixes until at room temperature H 2 O and CO 2 are essentially immiscible. This phenomenon is clearly illustrated in ?uid inclu- sions that have trapped single phase H 2 O–CO 2 mixtures above the relevant solvus. When viewed in a laboratory at room temperature, these ?uid inclusions are typically characterized by a so- called “double bubble.” The larger “bubble” is actually a globule of immiscible CO 2 ?uid within liquid H 2 O, whereas the smaller inner “bubble” comprises a mixture of H 2 O and CO 2 vapor (Figure 3.5c). The widespread occurrence of H 2 O- and CO 2 -bearing ?uid inclusions in the rock record attests to the ubiquity of aqueo-carbonic ?uids especially in the deep crust. Metamorphic waters are characterized by relat- ively high ? 18 O values (+10 to +30 permil) which are typically quite distinct compared to both Figure 3.5 (cont’d) (c) Microphotograph showing the typical appearance of a H 2 O–CO 2 -bearing ?uid inclusion. Prior to entrapment this ?uid was a homogeneous mixture of mutually soluble H 2 O and CO 2 . On cooling it subsequently exsolved to form immiscible fractions of H 2 O (outer ?uid wetting the walls of the inclusion) and CO 2 (inner globule of ?uid). A vapor bubble occurs within the inner CO 2 globule. ITOC03 09/03/2009 14:36 Page 138of a potential ore body. The study of hydrothermal ?uid movement has received a great deal of im- petus from the oil and gas industry, where it is particularly important to understand the path- ways of hydrocarbon-bearing brines in sedimentary sequences in order to explore for these types of deposits (see Chapter 5, section 5.4). Exploration for metallic hydrothermal ores has also bene?ted over the past few years from a better understand- ing of ?uid ?ow in the Earth’s crust, especially with respect to Archean lode gold ores and Mississippi Valley type Pb–Zn deposits (see Boxes 3.2 and 3.7 respectively). In Chapter 2 the nature and duration of hydrothermal ?uid ?ow associated with mag- matic intrusions was brie?y discussed. The pre- sent section looks at hydrothermal ?uid ?ow in the Earth’s crust in more general terms and also at a variety of scales. It emphasizes the role that deformation and the resulting structural features play, in terms of both movement and focusing of ?uid ?ow in the crust. It is these features that are of particular importance to the formation and location of metallic hydrothermal ores. 3.3.1 Factors affecting ?uid ?ow at a crustal scale The question of how large volumes of ?uid can move around at deep levels in the Earth’s crust, where rocks are highly compacted and have low intrinsic permeabilities, is one that has long taxed researchers in a variety of disciplines. Movement of ?uid is typically a response to either a thermal or a pressure gradient in the Earth’s crust. The lat- ter, in particular, is often related to deformation which causes both stress and strain to vary in the rocks being deformed. Deformation is widely regarded as having played a major role in control- ling ?uid ?ow throughout the crust (Sibson et al., 1975; Oliver, 1996) and its effects are discussed in more detail in section 3.3.3 below. Since crustal deformation is invariably linked to global tectonic processes it is clear why the latter are so intimately related to the formation of hydrother- mal ores (see Chapter 6). Another question that is pertinent to under- standing the nature of ?uid movement in the Earth’s crust is that of pervasive as opposed to channelized ?ow. The two scenarios can be described in terms of “intrinsic permeability,” where ?uids in?ltrate pervasively along grain boundaries and microcracks in a rock, and “hydraulic permeability,” where ?uid ?ows along major cracks that are suf?ciently interconnected to allow a distinct ?uid channelway to develop (Oliver, 1996). Pervasive ?uid ?ow can take place in porous rocks at relatively shallow levels in the crust, a situation that allows convective ?uid movement to occur. Figure 3.6 illustrates a vari- ety of tectonic scenarios at a crustal scale and the nature and distribution of major ?uid ?ow vec- tors associated with each. At relatively shallow crustal levels pervasive ?uid ?ow, along a porous aquifer, for example, often occurs in response to a gravity-driven hydraulic head formed by uplifted topography (Figure 3.6a). This is believed to be the dominant type of groundwater ?ow in contin- ental regions and ?ow rates of between 1 and 10 m yr -1 could develop depending on the perme- ability of the aquifer (Garven and Raffensperger, 1997). A similar style of long-range ?uid ?ow is now known to be associated with periods of orogenic compression which effectively squeeze the ?uid out of one package of rock into channel- ways (thrusts?) or high permeability aquifers in another (Figure 3.6b). Orogeny-driven ?uids have been implicated in the formation of the major Mississippi Valley type Pb–Zn ore province of the southeast USA (see Box 3.7). In oceanic crust ?uids are believed to ?ow in response to thermal gradients formed because of the high heat ?ow that characterizes the mid- ocean ridges (Figure 3.6c). Sea water is forced down into the crust along faults and, although essentially channelized, ?ows largely in response to convection, reappearing close to the zones of maximum heat ?ow. Thermally driven convect- ive ?uid ?ow can occur, as mentioned in Chap- ter 2, in close proximity to high level magmatic intrusions. It can also occur in deep intracratonic rift basins where high heat ?ow or ?uid density gradients result in convective ?ow, but only in aquifers where permeabilities are suf?ciently high (Figure 3.6d). Large-scale ?ow of ?uid at a crustal scale can also occur in association with the dilatancy (i.e. HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 139 ITOC03 09/03/2009 14:36 Page 139140 PART 2 HYDROTHERMAL PROCESSES (a) Gravity driven Maximum flow rate: 1–10 m yr –1 200 2 0 km Hydraulic head Intracratonic basin or rift (d) Thermally driven Maximum flow rate: 0.1–1 m yr –1 200 2 0 km (b) Orogeny driven Maximum flow rate: 0.1–1 m yr –1 2 0 20 km Compression Thrust terrane 300°C Mid-ocean ridge 2 0 (c) Thermally driven Ocean floor Sea water 2–3°C f f f 350°C isotherm f Exhalative vents f f 100 km (e) Dilatancy/fault driven Maximum flow rate: >10 m yr –1 Seismic pumping Normal fault Earthquake focus 5 5 0 km Uplifted region Figure 3.6 Various tectonic scenarios illustrating the mechanisms by which major ?uid movements in the Earth’s crust take place. (a) Gravity-driven ?uid ?ow in response to the creation of a hydrostatic head in an area of uplift. (b) Orogeny-driven ?uid ?ow in response to rock compaction in a fold and thrust belt. (c) Thermally driven ?uid ?ow in the ocean crust in response to high heat ?ow at mid-ocean ridges. (d) Thermally driven ?uid ?ow in a permeable unit within a basin or rift. (e) Dilatancy- or fault-driven ?uid ?ow along major, seismically active structural features by seismic pumping and fault valve mechanisms (modi?ed after Garven and Raffensperger, 1997). ITOC03 09/03/2009 14:36 Page 140HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 141 change in volume) of a rock mass that accom- panies faulting and seismic rupture (Figure 3.6e). Orogeny-driven ?uids may be partially associated with this process, but more typically dilatancy accompanies faulting and shearing and this par- ticular process is considered in more detail in section 3.3.3 below. 3.3.2 A note on hydrostatic versus lithostatic pressure gradients As mentioned previously, one of the main factors which controls ?uid migration pathways is the existence of a pressure differential in the host rock. In a completely dry rock the pressure at any depth is effectively provided by the weight of the mass of rock above that point and is known as lithostatic pressure. In Figure 3.7 the lithostatic pressure gradient is demonstrated by the line of slope 25 kPa m -1 , and re?ects the average density (i.e. about 2.5 g cm -3 ) of the rock sequence over that interval. The distribution of stress (i.e. force per unit area) in a rock at any point along a depth pro?le at lithostatic gradients will vary with orientation, and will be maximized in a vertical sense and minimized horizontally. By contrast, the pressure on a rock located at the bottom of the ocean will be given by the hydrostatic pressure gradient (Figure 3.7), which has a slope of about 10 kPa m -1 , re?ecting the lower density of water relative to rock (i.e. about 1.0 g cm -3 ). Stresses under hydrostatic pressure are distributed equ- ally in all directions, which accounts for the relative “weightlessness” of objects immersed in water. Uncompacted sediment on the ocean ?oor con- taining abundant pore water that is still in direct contact with the overlying ocean water column will also be pressurized along the hydrostatic gra- dient. As the sediment is buried and compaction 1200 4800 0 Depth (m) 3600 300 1200 Fluid pressure (kg cm –2 ) 2400 SS 600 900 SS SH SH SS “Overpressure” due to compaction disequilibrium 25 KPa m –1 10 KPa m –1 Hydrostatic Lithostatic Figure 3.7 Plot of depth versus ?uid pressure to illustrate the difference between hydrostatic and lithostatic pressure gradients. The dashed line shows the excesses in hydrostatic pressure (i.e. overpressure) that can accumulate in low permeability shale horizons (shown as SH relative to sandstone (SS) layers in the column on the left hand side) because of the inability of these rocks to ef?ciently drain off pore waters on compaction (after Hunt, 1979). ITOC03 09/03/2009 14:36 Page 141proceeds, water will be driven off. If, however, the pore water remains interconnected then the applied load on that rock will still be carried by the water and pressures will be made up of partly hydrostatic and partly lithostatic components and have values intermediate to the two gradients. When drainage of pore ?uid from the sediment is good then a condition known as compaction equi- librium will accompany the progressive burial and lithi?cation of that rock. If, however, the removal of water is impeded by low permeability then compaction will likewise be retarded (and porosity maintained at a higher value) and pres- sures will increase to values above hydrostatic, a condition known as “overpressuring.” Fine and coarse sediment will expel pore waters at differ- ent rates during burial, and will therefore com- pact along different pressure gradients. Fluid pressures will usually be signi?cantly higher in less permeable rock units such as shales relative to well drained rocks such as sandstones. Fig- ure 3.7 illustrates this effect with respect to a sequence of alternating sandstone and shale and plots the overpressures that occur in compacting shale horizons relative to sandstones which main- tain compaction equilibrium and hydrostatic pressures. Eventually, however, with increasing depth in normal crustal pro?les pore ?uid pres- sures increase from hydrostatic to lithostatic (so that ?uid pressure equals rock pressure) and this occurs at between 5 and 10 km depth depending on the nature of the rocks. The nature of hydro- static and lithostatic pressure gradients in the Earth’s crust, and the deviations from these end- member scenarios, are critical to understanding the nature of ?uid ?ow in all rock types. 3.3.3 Deformation and hydrothermal ?uid ?ow At progressively deeper levels in the crust, rock porosity is reduced and so is the ?uid volume con- tained within that rock. It is also more dif?cult for ?uids to move pervasively through a rock and under these conditions ?uid migration can only take place along channelways represented by structural discontinuities that form during deformation. Most of the ?uid movement that occurs at deeper crustal levels, and is relevant to the formation of mineral deposits, is located within faults that may, as a result, become loci for mineralization. Fault displacements are generally accomplished by increments of rapid movement that trigger earthquake events, usually in the upper 15 km (the seismogenic regime) of the crust (Sibson, 1994). The relationships between earth- quake events, fault propogation, and ?uid ?ow have become a major and important area of study that has relevance not only to ore genesis but to hydrology and earthquake prediction. It is, for example, well known that seismic events trigger a wide range of hydrological events that can vary from the sudden drying up of wells to signi?cant increases in the ?ow of rivers adjacent to faults (Muir Wood, 1994). The work of Sibson and co- workers (Sibson, 1986, 1987; Sibson et al., 1975, 1988) in particular has emphasized the import- ance of fault-driven ?uid ?ow (Figure 3.6e) to the formation of hydrothermal ore deposits and this is discussed in more detail below. Sibson et al. (1975) ?rst introduced the term seismic pumping to describe a theoretical model in which it was envisaged that the cyclicity of stress variations in and around a rupturing, seism- ically active fault system would affect local ?uid pressures and promote the ?ow of ?uids along the fault (Figure 3.8a). Prior to a seismic event friction between the opposing faces of a fault pre- vents rupture such that shear stress increases and the adjacent rock undergoes dilation (Figure 3.8a and b). The dilatant strain that develops in the rock around the fault causes cracks to form, into which ?uids ?ow. Consequently, ?uid pressures are predicted to fall in the build up to fault fail- ure. At the instant the fault ruptures and an earthquake occurrs, however, shear stress drops signi?cantly (?? in Figure 3.8b) but ?uid pressure increases, resulting in the upward expulsion of ?uids along the fault. As frictional forces are reinstated and shear stresses increase again, the entire cycle could repeat itself, causing episodic ?uid ?ow along fault-related channelways in the crust. The seismic pumping model provided an explanation and a mechanism for moving ?uids through the crust related to cyclical variations in both the stress and strain state of rocks around faults. 142 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 142HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 143 In reality the factors controlling channelized movement of hydrothermal solutions through the Earth’s crust are more complex than the mech- anisms outlined in the Sibson et al. (1975) model. The model is also somewhat unrealistic since there seems to be little or no evidence to sub- stantiate the sort of crack propagation in rocks adjacent to faults that was envisaged in the model (Sibson, 1994). It is nevertheless still apparent that stress/strain cycling in rocks in and around faults is a major control of crustal ?uid move- ment. The relevance of such processes to the formation of mesothermal lode-gold deposits has, for example, been demonstrated by Sibson et al. (1988). These types of deposits (see Box 3.2) are typically located within high angle (i.e. sub- vertical) reverse faults (Figure 3.9a). Such faults are an enigma since compressive stresses in the Earth’s crust generally form thrust faults that are shallowly inclined (i.e. 25–30° to a horizontal maximum principal compression). Friction the- ory suggests that reactivation of high angle faults in a horizontal compressive stress regime can only occur when the local ?uid pressure reaches or exceeds the lithostatic load. The subsequent model proposed by Sibson et al. (1988) envisages Inflow Shear stress (?) Fluid pressure Fluid flow around dilatant zone Outflow Time ?? Build-up of tectonic shear stress Onset of dilatancy Fluid fills dilatant cracks Collapse of dilatant zone (b) (a) Fault plane Stress ? 2 ? 1 ? 3 Fluids Surface Dilation zone Quartz- filled veins Earthquake focus Fault plane Limit of shear dislocation Fluids orientation Earthquake, seismic failure Figure 3.8 Early model explaining the episodic ?ow of ?uid along a seismically active fault zone. (a) Block diagram illustrating the geometry of a fault generated seismic pumping system. (b) Explanation of the changes in tectonic shear stress and ?uid pressure with time before and after a seismic event, and the accompanying ?uid ?ow direction around the dilatant zone. Maximum out?ow of ?uid from the zone of dilatancy occurs as the seismic failure event occurs and fractures close (after Sibson et al., 1975). ITOC03 09/03/2009 14:36 Page 143144 PART 2 HYDROTHERMAL PROCESSES that sites of mesothermal lode-gold mineraliza- tion are located in structures where fault rupture has been caused by the attainment of lithostatic ?uid pressures. Figure 3.9c illustrates how ?uid pressure in a seismically active fault builds up to levels approaching the lithostatic load because the upper crust represents an impervious cap to the ?uid reservoir beneath. Fault rupture occurs as ?uid pressures reach the lithostatic equivalent and the seismic event is accompanied by dilat- ancy and the development of a fracture perme- ability upwards and along the fault (Sibson et al., 1988). As the fault fails and shear stresses are substantially reduced, ?uids are also discharged into the open space created by the fault rupture itself. The dramatic decrease in ?uid pressure (back toward hydrostatic values) that results from the creation of open space could promote the precipitation of dissolved constituents within the hydrothermal solution, causing the fault to reseal itself. Once this has happened ?uid pressures are likely to increase again. This crack–seal mechan- ism is now known as the “fault valve model” and appears to explain many of the characteristics of mesothermal lode-gold deposits (Figure 3.9a and c). A slightly different mechanism of ?uid move- ment is envisaged with respect to strike slip fault- ing associated with extensional stresses. Dilational fault jogs are commonly found between the end of one rupture plane and the beginning of another in extensional en echelon fault arrays (Figure 3.9b), and such sites are particularly favorable for epi- thermal gold mineralization at high crustal levels (Sibson, 1987). Fault jogs are typically character- (a) ? 3 Principal compressive stress (? 1 ) Mid-crust Brittle Active metamorphism Fluids Mesothermal lode-gold deposit Dilatant fractures (b) (c) Fluids Dilational fault jog Epithermal gold deposit Veins En echelon strike slip faults Ductile Fault rupture Fault rupture Fault valve High angle reverse fault – orogenic lode gold Suction pump Strike-slip fault jog – epithermal gold Time P lith P hyd P fluid Fault valve Suction pump (Plan view) Figure 3.9 Models explaining the nature of hydrothermal ?uid ?ow in (a) high angle reverse faults in a horizontal compressive stress regime considered to be applicable to the formation of mesothermal lode-gold deposits such as the Mother Lode in California and in many Archean greenstone belts; and (b) en echelon strike-slip fault arrays associated with extensional stresses and considered applicable to some shallow level epithermal gold deposits. (c) Plot of ?uid pressure with time showing the cyclic ?uctuations envisaged for the fault valve model (applicable to high angle reverse fault-related mesothermal lode-gold deposits) and the suction pump model (applicable to strike-slip related fault jogs and shallow level epithermal gold mineralization). The diagrams are after Sibson (1987) and Sibson et al. (1988). ITOC03 09/03/2009 14:36 Page 144HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 145 ized by the development of extensional fractures which are preserved as a network of quartz veins. Fault jog related quartz vein arrays may also be mineralized, an example of which is the Martha Hill (or Waihi) gold deposit of the Coromandel Peninsula in New Zealand (Brathwaite and Faure, 2002). The development of extensional fractures in the fault jog results from the transfer of fault slip from one en echelon fault segment to the other during seismic activity along the fault trace. The opening up of open space at high crustal levels and the discharge of ?uid into these spaces results in a rapid ?uid pressure drop (Figure 3.9c) which could, at least within 2–3 km or so of the surface, also be accompanied by boil- ing of the ?uid. The mechanical energy released by boiling would result in further hydraulic frac- turing and brecciation, with enhanced ?uid circu- lation and mineral precipitation. This particular process is referred to as the “suction pump model” (Sibson, 1987) and appears to have appli- cability to strike slip fault systems and epither- mal ?uid ?ow at shallow crustal depths (Figure 3.9b and c). There is no doubt that deformation-controlled ?uid movement along major fault systems in the Earth’s crust is of major importance to the forma- tion of hydrothermal ore deposits over a consider- able range of crustal depths. This is con?rmed by the fact that many hydrothermal ores are located within structural discontinuities and also that exploration companies frequently target struc- tural features in their quest for mineral deposits. 3.3.4 Other factors affecting ?uid ?ow and mineral precipitation Previous discussion has centered on ?uid move- ment in large-scale structures such as faults that are tens to hundreds of kilometers in length. This section considers smaller-scale ?uid migration as evident at mesoscale (as individual quartz veins) or microscale (as ?uid inclusion traces in indivi- dual minerals) levels. At these scales effective ?ow of ?uid through a rock is determined by hydraulic conductivity, or the extent to which fractures along which ?uids might ?ow are interconnected (Odling, 1997; Cox et al., 2001). In a rock contain- ing very few fractures or cracks, and where poros- ity is low, permeability may be effectively zero and little or no ?uid will ?ow through the rock (Figure 3.10). As the fracture density increases, so too does the probability of two cracks inter- connecting. Eventually a point is reached, termed the percolation threshold, where permeability suddenly increases dramatically and ?uid ?ow across a ?nite volume of rock becomes possible (Figure 3.10; Odling, 1997). The attainment of a percolation threshold, at any scale in a rock mass, is clearly necessary if effective ?uid circulation is to take place. How do we know that a ?uid had passed through a rock? The evidence that a ?uid had once ?owed through a rock mass is provided either by alteration (see section 3.6 below) or by the presence of veins that Permeability Fracture density Percolation threshold Below percolation threshold Above percolation threshold Figure 3.10 The small scale considerations of ?uid ?ow in a rock depend on the creation of permeability and a percolation threshold, which is in turn a function of connectivity and fracture density. The percolation threshold is reached when one cluster of fractures becomes large enough to span the sample region (after Odling, 1997). ITOC03 09/03/2009 14:36 Page 145146 PART 2 HYDROTHERMAL PROCESSES are typically ?lled with quartz or a carbonate phase (i.e. calcite, dolomite, siderite, etc.), together with other less common gangue minerals. Veins are formed by an assemblage of minerals that precipitate from hot aqueous solutions as they passed through a fracture, effectively fossilizing and preserving the ?uid conduit in the rock record. Fracture ?ll will develop in one of two ways (Bjørlykke, 1994): 1 By diffusion of solids from the surrounds and precipitation within a fracture or open space (i.e. a vug). Since a mineral such as quartz has a lower solubility in aqueous solution at lower pressures and temperatures (see section 2.4.1 in Chapter 2), it follows that it is likely to precipitate in a frac- ture because the open space will exist under hydrostatic pressure gradients compared to the fracture walls which carry higher pressure litho- static loads. 2 By precipitation of minerals from ?uids ?owing through a fracture. In this case material can be brought in from a distant source and will preci- pitate in the open space as a function of many dif- ferent processes (see section 3.5.1 below), such as temperature decrease, rock alteration, solubility decrease, and boiling. The factors controlling precipitation of minerals are varied and complex and this topic is discussed in more detail with respect to ore-forming con- stituents in sections 3.4 and 3.5 below. Minerals also precipitate from ?uids ?owing pervasively through a porous rock mass and this is the mech- anism responsible for diagenesis, whereby sedi- mentary particles are cemented together and lithi?ed (i.e. turned into rock). Assuming equi- librium between pore ?uid and host rock, the volume of minerals precipitated from a ?uid circulating through a rock mass can be calculated from the relation (after Bjørlykke, 1994): V m = Ftsin ß(dT/dZ)? T /? [3.1] where V m is the volume of mineral precipitated; F is the ?uid ?ux; t is time (seconds); ß is the angle between direction of ?ow and isotherms in the rock; dT/dZ is the geothermal gradient; ? T is a function which re?ects solubility in terms of temperature, and ? is the density of the mineral being precipitated. A schematic illustration of mineral precipita- tion, and conversely of mineral dissolution, is shown in Figure 3.11, where pore ?uid is consid- ered to be convecting freely through a porous san- stone. Because the solubility of quartz is reduced down a temperature gradient it will tend to be precipitated along a cooling path, but dissolved along a heating path. The volumes of quartz pre- cipitated in such a situation can be calculated from equation [3.1] above. Conversely, a mineral such as calcite exhibits retrograde solubility (its solubility decreases as a function of increasing temperature) and will, therefore, behave in the opposite sense to quartz, tending to dissolve at the sites of quartz precipitation and vice versa Dissolution of quartz Precipitation of calcite Heating path Cooling path Precipitation of quartz Dissolution of calcite Pore fluid flux Decreasing calcite solubility Increasing quartz solubility Increasing temperature c c c c c q q q q Aquifer Figure 3.11 Schematic diagram showing a convecting pore ?uid circulating in a porous sandstone and the contrasting behavior of mineral precipitation and dissolution. In this model it is assumed that quartz solubility decreases, but calcite solubility increases, as ?uid temperatures decrease (after Byørlykke, 1994). ITOC03 09/03/2009 14:36 Page 146HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 147 (Figure 3.11). Different types of mineral precipit- ates in a sandstone aquifer can, therefore, provide information about the nature and hydrology of ?uid ?ow. In this type of environment calcula- tions will show that the precipitation of even small volumes of quartz or calcite requires huge ?uid ?uxes, so that mass transfer becomes effect- ive only when sediments have high permeabilit- ies, or ?uid ?uxes are very focused (Bjørlykke, 1994). Although cooling is important, precipita- tion of minerals from circulating ?uids is not only temperature-dependent and many other factors control the formation of ore constituents from hydrothermal solutions (see section 3.5 below). 3.4 FURTHER FACTORS AFFECTING METAL SOLUBILITY Solubility is de?ned as the “upper limit to the amount of dissolved metal that a hydrothermal ?uid can transport” (Wood and Samson, 1998). In Chapter 2 the concept of metal solubility in hydrothermal aqueous solutions was introduced and mention made of Holland’s (1972) classic experiments in which the solubility of metals such as Pb, Zn, Mn, and Fe was found to vary as an exponential function of the chloride ion (Cl - ) concentration of the ?uid. This work emphasized the fact that ef?cient transport of metals by hydrothermal ?uids in the Earth’s crust can be achieved only if aqueous solutions contain other dissolved ingredients with which the metals can bond, thereby promoting metal dissolution. For example, the solubility of a mineral such as galena (PbS) in pure water, even at high temper- atures, is extremely small. Wood and Samson (1998) calculated that the amount of Pb that could be dissolved in an aqueous solution (with 10 -3 molal H 2 S and in equilibrium with galena at 200°C) would be a trivial 47.4 ppb. If 5.5 wt% NaCl were added to this solution, however, the solubility of Pb at the same temperature would increase signi?cantly to 1038 ppm. In addition to Cl - , a variety of different ligands (electron donors or electronegative ions with lone pairs of valence electrons) will affect the solubil- ity of metals in solution. The ligand combines with a metallic ion by the formation of a coor- dinate bond, which is a type of covalent bond in which the shared electron pair is provided by one of the participating molecules (the ligand in this case) rather than the normal situation where the shared electrons are supplied from each of the two participating components. A ligand that promotes the formation of complexes by coordination bond- ing will typically increase the solubility of metals in aqueous ore-forming solutions. Considerations of how metals go into solution in hydrothermal ?uids can be made with respect to the concept of Lewis acids and bases. In the lat- ter concept the conventional de?nition of acids and bases (i.e. substances capable, respectively, of contributing H + ions to, or taking up H + ions from, a solution) is expanded to situations where H + ions are not present, and this has particular rel- evance to geological situations (Gill, 1996). A Lewis acid is de?ned as an atom or molecule that can accept a lone pair of valence electrons, whereas a Lewis base can donate an electron pair to a bond. “Hard” Lewis acids are strongly elec- tropositive metals (with high charges and/or small atomic radii) such as the alkali and alkaline earth metals (e.g. Na + , K + , Mg 2+ , Ca 2+ ) that form ionic bonds with strongly electronegative ele- ments like oxygen (O 2- ). “Soft” Lewis acids have an abundance of easily accessible electrons in their outer shells and prefer to form covalent bonds with soft bases. Soft acids are typi?ed by the chalcophile metals (i.e. those that have an af?nity with sulfur, such as Cu, Pb, Zn, Ag, Bi, Cd) which tend to form covalent bonds with lig- ands of low electronegativity, such as S 2- . The principle stating that in a competitive situation hard metals (acids or electron acceptors) will tend to complex with hard ligands (bases or electron donors), and soft metals with soft ligands, is commonly referred to as Pearson’s Principle, and underpins the solubility behavior of metals in hydrothermal solutions. A classi?cation of met- als and ligands in terms of the hard–soft break- down and applicable to ore-forming processes is presented in Table 3.2. The “borderline” category is added to accommodate the fact that some metals (such as Fe and Pb) can complex readily with both hard and soft ligands, forming a range of minerals such as sul?des (pyrite FeS 2 , galena PbS) ITOC03 09/03/2009 14:36 Page 147148 PART 2 HYDROTHERMAL PROCESSES Table 3.2 Classi?cation of some metals and ligands in terms of Lewis acid/base principles HARD METALS Li + Na + K + Rb + Cs + Be 2+ Sr 2+ Ba 2+ Fe 3+ Ce 4+ Sn 4+ Mo 4+ W 4+ V 4+ Mn 4+ As 5+ Sb 5+ U 6+ ? HARD LIGANDS NH 3 OH - - F - NO 3 - HCO 3 - CH 3 COO - CO 3 2- SO 4 2- PO 4 3- The major ligands in natural hydrothermal solutions are shown in bold. Caution is required in the application of the hard–soft model since the structure of water changes at higher temperatures and metal–ligand interaction will, likewise, change (Seward and Barnes, 1997). Source: after Pearson (1963). BORDERLINE Divalent transition metals (Zn 2+ Pb 2+ Fe 2+ etc.) ? BORDERLINE Cl - - Br - SOFT METALS Au + Ag + Cu + Hg 2+ Cd 2+ Sn 2+ Pt 2+ Pd 2+ Au 3+ Tl 3+ ? SOFT LIGANDS HS - - I - CN - H 2 S S 2 O 3 2- and carbonates (siderite FeCO 3 , cerrusite PbCO 3 ). Likewise Cl - can be an effective complexing agent for both hard and soft metals, and therefore pro- motes the solubility of a wide range of metals in hydrothermal aqueous solutions. The classi?cation of aqueous ionic species into hard and soft Lewis acids/bases assists consider- ably in understanding the nature of metal–ligand complexation, and, therefore, the controls on sol- ubility, in hydrothermal solutions (Seward and Barnes, 1997; Wood and Samson, 1998). In addi- tion, however, it is well known that temperature plays a very important role in determining the degree to which metals enter solution. In most cases stabilities of metal–ligand complexes such as PbCl + and ZnCl + will increase by several orders of magnitude as temperature increases from ambient values to 300°C (Seward and Barnes, 1997). This is illustrated, together with data for other metal–chloride complexes, in Figure 3.12. By contrast, an increase in pressure will have the opposite effect to temperature and the stabilities of metal–ligand complexes will tend to decrease because of bond dissociation and the formation of free metal ions. In general, however, the effect on solubilities of pressure increases are minimal and more than offset by the signi?cant changes asso- ciated with temperature increases (Seward and Barnes, 1997). 3.4.1 The important metal–ligand complexes in hydrothermal solutions Wood and Samson (1998) have provided an excellent summary of solubility data for a wide variety of metal–ligand complexes. These data are described below in terms of hard, borderline, and soft metals. The summary given below provides an indication of the most likely metal–ligand complex that will exist in a natural ?uid as a function of variables such as oxidation state, pH, temperature, and ?uid composition. However, it cannot and should not be used to try to simply predict the nature of metal speciation in any situation where the many parameters that control solubility are imperfectly known. The concept of hard–soft Lewis acids and bases is an idealized scenario predicated on the assumption that there is competition between metals and a variety of ligands. In natural situations metals may, for ex- ample, complex with the most suitable available base and the formation of actual metal–ligand complexes and their solubilities will be quite dif- ferent to those predicted on theoretical grounds. Hard metals Tungsten (W) Tungsten tends to occur in nature as the hexavalent aqueous cation W 6+ , although ITOC03 09/03/2009 14:36 Page 148HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 149 the pentavalent form W 5+ can also occur under more reducing conditions. These ions are hard Lewis acids and are likely to complex with hard bases such as O 2- , OH - , F - , and CO 3 2- , although the chloride ion is relatively unimportant as a complexing agent under most geological condi- tions. The majority of dissolved tungsten species in hydrothermal solutions occur as tungstates and have the forms WO 4 2- , HWO 4 - , and H 2 WO 4 . The stabilities of these tungstate species are obviously dependent on hydrogen ion concentration (i.e. on pH), such that H 2 WO 4 is only stable under acidic conditions, whereas WO 4 2- is stable at alkaline pH. Signi?cant quantities of tungsten can be car- ried in solution by these tungstate complexes and no other species are required to form most scheel- ite and wolframite deposits. Molybdenum (Mo) Molybdenum tends to be more easily reduced in nature than tungsten and typically occurs in a variety of aqueous valence states as Mo 6+ , Mo 5+ , Mo 4+ , and Mo 3+ . It is also a hard Lewis acid and like tungsten will complex with hard bases, in particular, O 2- and OH - . Chloride complexing takes place only under very acidic conditions and typically it is the oxyhy- droxide complexes (such as MoO 2 + and Mo(OH) 3 + ) that are the most likely to occur in natural aque- ous solutions. Uranium (U) Uranium is characterized by two valence states in nature, as the quadrivalent uranous ion, U 4+ , and the hexavalent uranyl ion, U 6+ . The uranous ion is generally characterized by very low solubilities in aqueous solutions and most ?uid-related transport takes place as U 6+ . As a hard acid, U 6+ will complex with hard bases such as O 2- , OH - , F - , HCO 3 - , and CO 3 2- . Borderline metals Arsenic (As) The trivalent cation As 3+ domin- ates the valence state of arsenic in nature and in solution the neutral H 3 AsO 3 complex is known –1 100 log ß 1 3 2 1 200 300 400 0 4 5 6 7 8 Temperature (°C) ZnCl + + UCl 3 MnCl + FeCl + AgCl° PbCl + Figure 3.12 Plot of the effective stability of a metal complex (expressed in terms of ß 1 , which is the equilibrium formation constant) versus temperature (after Seward and Barnes, 1997). The stabilities of metal–chloride complexes shown increase by several orders of magnitude as temperature increases, as will their solubilities in aqueous solution. ITOC03 09/03/2009 14:36 Page 149to be stable over wide ranges of temperature (above 200 °C), Eh, and pH. Other less protonated arsenate complexes are also stable at higher pH values (as for tungsten) and generally chloride complexes do not play a role in ?uid transport. Under conditions of high sul?de concentration and at low temperatures, As–sul?de complexes (such as AsS 2 (SH) 2- ) may become more important, although their exact stoichiometric proportions are unclear. Antimony (Sb) The aqueous geochemistry of antimony is similar to arsenic, with Sb 3+ predom- inating and Sb(OH) 3 occurring as the main stable complex in low-sul?de aqueous solutions. In the presence of high concentrations of reduced sul?de species in the ?uid a number of thioantimonite complexes (such as HSb 2 S 4 - ) are also likely to be stabilized and contribute to increasing antimony solubility. Iron (Fe) The majority of iron transported in aqueous solution through the Earth’s crust is in the ferrous or divalent form (Fe 2+ ), while ferric iron (Fe 3+ ), which forms under relatively oxidizing conditions, is much less soluble. With its inter- mediate Lewis acid properties, iron does not exhibit a preference for either hard or soft bases and a variety of metal–ligand complexes are implicated in the dissolution of ferrous iron in aqueous solutions, including Cl - , OH - , and HCO 3 - . Most experimental studies indicate that FeCl + and FeCl 2 are the main complexes involved in the hydrothermal transport of ferrous iron, especially at high temperatures and salinities. The free hydrated ion itself, Fe 2+ , is also implicated in ?uid transport where hydrolysis has taken place, as are Fe–bicarbonate complexes under more alkaline conditions. Fe–bisul?de complexes are generally not involved in the transport of iron in hydrother- mal solutions although they may be important in ocean ?oor exhalative environments (i.e. black smokers). Manganese (Mn) Manganese is also a borderline Lewis acid and, like iron, Mn 2+ complexation in hydrothermal solutions is likely to be dominated by the chloride ligand (i.e. MnCl + and MnCl 2 ), with hydroxide and bicarbonate complexes also contributing to Mn solubility. Tin (Sn) Tin is a metal that exhibits both hard acid quadrivalent (as Sn 4+ ) and borderline divalent (as Sn 2+ ) traits such that it can be solubilized by complexation with a number of different ligands. Under oxidizing conditions it has been found that the Sn 4+ –hydroxychloride complex, Sn(OH) 2 Cl 2 , is the dominant species, but that its solubility is low. Under more reducing conditions both Sn 4+ and Sn 2+ can complex with the simple chloride ion, forming very soluble complexes of the form SnCl n X-n (i.e. where X is the valence state, either 2 or 4, and n is the ligation number for the chloride complex). The divalent Sn 2+ –chloride complexes, formed under more reducing conditions, exhibit much higher solubilities than the quadrivalent Sn 4+ –hydroxychloride complex that exists in a more oxidized state. Sn–hydroxide complexes (Sn(OH) 4 and Sn(OH) 2 ) are very stable under alka- line, lower temperature conditions but their solub- ilities are again much lower than those exhibited by the dominant Sn–chloride complexes formed at higher temperatures and lower pH. Contrary to expectation, tin does not complex signi?cantly with ?uorine despite its association with highly fractionated ?uorite/topaz-bearing granites. Zinc (Zn) The dominant divalent zinc cation, Zn 2+ , is a borderline Lewis acid that complexes with a variety of ligands, including Cl - , HS - , OH - , HCO 3 - , and CO 3 2- . At relatively low temperatures and high pH, and in ?uids with high bisul?de concentrations but low salinities, a variety of Zn 2+ –bisul?de complexes are stable, including Zn(HS) 2 , Zn(HS) 3 - , and Zn(HS) 4 2- . In contrast, at higher temperatures and under more acidic condi- tions, a host of chloride complexes predominate, including Zn(Cl) + , Zn(Cl) 2 , Zn(Cl) 3 - , and Zn(Cl) 4 2- . Lead (Pb) Lead is the softest of the borderline Lewis acids and forms stronger bonds with Cl - and HS - ligands than with OH - or HCO 3 - . The valence state and complexing behavior of lead is very similar to that of zinc. Pb 2+ –bisul?de com- 150 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 150HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 151 plexes are more stable than chloride complexes in low temperature, neutral to alkaline, low salinity solutions. Different Pb 2+ –bisul?de complexes form as a function of pH, and these include Pb(HS) 2 under acidic conditions and Pb(HS) 3 - at higher pH (i.e. >6). At higher temperatures and lower pH, however, a variety of Pb 2+ –chloride complexes form and these include Pb(Cl) + , Pb(Cl) 2 , Pb(Cl) 3 - , and Pb(Cl) 4 2- . Pb 2+ –carbonate or Pb 2+ –bicarbonate complexes can form in CO 2 -bearing ?uids but their solubilities are typically low and they are unlikely, therefore, to have relevance in ore- forming processes. There is, however, compelling evidence that lead, and to a lesser extent zinc, are transported by organo-metallic complexes (see section 3.4.2 below), especially in relatively low temperature environments such as those related to the formation of Mississippi Valley type Pb–Zn deposits (see Box 3.7), where hydrothermal ?uids are known to contain hydrocarbons. The acetate ligand (CH 3 COO - ) is one organic molecule that has been implicated in base metal transport, although there are others, such as thiols or sulfur-bearing organic compounds, that may also be important. Soft metals Copper (Cu) In nature, copper occurs in both Cu + and Cu 2+ valence states, although the mono- valent state is predominant in most hydrothermal ?uids. It is a relatively soft acid and forms stable Cu + –chloride and Cu + –bisul?de complexes, as well as Cu + –hydroxide complexes if other ligands are not available. Movement of copper in hydrothermal solutions under a wide range of conditions is most likely to take place by Cu + –chloride complexation in the form CuCl 2 - . Although copper–bisul?de complexes will form in high-sulfur environments, they are not likely to play a signi?cant role in most hydrothermal ore-forming environments. Gold (Au) Gold is the softest metal ion consid- ered in this summary and prefers complexation with soft ligands such as the bisul?de ion, rather than chloride. Gold typically occurs in a monova- lent state (Au + ) in aqueous solutions and will form the complex AuHS at low pH and Au(HS) 2 - under weakly acidic to basic conditions. The bisul?de complexes predominate in most hydrothermal ?uids and are stable over a wide temperature range. Under very oxidizing, saline, and acidic con- ditions, it is possible to form both Au + – and Au 3+ –chloride complexes in the form of Au(Cl) 2 - and Au(Cl) 4 - respectively. The former complex might prevail in certain high temperature envir- onments such as porphyry coppers, whereas the latter is only likely to be important at low temper- atures in near surface settings. Au does form very soluble complexes with cyanide (Au(CN) 2 - ), which is the reason why cyanidation is used so effect- ively at low temperatures to recover gold from ore. The cyanide molecule is, however, unstable at higher temperatures and the complex has no relevance to hydrothermal gold transport. Silver (Ag) Silver is also a soft metal ion, but not as soft as gold. It is, therefore, more likely to form complexes with a borderline ligand such as chlo- ride than gold is, although solution chemistry will again be dominated by Ag + –bisul?de bonding. The predominant complexes are AgHS and Ag(HS) 2 - under acidic and neutral to alkaline conditions respectively. Several Ag + –chloride complexes are known to be stable over a wide range of condi- tions pertaining to hydrothermal ?uid ?ow and these include AgCl, Ag(Cl) 2 - , and Ag(Cl) 3 2- . Mercury (Hg) Mercury, occurring predominantly as Hg 2+ in aqueous solutions, is a soft metal ion and will complex preferentially with bisul?de ligands. Most mercury transport under neutral to alkaline and moderately reducing conditions is likely to take place by the formation of complexes such as Hg(HS) 2 , HgS(HS) - , and HgS 2 2- . Harder lig- ands probably do not play much of a role in most hydrothermal ?uids. Of interest and related to its low boiling point, however, is the fact that mer- cury can be transported as a vapor, or as elemental Hg in either aqueous solutions or hydrocarbon- bearing ?uids. In summary, it is apparent that the borderline chloride ion is the most important ligand that ITOC03 09/03/2009 14:36 Page 151152 PART 2 HYDROTHERMAL PROCESSES promotes the dissolution and transport of metals in hydrothermal ?uids, followed by bisul?de. A variety of other ligands such as OH - , HCO 3 - , CO 3 2- , F - , SO 4 2- , NO 3 - , NH 3 , and CH 3 COO - are of lesser importance because they are either stable under abnormal conditions or exist at very low concen- trations in natural ?uids. In addition to its ability to complex with a wide range of both hard and soft metal cations, Cl - is also the major anionic species in most natural ?uids. This ensures that many natural ?uids are very capable of transport- ing large quantities of metal in solution and it is for this reason that the circulation of hydrother- mal solutions in the Earth’s crust is such an important ore-forming process. Even though metal–ligand complexes may be readily stabilized in natural aqueous solutions, it is nevertheless apparent from both direct analysis and theoretical calculation that the actual concentrations of met- als dissolved in most hydrothermal ?uids are very low. This means that the formation of a viable ore deposit still requires large volumes of ?uid passing through a highly focused point in the Earth’s crust (i.e. high ?uid/rock ratios), as well as ef?cient precipitation mechanisms to take metals out of solution and concentrate them in the host rock. The compilation shown in Figure 3.13 shows the minimum concentrations of precious and base metals (i.e. the vertical dashed lines) estimated to be necessary for the formation of a deposit, and also the ranges shown for several examples where such data are available. Solutions that deposit Au and Ag typically contain much lower concentrations, in the range 1 ppb to 1 ppm, than those associated with Cu, Pb, and Zn ores. In the latter case massive sul?de deposits are formed from ?uids that typically carry 1–100ppm of metal, whereas the ?uids associated with MVT and porphyry deposits are relatively enriched with concentrations in the range 100–1000 ppm. 3.4.2 A brief note on metal–organic complexes A number of different hydrothermal ore types that are hosted in intracratonic and rift-related sedimentary sequences are known to have formed from ?uids containing signi?cant amounts of dis- solved organic compounds. It is also well known that crude oil and bitumen invariably contains minor amounts (<1%) of inorganic material that includes dissolved metals. It is not surprising, therefore, that considerable attention has recently been given to the roles that metal–organic com- plexes might have played in hydrothermal ore forming processes, especially in Mississippi Valley-type (MVT) Pb–Zn and stratiform, red-bed or shale-hosted base metal ores (see Parnell, 1994; Gize, 1999). Organic compounds in connate waters tend to occur either as neutral hydrocarbon molecules (alkanes, aromatics, etc.) or as charged anionic spe- cies such as the carboxylic acids (acetic acid, oxalic acid, etc.). Most neutral hydrocarbons (with the exception of methane, CH 4 ) have only limited aqueous solubility and this decreases exponenti- ally as the hydrocarbon molecule gets bigger and more complex. Organic acid anions, on the other hand, may have very high solubilities in water and the dominant aqueous anion, acetate (CH 3 COO - ), can exist at concentrations up to 2000mgl -1 (Hanor, 1994). Organometallic complexes (i.e. those in which a metal is covalently bonded to the carbon of an adjacent hydrocarbon molecule) only rarely occur in nature, whereas metal–organic complexes, in which a metal cation is attached to the electron-donor atom (such as oxygen, sulfur, or nitrogen) of an organic ligand, are widely stable in natural aqueous solutions, as well as in ore ?uids (Giordano and Kharaka, 1994). In ore ?uids metal–organic complexes will only exist if they are suf?ciently stable to be able to compete with the normal inorganic complexes discussed previously. Modeling has shown that Pb and Zn can be transported as metal–organic complexes involving acetate or oxalate (C 2 O 4 2- ) ligands (e.g. Zn(CH 3 COO - ) 2 and Pb(C 2 O 4 2- ) 2 2- ) in reduced ?uids with greater than 10 -5 molal H 2 S or HS - . In more oxidized ?uids, such as those more applicable to red-bed related stratiform base metal deposits, Pb and Zn would be transported predom- inantly as metal–chloride complexes, but Pb– and Zn–acetate complexes could still account for up to 5% of metal transport (Giordano and Kharaka, 1994). ITOC03 09/03/2009 14:36 Page 152HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 153 3.5 PRECIPITATION MECHANISMS FOR METALS IN SOLUTION The previous section has shown that it is possible to stabilize many different metal–ligand complexes under wide ranging conditions in natural hydro- thermal solutions, thereby promoting the kind of solubility levels required to make effective ore- forming ?uids. Once a metal is in solution, how- ever, it then needs to be extracted from that ?uid and concentrated in a portion of the Earth’s crust that is suf?ciently restricted and accessible to make an economically viable ore body. It is obvious that a wide range of precipitation mechanisms are likely to be effective since any mechanism that will destabilize a metal–ligand complex and, therefore, reduce the metal solubility, will cause it to be de- posited in the host rock through which the hydro- thermal solution is passing. The following sections examine the physical and chemical controls on –3 Base metal ores – 2– 1012345 0.001 Log concentration (ppm) Precious metal ores 0.001–0.01 0.0015 0.008 (Ag) 0.1–1.0 (Ag) Vein Skarn Porphyry Mississippi Valley Vein Massive Sulfide 100–1000 20–2200 540 1330 9.2 4.7 3.5 200 131 40–55 38–62 35 20–29 7.0 6.5 5.0 1.2 100–4600 80–4000 10900 9600 740–7700 Figure 3.13 Base and precious metal concentrations in ore- forming hydrothermal solutions, in each of ?ve major categories of deposit type. Base metals include Cu, Pb, and Zn (?lled dots), and precious metals comprise Au, Ag, and Hg (open circles). The numbers next to individual points or ranges refer to average concentrations or ranges in a variety of different situations, and are all after Seward and Barnes (1997). ITOC03 09/03/2009 14:36 Page 153metal precipitation, as well as adsorption and biologically mediated processes. The latter is par- ticularly important as it is now widely accepted that the geochemical pathways by which both metals and non-metals are distributed in the Earth’s crust are affected to varying degrees by the action of microorganisms. The concept of biomin- eralization is becoming increasingly important to the understanding of ore-forming processes and this topic is likely to be the subject of much more research in the future. 3.5.1 Physico-chemical factors affecting metal precipitation The basic principles that control ore deposition have been summarized by Barnes (1979b) and Seward and Barnes (1997). At shallow crustal levels ore deposition will take place by open space ?lling, whereas deeper down where porosity is restricted, replacement of existing minerals tends to occur. Decrease in temperature is the factor that, intuitively, is regarded as the most obvious way of promoting the precipitation of metals from hydrothermal ?uids. At depth, however, temperat- ure gradients across the structures within which ?uids are moving tend to be minimal and metal precipitation will be neither ef?cient nor well constrained to a particular trap zone. Deposition of metals in such a case is achieved more effect- ively by changing the properties or composition of the hydrothermal ?uid. If ore solution occurs by metal–chloride complexing then precipitation could occur very ef?ciently by increasing the pH of the ore ?uid (see Figure 3.14). An example is provided by the reaction of an acidic ore solution with a carbonate host rock and where precipitation is promoted by digestion and replacement of the host. Two coupled reactions illustrate the process in terms of calcite dissolution by an originally acidic, Zn 2+ –chloride-bearing ore solution: CaCO 3(SOLID) + 2H + ? Ca 2+ + H 2 CO 3(AQUEOUS) [3.2] and subsequent sphalerite precipitation caused by increase of pH in the ?uid as hydrogen ions are consumed: ZnCl n 2 - n + H 2 S (AQUEOUS) ? ZnS (SOLID) + 2H + + nCl [3.3] It is apparent from reactions 3.2 and 3.3 that pre- cipitation of metal–chloride complexes could also take place by increasing the H 2 S concentration of the ?uid (perhaps by ?uid mixing or bacterial sulfate reduction) or by decreasing the Cl - con- centration (again perhaps by mixing of the ore ?uid with dilute groundwater). As another option to calcite dissolution, an increase in ?uid pH could also be brought about by boiling of the ?uid, which results in acidic volatiles being partitioned preferentially into the vapor phase. There are, therefore, a number of ways of reducing the solub- ility of metal–chloride complexes and promoting metal precipitation. The factors that promote precipitation of metal– sul?de complexes may be somewhat different to those for chloride complexes (Barnes, 1979b). Although ?uid mixing (or dilution) and temperat- ure decreases will promote metal precipitation, it is oxidation of the ore ?uid (see Figure 3.14) that is particularly effective in decreasing the solubil- ities of metal–sul?de complexes. Oxidation (or loss of electrons) causes a decrease in the pH and also the total sul?de concentration, thereby pro- moting metal precipitation, as shown in reac- tion 3.4: Zn(HS) 3 - + 4O 2(AQUEOUS) ? ZnS (SOLID) + 3H + + 2SO 4 2- [3.4] Precipitation of ores from metal–sul?de com- plexes are controlled by oxidation, decreases in pH (or acidi?cation), and decreases in sul?de concentration in the ?uid, all tendencies that are opposite to those controlling metal–chloride disassociation. A more detailed discussion of the actual geological processes that affect ore deposi- tion controls is presented below, together with different scenarios where they might apply. An understanding of the geological factors relating to metal deposition is obviously very important to the exploration geologist who needs to have a ?rm grasp of the concepts that control where and why hydrothermal ore bodies form. 154 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 154Temperature Since the stabilities of many metal–ligand com- plexes increase as a function of temperature (Figure 3.12) it is clear that cooling of a ?uid will generally have the effect of promoting ore deposi- tion. Temperature decrease is particularly effective for destabilizing metal–chloride complexes be- cause their solubilities are much more sensitive to temperature changes than are those of equival- ent sul?de complexes. As mentioned previously, however, at deep crustal levels where ?uid and rock are in equilibrium and exist at the same tem- perature, thermal gradients are minimal and cool- ing is a slow and ineffectual process. Under these circumstances the concentration of metals and the formation of hydrothermal ores are generally not mediated by temperature decreases. In the near surface environment, however, rapid decreases in the temperature of an ore ?uid are much more likely to happen and in such environments cooling is likely to play an important role in ore formation. The ocean ?oor undoubtedly represents the prime example of where a dramatic reduction in the temperature of ore-forming ?uids plays the dominant role in controlling metal deposition. Volcanogenic massive sul?de deposits typically form at sites where hot (up to 350°C) brines with very high metal contents are vented onto the ocean ?oor as black smokers (see section 3.8.1 and Box 3.3). Precipitation of base metals (mainly Cu and Zn transported as chloride complexes; Scott, 1997) and, in certain cases, precious metals (Au trans- ported as a gold–bisul?de complex; Scott, 1997) is virtually instantaneous as the ore ?uids mix with an essentially in?nite volume of very cold (2–4°C) sea water. In this environment it is the very rapid cooling through virtually 350°C that causes a highly ef?cient precipitation of metal in the immediate vicinity of the exhalative vent. Black smokers are probably the most dramatic and ef?cient example of hydrothermal ore deposi- tion anywhere on Earth. The volcanic settings which give rise to the deposition of epithermal Au–Ag deposits (see Chapter 2, section 2.11 and Box 2.4) are also char- acterized by steep geothermal gradients and high conductive and convective heat loss by ?uids to the surrounding rocks. Cooling may, therefore, play a role in controlling metal solubilities although in such environments it is more likely to be the decrease in pressure and associated phase separa- tion (i.e. boiling and effervescence; see below) that plays the dominant role in ore deposition. Pressure Pressure variations do not dramatically effect the solubilities of metal–ligand complexes although it is clear that pressure increases will lead to a volume reduction which, in turn, promotes the dissociation of complexes to ionic species (Seward, 1981). In effect a decrease in pressure tends to favor an increase in solubility and, there- fore, works in the opposite sense to temperature. Pressure reduction is not a process that, on its own, is typically associated with ore formation. A major exception occurs, however, when pressure reduction is accompanied by an event such as boiling or effervescence. These speci?c processes are very important in ore deposition and are gen- erally promoted by decreases in pressure. In this sense pressure only has an indirect in?uence on ore formation. Phase separation (boiling and effervescence) Although one might intuitively expect that boil- ing is promoted by an increase in temperature, there is no doubt that, in the upper levels of the Earth’s crust, the transformation from liquid to vapor usually occurs because of a decrease in ?uid pressure (see trend A–A' in Figure 3.5a). At deeper levels in the crust a decrease in ?uid pressure could also be accompanied by effervescence, or the transition from a single phase H 2 O–CO 2 mixture to one where H 2 O and CO 2 unmix (see Figure 3.5c). Both these processes are potentially very important as mechanisms of precipitating metals from ore-forming solutions because they dramatically change the prevailing conditions under which metal–ligand complexes are stable. In section 3.3.3 the relationships between ?uid ?ow and deformation were discussed and the fault-valve and suction-pump models were HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 155 ITOC03 09/03/2009 14:36 Page 155described with reference to the formation of mesothermal lode-gold and epithermal Au–Ag deposits respectively. The metamorphic ?uids generally attributed to formation of mesothermal lode-gold deposits are characterized by single phase H 2 O–CO 2 mixtures and it is possible that the episodic decreases in ?uid pressure accom- panying crack–seal processes in a fault or shear (Figure 3.9a and c) could promote precipitation of ore components. This is because there is a tendency for the H 2 O–CO 2 solvus to move to higher temperatures with decreasing pressure (Brown, 1998). A typical mesothermal lode-gold ?uid comprising H 2 O + CO 2 at 350°C and with 2.6% NaCl, for example, will initially exist as a single homogeneous liquid just above the relevant solvus (shown as point A; Figure 3.5b). A pressure decrease will have the effect of moving the solvus plateau to higher temperatures, so that the ?uid enters the immiscible, or two phase, ?eld seg- regating CO 2 and H 2 O. CO 2 effervescence will undoubtedly affect the prevailing ?uid properties and likely promote precipitation of both gangue and ore components, as suggested in the fault- valve model. Similarly, in the near surface environment where en echelon strike slip faulting results in fault jog formation, the resulting rupture-related dilation will instantaneously reduce pressure (Figure 3.9c) to such an extent that a ?uid could move from a liquid to the vapor state. Boiling will, of course, only occur if the pressure reduction traverses the boiling point curve for the relev- ant hydrothermal ?uid composition (refer to phase diagrams in Figure 2.2). The boiling away of an aqueous vapor from a hydrothermal ?uid will result in the residual enrichment of solute, as well as partitioning of other volatile species into the gas phase, with probable increase of pH in the remaining ?uid. Such processes will dram- atically modify prevailing ?uid properties and the pH increase will, as mentioned previously, be particularly effective in precipitating metal– chloride complexes from solution. When accom- panied by temperature decreases, prolonged boiling will likely bring about solubility reduc- tion of most dissolved species. As suggested by Sibson (1987) with respect to epithermal ore- forming environments, boiling is likely to be a very ef?cient mechanism for ore deposition at shallow crustal levels. Fluid mixing/dilution The mixing of two ?uids is widely regarded as another important mechanism for reducing solubility in ore-forming solutions and promoting metal precipitation. This is particularly the case when a relatively hot, metal-charged ore ?uid mingles with a cooler, more dilute solution. Mixing of the two ?uids would result in cooling of the hotter with modi?cation of the pre- vailing ore ?uid properties and destabilization of existing metal–ligand complexes. There are several examples that point to the importance of ?uid mixing during ore deposition. In a classic stable isotope study, Sheppard et al. (1971) de- monstrated that precipitation of chalcopyrite at the interface between the potassic and phyllic alteration zones associated with porphyry copper deposits (see Box 2.1, Chapter 2) coincides with the zone where ?uids of dominantly magmatic origin mingled with those derived from a met- eoric source. A similar situation is known to exist with respect to precipitation of Au and Ag ores in epithermal deposits (Box 2.4, Chapter 2). Hedenquist and Aoki (1991) showed that mag- matic ?uids venting directly from a volcano with- out any mixing or dilution below the surface are less conducive to the formation of economic- ally viable deposits since ground preparation (or alteration) is minimized and the resultant ore accumulations at the surface have a low preserva- tion potential. By contrast, systems characterized by mixing of magmatic waters with a meteoric water carapace 1–2 km below the surface will more likely be associated with the formation of economically viable mineralization. This is because mixing of a hot, saline, metal-charged magmatic ?uid with a cooler, more dilute mete- oric water promotes acid leaching of the host rocks, increases their permeability, and forces the ?uids to condense and precipitate their dissolved metal solute. Hedenquist and Aoki (1991) sug- gested that interaction of magmatic ?uids with a meteoric ?uid blanket near the surface of active 156 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 156HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 157 The Olympic Dam deposit, some 500 km north of Adelaide in South Australia, was discovered in 1975 beneath 350 m of sedimentary cover. It represents one of the most specta- cular discoveries of modern times, containing around 2 bil- lion tons of Cu, U, Ag, and Au ore (Solomon and Groves, 1994). The ore body is hosted in a breccia complex within the apical portions of the 1588 Myr old (Creaser and Cooper, 1993) Roxby Downs granite, the latter being a K-, U-, and Th-enriched A-type or anorogenic granite. Miner- alization is generally attributed to mixing of two entirely different ?uids, one of possible magmatic derivation, and the other an oxidized meteoric ?uid (Reeve et al., 1990; Oreskes and Einaudi, 1992; Haynes et al., 1995). The breccias that host the ore comprise a variety of frag- ments, the majority of which are either granitic or hematitic, together with volcanic fragments, sedimentary rocks, and massive sul?de ore. The breccias are extremely iron rich and themselves comprise a huge iron resource made up of hematite and minor magnetite. Polymetallic mineralization, in the form of chalcocite, bornite, chalcopyrite, pitch- blende, argentite, gold, and a variety of rare earth element (REE) minerals, is most closely linked to the hematitic breccia. All breccias are cut by ma?c and felsic dykes that are approximately the same age as the Roxby Downs granite, indicating that magmatism, brecciation, and miner- alization were broadly coeval (Johnson and Cross, 1991). Ore genesis at Olympic Dam (Solomon and Groves, 1994) was initiated by the high level intrusion of fertile A-type granite, with accompanying ?uid exsolution and boiling, which gave rise to explosive volcanic activity and an early phase of brecciation. The ?uid involved here is inferred to be of magmatic origin and moderately oxidiz- ing. It was also clearly Fe-rich and precipitated an early generation of magnetite as well as REE-bearing minerals. Cooling of the magmatic system subsequently led to draw-down of groundwater (perhaps derived from a saline Fluid mixing and metal precipitation: the Olympic Dam iron oxide–copper–gold deposit, South Australia 1 TN 0 km MN Roxby Downs Granite Hematite quartz breccia Hematite breccia Whenan Shaft Granite breccia South Australia Olympic Dam Adelaide Roxby Downs Volcaniclastic rocks Figure 1 Simpli?ed, subsurface geological map of the Olympic Dam ore deposit (after Haynes et al., 1995). ITOC03 09/03/2009 14:36 Page 157playa lake above the deposit; Haynes et al., 1995) that in turn led to phreatic explosive activity, further breccia- tion, and hematite precipitation. The spatial association between hematite breccia and polymetallic mineralization indicates that a low temperature, highly oxidizing, meteoric ?uid was implicated in the dissolution and transport of substantial amounts of Cu, Fe, and U, possibly from the granite, or overlying volcanic rocks, or both. This ?uid encountered and mixed with the less oxidized magmatic solutions (containing reduced sulfur species), destabiliz- ing metal–ligand complexes and causing precipitation of metal sul?des together with iron oxides. Figure 2 illustrates a model showing the existence of two fundamentally different ?uid types at different crustal levels, with ore precipitation occurring at the interface between the ?uids. It is envisaged that ?uid circulation and mixing continued episodically for a considerable period of time, perhaps in part stimulated by the high radioelement (K, U, Th) content of the granite and the resultant heat generated by radioactive decay. The longevity of the ore-forming process is consistent with the complexity and enormous size of the ore deposit. Fluid mixing (a) Geological setting Playa lake Dykes Olympic Dam breccia complex Basalt Roxby Downs granite (b) Ore precipitation Ore leaching Oxidizing meteoric water Hotter, magmatic fluid Figure 2 Geological model for the formation of the Olympic Dam deposit (after Haynes et al., 1995). The geological setting and the distribution of a lower, hotter, moderately oxidizing ?uid of possible magmatic derivation, and an upper, saline, highly oxidizing meteoric ?uid, is shown in (a). Drawdown of meteoric ?uid together with uprising plumes of hotter water promote mixing of the ?uids and metal precipitation, as shown in (b). ITOC03 09/03/2009 14:36 Page 158HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 159 volcanic regions may be an important prerequis- ite to the formation of viable epithermal Au–Ag mineralization. In another example demonstrating the import- ance of mixed ?uids in ore formation, it has been suggested that sediment-hosted ore deposits asso- ciated with continental rifts environments are the product of the mixing of meteoric and connate ?uids. Using the active mineralizing systems of the Salton Sea geothermal system (see section 3.7 below) as an analogue, McKibben et al. (1988) have suggested that the mixing of upwelling high-salinity brines of essentially connate origin with descending meteoric waters of lower tem- perature and salinity may be a viable mechanism of promoting metal precipitation from the former. It has been suggested that this type of model might apply to the formation of several sediment- hosted ore deposits such as the stratiform Cu–Co ores of the central African Copperbelt, and the sedimentary exhalative (SEDEX) Zn–Pb ores of Mount Isa and Broken Hill in eastern Australia (McKibben et al., 1988; McKibben and Hardie, 1997). Mixing of two ?uids with distinct temperature, pH, and redox characteristics is a process that is likely to have widespread applicability to ore genesis in a variety of crustal settings. Another important is the Olympic Dam described in Box 3.1. Fluid/rock reactions (pH and Eh controls) A widespread manifestation of hydrothermal mineralization processes is the development of alteration mineral assemblages in and around the ?uid conduit. Alteration is caused by the reaction between ?uids and their wall rocks and is a com- plex process that has been extensively studied, both in ore genesis research and for the under- standing of metamorphic and metasomatic (mass transfer) processes in the crust. The interaction that occurs between a ?uid and its wall rock promotes metal precipitation because it is yet another process that changes the prevailing ?uid properties, especially in terms of acidity (pH) and redox state. Reactions describing the factors con- trolling metal precipitation were presented in sec- tion 3.5.1 above. Quanti?cation of the effects of changes in pH and redox state on, for example, gold solubility is demonstrated in Figure 3.14, where calculated gold solubility contours for the dominant Au(HS) 2 - complex, as well as for the less important AuCl 2 - complex, in a ?uid at 200 °C are shown. A reduced, neutral ?uid like that at A, subject to an oxidation event akin to the one in reaction [3.4] above, would have its Au(HS) 2 - solubility dramatically reduced, as shown by the oxidation trend in Figure 3.14. This would almost certainly result in signi?cant precipitation of Au from the ?uid. Similiarly, a decrease in ?uid pH from around 7 to 3 would also result in a reduction of gold solubility and deposition of the metal. As described earlier, the precipitation of ores from metal–chloride complexes may occur under different conditions altogether. In Figure 3.14 it is clear that the precipitation of Au from an oxidized, acidic ?uid like that at B, in which the AuCl 2 - complex is stable, will best be achieved by increasing pH (perhaps by consumption of H + dur- ing alteration) or by reduction (perhaps caused by sulfate reduction to sul?de). The relationship between alteration and hydro- thermal ore-forming processes is a very important one and for this reason the topic is covered in more detail in section 3.6 below. In this section the link between ?uid/rock reactions and the precipitation of metals from ore ?uids is also discussed. 3.5.2 Adsorption Although metal precipitation is generally con- sidered to have been instigated by changes to the prevailing ?uid properties and accompanying reduction in equilibrium solubility, it is evident that ores can also form by adsorption of metal onto an existing mineral surface. Metal deposi- tion by adsorption can occur from ?uids whose concentrations are below their saturation levels and the process may, therefore, be important in certain ore-forming environments. Adsorption is de?ned as the adherence of an ion in solution to the surface of a solid (or mineral) with which it is in contact. The process occurs because a mineral surface will inevitably contain charge inbalances ITOC03 09/03/2009 14:36 Page 159160 PART 2 HYDROTHERMAL PROCESSES –30 Log fO 2 –45 –48 –51 24681 01 2 pH –42 –39 –36 –33 100 ppb 1 ppb AuCl 2 – B Neutralization A Acidification 10 ppb 1 ppb 100 ppb Pyrite Pyrrhotite Au(HS) 2 – Magnetite Hematite Oxidation Reduction Figure 3.14 Log fO 2 –pH diagram showing the stability of iron oxide and sul?de minerals in relation to gold solubility contours for Au(HS) 2 - and AuCl 2 - complexes in a ?uid at 200 °C (also at saturated water vapor pressure, with ?S = 0.01 and a Cl - = 1.0). Oxidation of ?uid A across the pyrite–hematite phase boundary will result in a four order of magnitude decrease in the solubility of the Au(HS) 2 - complex and likely precipitation of gold. The same result would occur if the ?uid became more acidic and pH decreased by some 5 units. The converse would apply with respect to AuCl 2 - in ?uid B. In this case Au precipitation would occur in response to an increase in pH (neutralization) or reduction of the ?uid (perhaps caused by an increase in the activity of H 2 S). Phase diagram and contours from Wood (1998) and S. Wood, personal communication. created by the fact that metal cations will not always be fully coordinated with anions such as O 2- or S - . Sites of high charge density (either posit- ive or negative) on a mineral surface represent the locations where adsorption of oppositely charged ions is likely to occur. Zones of high charge density are represented by features such as lattice defect sites, fracture planes, and trace element substitution sites. In general, a mineral surface in contact with an acidic solution (i.e. a high activity of H + ) will contain an abundance of positive charge and is likely to adsorb anionic complexes. Conversely, a mineral surface in contact with an alkaline solution (with a high activity of OH - ) will show a tendency for a surplus negative charge and will adsorb cations. Experiments have shown that adsorption of many different metals onto the surface of a range of silicate, oxide, and sul?de minerals can be an ef?cient way of forming an ore deposit. A variety of parameters affect adsorptivity and the most important of these are the surface area and surface properties of the adsorbent, as well as the pH of the solution (Rose and Bianchi-Mosquera, 1993). Temperature tends to have an inverse effect on adsorption and for this reason it is typically a low temperature phenomenon. Fine grained clay particles, as well as other materials with a high ITOC03 09/03/2009 14:36 Page 160HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 161 surface area to volume aspect such as diatoma- ceous earth, are very ef?cient adsorbers and are widely used in ?ltration and the removal of toxic matter from industrial ef?uent. With respect to ore deposition, however, it is the oxide and sul- ?de minerals that appear to play an important role in the concentration of metals from hydro- thermal ?uids. Since most metals form cationic species in nature, it follows that adsorption will be promoted by an increase in solution pH. This trend is demonstrated in Figure 3.15a, where the experimentally derived adsorption ef?ciency of Cu onto goethite (FeO(OH)) is plotted for oxid- izing and reducing conditions. Under oxidizing conditions, such as those applicable to the forma- tion of red-bed hosted stratiform copper deposits (see Box 3.6), Cu will be adsorbed very ef?ciently from a CuCl + -bearing solution at pH greater than 6 (Rose and Bianchi-Mosquera, 1993). Under more reducing conditions Cu is only ef?ciently adsorbed under more alkaline conditions. Sim- ilar behavior to Cu is exhibited by other metals such as Pb, Zn, Co, and Ni (Figure 3.15b), although it is clear that the onset of ef?cient adsorption takes place at different pH values. Under the speci?c oxidizing conditions of the experiment, metals adsorb in the order Cu, Pb, Zn, Co, Ni with increasing pH. In a hydrothermal solution that is evolving along an acid neutral- ization trend caused, for example, by wall-rock alteration (see section 3.6 below) this sequence could account for any zonation evident in an ore deposit. Metal adsorption ef?ciency is clearly very sens- itive to pH, which is, in turn, mediated by the 100 % Au adsorbed 80 20 60 40 4681 0 (d) pH 100 % adsorption 80 20 60 40 468 1 4 (c) pH % adsorption 80 20 60 40 57 81 0 (b) pH % adsorption 100 20 60 40 57 81 0 (a) pH Pyrrhotite Pyrite AuHS° Mackinawite 10 12 2 Pyrrhotite (po) Sphalerite (sl) Pyrite (py) Pb (py) Cd (sl) Zn (po) Hg (po) 100 69 Cu Pb Zn Co Ni Goethite 80 69 Oxidized Reduced 25°C 1M NaCl CuCl + onto goethite Figure 3.15 (left) Plots of percentage adsorption versus pH for (a) copper as CuCl + onto goethite under oxidizing and reduced conditions and (b) a variety of transition metals onto goethite. Solutions in both cases are at 25 °C and 1 M NaCl, and data are from Rose and Bianchi-Mosquera (1993). (c) Plots of percentage adsorption versus pH for different metals with respect to various mineral sul?de substrates (after Jean and Bancroft, 1986). (d) Percentage adsorption for gold as AuHS onto different sul?de minerals as a function of pH. Solutions are at 25 °C and 0.1 M NaCl (after Widler and Seward, 2002). ITOC03 09/03/2009 14:36 Page 161presence of surface OH - -related charge distribu- tion. Experimental data suggest the existence of an adsorption edge where an increase from low to virtually complete adsorption occurs over a nar- row range of pH increase. Adsorption ef?ciency is not, on the other hand, much affected by the type of mineral surface. In addition to goethite, which is a good adsorbent under oxidizing conditions (Figure 3.15a and b), sul?de minerals also attract metals to their surfaces in more reducing envir- onments. Sul?de minerals are typically charac- terized by an intrinsic negative surface charge usually attributed to the presence of S 2- (Shuey, 1975). Jean and Bancroft (1986) demonstrated that comprehensive adsorption of metals like Hg, Pb, Zn, and Cd can take place under alkaline condi- tions onto phases such as pyrite, pyrrhotite, and sphalerite (Figure 3.15c). Mercury is, in fact, read- ily adsorbed onto sul?de mineral surfaces over almost the entire pH range. Interestingly, the role of adsorption of gold (as AuHS) from solution by sul?de minerals exhibits the converse pattern of behavior to the base metals. Widler and Seward (2002) showed that the most ef?cient adsorption of Au onto phases such as pyrite and pyrrhotite occurs under acidic condi- tions, with a marked decrease in adsorbence at neutral to alkaline pH ranges (Figure 3.15d). The reason for this particular trend is that the AuHS complex is itself only stable at low pH, whereas under neutral to alkaline conditions the negat- ively charged Au(HS) 2 - complex is stable. Since a negatively charged complex will be repulsed by a negatively charged mineral surface, adsorption will drop dramatically as pH increases. The actual process by which metal adsorption occurs is very complex. Jean and Bancroft (1985) and Knipe et al. (1992) have suggested a two-stage process that is schematically illustrated in Figure 3.16. It is envisaged that adsorption is initiated by physical adsorption where the metal–ligand com- plex is loosely held at the mineral surface by weak van der Waals forces. There is no charge transfer at this initial stage. The second stage of the pro- cess is chemical and is promoted by reduction, or electron gain, and formation of covalent-like bonds between the metal ion and mineral sub- strate. Adsorbed metal atoms may diffuse across the mineral surface to form clusters at sites of high charge density (Figure 3.16). Further growth of the cluster could occur either by electrochem- ical precipitation or by conventional precipitation around the earlier formed metal nuclei. Many sul?de minerals are good metal adsorbents and this fact clearly has an important bearing on the formation of different ore deposit types. Adsorp- 162 PART 2 HYDROTHERMAL PROCESSES 3 Growth 2 Chemical adsorpton 1 Physical adsorpton Covalent bonding L L Au Au Van der Waals forces Electron transfer e – (Zone of high charge density) Mineral surface Au cluster Figure 3.16 Diagram showing the steps involved in adsorbing metal ions onto a mineral surface (after Jean and Bancroft, 1985; Knipe et al., 1992). This case considers a gold–ligand (Au–L) complex that adheres initially by physical adsorption to a negatively charged sul?de mineral surface by weak Van der Waals forces and subsequently by chemical adsorption and covalent-like bonding to form a stronger attachment. Adsorbed gold ions can diffuse along the surface to form clusters such as that shown in the adjacent SEM image (inset; after Jean and Bancroft, 1985). Further growth of the site of metal accumulation can result in the formation of discrete gold particles that adhere onto sul?de grains such as pyrite or arsenopyrite. ITOC03 09/03/2009 14:36 Page 162HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 163 tion as a metal-depositing process in ore-forming environments may have been widely underestim- ated in terms of its importance. Adsorption of toxic metals onto a variety of different substrates is also likely to become increasingly important in combating pollution and for this reason also warrants much closer study in the future. As a ?nal point, it should be noted that the charge imbalances on mineral surfaces give rise to the fact that many natural materials act as semi- conductors. Sul?de mineral surfaces, for example, exist as both n-type (negative or electron satur- ated) and p-type (positive or electron depleted) semiconductors (Shuey, 1975) and this in?uences metal precipitation onto these surfaces in much the same way as described above. 3.5.3 Biologically mediated processes of metal precipitation It is well known that the evolution of all higher life forms since the end of the Precambrian is characterized by the ability of cells to interact chemically with inorganic metals such as silicon, calcium, and phosphorus to form shells, teeth, and bones. The interaction of microorganisms with a variety of other metals is also important for the formation of authigenic mineral phases during diagenetic and hydrothermal processes. The concept of biomineralization is increasingly being recognized as a signi?cant process in ore formation, especially with respect to deposit types such as banded iron-formations, VMS-SEDEX ores, phosphorites, and supergene enrichments of Cu and Au. The following section brie?y considers the basic principles of biomineralization and its signi?cance to ore-forming processes. The fundamental subdivision of life into prokaryotes and eukaryotes came about with an improvement in the understanding of cell micro- biology in the mid-twentieth century (Patterson, 1999). Prokaryotes are the more primitive micro- organism and the ancestral life form on Earth, developing at some stage between 4650 Ma and about 3500 Ma when they ?rst appear in the fossil record. They are single celled microorganisms that can be subdivided into two kingdoms, namely Bacteria (or Eubacteria) and Archaea (or Archaebacteria). Archaea exist in extreme envir- onments, such as volcanic hot-springs and ocean ?oor hydrothermal vents (thermophiles) or hyper- saline evaporitic waters (halophiles), and utilize strange metabolic pathways, involving the energy derived from hydrogen- and sulfur-based chem- ical reactions, to nourish themselves. Bacteria include a huge variety of microorganisms that occupy virtually every ecological niche on Earth. Bacteria perform many different functions: some are fermentative (Escherichia coli) whereas others either contain the enzymes for oxygen meta- bolism (i.e. they are aerobic) or lack such enzymes (anaerobic). Eukaryotes evolved from prokaryotes and con- tain much more complex cellular structures, including a discrete nucleus, chromosomes, and characteristic cycles of cell division (mitosis and meiosis) and sexual reproduction. They probably appeared during Paleoproterozoic times although they could also have formed somewhat earlier (Patterson, 1999). They are almost all aerobic and metabolize oxygen in discrete packages within the cell called mitochondria, which are semi- autonomous chemical entities, containing DNA that is different from that of its own cell nucleus and more akin to prokaryotic DNA. This has led to the suggestion that more advanced cellular structures evolved by symbiotic inclusion of aerobic bacteria (i.e. mitochondria) into primitive eukaryotic entities. All higher life forms (the Animalia, Plantae, Fungi, and Protista kingdoms) consist essentially of eukaryotic cells. Biomineralization A capacity to chemically interact with metals is an inherent characteristic of most microorgan- isms. Mediation by microorganisms in the forma- tion of a variety of authigenic mineral phases is referred to as “biomineralization” and the process has important implications for ore formation. The major animal phyla incorporated biomineraliza- tion as a “lifestyle” soon after their formation during the Cambrian explosion. It is this process that allowed the more advanced life forms to ITOC03 09/03/2009 14:36 Page 163develop an exoskeleton and teeth and, eventually, internal bones (the vertebrates). It is, however, apparent that biomineralization occurred much before the Cambrian Period and it has been sug- gested that the formation of magnetite in bacteria (the so-called magnetotactic bacteria) might rep- resent an ancestral template for unraveling the genetics of this process in higher life forms (Kir- schvink and Hagadorn, 2000). It is the bacterial kingdom that is implicated in many of the pro- cesses linked to biologically mediated precipita- tion of metals during ore formation. Biomineralization occurs in two ways (Lowenstam, 1981; Konhauser, 1998, 2003). The ?rst is termed “biologically induced biomineral- ization” and refers to situations where minerals form as a consequence of the activity of micro- organisms on their immediate environment. The second process is “biologically controlled biomineralization,” in which the microorganism directly controls the uptake of inorganic material, either within the cell (where elements diffuse through the cell wall) or at the interface between the cell and its surrounds. These two processes are brie?y described below with respect to iron, which is particularly relevant to many ore- forming environments. Biologically induced biomineralization is con- trolled by the same equilibrium principles that govern metal precipitation in inorganic chem- ical systems. The metabolism of microorganisms effects the properties (i.e. pH, Eh) of aqueous solu- tions in their immediate environment and can promote the precipitation of authigenic minerals as a function of their solubility products. More importantly, ionized cell surfaces can induce mineral nucleation in much the same way as discussed in the previous section on adsorption. A wide range of different minerals can be biolo- gically induced because the phases that form re?ect the composition of the aqueous solution in which the microbe exists. Ferric hydroxide (or ferrihydrite Fe(OH) 3 ) is one of the most common biominerals, forming in a variety of different envir- onments, and is often upgraded during diagenesis to more stable phases such as goethite and hem- atite. Precipitation of iron occurs in progressive stages, from Fe-staining of the cell walls, to nuclea- tion of small (<100 nm) ferric hydroxide grains and eventual replacement of bacterial colonies by the inorganic authigenic phase (Konhauser, 2003). There are a number of bacteria that passively induce ferric iron precipitation onto their cell walls to form ochreous accumulations, a common one being Leptothrix ochracea, as well as the Crenothrix, Clonothrix, and Metallogenium genera. Marine hydrothermal vent sites are characterized by abundant iron-depositing microbial popula- tions. It has also been suggested that Precambrian oceans were characterized by bacteria capable of binding both ferric iron and silica and that this biologically mediated process contributed to the formation of banded iron-formations (see Chap- ter 5). In the present-day surface environment, the bacterium Thiobacillus ferrooxidans is capable of oxidizing ferrous iron, creating low pH solutions that have environmental implications during the formation of acid mine drainage from sul?de mineral-bearing ore deposits and mines. The fundamental problem associated with acid mine drainage is the production of sulfate and toxic labile metals in very acid solutions, as expressed in terms of reaction [3.5]: FeS 2 + 8H 2 O - Fe 2+ + 2SO 4 2- + 16H + [3.5] This problem can, however, also be remediated by bacterial activity, as discussed in the section on sulfate-reducing bacteria (SRB) below. Biologically controlled biomineralization is characterized by the creation of an area internal to the cell into which ions speci?c to the require- ments of the microorganism are diffused, thereby creating localized conditions where concentra- tions are increased until saturation is reached. Minerals are thus able to form even though the aqueous milieu surrounding the microbe would not be able to precipitate the same phase. This process is responsible for the formation, for example, of linear arrays of magnetite crystals within magnetotactic bacteria, which enables them to navigate to preferred redox environ- ments. The initial uptake of metal in this case is as ferric iron, with subsequent reduction to the 164 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 164HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 165 ferrous state within the cell, but prior to transport into the magnetite bearing membrane. Fossil magnetotactic bacteria have been recognized from the 2000 Ma Gun?int Iron Formation, and this represents the oldest evidence for controlled biomineralization. In terms of ore formation, the role that many bacteria play is to catalyze the oxidation of soluble metals to insoluble oxide phases. One par- ticular group of anaerobic bacteria, the so-called “sulfate-reducing bacteria” (SRB), contribute to ore formation in a different way. SRB, such as Desulphovibrio desulphuricans, oxidize organic molecules by using sulfate as an electron accep- tor, a process that generates dissolved sul?de (HS - or H 2 S). On diagenesis the sul?de reacts with other metals or minerals to form sul?de phases that are either directly implicated in mineraliza- tion, or form the precursors to later ore-forming processes. These low temperature processes can result in the formation of many different sul?de minerals, including pyrite (via mackinawite and greigite) as well as nickel, lead, and zinc sul?des (Konhauser, 2003). In a modern setting SRB can be useful in the microbial treatment of pollution and acid mine drainage. With reference to reaction [3.5] above, SRB would catalyze the reduction of the sulfate to form sul?de, which in turn reacts with and stabilizes labile toxic metals. A bicar- bonate ion (HCO 3- ) is also commonly produced during these reactions, which has the effect of neutralizing the previously acid waters. This type of biological remediation of aqueous pollutants forms the basis for many operations that have to treat both acidic and metal toxic waters. Biomineralization is a process that has consid- erable applicability to many other ore-forming environments in addition to those involving iron. Manganese is another redox-sensitive element that is biogenically precipitated in much the same way as iron. Manganous oxides such as birness- ite and vernadite are the main phases formed in the presence of bacteria. Bacterial concentrations of Mn and Fe on the ocean ?oor have been im- plicated in the formation of manganese nodules (see Chapter 5). The formation of phosphate-rich sediments has also been linked, at least in part, to biogenic mediation, with apatite overgrowths having been observed replacing the remains of cyanobacteria. Large-scale phosphate deposits (phosphorites; see Chapter 5) forming along the eastern seaboard of Australia have been attributed to the progressive assimilation of sea water phos- phorus by benthic bacteria (Konhauser, 2003). The Figure 3.17 Microphotographs of (a) bacterial cell with amorphous ferric hydroxide replacing the cell walls, and (b) magnetotactic spirillum bacteria showing a chain of magnetite crystals developed along the longitudinal axis of the microorganism (photographs courtesy of Kurt Konhauser and Dennis Bazylinski). (a) (b) ITOC03 09/03/2009 14:36 Page 165amorphous silica that forms the characteristic siliceous sinters above epithermal gold deposits is also attributed to the presence of a diverse range of bacterial genera, the ?laments of which act as nucleation sites for silici?cation. The siderophile metals (Au and Pt) are also amenable to biogen- ically mediated concentration, particularly with respect to the formation of nuggets in placer and lateritic environments (Watterson, 1991; see Chapter 4). One of the most spectacular demon- strations of the role of bacteria in the concentra- tion of metals is derived from sulfur and lead isotopic determinations of sphalerite and galena ores from the world-class Navan deposit in Ireland. Although also characterized by replacement and epigenetic textural characteristics, the observation of pyritized worm tubes from a related deposit at Silvermines (Boyce et al., 1999) has shown that the Irish-type Zn–Pb–Ba ores are SEDEX types and related, at least in part, to syngenetic hydro- thermal venting on the sea ?oor (see section 3.8.2). Isotopic studies indicate that some 90% of the sul?des at Navan were derived by the bacterial reduction of Carboniferous sea water sulfate (Fallick et al., 2001). It is argued that without an ample supply of reduced sulfur the Navan deposit might not have formed to the size that it is, emphasizing the importance of biologically medi- ated ore-forming processes. Bacterial biomineralization is an extremely important process that has wide implications for the origin and evolution of life itself, and also for the formation of a wide variety of authigenic minerals in diagenetic and hydrothermal environ- ments. In terms of ore-forming processes its role has probably been underestimated and this is likely to be a fruitful area of research in the future. 3.6 MORE ON FLUID/ROCK INTERACTION – AN INTRODUCTION TO HYDROTHERMAL ALTERATION The passage of hydrothermal solutions through the Earth’s crust is invariably accompanied by alteration. Alteration of a rock by the ?uid pass- ing through it is marked by the development of a mineral assemblage that is different from the original one and re?ects the rock composition, as well as the properties and amount of ?uid that has traversed the system. Zones of alteration mark the pathways of hydrothermal ?uids through the crust and may represent useful guides for the exploration of many ore deposit types. The nature of hydrothermal alteration also provides an indi- cation of the ?uid properties associated with ore formation and there is commonly a close relation- ship between the chemical reactions involved in alteration and those responsible for metal deposi- tion. It is, therefore, appropriate that a summary of the characteristics of hydrothermal alteration accompanies any description of ore-forming pro- cesses. The subject is, however, a complex one and the following section is necessarily brief. More detailed accounts documenting the evolution of ideas regarding hydrothermal alteration are pro- vided in the relevant chapters of the three editions of H.L. Barnes’s Geochemistry of Hydrothermal Ore Deposits (namely Meyer and Hemley, 1967; Rose and Burt, 1979; Reed, 1997). Traditionally, alteration has been regarded as a process that, in its simplest form, involves both the hydrothermal ?uid itself (essentially water with its dissociated components H + and OH - ) and the dissolved constituents of the aqueous solu- tion. The low to moderate degrees of pervasive, isochemical alteration that characterize almost any rock evolving in the presence of a ?uid can largely be explained by hydrolysis or H + ion meta- somatism (see also Chapter 4, section 4.2.2). This process can be depicted in terms of simple chem- ical equilibria, or reactions, although it must be remembered that these are likely to be simpli?ca- tions of actual alteration processes. For example, one of the most common forms of alteration recognized in nature is the reaction of K-feldspar with water to form muscovite or sericite. This reaction can be expressed as follows: 3 /2KAlSi 3 O 8 + H + ? (k-fel) 1 /2KAl 3 Si 3 O 10 (OH) 2 + 3SiO 2 + K + [3.6] (musc) (qtz) The hydrolysis of K-feldspar to muscovite in terms of this reaction requires nothing more than the presence of H + ions in an aqueous solution. 166 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 166The reaction is isochemical and no new ingredi- ents need to be added to the system. If the reaction proceeds to the right, H + ions are consumed and the ?uid will become more basic, and will con- tinue to do so until the K-feldspar is used up. The reaction will also produce quartz as part of the alteration assemblage, as well as K + ions dissolved in the aqueous solution. If the products of reaction [3.6] are permitted to further react with H + ions in solution such that the system undergoes an increase in ?uid/rock ratio, then muscovite would react to form kaolin- ite, as follows: KAl 3 Si 3 O 10 (OH) 2 + H + + 3 /2H 2 O ? (musc) 3 /2Al 2 Si 2 O 5 (OH) 4 + K + [3.7] (kaol) This reaction illustrates the way in which increas- ing ?uid/rock ratios will change the alteration mineralogy by continued reaction with the rock after one set of mineral buffers have broken down (i.e. by consumption of K-feldspar in the case of reaction [3.6]). Alteration does not, however, necessarily result in the formation of only one mineral reactant, but generally forms an assem- blage of phases. Plagioclase, for example, occur- ring in the same rock that originally contained the K-feldspar, might also react with H + ions in solution to form pyrophyllite, as follows: 2NaAlSi 3 O 8 + 2H + ? Al 2 Si 4 O 10 (OH) 2 + 2SiO 2 + Na + (plag) (pyroph) (qtz) [3.8] Thus, reaction of a simple mineral assemblage with nothing but water could result in a multi- component alteration assemblage, even if the system remains isochemical. If consideration is now given to an open chem- ical system where the dissolved constituents of an aqueous solution are also involved in altera- tion, then the processes and their products will be still more complex. The isochemical reactions above ([3.6], [3.7], and [3.8]) describing hydrogen ion metasomatism pertain to situations where reactants and products have similar bulk composi- tions. All three reactions also re?ect acid–base exchange because an aqueous H + ion is consumed and replaced by cations from the reactants. The accumulation of aqueous base cations in the evolving ?uid will have an effect on the nature of downstream alteration, as they may themselves react with wall-rock. This type of alteration pro- cess is known as cation metasomatism. A wide variety of alteration products can form by cation metasomatism, examples of which are: 2NaAlSi 3 O 8 + 4(Mg,Fe) 2+ + 2(Fe,Al) 3+ + 10H 2 O ? (plag) (Mg,Fe) 4 (Fe,Al) 2 Si 2 O 10 (OH) 8 + 4SiO 2 + 2Na + + 12H + (chl) (qtz) [3.9] and 3CaMg(CO 3 ) 2 + 4SiO 2 + 6H + + 4H 2 O ? (dol) Mg 3 Si 4 O 10 (OH) 2 + 6H 2 CO 3 + 3Ca 2+ [3.10] (talc) In reaction [3.9] the presence of Mg, Fe, and Al cations in a ?uid which equilibrates with plagioclase will result in the chloritization of that mineral, with accompanying precipitation of quartz and signi?cant H + production. This type of reaction will result in the ?uid becoming more acidic during progressive alteration, which is con- trary to the case for hydrolysis reactions. Reaction [3.10] demonstrates the sort of process which takes place when a cation-bearing aqueous solu- tion interacts with a carbonate mineral, such as dolomite, to form a silicate mineral, such as talc. These processes are relevant to the formation of skarn ore deposits where fertile granites intrude carbonate country rocks, and are referred to as silication (see section 3.6.1 below). More recent considerations of alteration pro- cesses have suggested that they should be regarded not only in terms of chemical equilibria, but also as dynamic systems that evolve with time. Given the time scale of most geological events, the attainment of equilibrium between rock and ?uid is generally feasible, at least on a local scale. In dynamic systems, however, it is likely that kinetic considerations predominate over those of steady state equilibrium. The mineral assemblages HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 167 ITOC03 09/03/2009 14:36 Page 167that form as a result of hydrothermal alteration vary as a function of ?ve factors (Reed, 1997), namely: • temperature; • pressure; • host rock composition; • ?uid composition; and • the ratio of ?uid to rock in the alteration pro- cess (termed the ?uid/rock ratio). In nature these ?ve factors interact in a complex way to create alteration assemblages that re?ect all of them. The stabilities of mineral assem- blages are dictated largely by temperature and pressure, whereas the host rock composition controls, to a certain extent, which ingredients are available to make up a particular alteration assemblage. Fluid composition and temperature have a strong in?uence on the nature and extent to which the host rock is dissolved by the aqueous solution, as well as the nature and concentration of new constituents introduced into the system by the ?uid. The ?uid/rock ratio is considered to be the “master variable” (Reed, 1997) in alteration processes because it dictates the extent to which reactions will move to completion. Reed (1982, 1997) has shown that, because hydrothermal ?uids migrate along pathways that represent evolving chemical gradients, alteration should be viewed in a dynamic sense as a function of changing ?uid/rock ratio. This concept is schematically illustrated in Figure 3.18a, showing a fracture along which a hydrothermal ?uid is ?owing and adjacent to which alteration is taking place by diffusion-related reactions between ?uid and country rock. At any instant in time, alter- ation pinches out downstream along the fracture because of the progressively diminishing ability of the ?uid to react with the country rock as its enthalpy and its dissolved ingredients are effec- tively “used up” (e.g. an originally acidic ?uid may become progressively neutralized as its H + ions are consumed by acid–base exchange; see below). The sequence of alteration minerals out- wards or downwards of the fracture or vein re?ects the sequence of reactions that relate to decreas- ing ?uid/rock ratios (Figure 3.18a). The upstream portion of the vein has effectively “seen” more ?uid (and is, therefore, characterized by a higher ?uid/rock ratio) than the distal sections of the system, and this is re?ected in a variable alter- ation assemblage. The evolution of alteration assemblages in a dynamic hydrothermal system can be modeled by reacting a ?uid of known initial composition, pH, and temperature with a rock, using constraints imposed by established thermodynamic, chem- ical, and equilibrium data (Reed, 1997). The ?uid composition and pH is then changed as the con- stituents of chemical reactions are consumed (i.e. as a mineral buffer system breaks down). Figure 3.18b shows how the alteration assemblage changes in a rock of andesitic composition subject to reaction with a hydrothermal solution. The scenario is modeled in terms of a system with high initial ?uid/rock ratios and an acidic (low pH) ?uid, evolving to lower ?uid/rock ratios and alkaline (high pH) conditions. The plot shows that an acidic ?uid circulating with high ?uid/ rock ratios through the andesite will be charac- terized by an alteration assemblage comprising quartz + pyrophyllite +alunite, also referred to as an advanced argillic alteration assemblage (Figure 3.18b). As the alteration systematics evolve, the same rock, now in?ltrated by a more alkaline ?uid circulating at a lower ?uid/rock ratio, would be altered, initially to a sericite + chlorite assemblage and then to a chlorite + epidote + muscovite +albite assemblage. The latter assemblage is termed propylitic alteration and also matches the typical greenschist facies metamorphic assemblage. The expected distribu- tion of ore minerals in the evolving system is shown in Figure 3.18c. Viewing alteration assemblages in terms of dy- namic ?uid/rock interaction offers many advant- ages to the understanding of alteration processes, as well as to explaining the relationship between ?uid rock interactions and ore deposition. The implication of time into the models of Figure 3.18 provides a good explanation for zonation and paragenetic sequence, two hallmark character- istics of hydrothermal ore deposits, discussed in more detail below. An evolving alteration system that moves from high to low ?uid/rock ratios 168 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 168HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 169 Diffusion halo (a) Fracture Fluid flow Decreasing f/r Decreasing f/r 0.8 Relative proportion 0.6 0.4 0.2 (b) Andesitic host 300°C pH = 1 py chl al qz ep pyro par micr ab ser- chl propylitic ca act mus anh Advanced argillic –1 Log moles kg –1 –2 –3 –4 (c) –2 –4 –6 –8 (d) 1 pH hem bar py sl cp gn po mt pH HS – HCO 3 – SO 4 2– pH 1.0 0.0 –1.0 –2.0 Log f/r SO 4 2– 2 3 4 5 6 7 8 9 Log molality HS – HCO 3 – Figure 3.18 Models illustrating the dynamic nature of ?uid/rock interaction and hydrothermal alteration as a function of changing ?uid/rock ratio. (a) Conceptual model showing one side of a diffusion-related alteration halo around a fracture through which ?uid is passing from left to right. As ?uid reacts with its host rock it changes composition and, through time, alteration assemblages evolve both along and outward of the fracture system (i.e. with decreasing ?uid/rock ratios). (b) The occurrence and relative proportions of the principal alteration minerals forming when a relatively oxidized aqueous solution (of magmatic derivation at 300 °C and with initial pH = 1) reacts with a rock of andesitic composition, as a function of changing ?uid/rock ratio (ser, sericite; chl, chlorite; ca, calcite; act, actinolite; micr, microcline; ab, albite; ep, epidote; par, paragonite; mus, muscovite; pyro, pyrophyllite; al, alunite; py, pyrite; anh, anhydrite; qz, quartz). (c) The precipitation sequence of some of the major ore minerals as a function of changing ?uid/rock ratio for the same setting as in (b) (po, pyrrhotite; mt, magnetite; hem, hematite; bar, barite; sl, sphalerite; cp, chalcopyrite; gn, galena). (d) The evolution of ?uid pH, as well as concentrations of certain key ligands in the ?uid, as a function of changing ?uid/rock ratio (also for the same setting as b). All ?gures after Reed (1997). ITOC03 09/03/2009 14:36 Page 169(i.e. from left to right in Figure 3.18) re?ects a change from cation to hydrogen ion metasomatism, and from ?uid-dominated to rock-dominated en- vironments. The evolving system also dictates the compositional changes that occur in the ?uid itself, such as the stepwise increase in ?uid pH as H + ions are consumed by wall-rock reaction and stabilization of successive mineral assemblages (Figure 3.18d). The redox state of the ?uid also changes as the alteration system evolves. Figure 3.18d shows that with decreasing ?uid/rock ratio the concentration of SO 4 2- in the ?uid decreases as the HS - content increases. This is because SO 4 2- is reduced by reaction with ferrous wall-rocks, resulting in the precipitation of hematite in the alteration zone and fractionation of sul?de into the evolving aqueous solution (Reed, 1997). Sul- fate reduction is an important source of sul?de in many hydrothermal ore-forming systems and promotes base metal sul?de precipitation. It is the changes in pH and redox state of the ?uid that are so important in promoting the precipitation of metals from ore ?uids. There are many examples of situations where reaction between a hydrothermal ?uid and its wall rocks is the main cause of ore precipitation. This process can be recognized, for example, in samples that reveal textures illustrating the replacement of primary minerals in the host rock by ore minerals. Such a texture is distinct from situations where ore minerals are precipitated into open spaces such as fractures or vugs, where other mechanisms of ore deposition, such as ?uid mixing or boiling, might have prevailed. Skarn deposits (see section 2.10, Chapter 2) are a particularly good example of where alteration of carbonate rocks by acidic, metal-bearing solu- tions yields Ca–Mg silicate minerals accompan- ied by signi?cant polymetallic mineralization. Carbonate rocks do, of course, have the ability to neutralize acidic solutions and the pH increases that arise from ?uid/rock reactions in this environment promote ef?cient precipitation of metal sul?des (Reed, 1997). Likewise, greenstone belt hosted mesothermal lode-Au deposits are typically accompanied by extensive zones of quartz–albite–carbonate–muscovite–pyrite altera- tion in and around the shear zone or fault along which ?uids were focused (see section 3.9.1 below and Box 3.2). A common Au precipitation mecha- nism in these environments is the reaction of a H 2 O–CO 2 , sul?de-rich ?uid with ferrous host rocks (commonly basalt or banded iron-forma- tion) to form pyrite (Phillips, 1986). This causes destabilization of the dominant AuHS 2 - complex and precipitation of native Au together with the sul?de minerals. The intimate association between alteration and hydrothermal ore-forming processes is well established. The recognition of alteration styles and parageneses is now a standard tool in the array of techniques utilized by exploration geolo- gists and has important bearing on the nature of the ore deposit itself. 3.6.1 Types of alteration and their ore associations This section outlines some of the characteristics and ore associations of the main alteration styles associated with a variety of hydrothermal environments. Potassic alteration Potassic (or K-silicate) alteration is characterized by the formation of new K-feldspar and/or biotite, usually together with minor sericite, chlorite, and quartz. Accessory amounts of magnetite/ hematite and anhydrite may occur associated with the potassic alteration assemblage. It typi- cally represents the highest temperature form of alteration (500–600 °C) associated with porphyry Cu-type deposits, forming in the core of the system and usually within the granite intrusion itself. A degree of K + (cation) metasomatism in addition to hydrolysis is considered to have taken place during the formation of potassic alteration assemblages because studies of bulk composi- tions on a volume basis suggest that potassium has been added to the system. A variation of potassic alteration involving substantial addition of Na and Ca (called sodic and calcic alteration and characterized by abundant albite, epidote and actinolite) is documented in some porphyry sys- tems such as Yerington, Nevada, USA. 170 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 170The “Golden Mile” is a zone of exceptional gold mineral- ization located immediately to the east of the twin cities of Kalgoorlie–Boulder in Western Australia. Gold was dis- covered here in 1893 and well over 1200 tons of gold has been won from the area since then. Until 1975 mining was carried out in several underground operations, a situation which led to a steady decline in production due to increasing depths and decreasing grades. This situ- ation was reversed in the late 1980s when ownership was amalgamated and plans initiated to open a huge, low-grade open-pit operation known as the Fimiston or Super Pit. This is now the largest gold mining operation in Australia, producing well over 800 000 oz of gold per year. The Golden Mile is located in the Norseman–Wiluna greenstone belt of the Yilgarn Craton. Auriferous mineral- ization is structurally controlled and hosted mainly in steeply dipping shear zones or lodes that are centered on the Golden Mile Fault (Figure 1a). Hundreds of subsidiary lodes are concentrated in and around this system and are hosted by a 2675 Myr old ferruginous tholeiitic intrusion (sill) called the Golden Mile dolerite. Although the initial high-grade, largely underground, workings were concen- trated on the individual shear zones themselves, the system is accompanied by an extensive alteration halo of low grade mineralization which now forms the bulk of the ore in the Super Pit. It is the alteration of doleritic host rock by mineralizing, aqueo-carbonic hydrothermal solutions that is believed to be the main cause of the pervasive gold HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 171 Alteration and metal precipitation: the Golden Mile, Kalgoorlie, Western Australia – an Archean orogenic gold deposit (a) (b) N Metasediments Golden Mile dolerite Metavolcanics f f Golden Mile fault 01 km Lode System "Super Pit" f Boulder F. Eastern Trafalgar F. Western Lode System f Adelade F. Golden Pyke F. Kalgoorlie Quartz-pyrite- gold shear zone East Lode Carbonate zone Chlorite zone 0 100 m 50 Lake View Shaft Figure 1 (a) General geological outline of the Golden Mile, east of Kalgoorlie, Western Australia, and the approximate position of the “Super Pit,” currently Australia’s largest gold mining operation. (b) The nature of hydrothermal alteration in the Golden Mile dolerite. The East Lode was originally mined as a high grade underground deposit through the Lake View Shaft. The entire alteration zone now forms the bulk of the ore for the high tonnage–low grade operation in the Super Pit. Both diagrams are after Phillips (1986). ITOC03 09/03/2009 14:36 Page 171172 PART 2 HYDROTHERMAL PROCESSES mineralization in this deposit (Phillips, 1986) and this is discussed in more detail below. A schematic indication of the regional geology of the Golden Mile, and the approximate position of the Super Pit, is shown in Figure 1(a). The nature of the alteration halo around quartz–pyrite–gold mineralized shear zones (such as the East Lode at the Lake View mine, which has now been subsumed by the Super Pit) is shown in Figure 1(b). The Golden Mile dolerite was regionally metamorphosed to a dominantly actinolitic assemblage prior to and during folding and ductile deformation of the Norseman–Wiluna greenstone belt. Metamorphism was superseded by intro- duction of a reduced, low-salinity aqueo-carbonic ?uid that percolated along hundreds of shear zones cutting pre- ferentially through the relatively competent Golden Mile dolerite, progressively altering the host rock in and around them. Alteration is markedly zoned around individual shear zones, as well as on a broader scale, and is charac- terized by a paragenetic sequence that commences with early (metamorphic) actinolite and then progresses suc- cessively to chlorite-, carbonate-(siderite), and eventually pyrite-dominated assemblages (Phillips, 1986). Chlorite (with lesser carbonate, albite, and quartz) alteration forms a pervasive halo extending for several hundred meters around zones of shearing, and overprints the regional metamorphic assemblage. Carbonate assemblages re?ect a more intense manifestation of alteration (i.e. higher ?uid/rock ratio) and occur closer to individual shear zones (i.e. in halos extending up to 5 m away from the shear) or in more ferruginous parts of the differentiated dolerite sill. The shear zone itself is demarcated by a pyrite–quartz– muscovite–carbonate assemblage with which high-grade gold mineralization occurs. Phillips (1986) has argued that the alteration paragenesis is a product of a single meta- morphic ?uid that introduced CO 2 , K, S, and Au into the system, and that alteration and mineralization were syn- chronous. Precipitation of gold is considered to have been facilitated by the interaction of a reduced auriferous ?uid with Fe-rich host rocks, such as in the following reactions: FeCO 3 + 2H 2 S + 1 / 2 O 2 ? FeS 2 + CO 2 + 2H 2 O (i.e. siderite replaced by pyrite) and FeCO 3 + Au(HS) 2 - ? FeS 2 + CO 2 + H 2 O + Au (gold bisul?de complex destabilized by pyrite formation) This model emphasizes the role that host rock alteration plays on metal precipitation, as described in section 3.5.1 of this chapter. Although the process is undoubtedly very important in the formation of many hydrothermal ore deposits, it seldom takes place in isolation. At Golden Mile, for example, it is now recognized that the late-stage, high grade gold–tellurium–vanadium ores that characterize very rich vein systems such as the Oroya shoot were the product of the mixing of a (possibly magmatic) oxidized, SO 2 -bearing ?uid with the more reduced metamorphic ?uid that was responsible for the pervasive alteration described above (J. Walshe, Fluid mixing processes also appear to be broadly coin- cidental with the transition from a ductile to a brittle deformational regime, a changeover that is commonly associated with signi?cant mineralization in orogenic gold systems. Phyllic (or sericitic) alteration This alteration style is very common in a variety of hydrothermal ore deposits and typically forms over a wide temperature range by hydrolysis of feldspars to form sericite (?ne-grained white mica), with minor associated quartz, chlorite, and pyrite (see reaction [3.6] and Figure 3.18). Phyllic alteration is associated with porphyry Cu deposits, but also with mesothermal precious metal ores and volcanogenic massive sul?de deposits in felsic rocks. Propylitic alteration Propylitic alteration is probably the most wide- spread form of alteration in that it is essentially indistinguishable from the assemblages which form during regional greenschist metamorphism. It comprises mainly chlorite and epidote, together with lesser quantities of clinozoisite, calcite, zoisite, and albite. It is a mild form of alteration representing low to intermediate temperatures (200–350°C) and low ?uid/rock ratios (see Figure 3.18). This style of alteration tends to be isochemical and forms in response to H + metasom- atism. It characterizes the margins of porphyry Cu deposits as well as epithermal precious metal ores. Argillic alteration This alteration style is commonly subdivided into intermediate and advanced categories depending ITOC03 09/03/2009 14:36 Page 172on the intensity of host mineral breakdown. Intermediate argillic alteration affects mainly plagioclase feldspars and is characterized by the formation of clay minerals kaolinite and the smectite group (mainly montmorillonite). It typ- ically forms below about 250°C by H + meta- somatism and occurs on the fringes of porphyry systems. Advanced argillic alteration represents an extreme form of base leaching where rocks have been stripped of alkali elements by very acidic ?uids active in high ?uid/rock ratio envir- onments (see Figure 3.18). It is characterized by kaolinite, pyrophyllite, or dickite (depending on the temperature) and alunite together with lesser quartz, topaz, and tourmaline. It is commonly associated with near surface, epithermal precious metal deposits where alteration is associated with boiling ?uids and condensation of volatile-rich vapors to form extremely acidic solutions. Silication Silication is the conversion of a carbonate mineral or rock into a silicate mineral or rock (see reaction [3.10] above) and necessarily involves introduc- tion of additional components into the system (cation metasomatism). It is the main process which accompanies the prograde stage in the formation of polymetallic skarn deposits which develop when a fertile, acidic, magmatic ?uid in?ltrates a carbonate host rock. Carbonate rocks are a particularly ef?cient host for metal deposi- tion from hydrothermal solutions because of their ability to neutralize acidic ?uids and their “reactivity,” which enhances permeability and ?uid ?ow. Skarn reactions and the formation of “calc–silicate” mineral assemblages are very complex and develop over an extended temperat- ure range. Fluid/rock ratios also develop to very high values in skarn systems and these factors result in the precipitation of a diverse, polymetal- lic suite of ores. Silici?cation Silici?cation should not be confused with silica- tion and refers speci?cally to the formation of new quartz or amorphous silica minerals in a rock during alteration. Minor silici?cation develops in the alteration halos associated with many differ- ent ore deposit types and is usually a by-product of isochemical hydrolysis reactions where Si is locally derived (see reactions [3.6] and [3.8] above). The majority of fractures through which hydrothermal ?uids have passed are at least partially ?lled with quartz to form veins. The Si in these settings is usually derived by leaching of the country rocks through which the ?uids are circulating. Intense silici?cation, however, forms as a result of cation metasomatism, where sub- stantial addition of Si 4+ in solution is added to the system. This type of alteration is characteristic of the sinter zones in high level epithermal precious metal ore deposits. Carbonatization Carbonatization refers to the formation of car- bonate minerals (calcite, dolomite, magnesite, siderite, etc.) during alteration of a rock and is promoted by ?uids characterized by high partial pressures of carbon dioxide (P CO 2 ) and neutral to alkaline pH. Archean greenstone belt related lode-gold deposits represent an ore deposit type where carbonate alteration is virtually ubiquitous and is accompanied by an assemblage comprising quartz, muscovite, biotite, albite, and chlorite. It forms when reaction occurs between a low- salinity, CO 2 -rich ?uid and its host rock. The carbonate mineral that forms is a function of the composition of the host rock and could be dolomite in association with amphibolite, siderite in a banded iron-formation, or calcite in a granitic host. Greisenization The formation of a greisen is speci?c to the cupola zones of highly differentiated (S-type) granites that contain Sn and W mineralization, as well as signi?cant concentrations of other incompatible elements such as F, Li, and B. Greisens represent an alteration assemblage comprising mainly quartz, muscovite, and topaz, with lesser tour- maline and ?uorite, usually forming adjacent to quartz–cassiterite–wolframite veins. HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 173 ITOC03 09/03/2009 14:36 Page 173Hematitization Alteration that is associated with oxidizing ?uids often results in the formation of minerals with a high Fe 3+ /Fe 2+ ratio and, in particular, hematite with associated K-feldspar, sericite, chlorite, and epidote. In the magmatic-hydrothermal environ- ment, occurrences such as the granitoid-hosted, Olympic Dam-type Cu–Au–Fe–U deposit in South Australia are characterized by this style of alteration. It also typi?es other ore bodies, includ- ing the stratiform, sediment-hosted Cu–Co ores of the Central African Copperbelt. This style of alteration appears to be related to redox processes where highly saline, oxidizing ?uids come into contact with a more reduced host rock environ- ment, or mix with more reduced ?uids. 3.7 METAL ZONING AND PARAGENETIC SEQUENCE A characteristic feature of many hydrothermal ore deposits is the occurrence of a regular pattern of distribution, or zoning, of metals and minerals in space. Zoning can be observed at many differ- ent scales, ranging from regional patterns of metal distribution (at a scale of hundreds of kilometers), through a district scale, to individual ore-body related variations and even down to the level of a single vein or hand specimen (Smirnov, 1977; Guilbert and Park, 1986; Pirajno, 1992; Misra, 2000). A good example of regional metal zonation is the Andes metallogenic province, where the distributional pattern of ore deposits is related to continental scale subduction along the west- ern margin of South America. At a district scale, however, the distribution of metals is more tightly constrained and better de?ned, and com- monly characterized by a consistent pattern that is reproduced in other ore-forming environ- ments. District scale zoning is well exempli?ed by the Sn–W–Cu–Pb–Zn–Ag–Sb–U sequence of ores associated with the Cornubian batholith in southwest England (see Box 2.2, Chapter 2), a pat- tern that is reproduced in other granite-related ore-forming systems. Zonation at this and smaller scales is clearly related to the systematic evolu- tion of a hydrothermal ?uid (i.e. alteration) and the sequential precipitation of its contained metal budget by the types of processes discussed in sec- tion 3.6 above. Although the processes control- ling metal precipitation are complex and varied, the relatively consistent pattern of metal zona- tion in so many different ore-forming environ- ments means that the process should be amenable to rational and uni?ed explanation. Although there is still some ambiguity in its use, the term paragenesis is now widely applied to the association of minerals and metals that char- acterizes a particular ore type and, therefore, has a common origin (Guilbert and Park, 1986; Kutina et al., 1965). Because both space and time need to be considered in the understanding of zonation, the phrase paragenetic sequence is now used to describe the distribution in time of a set of genetic- ally related minerals or metals. The recognition and signi?cance of parageneses and paragenetic sequences in hydrothermal ore deposits came from early observations of ore-forming environ- ments by some of the founders of modern eco- nomic geology, notably Waldemar Lindgren (1933). On the basis of these empirical observations Emmons (1936) devised the concept of a “recon- structed vein.” This was an idealized composite vein that extended from deep in the crust up to the surface and contained a paragenetic sequence of ore components that is typical of the zoning patterns observed in many different hydro- thermal ore deposit types. The “Emmons recon- structed vein,” revised by Guilbert and Park (1986), is summarized in Table 3.3. Observations suggest that metals such as Mo, W, and Sn will precipitate early (or at deep levels) from a high temperature hydrothermal solution. Such solutions are sometimes referred to as hypothermal. They are followed, in the ideal pre- cipitation sequence, by Cu, and then Zn, Pb, Mn, and Ag, as the ?uid in?ltrates upwards in the crust and cools to form mesothermal solutions. The precious and volatile metals such as Au, Sb, and Hg are typically observed to represent the lat- est stages of the sequence, forming from still cooler epithermal solutions circulating near the surface. Extensive studies of hydrothermal ore deposits leave little doubt that the Emmons sequence shown in Table 3.3 has general applicab- 174 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 174ility. Many exceptions, however, do exist and there is little doubt that the scheme can be much improved and extended in the light of modern observations and theory. The wide ranging applicability of the Emmons sequence suggests that the secular precipitation of metals might simply be related to the decline in solubility that generally accompanies temper- ature decrease. Sections 3.5 and 3.6 above, how- ever, have shown that metal solubilities and precipitation are controlled by a variety of com- plexly interacting variables that are related to both the ?uid itself and its surrounds. Barnes (1975) ?rst showed that if metals are considered to have been transported as metal–sul?de com- plexes then the relative stabilities of these com- plexes, when corrected for differing concentration of metals in solution, closely match the Emmons precipitation sequence. On this basis he identi?ed the following paragenetic sequence: Fe – Ni – Sn – Cu – Zn – Pb – Ag – Au – Sb – Hg By contrast, if metal transport occurs by metal– chloride complexation then the above sequence no longer applies and a subset of these metals is more likely to be precipitated according to the sequence: Cu – Ag – Pb – Zn This suggests that paragenetic sequences are likely to be controlled by more than just a single variable. Susak and Crerar (1981) demonstrated that paragenetic sequences need to be considered in terms of classes of metals that are grouped in terms of mineral stoichiometry and the valence state of the aqueous metal species. They went on to show that the precipitation sequence of metals from within one particular class will be con- trolled essentially by the Gibbs free energies of the associated minerals. The precipitation of metals from different classes would, however, be controlled by the environment within which the metal-bearing solution occurs, and this could be a HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 175 Table 3.3 The “Emmons Reconstructed Vein” concept showing empirical observations that relate to typical patterns of zonation and paragenetic sequences in many hydrothermal ore deposits Vein Surface (epithermal) Intermediate (mesothermal) Deep (hypothermal) Source: after Guilbert and Park (1986). Metal Barren Mercury Antimony Gold–silver Barren Silver–manganese Lead Zinc Copper–arsenic–antimony Copper Molybdenum–tungsten–tin Barren Ore mineralogy (in bold), and related alteration assemblages Chalcedony, quartz, barite, ?uorite, and carbonate minerals Cinnabar: chalcedony, quartz, barite, ?uorite, and carbonate minerals Stibnite: quartz Gold, electrum, acanthite: quartz, chalcedony, adularia, alunite, carbonate minerals, silici?cation, some potassic, phyllic, and propylitic alteration Quartz and carbonate minerals Acanthite, rhodochrosite: quartz and carbonate minerals, some phyllic, argillic, propylitic alteration Galena: quartz with minor carbonate minerals Sphalerite: quartz with occasional carbonate minerals, advanced argillic alteration Chalcopyrite, tennantite–tetrahedrite: quartz, phyllic, propylitic, argillic alteration Chalcopyrite: quartz, phyllic alteration Molybdenite, huebnerite, scheelite, cassiterite: quartz, potassic alteration Potassic alteration, anhydrite, carbonate minerals ITOC03 09/03/2009 14:36 Page 175more complex, multivariable process. Thus, a simple and reproduceable zonation pattern or paragenetic sequence is likely to apply only to hydrothermal ore-forming environments that consist of metals from the same zoning class. Any exceptions and reversals to the typical Emmons sequence are likely to be explained in terms of either disequilibrium, or precipitation of metals from different classes. The predicted precipitation sequence for metals in speci?cally de?ned zonation classes is illus- trated in Figure 3.19. The precipitation sequence for metals in solution from the same class can be read on this diagram in terms of decreasing Gibbs free energy values for a speci?c temperature. Thus, the precipitation sequence for metals in the MS (q = 2) zoning class is Mn–Zn–Cd–Pb–Hg for any temperature, but this sequence would not necessarily prevail under conditions where a dif- ferent stoichiometry and valence state prevails. Replacement processes In the previous section the precipitation of min- erals and metals was discussed in the context of aqueous solutions passing through open spaces in a rock, especially vugs and veins at shallow crustal levels. Different conditions apply to metals that 176 PART 2 HYDROTHERMAL PROCESSES 100 0 50 –?G° (kcal mol –1 ) 30 20 10 100 150 200 250 300 350 T (°C) 40 50 60 70 80 90 Precipitation sequence for metals of the same zoning class WS 2 MnS (alabandite) ZnS (sphalerite) FeS 2 (pyrite) CdS (greenockite) SnS (herzenbergite) FeS (pyrrhotite) PbS (galena) Ag 2 S (argentite) HgS (cinnabar) AuS 2 Cu 2 S (chalcocite) f MS (q = 2) MS 2 (q = 2) M 2 S (q = 1) Zoning classes Figure 3.19 Plot of Gibbs free energy versus temperature for a variety of sul?de minerals categorized according to their stoichiometry (where M is metal and S is sul?de) and the valence state of the dissolved metal species (q). The precipitation sequence for metals in solution from the same zoning class is predicted in terms of decreasing values of free energy (after Susak and Crerar, 1981). ITOC03 09/03/2009 14:36 Page 176are precipitated during the replacement of min- erals deeper in the Earth’s crust. Replacement occurs when the original minerals in a rock are dissolved and near simultaneous precipitation of secondary minerals occurs (Seward and Barnes, 1997). In contrast to the more permeable near- surface environment where ore deposition takes place by open space ?lling, metal precipitation deeper in the crust occurs as ?uids percolate along poorly interconnected microfractures and pore spaces (see section 3.3.4). Thus, the process of replacement will only persist if porosity is maintained and if the accompanying ?uid–rock reactions are characterized by a reduction in the molar volume of the mineral being replaced. An increase in molar volume would effectively block the through-?ow of ?uid as microfractures and pores are ?lled by secondary mineral precipitates whose volumes are greater than the material being replaced. Comparison of molar volumes shows that cal- cite, for example, can be replaced by most calc– silicate and sul?de minerals, a feature that is commonly observed in skarn deposits. By contrast, oxide minerals such as magnetite and ilmenite are less likely to be completely sul?dized since the relevant reactions will involve an increase in molar volume. At best, only partial replacement of oxide by sul?de would occur since the vol- ume increase, together with the accompanying decrease in microporosity that accompanies the reaction, would limit ?uid ?ow. The sequence of precipitation of secondary or ore minerals during replacement processes can best be predicted in terms of decreasing molar volumes of the re- spective solids involved, as discussed by Seward and Barnes (1997). The mineral sequence shown below, with relevant molar volumes, is useful for the understanding of replacement processes and the prediction of replacement sequences. Replacement will only be effective when min- erals of smaller volume replace those with larger volume. In the sequence presented below minerals will only comprehensively replace those occur- ring to their left. In addition, mineral replacement will presumably occur more readily the greater the volume de?cit between original and second- ary mineral. microcline ‹ scheelite ‹ anhydrite ‹ calcite [54] [47] [46] [37] ‹ kaolinite ‹ galena ‹ molybdenite ‹ siderite [33] [31] [30] [29] ‹ cinnabar ‹ muscovite ‹ arsenopyrite [28] [28] [26] ‹ bornite ‹ pyrite ‹ sphalerite ‹ chalcopyrite [25] [24] [24] [22] ‹ pyrrhotite ‹ ilmenite ‹ hematite [18] [16] [15] ‹ magnetite ‹ chalcocite [15] [14] (This sequence is from Seward and Barnes (1979); [molar volumes] in cm 3 mol -1 .) 3.8 MODERN ANALOGUES OF ORE-FORMING PROCESSES – THE VMS–SEDEX CONTINUUM The term “volcanogenic massive sul?de,” or VMS, refers to a large family of mainly Cu–Zn (occasionally with minor Pb and Au) deposit types that formed during episodes of major orogenesis throughout Earth history (see Chapter 6). They are also referred to as VHMS deposits, where the acronym stands for “volcanic hosted massive sul?de.” VMS deposits occur in a variety of tectonic settings, but are typically related to precipitation of metals from hydrothermal solu- tions circulating in volcanically active submarine environments. Sedimentary exhalative (SEDEX) deposits, by contrast, are dominated by a Zn–Pb (with lesser Cu, but commonly Ba and Ag) metal association and are also related to hydrothermal ?uids venting onto the sea ?oor, but without an obvious or direct link to volcanism. Many of the large SEDEX deposits of the world are Proterozoic in age, although several examples, such as Red Dog (see Box 3.4) also formed in Phanerozoic times. Although there is generally no spatial or temporal link between SEDEX and VMS deposits, it is widely held that they rep- resent a continuum and are conceptually linked by the fact that they formed by the same basic processes (Gilmour, 1976; Plimer, 1978; Guilbert and Park, 1986; Kirkham and Roscoe, 1993; Misra, 2000). These processes are active and can be HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 177 ITOC03 09/03/2009 14:36 Page 177studied in modern day environments, as dis- cussed below. The notion of a continuum between VMS and SEDEX deposit types is, however, contentious and readers should be wary of over-interpreting the genetic link between the two deposit types. The principal VMS and SEDEX deposits appar- ently formed at different times of Earth history and in different tectonic settings. In addition, studies have shown that some of the stratiform base metal deposits, previously thought to be syngenetic and related to exhalative venting of hydrothermal solutions onto the sea ?oor, are more likely to be the product of replacement processes and, therefore, epigenetic in origin. A reinterpretation of this type has, for example, recently been made with respect to the Mount Isa deposits in Australia (Perkins, 1997). With this cautionary proviso in mind, it is nevertheless convenient to discuss VMS and SEDEX deposits as a conceptual continuum, at least insofar as the principal ore-forming processes are concerned. Both deposit types can be discussed in terms of modern analogues, which is advantageous in that the ore-forming processes can be studied directly. VMS deposits are discussed in the light of the spectacular discoveries on the ocean ?oors of “black smokers,” whereas SEDEX deposits are considered in terms of the rift-related hydrother- mal activity in the Red Sea and also around the Salton Sea in California. 3.8.1 “Black smokers” – a modern analogue for VMS deposit formation Even though it had been known since the 1960s that warm brines vented onto the ocean ?oor, it was the discovery in 1979 of “black smokers” at 21° N on the East Paci?c Rise (Francheteau et al., 1979) that provided spectacular insights into the nature of submarine hydrothermal activity and its potential impact on the understanding of ore deposition in such environments. Black smokers are described as hot (up to 400 °C), metal charged, reduced, and slightly acidic hydrothermal ?uids that vent onto the sea ?oor, usually in zones of extension and active volcanism along mid-ocean ridges (Figure 3.20). The ?uids originate essen- tially from cold (2°C), alkaline, oxidizing, and metal-de?cient sea water. They circulate through the basaltic ocean crust and, in so doing, scavenge metals to form the hydrothermal ?uids now observed at more than 100 black smoker sites in the Paci?c, Atlantic, and Indian Oceans as well as the Mediterranean Sea (Scott, 1997). Black smoker ?uids usually vent through tube-like structures, called chimneys (Figure 3.20), that are built out of a mixture of anhydrite, barite, and sul?de min- erals such as pyrite, pyrrhotite, chalcopyrite, and sphalerite, as well as gangue opaline silica. Metal-charged ?uids venting on the ocean ?oor point to the hydrothermal exhalative processes by which most, if not all, VMS and SEDEX can be explained. Black smokers venting along mid- ocean ridges represent a direct analogy for one speci?c type of the VMS family of deposits, namely the ophiolite-hosted Cu–Zn deposits such as those of the Troodos Massif, Cyprus (Box 3.3). An ophiolite represents a section of a mid-ocean ridge spreading center that is preserved by obduction onto continental material, and is characterized by a sheeted dyke complex, pillowed basalt, and pelagic sediment. A lens of massive sul?de ore, comprising mainly pyrite, chalcopyrite, and sphalerite is located at the ocean ?oor interface and this is underlain by a pipe-like zone of dis- seminated sul?des and intense chloritic alter- ation representing the conduit along which ?uids passed on their way to the ocean ?oor (Figure 3.21a). The geometry of this ore deposit type is fairly typical of most other VMS deposits (Figure 3.21b) even though they may form in other tec- tonic settings such as island arcs, back-arc basins, and fore-arc troughs. Other well known Cyprus- type VMS deposits include Lokken in Norway and Outokumpu in Finland. Famous examples of VMS deposits formed in other tectonic settings include the greenstone belt hosted Kidd Creek and Mattabi deposits in the Archean Superior Province of Canada, Besshi and the numerous Kuroko-type deposits of the Japanese island arc, Rio Tinto and Neves Corvo in the Iberian Pyrite Belt of Spain and Portugal, the Buchans and Bathurst deposits of Newfoundland, Jerome in Arizona, and the Mount Lyell and Hellyer depo- sits in Tasmania. 178 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 178HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 179 There is now general consensus that submarine venting represents a very good analogue for the processes that applied during the formation of most of the actual VMS deposits preserved or fossilized in the Earth’s crust. Excellent reviews of the characteristics of actual VMS deposits and the nature of their ore-forming processes are provided by Franklin et al. (1981), Hekinian and Fouquet (1985), Lydon (1988), Large (1992), and Scott (1997). Studies of many actual deposits con?rm that the ?uids involved are derived dom- inantly from sea water, although ?uid inclusion and stable isotope studies suggest that in certain cases a minor magmatic ?uid component may have been incorporated into the circulating ?uid system. The main source of metals is believed Anhydrite with some sul?de Black smoker Chimneys Collapsed chimney Anhydrite Impermeable crust Sea floor basalt Mainly sul?des Alkaline (pH ~7–8) Oxidizing SO 4 -rich Metal-deficient Hydrothermal fluid Hot (up to 400°C) Acidic (pH ~4–6) Reducing H 2 S-rich Metal-rich (Fe, Mn, Zn, Cu) Sea water Cold (2°C) Extension Extension Magma Hydrothermal fluid Figure 3.20 Conceptual diagram illustrating the ?uid characteristics and circulation pattern in mid-ocean ridge environments that give rise to the formation of “black smokers” on the sea ?oor. Inset is a cross section of an exhalative vent site showing the construction of anhydrite–sul?de chimneys on top of a mound of massive sul?de mineralization (after Lydon 1988; Scott, 1997). ITOC03 09/03/2009 14:36 Page 179to have been the volcanic rocks through which the sea water was percolating, and evidence in support of this comes from mass balance consid- erations and the fact that metal assemblages in different VMS deposits are consistent with the expected metal contents and ratios in the associ- ated primary igneous rock. For example, Cyprus- type VMS deposits, which re?ect leaching of a dominantly ma?c volcanic source rock, are typi?ed by a Cu + Zn metal association (Franklin, 1993) and this is explained by the fact that ma?c volcanics are characterized by much higher Cu and Zn contents than their felsic equivalents. However, the reverse is true of Pb (see Table 1.2, Chapter 1) and this helps to explain why the Kuroko deposits of Japan, for example, described as a Zn–Pb–Cu variant of the VMS family (Pirajno, 1992; Franklin, 1993; Misra, 2000) and associated with arc-related bimodal (i.e. felsic and ma?c) volcanism, have a different metal association. By contrast, the source of sulfur is essentially from the sulfate component of the sea water itself with reduction of sulfate to sul?de occurring during ?uid–rock interaction prior to venting (Ohmoto et al., 1983). This is demonstrated in the compilation by Large (1992), which shows similar trends in sulfur isotope variations of VMS sul?de ores and sea water derived sulfate minerals over time (at least for Phanerozoic deposits). 180 PART 2 HYDROTHERMAL PROCESSES Sul?de-bearing epiclastic (b) (a) Mafic pillow lavas Pelagic sediments Oxidized zone Massive sul?de ore zone Stockwork ore zone (disseminated sul?des and chloritic alteration) Sheeted dyke complex Pyritic siltstone 0 20 m Bedded carbonate-sul?de lens Quartz-hematite + magnetite exhalite – Massive sul?de Stringer sul?de Figure 3.21 (a) Section through a typical ophiolite-hosted, Cyprus- type VMS deposit. Footwall rocks may consist of a sheeted dyke complex and the associated volcanics are often pillowed and have a tholeiitic composition. After Hutchinson and Searle (1971). (b) Section showing the characteristics of VMS deposits other than the ophiolite-hosted type. Associated volcanics may be intermediate or felsic in composition and a closer lateral link to chemical and epiclastic sediments is often apparent. The lens of massive sul?de ore formed on the ocean ?oor, underlain by a stockwork zone of disseminated sul?des and intensely altered volcanic rock, is typical of VMS deposits in general (after Large, 1992). ITOC03 09/03/2009 14:36 Page 180HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 181 The 91 Myr old Troodos ophiolite complex in Cyprus represents an obducted and exposed fragment of oceanic crust that has been extremely well studied. It is also famous as a site of copper mining since the Bronze Age and contains some 90 small to intermediate sized Cu–Zn VMS deposits hosted in pillowed basalts, mainly along the northern edge of the complex (Figure 1). The ophiolite is well exposed and comprises a cross section of the ocean crust, from lower ma?c and ultrama?c cumulates (gabbro and harzburgite that contain podiform chromitite Exhalative venting and “black smokers” on the sea ?oor: the VMS deposits of the Troodos ophiolite, Cyprus Mediterranean Sea Nicosia Mavrovouni Cyprus Troodos complex Paphos Limassol Mt Olympus Skouriotissa Alestos/Memi Larnaca Famagusta Basaltic pillow lavas Sheeted dyke complex Gabbroic/harzburgitic cumulates VMS deposits (Cu-Zn) 0 10 20 30 km Figure 1 Generalized geology of the Troodos ophiolite complex in Cyprus and the distribution of VMS style Cu–Zn mineralization. Low-grade ore Lower pillow lava Upper pillow lava Upper pillow lava High-grade ore Sediments and "umber" Upper pillow lava f f f f 0 100 meters Figure 2 The nature and geometry of VMS mineralization in typical Cyprus deposits. Mineralization can be concentrated at the interface between successive basaltic lava ?ows, as shown, or it can occur between the upper pillow lavas and the umber sediments (after Constantinou and Govett, 1973). ITOC03 09/03/2009 14:36 Page 181VMS deposits are characterized by well devel- oped metal zonation patterns de?ned by a typical sequence from Fe to Fe–Cu to Cu–Pb–Zn to Pb–Zn–Ba in an upward and lateral sense. At a ?rst approximation this sequence re?ects the variable solubilities of these metals at progress- ively lower temperatures and could be seen as a paragenetic sequence. In detail, however, the development of this zonation is more complex and re?ects the evolution of ?uids and the growth mechanism of the massive sul?de mound with time, as described in Figure 3.22a. In this model it is envisaged that the temperature of the ore ?uid increases with time as the deposit grows. Low temperatures would not be able to dissolve much in the way of base metals, although such ?uids could transport sulfate complexes and precipitate anhydrite or barite on mixing with sea water (the so-called “white smokers” seen associated with low temperature vents). At temperatures approaching 250 °C, however, the solubilities of Pb and Zn as chloride complexes would be high, reaching 100 ppm under the conditions applicable to Figure 3.22b. Copper would be poorly soluble, deposits; see Chapter 1), overlain by a sheeted dyke complex and pillowed basalts. The largest deposits are Mavrovouni (15 million tons) and Skouriotissa (6 million tons), although several others have been mined in the recent past (Herzig and Hannington, 1995). Mineralization in the Cyprus deposits comprises massive pyrite and chalcopyrite, with lesser sphalerite, hosted in pillow lavas and associated with zones of intense silici?ca- tion and chlorite alteration (Figure 2; Constantinou and Govett, 1973). The lavas are overlain by shales and cherts that are Fe- and Mn-rich (referred to as umber), and may also contain concentrations of Au. The hydrothermal ?uid responsible for mineralization in the Troodos VMS ores was modi?ed sea water that discharged through black smoker vents located along ridge axis parallel faults at temperatures of 300–350 °C (Spooner, 1980). The characteristics of the Cyprus deposits have been com- pared directly with the TAG site, an active exhalative vent system on the mid-Atlantic ridge (Eddy et al., 1998). It is also evident, however, that VMS mineralization can be associated with magmatic activity that is removed from that occurring along the ridge axis. The intrusion of dis- crete gabbroic plutons, and the formation of seamounts and sea ?oor calderas, for example, are features that pro- but some Au could be transported as a bisul?de complex. Discharge of such ?uids would result in precipitation of barite/anhydrite together with sphalerite and galena, as well as minor gold (Stage 1, Figure 3.22a). As ?uids evolve to higher temperatures (250–300°C) they are capable of containing signi?cant Cu as a chloride complex and chalcopyrite will be precipitated in the foot- wall stockwork zone and at the base of the mas- sive sul?de mound (Stage 2, Figure 3.22a). In the latter area it has been observed that chalcopyrite replaces sphalerite. Since the solvent capacity of a 250°C ?uid with respect to both Zn and Pb is high, it is feasible for the earlier formed sul- ?de minerals to be dissolved and reprecipitated further up in the mound, or distal to it. This zone-re?ning type of process accounts for the development of high grade massive sul?de cap- pings seen in certain VMS deposits. At still higher temperatures (300–350 °C) both Cu and Au are in solution as chloride complexes, ultimately giving rise to further chalcopyrite precipitation, together with Au and pyrite. In this model the sequence Fe–Cu–Pb–Zn–Ba is in fact the reverse of the mote ?uid circulation and VMS styles of mineralization in areas that can be well away from the ridge axis. This observation has obvious exploration signi?cance, and the Alestos and Memi deposits in Cyprus are two examples of classic VMS type mineralization associated with off-axis mineralization (Eddy et al., 1998). 182 PART 2 HYDROTHERMAL PROCESSES Figure 3 Pillow lavas exposed adjacent to a small VMS-type Cu–Zn deposit in the Troodos ophiolite, Cyprus (photograph by Carl Anhaeusser). ITOC03 09/03/2009 14:36 Page 182HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 183 paragenetic sequence predicted simply in terms of solubility variations with declining tempera- ture. This scheme also predicts that VMS deposits could grow by input of metals from below and not necessarily by sul?de particles sedimenting out from the sea water, or aggregation of collapsed chimneys. Finally, it should be emphasized that biologically mediated precipitation mechanisms, such as the involvement of sulfate-reducing bacteria in exhalative vent processes, are likely 150 Cu, Pb and Zn solubilty (ppm) 1000 100 10 1 0.1 0.01 200 250 300 350 400 Temperature (°C) 100 10 1 0.1 0.001 0.01 Au solubilty (ppb) Zn-Pb-Au zone Cu zone Cu-Au zone Pyrite zone Au PbS CuFeS 2 ZnS (b) (a) Stage 1 150–250°C Massive sphalerite, galena Anhydrite or barite Zn, Pb stockwork Stage 2 Massive chalcopyrite 250–300°C Cu stockwork (with Au) Stage 3 Fe-Cu-Pb-Zn-Ba zoning Massive pyrite Enriched Pb, Zn zone 300–350°C Alteration zone, disseminated sul?des Figure 3.22 (a) Secular model explaining the evolution of ?uids, growth of massive sul?de mounds, zoning, and paragenetic sequence for VMS deposits. (b) Plot of temperature versus Cu, Pb, Zn, and Au solubilities appropriate to ?uid conditions during VMS ore deposition (pH = 4; 1 M NaCl; aH 2 S = 0.001; SO 4 /H 2 S = 0.01). Both diagrams after Large (1992). ITOC03 09/03/2009 14:36 Page 183to be important contributing factors during the formation of both VMS and SEDEX deposits. Large-scale precipitation of sul?des by bacterial reduction of sea water sulfate has, for example, been suggested for the Irish Zn–Pb–Ba deposits (see section 3.5.3 above; Fallick et al., 2001). Not all VMS deposits are characterized by a classic lens shaped sul?de mound and a well de?ned footwall alteration zone. Some have ?at or tabular morphologies and appear to be distal from a vent, showing little or no evidence of foot- wall alteration/mineralization. Others exhibit indications of massive sul?de lenses stacked one on top of the other or signi?cant metal deposition below the sea water–rock interface (Large, 1992). The morphology of atypical VMS deposits may largely be a function of the physical properties of the ore ?uids themselves as they in?ltrate up through the footwall volcanics and then vent onto the sea ?oor. Sato (1972) has suggested that sub- marine ?uids would be either more or less dense than sea water as a function of their temperature, salinity, and degree of mixing with cold sea water. Figure 3.23a (Type I) shows a situation where an ore solution is more dense than sea water, giving rise to concentrated metal precipitation that remains proximal to the site of venting. This mechanism could explain the formation of high grade tabular VMS deposits. If the ore solution and sea water have similar densities venting will occur but metals will be precipitated close to the site or in nearby topographic depressions (Type II, Figure 3.23a). A different product would result if the black smoker ?uids were much less dense than sea water (Type III, Figure 3.23a), giving rise to buoyant metal-rich plumes that disperse met- als into more distal marine sediments. Although this is not likely to be a particularly ef?cient pro- cess it remains a possible mechanism for explain- ing distal, low grade deposits associated with a higher proportion of ocean ?oor sediment. Another factor that may be important in considerations of exactly where in the system metals accumulate relates to whether or not the hydrothermal solutions undergo boiling. The phase change from liquid to vapor is strongly pressure-dependent and even though many black smoker ?uids have been observed venting at temperatures as high as 400°C, they appear on the sea ?oor as ?uids because of the high con?n- ing pressures that exist deep in the oceans. This is con?rmed in Figure 3.23b, where the boiling point curve for sea water is plotted relative to deep (represented by 21° N East Paci?c Rise) and shallow (represented by the Juan de Fuca ridge) ocean ?oor settings. Any 350°C hydrothermal solution rising adiabatically through the crust would intersect the deep ocean ?oor as a liquid since it would not yet have intersected the boiling point curve. Metal precipitation on the sea ?oor from the black smoker would, in this case, occur by rapid quenching of the ore solution as it mixed with a virtually limitless volume of cold sea water. By contrast, the same ?uid moving up toward a shallower sea ?oor is likely to intersect the boiling point curve before it vents to surface (Figure 3.23b). Boiling below the ocean ?oor is likely to promote brecciation of the footwall and local metal precipitation, possibly resulting in extensive stockwork mineralization. Classic VMS deposition is unlikely to occur and ore bodies may appear stacked one on top of the other. Although there are many variants in the family of VMS ores the processes applicable to the submarine exhalative model appear capable of explaining many of the features of this very important class of deposits. 3.8.2 The Salton Sea and Red Sea geothermal systems – modern analogues for SEDEX mineralization processes SEDEX deposits contain more than half of the world’s known resources of Pb and Zn and are generally represented by bigger and richer deposits than the VMS category, although there may be fewer of them. They are typically formed within intracratonic rift basins and are hosted by marine clastic or chemical sediments with little or no direct association with volcanic rocks. Most of the world’s giant SEDEX deposits are Paleo- to Mesoproterozoic in age and well known examples include HYC (McArthur River), Mount Isa and Broken Hill in eastern Australia, Sullivan in British Columbia, Aggeneys and Gamsberg in South Africa, and Rajpura-Dariba in India. A 184 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 184HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 185 signi?cant group of deposits is Paleozoic in age and these include Meggen and Rammelsberg in Germany, Navan in Ireland, Red Dog in Alaska, and Howard’s Pass in eastern Canada. Modern analogues for the genesis of SEDEX deposits have been described from the Salton Sea and Gulf of California (Figure 3.24) and also the Red Sea. The study of these hydrothermal 0 Pressure (bars) 100 200 300 400 100 200 300 400 500 Temperature (°C) 1000 2000 3000 4000 Depth (meters) (b) 500 5000 Liquid Vapor Sea floor Juan de Fuca Ridge Sea floor 21°N “Critical point” sea water Deep water “Black smoker” and metal deposition on sea floor Boiling and metal deposition below sea floor Sea water (3.5 wt% NaCl) boiling curve Cu Cu–Pb–Zn Pb–Zn–Au–Ag Shallow water 0 Temperature (°C) 100 200 300 0.90 0.95 1.00 1.05 Density (g cm –3 ) (a) 0 0.85 2 m 1 m 2.5 m 5 m 3 m II III Density of seawater at 10°C I Type III Type II Type I Adiabatic uprise of 350°C hydrothermal fluid Figure 3.23 (a) Plot of temperature versus density for a variety of possible hydrothermal solutions (where 1 M, 3 M, etc. refer to the NaCl molality or the salinity of the solution) relative to sea water at 10 °C. Different solutions may be either less or more dense than sea water. At one extreme they would form buoyant plumes capable of distributing metals some distance away from the site of venting (Type III), whereas at the other they would precipitate metals proximally to the vent (Type I). After Sato (1972). (b) Pressure (or depth) versus temperature plot showing the sea water (with 3.5 wt% NaCl) boiling curve relative to deep (21° N East Paci?c Rise) and shallow (Juan de Fuca ridge) sea ?oor. Because of the strong pressure dependency on boiling, a 350 °C hydrothermal ?uid rising adiabatically in the crust would intersect a deep ocean ?oor as a liquid and would vent as a black smoker precipitating metals into the sea. The same ?uid relative to a shallower ocean would boil before intersecting the ?oor such that only a vapor phase would reach the surface. Metal precipitation in such a case will likely occur beneath the ocean ?oor. After Delaney and Cosens (1982). ITOC03 09/03/2009 14:36 Page 185systems reveals both syn-sedimentary exhalative and replacement processes and can, to a certain extent, be used to discount the often controversial views that prevail regarding the origin of sediment- hosted stratiform deposits. Salton Sea geothermal system The East Paci?c Rise intersects the North American plate at the head of the Gulf of California (Figure 3.24a). At this point the prevail- ing stresses along this mid-ocean ridge are trans- ferred onto the continent to form the essentially dextral San Andreas fault system. The fault cuts through a dry intermontane region called Imperial Valley, which is characterized by active, rift- related sedimentation. Early settlers discovered a huge salt pan in this valley, named the Salton Sink, that was mined for a few years in the late 1800s. In 1905, however, a canal, designed to bring irrigation water from the Colorado River to settler farmers in Imperial Valley, was breached by spring ?oods and the entire volume of this large river was channeled into the Salton Sink, at that time close to 100 meters below sea level. It was two years before the breach was plugged and the Colorado River reverted to its natural course. By that time the salt pan had become the Salton Sea, a huge inland lake that covered an area of several hundred square kilometers. The Salton Sea soon became brackish as the lake waters dissolved the salt on its ?oor. Evaporation also increased the salinity of Salton Sea waters to the Na–Cl–SO 4 dominated composition that is observed today (McKibben and Hardie, 1997). Over the past century water from the Salton Sea has been percolating downwards along fault- related pathways and through earlier formed lacustrine sediments to form an extensive hydro- thermal system from which geothermal power is now derived. In addition, hot metalliferous brines, also derived directly from Salton Sea waters and heated by a combination of the high prevailing 186 PART 2 HYDROTHERMAL PROCESSES SAN ANDREAS FAULT (a) Pacific plate North American plate 30° USA Mexico N 0 500 km (b) N Basement outcrops USA Mexico 02 0 km Colorado River Primary transform fault Active transform extension Inactive fault zone Spreading center USA Mexico SALTON SEA San Andreas fault Cerro Prieto Volcano EAST PACIFIC RISE Salton Sea geothermal system Gulf of California Figure 3.24 (left) The Salton Sea Geothermal System (SSGS) and its location (inset) relative to the Gulf of California transform fault region and the dextral San Andreas fault system (after McKibben and Hardie, 1997). ITOC03 09/03/2009 14:36 Page 186HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 187 geothermal gradient and active volcanism in the region, vent from numerous sites, and it is these that are of particular interest to the study of ore-forming processes (White, 1963; Skinner et al., 1969). The combination of well understood ?uid compositional and circulation pathways, set in an active seismic and volcanic, intracon- tinental rift environment, has made the Salton Sea geothermal system a classic example of an ore-forming environment whose evolution can be observed, and from which many inferences about ancient sediment-hosted ore-forming processes can be made. In addition to its obvious applicabil- ity to the understanding of SEDEX ore-forming processes, the processes observed in this environ- ment might also be relevant to the formation of stratiform red-bed hosted Cu deposits such as the Kupferschiefer ores of central Europe and the Central African Copperbelt (McKibben et al., 1988; McKibben and Hardie, 1997). Although the Salton Sea geothermal system does not exist in a submarine environment, the geometry of the system and the processes involved are considered to be useful for the understanding of sediment-hosted hydrothermal base metal deposits in which ?uids circulated during and soon after deposition. A variety of conditions need to prevail in order to make SEDEX ore-forming processes viable. These include an abundant ?uid source, a rich source of metals and complexing agents, active rifting, and deformation induced ?uid ?ow. Detailed study of geothermal wells around the Salton Sea has shown that at depths of 1000–3000 meters a hot (350 °C), dense (up to 26 wt% dissolved solids), Na–Ca–K–Cl brine exists (McKibben et al., 1988). This ?uid also contains signi?cant concentrations of metals, in particular Fe, Mn, Pb, Zn, and Cu, derived from local lacus- trine sediments. Precipitation of these metals has been induced by mixing of the hot metal charged brine with cooler, dilute surface waters about 1000 meters below the surface, to form sediment- hosted base metal sul?de veins and associated chloritic alteration. This is clearly not an exhalat- ive system and the ores that form will appear epigenetic in character. Ores that form in this environment could also have a stratiform geo- metry, with metals being deposited by replacement of pre-existing phases such as diagenetic cements or easily dissolvable detrital minerals. In the Salton Sea setting it is also conceivable, especially in the event of very energetic ?uid circulation, that the metal-charged brine could intersect the lake ?oor and vent ore-bearing solutions in much the same way that black smokers are exhaled along the mid-ocean ridges. It seems feasible, there- fore, for SEDEX ore bodies to appear syngenetic and exhalative in sediments that were deposited at the time of ?uid venting, but also distinctly epigenetic relative to the footwall sedimentary sequence. A model illustrating the main features of SEDEX deposits and incorporating the concept that syngenetic exhalative and epigenetic replace- ment type ores could be coeval and form part of the same system is presented in Figure 3.25. The Red Sea and the VMS–SEDEX continuum The extension of the East Paci?c Rise (mid-ocean ridge) onto continental North America to form the San Andreas fault system (Figure 3.24) pro- vides support for the view that a conceptual con- tinuum exists between VMS and SEDEX styles of mineralization. Present day ocean ?oor black smokers at 21° N on the East Paci?c Rise and the active, metal-charged geothermal systems of the Salton Sea perhaps represent extremes in the range of VMS and SEDEX ore-forming systems. The Guaymas Basin in the Gulf of California is also known to contain terrigenous sediment- hosted styles of VMS mineralization, also known as Besshi-type ores, from the type location in Japan. These intermediate styles of mineraliza- tion also provide support for the view that VMS and SEDEX styles of base/precious metal mineral- ization can be linked in terms of both tectonic set- ting and hydrothermal systems. Perhaps the best example of the conceptual link between VMS and SEDEX deposit types, however, is provided by the Red Sea rift, separating Africa and Arabia. The Red Sea is an ocean basin in the early stages of its development, containing active geothermal venting and deposition of base metal deposits in clayey muds accumulating in depressions on the sea ?oor (Bischoff, 1969). Mineralization is particularly well developed at a site known as the ITOC03 09/03/2009 14:36 Page 187188 PART 2 HYDROTHERMAL PROCESSES Vent complex Syn-sedimentary growth fault f Fault breccia and feeder pipe Replacement ore Fault scarp Exhalative ore Distal sediments, barite, chert Permeable units Fluid ingress f f f Figure 3.25 Diagram illustrating the setting for the formation of SEDEX-type Pb–Zn ores and a scenario which incorporates both exhalative and replacement concepts for the formation of these ores (after compilations by Goodfellow et al., 1993; Misra, 2000). The Red Dog mine in northwest Alaska is presently the world’s largest producer of zinc concentrate, contributing about 6% of annual global supply. It is a very rich deposit and in 2002 had reserves of close to 100 million tons of ore grading 18% zinc, 5% lead, and 85 g ton -1 silver ( It was discovered in 1968 by an astute bush pilot who noticed a color anomaly in the rocks caused by oxidation of the sul?de ores. The Red Dog ores are hosted in Carboniferous black shales and limestone of the Kuna Formation in the De Long mountains of the Brooks Range, northern Alaska. The host succession has been deformed into a sequence of thrust allochthons during a Jurassic orogeny. Figure 1 shows an interpretive section through a Devonian– Carboniferous sequence of sediments in the Brooks Range prior to thrusting. The sequence of sediments, including Sedimentary exhalative (SEDEX) processes: the Red Dog Zn–Pb–Ag deposit, Alaska Conglomerate Chert Limestone Black shale Sandstone Devonian–Carboniferous boundary Red Dog sequence Oceanic side Continental side ? Sea level Figure 1 Interpretive, pre-thrusting, cross section through Devonian–Carboniferous sediments in the Brooks Range, illustrating the depositional setting of host rocks at the Red Dog mine. Syn-sedimentary growth faults and abrupt lateral facies variations are also a feature of the regional setting (after Moore et al., 1986). ITOC03 09/03/2009 14:36 Page 188HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 189 Atlantis II Deep where muds are estimated to con- tain around 50 million tons of rich Zn–Cu–Pb– Ag–Au mineralization. Although the geological and tectonic setting of the Red Sea mineralization is analogous to a mid-ocean ridge VMS setting, the absence of black smokers and the presence of both clastic and carbonate sediment hosts to the mineralization are features perhaps more akin to SEDEX ores (Pirajno, 1992). 3.9 MINERAL DEPOSITS ASSOCIATED WITH AQUEO- CARBONIC METAMORPHIC FLUIDS A signi?cant proportion of the world’s gold deposits are genetically linked to the formation of metamorphic ?uids that are typically character- ized by low salinity, near neutral pH, and mixed H 2 O–CO 2 compositions. These gold deposits, of which there are many different types that formed Figure 2 Simpli?ed block diagram showing the environment of ore formation in the Carboniferous for the Red Dog deposit (after Moore et al., 1986). Noatak sandstone Turbidites Site of exhalative venting Kayak shale Red Dog Kivalina unit f f f f limestones and turbiditic assemblages, is consistent with deposition in a restricted ocean basin, with no indication of volcanism in the immediate environs. Abrupt lateral facies changes in the succession point to the presence of syn-sedimentary growth faults that would have facilitated ?uid circulation through the basin. Mineralization is stratabound and, despite the thrusting, is relatively ?at-lying (Moore et al., 1986). It is marked by laterally extensive zones of silici?cation (chert and chalcedony) within which poorly bedded sul?de minerals (sphalerite, pyrite–marcasite, and galena) occur. Barite is closely associated with the ore and tends to cap the mineralized zones. An interesting feature of the ore zones is the occurrence of sinuous cylindrical structures interpreted to be worm tubes, similar to biotic structures observed in present day exhalative vent environments. This evidence supports the notion that silica and sul?de precipitation took place at the rock–water interface on the ocean ?oor and was, therefore, coeval with the deposi- tion of the sediments. There is also, however, evidence for sul?de replacement of sediments at Red Dog, indicating that the dual processes of exhalation and replacement were taking place (see section 3.8.2 of this chapter). Figure 2 presents a simpli?ed model of the ore-forming environment at Red Dog. Fluids circulating through Carboniferous organic-rich marine sediments scavenged metals from these host rocks and were vented onto the ocean ?oor at sights created by syn-sedimentary growth faults. Mineralization at Red Dog is massive, brecciated, and poorly bedded, suggesting that sul?des were pre- cipitated fairly close to the vent sites (Moore et al., 1986). Variable sulfur isotope signatures in sul?de minerals sug- gest that precipitation was a product of mixing a buoyant hydrothermal plume with sea water that was becoming progressively more oxic over time. It is suggested that mineralization ceased when the sea water became suf?ci- ently oxidizing to favor barite precipitation over sul?des (Moore et al., 1986). ITOC03 09/03/2009 14:36 Page 189over the entire span of Earth history, are asso- ciated with regionally metamorphosed terranes resulting from compression at convergent plate margins, and have been termed orogenic gold deposits (Groves et al., 1998). They are also widely referred to, especially in the older literature, as “mesothermal” or “lode-gold” deposits, as they commonly precipitated at around 300 °C, forming quartz-vein or fracture dominated ore systems. Although orogenic gold deposits are commonly associated with Archean granite–greenstone ter- ranes (see Chapter 6), important examples of oro- genic gold ores are also found hosted in a variety of Proterozoic and Phanerozoic settings. The dom- inant and characteristic genetic features that link all orogenic gold deposits are a synchroneity with major accretionary or collisional orogenic episodes and the production of metamorphic – and in some cases magmatic – ?uids that pre- cipitate metals at various crustal levels along deep-seated shear and fracture zones (Figure 3.26; mega- and mesoscale models). It is widely recog- nized that these deposits were formed over a range of P–T conditions, occurring in granulite to greenschist facies host environments and ductile through to brittle structural regimes (Colvine, 1989; Groves, 1993). In the sections that follow, a variety of orogenic gold deposits from throughout geological time are described. In addition, Carlin-type deposits, as well as the hydrothermal component of mineral- ization in quartz pebble conglomerate hosted gold ores such as the Witwatersrand and Tarkwaian deposits, are described, as they share several com- mon features with orogenic gold deposits. 3.9.1 Orogenic gold deposits There are many different types of orogenic gold deposits and their classi?cation into a single 190 PART 2 HYDROTHERMAL PROCESSES Carbonation chloritization sul?dation Paleosurface First order fault zone Archean granite greenstone terrane Second/ third order fault zone Current surface Oxidized magma? Internal granitoids Granulite facies Metamorphic devolatilization Brittle-ductile transition Metamorphic devolatilization H 2 O–NaCl–CO 2 +CH 4 fluid inclusions 1 Phase separation 2 Fluid-rock reaction (e.g. sul?dation) Fault Proximal alteration zone Distal alteration zone Megascale Mesoscale Ore deposition Figure 3.26 Schematic illustrations showing the principal features of Archean orogenic gold deposits – many of which also apply to Proterozoic and Phanerozoic examples. The main features at a megascale are that the deposits are associated with a convergent plate margin where metamorphism results in the production of ?uids that are focused along major structural discontinuities. At a mesoscale major shear zones within or along the margins of greenstone belts represent district-scale hosts to mineralization that are accompanied by broad zones of intense alteration. Individual deposits occur along second- and third-order structures. Ore deposition is a function of ?uid rock reaction and/or H 2 O–CO 2 phase separation (after Hagemann and Cassidy, 2000). ITOC03 09/03/2009 14:36 Page 190HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 191 category is bound to be controversial. From a descriptive and organizational point of view, how- ever, the classi?cation is appropriate and recogni- tion of the different subtypes that undoubtedly do exist can perhaps best be made in terms of geolo- gical time and crustal evolution. Archean The Archean was a major gold metallogenic epoch and many important gold deposits occur in the granite–greenstone terranes of the world, examples of which include the Superior Prov- ince of Canada, the Yilgarn Craton of Western Australia, and the Zimbabwe Craton. A feature of Archean orogenic gold deposits is that they can be hosted in a variety of greenstone belt associated lithologies, including ma?c volcanics, metasedi- ments, and banded iron-formations, as well as in felsic plutonic rocks. Despite the range of host rock lithologies, they are invariably found asso- ciated with zones of high strain that are manifest as brittle, brittle–ductile or ductile deformation, depending on the crustal depth at which ?uids were circulating. Reviews of the nature and char- acteristics of Archean deposits can be found in McCuaig and Kerrich (1998) and Hagemann and Cassidy (2000). Figure 3.26 illustrates the main features of Archean orogenic gold deposits. At a megascale the energy required to form huge zones of ?uid ?ow, alteration, and mineralization, such as those preserved in the Abitibi (Superior) and Norseman–Wiluna (Yilgarn) greenstone belts, is ultimately derived from plate subduction and collision. Crustal thickening, deformation, meta- morphism, and synorogenic magmatism all play important roles in the ultimate origin of ?uids and their focused ?ow upwards into the crust. A major part of the ?uids implicated in mineraliza- tion appears to have been derived from the dehydration that accompanies regional prograde metamorphism. These ?uids are fairly reduced (fO 2 around the quartz–fayalite–magnetite or QFM buffer) and gold is preferentially transported as the Au(HS) 2 - complex (Hagemann and Cassidy, 2000; Ridley and Diamond, 2000). Derivation of mineralizing ?uids from oxidized granitic magma is also considered likely, although such ?uids are probably restricted to environments where a spatial link between mineralization and granite intrusion is evident, or an association between Au and metals such as Mo, W, or Cu occurs (Burrows and Spooner, 1987). Gold mineralization is broadly synchronous with the regional peak of metamorphism, although this peak may be in the lower crust rather than at the site of deposition in the upper crust. Fluids tend to be focused along major structural discontinuities that have strike lengths of tens to hundreds of kilometers and extend to considerable depths in the crust. Examples of well mineralized structural systems include the Larder Lake–Cadillac fault zone in the Abitibi belt and the Boulder–Lefroy fault zone in the Yilgarn craton. Individual deposits and mines are typically found along second- and third-order structures within these major discontinuities. Movement of ?uid along these structures is explained by the fault-valve model, as discussed in section 3.3.3 above. The sites of mineralization are marked, at a regional scale, by zones of intense carbonation and chloritization that can extend for hundreds of meters around the conduit (see Box 3.2). Alteration plays a major role in metal precipitation, and this is particularly the case for intense wall-rock alteration and sul?dization. H 2 O–CO 2 phase separation is also considered to be an important process in gold deposition and may explain the siting of rich accumulations of gold in quartz veins (i.e. the bonanza lodes). The latter mechanism may also explain the com- mon association of gold mineralization with the brittle–ductile transition, as this zone coincides broadly with the relevant P–T conditions along parts of the H 2 O–CO 2 solvus (see Figure 3.5c). Mixing of metamorphic with meteoric ?uids may explain gold precipitation at high levels in the crust. Proterozoic The Proterozoic Eon is not as important as either its Archean or Phanerozoic counterparts in terms of gold mineralization (see Chapter 6). There are, nevertheless, several important examples of oro- genic gold deposits from this period of time and these include Ashanti–Obuasi and several other ITOC03 09/03/2009 14:36 Page 191deposits associated with the Birimian orogeny of West Africa, Telfer in Western Australia, Homestake in South Dakota, USA, Omai on the Guayanian craton of South America, and the Sabie–Pilgrim’s Rest gold?eld in South Africa. All these examples are associated with periods of major orogenesis and largely involve metamor- phic ?uids very similar in character to those applicable to the formation of Archean orogenic gold deposits. Gold deposition is generally late orogenic and is hosted in major, high-angle thrust faults (Partington and Williams, 2000). Granite intrusions, and possibly a magmatic ?uid compon- ent, are associated with many of the Proterozoic deposits and, in cases such as Telfer (Rowins et al., 1997) and Sabie–Pilgrim’s Rest (Boer et al., 1995), a polymetallic association of Cu, Co, and Bi, in addition to Au, is evident. In general, however, the characteristics of Proterozoic oro- genic gold deposits are very similar to those of the Archean. Phanerozoic Orogenic gold deposits of the Phanerozoic Eon were formed during two mineralization episodes, one in the Silurian–Devonian at around 450– 350Ma and the other in the Cretaceous– Paleogene at around 150–50Ma (Bierlein and Crowe, 2000). These gold deposits are again asso- ciated with convergent plate margins and hosted in compressional to transpressional shear zones that typically cut through thick marine shale sequences that have been metamorphosed to greenschist facies grades. They are often referred to in the older literature as “slate-belt hosted” gold deposits. There are many important examples of Phanerozoic orogenic gold deposits and these include, as Paleozoic examples, Bendigo, Stawell, and Ballarat in the Lachlan orogen of southeastern Australia, Haile in the Carolina slate belt, USA, Muruntau in Uzbekistan, the Urals of central Russia, and the Meguma terrane of Nova Scotia, Canada. Younger, Mesozoic examples include the Juneau gold belt of southern Alaska and other portions of the Cordilleran orogen of Canada and the western USA, including the Mother Lode of California (Bierlein and Crowe, 2000). 3.9.2 Carlin-type gold deposits Although several of the features of Carlin-type ores mirror those of orogenic gold deposits, they are generally classi?ed separately because of one very notable difference. While linked to orogenic activity and hosted along major, deep-crustal structures, they are not related to compression but formed after the onset of extensional forces that followed earlier, subduction-related pro- cesses (Hofstra and Cline, 2000). Carlin-type ores are located in Nevada, USA, and formed over a short period of time in the Paleogene Period, between 42 and 30 Ma. In this sense they are regarded as an essentially unique deposit type, although it is likely that similar ores will be located elsewhere, such as in China. In Nevada they are hosted in Paleozoic carbonate rocks that have been subjected to severe crustal shortening and west to east directed thrusting (Figure 3.27a and Box 3.5). The carbonate sequences are struc- turally overlain by less permeable siliciclastic rocks and have acted as aquifers along which metamorphically derived ?uids have ?owed. During mid-Miocene times host rocks were sub- jected to uplift and extension, which thinned the crust and gave rise to the Basin and Range faulting and associated ?uid circulation along the newly formed dilatant structures. Fluids associated with Carlin deposits are very similar to the metamorphic ?uids implicated in orogenic gold deposits, comprising low salinity, moderately acidic, reduced H 2 O–CO 2 solutions in which gold is transported as either Au(HS) 2 - or Au(HS). Fluid temperatures are cooler than those associated with orogenic ores and fall in the range 150–250 °C, which has led to the suggestion that they may, however, be meteoric rather than meta- morphic in origin. Gold deposition occurs where normal faults intersect a less permeable cap rock, usually at a shale/limestone contact and in the crests of fault-propagated anticlines (Figure 3.27b). The precipitation mechanism is believed to be related to neutralization of the ore ?uid during carbonate dissolution, although argillic alteration, silici?cation, and sul?dization also played a role in the ore-forming process. In Carlin-type deposits the gold occurs as micrometer-sized particles or 192 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 192Archean crust (b) Golconda allochthon Early compression Oceanic crust Later extension Twin Creeks Carbonate sequence Basin and Range faults Proterozoic–Cambrian sequence 42° 0 120° 100 40° 118° 116° 112° 114° (a) Utah Arizona Getchell Carlin Battle Mt Eureka Golgonda thrust Roberts Mt thrust Sevier Orogenic Belt Eastern limit of magmatism Magmatic arc California Oregon Idaho Nevada Roberts Mountain allochthon Figure 3.27 (a) Map showing the distribution of Carlin-type gold deposits in Nevada and the geological framework. (b) Cross section illustrating the fundamental controls behind the location of Carlin-type gold deposits (after Hofstra and Cline, 2000). ITOC03 09/03/2009 14:36 Page 193194 PART 2 HYDROTHERMAL PROCESSES Twin Creeks mine, formed in 1993 after the merger of the Chimney Creek and Rabbit Creek mines, is one of the largest Carlin-type deposits in Nevada, with reserves in 1999 of some 90 million tons of ore at an average gold grade of 2.5 g ton -1 . Most of the gold at the mine is extracted from the huge Mega Pit which produces some 280 000 tons of ore per day. Although gold at the Mega Pit is hosted essen- tially in limestones of the Ordovician Comus Formation, mineralization elsewhere on the mine (i.e. the Vista Pit) also occurs in overlying sequences that form a package of thrusted allochthons (see Figure 3.27a and b) emplaced in Carboniferous times. Gold mineralization occurs mainly in the arsenic-rich rims to pyrite grains formed during hydro- thermal alteration (i.e. decalci?cation, dolomitization, silici?cation, and sul?dization) of the Comus Formation limestones (Thoreson et al., 2000). In the Mega Pit the mineralized limestones are capped by the Roberts Mountain thrust which folded the under- lying limestones and also possibly acted as a cap that limited further egress of hydrothermal ?uids and con- centrated mineralization within the Comus Formation. Figure 1 shows a cross section through the Mega Pit, illus- trating the relationships between the early thrusting and folding of Comus limestones, and the later extensional faults along which mineralizing ?uids are thought to have been focused. In this section line gold mineraliza- tion is concentrated largely within the core of a prominent eastward verging recumbent antiform. Circulation of orogeny-driven aqueo-carbonic ?uids: Twin Creeks – a Carlin-type gold deposit, Nevada 0 100 meters Paleocene cover Roberts Mountain thrust Upper Sill Main Sill West Side Fault Gold mineralization (in fold core) Upper Sill Black Nasty Fault Mega Pit outline Comus Formation (folded during thrust event) DZ Fault Upper Sill f f f f f Figure 1 Cross section looking north of the Mega Pit at the Twin Creeks mine (after Thoreson et al., 2000). ITOC03 09/03/2009 14:36 Page 194HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 195 in solid solution in arsenic rich pyrite, marcasite, or arsenopyrite. 3.9.3 Quartz pebble conglomerate hosted gold deposits Quartz pebble conglomerates of Archean and Paleoproterozoic ages represent a very important source of gold, although production over the past century has been dominated by the Witwatersrand Basin in South Africa. The seven major gold?elds associated with the Witwatersrand sequences have produced a staggering 48 000 tons of gold metal – some 35% of all the gold produced in the history of mankind – since 1886 (Phillips and Law, 2000). Other conglomerate-hosted gold deposits such as those of the Tarkwa (Ghana) and Jacobina (Brazil) basins are relatively unimportant by comparison. Irrespective of their production statistics, quartz pebble conglomerate hosted gold deposits are characterized by considerable controversy regard- ing the origin of their gold mineralization. Argu- ments revolve essentially around whether gold was introduced as detrital particles during sedi- mentation (a placer process; see Chapter 5) or precipitated from hydrothermal solutions circu- lating through the sediments at some stage after sediment deposition. There is little doubt, how- ever, that in all three cases (i.e. the Witwatersrand, Mineralization at Twin Creeks, as elsewhere in the Carlin district of Nevada, is considered to have been a result of ?uid circulation between 42 and 30Ma (Eocene– Oligocene) during the onset of extensional (Basin and Range) tectonism and mantle plume activity (Hofstra and Cline, 2000). Gold deposition, therefore, took place long after deposition of the host rocks (in the Ordovician) and their subsequent deformation (thrusting and folding) in Carboniferous times. Although subduction-related mag- matism took place in the region between 43 and 34 Ma, no intrusives are directly implicated in the formation of any Carlin-type deposit and most workers refer to a metamorphic ?uid as the mineralizing agent. This ?uid was a moderately acidic, reduced, H 2 O–CO 2 solution at around 150–250 °C in which gold was transported as bisul?de complexes (Hofstra and Cline, 2000). Gold pre- cipitation is believed to have accompanied cooling and neutralization of the ?uid as it interacted with reactive Tarkwa, and Jacobina) the host sediments under- went burial and regional metamorphism and that the associated circulation of metamorphic ?uids resulted in a clearly epigenetic component of mineralization. Fluid inclusion studies indic- ate that certain of these ?uids have a mixed H 2 O–CO 2 composition and have some similarit- ies to those involved with orogenic gold deposits. One view of the origin of gold mineralization in the Witwatersrand sediments is that it occurs along major unconformities, represented by con- glomerate deposition, which represent zones of intense ?uid ?ow and alteration (Figure 3.28a). Alteration is the product of ?uid–rock reactions involving progressive neutralization of an acid ?uid (see section 3.5 above), with the most intense alteration being characterized by the assem- blage pyrophyllite–chloritoid–muscovite–chlorite (Barnicoat et al., 1997; Philips and Law, 2000). In the Witwatersrand Basin, however, gold miner- alization is characterized by a long-lived paragen- etic sequence (Robb et al., 1997; Frimmel and Minter, 2002) and is likely to be related to more than just one-pass ?uid–rock interaction. The existence of a prominent alteration halo around zones of pervasive ?uid ?ow is nevertheless one of the reasons why the Witwatersrand gold deposits are so large. At the scale of a single deposit similar effects can be seen, as exempli?ed country rocks such as the Comus Formation limestones. Circulation of mineralizing ?uids was focused essentially along major extensional faults, which is the main reason why major deposits are located along well de?ned struc- tural trend lines. Individual deposits tend to occur beneath a less permeable, often siliciclastic, cap rock (such as the Roberts Mountain allochthon at Twin Creeks) causing the ?uids ascending along the faults to ?ow laterally into more reactive sequences such as the Comus Formation. Carlin-type deposits are unique in the broader family of orogenic gold deposits and are perhaps intermediate in character between more typical compression-related orogenic gold ores and low-sul?dation epithermal gold deposits (see Chapter 2). There is, however, no reason why similar styles of epigenetic, sediment-hosted deposits, containing “invisible” or micrometer-sized gold in sul?des, should not be located in other parts of the world besides Nevada. ITOC03 09/03/2009 14:36 Page 195196 PART 2 HYDROTHERMAL PROCESSES Gold-bearing conglomerate packages Elsburg Kimberley Bird Main Elsburg sequence S Kimberley sequence Main sequence Bird sequence Central Rand Group West Rand Group Alteration assemblages (+Qtz) Pyrophyllite, chloritoid, muscovite, chlorite Muscovite, chlorite Albite, epidote, calcite, muscovite, chlorite, paragonite Stratigraphic boundaries Principal zones of fluid flow 20 km 500 m Witwatersrand basin (a) (b) Gold-bearing conglomerate W E Shear zones Shear zones 0 metres 2 Cr mica, tourmaline, rutile, muscovite alteration Zone of sul?dation (pyrite) Barren conglomerate N ITOC03 09/03/2009 14:36 Page 196HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 197 by the João Belo reef in the Jacobina Basin of Brazil (Figure 3.28b). Here gold mineralization coincides with a section of the conglomerate horizon that is cut by a number of bedding-parallel shear zones along which ?uids have ?owed resulting in fuchsite (Cr–mica)–tourmaline–rutile–muscovite –pyrite alteration (Milési et al., 2002). Outside this alteration zone the conglomerate is reported to be barren. The nature of the ore-forming ?uid associated with the Witwatersrand, Jacobina, and Tarkwa auriferous conglomerates has been described as mixed H 2 O–CO 2 , fairly reducing, near-neutral, and with low salinity (Philips and Law, 2000). Barnicoat et al. (1997) preferred a more oxidized, acidic ?uid for the Witwatersrand Basin whereas Klemd et al. (1993) suggested that the ore ?uids associated with the Tarkwaian conglomerates were dominantly high density carbonic with signi- ?cant contained methane and nitrogen. Detailed ?uid inclusion studies (Drennan et al., 1999; Frimmel et al., 1999) have indicated that in the Witwatersrand Basin different ?uids are implicated in at least two distinct episodes of mineralization, and these include mixed H 2 O–CO 2 compositions as well as aqueous, high-salinity ?uids. The origin of these ?uids is also contentious and options range from devolatilization of greenschist to amphibolite facies ma?c rocks beneath the sedi- mentary basins, to derivation by prograde meta- morphic reactions within the basin itself (Stevens et al., 1997 and Figure 3.4). The hydrothermal ore-forming processes of the Witwatersrand Basin in particular are long-lived and complex and this is the reason why there has been, and still is, so much debate on the genesis of these ores. It is, for example, now reasonably well established that the basin was struck by a large meteorite at 2025 Ma and that the catastrophic exhumation of the gold-bearing sedimentary host rocks, as well as the likely circulation of related meteoric ?uids, must also have played an important role in the ore formation process (Gibson and Reimold, 1999). 3.10 ORE DEPOSITS ASSOCIATED WITH CONNATE FLUIDS A number of sediment-hosted base metal deposits are genetically linked to the circulation of con- nate ?uids during diagenesis. In these deposit types metal transport and deposition is generally restricted to the sedimentary sequence through which the connate ?uids circulate. Ores included in this category are the important stratiform, sediment-hosted copper deposits (abbreviated to SSC, but also called red-bed copper deposits) and the family of Pb–Zn ores, usually associated with carbonate sediments (although some are sandstone-hosted), known as Mississippi Valley type (or MVT) deposits. Although the two deposit types are different in several respects, they both owe their origins to circulating basinal brines. Sverjensky (1989) has suggested that the major differences between SSC and MVT deposits relate mainly to the type of sediment through which the ?uids have traveled prior to ore deposition and also to the contrasting properties of Cu, Pb, and Zn in hydrothermal solutions. It is suggested that a basinal brine channeled into an oxidized (red-bed) aquifer would scavenge all three base metals from fertile detritus in the basin, but that the resulting ore ?uid would be saturated only with respect to Cu and undersaturated relative to Pb and Zn. The tendency in such an environment would be for ?uids to preferentially deposit copper sul?de min- erals while the higher solubility metals remain in solution. Accordingly, SSC deposits tend to be characterized by a Cu > (Pb + Zn) metal associ- ation (Figure 3.29). By contrast, the same original ?uid channeled into a relatively reduced carbon- ate aquifer would evolve to Zn and Pb saturated conditions in the presence of dissolved sulfur species and ore deposition would yield sphalerite > galena >>chalcopyrite, as is typical of most Figure 3.28 (Opposite) (a) Simpli?ed north–south section showing the pattern of alteration at a regional scale associated with gold-bearing conglomerate packages in the Central Rand Group of the Witwatersrand Basin (after Barnicoat et al., 1997); (b) Cross section through the João Belo reef in the Jacobina Basin showing the relationship between shear zone related ?uid conduits, alteration, and gold mineralization at a deposit scale (after Milési et al., 2002). ITOC03 09/03/2009 14:36 Page 197MVT deposits. In the situation where the original basinal brine passed through a quartzose sand- stone aquifer, the ?uid would not evolve signi?c- antly from its original metal carrying capacity, with saturated levels of Cu and Pb but undersatur- ated relative to Zn. Maintenance of the oxidation state of the ?uid at levels appropriate to hematite– magnetite coexistence would nevertheless limit the mobility of copper and resulting ores would be galena-rich, consistent with the fact that sandstone-hosted MVT deposits tend to have Pb > Zn >> Cu metal ratios. It is interesting to note that in the Central African Copperbelt, for ex- ample, Cu(–Co) mineralization is typically hosted in clastic sediments such as arkose and shale, whereas Zn–Pb mineralization is found associ- ated with carbonate rocks. Some of the processes relevant to the formation of SSC and MVT deposits are discussed in more detail below. 3.10.1 Stratiform sediment-hosted copper (SSC) deposits SSC deposits worldwide rank second only to por- phyry copper deposits in terms of copper produc- tion and they represent the most important global source of cobalt, as well as containing resources of many other metals such as Pb, Zn, Ag, U, Au, PGE, and Re. There are several deposits of varying ages around the world, but these are overshadowed by the huge resources of only two regions, the Neo- proterozoic Central African Copperbelt (Box 3.6) and the Permo-Triassic Kupferschiefer (or “copper shale”) of central Europe. Other important de- posits include White Pine in Michigan, Udokan and Dzhezhkazgan in Kazakhstan, Corocoro in Bolivia, and the Dongchuan district of China (Misra, 2000). The formation of major SSC de- posits seems to coincide with periods of super- continent amalgamation such as Rodinia in the Neoproterozoic and Pangea in the Permo-Triassic (see Chapter 6). SSC deposits are typically hosted within an intracontinental, rift-related sedimentary sequence. The early part of the sequence was either depos- ited originally as an oxidized (red-bed), aeolian to evaporitic assemblage, or was rapidly oxidized during burial and diagenesis. This sequence is overlain by a shallow marine transgression that deposited a more reduced assemblage of shales, carbonates, and evaporites. Basin-derived ?uid circulation was promoted by the high heat ?ow conditions accompanying rapid rifting and sub- 198 PART 2 HYDROTHERMAL PROCESSES Basinal brine source T = 100–150°C pH = 4–6 fO 2 = magnetite/ hematite Zn = 5 ppm Pb = 1 ppm Cu = 0.1 ppm Carbonate aquifer system (reduced) Quartz sandstone aquifer system Evaporite + oxidized red bed aquifer system Pb-rich ores Pb > Zn >> Cu Zn-rich ores Zn > Pb >> Cu Cu-rich ores Cu > Pb + Zn Upper Mississippi Valley Examples Pine Point Laisvall Southeast Missouri Kupferscheifer Central African Copperbelt White Pine Figure 3.29 Diagrammatic representation showing the relationship between a single basinal brine and the conditions under which it might form SSC, as well as carbonate- and sandstone-hosted, MVT deposits (after Sverjensky, 1989; Metcalfe et al., 1994). ITOC03 09/03/2009 14:36 Page 198HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 199 sidence, and the very permeable environment created by porous clastic sediments and active growth faults. The connate ?uids circulating at this stage of basin evolution were saline, relat- ively oxidized, and pH neutral. Metals, in particu- lar copper, are believed to have been leached from detrital minerals such as magnetite, biotite, hornblende, and pyroxene which themselves may have been derived from erosion of a fertile base- ment assemblage (a Paleoproterozoic magmatic arc in the case, for example, of the Central African Copperbelt; Rainault et al., 1999; and Box 3.6). The aqueous transport of copper under these conditions was as a cuprous–chloride complex, probably CuCl 3 2- (Rose, 1976). Metal deposition occurred at a redox interface where the oxidized, metal-charged connate ?uids intersected overly- ing or laterally equivalent reduced sediments or ?uids. The ore deposits are generally zoned, usually at a district scale, and typical zonation is characterized by the sequence barren/hematite – native copper – chalcocite – bornite – chalcopyrite – Pb/Zn/Co sul?des – pyrite. SSC deposits can be subdivided into two sub- types (Kirkham, 1989). The ?rst, less important, subtype is hosted in continental red-beds that were probably originally oxidized, with ores pre- cipitated around locally developed zones of reduc- tion (for example, Dzhezhkazgan in Kazakhstan and Corocoro in Bolivia). The second subtype, represented by the much larger Kupferschiefer and Copperbelt deposits, is characterized by metal deposition in more reduced shallow marine sequences that were partially oxidized subse- quent to deposition, and where oxidized (red-bed) sediments occur stratigraphically beneath the ore zone. In both cases, however, ore ?uids are con- sidered to have interacted with the dominantly clastic, oxidized sedimentary sequence and it is this feature that determines the nature and prop- erties of the hydrothermal solutions. Such ?uids are likely to have been characterized by low tem- peratures (<150°C and possibly even <100°C), a neutral range of pH (5–9) constrained by carbon- ate or silicate equilibria (K-feldspar/illite), oxidized conditions (hematite and other ferric-oxyhydrox- ides stable), and moderate salinities (typically up to 20 wt% NaCl equivalent), re?ecting interac- tion of the ?uids with evaporitic rocks. Rose (1976) showed that under these conditions the CuCl 3 2- complex would have been particularly stable (Figure 3.30), although other aqueous spe- cies, such as CuSO 4 or Cu(OH) 2 2+ could also have existed. Solubility of copper as a chloride com- plex under such conditions has been estimated to be as high as 35–100 ppm (Haynes, 1986a). In Figure 3.30 it is evident that the stability ?eld of the CuCl 3 2- complex largely coincides with that of hematite stability, and that the aqueous sulfur species would be SO 4 2- rather than H 2 S or HS - . Rose (1976) points out that the conditions de- scribed above are particularly well suited to copper mobilization. By comparison, copper is relatively insoluble in normal meteoric waters and it is also unlikely to be easily leached from more reduced sediments where copper sul?des are stable. An extensive red-bed sedimentary environment where fertile detrital minerals readily break down to form a source of copper metal, and where stable CuCl 3 2- complexes result in high copper solubilities, is, therefore, ideal for the formation of Cu > (Pb + Zn) SSC deposits. The formation of SSC deposits has, in the past, been a controversial topic and models supporting syngenetic as opposed to diagenetic origins have been much debated (Garlick, 1982; Fleischer, 1984; Jowett et al., 1987; Annels, 1989; Sweeney et al., 1991). The syngenetic view, in which sul- ?des were believed to have precipitated directly out of anoxic sea water in processes akin perhaps to those giving rise to Mn deposition in the present day Black Sea (see Chapter 5), have now largely been superseded by models that view ore genesis in terms of either early or late diagenesis. A diagenetic origin is substantiated by many of the prominent features of SSC deposits, especially the broadly transgressive nature of ore zones relat- ive to lithologies, the clear relationship of metal deposition to a redox front, and replacement tex- tures in the sul?de mineral paragenetic sequence. There are, nevertheless, several processes that can be called upon to explain ore deposition and metal zonation in SSC deposits. It is evident in the Kupferschiefer deposits, for example, that metal deposition is spatially associated with a reddish oxidized zone (termed ITOC03 09/03/2009 14:36 Page 199the “Rote Fäule”) that transgresses different lithologies and comprises hematite-replacing early diagenetic sul?des. The oxidized zone repres- ents a chemical front in the footwall of the ore zone and against which a zoned sequence of base metal sul?des (chalcocite–bornite–chalcopyrite– galena–sphalerite–pyrite) is arranged. Much of the metal concentration is nevertheless con- centrated within the organic-rich shale (i.e. the Kupferschiefer itself) since it is this unit that contained concentrations of early diagenetic framboidal iron sul?des (such as marcasite or pyrite) formed by bacterial reduction of sulfate to sul?de. Sawlowicz (1992) has documented evid- ence of copper sul?de replacement of framboids in the Kupferschiefer, and the applicability of early chalcocite replacement of biogenic sul?de material to SSC deposits in general has been dis- cussed by Haynes (1986b). These features suggest low temperature processes and point to metal deposition during the early stages of diagenesis of the sedimentary succession. A schematic rep- resentation of the zonal distribution of metals against the Rote Fäule in the Kupferscheifer ores is shown in Figure 3.31a, together with a model by Metcalfe et al. (1994) that demonstrates the effect of changing the redox state of the ore ?uid on sequential metal precipitation (Figure 3.31b). In 200 PART 2 HYDROTHERMAL PROCESSES Hematite –0.4 0.4 0.0 91 1 Eh (volts) pH 3 5 7 0.8 Cu 2+ CuO O 2 H 2 O CuCl 3 2– Cu 4 (OH) 6 Cl 2 Cu Cu 2 O H 2 S SO 4 2– H 2 O H 2 Cu 2 S HS – Figure 3.30 Eh–pH plot for the system Cu–O–H–S–Cl at 25 °C (with ?S = 10 -4 m and Cl - = 0.5m). Also shown are the stability ?elds for hematite, various sulfur species, and the cuprous-chloride complex for conditions compatible with the formation of SSC deposits (after Rose, 1976). ITOC03 09/03/2009 14:36 Page 200HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 201 the latter metal precipitation is achieved by mix- ing the oxidized ore-bearing ?uid with a reduced ?uid, rather than reacting the ?uid with a reduced sediment. The effect is nevertheless the same and metals are precipitated in the sequence Cu–Pb– Zn as a function of increased mixing of the two ?uids. This is exactly the sequence observed in the Kupferschiefer ores and con?rms that this type of zoning could be a result of sequential metal precipitation from mixed connate ?uids. There are other ways of explaining the patterns of metal zoning and paragenetic sequence in SSC- type deposits. Alternative proposals prefer a late diagenetic origin for metal precipitation in SSC deposits where sulfate reduction to sul?de occurs at higher temperatures and is not bacterially induced. This could be achieved by ?uid mixing (as suggested in Figure 3.31b) or simply by interac- tion between oxidized ?uid and reduced country rock. Ripley et al. (1985) have proposed a scheme in which the pattern of zoning in SSC type ores is explained in terms of either ?uid mixing or ?uid- rock reaction. In a plot of Fe 2+ versus Cu + activit- ies (Figure 3.32) it is evident that a ?uid with a high initial Cu content will become progressively more dilute as copper sul?des precipitate. This 100 0 0 % total metal in solution 80 20 60 40 10 20 30 40 50 (b) (a) Cu Pb Zn Fe 2+ Ore Rote Fäule Fe 3+ Zn Pb Cu % reduced water in mixture Figure 3.31 (a) Section showing metal zonation in typical Kupferschiefer ores in relation to the transgressive, oxidized “Rote Fäule” zone (after Jowett et al., 1987). (b) Model showing the effects of mixing a typical SSC-type ore ?uid with a reduced ?uid on metal precipitation. The sequence of deposition is the same as the zonal pattern observed in the Kupferschiefer (after Metcalfe et al., 1994). ITOC03 09/03/2009 14:36 Page 201would produce a paragenetic sequence and zona- tion pattern that re?ects a decreasing Cu content in the sul?de ores, such as chalcocite – bornite – chalcopyrite – pyrite. Another feature of SSC deposits is that they contain appreciable amounts of other metals, such as Ag, Pb, and Zn (Kupferschiefer) and Co (Copperbelt), all of which form part of a broadly zonal arrangement of ores. Such patterns may be dif?cult to explain in terms of solubility contrasts when metal concentrations are well below their saturation levels, and in such cases a mechanism such as adsorption may be relevant (see section 3.5 above). Figure 3.15b, for example, shows the variations in adsorption ef?ciency as a function of pH. It is feasible that variable adsorption ef?cien- cies could control the uptake of different metals from red-beds into connate ?uids as well as their subsequent deposition into nearby reducing sedi- ments. This could explain regional zonation pat- terns in SSC ores, although the details of how this might relate ultimately to paragenetic sequences are likely to be complex (Rose and Bianchi- Mosquera, 1993). 3.10.2 Mississippi Valley type (MVT) Pb–Zn deposits MVT deposits, like SEDEX and SSC ores, owe their origin to ?uid circulation and metal transport/ deposition within sedimentary basins. Unlike the syngenetic to diagenetic time frame of SEDEX and SSC deposit formation, however, MVT ores are distinctly epigenetic and metals can be de- posited tens of millions of years after sediment deposition. The deposits form from relatively low temperature ?uids (<150°C) and are broadly stratabound, mainly carbonate-hosted, and dom- inated by sphalerite and galena with associated ?uorite and barite. Most MVT deposits contain signi?cantly more sphalerite than galena, with the notable exception of the Viburnum Trend in southeast Missouri, where the reverse applies. The sandstone-hosted MVT deposits (such as Laisvall and some deposits in southeast Missouri) represent a conceptually intermediate category between SSC and carbonate-hosted MVT ores (Figure 3.29) and are characterized by Pb > Zn >> Cu. MVT deposits contain a signi?cant propor- tion of the world’s Zn and Pb reserves and there are many other deposits around the world in addi- tion to the numerous mines in the MVT “type area” of the southeastern USA. In the latter region the main metal-producing districts include the Viburnum Trend of southeast Missouri (which is mainly a Pb-producing region), the Tri-State (i.e. Oklahoma, Kansas, Missouri) region, the Illinois– Wisconsin region and the Mascot–Jefferson City area of east Tennessee. Other well known MVT districts or deposits elsewhere in the world include the Pine Point and Polaris deposits of northern Canada, the Silesian district of Poland, Mechernich in Austria, the Lennard Shelf district, Sorby Hills and Coxco in Australia, and Pering in South Africa (Misra, 2000). 202 PART 2 HYDROTHERMAL PROCESSES –10 0 Log [?(Cu + )/? (H + )] –8 2468 –6 Chalcocite Log [?(Fe 2+ )/? (H + )] T = 100°C Log f O 2 = –54 Chalcopyrite Bornite Hematite saturation Pyrite Figure 3.32 Plot of Cu + /H + versus Fe + /H + activity ratios to explain the zoning and paragenetic sequence that is often observed in SSC deposits when a Cu- bearing ?uid reacts with a pyritic shale. In this situation the original copper content of the ?uid is high and then decreases as a function of Cu-sul?de precipitation to form the paragenetic sequence chalcocite–bornite–chalcopyrite–pyrite (after Ripley et al., 1985). ITOC03 09/03/2009 14:36 Page 202HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 203 The stratiform, sediment-hosted Cu–Co deposits hosted within the Neoproterozoic Katangan sequence of Zambia and the Democratic Republic of the Congo (DRC) form the Central African Copperbelt, undoubtedly one of the great metallogenic provinces of the world. Several world class deposits (such as Nchanga, Konkola, and Tenke Fungurume), together with dozens of smaller mines, combine to make this region one of the foremost copper- producing areas of the world, as well as its most important Co supplier. Although ?gures for the entire Central African Copperbelt are unreliable, it has been estimated that the region still contains resources in excess of 150 million tons of Cu metal and 8 million tons of Co metal (Misra, 2000). Circulation of sediment-hosted connate ?uids: 1 The Central African Copperbelt Less than 570 Ma Glacial diamictite Glacial diamictite (circa 750 Ma) Less than 880 Ma Stratiform Cu–Co mineralization Nchanga granite (880 Ma) Gneiss Muva metasediments Lufubu metavolcanics Paleoproterozoic basement (2050–1870 Ma) Katangan sequence Upper Kundelungu Lower Kundelungu Mwashya Roan Group Figure 1 Simpli?ed stratigraphic pro?le through the rocks of the Central African Copperbelt showing the relationship between the Paleoproterozoic granite–gneiss and volcano-sedimentary basement and the overlying Katangan sequence. Stratiform Cu–Co mineralization is largely hosted in a variety of sedimentary lithotypes in the lower part of the Roan Group, but does also occur higher up in the succession and up into the Mwashya Group. The age constraints used in this compilation are after Rainaud et al. (2002a), Master et al. (2002), and Key et al. (2001). ITOC03 09/03/2009 14:36 Page 203The Katangan sediments were deposited in an intra- cratonic rift formed on a Paleoproterozoic basement comprising 2050–1870Myr old granite–gneiss, and meta-volcanosedimentary sequences (Figure 1; Unrug, 1988). The basement is thought to represent a deformed but fertile magmatic arc that contained the ultimate source of Cu (in porphyry type intrusions; Rainaud et al., 1999), and possibly Co too. Initiation of rifting and sediment deposition followed soon after intrusion of the Nchanga granite at 880 Ma. Initially sedimentation was dominated by siliciclastic deposition, but this was followed by playa lake and restricted marine incursions to form evaporites, shales, and carbonate rocks. A major glacial diamictite (the “grand conglomerat”) occurs at the top of the Mwashya Group, the age of which (circa 750 Ma) makes it a likely correlative of the mid-Neoproterozoic “Snowball Earth” glaciation (Robb et al., 2002). The uppermost parts of the Katangan sequence (i.e. the Upper Kundelungu Group and its correlatives) now appear to be substantially younger (i.e. less than 565 Ma) than the underlying portions and are regarded as a separate basin that formed as a foreland to the Pan-African orogeny (Wendorff, 2001). Mineralization in the Central African Copperbelt is a much debated and contentious issue. Ideas on ore genesis in the 1960s and 1970s were dominated by syngenetic models, where metals were accumulated essentially at the same time as the sediments were being deposited (Fleisher et al., 1976). More recent work has shown that mineraliza- tion is more likely to be related to later ?uids circulating through the sediments, either during diagenesis and com- paction of the sequence (Sweeney et al., 1991) or signi?c- antly later during deformation and metamorphism (Unrug, 1988). Although much work is still required, it seems likely that metals were sequentially precipitated by processes related to reduction–oxidation reactions between sedi- mentary host rocks and a fertile connate ?uid. A highly oxidized, saline, and sulfate-rich connate ?uid, derived by interaction with evaporites in the Roan Group, is regarded as the agent that scavenged Cu and Co from the sedi- ments themselves. The Cu may have been introduced into the basin in Cu-rich magnetite or ma?c mineral detritus eroded from the fertile Paleoproterozoic magmatic arc in the hinterland (S. Master, personal communication). Co, on the other hand, may have been derived from penecontemporaneous sills that intrude the upper Roan Group (Annels, 1984). Cu and Co would be highly soluble as chloride complexes in a connate ?uid of this type (see section 3.10.1 of this chapter) and it is likely that very ef?cient leaching of these metals during diagenesis and oxidative alteration of the host rocks would have occurred. Precipitation of these metals would occur as the ?uid reacted, either with more reduced strata (containing organic carbon or diagenetic framboidal pyrite), or with a second, more reduced, ?uid. Metal zonation, which is a feature of mineralization throughout the Copperbelt, could, at least in part, be a product of the variable reduc- tion potentials of metals such as Cu, Co, Pb, and Zn. An early or late diagenetic model for the origin and timing of stratiform mineralization in the Central African Copperbelt is consistent with the fact that it pre-dates the major period of compressive deformation that affected the entire region during the Lu?lian orogeny (590–510 Ma; Rainaud et al., 2002b). Figure 2 shows the folded nature of the host rock strata at the Chambishi Mine in Zambia; the stratiform Cu–sul?de ore here is folded and cut by an axial planar cleavage. The majority of MVT deposits worldwide formed in the Phanerozoic Eon, but more speci?cally, in Devonian to Permian times; some deposits, such as Pine Point and the Silesian district, also formed during the Cretaceous– Paleogene period (Leach et al., 2001). The former, far more important, metallogenic epoch coincides with the assembly of Pangea when major com- pressional orogenies were active over many parts of the Earth (see Chapter 6). The later period is more speci?cally related to the Alpine and Laramide orogenies and, likewise, associated with com- pressional tectonic regimes. The timing of MVT deposits in relation to associated orogenic activity 204 PART 2 HYDROTHERMAL PROCESSES Figure 2 Folded metasediments hosting stratiform Cu–Co sul?de ore at the Chambishi Mine, Zambia. The bedding parallel ore is cut by an axial planar cleavage (photograph by Lynnette Greyling). ITOC03 09/03/2009 14:36 Page 204HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 205 provides an important clue as to the causes of ?uid ?ow and hydrothermal activity involved in their formation. It is generally accepted that topo- graphically driven ?uid pathways were critical to the development of large MVT ore districts, and that the carbonate host rocks maintained a hydro- logical connection to orogenic belts active during the period of ore deposition (Rickard et al., 1979; Oliver, 1986; Duane and De Wit, 1988). Other features believed to conceptually link most MVT deposits include a low-latitudinal setting, where high rainfall ensured an adequate ?uid reservoir, and the presence, somewhere in the ?uid ?ow system, of a high-evaporation sabkha environ- ment that resulted in the high ?uid salinities necessary for viable levels of metal solubility and transport (Leach et al., 2001). Figure 3.33 illustrates the main characteristics of MVT deposits. Regional ?uid ?ow is stimulated by compressional orogeny that results in thrust faulting and uplift and this, in turn, creates a topographic head and ?uid ?ow down a hydrolo- gical gradient. Fluid ?ow in these tectonic settings occurs over distances of hundreds of kilometers and such ?uids are implicated in the migration of hydrocarbons as well as metals (Rickard, 1976; Oliver, 1986). At the site of metal deposition MVT ores are sometimes bedded and focused along conformable dolostone/limestone interfaces, but are more commonly associated with discordant, dissolution-related zones of brecciation. The hy- drothermal ?uids that are linked to MVT deposit formation are typically low-temperature (100– 150 °C), high-salinity (>15 wt% NaCl equivalent) brines, with appreciable SO 4 2- , CO 2 and CH 4 and associated organic compounds and oil-like Tectonic compresson 0 100 km Topographically elevated orogen Foreland basin MVT deposit Probable trace of pre-mineralization fault f Dolostone or limestone Bedded-type orebody meters 0 40 Discordant-type orebody Mineralized breccia zones f Figure 3.33 Diagram illustrating the concept of hydrological continuity between a compressional orogenic belt and a foreland sedimentary basin through which orogenically and topographically driven ?uids ?ow, and within which MVT Zn–Pb deposits form. Inset shows characteristics of bedded and discordant mineralization at a deposit scale. Mineralization may be bedded and associated with dolostone/limestone sequences, or in discordant carbonate breccia zones (after Garven et al., 1993). ITOC03 09/03/2009 14:36 Page 205droplets (Gize and Hoering, 1980; Roedder, 1984). These compositional characteristics are not unlike oil-?eld brines and re?ect protracted ?uid resid- ence times in the sedimentary basin. In many MVT deposits ores occur as cement between carbonate breccia fragments suggesting that brecciation occurred either before (perhaps as a karst related feature) or during metal deposition (Anderson, 1983). Hydrothermal dissolution of carbonate requires acidic ?uids and could occur by a reaction such as: CaMg(CO 3 ) 2 + 4H + ? Ca 2+ + Mg 2+ + 2H 2 O + 2CO 2 [3.11] The ?uid itself might originally have been acidic, or hydrogen ions were produced by precipitation of metal sul?des according to the reaction: H 2 S + Zn 2+ ? ZnS + 2H + [3.12] The production of additional Ca 2+ , Mg 2+ , and CO 2 in the ?uid as a result of carbonate dissolution provides the ingredients for later precipitation of calcite and dolomite, a feature that is observed in some MVT deposits where secondary carbonate gangue minerals cement previously precipitated ore sul?des (Misra, 2000). The rather unusual nature of the ore ?uid associated with MVT ore formation raises some interesting questions concerning metal transport and deposition (Sverjensky, 1986). Zn and Pb are borderline Lewis acids (see section 3.4.1 above) and can complex with both chloride and bisul?de ligands. At high temperatures and under acidic conditions the metal–chloride complex tends to be more stable, whereas at lower temperatures 206 PART 2 HYDROTHERMAL PROCESSES –48 Log f O 2 –54 –56 –58 34567 1 0 pH –52 –50 HS – 9 8 11 SO 2– 4 H 2 S Pyrite Hematite Magnetite T = 100°C NaCl = 3 m ?S = 10 –2 m Calcite stable Pb relatively soluble ? ?Pb = 10 –5 m Reduced S = 10 –5 m Sufficient reduced S to form sul?de ore Pb relatively insoluble Figure 3.34 fO 2 –pH plot showing key mineral and aqueous species stabilities under the conditions shown. Solubility contours of ?Pb = 10 -5 m and reduced sulfur = 10 -5 m are also shown. Higher solubilities lie to the left of and above the Pb contour, and below the sulfur contour (after Anderson, 1975). ITOC03 09/03/2009 14:36 Page 206HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 207 and neutral to alkaline pH the metal–bisul?de complexes dominate. There seems little doubt, given the prevailing conditions, that metal– chloride complexes predominate during MVT ore formation (Sverjensky, 1986). Anderson (1975) pointed out that viable Zn and Pb solubilities are, however, only achieved when ?uids are either relatively oxidized (i.e. the sulfur in the ?uid is transported as SO 4 2- and not as H 2 S or HS - ), or at low pH. Figure 3.34 shows that if a ?uid was suf?ciently oxidized for SO 4 2- to have been stable then high Pb solubilities could have occurred over a range of pH. That the sites of MVT ore deposi- tion are characterized by sul?de minerals, how- ever, implies that metal precipitation must have been accompanied by the reduction of sulfate to sul?de. In some MVT deposits, such as the Pine Point district, such a process is feasible and ore deposition has been attributed to the mixing of a sulfur-de?cient ore ?uid with another carry- ing signi?cant reduced sulfur (i.e. H 2 S or HS - ) at the site of ore deposition (Beales and Jackson, 1966). Alternatively, the ore ?uid might itself originally contain signi?cant SO 4 2- , with reduc- tion and metal precipitation occurring, again either by mixing with a reduced ?uid, or simply by reaction of the ?uid with organic matter. In many MVT deposits, however, textures and geo- chemistry are incompatible with a ?uid mixing model and ores appear to have precipitated over a protracted period of time. If ?uid mixing and/or sulfate reduction did not occur then the ore ?uid itself would have needed suf?cient reduced sulfur to form sul?de minerals (i.e. in excess of 10 -5 m; see Figure 3.34). In a reduced ?uid metal–chloride complexes will only be stable under acidic con- ditions (low pH) and it has been suggested that this could not apply to MVT ?uids that have equilibrated with carbonate host rocks. It appears, however, that the CO 2 contents of ?uids asso- ciated with some MVT deposits are suf?ciently high to form low pH solutions (Sverjensky, 1981; Jones and Kesler, 1992). Such acidic ?uids will promote the stability of metal–chloride com- plexes resulting in high metal solubilities. Such ?uids also promote ground preparation by creat- ing dissolution breccias at the site of ore deposi- tion. This in turn results in metal precipitation as the ore ?uid is neutralized (equation [3.11] above) by reaction with the carbonate host rock. The Viburnum Trend, also known as the “New Lead Belt,” in southeast Missouri is the world’s largest lead-producing metallogenic province. In the mid-1980s production plus total reserves from this remarkable, 65km long, belt of mineralization amounted to some 540 million tons of ore at an average grade of 6% Pb and 1% Zn (Misra, 2000). The Viburnum Trend occurs to the west of the St Francois Mountains, a Precambrian basement outlier that is uncon- formably overlain by a Cambro–Ordovician sequence of siliciclastic and carbonate sediments. The host rocks to Pb–Zn ore are mainly Cambrian dolostones of the Bonneterre Formation, although mineralization took place more than 200 million years later during Carboniferous times. The Pb–Zn ores of the Viburnum Trend are regarded as a classic example of Mississippi Valley type (MVT) deposits. The galena–sphalerite dominated mineralization of the Viburnum Trend is hosted within a sequence of carbonate rocks that formed during a shallow marine incursion over a continental shelf (Larsen, 1977). A stromatolitic reef developed around the northern and western margins of a positive feature (the St Francois Mountains) on this shelf, resulting in the formation of biostomal dolomites that are underlain by the porous Lamotte sandstone and over- lain by the relatively impervious Davis shale (Figure 2). Accurate age dating of the actual mineralization in the Viburnum Trend, and indeed for much of the related Pb–Zn MVT styles of mineralization throughout the south- eastern USA, indicates that it formed during a relatively brief interval of time in the late Carboniferous (i.e. between about 330 and 300 Ma; Leach et al., 2001). This interval coincides with a period of Earth history characterized Circulation of sediment-hosted connate ?uids: 2 The Viburnum Trend, Missouri ITOC03 09/03/2009 14:36 Page 207208 PART 2 HYDROTHERMAL PROCESSES 01 0 km Legend Outcrops of Precambrian basement Cambrian and Ordovician sedimentary rocks Intensive mineralization Weak mineralization Major faults Missouri Indian Creek Mine Potosi No. 29 No. 27 No. 28 Viburnum Mines Magmont Mine Buick Mine Brushy Creek Mine Fletcher Mine Ozark Lead Mine St Francois Mountains Viburnum Trend Old Lead Belt Figure 1 Geological setting of the Viburnum Trend Pb–Zn mineralization and the major mines in this belt (after Kisvarsanyi, 1977). ITOC03 09/03/2009 14:36 Page 208HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 209 3.11 ORE DEPOSITS ASSOCIATED WITH NEAR SURFACE METEORIC FLUIDS (GROUNDWATER) Few hydrothermal ore deposits are linked to the circulation of ambient meteoric ?uids, or ground- waters, in the near surface environment since metals are generally not transported in these types of ?uids. A notable exception is the family of sediment-hosted uranium deposit types associ- ated with low temperature meteoric ?uid ?ow (Nash et al., 1981). One category of these deposit types is the calcrete-hosted, uranium–vanadium ores that form in arid regions. These deposits are related to high rates of evaporation in sur?cial environments and, consequently, are discussed in Chapter 4. The other category, the sandstone- hosted uranium ores, is related to groundwater ?ow and redox precipitation mechanisms, and is discussed below. 3.11.1 A brief note on the aqueous transport and deposition of uranium In nature uranium occurs in two valence states, as the uranous (U 4+ ) ion and the uranyl (U 6+ ) ion. In the magmatic environment uranium occurs essentially as U 4+ and in this form is a highly incompatible trace element that occurs in only a few accessory minerals (zircon, monazite, apatite, sphene, etc.) and is concentrated into residual melts. The uranous ion is, however, readily oxi- dized to U 6+ in meteoric waters. In most natural aqueous solutions uranous complexes are insol- uble, but U 6+ forms stable complexes with ?uoride (under acidic conditions at pH <4), phosphate (under near neutral conditions at 5 < pH < 7.5) and carbonate (under alkaline conditions at pH > 8) ligands. The oxidized form of uranium is, there- fore, easily transported over a wide range of by major compressional orogeny, during which time the amalgamation of the supercontinent Pangea was taking place (see Chapter 6). In the southeastern USA this event is recorded by the Ouachita collisional orogen and it is suggested that it was this event that stimulated ?uid ?ow throughout the region. Orogeny- or topography-driven ?uid ?ow (see sections 3.3.1 and 3.10.2 of this chapter) focused hydrothermal solutions into connected paleo- aquifers such as the Lamotte sandstone, with subsequent metal precipitation occurring in response to features such as ?uid interaction and dissolution of carbonate rocks. An interesting feature of many of the Viburnum Trend deposits W Davis shale Clastic carbonate Lamotte sandstone 0 300 meters Pb-Zn sul?de ore Fletcher mine No. 30 shaft Precambrian basement Stromatolitic reef Bonneterre Formation E Figure 2 West–east section through the Fletcher Mine, Viburnum Trend, showing the close association between mineralization, basement topographic highs, and stratal pinch-outs (after Paarlberg and Evans, 1977). is the close association between basement topography, overlying sedimentary strata pinch-outs, and mineraliza- tion. Figure 2 shows a west–east section through the Fletcher Mine where mineralization is constrained to a zone in the Bonneterre dolostones that is directly above a basement topographic high. Stratal pinch-outs against this buttress suggest that ?uids, perhaps originally ?owing through underlying successions such as the Lamotte sandstone (where some mineralization does occur) or the lower stromatolitic reefs, might have been forced upwards into the clastic carbonates, and that this was an important factor in?uencing metal precipitation. ITOC03 09/03/2009 14:36 Page 209pH conditions, whereas the reduced form of the metal is generally insoluble. Figure 3.35, an Eh–pH diagram for the system U–O 2 –CO 2 –H 2 O at 25°C, shows that for most meteoric waters in the near neutral pH range, the dominant aqueous species are likely to be U 6+ –oxide or –carbonate complexes. Uranium can be precipitated by low- ering Eh of the ?uid to form a uranous oxide, in the form of either uraninite or pitchblende. Consequently, many low temperature uranium deposit types are a product of oxidation–reduction processes where the metal is transported as U 6+ and precipitated by reduction. In the presence of high concentrations of other ligands, uranium can be transported as a variety of different complexes, and at higher temperatures (>100°C) and under acidic conditions a molecule such as UO 2 Cl + may be stable (Kojima et al., 1994). In addition to redox controls, uranium precipitation may also be pro- moted by ?uid mixing, changes in ?uid pH, and adsorption. A more complete overview of ura- nium ore-forming processes is presented in Nash et al. (1981). 3.11.2 Sandstone-hosted uranium deposits Most of the world’s productive sandstone-hosted uranium deposits occur in the USA. These de- posits occur mainly in three regions, the Colorado Plateau region (around the mutual corners of Utah, Colorado, New Mexico, and Arizona states), south Texas, and the Wyoming–South Dakota region. Smaller, generally non-viable, deposits are known from elsewhere, such as the Permian Beaufort Group sandstones of the Karoo sequence in South Africa and the Lodève deposits in France. The deposits are generally associated with Paleo- zoic to Mesozoic ?uvio-lacustrine sandstones and arkoses and are stratabound, although different geometrical variants exist. Most of the deposits in the Colorado Plateau region are tabular and asso- ciated with either organic material or vanadium (Northrop and Goldhaber, 1990). Deposits in Wyoming and south Texas, by contrast, are sinu- ous in plan and have a crescent shape in cross section. Although still stratabound these ores are discordant to bedding and are typically formed at the interface between oxidized and reduced por- tions of the same sandstone aquifer (Reynolds and Goldhaber, 1983; Goldhaber et al., 1983). It is the latter type that is commonly referred to as roll- front uranium deposits. The genesis of the two subtypes is quite different and they are discussed separately. Colorado Plateau (tabular) uranium–vanadium type Important features of tabular sandstone-hosted deposits are that they can occur stacked one on top of the other, and that the ore zones are bounded above and below by sandstones contain- ing a dolomitic cement. The generally accepted model for the formation of these deposit types is based on work done in the Henry Basin of Utah, where stacked U–V ore bodies are considered typical of the Colorado Plateau region as a whole 210 PART 2 HYDROTHERMAL PROCESSES 1.0 Eh (volts) 0 2468 0.2 pH T = 25°C P CO 2 = 10 –2 atm 10 12 –0.4 –0.2 0.4 0.6 0.8 U 4+ UO 2 2+ Uraninite UO 2 UO 2 CO 3 ° UO 2 (CO 3 ) 2 2– UO 2 (CO 3 ) 3 4– H 2 O O 2 H 2 O H 2 Figure 3.35 Eh–pH diagram showing relevant aqueous uranium species for the conditions speci?ed. For most meteoric waters in the near neutral pH range, the dominant aqueous species are likely to be U 6+ -oxide or -carbonate complexes. These will be precipitated from solution by a reduction of Eh to form U 4+ -oxide, or uraninite (after Langmuir, 1978). ITOC03 09/03/2009 14:36 Page 210HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 211 (Northrop and Goldhaber, 1990). The ore bodies of this area occur predominantly within the Jurassic, sandstone-dominated Salt Wash Member of the Morrison Formation, although on a regional scale they are slightly discordant with respect to bed- ding (Figure 3.36a). The ores are believed to have formed at the interface between two discrete, low temperature meteoric ?uids. One of these ?uids, a relatively stagnant but saline basinal brine con- taining Na + , Mg 2+ , and Ca 2+ cations and Cl - and SO 4 2- anions, underlies a low salinity, meteoric ?uid that ?ows readily along aquifer horizons bringing with it highly soluble metal species such as UO 2 (CO 3 ) 2 2- and VO + . Mixing of these two ?uids results in precipitation of metals in subhor- izontal tabular zones re?ecting the ?uid interface (Figure 3.36b). The main uranium ore mineral in these deposits is a uranous silicate, cof?nite (U(SiO 4 ) 1-x (OH) 4x ), which is associated either with syn-sedimentary organic (plant) debris, or together with a vanadium-rich chlorite in the intergranular sandstone pore spaces. Fluid mixing and metal precipitation occurred within a few hundred meters of the surface such that ?uid temperatures were unlikely to have exceeded 30–40 °C. The stagnant basinal brine is considered to have been locally derived, but inter- acted with evaporitic sequences, which explains its high Mg/Ca ratios. The overlying meteoric ?uid was essentially groundwater that scavenged U 6+ and V 4+ , as well as other oxide-soluble metals such as Cu, Co, As, Se, and Mo, from overlying, uranium-enriched tuffaceous rocks, or from breakdown of detrital Fe–Ti oxide minerals in the aquifer sediments through which it ?owed. Mixing of the two ?uids and associated diagenetic reactions ultimately gave rise to ore formation, the details of which are presented in Northrop and Goldhaber (1990). One of the products of ?uid mixing in this environment was a dolom- ite cement that precipitated from pH-related (Ca,Mg)CO 3 oversaturation within the basinal brine. The pH changes in the mixing zone appear to have been a product of the formation of vanadium-rich clays (smectite–chlorite) during compaction and sediment diagenesis. Vanadium is incorporated into the dioctahedral interlayer hydroxide sites (as V(OH) 3 ) of the clay minerals, which reduces the [OH - ] activity of the ?uid and decreases its pH. The same chemical controls are also implicated in the formation of cof?nite ore, although the actual processes are more complex and involved two stages. The uranyl–dicarbonate complex most likely relevant to this meteoric ?uid is stable over only a limited pH range (around 7–8 in the conditions applicable to Figure 3.35). A decrease in ?uid pH caused by ?uid mix- ing and V-clay formation causes a lowering of UO 2 (CO 3 ) 2 2- solubility which promotes the ad- sorption of UO 2 + molecules onto the surface of quartz grains. Once the ?uid is reduced (probably by bacterially induced SO 4 reduction and produc- tion of H 2 S) to stabilize U 4+ complexes, bonding between the uranous ion and silica leads to the formation of cof?nite (Northrop and Goldhaber, 1990). It should be noted that organic matter played an important role during the ore-forming process, as is evident from the very close asso- ciation between cof?nite and plant debris. In addition to the fact that organic matter is itself a reductant, it supplies a source of nutrient for sul- fate-reducing bacteria that in turn supply biogenic H 2 S for reduction of U 6+ and V 4+ to their insoluble lower valency forms. In the Grants mineral belt of the Colorado Plateau region uranium mineraliza- tion is associated with organic matter in the form of humate that was introduced epigenetically into the sediments by organic acid rich ?uids. The model illustrated in Figure 3.36b is particu- larly applicable to the tabular, stacked ores of the Colorado Plateau region in which there is a uranium–vanadium association. It suggests that the stacked nature of the tabular ore bodies could be the product of a two-?uid interface that episod- ically migrates upwards with time. The dominant control on ore formation is pH, although adsorp- tion, bacterial mediation, and redox reactions are also important processes. Evidence for the migration of the low pH zone is provided by the fact that the footwall zones to each ore horizon retain evidence of ore-related dolomite crystals partly dissolved by the acidic ?uids as they move upwards through the sequence. Roll-front type Although somewhat similar to tabular deposits, roll-front ores form as a consequence of very ITOC03 09/03/2009 14:36 Page 211212 PART 2 HYDROTHERMAL PROCESSES Brine (a) Middle silt Paleo- hydrological gradient Salt Wash Member Tidwell Member Tony M. orebody Frank M. orebody Mudstone Brushy Basin Member Sandstone Middle silt Tidwell Member Morrisom Formation Salt Wash Member Scale 15 m Meteoric water (Ca 2+ , HCO 3 , VO + , UO 2 (CO 3 ) 2– ) 2 Dolomite Zone of mixing Brine (Na + , Mg 2+ , Ca 2+ , Cl – , SO 4 2– ) (b) Salt Wash Member t = 1 (initial mixing of meteoric/brine waters) pH 7 V-Clay t = 2 (upward migration of mixing zone) pH 7 Tidwell Member Meteoric water Brine Organic debris V-Clay Coffinite Partly dissolved residual dolomite V-clay t = 3 (continued upward migration of mixing zone) pH 7 Ore zone 2 V-Clay Coffinite V-clay Ore zone 1 Ore zone 1 { Stacked mineralized intervals Figure 3.36 (a) The orebody geometry and nature of host rocks for the tabular uranium–vanadium deposits of the Henry Basin area of Utah, considered typical of the Colorado Plateau region as a whole. (b) Secular genetic model for the tabular, stacked orebody geometries of Colorado Plateau type deposits, involving an upward-migrating mixing zone between a lower stagnant brine and an overlying, ore-bearing, aquifer-focused meteoric ?uid (after Northrop and Goldhaber, 1990). ITOC03 09/03/2009 14:36 Page 212HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 213 different chemical processes. In its simplest form the ore genesis model for roll-front deposits that best accommodates their geometry, in both plan and section, is one where an oxidized, meteoric ?uid transporting soluble uranyl–carbonate com- plexes ?ows along a sandstone aquifer and precip- itates uranium ore at a redox front (Figure 3.37a). Up-dip of the roll-front the sandstones are altered (detrital silicate minerals are altered to clays, Fe–Ti oxide phases are leached) and oxidized (sec- ondary hematite has formed, organic carbon is biodegraded). The redox front itself is character- ized by an inner alteration envelope (marked by goethite, siderite, pyrite, or marcasite) and an outer ore zone (comprising pitchblende and/or cof?nite with pyrite and some organic carbon). Down-dip of the redox front the sandstone is relatively unaltered and contains a more reduced assemblage of pre-ore pyrite, calcite, and organic matter, with detrital mineral phases relatively intact (Figure 3.37a). In this model the redox front can be regarded as the position in space and time where the meteoric ?uid has lost its capacity to oxidize the sandstone through which it is per- colating. Eh of the ?uid changes dramatically at the redox front and soluble uranyl–carbonate complexes are destabilized with precipitation of uranous oxide or silicate minerals. Other soluble metal–oxide complexes also present in the ?uid (such as V, As, Se, Mo, Cu, Co) are likewise reduced and precipitate as various minerals, either before or after uranium depending on their relative solubilities as a function of Eh. The roll- front does not remain static and it will migrate down the paleoslope as the meteoric ?uid is recharged and the process of progressive, down- dip oxidation of the sandstone evolves with time. The down-dip migration of the redox front can be equated to the process of zone-re?ning and the ore body is “frozen in” only once the paleohydro- logical regime changes and ?uid ?ow ceases. A modi?cation of this scheme has been proposed by Goldhaber et al. (1978, 1983), who noted that some roll-front deposits are spatially associated with syn-depositional faults, and that the uranium ores are superseded by paragenetically later sul?de minerals. A ?uid mixing model was proposed to accommodate these features and it is envisaged (a) Goethite siderite Se, S Altered (oxidized) Solution flow Shale (impermeable) Hematite Calcite leached Metals leached Organic C destroyed Feldspars altered Fe-Mg silicates altered Shale (impermeable) Sandstone (aquifer) Unaltered (reduced) Uranium ore zone Pitchblende/coffinite Ore-stage pyrite Calcite Organic C V, Mo, Co, Cu As, Se Mo halo Unaltered silicates Pre-ore pyrite, calcite and carbonaceous matter preserved Relevant ionic species in solution UO 2 (CO 3 ) 2 2– UO 2 (CO 3 ) 3 4– V 4 O 12 4– AsO 4 2– SeO 3 2– MoO 4 2– (b) Aquifer Euhedral/framboidal pyrite Pre-ore sul?dation H 2 S f Roll-front formation Mixing zone Alteration envelope O 2 Alteration envelope Oxidized Reduced f Figure 3.37 (a) Cross section through a typical roll front type sandstone-hosted uranium deposit, such as those in Wyoming, South Dakota, and south Texas. Uranium mineralization occurs at the interface (or redox front) between the altered/oxidized and unaltered/reduced portions of the same aquifer (after De Voto, 1978; Misra, 2000). (b) A two-?uid mixing model for roll front deposits. Initially a reduced ?uid migrates up a syn-depositional fault (f) and sul?dizes the sandstone (top box). This is followed by down-dip migration of an oxygenated, uranium-bearing meteoric ?uid and the precipitation of uranium ore at the redox front formed in the mixing zone (bottom box). This model is after Goldhaber et al. (1978). ITOC03 09/03/2009 14:36 Page 213that a reduced ?uid migrated up the fault and sul?dized the sandstone aquifer for several hun- dred meters on either side (Figure 3.37b). This stage of ground preparation created the reduced environment responsible for locating the ore- related redox front. Evidence for this additional stage in the mineralization process is provided by euhedral pyrite (in deposits that do not contain organic matter) or framboidal pyrite (in those deposits that do). This stage was followed by, or was largely coincidental with, down-dip migra- There are many different types of ?uids circulating through the Earth’s crust and these have given rise to a wide variety of hydrothermal ore deposit types that exist in virtually every tectonic setting and have formed over most of Earth history. Juvenile ?uids exsolve from magmas, in particular those with felsic compositions, and give rise to granite (sensu lato) related ore deposits that include por- phyry Cu–Mo, skarn and greisen related Sn–W, intrusion linked Fe oxide–Cu–Au, and high sul- phidation Au–Ag ores (discussed in Chapter 2). Metamorphic ?uids, derived from volatiles liber- ated during prograde mineral reactions, are typ- ically aqueo-carbonic in composition and are associated worldwide with orogenic gold deposits that are particularly well developed in Archean and Phanerozoic rocks. Connate ?uids formed during diagenesis interact with either reduced (forming Pb–Zn dominant ores) or oxidized (form- ing Cu dominant ores) sedimentary environments. These ?uids are implicated in the formation of Mississippi Valley type (Pb–Zn) and stratiform sediment-hosted copper ores. Circulation of near surface meteoric waters can dissolve labile con- stituents such as the uranyl ion, giving rise to a variety of different sediment-hosted uranium deposits. Finally, sea water circulating through the oceanic crust in the vicinity of (ridge-related) fracturing and volcanic activity is vented onto the sea ?oor as black smokers, providing a modern analogue for the formation of Cu–Zn-dominated VMS deposits. Similar exhalative processes also tion of an oxidized, uranium-bearing meteoric ?uid and the subsequent precipitation of uranium ore at a redox front that, in this case, is repres- ented by a zone of ?uid mixing (Figure 3.37b). In certain cases, such as the deposits of south Texas, continued emanation up the fault of sul?de- bearing solutions from deep within the basin re- sulted in post-mineralization sul?dization and the growth of late-stage pyrite and marcasite that overprinted even the altered up-dip sections of the sandstone aquifer. occur in different tectonic settings and with dif- ferent metal assemblages, giving rise to sediment- hosted, Zn–Pb-dominated SEDEX type deposits. The various hydrothermal deposits described in Chapters 2 and 3 and their relationship to the different ?uid types discussed in these sections are summarized in Figure 3.38. Formation of hydrothermal ore deposits is linked not only to the generation of signi?cant volumes of ?uid in the Earth’s crust, but also to its ability to circulate through rock and be focused into structural conduits (shear zones, faults, brec- cias, etc.) created during deformation. The ability of hydrothermal ?uids to dissolve metals provides the means whereby ore-forming constituents are concentrated in this medium. Temperature and composition of hydrothermal ?uids (in particular the presence and abundance of dissolved ligands able to complex with different metals), together with pH and fO 2 , control the metal-carrying cap- ability of any given ?uid. Precipitation of metals is governed by a reduction in solubility which can be caused by either compositional changes (inter- action between ?uid and rock, or mixing with another ?uid), or changes in the physical parame- ters (P and T) of the ?uid itself. Economically viable hydrothermal ore deposits occur when a large volume of ?uid with a high metal-carrying capacity is focused into a geological location that is both localized and accessible, and where ef?- cient precipitation mechanisms can be sustained for a substantial period of time. 214 PART 2 HYDROTHERMAL PROCESSES ITOC03 09/03/2009 14:36 Page 214HYDROTHERMAL ORE-FORMING PROCESSES CHAPTER 3 215 For those readers wishing to delve further into hydro- thermal ore-forming processes, the following is a selec- tion of references to books and journal special issues. Geochemistry of Hydrothermal Ore Deposits, ed. H.L. Barnes. 1st edn 1967. Holt, Rinehart and Winston Inc., pp. 34–76. 2nd edn 1979. John Wiley and Sons, pp. 71–136. 3rd edn 1997. John Wiley and Sons, pp. 63–124 and 737–96. Gill, R. (1996) Chemical Fundamentals of Geology. London: Chapman & Hall, 290 pp. Guilbert, J.M. and Park, C.F. (1986) The Geology of Ore Deposits. London: W.H. Freeman and Co., 985 pp. Pirajno, F. (1992) Hydrothermal Mineral Deposits. New York: Springer-Verlag, 709 pp. Misra, K.C. (2000) Understanding Mineral Deposits. Dordrecht: Kluwer Academic Publishers, 845 pp. Reviews in Economic Geology: Volume 8: Barrie, C.T. and Hannington, M.D. (eds) (1999) Volcanic-associated Massive Sul?de Deposits: Processes and Examples in Modern and Ancient Settings. El Paso, TX: Society of Economic Geologists, 408 pp. Reviews in Economic Geology: Volume 10: Richards, J.P. and Larson, P.B. (eds) (1999) Techniques in Hydrothermal Ore Deposits Geology. El Paso, TX: Society of Economic Geologists, 256 pp. Reviews in Economic Geology: Volume 13: Hagemann, S.P. and Brown, P.E. (eds) (2000) Gold in 2000. El Paso, TX: Society of Economic Geologists, 559 pp. Wolf, K.H. (ed.) (1976) Handbook of Strata-bound and Stratiform Ore Deposits, volumes 1–14. New York: Elsevier. Figure 3.38 Diagram illustrating the relationship between different ?uid types and various hydrothermal ore deposit types. The diagram is relevant to both Chapters 2 and 3, and is modi?ed after Skinner (1997). Connate water water Magmatic Hydrothermal solution Deep circulation Shallow circulation Groundwater/sea water Meteoric VMS/SEDEX Kuroko, Cyprus Broken Hill, Red Dog Sandstone-hosted/surficial uranium Colorado plateau, Yeelirrie Stratiform diagenetic Cu Zambian Copperbelt, Kupferschiefer Mississippi Valley Viburnum trend Skarn King Island, MacTung Orogenic gold Kalgoorlie, Carlin Epithermal Kasuga, Hishikari Iron oxide-Cu- Au Carajas, Olympic Dam Porphyry Cu/Mo Escondida, Henderson Sn–W greisens Cornwall Approximate depth WATER HYDROTHERMAL ORE DEPOSIT TYPES Chapter 2 Chapter 3 Epigenetic Au in quartz-pebble conglomerate Witwatersrand, Sierra Jacobina Metamorphic water ITOC03 09/03/2009 14:36 Page 215ITOC03 09/03/2009 14:36 Page 216Sedimentary/Sur?cial Processes ITOC04 09/03/2009 14:35 Page 217ITOC04 09/03/2009 14:35 Page 2184.1 INTRODUCTION Once metals have been concentrated in the Earth’s crust and then exposed at its surface, they are commonly subjected to further concentra- tion by chemical weathering. The relationship between weathering and ore formation is often a key ingredient that leads to the creation of a viable deposit and many ores would not be mineable were it not for the fact that grade enhancement commonly occurs in the sur?cial environment. There are several deposit types where the ?nal enrichment stage is related to sur?cial weathering processes. Some of these deposit types are economically very important and contain ores, such as bauxite, that do not occur in any other form. Ore-forming processes in the sur?cial environment are intimately asso- ciated with pedogenesis, or soil formation, which is a complex, multifaceted process re?ecting local climate and the dominantly chemical interactions between rock and atmosphere. Soil itself is an extremely valuable resource since the world’s food production is largely dependent on its exist- ence and preservation. This chapter examines the concentration of metals in soil, as well as in the associated regolith (i.e. all the unconsolidated material that rests on top of solid, unaltered rock). Many different metals are enriched in the sur?cial environment, the most important of which, from a metallogenic viewpoint, include Al, Ni, Mn, Fe, Cu, Au, Pt, and U. Emphasis is placed here on laterites, with their associated enrichments of Al, Ni, Au, and platinum group elements (PGE). In addition, clay deposits, cal- cretes, and associated uranium mineralization, as well as the supergene enrichment of Cu in porphyry-type deposits, are discussed. Sur?cial enrichment of metals takes place in many other settings too, and is a key process, for example, in the formation of viable iron ores associated with banded iron-formations. The latter are described in more detail in Chapter 5 (Box 5.2). Sur?cial and supergene ore-forming processes Box 4.1 Lateritization – the Los Pijiguaos bauxite deposit, Venezuela Box 4.2 Supergene and “exotic” copper – the porphyry copper giants of northern Chile PRINCIPLES OF CHEMICAL WEATHERING dissolution and hydration hydrolysis and acid hydrolysis oxidation–reduction cation exchange FORMATION OF LATERITIC SOIL/REGOLITH PROFILES BAUXITE ORE FORMATION NICKEL, GOLD, AND PGE IN LATERITES CLAY DEPOSITS CALCRETES AND SURFICIAL URANIUM DEPOSITS SUPERGENE ENRICHMENT OF COPPER AND OTHER METALS ITOC04 09/03/2009 14:35 Page 2194.2 PRINCIPLES OF CHEMICAL WEATHERING From a metallogenic point of view chemical weathering can be subdivided into three processes: • Dissolution of rock material and the transport/ removal of soluble ions and molecules by aqueous solutions. Some of the principles applicable to this process at higher temperatures were dis- cussed in Chapters 2 and 3. • Production of new minerals, in particular clays, oxides and hydroxides, and carbonates. Again, this topic was discussed brie?y in the section on hydrothermal alteration in Chapter 3. • Accumulation of unaltered (low solubility) residual material such as silica, alumina, and gold. The main chemical processes that contribute to weathering include dissolution, oxidation, hydrolysis, and acid hydrolysis (Leeder, 1999). Once weathering has commenced and ?ne clay particles have been produced, cation exchange further promotes the breakdown of minerals in the weathering zone. Each of these processes is relevant to ore formation in the sur?cial environ- ment and is discussed brie?y in turn below. 4.2.1 Dissolution and hydration Certain natural materials such as halite (NaCl) and other evaporitic minerals, as well as the car- bonate minerals (calcite, siderite, dolomite, etc.), tend to dissolve relatively easily and completely in normal to acidic groundwaters. This type of dissolution contrasts with the breakdown of most rock-forming silicate minerals which dissolve less easily and do so incongruently (i.e. only cer- tain components of the mineral go into solution). The relative solubilities of different elements in surface waters depends on a variety of factors, but can be qualitatively predicted (Figure 4.1) in terms of their ionic potential (or the ratio of ionic charge to ionic radius). Cations with low ionic potentials (<3) are easily hydrated and are mobile under a range of conditions, although they will precipitate under alkaline conditions and are readily adsorbed by clay particles. Similarly, anions with high ionic potentials (>10) form solu- ble complexes and dissolve easily, but will precip- itate together with alkali elements. Ions with intermediate values (ionic potentials between 3 and 10) tend to be relatively insoluble and precip- itate readily as hydroxides. Over the pH range at which most groundwaters exist (5–9), silicon is more soluble than aluminum (Figure 4.2) and consequently chemical weather- ing will tend to leach Si, leaving behind a residual concentration of immobile Al and ferric oxides/ hydroxides. This is typical of soil formation pro- cesses in tropical, high rainfall areas and yields lateritic soil pro?les, which can also contain con- centrations of bauxite (aluminum ore) and Ni. Lateritic soils will not, however, form under acidic conditions (pH < 5) as Al is more soluble than Si (Figure 4.2) and the resultant soils (pod- zols) are silica-enriched and typically depleted in Al and Fe. Another process that contributes to the dis- solution of minerals in the weathering zone is hydration. In aqueous solutions water molecules cluster around ionic species as a result of their charge polarity and this contributes to the ef?cacy of water as a solvent of ionic compounds. Hydration of minerals can also occur directly (Bland and Rolls, 1998), good examples of which include the formation of gypsum (CaSO 4 .2H 2 O) from anhydrite (CaSO 4 ) and the incorporation of water into the structure of clays such as montmorillonite. Mineral hydration results in expansion of the lattice structure and assists in the physical and chemical breakdown of the material. 4.2.2 Hydrolysis and acid hydrolysis Hydrolysis is de?ned as a chemical reaction in which one or both of the O–H bonds in the water molecule is broken (Gill, 1996). Such reactions are important in weathering. One example occurs during the breakdown of aluminosilicate min- erals such as feldspar, and also the liberation of silicon as silicic acid into solution, as shown by the reaction: Si 4+ + 4H 2 O ? H 4 SiO 4 + 4H + [4.1] Although quartz itself is generally insoluble, it is feasible to mobilize silica over a wide range of 220 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC04 09/03/2009 14:35 Page 220SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 221 (i.e. those with a net charge defect) react with acids (H + ions) in solution. This process displaces metallic cations from the crystal lattice (see sec- tion 4.4 below), which then either go into solution or precipitate. The process is most active at sur- faces exposed by fractures, cleavages, and lattice defect sites. 4.2.3 Oxidation Oxidation (and reduction) refers essentially to chemical processes that involve the transfer of electrons. In the sur?cial environment oxygen, present in either water or the air, is the most common oxidizing agent. The element most commonly oxidized in the sur?cial environment is probably iron, which is converted from the ferrous (Fe 2+ ) to the ferric (Fe 3+ ) valence state by oxidation (loss of electrons). An example pH (see Figure 4.2b) by hydrolysis reactions that result in the formation of relatively soluble silicic acid (H 4 SiO 4 ). Another example relates to the hydrolysate elements such as Fe and Al (Figure 4.1) that are relatively soluble in acidic solutions, but will precipitate as a result of hydrolysis. The hydrolysis of aluminum, yielding an aluminum hydroxide precipitate, is illustrated by reaction [4.2]: Al 3+ + 3H 2 O ? Al(OH) 3 + 3H + [4.2] It is this type of process that results, for example, in the concentration of aluminum (as gibbsite Al(OH) 3 ) and ferric iron (as goethite FeO(OH)) in lateritic soils. Acid hydrolysis refers to the processes whereby silicate minerals break down in the weathering zone. In this process activated mineral surfaces 6 0 0.5 Cation charge (valency) 3 2 1.0 1.5 2.0 1 4 5 Oxy-anions Insoluble hydrolysates Soluble cations Ba Pb Sr Na KR b Ca Cu Li REE UTh Cu Mn Mg Fe Ni Sn Si B Be Al Fe V Cr C NPV Ionic potential = 10 Ionic potential = 3 SM o Cation Oxygen Hydrogen Mobile as hydrated cations (precipitate under alkaline conditions, adsorbed on clays) Soluble oxy-anion complexes (precipitate with alkali elements) Insoluble hydroxides immobile in acid and alkali conditions Ionic radius (Å) Ti Zr Mn Figure 4.1 Simpli?ed scheme on the basis of ionic potential (ionic charge/ionic radius) showing the relative mobility of selected ions in aqueous solutions in the sur?cial environment (modi?ed after Leeder, 1999). ITOC04 09/03/2009 14:35 Page 221of the role of oxidation in chemical weather- ing is provided by the relative instability of biotite compared to muscovite. Biotite has the formula K + [(Mg 2+ , Fe 2+ ) 3 (Si 3 Al)O 10 (OH) 2 ] - and is much more easily weathered than muscovite K + [(Al 2 )(Si 3 Al)O 10 (OH) 2 ] - because of the ease with which the ferrous iron can be oxidized to ferric iron. Weathering and the resulting oxidation of Fe in biotite leads to a charge inbalance that destabi- lizes the mineral, a process that is less likely to happen in muscovite since it contains no iron in its lattice. The presence of iron in minerals such as olivine and the orthopyroxenes is one of the main reasons why they are so unstable in the weathering zone. 4.2.4 Cation exchange Clay particles are often colloidal in nature (i.e. they have diameters < 2 µm) and are characterized by a net negative surface charge brought about by the replacement of Si 4+ by Al 3+ in the clay lattice. The negative charge is neutralized by adsorption of cations onto the surface of the colloids (see Figure 3.16, Chapter 3). The adsorbed cations may be exchanged for others when water passes through weathered material containing clay colloids and this has an effect on mineral stabilities as well as the nature of leaching and precipitation in regolith pro?les. A simpli?ed scheme (after Bland and Rolls, 1998) illustrating the effect of cation exchange at 222 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 6 –8 5 Log molar concentration –6 79 –4 pH (Insoluble) Below pH = 4.5 Al more soluble than Si Al 3 soluble 5 50 100 150 200 250 67891 01 1 pH SiO 2 (amorphous) H 2 SiO 4 H 3 SiO 4 – Si concentration (mg l –1 ) (a) (b) Alumina Al(OH) 3 insoluble Al(OH) 4– soluble (Soluble) Silica 300 25°C Figure 4.2 (a) The solubility of Si and Al as a function of pH (after Raiswell et al., 1980). (b) The solubility of amorphous silica at 25 °C (after Bland and Rolls, 1998). ITOC04 09/03/2009 14:35 Page 222SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 223 the surface of a colloidal particle is shown in reac- tion [4.3] below. In this example cation exchange promotes the solution of Ca in groundwaters: COLLOID – Ca 2+ + 2H + (aq) › COLLOID – H + + Ca 2+ (aq) [4.3] | H + 4.3 LATERITIC DEPOSITS 4.3.1 Laterite formation Laterite is de?ned as the product of intense weathering in humid, warm, intertropical regions of the world, and is typically rich in kaolinitic clay as well as Fe- and Al-oxides/oxyhydroxides. Laterites are generally well layered, due to altern- ating downward percolation of rainwater and upward movement of moisture in the regolith during seasonal dry spells, and are often capped by some form of duricrust. Laterites are econom- ically important as they represent the principal environment within which aluminum ores (baux- ite) occur. They can also contain signi?cant concentrations of other metals such as Ni, Mn, and Au, as well as Cu and PGE. Laterites form on stable continental land masses, over long periods of time. In parts of Africa, South America, and Australia laterization has been ongoing for over 100 million years (Butt et al., 2000), and, conse- quently, they are characterized by thick regolith pro?les (up to 150 m) in which intense leaching of most elements has occurred such that they no longer re?ect the rock compositions from which they were derived. As with most soil/regolith pro?les the main processes involved in lateritiza- tion can be subdivided into those of eluviation, where clays and solutes are removed from a par- ticular horizon, and illuviation, where material is accumulated, usually at a lower level. A gener- alized lateritic regolith pro?le is shown in Figure 4.3, together with an indication of the pattern of horizontally orientated leaching and retention of a variety of different elements in the regolith zones (Butt et al., 2000). The base of a lateritic regolith pro?le is charac- terized by the saprolith zone, which is highly weathered rock where the primary texture and fabric is still preserved. Given that the ?uids involved in this type of weathering are typically oxidizing and slightly acidic, the lowermost saprock zone (Figure 4.3) is characterized by the destabilization of sul?des and carbonates and the associated leaching of most chalcophile metals Released in the lower saprolite Aluminosilicates Ferromagnesians (pyroxene, olivine amphiboles, chlorite, biotite) Ca, Cs, K, Na, Rb Ca, Mg Si, Al (kaolinite); Ba (barite) Fe, Ni, Co, Cr, Ga, Mn, Ti, V (Fe and Mn oxides) Released in upper saprolite Aluminosilicates (muscovite) Ferromagnesians (chlorite, talc, amphibole) Smectite clays Cs, K, Rb Si, Al (kaolinite) Ca, Mg, Na Si, Al (kaolinite) Carbonates Released at weathering front Sul?des As, Au, Cd, Co, Cu, Mo, Ni, Zn, S Ca, Mg, Fe, Mn, Sr As, Cu, Ni, Pb, Sb, Zn, (Fe oxides; sulfates, arsenates, carbonates, alunite-jarosite) Released in the mottled and ferruginous zones Aluminosilicates (muscovite, kaolinite) iron oxides; gold K, Rb, Cs Trace elements; Au Si, Al (kaolinite) Lateritic gravel Lateritic duricrust Lag (a) Generalized lateritic regolith profile Soil Residuum Mottled zone (kaolinite matrix) Plasmic zone, mainly kaolinite and goethite (Primary fabric destroyed) Regolith Pedolith Saprolith Saprolite Saprock <20% weatherable minerals altered (Primary fabric preserved) >20% weatherable minerals altered Unaltered bedrock Cementation front Weathering front (b) Element mobility in typical lateritic profiles Host minerals Leached Partly retained (in secondary minerals) Fe laterite Mg, Li Fe, Ni, Co, Cr, Ga, Mn, Ti, V (Fe and Mn oxides) Figure 4.3 (a) A generalized lateritic regolith pro?le showing the different horizons and the terminology used in their description. (b) Generalized pattern of element mobility in lateritic regoliths (after Butt et al., 2000). ITOC04 09/03/2009 14:35 Page 223and alkaline/alkaline earth elements. The lower saprolite zone is characterized by the destruction of feldspars and ferromagnesian minerals, with Si and Al retained in clay minerals (kaolinite and halloysite). Fe oxides/oxyhydroxides also form in this zone with partial retention of some trans- ition metals in phases such as hematite and goethite. The mid- to upper-saprolite zone sees alteration of all but the most resistant of minerals as well as destruction of earlier formed secondary minerals such as chlorite and smectite. Only min- erals such as muscovite and talc tend to survive intact through this zone. The upper part of the regolith pro?le, the pedolith zone, is character- ized by complete destruction of rock fabric and leaching of all but the most stable elements. This zone is dominated compositionally by Si, Al, and ferric Fe occurring mainly in kaolinite, quartz, and hematite/goethite. A ferruginous residual zone is best developed over ma?c/ultrama?c bedrock, whereas kaolinite is more abundant over felsic lithologies. The zone is also characterized by accretionary and dissolution textures such as pisoliths and nodules, or kaolinite replacement by gibbsite or amorphous silica. Some metals (Au, Cr, V, Sc, Ga) tend to be associated, by adsorption, with Fe oxides/oxyhydroxides in the ferruginous residuum, whereas other elements are retained in the upper zones simply because their mineral hosts are particularly stable (i.e. Zr, Hf in zircon, Cr in chromite, Ti in rutile, etc.). Laterites can be subdivided into ferruginous (ferricretes) and aluminous (alcrete or bauxite) varieties. Ferricretes are characterized by a ferru- ginous residuum in the upper zone, where Fe 2 O 3 contents can be as high as 20%. They tend to form under speci?c climatic conditions where rainfall is less than about 1700 mm per year but average temperatures are high (Tardy, 1992). Under higher rainfall conditions ferricrete dissolution tends to occur and aluminum hydroxide (gibbsite Al(OH) 3 ) accumulates. 4.3.2 Bauxite ore formation Bauxitic ore, in the form of the minerals gibbsite/ boehmite and diaspore, is the principal source of aluminum metal, demand for which increased dramatically in the latter half of the twentieth century. The accumulation of an alumina-rich residuum, as opposed to one enriched in iron, in the upper zone of a lateritic pro?le is a function of higher rainfall, but also lower average temperat- ures (around 22 °C rather than 28 °C for ferric- retes) and higher humidity (Tardy, 1992). Actual alumina enrichment in the upper parts of laterite pro?les is due, at least in part, to relatively high Si mobility compared to Al, and probably re?ects near neutral pH conditions (between 4.5 and 9; see Figure 4.2a). This results in incongruent dissolu- tion of minerals such as felspar and kaolinite, where Si is leached in preference to Al, yielding a gibbsite-like residue. This process is shown schematically (after Bland and Rolls, 1998) as: feldspar – (loss of Si) › kaolinite – (loss of Si) › gibbsite (Al(OH) 3 ) [4.4] Seasonal climatic variations are also considered important to the formation of bauxitic ores as the alternation of wet and dry spells promotes ?uctuations in groundwater levels and, hence, dissolution and mass transfer. Variations in bauxitic pro?les, as well as transformation from hydrated gibbsite to the relatively dehydrated version, boehmite, or to diaspore (AlO(OH)), re- sult from such ?uctuations. The mineralogical pro?les in zones of bauxite mineralization may be quite variable. In humid, equatorial laterite zones, hydrated minerals such as gibbsite and goethite predominate, whereas in seasonally contrasted climates the ores are relatively dehydrated and boehmite–hematite assemblages form (Tardy, 1992). The redistribution of iron, and the segregation of Al and Fe, is a necessary process in bauxite formation because ferruginous minerals tend to contaminate the ore. High quality bauxitic ores require that both Fe and Si be removed, but not alumina, whereas ferricretes and conventional laterites are characterized by different combina- tions of element leaching. The interplay of Eh and pH is critical to the formation of high quality bauxitic ores as discussed in some detail by Norton (1973). The essential features of Norton’s model are shown in Figure 4.4. In terms of Eh and 224 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC04 09/03/2009 14:35 Page 224SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 225 pH, Si is leachable by hydrolysis (see equation 4.1 above) and is increasingly soluble at higher pH, whereas Al is most soluble at either very low or high pH (Figure 4.2a). Fe is most mobile as ferrous iron at low Eh and pH. In lateritic environments where concentrations of both Fe and Al occur, it is the rather special conditions where these two metals are segregated that provide the means for high quality bauxite formation. The plotting of Fe and Al solubility contours in Figure 4.4 allows one to draw an isosolubility curve where Al and Fe are equally soluble. Above this curve, at high Eh and pH, Al minerals such as gibbsite will be more soluble (and Al preferentially leached) than below this curve where Fe minerals such as hematite or goethite will be more soluble (and Fe preferentially leached). Using these constraints, several situations can be described that are relevant to laterite formation. The area of Eh–pH space encompassing ?elds 1 and 2 on Figure 4.4 will be characterized by leaching of both Fe and Al and will not result in laterite formation, but more likely Si-enriched podzol soils. Fields 3 and 4 are characterized by restricted solubility of Fe and Al and laterites and bauxites will only form if the bedrock composition itself has high iron or alu- minum contents. In ?eld 3 bauxites are unlikely to form because of preferential leaching of Al from the soil, although over a protracted time period and given the right climatic conditions, high Fe laterites will form under these conditions. Groundwater solutions forming in ?eld 5 will contain more Fe than Al and laterites will form in this environment. The optimum conditions for bauxite formation are provided in ?eld 6 where groundwater solutions will preferentially remove Fe. In this ?eld Al hydrolysates are stable, espe- cially at pH between 5 and 7, and gibbsite will accumulate. This model shows that both the ini- tial and subsequent Eh and pH of the groundwater solutions involved will impose important con- trols on the relative mobilities of Al and Fe and, consequently, on the dissolution and precipita- tion of relevant minerals during laterite and baux- ite formation. It should be noted that disequilibria and the role of organic acids in element mobility might complicate some of the generalizations made above and could account for the variability that is sometimes observed in actual weathering pro?les. The huge bauxite deposits of Jamaica represent a good example of the controls involved in the formation of aluminum ores. Thick deposits of bentonitic volcanic ash were laid down on a limestone bedrock and the former are considered to be the ultimate source of residual alumina in the bauxite deposits. Desilication of volcanic glass and other silicate minerals by rapid and ef?cient drainage through the underlying karst limestone is considered to have been responsible for the gibbsite-dominated ores (Comer, 1974). In Venezuela, by contrast, lateritic bauxites are derived by deep weathering of a granitic bedrock (Meyer et al., 2002) during prolonged uplift and erosion of the Guyana Shield starting about 35 million years ago (see Box 4.1). 1.0 –0.8 Eh (volts) –0.2 –0.4 12 –0.6 0.0 0.2 pH 4 6 8 10 0.4 0.6 0.8 1.2 2 H 2 O H 2 Hematite Magnetite O 2 H 2 O Equal solubility of Al and Fe 5 1 6 2 4 3 ?Al = 10 –6 10 –8 10 –10 ?Fe = 10 –6 10 –8 10 –4 Fe minerals more soluble than Al minerals Al minerals more soluble than Fe minerals 25°C Figure 4.4 Eh–pH diagram showing conditions relevant to the formation of laterites and bauxite ore. The solubility contours are in mol l -1 and assume equilibrium of the solution with gibbsite and hematite/goethite (after Norton, 1973). ITOC04 09/03/2009 14:35 Page 225226 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Approximately 90% of the world’s bauxite, the principal source of alumina, is obtained from laterites formed by deep weathering of continental planation surfaces exposed to hot humid climates. Global production of bauxite has risen signi?cantly in recent years and is presently at around 125 million tons per annum (Meyer et al., 2002). One of the largest bauxite deposits in the world is at Los Pijiguaos in Venezuela, where annual production is about 5 million tons from a total resource that could be as much as 6 billion tons. The Los Pijiguaos bauxite deposit formed from a Mesoproterozoic, metaluminous Rapakivi-type granite (the Parguaza granite) that was deeply eroded and weathered during Cretaceous–Paleogene times to form the Nuria planation surface (Figure 1). This profound weathering event gave rise to the development of thick lateritic soil cover in which economic concentrations of bauxite (around 50 wt% Al 2 O 3 ) form a mineable zone that aver- ages 8 m thick. The complete laterite pro?le comprises an upper concretionary zone (bauxite), underlain by a mottled or saprolitic layer and then the granitic bedrock. The bauxite layer itself can be subdivided into four zones comprising various concretionary and pisolitic textures. The deposit as a whole is subdivided into blocks that are de?ned by ?uvial channels that cut through the deposit subsequent to its formation (Figure 1). The bauxite at Los Pijiguaos is made up principally of gibbsite with lesser quartz and rare kaolinite. The original granite textures in the bauxite layer have been almost entirely destroyed, with feldspars, and to a lesser extent quartz, being largely replaced by gibbsite (Figure 3). An increasing degree of lateritization is marked by systematic compositional changes in the rock. This is demonstrated in the Al 2 O 3 –SiO 2 –Fe 2 O 3 ternary plot in Figure 2 where the process of lateritization is characterized by an increase in the alumina content. Quartz and gibbsite are inversely correlated in the bauxite ore and iron contents remain relatively constant, with an increase in Fe 2 O 3 in the most highly lateritized samples. The bauxite at Los Pijiguaos is a product of desilication, hydration and residual Al 2 O 3 concentration of the Parguaza granite, and involves a mass loss of some 60% (Meyer et al., 2002). These processes produce a high grade and high purity gibbsitic ore from which alumina can be extracted by low temperature diges- tion, making Los Pijiguaos one of the most commercially pro?table bauxite deposits in the world. Lateritization: Bauxite: the Los Pijiguaos deposit, Venezuela 10 0 km 1000 Altitude (m) 800 600 400 200 0 North South Kamarata Surface (Jurassic) Nuria Surface (Cretaceous) Incised drainage Los Pijiguaos Bauxite deposit Los Llanos Surface (Pliocene–Pleistocene) Figure 1 Schematic pro?le through the Los Pijiguaos bauxite deposit showing the concentration of laterite on a dissected late Cretaceous planation surface (after Meyer et al., 2002). ITOC04 09/03/2009 14:35 Page 226SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 227 4.3.3 Nickel laterites Laterites that are developed on ultrama?c rocks containing abundant olivine and orthopyroxene and, hence, with high Ni contents (i.e. 0.1 to 0.3%; see Chapter 2), commonly develop concen- trations of Ni silicate or oxide that are up to 10 times the original concentration. Given the relat- ive ease of mining of this type of ore compared to hard-rock magmatic sul?de ores, as well as recent developments in extractive metallurgical processes, Ni laterites are a much sought after commodity. The principles by which these ores form are relatively straightforward and basically involve eluviation of Ni from the uppermost lat- eritic residuum and concentration in underlying saprolitic illuvium as nickeliferous talc, serpen- tine, or smectite, or less commonly together with goethite. The primary olivine and orthopyroxene minerals that are the source of the Ni form the main ingre- dients of an ultrama?c rock that may originally Figure 2 (a) Ternary Al 2 O 3 –SiO 2 –Fe 2 O 3 plot showing the residual increase in alumina content with progressive lateritization in the Los Pijiguaos bauxite ores. PG is the composition of the unaltered granite. Fe 2 O 3 50 Al 2 O 3 50 Weak lateritization Moderate lateritization Strong lateritization SiO 2 Kaolinization PG Bauxitization Figure 3 Photomicrograph showing the replacement of quartz and feldspar by Kaolinite and gibbsite in bauxite ores from Los Pijiguaos (photograph courtesy of Michael Meyer). ITOC04 09/03/2009 14:35 Page 227have been part of an obducted ophiolite complex (in the case of the New Caledonia deposits in the south Paci?c) or a layered ma?c intrusion (such as at Niquelandia in Brazil). These minerals are rapidly serpentinized, in some cases prior to lateritization by interaction with sea water or during low-grade metamorphism or alteration. The alteration of olivine is by hydration to amor- phous silica, serpentine, and limonite, with the serpentine able to accommodate as much or more of the Ni as the original olivine. Most nickel laterites form on bedrock that is already largely altered to serpentinite (Golightly, 1981). In later- itic regoliths the groundwaters become progres- sively less acidic with depth (pH increases downwards from around 5 to 8.5; Golightly, 1981) and bicarbonate is the main anion in solution. Olivine breaks down completely under such conditions and this is followed by orthopyroxene, serpentine, chlorite, and talc. An important reaction in this process is given in [4.5] below (after Golightly, 1981) where olivine breaks down to smectitic clays (saponite–nontronite) and goethite: 4(Fe 2 ,Mg 2 )SiO 4 + 8H + + 4O 2 ? olivine (Fe 2 ,Mg 3 )Si 4 O 10 (OH) 2 + 6FeO(OH) + 5Mg 2+ [4.5] smectite goethite Once smectite, and also serpentine and talc, are present in the regolith, further concentration of Ni takes place largely by cation exchange, mainly with Mg 2+ . This results in the formation of a variety of unusual Ni enriched phyllosilicate minerals such as kerolite (Ni–talc), nepouite (Ni–serpentine), and pimelite (Ni–smectite). An example of such an ion exchange reaction, where nepouite forms from serpentine, is shown in [4.6] below (after Golightly, 1981): Mg 3 Si 2 O 5 (OH) 4 + 3Ni 2+ (aq) ? serpentine Ni 3 Si 2 O 5 (OH) 4 + 3Mg 2+ (aq) [4.6] nepouite The collective term applied to Ni–phyllosilicate ore minerals in lateritic environments is “gar- nierite,” after the Frenchman Jules Garnier who discovered the enormous Ni laterite ores of New Caledonia in the nineteenth century. Some Ni concentration is also associated with goethite, although the exact nature of the ?xation mech- anism is not known. It is possible that adsorption of Ni onto goethitic colloids might occur at neutral to slightly alkaline pH, as discussed pre- viously in Chapter 3 and shown in Figure 3.15b. The limonite zone in the upper part of most lat- erite pro?les (Figure 4.5) is, however, generally depleted in Ni. Enormous resources of Ni ore are located in ultrama?c rooted lateritic regoliths in many parts of the world. The thickest laterite and the richest concentrations of garnierite tend to occur where the bedrock is characterized by closely spaced jointing, as this is where maximum groundwater circulation and ?uid–rock interaction takes place. The best Ni laterite concentrations also appear to display a topographic control and tend to occur either beneath a hill or on the edge of a plateau or terrace (Golightly, 1981). This is because these deposits are sensitive to surface erosion and to ?uctuations in water table levels that in turn con- trol the distribution of zones of eluviation and illuviation. 4.3.4 Gold in laterites It is well known that gold occurs in the upper pedolithic portions of laterite weathering zones in many parts of the world, including Brazil, west Africa and Western Australia (Nahon et al., 1992). This gold takes many forms, ranging from large, rounded nugget-like particles and gold dendrites in cracks and joints, to small crystals in pore spaces. In the Yilgarn Craton of Western Australia, for example, the recovery of gold from lateritic pro?les is a major activity and some spec- tacular nuggets, occasionally over 1 kg in mass, have been found. Whereas the primary source of gold in these environments is high in silver (strictly a Au–Ag alloy with typically 5–10% Ag as well as minor Hg and other impurities), the gold that is concentrated in lateritic pro?les has been chemically puri?ed. This suggests that dif- ferential mobilization and decoupling of Au and 228 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC04 09/03/2009 14:35 Page 228SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 229 Ag takes place as meteoric ?uids percolate through the weathering zone. It is only under certain very speci?c conditions in the near surface environment that Au and Ag can be substantially mobilized. Actual measure- ments of near surface groundwaters in lateritic environments indicate that aqueous solutions are relatively acidic (pH < 5), oxidizing, and in Western Australia have low to moderate salinities because of a marine in?uence on rainfall (Mann, 1984). With increasing depth, groundwater pH tends to increase (i.e. become more alkaline) and the ?uids become more reducing as the ferric/fer- rous ratio decreases (Figure 4.6). The process of lateritization, described in section 4.3.1 above, is controlled essentially by hydrolysis as groundwa- ters react with the regolith and iron is oxidized. Two reactions re?ect the nature of the process and indicate why groundwaters become more acidic upwards in lateritic pro?les. Reaction [4.7] describes the initial breakdown, by hydrolysis, of sul?de phases such as pyrite in the lower sections of the weathering zone close to bedrock, and the production of hydrogen ions: 2FeS 2 + 2H 2 O + 7O 2 ? 2Fe 2+ + 4SO 4 2- + 4H + [4.7] Reaction [4.8] illustrates the oxidation of ferrous iron in the vicinity of the water table, to form goethite in the ferruginous pedolith, with further production of hydrogen ions (after Mann, 1984): 2Fe 2+ + 3H 2 O + O 2 ? 2FeOOH + 4H + [4.8] goethite These reactions illustrate some of the processes active during lateritization and explain why groundwaters are likely to be acidic in such envir- onments. Laterites formed over felsic rocks are prone to be more acidic than those over a ma?c substrate, as the latter will have pH buffered by bicarbonate ions formed during alteration (Mann, 1984). Experimental work (Cloke and Kelly, 1964; Mann, 1984) indicates that at low pH and high Eh, and in the presence of Cl - ions, gold in the sur?cial environment can go into solution as an AuCl 4 - complex (Figure 4.7). Substantial Au dis- solution only occurs, however, at relatively high Eh Zone of residual laterite 1–20 meters (eluviation) Zone of altered peridotite and nickel concentration 1–20 meters (illuviation) Zone of fresh peridotite Iron crust Pisolites Red limonite Yellow limonite Ore with boulders (Garnierite, Amorphous silica) Rocky ore Bedrock Pedolith Saprolith 1 2 3 4 5 6 7 Ni, Cr 2 O 3 , Al 2 O 3 (wt%) Ni Cr 2 O 3 SiO 2 MgO Al 2 O 3 Fe 2 O 3 Ni 10 20 30 40 50 60 70 80 MgO, Fe 2 O 3 , SiO 2 (wt%) Surface Figure 4.5 Descriptive pro?le and Ni ore distribution in a lateritic regolith typical of the New Caledonian deposits. The chemical pro?le clearly distinguishes the ferruginous/aluminous residual zone where Si, Mg, and Ni are leached, from the saprolith where illuviation has resulted in concentration of Ni (after Troly et al., 1979; Guilbert and Park, 1986). ITOC04 09/03/2009 14:35 Page 229when Fe 2+ is completely oxidized in the presence of free oxygen. By contrast, silver will go into solution more readily, and under more reducing conditions, as several complexes, including AgCl (the most stable) and AgCl 2 - or AgCl 3 2- (Figure 4.7). Accordingly, particles of Au–Ag alloy in the sur?cial environment will be more likely to be leached of their Ag over a broader range of conditions. A model for gold concentration in laterites is shown in Figure 4.6 and is discussed below in terms of gold– and silver–chloride speciation, as well as the likely Eh–pH conditions of a typical lateritic pro?le, shown in Figure 4.7. Using these 230 PART 3 SEDIMENTARY/SURFICIAL PROCESSES AgCl Dissolution of Au, Ag Reprecipitation of Au Pedolith O 2 O 2 AgCl Water table Low pH High Eh AgCl Fe 2+ Fe 2+ Felsic Neutral pH Low Eh Mafic Fe 2+ Bedrock Saprolith Quartz vein – gold source (a) Higher purity gold rim (Ag-depleted) Peripheral void Primary Au–Ag alloy Iron oxyhydroxide (b) Iron oxyhydroxide replacement halo, containing gold (Ag-depleted) AuCl 4 – Figure 4.6 The nature and characteristics of gold and silver redistribution in a typical lateritic pro?le. The inset diagram shows two mechanisms whereby a primary Au–Ag alloy particle could be broken down during lateritization and silver preferentially removed (modi?ed after Mann, 1984; Nahon et al., 1992). ITOC04 09/03/2009 14:35 Page 230SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 231 constraints it is apparent that primary Au–Ag alloy particles in the near surface regolith above the water table will be readily dissolved in the acidic, oxygenated, and moderately saline ground- waters prevailing in this environment. Mann (1984) has shown that the breakdown of the par- ticle and dissolution of Au and Ag actually takes place by replacement of the alloy by an iron oxy- hydroxide phase (such as goethite). The goethite itself, however, can be seen to contain tiny par- ticles of high purity gold (Figure 4.6 inset b) suggest- ing that the latter metal is rapidly reprecipitated as goethite forms, whereas silver remains in solu- tion and is transported away. This is explained by the fact that high gold solubilities are only maintained at high Eh and that the metal will precipitate with only a slight decrease in Eh under mildly reducing conditions, as described by the following reaction: AuCl 4 - + 3Fe 2+ + 6H 2 O ? Au + 3FeOOH + 4Cl - + 9H + [4.9] A decrease in AuCl 4 - solubility and precipitation of gold is, therefore, promoted by interaction of the ?uid with a more reducing regolith in which Fe 2+ and Mn 2+ contents are higher with increasing depth in the pro?le and below the water table. Redox reactions, therefore, control the precipita- tion of gold in this environment and explain its association with minerals such as goethite and manganite. It is clear in Figure 4.6 that under the same conditions in which gold precipitates, silver as AgCl remains in solution and will be trans- ported away from the site of gold deposition. Primary Au–Ag alloy particles that are not closely associated with an iron or manganese oxy- hydroxide tend to show a preferential depletion in silver around their edges and the formation of higher purity gold halos (Figure 4.6 inset a). Such features are a product either of silver diffusion from the outer margins of the particle, or of com- plete dissolution of the entire alloy and in situ reprecipitation of the gold. It should also be noted that the ef?cacy of chloride ion ligands in promot- ing gold and silver dissolution is particularly appropriate to laterites that presently occur in arid climates, such as Western Australia. In humid, tropical regions chloride ion activity is reduced by high rainfall dilution and other compounds, such Eh (volts) –0.4 12 0 pH 4 6 8 10 0.4 0.8 1.2 2 H 2 O H 2 14 O 2 H 2 O AuCl 4 – Fe 3+ Fe 2+ Au Pt Pd Pt Pt O 2 . × H 2 O Pd AgCl Ag PdCl 2– 4 PtCl 2– 4 Pt (OH) 2 Pd (OH) 2 Lake water River water Lateritic profile Figure 4.7 Eh–pH diagram showing the characteristics of a lateritic pro?le in relation to lacustrine and riverine waters. Also shown are the redox potentials and stabilities of Au, Ag, Pt, and Pd as chloride complexes, illustrating the range of Eh that controls solubility of the various metals (information compiled from data in Fuchs and Rose, 1974; Mann, 1984; Bowles, 1986). ITOC04 09/03/2009 14:35 Page 231as organic humic and fulvic acids, may play a more important role in gold dissolution (Nahon et al., 1992). Finally, it should be noted that microorganisms play an important role in the concentration of gold in lateritic soils. Secondary gold, in the form of the nuggets found either in laterites or in other environments where placer gold deposits occur, are sometimes characterized by spherical morphologies consistent with the size and shape of bacteria (Watterson, 1991). Southam and Beveridge (1994) have shown experimentally that Bacillus subtilis, a common bacterium in soils, is capable of accumulating gold by diffusion across the cell wall and precipitation within the cytoplasm. It is suggested that gold was stabilized internally as an organo-gold complex and then precipitated in colloidal form once a critical con- centration had been reached. Subsequent diage- nesis of sediment containing such gold-enriched microorganisms would result in recrystallization and coalescence of gold to form nugget like shapes. 4.3.5 A note on platinum group element (PGE) enrichment in laterites In addition to gold in laterites, it is now widely appreciated that PGE can also be concentrated in weathering pro?les. Laterites in Sierra Leone, for example, are known to contain PGE deposits (Bowles, 1986) and signi?cant concentrations of these precious metals also occur in non-lateritic soils such as over the Stillwater Complex in Montana and in Quebec (Fuchs and Rose, 1974; Wood and Vlassopoulos, 1990). Evidence for the mobilization of PGE in the weathering zone is provided by the existence of well formed, crys- talline Pt–Fe or Os–Ir–Ru alloys in pedolith, and in some cases PGE-sul?de minerals, that are com- positionally different, and often larger, than those in the source rock (Bowles, 1986). In the case of laterite-hosted PGE concentrations, it is consid- ered likely that the controls on Pt and Pd solubil- ity were similar to those affecting Au and Ag and that these precious metals readily go into solution as chloride complexes in acidic, oxidized environ- ments (Bowles, 1986). Figure 4.7 shows that PdCl 4 2- is stable over a broader range of Eh and pH than PtCl 4 2- , a feature that is consistent with observa- tions that palladium is more readily dissolved and concentrated in groundwaters than is platinum (Wood and Vlassopoulos, 1990). Like gold and silver, platinum and palladium tend to become decoupled in the weathering environment, with Pd going into solution and Pt concentrating in soils and sediment. Similarly, redox reactions control the precipitation of PGE in this environ- ment, with Pt, like Au, precipitating before Pd as Eh is reduced. In non-lateritic environments, where groundwaters are near neutral to alkaline and more reducing, PGE will not be transported as chloride complexes but more likely as neut- ral hydroxide complexes (PdOH 2 and PtOH 2 ). Decoupling of Pt and Pd still occurs in non- lateritic weathering pro?les, although it is sug- gested that whereas Pd is dispersed in solution, Pt, together with Au, is less easily mobilized, perhaps as colloidal particles (Wood and Vlassopoulos, 1990). Finally, it should be noted that there is evidence pointing to an effective interaction be- tween PGE and organic acids, in particular humic acid, in weathering zones (Wood, 1990; Bowles et al., 1994). Pt and Pd are reasonably soluble in natural solutions containing humic acid, although the exact nature of the complex involved is not known. Humate is also thought to promote PGE transport as colloidal suspensions, whereas solid organic debris may also be responsible for precip- itation of precious metals. A spectacular example of where both Au and PGE have been concentrated into a deeply weath- ered lateritic pro?le is the Serra Palada deposit in the Carajás mineral province of Brazil. This deposit was discovered and exploited exclusively by artisinal miners who took advantage of the rich and easily extractable laterite-hosted ore. Average grades are reported to be around 15 g ton -1 Au, 4 g ton -1 Pd, and 2 g ton -1 Pt, with the primary ore body thought to resemble other Fe oxide–Cu–Au deposits (such as the Olympic Dam deposit; see Box 3.1, Chapter 3) found in the Carajás province (Grainger et al., 2002). Weathering has resulted in a lateritic pro?le dominated by kaolinite, manganese oxides, and Fe oxides/oxyhydroxides. Amorphous carbon is present as the weathered residue of an originally carbonaceous siltstone. 232 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC04 09/03/2009 14:35 Page 232SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 233 Most of the ore constituents have been extens- ively remobilized during lateritization, with base metals occurring in Mn oxides and Au–Pt–Pd occur- ring together with amorphous carbon and Fe–Mn oxides/oxyhydroxides. Some of the ore also occurs as nuggets of Au and PGE alloy, some of which are reported to weigh as much as 60 kg (Grainger et al., 2002). 4.4 CLAY DEPOSITS Clay minerals are volumetrically the most ab- undant product of weathering and either occur in situ or have been transported to a site of deposi- tion. They are also economically important and used for a variety of industrial applications includ- ing paper, ceramics, ?ltration, and lubricants. Clay formed during weathering re?ects both the nature of the source material and the weather- ing conditions. Nevertheless, very similar clay mineral assemblages ultimately form from both felsic and ma?c rocks (Figure 4.8). Temperature and rainfall in particular, as well as local Eh–pH conditions, determine the nature and rates of clay-forming processes. Some of the more important clay minerals considered here include kaolinite, illite and the smectite group (includ- ing montmorillonite). Kaolinite is formed under humid conditions by acid hydrolysis of feldspar- bearing rocks (e.g. in laterites), whereas illite forms under more alkaline conditions by weather- ing of feldspars and micas. The smectite clays commonly weather from intermediate to basic rocks under alkaline conditions and are highly expandable, containing intracrystalline layers of water and exchangeable cations. It should be noted that clay minerals are not only the result of weathering processes but can also form as the products of low temperature hydrothermal altera- tion (see Chapter 3). Clay formation is initiated during chemical weathering by acid hydrolysis (see section 4.2.2 above and Figure 4.9). A typical reaction that describes the process is shown below: (Ca,Na)Al 2 Si 2 O 8 + 2H + + H 2 O ? plagioclase Al 2 Si 2 O 5 (OH) 4 + Na + + Ca 2+ [4.10] kaolinite 80 02 0 Relative clay mineral content (%) 20 40 80 40 60 Mean annual rainfall (inches) 60 MAFIC ROCKS Vermiculite Kaolinite Smectite Gibbsite (b) 80 0 20 Relative clay mineral content (%) 20 40 80 40 60 Mean annual rainfall (inches) 60 FELSIC ROCKS Vermiculite Kaolinite Smectite Gibbsite (a) Illite Figure 4.8 The effect of rainfall on clay mineral formation in (a) felsic and (b) ma?c rocks (after Barshad, 1966). ITOC04 09/03/2009 14:35 Page 233In the above reaction plagioclase feldspar is bro- ken down during weathering to form kaolinite, with the release into solution of Na and Ca ions. The actual mechanisms of weathering are undoubtedly more complex and may involve several incremental stages during the formation of clays. A schematic diagram illustrating the various stages involved during the weathering and breakdown of K-feldspar by acid hydrolysis is shown in Figure 4.9. Alteration by acid hydrolysis is most relevant at acidic pH and the process is kinetically less effective under more alkaline conditions, where other processes such as Si- hydroxylation may be more ef?cient. The mobility of different elements in the sur?cial environment varies considerably (Figure 4.1) and this has led to the recognition of a hierarchy of mobility, as follows: Ca > Na > Mg > Si > K > Al = Fe [4.11] The alkaline and alkali earth elements are typ- ically the most soluble in groundwaters, with Al and ferric Fe being relatively immobile. The combination of kinetic and thermodynamic con- straints on clay mineral formation during weather- ing results in a well de?ned sequence of reaction pathways that can be related, for example, to climatic conditions. The effects of rainfall on the formation of clay mineral assemblages is shown in Figure 4.8, for both felsic and ma?c parental rock compositions. The general pattern is that smectitic clays tend to form under relatively arid to semi-arid conditions, whereas kaolinite tends to dominate in wetter climates. This pattern may also re?ect a paragenetic sequence of clay forma- tion, with smectites forming relatively early on and being superseded, or replaced, by kaolinite or vermiculite. 234 PART 3 SEDIMENTARY/SURFICIAL PROCESSES H + (c) H 2 O H + Si H + H + In solution as silicic acid K + O – – – – (b) K + K + H + H 2 O H + – – H + H + H + – – Exchange of K + for H + at surface (protonation) Defect lattice site In solution (a) K + K + H + H + O Si Al O – – Capillary fringe of H 2 O (with H + ) – K-feldspar – Figure 4.9 (left) Representation of the stages during the progressive weathering and breakdown of K-feldspar by acid hydrolysis, leading to the formation of clay minerals (modi?ed after Leeder, 1999). ITOC04 09/03/2009 14:35 Page 234SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 235 4.4.1 The kaolinite (china clay) deposits of Cornwall The Cornubian batholith of southwest England, so well known for its polymetallic magmatic- hydrothermal mineralization (Box 2.2 in Chap- ter 2), also hosts some of the largest and best quality kaolinite (or china clay) deposits in the world. Of considerable demand in the paper and ceramic industries, the kaolinite resource of Cornwall surpasses the base metal resource in value, and considerable reserves of high quality china clay still exist. The formation of the Cornish kaolinite deposits is complex and controversial. Some workers have argued in favor of a weathering-related origin (Sheppard, 1977), but this has been countered by proponents of a hydrothermal process who main- tain that clay formation represents the end stage of a very long-lived paragenetic sequence initiated soon after granite emplacement and terminated with circulation of low temperature meteoric ?uids during clay formation (Alderton, 1993). The actual processes of kaolinite formation are essen- tially the same irrespective of the actual ?uid origin and on balance it seems likely that both a late hydrothermal process and a tertiary weather- ing event contributed to clay formation. Zones of intense kaolinization are best devel- oped in areas where greisen altered Sn–W ore veins and major NW–SE trending faulting have occurred. This suggests a genetic link to previ- ous, higher temperature, hydrothermal processes, but could also point to groundwater circulation focused into sites previously accessed by high temperature ?uids. Kaolinite itself is preferenti- ally developed from the plagioclase feldspar com- ponent of the granite, while K–feldspar remains relatively stable except in zones of extreme alteration/weathering. This is consistent with the mobility hierarchy described in [4.11] above, where Ca and Na are more easily leached than K during acid hydrolysis. The process is also demon- strated in reaction [4.10] although, as mentioned previously, the actual mechanism is likely to be more complex. Many workers have described an intermediate stage of either muscovite or smec- tite formation prior to the ultimate formation of kaolinite (Exley, 1976; Durrance and Bristow, 1986). This, too, is consistent with the pattern of progressive clay mineral formation illustrated as a function of rainfall in Figure 4.8. It seems likely, therefore, that acid hydrolysis type reactions gave rise to early alteration of plagioclase feldspars to form sericite (?ne grained muscovite), as well as incipient argillic alteration. Exhumation of the batholith during Mesozoic times, together with further meteoric ?uid circulation and deep weathering, continued the alteration process to form zones of intense kaolinization, particularly in areas where previous alteration had taken place. It appears likely that kaolinization was also accompanied by leaching of Fe from the altered granite, a process that contributed to the very high purity and “whiteness” of the Cornish china clay. 4.5 CALCRETE-HOSTED DEPOSITS Most of the laterally extensive surface or near- surface calcrete (or caliche) layers that typify arid environments around the world are referred to as “pedogenic calcrete” because they represent calci?ed soils (Klappa, 1983). Calcrete is de?ned simply as an accumulation of ?ne grained calcite (CaCO 3 ) in the vadose zone (i.e. above the water table) during a combination of pedogenic (soil forming) and diagenetic (lithi?cation) processes. The solution and precipitation of calcite in the Figure 4.10 Strongly kaolinized granite in Cornwall, being mined for “china clay.” ITOC04 09/03/2009 14:35 Page 235sur?cial environment is represented by the fol- lowing general equation (after Klappa, 1983): Ca 2+ + 2HCO 3 - ? CaCO 3 + CO 2 + H 2 O [4.12] The solubility of CaCO 3 increases with decreas- ing temperature, with lower pH, and with increas- ing partial pressure of CO 2 . Thus, extraction of CO 2 from an aqueous solution will precipitate CaCO 3 . Precipitation of calcite may also result when water is removed from a soil pro?le during evaporation and evapotranspiration, resulting in an increase in the concentration of aqueous ions in the solution. A simpli?ed classi?cation of calcrete types is shown in Figure 4.11 where it is evident that, in addition to the very extensive pedogenic calcretes, a more locally distributed non-pedogenic variant, termed valley or channel calcrete, also occurs. It is the latter variety that is particularly important as the host rock for sur?cial uranium deposits. 4.5.1 Calcrete-hosted or sur?cial uranium deposits Important uranium resources have been discov- ered in channelized calcretes from arid regions in Australia and Namibia. The Yeelirrie (Western Australia) and Langer Heinrich (Namibia) deposits represent well known examples of sur?cial ura- nium ores formed by accumulations of a bright yellow potassium–uranium vanadate mineral, carnotite (K 2 (UO 2 ) 2 (V 2 O 8 ).3H 2 O), within calcretized ?uvial drainage channels (Carlisle, 1983). Since carnotite is a uranyl (U 6+ )–vanadate, reduction of hexavalent uranium is not the main process involved in the precipitation and concentration of uranium ores (see Chapter 3). Rather, the forma- tion of this type of uranium deposit is related to high rates of groundwater evaporation and the resultant decrease of aqueous carbonate, vana- dium, and uranium solubilities within a few meters of the surface. Calcretized ?uvial channels represent the remnants of rivers from a previous higher rainfall interval. Such channels occasion- ally drained a uranium-fertile source region and, where preserved, represent zones of focused groundwater ?ow within which mineral precip- itation resulted in concentration of uranium ore. In this process uranium and other components are required to remain in solution until they reach a zone where carnotite can be precipitated by evaporation of the groundwater. At Yeelirrie in Western Australia, ?uvial 236 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Non-pedogenic groundwater calcrete H 2 O CO 2 liquid H 2 O CO 2 vapor Soil moisture zone Gravitational water zone Capillary fringe Water table Calcretes Vadose Groundwater Phreatic Soil forming processes Descending water Capillary transport Process Calcrete classification Predominant transport in solution Surficial transport Vertical redistribution (Minor uranium) (Essentially no uranium) Lateral transport (Largest uranium potential) Non-pedogenic superficial calcrete Pedogenic calcrete Valley (channel) calcrete Deltaic calcrete Lake margin calcrete Figure 4.11 Simpli?ed classi?cation scheme for calcretes and a summary of the processes by which they form. An indication of the suitability of calcretes for hosting sur?cial uranium deposits is also provided (after Carlisle, 1983). ITOC04 09/03/2009 14:35 Page 236SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 237 channels started to incise the bedrock during Paleogene times, although the period of aridity commenced only in the late Pliocene. Active cal- crete deposition and carnotite ore formation are thought to have occurred in only the past 500 000 years (Carlisle, 1983). Calcrete crops out along the axis of a paleochannel that can be traced for over 100 km and within which a carnotite bearing orebody, some 6 km long, 0.5 km wide, and up to 8 m thick with a resource of about 46 000 tons of U 3 O 8 , is de?ned (Figure 4.12a). The Langer Heinrich deposit in Namibia is remarkably simi- lar in most respects to Yeelirrie. The main ingredients of carnotite, U, K, and V, are derived locally from weathered granites (for the uranium and potassium) and possibly more ma?c rocks for the vanadium. In relatively car- bonated groundwaters under near neutral con- ditions, U 6+ would be transported as a carbonate complex (see Figure 3.36), whereas vanadium was probably in solution as V 4+ , although the exact species and complexing agent is not known (b) Alluvial plain Water table pH 4.5–7.0 pH 6.0–7.0 Weathered Archean granite Calcrete channel Basement high Salinity increase pH 7.0–8.5 Groundwater flow Increasing Eh Zone of carnotite mineralization High evaporation (CO 2 + H 2 O) Calcrete channel Archean granite Yeelirrie Yeelirrie Orebody Watershed Lake Miranda 10 km (a) Channel Figure 4.12 (a) Geological setting of the Yeelirrie carnotite deposit hosted in channelized calcrete, Western Australia (after Mann and Deutscher, 1978; Carlisle, 1983). (b) Model depicting the setting and processes involved in the formation of carnotite deposits in calcretized channels (after Carlisle, 1983). ITOC04 09/03/2009 14:35 Page 237(Mann and Deutscher, 1978). Focused ground- water ?ow introduced the ore-forming ingredients into the channel, with the combination of high evaporation rates and calcite precipitation ensur- ing that the solution evolved toward higher salin- ities and pH along the ?ow path (Figure 4.12b). In detail, however, the actual precipitation of carnotite is a complex process and appears to be related either to evaporation and decomplexation of uranyl–carbonate complexes, or to oxidation of V 4+ to V 5+ , or both (Mann and Deutscher, 1978). A general equation for the precipitation of carnotite can be written as follows (after Carlisle, 1983): 2UO 2 2+ + 2H 2 VO 4 - + 2K + + 3H 2 O ? K 2 (UO 2 ) 2 (V 2 O 8 ).3H 2 O + 4H + [4.13] carnotite Evaporation on its own removes CO 2 from solution and drives equilibrium equations such as [4.12] and [4.13] above to the right, promoting pre- cipitation of calcite and carnotite. Precipitation of carnotite, however, produces hydrogen ions (see equation [4.13]), which lowers pH and increases the solubility of calcite. Dissolution of calcite provides additional CO 3 2- to the solution, which in turn increases uranium solubility and favors dissolution of carnotite. The processes described in terms of equations [4.12] and [4.13] appear, therefore, to be counter-productive and this is evident because the orebodies themselves provide textural evidence for repeated dissolution and reprecipitation of ore and gangue minerals. In reality, however, the very high rates of evapora- tion increase the Ca 2+ and Mg 2+ concentrations of the groundwater to such an extent that calcite (as well as dolomite) is forced to precipitate and carbonate activities are decreased. Formation of carbonate minerals, under conditions in which they would not otherwise have precipitated, also promotes the destabilization of uranyl–carbonate complexes and eventually results in carnotite formation coeval with calcretization. Mann and Deutscher (1978) have suggested that oxidation of vanadium may also play a role in carnotite formation, since V 4+ is soluble in mildly reducing waters at near neutral pH, but is precipitated as V 5+ with increasing Eh. They envisage that vana- dium might have been transported separately in deeper groundwaters ?owing below the cal- cretized channel and that this ?uid mixed with the sur?cial, oxidized, U- and K-bearing ?uid to form carnotite in a mixing zone. The migration of the lower V-bearing ?uid to the near surface environment is caused by high points in the basement topography (Figure 4.12b) which force groundwaters upwards into the evaporative zone to mix with oxidized ?uids, or be directly oxid- ized by interaction with the atmosphere. This notion is consistent with the common observa- tion that carnotite mineralization is spatially related to pinch-outs and constrictions in the ?uid ?ow path. 4.6 SUPERGENE ENRICHMENT OF CU AND OTHER METALS IN NEAR SURFACE DEPOSITS The processes of weathering can be responsible for the in situ enrichment of Cu, as well as other metals such as Zn, Ag, and Au, in many deposits that occur at or near the surface. The process is generally referred to as supergene enrichment and is a product of oxidation and hydrolysis of sul?de minerals in the upper portions of weather- ing pro?les. It is an extremely important process in the formation of low-grade porphyry copper deposits because the presence of an enriched, easily extractable supergene blanket of secondary copper ore minerals above the primary or hypo- gene ore is often the one factor that makes them economically viable. The processes involved in the formation of supergene mineralization are similar to those already discussed in section 4.3.4 above and applicable to the concentration of gold nuggets or platinum group elements in lateritic weathering zones. This section focuses speci- ?cally on copper and the formation of supergene enrichments in porphyry-type deposits, although the principles also apply to other deposit types, in particular VMS and SEDEX base metal ores. 4.6.1 Supergene oxidation of copper deposits Unlike the processes applicable to gold and nickel, copper enrichments are not speci?c to lateritic environments and supergene copper deposits occur in any sur?cial environment where oxidized, 238 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC04 09/03/2009 14:35 Page 238SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 239 Primary, unaltered hypogene ore Copper carbonates, oxides, silicates, etc. Secondary sul?de enrichment Water table (redox barrier) Oxidized ores Leached zone Leached capping (gossan) Mineralized vein Water seepage Weathered surface Hydrated iron oxides MINERALS Cu 2+ Chalcocite, covellite, bornite Chalcopyrite, pyrite Reducing oxidizing Supergene enrichment Cu 2+ WEATHERING ZONES Illuvial Eluvial acidic groundwaters are able to destabilize sul?de minerals and leach copper. The principles involved in this process are illustrated in Figure 4.13 and also summarized in terms of the reaction below: 4CuFeS 2 + 17O 2 + 10H 2 O ? chalcopyrite 4Fe(OH) 3 + 4Cu 2+ + 8SO 4 2- + 8H + [4.14] goethite In most porphyry copper environments pyrite is the dominant sul?de mineral and its hydrolysis and oxidation dictates the production of hydrogen ions (i.e. the decrease in pH) in the weathering zone (see equations [4.7] and [4.8] above). Pyrite breakdown is also accompanied by the formation of goethite in the regolith and the liberation of SO 4 2- . Chalcopyrite is the major copper–iron sul?de mineral and its breakdown, described in reaction [4.14], produces soluble cuprous ions that are dissolved in groundwater solutions. In the regolith pro?le, therefore, oxidizing, acidic groundwaters leach the primary, hypogene ore- body of its metals, leaving behind an eluviated zone that may, if the intensity of leaching is not too severe, be residually enriched in iron, as hematite or goethite/limonite. The upper clay- and Fe–oxyhydroxide-rich capping, which also contains the skeletal outlines of the original sul?de minerals, is referred to as a gossan. Gossans are very useful indicators of the previous existence of sul?dic ore, not only in porphyry systems, but in many other ore-forming environ- ments too. The soluble copper ions percolate downwards in the regolith pro?le and encounter progressively more reducing conditions, either as a function of the neutralization of acid solutions by the host rock, or at the water table. Copper is then precipitated as various secondary minerals, the compositions of which re?ect the ground- water composition as well as local pH and Eh in the supergene zone. The so-called “copper-oxide” minerals that precipitate here are actually com- positionally and mineralogically very complex and can include a variety of copper –carbonate, –silicate, –phosphate, –sulfate, – arsenate, as well as –oxyhydroxide phases (Chávez, 2000). In addition, Cu also replaces pre-existing sul?de minerals (i.e. pyrite and chalcopyrite) in more reduced zones where the latter minerals are still stable. Cu typically replaces the Fe in hypogene minerals such as pyrite and chalcopyrite, to form a suite of Figure 4.13 Schematic section through a copper deposit showing the typical pattern of an upper, oxidized horizon (the leached or eluvial zone) overlying a more reduced zone of metal accumulation (the supergene blanket or illuvial zone). The uppermost zone of ferruginous material, often containing the skeletal outlines of original sul?de minerals, is known as gossan. The redox barrier may be the water table or simply a rock buffer (after Webb, 1995). ITOC04 09/03/2009 14:35 Page 239Cu-enriched sul?de phases including chalcocite (Cu 2 S), covellite (CuS), and bornite (Cu 5 FeS 4 ). A more detailed illustration of the distribution of stable mineral assemblages in a typical super- gene weathering pro?le over a porphyry copper deposit is shown in Figure 4.14. This simulation integrates the effects of time (or more accurately the evolution of the chemical system in terms of ?uid/rock ratio) in much the same way that altera- tion processes were described in dynamic terms as a function of evolving ?uid/rock ratios in sec- tion 3.6 of Chapter 3. The primary, hypogene ore is considered to be hosted in a granite and com- prises pyrite + chalcopyrite that is weathered by a ?ux of slightly acidic rainfall. Within about 6000 years (period 1 in Figure 4.14) the primary sul?des are totally dissolved from the leached zone and copper is reprecipitated as chalcocite and covel- lite by replacement of primary sul?des in the supergene blanket, as shown in reaction [4.15]: CuFeS 2 + 3Cu 2+ ? 2Cu 2 S + Fe 2+ [4.15] chalcopyrite chalcocite The stable assemblage in the leached (gossanous) zone during stage 1, and also into stage 2, is hematite/goethite, alunite, and quartz. The dis- solution and/or replacement of primary sul?des is differential and chalcopyrite tends to disappear from both leached and supergene zones before pyrite, which is only removed completely from the upper zone by stage 2. With time, and as the ?uid/rock ratio increases, the leached zone devel- ops gibbsite, muscovite/sericite, and minor clay minerals (stage 3), eventually evolving into a unit dominated by sericite and kaolinite (stage 4). 240 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 5000 10 000 Time (years) LEACHED ZONE Hematite/Goethite Alunite Ca-Nontronite Gibbsite Kaolinite Muscovite Quartz SUPERGENE BLANKET Bornite Chalcocite Covellite Pyrite Alunite Ca-Nontronite Kaolinite Muscovite Quartz PRIMARY HYPOGENE ORE Chalcopyrite Pyrite Albite Biotite Ca-Nontronite Epidote K-Feldspar Muscovite Na-Nontronite Quartz 1 2 3 4 Supergene blanket formation Stages Figure 4.14 Illustration of the stable mineral assemblages modeled as a function of time in the weathering zones above a porphyry copper deposit. The primary, hypogene ore (least altered) is compared to the overlying supergene blanket and the oxidized, leached zone. Note that with time (from stages 1 to 4) the mineral assemblages evolve as certain phases are consumed and others precipitated (after Ague and Brimhall, 1989). ITOC04 09/03/2009 14:35 Page 240SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 241 Alunite disappears and SO 4 2- is removed from the weathering zone. The pH of groundwaters in the leached zone starts off low (acidic) during the sul?de dissolution stage and then evolves to slightly more alkaline conditions with time. The supergene blanket initially develops an assem- blage comprising covellite and chalcocite, as the primary sul?des are replaced, together with quartz and alunite. Once chalcopyrite is con- sumed from the leached zone, the downward migrating ?ux of Cu ions in solution is reduced and bornite develops later in the paragenetic sequence (i.e. from stage 2 onwards). Again in this zone, the clay minerals and muscovite/sericite are stabilized by progressive hydrolysis of primary feldspars and micas and, ultimately, the stable mineral assemblage is not unlike that of the overlying leached zone except that copper sul?de (bornite) is still stable. In the primary host rock weathering has only limited effects, although over time relatively unstable minerals such as plagioclase, biotite, and magnetite will tend to disappear. It is interesting to note that sulfate (SO 4 2- ), produced in abundance during the oxida- tion of primary sul?des (see reaction [4.14] above), is generally transported away from the in situ weathering zone, and very little sulfate reduction (either organic or inorganic) tends to occur in the supergene blanket, or below the water table. This is another reason why low-S copper sul?des such as covellite and chalcocite form in this environment. In addition to the formation of high metal/ sulfur copper sul?de minerals at a redox barrier in the supergene blanket, it is apparent that a diverse suite of secondary copper minerals can form at the base of the oxidized zone (Figure 4.13). This suite includes oxides such as cuprite and tenorite (the former usually associated with native Cu), carbonates such as malachite and azurite, sulfates such as brochantite, antlerite, and chal- canthite, chlorides such as atacamite, silicates such as chrysocolla, and phosphates such as libethinite. The precipitation of these oxidized secondary minerals is generally due to direct pre- cipitation from groundwaters that are saturated with respect to one or more of the various com- ponents that make up this suite of minerals. Fig- ure 4.15 shows an Eh–pH diagram that identi?es the stability ?elds of several of the copper min- erals encountered in supergene enrichment zones above porphyry copper deposits. Under relatively low Eh conditions, below the water table, the high metal/sulfur copper sul?des are stable, whereas at or close to the water table, cuprite and native Cu form. Above the water table the stability ?elds of a range of secondary copper minerals (note only a few are shown) occur and these are controlled essentially by pH. This is because it is essentially the pH that controls the solubilities of complex- ing ligands (i.e. CO 3 2- , OH - , SO 4 2- , PO 4 2- , etc.) and determines which of the relevant copper complexes are likely to be saturated in any given environment. The extent to which leached and supergene zones develop, and the nature of the secondary minerals that form, depends to a cer- tain extent on the amount of primary, hypogene sul?des in the host rock and the acidity generated in the groundwaters as a consequence of their oxidation/weathering. Weathering of a host rock with a low sul?de content will typically result in minimal acidity, limited mobility of Fe and other metals, and formation of secondary copper min- erals that are stable in the near neutral to slightly acidic pH range (Chávez, 2000). Conversely, weathering of protores with a high sul?de mineral content will result in more extreme acidity and the formation of secondary copper minerals stable only at lower pH. Local conditions also play an important role, however, and supergene enrich- ment in the proximity of a limestone, for ex- ample, will result in local groundwaters with a high CO 3 2- content, resulting in the stabilization of minerals such as malachite or azurite under neutral to alkaline conditions (Figure 4.15 and equation [4.16] below). 2Cu 2+ + CO 3 2- + 2OH - ? Cu 2 (OH) 2 CO 3 [4.16] malachite Probably the most spectacular examples of super- gene enrichment are associated with the giant porphyry copper deposits of northern Chile, such as Chuquicamata, El Salvador, and El Abra. In addition to world-class hypogene orebodies at depth, these deposits have been subjected to a complex Paleogene to Neogene geomorphological evolu- tion that resulted in very signi?cant supergene ITOC04 09/03/2009 14:35 Page 241enrichment (Sillitoe and McKee, 1996). The super- gene and “exotic” ores at Chuquicamata and El Salvador are described in more detail in Box 4.2. A note on supergene enrichment of other metals Zones of supergene enrichment are also found in the sur?cial environment above any exposed metal orebody. In porphyry systems, molybde- num is normally comprehensively removed from regolith pro?les, although it can accumulate under very oxidizing conditions as ferrimolyb- dite, and under alkaline conditions as powellite (Ca–molybdate). Relative to Cu, the other base metals do not form as commonly as secondary mineral enrichments. Zn is often dispersed in groundwaters, whereas Pb tends to be relatively immobile. These two metals do, however, form a range of sulfate and carbonate supergene minerals, such as anglesite (PbSO 4 ), cerrusite (PbCO 3 ), and smithsonite (ZnCO 3 ), under certain conditions With new technologies available in extractive metallurgy, non-sul?de zinc ores, especially those comprising willemite (Zn 2 SiO 4 ), now represent an important category of mineralization associ- ated with oxidation of stratiform, sediment- hosted base metal deposits. Examples include the Skorpion Zn mine in Namibia and Vazante in Brazil. Silver tends to behave in much the same way as copper, and acanthite (Ag 2 S) is readily oxid- ized in acidic groundwaters, with Ag being rep- recipitated as the native metal, or as Ag–halides 242 PART 3 SEDIMENTARY/SURFICIAL PROCESSES –1.0 Eh (volts) –0.5 12 0.0 pH 4 6 8 10 0.5 1.0 2 0 14 25°C 1 atmosphere Chalcanthite 10 –1 10 –40 10 –55 10 –83 pO 2 (atmospheres) Tenorite (CuO) Antlerite Brochantite Malachite Cu 2+ CuSO 4 5H 2 O Cu 3 SO 4 (OH) 4 Cu 2 (OH) 2 CO 3 Cu 4 SO 4 (OH) 6 Top of water table Native Cu Bornite (Cu 5 Fe S 4 ), Chalcopyrite (CuFeS 2 ) Covellite (CuS) Chalcocite (Cu 2 S) Cuprite (Cu 2 O) H 2 H 2 O H 2 O O 2 Figure 4.15 Eh–pH diagram showing the stability ?elds of selected copper minerals at 25 °C and 1 atmosphere (after Guilbert and Park, 1986). ITOC04 09/03/2009 14:35 Page 242SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 243 Porphyry coppers are high tonnage–low grade deposits. The average grade of primary sul?de, or hypogene, mineralization is typically so low that they are only mar- ginally economic. Fortunately, many porphyry copper deposits are capped by a blanket of enriched ore formed by supergene processes. The supergene blanket, usually accessible during the early stages of mining, contributes signi?cantly to the overall viability of the mining opera- tion. At Chuquicamata, for example, which is the largest copper deposit in the world, with a total reserve of some 11.4 billion tons of ore at 0.76 wt% copper, the supergene blanket comprised a major proportion of the ore body that has now largely been mined out. The supergene blanket is made up of a barren leached zone, an upper copper “oxide” zone of antlerite, brochantite, atacamite, and chrysocolla, and an underlying copper sul?de zone made up mainly of chalcocite (Ossandón et al., 2001). In addition to the in situ supergene ores, some of the giant porphyry copper deposits of northern Chile, such as Chuquicamata, El Salvador, and El Abra, also contain “exotic” copper oxide mineralization. This is secondary ore that has been transported laterally from the leached portion of the supergene blanket and precipitated in Supergene processes: supergene and “exotic” mineralization in the porphyry copper giants of northern Chile Chuquicamata 3600N section 500 0 m Leached Supergene (oxides) Quartz-sericite alteration (phyllic) West fault Potassic alteration Supergene (sul?des) Leached Pit outline 1996 f f Original topography Chloritic alteration (propylitic) (a) Figure 1 (a) Cross section through the Chuquicamata mine showing the distribution of the supergene blanket (represented by the leached cap, the copper “oxide” zone, and supergene chalcocite ore) in relation to the primary hypogene ore (after Ossandón et al., 2001). the drainage network surrounding the deposits. Again at Chuquicamata, exotic copper oxide mineralization is located in a gravel ?lled paleochannel, extending from the main pit to the South or Mina Sur deposit, a distance of about 7 km. Some 300 million tons of exotic ore was deposited in these gravels as chrysocolla and copper wad, a Cu-rich, K-bearing Mn oxyhydrate (Mote et al., 2001). Similar exotic copper mineralization occurs at El Salvador (the Damiana and Quebrada Turquesa deposits) and also at El Abra. Supergene and exotic mineralization is related to the formation of acidic groundwaters that take up Cu into solution and reprecipitate it elsewhere. Secondary copper is either reprecipitated in the chalcocite rich supergene blanket beneath the leached zone (Figure 1a), or trans- ported laterally by groundwaters moving through paleo- drainage channels to form the distal exotic ore bodies of chrysocolla and copper wad (Figure 1b). The high degree of preservation of secondary copper ores in northern Chile is related to episodic tectonic uplift in combination with the pattern of global climatic rainfall and water- table ?uctuations (Mote et al., 2001). Accurate dating of copper wad and alteration assemblages in supergene ITOC04 09/03/2009 14:35 Page 243Figure 2 Exotic Cu-Mn oxide mineralization (wad) in gravels exposed in a bench cut through the Damiana ore body, El Salvador (photo courtesy of George Brimhall). 244 PART 3 SEDIMENTARY/SURFICIAL PROCESSES mineralized zones of northern Chile indicates that sur?cial processes were long-lived and episodic. Some supergene mineralization is preserved from just a few million years after the hypogene ores formed at the Eocene–Oligocene boundary (around 35 Ma). The main pulses of supergene ore formation, however, occurred toward the end of the Oligocene (20–25 Ma) and in the mid-Miocene (12–15 Ma). These pulses coincided broadly with periods of relatively high rainfall which pro- moted weathering, leaching of copper, and groundwater ?ow, as well as the deposition of gravels in proximal drainage channels. The subsequent preservation of both supergene and exotic styles of mineralization is due to the drop in erosion rates (i.e. reduction in uplift) and, more importantly, the onset of hyperaridity and limited ground- water ?ow, as the Atacama desert formed from the mid- Miocene onwards (Alpers and Brimhall, 1989). 5 0 km S26°15' Exotic deposits in gravels Gravels W69°37' Rio Seco Quebrada Turquesa El Salvador Townsite Drainage courses El Salvador District Damiana N Open Pit (b) Figure 1 (cont’d) (b) Simpli?ed map of the El Salvador district showing the distribution of gravel-?lled paleochannels and exotic deposits extending radially away from the main pit (after Mote et al., 2001). ITOC04 09/03/2009 14:35 Page 244SURFICIAL AND SUPERGENE ORE-FORMING PROCESSES CHAPTER 4 245 (AgCl, AgBr, AgI) in regions characterized by arid climates. In the Atacama desert of Chile there have been spectacular concentrations of super- gene silver ores discovered, including a 20 ton The chemical processes that contribute to weath- ering include hydration and dissolution, hydro- lysis and acid hydrolysis, oxidation and cation exchange. Sur?cial and supergene ore-forming processes are related essentially to pedogenesis, which can be simpli?ed into an upper zone of eluviation (leaching of labile constituents and residual concentration of immobile elements) and an underlying zone of illuviation (precipitation of labile constituents from above). Laterites are the product of intense weathering in humid, warm intertropical regions and are important hosts to bauxite ores, as well as concentrations of metals such as Ni, Au, and the PGE. Residual concentra- tions of alumina, and the formation of bauxitic ores, occur in high rainfall areas where Eh and pH are such that both Si and Fe in laterites are more soluble than Al. Ni enrichments occur above ultrama?c intrusions in the illuviated lat- erite zone where the metal is concentrated in phyllosilicate minerals by cation exchange. Au and Pt enrichments also occur in laterites above previously mineralized terranes, forming in the presence of highly oxidized, acidic, and saline Bland, W. and Rolls, D. (1998) Weathering: An Intro- duction to the Scienti?c Principles. London: Arnold, 271 pp. Guilbert, J.M. and Park, C.F. (1986) The Geology of Ore Deposits. New York: W.H. Freeman and Co., 985 pp. (Chapter 17). Leeder, M. (1999) Sedimentology and Sedimentary Basins: From Turbulence to Tectonics. Oxford: Black- well Science, 592 pp. (Chapter 2). Martini, I.P. and Chesworth, W. (1992) Weathering, Soils and Paleosols. Developments in Earth Surface Processes 2. New York: Elsevier, 618 pp. Williams, P.A. (1990) Oxide Zone Geochemistry. New York: Ellis Horwood, 286 pp. Wilson, R.C.L. (1983) Residual Deposits: Surface Related Weathering Processes and Materials. London: The Geological Society of London and Blackwell Scienti?c Publications, 258 pp. aggregate of embolite (Ag(Cl,Br)) and native silver (Guilbert and Park, 1986). The concentration of gold, with speci?c reference to lateritic regolith pro?les, is discussed in section 4.3.4 above. groundwaters. Fixation of the precious metals occurs by reduction or adsorption, in the presence of carbonaceous matter or Fe and Mn oxyhydrox- ides. Oxidized, acidic groundwaters are capable of leaching metals, not only during laterite forma- tion, but in any environment where such ?uids are present. Supergene enrichment of Cu can be very important in the sur?cial environment above hypogene porphyry styles of mineralization. Enrichment of copper again occurs in the illuvi- ated zone, although in this case both Cu–sul?de minerals (in the relatively reduced zone beneath the water table) and Cu–oxide minerals (above the water table) form. The formation of viable clay deposits, such as kaolinite, is a product of progressive acid hydro- lysis, essentially of plagioclase feldspar. Calcrete is a pedogenic product of high evaporation envir- onments, but can also form in paleo-drainage channels in now arid climatic regions. In uranium- fertile drainage systems calcrete channels rep- resent zones where concentrations of secondary uranium minerals precipitate together with calcite by evaporation-induced processes. ITOC04 09/03/2009 14:35 Page 245246 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 5.1 INTRODUCTION Sedimentary rocks host a signi?cant proportion of the global inventory of mineral deposits and also contain the world’s fossil fuel resources. Previous chapters have considered a variety of ore-forming processes that have resulted in the epigenetic con- centration of metals and minerals in sedimentary rocks. This chapter will concentrate on processes that are syngenetic with respect to the host sediments and where the ores are themselves sediments or part of the sedimentary sequence. Processes to be discussed include the accumula- tion of heavy, detrital minerals and the formation of placer deposits, the deposition of organic-rich black shales, and the precipitation mechanisms that give rise to Fe, Mn, and P concentrations in chemical sediments. In addition, there is a brief overview of the origins of oil and gas deposits, as well as of coali?cation processes, because the organic source material for these deposits re?ects the local depositional environment and ore forma- tion was broadly coeval with sediment accu- mulation. Although it might be argued that the migration of oil through sediments involves later ?uid ?ow through the depository, such processes are generally early diagenetic in character and therefore warrant inclusion in the present chapter. By contrast, syngenetic, clastic sediment- hosted SEDEX Pb–Zn–Ag, sandstone-hosted U ores, and late diagenetic red-bed Cu–(Ag–Co) type deposits are discussed in Chapter 3, since they involve processes that are more hydrothermal in character, or have their metals originating from outside the sedimentary host rocks. These Sedimentary ore-forming processes Box 5.1 Placer processes: the alluvial diamond deposits of the Orange River, southern Africa Box 5.2 Chemical sedimentation: banded iron- formations: the Mount Whaleback iron ore deposit, Hamersley Province, Western Australia Box 5.3 Fossil fuels: oil and gas; the Arabian (Persian) Gulf, Middle East SEDIMENTARY BASINS AND THEIR TECTONIC SETTINGS CLASTIC SEDIMENTATION AND HEAVY MINERAL CONCENTRATIONS sorting mechanisms relevant to placer formation application of sorting mechanisms to placer deposits sediment sorting in beach and aeolian environments CHEMICAL SEDIMENTATION AND ORE FORMATION ironstones and banded iron-formations bedded manganese deposits phosphorites black shales ocean ?oor manganese nodules evaporites FORMATION OF FOSSIL FUELS – A BRIEF OVERVIEW oil and gas formation coali?cation oil shales and tar sands gas hydrates ITOC05 09/03/2009 14:34 Page 246subdivisions are somewhat arbitrary and re?ect more the organizational structure of this book than a rigorous genetic classi?cation. Modern trends in the application of basin ana- lysis techniques to exploration have undoubtedly been set in the oil industry, where, in recent decades, major advances have been made in understanding the relationships between sedi- mentology, tectonics, and the maturation of organic material. The evolution of oil and gas within this cycle can be effectively assessed in terms of organic chemistry and ?uid ?ow patterns in the rock (Eidel, 1991). This level of under- standing requires a high degree of integration of numerous earth science disciplines (including paleontology, stratigraphy, sedimentology, struc- tural geology, plate tectonics, geohydrology, organic geochemistry, and others), an approach that is increasingly being applied to all types of explora- tion in sedimentary basins (Force et al., 1991). As with the fossil fuels, ore-forming processes that give rise to placer deposits and metal enrich- ments associated with chemical sediments and diagenetic ?uid ?ow are intimately related to the origin, tectonic setting, and evolution of the sedi- mentary host rocks. Most major syn-sedimentary ore deposits tend to occur in a limited range of basin types (Eidel, 1991). Passive continental margin chemical sedimentary ores (banded iron- formations and ironstones, bedded Mn deposits, and phosphate ores), as well as shoreline and ?uvial placer deposits (gold, cassiterite, diamonds, and zircon–ilmenite–rutile black sands), re?ect ore-forming processes that prevail in cratonic set- tings, where features such as protracted stability and high depositional energies for reworking of sediment load apply. Sediment-hosted ores in which late diagenetic or epigenetic processes are responsible for metal accumulation (such as red-bed Cu–(Ag–Co) deposits and some SEDEX Pb–Zn–Ag ores; see Chapter 3) tend to be associ- ated with rift basins where rates of deposition are high and active faulting promotes the circulation of connate and meteoric ?uids. It is also notewor- thy that many oil and gas ?elds are also associated with rift-related cratonic basins. The relationships between sedimentation and ore formation are multifaceted and only selected topics are discussed in this chapter. Additional information is available from the works listed at the end of this chapter. 5.2 CLASTIC SEDIMENTATION AND HEAVY MINERAL CONCENTRATION – PLACER DEPOSITS A placer deposit is one in which dense (or “heavy”) detrital minerals are concentrated dur- ing sediment deposition. They are an important class of deposit type and can contain a wide variety of minerals and metals, including gold, uraninite, diamond, cassiterite, ilmenite, rutile, and zircon. Well known examples of placer deposits include the late Archean Witwatersrand and Huronian basins (Au and U) in South Africa and Canada respectively, although the origin of these ores is controversial. Geologically more recent occurrences include the diamond placers of the Orange River system (see Box 5.1) and the western coastline of southern Africa, the cassiterite (Sn) placers of the west coast of peninsula Malaysia, and the beach-related “black sand” placers (Ti, Zr, Th) of Western Australia and New South Wales, South Africa, Florida, and India. The ?uid dynamic processes involved in placer formation are invariably very complex. It is exceedingly dif?cult to predict, for either explora- tion or mining purposes, where heavy mineral concentrations occur. This section will only touch brie?y on what has become a highly math- ematical subject that has applicability to geomor- phology and ?ood control engineering, in addition to ore deposit geology. More detailed reviews of the principles involved in clastic sedimentation and placer formation can be found in Miall (1978), Slingerland (1984), Slingerland and Smith (1986), Force (1991), and Pye (1994). 5.2.1 Basic principles The formation of placer deposits is essentially a process of sorting light from heavy minerals during sedimentation. In nature heavy mineral concentration occurs at a variety of scales, rang- ing from regional systems (alluvial fans, beaches, etc.), through intermediate features (the inner SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 247 ITOC05 09/03/2009 14:34 Page 247bank of a river bend or a point bar), to small-scale features (bedding laminae or cross-bed foresets). An experimental simulation of placer processes is illustrated in Figure 5.1, where a heavy mineral is fed via a “tributary” into the “trunk river.” The con?uence of two orthogonal ?ow streams results in the creation of a vortex or eddy downstream of the junction, with the accompanying erosion of the bed removing light particles and heavy min- erals concentrated in the scoured zone. The transportation and deposition of sediment in ?uvial and related systems is a complex topic (see overviews in Selley, 1988; Friedman et al., 1992; Pye, 1994; Allen, 1997; Leeder, 1999), the basic principles of which are essential to the understanding of placer processes. To complicate matters, sediments can also be sorted by the action of wind, and this process has applicability to the formation of placers in, for example, beach environments (Kocurek, 1996). One of the parameters used to quantify the con- ditions of ?uid motion is the Reynolds number, which is a dimensionless ratio identifying ?uid ?ow as either laminar and stable, or turbulent and unstable. The Reynolds equation is expressed as: Re = UL? f /? [5.1] where Re is the dimensionless Reynolds number; U is the ?uid velocity; L is the length over which the ?uid is ?owing; ? f is the ?uid density; and ? is the ?uid (molecular) viscosity. For low Reynolds numbers ?ow is laminar and vice versa for turbulent ?ow (Figure 5.2a). The behavior of a particle in a river channel will obvi- ously depend to a large extent on the type of ?ow it is being subjected to. Flow in natural stream channels is predominantly turbulent, the detailed anatomy of which is shown schematically in Figure 5.2b. Three layers of ?ow can be identi?ed. The bottom zone is the non-turbulent viscous sublayer, which is very thin and and may break down altogether in cases where the channel ?oor is very rough and turbulence is generated by the upward protrusion of clasts from the bed load. Above it is the turbulence generation sublayer, where shear stresses are high and eddies are gener- ated. The remainder of the stream pro?le is the outer or core layer, which has the highest ?ow velocities. A shear stress (? o in Figure 5.2b) is imposed on the bed load by the moving ?uid and is a function of ?uid density, the slope of the stream bed, and ?ow depth. The curved velocity pro?le of the channel section is caused by fric- tional drag of the ?uid against the bed. The types of ?uid ?ow in water (or air) de?ne the character and ef?ciency of mass (sediment) transport. A particle or grain will move through a ?uid as a function of its size, shape, and density, as well as the velocity and viscosity of the ?uid itself. In water, a particle at any instant will move in one of three ways: the heaviest particles (boulders, gravel) roll or slide along the channel ?oor to form the bedload (or traction carpet); intermediate sized particles (sand) effectively bounce along with the current (a process known as saltation); while the ?nest or lightest material (silt and clay) will be carried in suspension by the current (Figure 5.3a). In air the types of movement are similar, but the lower density and viscosity of air relative to water dictate that moving particles are smaller, but their motion is more vigorous (Figure 5.3b). The type of particles moving in a ?uvial channel by saltation and suspension is partly a 248 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Figure 5.1 Results of a ?ume experiment simulating (a) ?uid ?ow at the con?uence of a tributary with its trunk river and (b) the resulting distribution of heavy minerals (after Best and Brayshaw, 1985). (a) (b) Vortex Scour ITOC05 09/03/2009 14:34 Page 248(b) (a) Laminar flow Turbulent flow Outer (core) region Turbulence-generation layer Viscous sublayer Mean velocity profile J ? 0 S Surface Channel Bed Figure 5.2 (a) Schematic illustration, using streamlines, of the nature of laminar and turbulent ?uid ?ow. (b) Internal structure of turbulent ?uid ?ow in a natural channel. ? o is the boundary shear stress imposed by the ?uid on its bed and is a function of J (the ?ow depth) and S (the bed slope), as well as ?uid density (after Slingerland and Smith, 1986). (a) Water (b) Air Sliding Rolling Saltation Suspension Saltation Suspension Descending saltation Creep Figure 5.3 Illustration of the different mechanisms of sediment transport in water (a) and air (b) (after Allen, 1994). ITOC05 09/03/2009 14:34 Page 249function of the nature of the ?uid ?ow (as de?ned by the Reynolds number), whereas bed load movement is determined by shear stress at the boundary layer (Figure 5.2b) and the character- istics of the particles themselves. Combining ?uid ?ow (hydrodynamic) and physical mass transport parameters provides a useful semi-quantitative indication of the processes of sedimentation, a technique ?rst presented diagrammatically by Hjulström (Figure 5.4; after Sundborg, 1956). In this diagram the conditions under which either erosion, transportation, or deposition will take place are shown as a function of ?ow velocity and grain size. Deposition occurs either as ?ow velo- city decreases or grain size increases (or both), and these parameters are very relevant to the forma- tion of placer deposits. 5.2.2 Hydraulic sorting mechanisms relevant to placer formation Slingerland and Smith (1986) divided the mech- anisms of sorting into four types, namely: • free or hindered settling of grains; • entrainment of grains from a granular bed load by ?owing water; • shearing of grains in a moving ?uidized bed; • differential transport of grains by ?owing water. Each of these is discussed in more detail below with respect to their roles in the formation of placer deposits. Settling The settling velocity of a perfectly spherical par- ticle in a low Reynolds number (non-turbulent) ?uid can be determined using Stokes’ Law, which is expressed as follows: V = gd 2 (? p -? f )/18? [5.2] where V is the particle settling velocity; g is the acceleration due to gravity; d is the particle diameter; ? p and ? f are the densities of particle and ?uid respectively; and ? is the ?uid (molecular) viscosity. The relationship indicates that particle settling velocities in the same ?uid medium are propor- tional to particle diameters (squared) and densit- ies. Figure 5.5 shows that, in terms of Stokes’ Law, different sized grains of quartz, pyrite, and gold may settle at the same velocity, a condition that is referred to as hydraulic (or settling) equi- valence. This suggests that particles of differing size and density could conceivably reside to- gether in the same sedimentary layer, to form a rock such as a conglomerate. Hydraulic equi- valence is sometimes mistakenly taken as an explanation for the concentration of small, heavy detrital minerals in a coarse sedimentary rock. In fact the type of settling equivalence predicted by Stokes’ Law is a poor approximation of the complex and dynamic sorting mechanisms taking place during actual placer formation and other processes are required to explain the situation in nature. 250 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 512 0.5 64 Current speed (cm s –1 ) 2 1 8 41 Particle diameter (mm) 16 0.25 0.06 0.004 0.01 4 16 32 64 128 256 Sand Gravel Silt EROSION TRANSPORTATION DEPOSITION Cohesive bed Non-cohesive bed Figure 5.4 A Hjulström diagram showing how sedimentary processes can be assessed in terms of hydrodynamic (?ow velocity) and physical (grain size) parameters. Critical conditions for deposition are shown, as well as those for erosion and transportation for two situations in which cohesive and non-cohesive channel beds apply (after Sundborg, 1956; Friedman et al., 1992). ITOC05 09/03/2009 14:34 Page 250SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 251 There are several reasons why a simple Stokesian settling is inadequate as an explanation for placer forming process. These include the fact that stream ?ow is generally turbulent (it has a high Reynolds number), particles are not spher- ical, and particle sizes may be either too big or too small for the relationship in equation [5.2] to be upheld. In addition, if the concentrations of grains in the ?uid is high (>5%) then settling is no longer unhindered and settling velocities are retarded by grain–grain collisions and current counter?ow. The random instability of turbulent ?ow makes it virtually impossible to predict particle settling velocities and there is no completely satisfactory model for simulating this condition (Slingerland and Smith, 1986). Likewise, particle shape has an important effect on settling velocity. A tabular biotite grain, for example, will settle between 4 and 12 times slower than a quartz grain of equi- valent diameter. Big grains have large coef?cients of drag in a ?uid and accordingly their settling velo- cities vary as a function, not of the square of the diameter (equation [5.2]), but of the square root (Figure 5.5). Smaller grains are, therefore, much more effectively sorted by settling than are big grains. Although the ratios of particle diameters (quartz:pyrite:gold approximately 32:2:1) re?ect- ing hydraulic equivalence in Figure 5.5 appear to be reasonably consistent with what one might expect in a gold-bearing conglomerate such as in the Witwatersrand Basin, the diameter of gold itself in these deposits is typically much too small to be explainable by Stokesian-type settling. For that component of Witwatersrand gold that is detrital, it is, therefore, likely that some other concentra- tion mechanisms, such as entrainment (see below) might have applied (Frimmel and Minter, 2002). Particle settling in nature is a dif?cult para- meter to quantify because of the wide range of variables likely to affect it. A more realistic expression of particle settling velocity, which takes into account the frictional drag that arises from different shapes, is provided by the following expression (after Slingerland and Smith, 1986): V = [4(? p – ? f )gd/3? f C d ] 1 / 2 [5.3] where V is the particle settling velocity; g is the acceleration due to gravity; d is the particle dia- meter; ? p and ? f are the densities of particle and ?uid respectively; and C d is the coef?cient of drag and is de?ned as 24/Reynolds number. Settling does play a role in the formation of placer deposits, but on its own is of little use in understanding the processes by which they form. The existence of a state of hydraulic equivalence Quartz Pyrite Gold d = 16 ? = 2 d = 1 ? = 5 d = 0.5 ? = 17 V P V Q V G == Water ? f = 1 Hydraulic equivalence Velocity ? d 2 . (? p – ? f ) (for gold) Velocity ? d 1/2 . (? p – ? f ) (for quartz and pyrite) Figure 5.5 Illustration showing the principle of hydraulic equivalence for particles settling according to Stokes’ Law. The settling velocities of quartz, pyrite, and gold, with radii and densities as shown, are the same, indicating that they would settle out of a non-turbulent column of water into the same sedimentary layer. ITOC05 09/03/2009 14:34 Page 251will not explain how heavy detrital minerals are sorted or concentrated in dynamic river or beach systems. Hydraulic equivalence describes a con- dition of equal settling velocities and accounts for unsorted accumulation of heavy minerals in coarser grained sediment (conglomerate). By contrast, if settling velocities are not equal and settling occurs in a system that is ?owing (i.e. a river), then heavy minerals will be segregated laterally downstream as a function of size and density (see section on transport sorting below). In any event it is the movement or ?ow of the ?uid medium that has the dominant role to play in placer formation. Entrainment Entrainment sorting refers to the ability of a ?uid in contact with bed load particles to dislocate certain grains from that bed and move them fur- ther downstream. As with settling, a considerable amount of effort has been spent by ?uid dynam- icists to quantify the criteria for entrainment in terms of variables such as the hydraulic/?ow conditions, particle size, shape and density. In the channel cross section shown in Figure 5.2b, the concept of a ?uid force (referred to as ? o , the boundary shear stress) acting on the bed load is demonstrated. It is clear that ? o must exceed the forces that keep any given particle in place (i.e. size, mass, shape, friction) before the particle can start to move. The critical shear stress required to initiate movement of any given particle is known as the Shields entrainment function, or simply the Shields parameter (?), a detailed description and derivation of which is provided in Slingerland and Smith (1986). The Shields parameter is ex- pressed as: ?=? c /[(? p -? f )gd] [5.4] where ? is the dimensionless Shields parameter; g is the acceleration due to gravity; d is the particle diameter; ? p and ? f are the densities of particle and ?uid respectively; and ? c is the critical boundary shear stress. In relatively simple situations the Shields cri- terion can be shown to simulate the movement of bed load particles in a channel reasonably well, and Figure 5.6a illustrates the critical conditions that differentiate between entrainment of a par- ticle into the channel and non-movement of the grain. For a uniform bed load, particle size is the dominant control and sand, for example, will have a lower entrainment threshold than gravel. Again, however, a uniform bed load seldom applies and there are many factors which complic- ate entrainment sorting. The bed load roughness, or, put another way, the extent to which a particle sits proud of the bed, will clearly have an effect on the entrainment threshold. In Figure 5.6a this variable is illustrated and quanti?ed by the ratio p/D. An increase in this value equates with grains which jut out and this can be seen to have the log- ical effect of decreasing the entrainment thresh- old. Other factors which complicate entrainment sorting include the clast shape (sphericity tends to make entrainment easier), bed consolidation, and the range of grain sizes in any given bed load. The latter point is particularly important in the situ- ation where sediment is made up of a bimodal assemblage of large and small grains. In this case small grains, even if they are relatively light, will not be entrained because they rapidly become entrapped within the much larger particles and are no longer available for entrainment. The effects of entrainment on sorting and placer accumulation are shown in Figure 5.6b. In this diagram (similar to the Shields diagram but where ? is replaced by ? c and Re* is replaced by grain diameter) the critical boundary shear stress for entrainment increases as a function of grain density. This indicates that, for a uniformly sized bed and with all other factors being equal, lighter particles will be effectively entrained at lower shear stresses and, therefore, winnowed away, leaving a residual concentration of heavier par- ticles. It is, therefore, quite feasible to entrain quartz from the bed load of a stream and leave behind a residual accumulation of heavy minerals that could be preserved as a placer deposit. Thin laminae of heavy minerals in a ?uvial setting, or along cross-bed foresets, are likely to be the result of this type of sorting process. Entrainment pro- cesses also apply to wind-blown sedimentation and sorting, but the magnitude of the critical 252 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 252SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 253 thresholds will be quite different because of the quite different mechanical conditions that per- tain to the aeolian environment. Shear sorting Shear sorting of grains is a process that only applies to the concentrated ?ow of suspended particles in a ?uidized bed. During the movement of suspended particles in a dense granular dis- persion, grain collisions create a net force that is perpendicular to the plane of shearing and disperses the granules toward the free surface (i.e. upwards). Counter-intuitively, the dispersive pressure is greater on large and dense grains within the same horizon of ?ow so that these grains migrate upwards relative to smaller and lighter particles. The same effect can be rational- ized in terms of a concept known as kinetic sieving, where smaller grains simply fall between larger ones (Slingerland, 1984). The effects of shear sorting have been quanti- ?ed for sediment populations of mixed sizes and densities by Sallenger (1979), who showed that 10 0 ? Re* 10 4 Non-movement 10 3 10 2 10 1 10 0 10 –1 10 –2 Entrainment 0.6 0.4 0.2 p/D = 0 D p (a) 10 3 ? c d (cm) 10 1 10 0 10 –1 10 –2 10 0 10 –1 (b) 10 –3 10 1 10 2 Gold (? = 19) Ilmenite (? = 5) Quartz (? = 2.5) Figure 5.6 (a) Shields diagram showing the threshold conditions between entrainment and non- movement of a grain for a bed load of uniform size in terms of the Shields parameter (de?ned in equation [5.6]) and the grain Reynolds number (Re* or the Reynolds number as it pertains to a single particle). The effects of increasing the protrusion of a particle above the bed ?oor (i.e. increasing the ratio p/D) will have the effect of shifting the entrainment threshold: the higher a particle protrudes above the bed, the easier it will be to entrain. (b) Diagram showing the effects of grain density (?) on the critical boundary shear stress (? c ). Greater shear stresses are required to entrain denser particles. Details after Slingerland and Smith (1986) and Reid and Frostick (1994). ITOC05 09/03/2009 14:34 Page 253two grains of different densities (?) coming to rest in the same horizon would have relative sizes (d) given by: d h = d l (? l /? h ) 1 / 2 [5.5] where d h and d l are the diameters of heavy and light fractions and ? h and ? l are the densities of heavy and light particles respectively. An illustration of shear sorting in a concen- trated granular dispersion comprising quartz and magnetite, in the proportion 90:10, is shown in Figure 5.7. The horizon at a relative height of between 0.75 and 0.5 units from the surface is seen to contain magnetite concentrations up to double the initial concentration. The mechanism is, therefore, one of the few processes that will explain heavy mineral concentrations in an elev- ated horizon of the sediment strata and could also explain inversely graded particle concentra- tions. Shear sorting can apply only to concen- trated suspended particle loads such as beach swash zones and wind-related dune formation. The process could be applicable to the concen- tration of Ti–Zr–Th black sand placers in these environments. Transport sorting Transport sorting is by far the most important sorting process and the one most applicable to the broadest range of environments in which placer deposits form. The quanti?cation of transport sorting is a complex process, but conceptually it refers simply to the differential transport rates that exist during movement of particles in a ?owing ?uid medium. It is complex because it incorporates two distinct components, namely the varying rates of movement of grains both in the bed load (determined by entrainment), and in suspension (determined by settling). The previous discussion of settling considered the concept only in terms of unhindered, non- turbulent ?ow and it was emphasized that it had limited use in understanding the dynamics of placer-forming processes. The concept of sus- pension sorting, however, is an extension to the rather simplistic considerations of the previous section and refers to the fractionation of grains with different settling velocities into different levels above the bed in a turbulent channel ?ow system. Subsequent to its deposition down- stream, sediment sorted in this fashion can result in substantial heavy mineral enrichments. It is discussed here, rather than earlier, because it is one of the two components that contribute to the determination of transport sorting. The concentrations of heavy (h) and light (l) particles that coexist at any point in the channel ?ow system can be quanti?ed in terms of an equa- tion derived in Slingerland (1984): (C/C a ) h = (C/C a ) l V h /V l [5.6] where C is the concentration at a given level in the channel ?ow; C a is a reference concentration; and V h and V l are the settling velocities of heavy and light particles respectively. 254 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 0.00 1.00 2.0 Relative height of granular bed 0.50 0.75 0.25 1.0 0.5 0.25 0.125 Grain size (mm) 1.0 0.125 0.00 1.00 0 0.50 0.75 0.25 10 20 30 Magnetite concentration (%) Fine quartz Coarse quartz Zone of magnetite enrichment Figure 5.7 Quartz grain size and magnetite concentrations in a shear sorted granular dispersion comprising quartz and magnetite in the proportion 90:10. The relative depth of the granular bed is on the ordinate; original magnetite concentration (10%) is shown as the vertical dashed line (after Slingerland, 1984). ITOC05 09/03/2009 14:34 Page 254SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 255 The above equation predicts that particles in a turbulent ?owing channel will be sorted vertic- ally according to their settling velocities which, in turn according to Stokes’ Law, are determined by their relative sizes and densities. The con- cept is pertinent to transport sorting because the effects of suspension sorting are only relevant downstream once deposition of a particular sedi- ment horizon, with its possible enrichment of a hydraulically equivalent mineral, has taken place. Note that the results of suspension sorting are contrary to those obtained by shear sorting in a concentrated ?uidized bed. The other component of transport sorting applies to the movement of bed load and is dis- cussed previously in the section on entrainment. The entrainment threshold of a particle sitting on the bed is determined by the Shields parameter (equation [5.4]) but is also strongly in?uenced by bed roughness. Consequently sediment transport rates would be expected to decrease as bed rough- ness increases. Slingerland (1984) has computer modeled transport sorting processes using an ini- tial sediment comprising quartz and magnetite, in the proportion 90:10, and mean grain diameters of about 0.4 and 0.2 mm respectively. The results con?rm that for a given shear velocity (another measure of the boundary shear stress, as shown in Figure 5.2b) all sizes of quartz and magnetite exhibit a decrease in transport rate as bed rough- ness increases. In addition, and as shown in Figure 5.8a, the proportion of magnetite in the moving sediment load increases as a direct function of shear velocity for a given bed roughness (trend a–a') but decreases as a function of roughness for a given shear velocity (trend b–b'). The effects of transport sorting have also been demonstrated experimentally in a ?ume, where sediment transport is simulated using simpli?ed, scaled down parameters. Figure 5.8b shows that a positive relationship exists between transport velocity and shear velocity for any particle size and bed roughness. However, for any measure of bed roughness (K) larger grains move faster than smaller ones, a feature that is explained by the fact that smaller grains progress less easily over a rough bed because of trapping and shielding. Thus, for any given shear velocity and bed rough- ness, the largest grains have the fastest transport rates (trend c–c'–c¨ in Figure 5.8b). Conversely, the slowest transport rates for a given shear velo- city are a feature of the smallest particles (i.e. 0.21 mm in this particular experiment) moving over the roughest bed (K = 0.78 in Figure 5.8b). Clearly, transport sorting is a complex process which can have quite different outcomes from one part of a channel ?ow system to another. A bed of sediment that is being deposited at any given moment may undergo enrichment in heavy minerals if the combination of shear velocity and the sizes of heavy and light particles relative to bed roughness mitigate against entrainment, thereby resulting in the formation of a lag deposit. If, on the other hand, the same particles are sub- sequently entrained into the sediment load under a different ?ow regime, they might then be sub- jected to suspension sorting which would result in heavy mineral enrichment at a completely different downstream location. 5.2.3 Application of sorting principles to placer deposits Enrichment of heavy minerals by grain-sorting mechanisms occurs on all scales in nature, from single grain laminae on cross-bed foresets, to large regional scale concentrations that have accumu- lated in a particular sedimentary environment such as an alluvial fan or beach system. It is tempting to suggest that small scale systems might be more easily explained in terms of only one of the mechanisms discussed above, and that regional systems were more complex, involving several mechanisms acting in concert. This ration- ale does not hold, however, and even small scale systems can be a product of complex interactions; the following examples demonstrate the applica- tion of sorting mechanisms to different scales of deposition. Small scale As an example of heavy mineral concentration at a small scale, the mechanisms of grain sorting in and around a dune or ripple migrating along the bed of a stream are shown in Figure 5.9a. The dune ITOC05 09/03/2009 14:34 Page 255crest is characterized by high shear velocities and non-turbulent ?ow which increases the Shields factor and promotes entrainment of larger, lighter grains and residual concentration of the heavier particles. The dune foreset, on the other hand, is likely to receive heavy mineral concentrations by shear sorting of high concentration grain avalanches down the slope of the advancing dune. The trough or scour forming ahead of the dune is likely to receive heavy grain concentrations by settling sorting in a locally turbulent micro- environment. It is, therefore, evident that several 256 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 50 12 Vg (cm s –1 ) 10 20 468 U* (cm s –1 ) 21 0 10 0 Roughness (mm) 4 2 8 10 20 60 50 U* (cm s –1 ) 70 14 40 30 K = 0.21 K = 0.39 K = 0.78 Grain diameter 0.78 mm 0.39 mm 0.21 mm K = Bed roughness, mm c'' c' c 40 30 6 (b) (a) 1 2 3 4 5 6 7 aa ’ b’ b Magnetite concentration (%) in sediment load Approximate limit of magnetite transport Lag deposits Figure 5.8 (a) Computer simulation showing the effects of transport sorting in terms of bed roughness (measured by the height of clasts protruding from the bed in millimeters) and shear velocity U*. The plot shows the approximate limit for transport of magnetite and the magnetite concentrations (%) in the sediment load. Maximum enrichment of magnetite on the bed (i.e. the formation of a lag deposit) would occur above this limit under conditions of low shear velocity and high bed roughness (after Slingerland, 1984). (b) Results of a ?ume experiment showing the effects on grain transport velocities (V g ) of shear velocity (U*), bed roughness, and grain diameter (after Meland and Norman, 1966). ITOC05 09/03/2009 14:34 Page 256High Au/U ratio Low Au/U ratio (a) 0 Small scale 20 cm Stream flow c Settling sorting (settling sorting of suspended particles in turbulent zone) c Bed a b Shear sorting (concentrated granular avalanches down slope) b a Entrainment sorting (high shear velocity and laminar flow) (b) Intermediate scale A meander Channel Point bar Divergent flow A’ A Convergent flow Older alluvium 03 0 meters A’ (c) Large scale Braided channels Alluvial fan Basal – Steyn entry front N Welkom gold field 0 km 2 Figure 5.9 Heavy mineral sorting processes at various scales and in different sedimentary environments: (a) at a small scale along laminae associated with dunes and ripples forming along a stream bed (after Slingerland, 1984); (b) at an intermediate scale in an aggrading point bar forming along the convex bank of a meander channel (after Smith and Beukes, 1983); and (c) at a large scale on a large braided, alluvial fan complex such as the Welkom gold?eld, Witwatersrand Basin (after Minter, 1978). ITOC05 09/03/2009 14:34 Page 257sorting mechanisms apply even to the small scale heavy mineral concentrations that one often sees in and around dune and ripple features in river- or beach-related sediments. Intermediate scale An example of heavy mineral concentration at an intermediate scale is provided by the develop- ment of point bars along the convex bank of a meandering river channel. Such sites are well known to miners dredging river channels for accumulations of minerals such as cassiterite or gold. Meandering river channels are also prime targets during the exploitation of placer diamonds along the Orange River in southern Africa (see Box 5.1). Figure 5.9b shows the geometry and ?ow patterns associated with the aggradation of a point bar in a meander channel and its migration toward the opposite concave bank, which is being subjected to erosion. Heavy mineral concentra- tions actually form in degraded scours along the bottom of the channel itself. Both components of transport sorting seem to contribute to the enrichment process, namely settling of grains in the highly turbulent environment formed by con- vergence of disparate ?ow orientations, and entrainment, since shear stresses are above the thresholds for most of the bed load. Transport sorting, therefore, results in placer accumulations which are then amenable to preservation as the point bar sediment migrates over the accumu- lated heavy minerals. This process has been applied to the concentration of gold and uraninite in point bars formed on the ?uvial fan deltas of the Witwatersrand basin (Smith and Minter, 1980). Large scale Large scale sorting of heavy minerals over areas of tens of square kilometers has been described from deltaic fan conglomerates of the Welkom gold?eld in the Witwatersrand basin, South Africa. Figure 5.9c shows the relative distribu- tions of gold and uraninite (as the Au/U ratio) in the composite Basal–Steyn placer with respect to the sediment entry point and major ?uvial braid channels on the fan delta complex. Close to the entry point channel ?ll and longitudinal bars comprise coarse pebble conglomerate, with max- imum clast sizes in the range 20–40 mm in dia- meter. By contrast, the contained gold particles are much ?ner and range between about 0.5 and 0.005 mm in size, although accurate size distribu- tion patterns are not known. Several kilometers downstream the fan delta is built of less deeply channeled quartz–arenite. The Au/ U ratio decreases downstream and this is a function of both a decrease in gold and an increase in uraninite down the paleoslope (Minter, 1978). This has been inter- preted to re?ect transport sorting which results in net deposition of gold in more proximal locations of the fan, with the less dense uraninite being more effectively concentrated in distal parts. Transport sorting ensures that clast sizes (and also, therefore, bed roughness), as well as shear velocity, decrease exponentially down slope. Uraninite is signi?cantly larger, and also lighter, than gold and will exhibit higher grain transport velocities, and, therefore, be carried a greater dis- tance down slope. In addition, gold will be trans- ported less easily over the proximal, high bed roughness portions of the fan delta complex and will, therefore, be trapped more effectively in these areas compared to uraninite. Many of the laterally extensive Witwatersrand placers rep- resent regional, low dip angle, unconformity sur- faces where the fan delta complexes are being degraded very slowly and the ?uid regime is con- sequently very consistent over large areas. This type of setting represents an excellent site for the formation of very large placer deposits because signi?cant volumes of sediment can be subjected to a consistent set of transport sorting mechan- isms (Slingerland and Smith, 1986). 5.2.4 A note concerning sediment sorting in beach and eolian environments Much of the previous discussion applies to pro- cesses pertaining to unidirectional water ?ow and is, therefore, mainly relevant to ?uvial environ- ments. As mentioned previously, however, many important placer deposits are associated with sediments deposited in shoreline environments 258 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 258SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 259 Diamond is the hardest known natural substance. High quality (i.e. based on clarity, colour and a lack of ?aws) diamonds are used for jewelry, whereas ?awed and poorly crystalline stones are used as abrasives in a wide range of cutting tools. The speci?c gravity (3.5) and hardness of diamond ensure that it is concentrated together with other heavy minerals in both ?uvial and marine placer deposits. Over much of recorded history India was the only country to produce diamonds from alluvial sources, but in the 1860s diamonds were discovered in South Africa, spe- ci?cally in gravels of the Orange River and its tributaries. This led to the discovery of the primary kimberlitic source of diamonds around the town of Kimberley, and then in 1908 of the huge beach placers along the west coast of southern Africa. The alluvial and beach diamond placers of South Africa and Namibia are the product of deep erosion of kimberlites on the Kalahari Craton in Cenozoic times. The erosion of diamondiferous kimberlite liberated the diamonds onto the land surface for subsequent redis- tribution into the very sizable catchment of the Orange River drainage system and its precursors. This drainage Placer processes: the alluvial diamond deposits of the Orange River, southern Africa Namibia 400 0 km Cape Town Port Elizabeth Diamondiferous kimberlite Indian Ocean Alluvial diamond deposits Beach-related diamond placers Atlantic Ocean Offshore diamond concessions Orange River River Orange Johannesburg Durban Lesotho Swazi- land South Africa Vaal River Hartz River Kimberley Botswana Zimbabwe Kalahari Craton Mozambique Kalahari Craton N Paleo-drainage channels Figure 1 Map showing the Orange River drainage system in relation to the distribution of deeply eroded kimberlites on the Kalahari Craton. The distribution of alluvial diamond deposits along major river channels and paleo-channels, as well as the beach placers along the west coast of southern Africa, are also shown (after Lynn et al., 1998). ITOC05 09/03/2009 14:34 Page 259260 PART 3 SEDIMENTARY/SURFICIAL PROCESSES IV N V V V V Orange River Gravel terraces Upper terrace Intermediate terraces Lower terrace III V I II I I I I I I V V V III III IV Exposed basement III Namibia 2000 0 meters 1000 I South Africa Tributary III Orange River Figure 2 Map showing the distribution of gravel terraces in the vicinity of the Bloeddrift diamond mine along the lower reaches of the Orange River (after Van Wyk and Pienaar, 1986). ?ows westwards and exits into the southern Atlantic at Alexander Bay–Oranjemund at the South Africa–Namibia border. Diamonds are trapped in gravel terraces that represent preserved sections of river sediment, abandoned by the present river as it migrates laterally or incises down- wards (Lynn et al., 1998). Diamonds are also washed out into the ocean and redistributed by long-shore currents to be concentrated in gravels either beneath the wave base, or in remnant beaches re?ecting previous ?uctuations of sea level. In the upper reaches of the Orange River system (which includes the Vaal River) diamonds are preferentially con- centrated in areas where the rivers ?ow over resistant bedrock, such as the Ventersdorp lavas, where good gully ITOC05 09/03/2009 14:34 Page 260SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 261 where sediment sorting is largely controlled by the dynamics of waves and by tidal ?uctuations. It is also possible in such environments that sorting processes could be in?uenced by wind action and that the latter might interact with water-borne sediment dispersal patterns. In fact, heavy mineral concentrations in some of the enormous beach-related “black sand” (Ti–Zr) placer deposits of the world, such as Richard’s Bay in South Africa, are a likely product of both beach- and wind-related sorting processes. This section brie?y considers the physical processes involved in such environments and emphasizes some of the differences that exist compared to the ?uvial system. Beaches The sorting of sediment in beach environments occurs at a number of scales, ranging from long- shore and orthogonal processes (related to mass transport parallel and at right angles to the shore line) to swash zone-related features attributed to the “to and fro” motion of waves breaking on the shore. A detailed overview of these various processes is presented in Hardisty (1994) and Allen (1997). Beach-related placer deposits appear to be mainly related to processes occurring in the swash zone and it is on this environment that the present discussion will focus. The origin and nature of ocean waves is a quanti?able topic of study and the effects of wave action on sediment bed forms are well under- stood. Beaches represent the interplay between sediment supply, wave energy, and shoreline gra- dient. Sediment transfer on beaches is in?uenced largely by tides and currents, whereas sediment sorting is dominated by waves and swash pro- cesses. Figure 5.10a illustrates the spectrum of wave-dominated shoreline types, emphasizing the differences between a low energy, shallow gra- dient beach comprising mud and silt, and a steep, highly energetic environment in which gravel beaches are deposited. An intermediate situation re?ects the setting in which sand-dominated beaches are likely to form. and pothole trap sites can form. Diamonds are concen- trated as residual lag deposits in such traps. Concentration is enhanced by the extreme hardness of diamond as other less resistant minerals undergo attrition and are more easily washed out of irregularities in the bedrock (Lynn et al., 1998). In the lower, more mature reaches of the Orange River, a number of gravel terraces occur. These contain large, low grade accumulations of diamonds and were deposited in lower Miocene to upper Pleistocene times, occuring at different elevations relative to the present day channel (Van Wyk and Pienaar, 1986). As many as ?ve terraces can be identi?ed, each characterized by diagnostic fossil faunal and ?oral assemblages. Diamonds derived from gravels of the lower reaches of the Orange River are 97% gem quality since it is the un?awed diamonds that typically resist the rigors of mechanical transport and abrasion. Most of the diamonds are between 0.85 and 1.30 carats in size, but occasionally very large stones (60–100 carats) are also recovered. The single most important feature in the concentration of diamonds in the terrace gravels is bedrock irregularity. The development of large potholes, plunge pools, and bedrock ribbing are all features that impinge on grade distribution in the gravels. The lowermost sections of the gravel pro?le are typically the highest grade. Bedrock Figure 3 Diamondiferous gravels related to the paleo-Orange River system, deposited unconformably on irregular, potholed bedrock (foreground). depressions can be extremely productive, with grades of over 40 carats per 100 tons of ore being reported (Van Wyk and Pienaar, 1986). Large terraces are generally higher grade and contain bigger stones in their upstream portions, attesting to ef?cient trapping and sorting mechanisms prior to preservation. ITOC05 09/03/2009 14:34 Page 261Although the type of beach forming is import- ant to whether a placer deposit is likely to develop, environmental considerations alone are unlikely to explain the dominant processes in beach placer formation. It is apparent from the limited amount of research that has been carried out in this ?eld that factors such as source fertility and swash processes are a more important consideration (Komar and Wang, 1984; Komar, 1989; Hughes et al., 2000). Source fertility is self-evident and dictates whether a placer deposit is likely to con- tain ilmenite–zircon or diamond concentrations, or none at all. By contrast, the ?ow dynamics of a swash zone are more dif?cult to evaluate. The application of these processes to the concen- tration of heavy minerals on a beach has been studied by Hughes et al. (2000). An illustration of modeled swash dynamics is presented in Figure 5.10b and shows that motion of the swash front is symmetric about the surge and backwash, but the water velocity is asymmetric because local ?ow reversal occurs before the front has reached its ?nal landward advance. The dynamics of the swash zone have been used to evaluate mineral- sorting mechanisms in the beach placers of south- east Australia and show that neither settling nor entrainment is likely to be important in the concentration of heavy minerals. Settling is dis- counted because swash zones are typically not deep enough to allow effective settling of heavy minerals to occur. Entrainment, likewise, was ineffective as a sorting mechanism because the large bed shear stresses prevailing imply that most, if not all, the minerals would be in motion with little or no prospect of selective entrainment (Hughes et al., 2000). The dominant process appears rather to have been shear sorting, where the dispersive (upward) pressures acting on large, dense grains moving in a concentrated sheet ?ow were larger than those applicable to smaller, less dense grains. In a study of the heavy mineral placers of the Oregon, USA, coastline, Komar and Wang (1984) were also able to discount grain settling as a sorting mechanism because the set- tling velocities of light and heavy mineral frac- tions were found to be similar. In contrast to the Australian situation, this study found that select- ive grain entrainment was likely to have been the 262 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 6 –6 –1 X s (m), U s and U (m s –1 ) –2 –4 4 012 7 Time (s) 45 36 2 0 X s -motion of the swash front U s -velocity of the swash front U-water velocity within swash (b) Seaward flow Landward flow Increasing abundance of sediment supply Increasing grain size, gradient of shoreline and nearshore wave energy Steep beach Gravel Wave break Reflection of wave energy Sand Wave break Dissipation of wave energy Shallow profile Dissipation of wave energy Trough Bar Absence of wave breaking (a) Mixed mud/silt Figure 5.10 (a) Factors controlling the nature of wave dominated beaches and shorelines (after Reading and Collinson, 1996). (b) Mathematical model illustrating the dynamics of the swash zone created for a 0.5 m high breaker at the shoreline (after Hughes et al., 2000). ITOC05 09/03/2009 14:34 Page 262SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 263 most ef?cient concentration process and the one that also accounted for lateral variations observed in the distribution of heavy minerals. Figure 5.11a shows the variable distribution of heavy minerals as a function of distance offshore, a local char- acteristic that demands a sorting process capable of laterally sorting the mineral components. Figure 5.11b illustrates the calculated threshold stresses for the various minerals examined in the Oregon beach placers as a function of the meas- ured concentration factors and shows that the densest and smallest grains were the most effect- ively concentrated. This may be due largely to the fact that these grains were the most dif?cult to entrain and, consequently, were residually concentrated in the near shore environment. The fact that entrainment sorting seems to apply to the Oregon placers, but not those formed along the Australian coastline, is possibly a function of the higher wave energy in the latter area and the development of conditions favoring shear sorting over entrainment (Hughes et al., 2000). Although the mechanics of the sorting processes are sim- ilar, the detailed patterns of heavy mineral distri- bution on beaches may be quite different to those in ?uvial systems where transport sorting is the dominant process. Wind-borne sediment transport As illustrated in Figure 5.3, the transport of sediment by wind, although similar in principle to that by water, differs in detail because of the much lower viscosity and density of air and the generally higher kinetic energies of eolian trans- port. Grains transported by wind are subjected to different entrainment criteria than those moved subaqueously, and are also subjected to more dramatic ballistic effects due to their energetic motion. In equation [5.4], the critical shear stress required to initiate entrainment of a given par- ticle is seen to vary linearly as a function of its density and size. Bagnold (1941) showed that for eolian sediment transport the critical shear stress required to initiate entrainment varied as a function of the square root of density and size. This implies that it is easier to move small par- ticles by wind than it is to move them by water, which explains the abundance of widespread, coarse grained lag deposits in desert and beach environments. Such deposits form by the de?ation of a surface initially comprising multiple sized particles, by winnowing away the ?ne grained material and leaving behind the gravel residue. Wind-related sediment transport characteristics 50 0 0 % heavy minerals in sediment 30 20 40 10 20 60 Offshore distance (m) 40 50 30 10 (a) 0.1 1 Concentration factor 100 10 1000 23 7 Threshold stress, ? (dynes cm –2 ) 56 4 1 (b) Augite Hornblende d = 0.21 mm ? = 3 g cm –3 d = 0.15 mm ? = 4.7 g cm –3 Garnet Ilmenite Concentrated in placer Moves offshore Zircon Ilmenite Garnet Hypersthene Augite Epidote Hornblende Increasingly difficult to entrain Quartz Figure 5.11 (a) Plot showing the changes in heavy mineral contents along a pro?le of beach sediment in Oregon as a function of distance offshore. (b) Plot of concentration factors of the principal minerals in an Oregon beach placer deposit as a function of calculated threshold stress, illustrating the likelihood of selective entrainment processes during the formation of the deposit (after Komar and Wang, 1984). ITOC05 09/03/2009 14:34 Page 263are shown in Figure 5.12 and can be compared with the Hjulström diagram of Figure 5.4 which pertains to subaqueous sediment transport. The extent to which beach-related placer deposits are modi?ed by wind-blown sediment dispersal is a question that has not received much attention, despite the fact that such interactions are almost certain to have taken place (Chan and Kocurek, 1988). At the Richard’s Bay deposits, for example, a substantial proportion of the Ti–Zr heavy minerals are extracted from eolian dunes even though much of the original concentration took place in the swash zone of the adjacent beaches. 5.2.5 Numerical simulation of placer processes Sophisticated computer simulations of complex natural placer processes have now been developed with a view to improving the prediction of heavy mineral distribution patterns in placer deposits. An example is the MIDAS (Model Investigating Density And Size sorting) program that predicts the transport and sorting of heterogeneous size– density graded sediment under natural conditions (Van Niekerk et al., 1992; Vogel et al., 1992). It has been applied to the problem of grade control in the Witwatersrand basin and can predict gold distribution patterns in the host conglomerates with remarkable accuracy (Nami and James, 1987). The conglomerates that host concentrations of detrital gold in the Witwatersrand basin were deposited in braided alluvial fan systems, the main elements of which comprise channel and overbank ?ow deposits. The geometry of this environment is shown in Figure 5.13a, which illustrates that ?uid ?ow can be resolved into two components, a higher velocity ?ow contained within the channel, and a lower velocity ?ow in the overbank plain that is transverse to that in the channel. Hydrodynamic considerations suggest that the ?ne grained detrital gold particles in the Witwatersrand deposits were transported in suspension as the prevailing hydraulic conditions exceeded those required to entrain and suspend gold in the channel. The model considers the dis- tribution of gold, therefore, in terms of an inter- action between ?ow in the channel where gold is being entrained, and ?ow on the overbank plain where suspended particles will tend to settle because of reduced transport capacity. The MIDAS program integrates the effects of ?ow velocity, sediment concentration, and the various parameters of transport sorting for both the lon- gitudinal ?ow in the channel and the transverse ?ow in the overbank plain. The results of com- puter simulations are compared to actual gold grade distributions in two examples from the Carbon Leader Reef (Figure 5.13b) and exhibit a remarkable degree of correlation, even when the bed morphology is complex. The fact that the highest gold grades are often found along the edges of channels has long been known, but has lacked a feasible explanation. The simulation suggests that the highest gold grades coincide with zones of ?ow interaction and that gold 264 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Shear velocity (m s –1 ) 0.5 2.0 Grain diameter (mm) 0.1 0.001 0.01 1.0 Clay 1 0.1 Silt Sand Long-term suspension Bedload creep No transport Short-term suspension Modified saltation Entrainment threshold Figure 5.12 Wind blown (eolian) sediment transport characteristics as a function of shear velocity and grain size (after Tsoar and Pye, 1987). ITOC05 09/03/2009 14:34 Page 264SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 265 500 1000 1000 (b) g t –1 500 Zones of turbulence Water surface Channel Longitudinal flow direction with suspension transport Overbank plain Transverse flow direction (a) Carbon Leader Reef Measured Simulated g t –1 Figure 5.13 (a) Geometry and ?ow characteristics of a braided fan delta complex used to simulate gold distribution patterns in Witwatersrand placer deposits. (b) Comparisons of computer simulations and actual gold distribution patterns in two different pro?les across the Carbon leader reef. Channel edge gold concentrations are reasonably well modeled in terms of transport sorting (after Nami and James, 1987). ITOC05 09/03/2009 14:34 Page 265distribution patterns in the Witwatersrand de- posits are a function, at least in part, of transport sorting. It should be emphasized, however, that this relationship exists despite the fact that it is well known that most of the presently observed gold in the basin has been remobilized and is secondary in nature. This is one of the reasons why the origin of Witwatersrand ores is still so controversial. 5.3 CHEMICAL SEDIMENTATION – BANDED IRON- FORMATIONS, PHOSPHORITES, AND EVAPORITES In contrast to the mechanical processes of sedimentation discussed above, where clastic sediment is sorted and deposited by water and wind, chemical sedimentation refers to the pre- cipitation of dissolved components from solution, essentially out of sea water or brine. A wide variety of rocks are formed by the compaction and lithi?cation of chemical precipitates and these include carbonate sediments (limestone and dolomite), siliceous sediments (chert) and iron-rich sediments (ironstones and banded iron- formations), as well as less voluminous accumu- lations of manganese oxides, phosphates, and barite. The majority of the world’s Fe, Mn, and phosphate resources, all extremely important and strategic commodities, are the products of chemical sedimentation and are hosted in chem- ical sediments. In addition, rocks such as lime- stone, comprising essentially CaCO 3 , have great value as the primary raw material for the manu- facture of cement. Furthermore, sediments known as evaporites, in which chemical precipitation is promoted by evaporation, contain the main economically viable concentrations of elements such as K, Na, Ca, Mg, Li, I, Br, and Cl, as well as compounds such as borates, nitrates, and sulfates, all of which are widely used in the chemical and agricultural industries. Most chemical sediments form in marine or marginal marine environments. The continental shelves, together with intratidal and lagoonal settings, represent the geological settings where chemical sediments and associated deposits are generally located. The chemical processes by which ore concentrations form are complex and controlled by parameters such as oxidation– reduction and pH, as well as climate, paleolatit- ude, and biological–atmospheric evolution. This section will ?rst consider the processes associated with the formation of ironstones, banded iron- formations, and bedded manganese oxide ores. This is followed by a discussion of phosphorites and also the formation of evaporites, carbonace- ous “black shales,” and manganese nodules. 5.3.1 Ironstones and banded iron-formations In terms of morphology, texture and mineralogy, there are three main types of iron ores and it is instructive to consider each of these indivi- dually since they illustrate different ore-forming process. The three types, in order of increasing importance, are bog iron deposits, ironstone deposits, and banded iron-formations, the latter generally referred to by the acronym BIF. Bog iron ores Bog ores form principally in lakes and swamps of the glaciated tundra regions of the northern hemisphere, such as northern Canada and Scandanavia. The deposits are typically small and thin, and comprise concentrations of goethite and limonite (Fe–oxyhydroxides) associated with organic-rich shale. They formed in the recent geological past, and in some places are still doing so at present (Stanton, 1972). The principal con- centration mechanism for iron in bog ores, as well as other iron ore deposits, is related to the fact that the metal occurs in two valence states, namely Fe 2+ (the ferrous ion) which is generally soluble in surface waters, and Fe 3+ (the ferric ion) which is insoluble and precipitates out of surface solutions. Concentration of iron occurs when aqueous solutions containing labile Fe 2+ , in a relatively reduced environment, are oxidized, with the subsequent formation and precipitation of Fe 3+ . Figure 5.14 illustrates the principle with respect to bog iron ores accumulating at the inter- face between oxygenated surface waters ?owing along an aquifer and the reduced iron-rich solu- tions percolating downwards through a swamp. 266 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 266SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 267 At this interface ferrous iron in solution is oxid- ized to ferric iron which precipitates as limonite or goethite. Ironstone deposits Ironstones are typically Phanerozoic in age and are widespread in occurrence, representing an important source of iron in the eastern USA and western Europe, particularly in the early part of the twentieth century. Ironstone deposits are commonly referred to as minette or Lorraine-type iron ores, well known occurrences of which were found in the Jurassic sediments of England and the Alsace-Lorraine region of France and Germany. In North America the Silurian Clinton type iron ores of Kentucky and Alabama are analogues of the younger European deposits. Ironstones form in shallow marine and deltaic environments and typically consist of goethite and hematite that has been rolled into oolites or pellets, suggesting the action of mechanical abrasion. The deposits contain little or no chert, but are commonly associated with Fe-rich silicate minerals such as glauconite and chamosite. The environment of deposition suggests that the iron was introduced from a continental source via a ?uvial system in which the metal was either in solution as Fe 2+ or transported as a colloid (see Chapter 3). Ironstones are often linked strati- graphically to organic-rich black shales and, in certain cases, seem to develop in strata represent- ing periods of major marine transgression and continental margin ?ooding. They do not occur randomly in time and show distinct peaks in the Ordovician–Silurian and again in the Jurassic periods. Their formation appears, therefore, to be related to a pattern of global tectonic cyclicity and speci?cally to times of continental dispersal and sea-level highstand (see Chapter 6), as well as periods of warmer climate and increased rates of chemical sedimentation (Van Houten and Authur, 1989). The origin of ironstones is complex and con- troversial and needs to account for both the Fe concentration processes and the formation of the ubiquitous oolites that typify these ores (Young and Taylor, 1989). A model for the formation of minette-type ironstones, after Siehl and Thein (1989), is presented in Figure 5.15. Iron is thought to have been concentrated initially on continents that were being subjected to deep weathering and erosion in a warm, humid climatic regime. Highly oxidized lateritic soils forming in such an envir- onment would have been the sites of iron enrich- ment as insoluble Fe 3+ remained in situ while other components of the regolith were leached away (see Chapter 4). In addition, laterites were also the sites where iron ooids formed in response to low temperature chemical and biogenically mediated processes. The lateritic soils and ooids were then transferred into a shallow marine Fe 2+ Swamp Aquifer with oxygenated groundwater Fe 2+ Fe 2+ Fe 3+ accumulation (as limonite Fe (OH) 3 ) Bog iron ore at redox interface Reduced carbonaceous shale Figure 5.14 The development of limonitic concentrations in the formation of a bog iron ore where a reduced solution transporting ferrous iron interacts with oxidized groundwater ?owing along an aquifer (after Stanton, 1972). ITOC05 09/03/2009 14:34 Page 267environment, either by ?ooding during trans- gression or by erosion during regression, to be reworked and concentrated in ?uvio-deltaic or littoral settings. Note that the chemically or biogenically formed pedogenic ooids are different to those that arise from the mechanical abrasion of particles in the swash zones of shallow mar- ine environments. Subsequent diagenesis of the ironstone accumulations resulted in further post-depositional modi?cations to the texture and mineralogy of the ores. Unlike the relatively simple processes invoked to explain the forma- tion of bog ores, ironstones appear to require a combination of speci?c environmental condi- tions, as well as a variety of processes (including oxidation–reduction, diagenesis, mechanical sedi- mentation, and microbial activity), to form sub- stantial deposits. Banded iron-formations (BIFs) The term banded iron-formation is somewhat controversial because of the connotation it has with respect to stratigraphic terminology. For this reason the term is hyphenated, and applies strictly to a bedded chemical sediment compris- ing alternating layers rich in iron and chert (Klein and Beukes, 1993). BIFs represent the most import- ant global source of iron ore and far outweigh the ironstone and bog iron ores in terms of reserves and total production. Whereas the latter occur predominantly in Phanerozoic rock sequences, BIFs are much older and formed in essentially three periods of Archean and Proterozoic Earth history, namely 3500–3000 Ma, 2500–2000 Ma, and 1000 –500 Ma (Figure 5.16). These three classes equate broadly with different tectonic settings and are referred to as, respectively, Algoma, Lake Superior, and Rapitan types. Algoma type BIFs are associated with volcanic arcs and are typically found in Archean greenstone belts. These deposits tend to be fairly small but they are mined in places such as the Abitibi greenstone belt of Ontario, Canada. The majority of Lake Superior, or simply Superior, type BIFs are located on stable continental platforms and were mainly deposited in Paleoproterozoic times. They represent by far the most important category of iron ore deposits (Figure 5.16) and most of the major currently pro- ducing iron ore districts of the world fall into this category. Examples include the Hamersley Basin of Western Australia (see Box 5.2), the Transvaal Basin of South Africa, the “Quadrilatero Ferrifero” of Brazil, the Labrador trough of Canada, the Krivoy Rog–Kursk deposits of the Ukraine, the Singhbhum region of India, and the type area in the Lake Superior region of the USA. Finally, the Rapitan type iron ores represent a rather unusual 268 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Black shale . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . .. Reworked ironstone (mechanically re- worked and sorted) Delta Laterite Fe concentration (pedogenic ooid formation) River Transgression Sea-level highstand Figure 5.15 Simpli?ed environmental model for the formation of oolitic ironstone ores. The model invokes initial Fe enrichment and pedogenic ooid formation in lateritic soils on the continental edge and transfer of this “protore” to a marginal marine setting with subsequent mechanical abrasion and reworking/concentration of ooids (after Siehl and Thein, 1989). ITOC05 09/03/2009 14:34 Page 268SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 269 occurrence of iron ores associated with glacio- genic sediments formed during the major Neo- proterozoic ice ages. The type occurrence is the Rapitan Group in the McKenzie Mountains of northwest Canada. In addition to the above tectonic classi?cation, BIFs have also been categorized in terms of the mineralogy of the associated iron phases. Although in most BIFs the iron mineral is an oxide phase (hematite or magnetite), carbonate (siderite), silicate (greenalite and minnesotaite), and sul?de (pyrite) iron minerals also occur, together with chert or carbonaceous shale. This observation led James (1954) to suggest a facies concept for BIF formation whereby the progres- sion from oxide through carbonate to sul?de phases was considered to re?ect precipitation of the relevant iron minerals in successively more reducing environments (iron silicate phases are stable over a wide range of Eh and do not, there- fore, conform to this simple progression). A con- sideration of the stability of the main iron phases as a function of Eh and pH con?rms that the redox state of the depositional environment plays an important role in determining the mineralogy of BIFs, although the situation is undoubtedly more complex than the facies concept would suggest. Figure 5.17a shows that ferrous iron is stable in solutions that are acidic and reducing, but that for a given pH, oxidation (or an increase in the Eh) will stabilize hematite (Fe 2 O 3 or ferric oxide), which is the principal iron phase over a wide 7.0 0.0 4.0 Tons (×10 14 ) of BIF 5.0 4.0 6.0 3.5 3.0 2.5 0.0 0.5 2.0 1.0 1.5 1.0 2.0 3.0 Algoma Rapitan Superior Ice sheet Magmatic arc “Black smoker” Continental crust Time before present (Ga) Sea level Continental platform Oceanic crust Figure 5.16 Tectonic and environmental model showing the depositional settings for Algoma, Superior, and Rapitan type BIFs (after Clemmey, 1985; Maynard, 1991). The inset histogram illustrates the approximate tonnages of BIF resource for each of the three major types as a function of time (after Holland, 1984). ITOC05 09/03/2009 14:34 Page 269range of geologically pertinent conditions. Siderite and pyrite are only stable under reducing con- ditions although these ?elds would obviously expand if the activities of total carbonate or sulphur in solution were increased. Similarly, the phase diagram indicates that magnetite is only stable under reducing and alkaline conditions, but this ?eld expands well into the range of neutral pH if the activities of carbonate and sulphur are lower than the prevailing conditions for this diagram. Banded iron-formations are chemical sediments in which the major components, Fe and Si, appear to have been derived from the ocean itself, rather than from a continental source, as in the case for ironstones. This is evident from the lack of alum- inous and silicate mineral particulate matter in BIFs. Models for the formation of these rocks are controversial and hampered by a lack of modern analogues. Features which need to be explained include the origin of the Fe and Si, their transport and precipitation mechanisms, the cause of the delicate silica- and iron-rich banding at various scales, and their formation during certain time periods, in particular between around 2500 and 2000 Ma. The source of iron is dominantly from the ocean itself, through either direct introduction of Fe 2+ from hydrothermal exhalations (black smokers; see Chapter 3) on the sea ?oor or dis- solution of oceanic crust. Rare earth element pat- terns of BIFs suggest that hydrothermal venting and volcanic activity are likely to have contrib- uted signi?cantly to the source of Fe 2+ in Algoma 270 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 1.0 –0.8 0 Eh (volts) 46 pH 2 8 10 14 12 –0.6 –0.4 –0.2 0.8 0.6 0.4 0.2 0.0 1.0 Fe 3+ Siderite FeCO 3 Magnetite Fe 3 O 4 Pyrite FeS 2 Lower limit for water stability Fe 2+ Hematite Fe 2 O 3 Upper limit for water stability (a) –0.8 0 Eh (volts) 46 pH 2 8 10 14 12 –0.4 –0.2 0.8 0.6 0.4 0.2 0.0 1.0 Pyrolusite MnO 2 Lower limit for water stability Mn 2+ Manganite Mn 2 O 3 Upper limit for water stability –0.6 Hausmannite Mn 3 O 4 Rhodochrosite MnCO 3 (b) Figure 5.17 (a) Eh–pH diagram showing the stabilities of common iron minerals. The conditions that apply to this particular phase diagram are: T = 25 °C, P total = 1 bar, molarities of Fe, S, and CO 3 are, respectively, 10 -6 , 10 -6 , and 1. (b) Eh–pH diagram showing the stabilities of common manganese minerals. Identical conditions apply, but with the molarity of Mn = 10 -6 (diagrams modi?ed after Garrels and Christ, 1965; Krauskopf and Bird, 1995). Note that the manganese oxides (MnO 2 and Mn 2 O 3 ) are stable at higher Eh than the equivalent ferric oxide (hematite), and would only form, therefore, under more oxidizing conditions. ITOC05 09/03/2009 14:34 Page 270SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 271 type BIFs, but that this in?uence diminished with time such that Rapitan ores re?ect a direct ocean water character (Misra, 2000). However, the enormous volumes of iron required to form the Lake Superior type BIFs, given that they appear to be spatially unrelated to volcanic or exhalative sea-?oor activity, remains something of a prob- lem. The oceans themselves are likely to represent an adequate source of Si since the solubility of amorphous silica (as Si(OH) 4 ) is relatively high (about 120mgl -1 ). It seems likely that the Archean–Paleoproterozoic oceans would have been saturated with respect to silica, ensuring a steady precipitation and accumulation of siliceous matter on the ocean ?oor during this early stage of Earth history. This pattern is likely to have changed in later geological times when the organ- isms that use silica to build an exoskeleton (i.e. siliceous protozoans such as radiolaria) evolved and the element was extracted from the water col- umn. The modern oceans, for example, typically contain <10mg l -1 dissolved silica and are markedly undersaturated (MacKenzie, 1975). The decrease of available Si in the oceans as a function of biological evolution contributes to an explana- tion for the changing character of iron ores, from BIFs to ironstones, with time. The transport and precipitation mechanisms involved in BIF genesis have traditionally been explained in terms of the Fe 2+ upwelling model, analogous to the observations of phosphorus upwelling onto the continental shelves (see sec- tion 5.3.3 below). Figure 5.18 illustrates the main elements of the model, after Klein and Beukes (1993). Ferrous iron from deep, reduced ocean levels is introduced to shallower shelf environ- ments by upwelling currents similiar to those seen along the western coastlines of continents such as Africa and the Americas today. These currents interact with the shallower waters and Fe 2+ is oxidized, with subsequent hydrolysis and precipitation of ferric hydroxide (Fe(OH) 3 ). In marine environments where CO 3 activities are suf?ciently high, FeCO 3 would precipitate. Oxidation takes place at shallow water levels where sunlight can penetrate and is due either to production of oxygen by photosynthesizing organisms living in the photic zone, or to photons of ultraviolet and blue light, which induce photo- oxidation of ferrous to ferric ions (Cairns-Smith, Fe 3+ Si Si Fe-oxides (hydroxides) Si FeCO 3 Fe 3+ Fe 2+ Si O 2 Si Si Upwelling and hydrothermal input >150 m Bottom Redox interface 100 m Photic limit 50 m Surface Reducing Oxidizing Photosynthesis Hematite Magnetite Siderite White Fe-poor chert Si O 2 Photo-oxidation Ultraviolet radiation C Black Carbon supply Si Si C Banded iron-formation Figure 5.18 Model invoking upwelling and oxidation of ferrous iron from an oceanic source to explain the depositional environment for BIFs. Oxidation of ferrous iron and precipitation of ferric iron compounds occurs at a diffuse redox interface formed by the production of oxygen in the upper water levels, either by photosynthesizing organisms or by ultraviolet radiation induced photo-oxidation, or both. The lateral zonation of BIF facies (i.e. siderite–magnetite–hematite) shown here differs from the simple scheme envisaged by James (1954). Diagram modi?ed after Klein and Beukes (1993). ITOC05 09/03/2009 14:34 Page 2711978). Precipitation of silica is problematical, but it has been suggested that evaporation from the ocean surface could promote local silica over- saturation and thereby promote precipitation as a gel which, on compaction, is transformed to chert. The explanation of banding in BIFs remains a problem and suggestions range from diurnal cycles (where photosynthesis and photo-oxidation shut down at night) for very thin lamellae, to seasonal cycles (where summer seasons promote biological activity and enhance iron precipitation) for mesobands. Other mechanisms that may have contributed to the formation of banding in BIFs include differential ?occulation of Fe and Si col- loids as a function of water chemistry, and some form of biological mediation. Studies of biominer- alization processes (see Chapter 3) in modern day environments have shown that the ?lamentous bacterium Chloro?exus is capable of binding either silica or iron, and occasionally both silica and iron, to the cell walls (Konhauser and Ferris, 1996). This form of induced biomineralization is considered to apply to the formation of some Precambrian BIFs, a notion supported by the presence of spheroidal structures in BIFs inter- preted as microfossils (LaBerge, 1973). During the late Archean and Paleoproterozoic, early marine planktonic bacteria that photosynthetically pro- duced oxygen may have been responsible for the oxidation of ferrous iron in the ocean waters and precipitation of ferric hydroxides. At the same time oversaturation of the oceans with respect to silica would have led to continuous biologically mediated precipitation of chert, where cell walls acted as the templates for nucleation of silica (Konhauser and Ferris, 1996). In this model, the banding in iron-formations is attributed either to episodic depletion of ferrous iron or to periodic (seasonal) diminished activity of the oxygen pro- ducing planktonic bacteria. The formation of BIFs in discrete time periods (Figure 5.16) is another feature that lacks detailed explanation, but is almost certainly related to atmospheric and biological evolution, as well as to the pattern of global tectonic cycles (see Chap- ter 6). The enormous Superior type BIFs formed on stable continental shelves during periods of marine transgression and sea-level highstand, conditions which re?ect tectonic cycles of contin- ental dispersion and active, elevated mid-ocean ridges. These conditions, coupled with the relev- ant states of biological and ocean–atmosphere evolution, were unique to the Paleoproterozoic and help to explain the predominance of BIFs at this time. Subsequent to oxyatmoinversion (i.e. the period in the Paleoproterozoic when oxygen levels in the atmosphere increased sub- stantially; see Chapter 6) the atmosphere and oceans became more oxidized and iron existed largely as insoluble Fe 3+ , such that the likelihood of Lake Superior type BIFs forming diminished. Rapitan type ores, on the other hand, appear to be intimately associated with widely distributed Neoproterozoic glaciogenic sediments and the near-global glaciations that characterized this time period (i.e. the so-called “Snowball Earth” period). The large-scale covering of the oceans with an ice-cap would have isolated them from the atmosphere and reintroduced the sort of reducing conditions in much of the ocean water column that only existed prior to oxyatmoinver- sion. It is widely held that the development of the Rapitan type iron ores is related to this near global oceanic anoxia, which, in turn, led to enrichment of the oceans in ferrous iron. Inter- and post- glacial melting of the ice-cap resulted in glacio- genic sedimentation, reoxidation of the oceans and associated precipitation of ferric iron. Such a scenario has been used, for example, to explain the formation of the Braemar BIFs and diamictic ironstones associated with Sturtian aged (circa 750 Ma) glaciogenic sediments in the Adelaide geosyncline of South Australia (Lottermoser and Ashley, 2000). 5.3.2 Bedded manganese deposits Most of the world’s sedimentary manganese deposits formed in environments similar to those in which BIF and ironstone ores also formed. Superior type BIFs are sometimes closely asso- ciated with manganese ores, the prime example being the world’s largest exploited Mn deposits in the Kalahari manganese ?eld of South Africa. In this case, enormous reserves of bedded manganese oxide ores were formed within the 272 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 272SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 273 The Archean to Paleoproterozoic Hamersley Province of Western Australia contains huge volumes of Superior-type banded iron-formations (BIF) that host some of the world’s largest iron deposits. Western Australia is the world’s largest exporter of iron ore, the bulk of which comes from enriched BIF in the Hamersley Province. Several large mines, including Mesa J, Tom Price, Paraburdoo, and Mount Whaleback, together with smaller operations, com- bine to produce over 170 million tons of export quality iron ore annually, much of which goes to Japan, South Korea, and China (Kneeshaw, 2002). The Mount Whaleback deposit is the largest single iron ore deposit in Australia, with an original ore resource of some 1.8 billion tons. The deposit is located in the southeastern portion of the Hamersley Province in an area that was episodically deformed, with the most intense folding attributed to the Ophthalmian orogeny between 2400 and 2200 Ma (Figure 1). The ore is hosted in BIF and ferruginous shales of the Dales Gorge Member of the Brockman Iron Formation. The latter unit is the middle of three major iron-formation sequences in the Hamersley Province, forming between 2540 and 2450 Ma (Barley et al., 1997). This period of Paleoproterozoic Earth history was a time when similar chemical sediments were being deposited in continental margin or platformal settings else- where in the world, such as South Africa, Brazil, and the USA (see section 5.3.1 of this chapter). These countries also contain large BIF-hosted iron ore deposits that are very similar to those of the Hamersley Province. The BIF host rock in all these deposits was originally a bedded chert– magnetite/ hematite chemical sediment that typically con- tains around 25–30% Fe. In order to make an economically viable ore deposit the BIF needs to be upgraded so that the Fe content is at least double this value, a process that is achieved by either selectively leaching or replacing the silica in the rock. The processes whereby BIF is enriched to form iron ore are complex and episodic. There is consider- able controversy in the literature regarding the nature of the enrichment process and both supergene and hypogene processes have been advocated. The conventional model of ore formation at Mount Whaleback, and indeed most other deposits of the Hamersley Province, is that BIF was exposed and sub- jected to supergene alteration after the main period of deformation and folding of the host rock. This process involved in situ replacement of chert by goethite and oxidation of magnetite to hematite (a process called martitization). The actual processes involved are some- what akin to those taking place during the formation of laterites (see section 4.3 in Chapter 4), although rather than simple leaching of silica and residual enrichment of iron, the chert is largely replaced by Fe-oxyhydroxides. After supergene alteration the resulting goethite–martite rock was then buried and metamorphosed to form micro- crystalline hematite–martite ore. This model, envisaging a combination of supergene and metamorphic processes, was initially attributed to Morris (1985) and is believed to be consistent with the characteristics of most of the enriched BIF deposits of the region. These characteristics include the high Fe grades (65% Fe) of the enriched BIF, the lumpy character of the ore, and low phosphorous con- tents (around 0.05%), all features that are highly desirable in the marketing of the ore (Kneeshaw, 2002). Exposure of the BIF and supergene alteration apparently took place prior to 2200 Ma (Martin et al., 1998a), which is the maximum age of deposition of conglomerates adjac- ent to the Hamersley Province (i.e. the Wyloo Group; Figure 1) within which detrital fragments of hematite ore are found. Fluid inclusion and stable isotope data have also suggested that the ?uids involved with the early phases of hematite–martite ore formation were hot (up to 400 °C), saline hydrothermal solutions with characteristics incon- sistent with a supergene or near-surface setting. These and other observations, such as the mineralogy of the ore assemblage at the Tom Price mine, have resulted in a number of workers suggesting a hypogene origin for the enriched BIF-hosted hematite–martite ores (Barley et al., 1999; Powell et al., 1999). One suggestion is that hot, oxidizing basinal ?uids were focused into low angle thrust faults during the main stage of deformation and folding (i.e. “orogenic ?uids”; see Chapter 3) and that it was these ?uids that were initially responsible for silica dissolution and hematite recrystallization in the BIF. One of the features that appears to favor a hypogene origin for the Hamersley iron ores is that they can extend to considerable depths (over 500 m) below the surface. This feature on its own does not, however, militate against supergene processes. The Morris et al. (1980) model for Mount Whaleback envisages supergene alteration by electrochemical processes (Figure 2). The high elec- trical conductivity of magnetite-rich layers in BIF enables oxidation potentials to extend to signi?cant depths, with an electrical circuit being completed by ionic conduction through groundwaters circulating along suitably positioned fault systems. A cathode is set up at surface where electrons are consumed in oxygenated groundwaters. The anode Chemical sedimentation: banded iron-formations: the Mount Whaleback iron ore deposit, Hamersley Province, Western Australia ITOC05 09/03/2009 14:34 Page 273274 PART 3 SEDIMENTARY/SURFICIAL PROCESSES + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Wyloo Group Port Hedland Perth Location map Western Australia Australia Pilbara province Yarrie Marble Bar Nullagine Archean granite– greenstone terrace Port Hedland Roebourne Mesa J Dampler Hamersley Group Paraburdoo Tom Price Hamersley province Mt Whaleback Fortescue Group + + + + + + + + + + + + + + + Figure 1 Simpli?ed geology and location of some of the major iron ore deposits in the Hamersley Province, Western Australia (after Kneeshaw, 2002). ITOC05 09/03/2009 14:34 Page 274SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 275 Transvaal basin at stratigraphic levels just above those where the equally substantial BIF-hosted iron ores occur. This is not unexpected given the remarkably similar chemical properties of Fe and Mn. The geochemical behavior of manganese, like iron, is controlled by oxidation potential and it exists as Mn 2+ , which is soluble under reducing and acidic conditions, as well as Mn 3+ and Mn 4+ , which are less soluble and stabilize as manganese oxides under relatively oxidizing and alkaline conditions (Figure 5.17b). Pyrolusite, or MnO 2 , is the dominant oxide phase at high Eh and over a range of pH. Comparison of the manganese with the iron phase diagram (Figure 5.17a and b) shows that higher oxidation potentials are required to stabilize pyrolusite than hematite. The fact that BIF Hanging wall shale Low grade ore High grade ore Footwall shale Goethite–martite Cathode 4e – +O 2 +H 2 O 4OH – Anode Anode e – e – Water table Ground water f Fe 2+ Fe 3+ +e – Fe 3+ +3H 2 O Fe(OH) 3 +3H + 100 m 100 m Fault zone f Present surface Figure 2 Simpli?ed cross section through the Mount Whaleback iron ore deposit showing the extent of enriched BIF-hosted ore and the postulated nature of iron enrichment by an electrolytically mediated process (after Morris et al., 1980). exists at depth where ferrous ions are oxidized to ferric ions and then precipitated as an iron-oxyhydroxide such as goethite. This electrochemical process explains alteration processes at signi?cant depths, but does not preclude the existence of normal supergene processes, or lateritization, in the near surface environment (Morris et al., 1980). Finally, it should be noted that the Hamersley Province is characterized by a signi?cant component (about 35%) of much younger mineralization that clearly is related to supergene processes active during a cycle of Paleogene erosion and weathering. These iron deposits are referred to as channel and detrital ores and are located in mature river courses and colluvial fans, both of which are spatially linked to bedded ores (Kneeshaw, 2002). Although the origin of the enriched BIF ores remains controversial, the region has been subjected to a long and involved geo- logical history and the ore-forming processes are expected to be complex and episodic. ITOC05 09/03/2009 14:34 Page 275Fe 2+ oxidizes more readily than Mn 2+ means that iron can precipitate while Mn remains in solu- tion, and provides a possible explanation for the observation that iron ores may be spatially (or stratigraphically) separated from manganese ores, as in the case of the Transvaal basin. In the latter case, early precipitation of ferric iron out of solu- tion to form huge volumes of BIF (within which the Sishen and Thabazimbi deposits occur) would have depleted the water column of iron so that later precipitation, at a higher stratigraphic level and under more oxidizing conditions, resulted in the formation of Fe-depleted manganese oxide ores. Manganese deposits are located in sediments of variable age, from the Paleoproterozoic to recent. Some of the biggest deposits are Phanerozoic in age, such as the Cretaceous Groote Eylandt manganese oxide ores in northern Australia, and the Molango district of Mexico where the ores are made up dominantly of rhodochrosite (MnCO 3 ). An interesting modern day analogue for the form- ation of sedimentary manganese ores is provided by the Black Sea, where active sedimentation results in ongoing accumulation of MnO 2 . The Black Sea is markedly strati?ed in terms of water density and composition and the waters below about 200 m depth are euxinic (i.e. highly reduced, with H 2 S stable). Pyritic muds accumulate on the sea ?oor and this effectively depletes the entire water column of iron. Mn by contrast is concen- trated in the deeper waters, since it exists in solu- tion as Mn 2+ , but is depleted in the upper oxidized 200 m where it precipitates and settles out. The depth–concentration pro?les for Mn and Fe in the Black Sea are illustrated in Figure 5.19a, which shows the zone of manganese enrichment that forms just below the redox interface where mixing between high-Mn euxinic and oxidized surface waters occurs. At the interface Mn 2+ is oxidized to Mn 3+ and particulate Mn forms and then settles at deeper levels where it can redis- solve. However, where the enriched mixing zone intersects shallow sea ?oor, accumulation of either MnO 2 or MnCO 3 (depending on the local oxidation potential; see Figure 5.17b) takes place on the substrate. Mineralization is further enhanced if the sites of accumulation are pro- tected from dilution by clastic sediment input, as 276 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Figure 5.19 (a) Depth concentration pro?le for Fe and Mn in the Black Sea. Fe is depleted in the water column by pyrite accumulation on the sea ?oor, whereas Mn occurs to moderately high levels in solution in the dominantly reduced water column. Maximum enrichment of Mn occurs just below the redox interface where Mn 2+ is oxidized to Mn 3+ and precipitates. (b) The distribution of Mn in sea ?oor sediments of the Black Sea. Accumulation of Mn occurs at the intersection of the zone of Mn precipitation with the sea ?oor (diagrams modi?ed after Force and Maynard, 1991). 0 –1500 Depth relative to redox interface (m) –1000 –500 100 Concentration (ppb) 200 300 500 Oxidized (O 2 zone) Approximate shelf break (H 2 S zone) Reduced/euxinic Mn 2+ Fe Shoreline Euxinic-facies sediments Mn < 0.05% Fe > 5% Black Sea Zone of MnO 2 enrichment (no dilution by clastic sediment input) Oxic facies sediments Mn ~0.15% Fe ~5% Current (a) (b) Detrital source material Mn 0.1% Fe 5% ITOC05 09/03/2009 14:34 Page 276SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 277 shown for the north-central portion of the Black Sea (Figure 5.19b). 5.3.3 Phosphorites Phosphorus is a very important element that is essential for the growth and development of most living organisms. It is a basic ingredient of some proteins but, more importantly, it is a building block for the bones and teeth of vertebrates, as well as the shells and chitinous exoskeletons of many invertebrates. Plants also require it for growth and it is for this reason that phosphate is such an important ingredient of fertilizer. In order to sustain global food requirements the manufacture of arti?cial fertilizers has become an enormous industry that requires approximately 150 million tons of phosphate per year, the raw material for which comes predominantly from phosphorus-rich sediments or phosphorites (i.e. a sediment containing more than 15–20% P 2 O 5 ). Phosphorites generally form in much the same environmental niches as do banded iron- formations, ironstones, and bedded manganese ores, namely along continental shelves and in shallow marginal marine settings such as lagoons and deltas. The processes of formation are, there- fore, very similar to those for Fe and Mn deposits and essentially involve the upwelling of cold ocean water containing above-normal concentra- tions of metals, and their subsequent precipita- tion onto the continental shelves. In fact one of the earliest published accounts of the upwelling hypothesis, attributed to the Russian scientist A.V. Kazakov in the 1930s, applied to the forma- tion of phosphate-rich sediments in a continental shelf setting, and the concept was only later applied to iron and manganese deposits. Deep sea drilling and ocean ?oor exploration have shown that phosphate concentrations are presently form- ing off-shore along the coastlines of Namibia and Chile–Peru, and are related to upwelling of deep, cold currents carrying small amounts of phos- phate in solution. The same environments are also well known as proli?c ?shing grounds, a feature which substantiates the relationship between the supply of phosphatic nutrients and biological productivity. When the idea of oceanic upwelling was ?rst introduced, it was suggested that precipitation of phosphorus onto the sea ?oor in areas of upwelling was inorganic and related to decreasing solubility in the near shore environment. This notion is over-simpli?ed and it is now known that the phosphorus cycle in the oceans is largely controlled by organic processes. A role for micro- organisms in the deposition of phosphate is also considered likely (Trudinger, 1976). Phosphorus dissolves in sea water as either PO 4 3- (stable under very alkaline conditions), HPO 4 2- (the dominant anionic complex), or H 2 PO 4 - (stable under acidic conditions) complexes. These anionic complexes are absorbed by the organisms living in shallow marine environments to form their shells, bones, and teeth. The subsequent accumulation of phos- phorus on the sea ?oor, therefore, is not related to a chemical redox reaction, as in the cases of Fe and Mn, but occurs after the host organism dies and settles to the ocean ?oor. Phosphorus is released from the decaying organism to form a calcium phosphate compound, which then converts to an impure, cryptocrystalline form of apatite (Ca 5 (PO 4 ) 3 [F,OH]) or collophane (a carbonate–apatite). The concentration of calcium phosphate on the sea ?oor is described in terms of the reaction: 3Ca 2+ + H 2 PO 4 ? Ca 3 (PO 4 ) 2 + 2H + [5.7] which suggests that its precipitation is dependent, among other things, on pH. Figure 5.20 shows apatite solubility in sea water as a function of pH and oxalate content of the solution (oxalic acid is H 2 C 2 O 4 ). It is apparent that apatite solubility is signi?cantly higher under more acidic conditions and, on this basis, that apatite formation is more likely to occur under the relatively alkaline, warm-water conditions that apply to continental shelves. Deeper, colder and more acidic waters will promote phosphate solubility. The formation of apatite/collophane in phos- phatic sediments is, however, not a straight- forward process and there has been considerable debate as to whether it develops as a primary precipitate at the sea water–sediment interface (i.e. where the organic material is decaying) or as a ITOC05 09/03/2009 14:34 Page 277product of diagenesis during later compaction and dewatering of the sediment. This debate has considerable bearing on the origin and formation of phosphorite ores and evidence seems to exist for both processes. Typical concentrations of dissolved phosphate in the deep portions of the present day oceans are low (about 50–100 ppb; Bentor, 1980) and the shallow marine environ- ment is even more strongly depleted because of biological uptake. Given the low present day concentrations of phosphate in solution, even in the deep ocean, it seems very unlikely that sea water is saturated with respect to apatite. This, in turn, suggests that the primary precipitation of apatite at the sea water–sediment interface is also unlikely. Although this situation applies gener- ally to the present day oceans, it may not, how- ever, have prevailed in previous geological eras when factors such as ocean chemistry and biolo- gical evolutionary development were different. It may also not apply to restricted, but biologic- ally productive, environments (such as a lagoon) where high rates of organic decay can produce localized saturation of phosphate in sea water, as in the case of the southern African example described below. Sheldon (1980) has suggested that phosphate concentrations in the Precambrian oceans might have been higher than at present because the lower-O 2 , higher-CO 2 atmospheric conditions prevailing at that time would have meant a more acidic sea-water composition and higher phos- phate solubility. It is also apparent that oceans that existed before the development of phosphate- dependent organisms did not witness biological phosphorus depletion. By the same token, oceans at this time could not produce the environments in which biologically mediated deposition and concentration of apatite or collophane might occur. Substantial Precambrian phosphorite deposits, therefore, do not exist, and the few that do occur are likely to have formed from the direct pre- cipitation of phosphate from sea water. As the atmosphere evolved, however, any event that coincided with a substantial proliferation of life, such as the Cambrian explosion, could have had the effect of removing CO 2 from the oceans and atmosphere, increasing pH in the shallow water environment and further promoting the direct precipitation of apatite from a saturated sea water column. A more ef?cient form of concentration, however, would have been through the decay of phosphate-dependent life forms and the forma- tion of phosphorites in association with organic- rich sediments. This is probably the reason why all large phosphorite deposits are Phanerozoic in age. It is pertinent to note that the Precambrian– Cambrian boundary, for example, coincides with a period in Earth history when substantial deposi- tion of phosphoritic sediments took place (Cook and Shergold, 1984), good examples of which include the deposits of Mongolia, southeast Asia and Australia. By contrast, formation of Mesozoic and Cenozoic phosphorite deposits must have taken place essentially during the diagenetic stage of sediment evolution, since the phosphate con- centration levels in the water column were too low to allow for direct apatite or collophane pre- cipitation. In this scenario, the concentration of phosphate in the organic-rich host sediments builds up progressively with time such that the interstitial waters would eventually become highly enriched in phosphate (measurements of up to several thousands ppb HPO 4 2- in solution have been recorded; Bentor, 1980). This represents an ideal environment for the formation of diagen- 278 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 0.001 Solubility of apatite (moles liter –1 ) Total oxalate concentration (moles liter –1 ) 10 –1 0.01 0.1 0.0001 10 –6 10 –5 10 –4 10 –3 10 –2 pH 8 pH 7 pH 6 pH 5 Figure 5.20 Estimates of apatite solubility in sea water (expressed in terms of phosphate concentration) as a function of solution pH and oxalate concentration (after Schwartz, 1971). ITOC05 09/03/2009 14:34 Page 278SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 279 etic apatite as the sediments are compacted and lithi?ed. Such processes would appear to pertain to deposits such as the Cretaceous–Paleogene de- posits of Morocco and possibly also the oolitic phos- phorites of Florida. In the latter case, however, subsequent mechanical abrasion and sedimentary transport of diagenetic aggregations resulted in deposits that are reworked and allochthonous (Riggs, 1979). Other factors, such as global climatic patterns and eustatic sea-level changes, are also known to have played a role in the formation of phosphor- ites (Cook and McElhinny, 1979; Sheldon, 1980). Figure 5.21a shows that there is a broad correla- tion between the number of phosphate deposits that have formed and periods of sea-level high- stand. The explanation offered is that elevated sea levels and ?ooding of the continental shelves enhanced circulation and upwelling. If enhanced upwelling occurred in an equatorially aligned sea way (Figure 5.21b) then phosphorite formation would have been promoted by the biogenic concentration and/or solubility decrease that accompanied the existence of such a biologically productive environment. The late Mesozoic to Cenozoic phosphorite deposits of North Africa, deposited in the Tethyan sea way, may well repres- ent examples of this control mechanism. Figure 5.21a shows that there is also a correlation be- tween glacial events and phosphogenesis. In this case the lack of vertical mixing in the oceans during stable periods results in a phosphate build- up in the deep water sink. The development of strong trade wind systems between glacial and non-glacial events in mature, longitudinally ori- entated oceans would promote the establishment of major oceanic gyrals which would, in turn, result in strong upwelling along the western edges of continental land masses (Figure 5.21c). This mechanism could explain the development of the Permian Phosphoria Formation, probably the world’s largest accumulation of phosphate ores, along the western margin of continental USA in Pangean times. This scenario also obviously applies to the present day Atlantic ocean and accounts for the pattern of upwelling and phos- phate deposition along the western margin of Africa (see below). A model for phosphogenesis based on present day deposition Birch (1980) studied the nature and origin of phos- phorites forming off the western and southern continental margins of South Africa and Namibia. Exploration of the continental shelf in this region reveals two types of phosphorite (Figure 5.22a). One variety developed mainly along the Namibian coastline and comprises oolitic apatite/collophane ores derived by accretionary growth arising from the direct precipitation of phosphate from sea water. The other type, formed mainly along the South African coastline, consists of phosphatic replacement of fossiliferous limestone and is diagenetic in origin. Both types of phosphorite are geologically recent (Eocene to Miocene) in age and are believed to have formed at the same time. A model for this type of penecontemporaneous ore formation is shown in Figure 5.22b. Diffuse upwelling along the outer reaches of the shelf results in reduced nutrient supply and biological productivity, and consequently the rate of phos- phate release during decay of organisms on the sea ?oor is also low. Apatite is undersaturated and cannot precipitate directly from the sea water, but subsequent diagenesis can lead to replacement of limestone by calcium phosphate. By contrast, in the near shore environment very high biological productivity in restricted shallow embayments and lagoons is maintained by wind-induced nutrient upwelling. Mass mortality of life forms in these restricted environments (especially dur- ing summer low tides) leads to the accumulation of abundant organic matter on the substrate with local increases in dissolved phosphate content to saturation levels. Varve-like layers of pure apatite as well as accretionary oolitic growth provide the textural support for a mode of formation by direct precipitation of phosphate at the sea water– sediment interface. 5.3.4 Black shales Shales that are rich in organic matter are econom- ically important because they are often enriched in a large variety of metals, including V, Cr, Co, Ni, Ti, Cu, Pb, Zn, Mo, U, Ag, Sb, Tl, Se, and Cd. ITOC05 09/03/2009 14:34 Page 279280 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Million years 0 600 400 500 300 200 100 Cenozoic Paleozoic Mesozoic (a) 20 10 0 Phosphate deposits Glacial events 200 0 –200 Sea level (m) (b) Latitudinal model Continent Equatorial upwelling Tethys- like seaway Continent (c) Longitudinal model Atlantic- like ocean Continent Northern gyre Southern gyre Trade wind upwelling Trade wind upwelling Trade winds Phosphogenesis Continent Figure 5.21 (a) Generalized correlations between the development of phosphate deposits, glacial events and eustatic sea level changes in the Phanerozoic Eon. (b) Model for phosphate deposition related to equatorial upwelling in a latitudinally orientated seaway such as Mesozoic Tethys. (c) Model for phosphate deposition related to trade wind induced circulation patterns in a mature longitudinally orientated ocean (after Sheldon, 1980). ITOC05 09/03/2009 14:34 Page 280SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 281 Diagenetic phosphate-rich rocks Shelf break Dilute upwelling low biological productivity low dissolved PO 4 content 20° 25° 30° 35° 15° 20° 25° 30° (a) Walvis Bay Cape Town Port Elizabeth Agulhas Bank South Africa Durban Directly precipitated phosphate-rich rocks Namibia Orange River 100 200 300 0 km (b) Wind “Pavement” of diagenetic phosphate-rich rocks Phosphatic varieties formed by direct precipitation High biological productivity, high rates of mass mortality, apatite saturation in sea water Figure 5.22 (a) Map showing the distribution of phosphate-rich sediments along the continental margins of South Africa and Namibia. (b) Model showing contrasting modes of phosphate deposition in the outer shelf (where diagenetic accumulation of apatite is taking place) and in restricted near-shore environments characterized by high biological productivity and seasonal mass mortality (where precipitation of apatite directly from sea water is occurring). Diagrams after Birch (1980). ITOC05 09/03/2009 14:34 Page 281Although few metalliferous black shales are presently mined (with the exception of China where several small deposits are exploited) they represent a signi?cant metal resource and prob- ably also a source of ore components for younger hydrothermal deposits. As mentioned previously, many black shales are spatially associated with ironstones (Figure 5.15) and form in the same environments. At a global scale, they formed penecontemporaneously with ironstones, and their major periods of development can be linked to the ?rst order eustatic sea-level highstands in the Ordovician–Devonian and Jurassic–Paleogene intervals (see Chapter 6). In this respect they are also linked to the Ocean Anoxic Events (OAEs) that are now well documented in the Phanerozoic Eon. Sea-level highstand resulted in continental margin ?ooding and transgression of muddy sediment across the shelf. The deeper water con- ditions that generally applied resulted in poorer circulation, a reduction in oxygen levels, and widespread oceanic stagnation (Van Houten and Authur, 1989), providing the kind of depositional environment favorable for metalliferous black shale formation. Well known examples of metalliferous black shales include the Cambro – Ordovician Alum shale of Scandanavia, which is particularly enriched in uranium, and the Devonian New Albany shale of Indiana, USA, which has signi?cant concentra- tions of Pb as well as V, Cu, Mo, and Ni. The euxinic shaley sediments forming on the ?oor of the Black Sea represent an example of manganese, as well as Co, Cu, Ni, and V, enrichment, in a present day sedimentary basin (Holland, 1984). Black shales are characterized by signi?cant quantities of organic carbon which is preserved from degradation/oxidation by the almost total lack of free oxygen in the immediate environment of deposition. This environment is not only anoxic but also euxinic (where reduced sulfur species are stable) and it is this combination, together with a lack of clastic dilution, that provides the optimal conditions for development of metalliferous black shales (Leventhal, 1993). A sediment organic con- tent of even 1% will bacterially and chemically deplete the oxygen content of the immediate environment to virtually zero. This microenvir- onment promotes the growth of sulfate-reducing bacteria which in turn produce HS - or H 2 S (depending on the local pH) and the development of euxinic conditions. The biologically mediated development of euxinic waters can be described in terms of the following reaction: R(CH 2 O) 2 + SO 4 2- ? R + 2HCO 3 - + H 2 S [5.8] where R(CH 2 O) 2 is an abbreviated representation of a complex (organic) carbohydrate molecule. The subsequent concentration of metals in this environment is brought about by the af?nity of positively charged metals in solution in the sea water for the thio complex, as demonstrated in the following simpli?ed reaction: HS - + Me 2+ ? MeS + H + [5.9] where Me is a metal. The above reactions imply a correlation between organic carbon, sulfur, and metal contents in black shales since the higher the content of sulfate-reducing bacteria, the more sul?de is produced and the greater the degree of extraction of metal from the sea water column would be. Such a correlation is well demonstrated in Figure 5.23a, for Devonian black shales in the eastern USA, and lends support to the processes outlined above. The source and metal content of black shales will obviously vary from place to place and this accounts for the observation that different horizons have characteristic metal signatures. In oceanic settings submarine exhalative vents, or black smokers, represent likely metal sources, whereas in continental margin settings (such as the Black Sea) metals are derived from terrigenous clastic input. Figure 5.23b summarizes the geolo- gical and environmental conditions necessary for the optimal development of a metalliferous black shale. This type of sedimentary rock represents a potentially important source of exploitable met- als in the future. 5.3.5 A note on ocean ?oor manganese nodules Since their discovery during the 1872–6 oceano- graphic expedition by HMS Challenger, it has 282 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 282SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 283 become apparent that a vast resource of Mn, Fe, Cu, and Ni (as well as lesser amounts of Co, Zn, and other metals) is contained within the (ferro)- manganese nodules that occur in the deep, pelagic portions of all the major oceans. Manganese nod- ules typically occur in parts of the ocean basins where sediment accumulation rates are very low (less than 7 meters per million years; Heath, 1981) 0 0 0 Molybdenum (ppm) 10 5 15 12 1 0 Elemental carbon (wt%) 45 36 7 8 9 20 25 30 35 40 10 20 30 40 50 60 70 0 1 2 3 4 5 6 (a) Sulfur (wt%) Uranium (ppm) (b) Marine Basin/Sink Anoxic Environment Euxinic Environment Metal rich black shale Sulfate reduction in water column and sediments High S/C Burial and time Good preservation of orga nic matter and sul?des Low or no O 2 in water column sulfate reduction in sediment Source of SO 4 2– Stratified water column Restricted basin Slow sedimentation rate (less clastic dilution) H 2 S present Organic Matter Mafic (Ni, Cr, Co, Y, Cu, Fe, Mn, Ag, Zn) Metal Source Felsic (Th, U, REE, Pb, Ba) Marine source Terrestrial source Figure 5.23 (a) Plots of carbon content versus sulfur, molybdenum, and uranium for Devonian black shales from the eastern USA (after compilations by Holland, 1984). (b) Schematic outline of the geological and environmental conditions required for the optimal formation of metalliferous black shales (after Leventhal, 1993). and are absent from areas of rapid sedimentation, as well as from equatorial and high-latitude regions of high biological productivity and chem- ical sedimentation. They appear to be best devel- oped in the Paci?c Ocean, where exploration has identi?ed large areas of proli?c nodule formation (up to 100 nodules per square meter). These areas also contain nodules that have higher than ITOC05 09/03/2009 14:34 Page 283normal metal contents, with up to 2 wt% com- bined Cu + Ni and substantial Co and Zn (Fig- ure 5.24). One such area is the Clarion–Clipperton region, to the west of Mexico. Manganese nodules are typically ovoid, a few centimeters in diameter, and have the appearance on the sea ?oor of a burnt potato. In cross section they are complexly layered, often partly con- centric, and represent concretions that have apparently nucleated around a clastic or biogenic fragment. They often exhibit internal radiating shrinkage cracks re?ecting a change in mineral- ogy from hydrated Fe and Mn oxyhydroxides to a variety of more compact, but chemically com- plex, Fe and Mn oxides such as vernadite, birnes- site, todorokite, goethite, and hematite (Heath, 1981). The formation of manganese nodules is a complex process and still not fully understood. Metals are undoubtedly derived from the sea water itself, but ultimately come from various sources including submarine exhalative vents, clastic and volcanic input, and diagenesis of pelagic sediments. The formation of the nodules and the concentration of metals within them have been attributed to one or more of the follow- ing mechanisms: • Settling of clay and biogenic (fecal) debris that con- tained either absorbed or ingested metals obtained from the sea water, with subsequent release of metals during dissolution and decomposition. • Direct precipitation of metals from sea water onto a suitable substrate which forms the nucleus around which concretionary growth takes place. • Upward diffusion of metals in ocean bottom sediment pore waters. • Authigenic reactions in ocean bottom sedi- ments during alteration and compaction. • Bacterial activity and the oxidation of transi- tion metals. In this regard, certain types of bac- teria (such as Metallogenium) are known to be particularly prevalent in the oxidation and subse- quent precipitation of labile Mn 2+ and have been detected in manganese nodules from the Atlantic Ocean (Trudinger, 1976). 284 PART 3 SEDIMENTARY/SURFICIAL PROCESSES East Pacific Rise 40°S 60°W 100° 140° 180° 140° 100°E 20° 0° 20° 40° 60°N Australia North America Clarion- Clipperton region 1.0 0.5 1.0 0.5 Equator 1.0 Russia Wt% Mn 25 Fe 7 Ni 1.1 Cu 1.1 Co 0.2 Zn 0.1 0.4 Figure 5.24 Map showing sub- equatorial zones of base metal enriched manganese nodule concentration in the Paci?c Ocean, contoured with respect to the copper content (in wt%) of nodules. Nodule-free areas occur close to the continents, where sedimentation rates are high, as well as along the equator and East Paci?c Rise, where biological productivity is high. The Clarion–Clipperton region, a particularly enriched zone, is boxed (after Heath, 1981). ITOC05 09/03/2009 14:34 Page 284SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 285 • Preferential uptake of dissolved transition metals into speci?c Fe and Mn oxyhydroxide minerals, such as the af?nity of Cu and Ni for todorokite (a multi-element Mn–oxyhydroxide). The current level of uncertainty about the processes involved in the formation of deep ocean manganese nodules is re?ected in the variety of suggested mechanisms listed above. Many of these suggestions are not mutually exclusive and it is likely that nodule formation is a long-lived and complex process. The biggest problem with the exploitation of manganese nodules at present is not so much technical as it is political and eco- nomic. The technologies required for exploiting manganese nodules at depth in several kilometers of ocean water could be overcome, but the formu- lation of an internationally acceptable Law of the Sea is a much bigger and more contentious issue. Manganese nodules, nevertheless, represent an enormous metal resource for the future. 5.3.6 Evaporites The bulk of the world’s production of rock salt (halite), as well as of potash, borates, and nitrates for agricultural fertilizers, comes from rocks known as evaporites. These are chemical pre- cipitates that form as a result of evaporation of a brine, usually derived from sea water. There are two main environments in which evaporites form. The major one, in which the so-called “saline giants” of the world formed, is marginal marine in setting and represented by large lagoons or restricted embayments into which periodic or continual sea water recharge occurs. This type of deposit may be both laterally extensive and thick, but is characterized by a relatively limited range of mineral precipitants derived from sea water, which itself has remained fairly constant in terms of bulk composition over much of geological time. The second setting is intracontinental and lacustrine and results in much smaller, thinner deposits which are characterized by a more diverse range of mineral precipitants since continental ?uvial input introduces a broader range of dissolved ingredients to the lake waters. Examples of deposits representing the former include the Permian Zechstein Formation, which formed in a shallow sea covering large areas of the United Kingdom, The Netherlands, Germany, and Poland, and also the Silurian Salina Formation of New York and Michigan states in the USA. Typical products mined from these deposits include halite, potash, and sulfates. Probably the largest known evaporite sequence occurs in late Miocene strata beneath the ?oor of the Mediter- ranean Sea, where deposits extend laterally for over 2000 km and may be up to 2 km thick. Much of this resource is obviously unexploitable, but is so large that its formation must have caused a signi?cant reduction in oceanic salinity at the time (Kendall and Harwood, 1996). Examples of intracontinental lacustrine deposits include the dry lakes of Chile, Death Valley in California, and the Great Salt Lake of Utah. In addition to the products mentioned above, these deposits are also important producers of borates (from California) and nitrates (from Chile), as well as Mg, Br, and Li (from Utah). The principles behind the formation of evap- orite deposits are relatively simple in theory, although actual deposits can be complex and exhibit a large variety of different mineral salts and paragenetic sequences. As sea water or brine evaporates and water vapor is removed into the atmosphere, the salinity (i.e. the total content of dissolved salts in solution) of the residual solution increases and individual salts precipitate as their respective solubility limits are reached. The order or sequence of precipitation re?ects the scale of increasing solubility at a given temper- ature, such that the salts with the lower solubil- ities precipitate ?rst, followed consecutively by salts with progressively higher solubilities. The relative quantities of precipitated products in an evaporite deposit are also constrained by the solubility limits of the various salts dissolved in the brine solution. These considerations seem to suggest a relatively straightforward precipita- tion process and uniform paragenetic sequence. In fact the chemistry of brine solutions is very com- plex and factors such as convection dynamics, post-precipitation diagenesis, and equilibrium as opposed to fractional crystallization combine to ensure considerable diversity in the nature of these deposits. ITOC05 09/03/2009 14:34 Page 285Marine evaporites that formed from sea water that was well mixed and had a constant composi- tion over much of geological time tend to be dom- inated by the same assemblage of major mineral precipitates, namely halite (NaCl), gypsum/ anhydrite (CaSO 4 /CaSO 4 .2H 2 O), and sylvite (KCl). Evaporite deposits can, however, also contain minor accumulations of other salts, especially those representing the late stage precipitation of high solubility compounds. The latter typically comprise hydrated, multi-element, Mg, Br, Sr, K chlorides and borates, and are called bitterns. The preservation of bitterns in evaporite deposits appears to be controlled largely by geological fac- tors. Figure 5.25a shows a schematic cross section through a shallow, marginal marine setting in 286 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Open ocean 1.05–1.1 1.1–1.2 (a) 1.2–1.3 NaCl Algal mound CaSO 4 KCl and bitterns CaCO 3 Land Sea water density (in g cm –3 ) increases Density of sea water Volume (liters) 1000 1.0 400 200 500 1.4 1.1 1.2 1.3 900 800 700 600 300 100 First Last Sequence of precipitation Bitterns Volume of seawater Mineral precipitate Calcite (CaCO 3 ) Gypsum/anhydrite (CaSO 4 /CaSO 4 .2H 2 O) Halite (NaCl) Epsomite (MgSO 4 .7H 2 O) Kainite (KMgClSO 4 .3H 2 O) Sylvite (KCl) Carnallite (KMgCl 3 .6H 2 O) Borates and celestite (SrSO 4 ) g cm –3 (b) 0.11 1.20 26.90 0.03 0.09 2.25 0.70 3.18 34.48 0.01 0.02 CaCO 3 CaSO 4 NaCl NaBr MgBr 2 MgSO 4 KCl MgCl 2 borates, sulfates Sea water g kg –1 Total Figure 5.25 (a) Schematic cross section showing the important features necessary for the formation of large marine evaporite sequences. (b) Paragenetic sequence for an evaporite assemblage from typical sea water containing the ingredients shown in the left hand column. The amount of sea water (per 1000 liter volume) that has to evaporate in order to consecutively precipitate the observed sequence of mineral salts is shown by the curve adjacent to the paragenetic sequence (diagrams modi?ed after Guilbert and Park, 1986). ITOC05 09/03/2009 14:34 Page 286SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 287 which an evaporite sequence is being deposited. Sea water recharge ensures replenishment of the solution and sustained mineral precipitation from the brine. Intense evaporation results in marked density strati?cation and accumulation of highly saline, dense brines in the bottom waters. In environments where there is a constriction at the entrance of the basin, there is little opportunity for this dense brine to discharge back into the deep parts of the open ocean and, hence, the high solubility components of the solution remain within the basin to eventually precipitate out as bitterns once the solubility limits are exceeded. Bitterns do not accumulate if re?ux of the dense bottom water brines to the open ocean occurs because the high solubility components of the solution are then ?ushed out of the basin. In the latter case, if the bitterns do not precipitate, the evaporite mineral assemblage may be trunc- ated or incomplete and the deposit less diverse in terms of the commodities it could exploit. Figure 5.25b illustrates the typical mineral paragenetic sequence for evaporite deposits as a function of precipitation from sea water, the composition of which is shown on the left hand side of the ?gure. The proportion of sea water remaining after evap- oration (as a function of a 1000 liter volume) and its increasing density are also shown in relation to the precipitation sequence. The formation of the giant Phanerozoic saline deposits is a contentious issue, mainly because modern analogues do not exist and ancient evaporite sequences are generally altered and tectonically modi?ed. The evaporite deposits on the ?oor of the Mediterranean Sea are now believed to have formed in a sabkha environment, implying that much of the Mediterranean basin was subaerial and desiccated during late Miocene times (Kendall and Harwood, 1996). Evaporitic minerals which form in a sabkha environment precipitate from groundwater brines that evapor- ate in the capillary zone immediately above the water table. Recharge of the brine occurs either by periodic ?ooding or by sea water seepage into the sabkha basin. A good example of a complete evap- orate deposit formed as part of a Phanerozoic saline giant is the Boulby Mine in the Zechstein of northeast England. 5.4 FOSSIL FUELS – OIL/GAS FORMATION AND COALIFICATION A fossil fuel is formed from the altered remains of organic matter (plant and animal) trapped in sedi- mentary rock. The hydrocarbon compounds that form during the burial and degradation of organic material retain a signi?cant proportion of the chemical energy imparted into the original living organism by the sun. This energy is harnessed in many different ways by burning or combust- ing the fuel. The main fossil fuels are coal and petroleum, the latter being a general term that encompasses crude oil and natural gas (mainly methane), as well as solid hydrocarbons. Although most ore deposit books do not include descriptions of fossil fuel occurrences, it is felt that any text dealing with ore-forming processes would be incomplete without a brief discussion on the accumulation of hydrocarbons in the Earth’s crust. There is an enormous volume of published literature on the origin and nature of fossil fuel deposits and, consequently, the follow- ing section presents only brief overviews relevant to oil/gas formation and coali?cation processes. Mention is also made of tar sands, oil shales, and gas-hydrates, all of which represent potential fuel resources in the future. 5.4.1 Basic principles The global carbon cycle is complex, involving both circulation, mainly in the form of gaseous CO 2 and CH 4 , and storage, as reduced carbon or carbonate minerals, in different reservoirs such as the deep ocean and sedimentary rocks. Organic processes (mainly photosynthesis) ensure that a small proportion of the carbon in the global cycle is combined with hydrogen and oxygen to form the molecular building blocks of the various biota that inhabit the Earth’s surface. The areas of greatest organic productivity on Earth are repres- ented on the continents by tropical vegetative zones, and in the oceans by areas of cold current upwelling. As a ?rst approximation, therefore, it is logical to expect that fossil fuels are most likely to have formed in these two environments. Indeed, coals develop mainly in continental ITOC05 09/03/2009 14:34 Page 287settings, from accumulation of vegetation in either humid, tropical swamp environments or, as in the case of younger coal seams, more temperate to polar regions. Oil and gas deposits, by contrast, develop mainly from the accumulation of phyto- and zooplankton in marine settings. When organisms die they decompose by bac- terial decay and/or oxidation, and rapidly break down to relatively simple molecular constituents such as CO 2 and CH 4 . Such a process would not be conducive to the formation of fossil fuels, which requires preservation of much of the complex organic hydrocarbon material. Suitable source rocks from which fossil fuels accumulate must, therefore, have formed in reducing environments where sedimentation rates are neither too slow (in which case oxidation and molecular break- down might occur) nor too rapid (in which case dilution of the total organic matter content of the rock would occur). The prevailing theory for the origin of both petroleum and coal is based on the notion that organisms are buried during the sedimentary process and then subjected to a series of alteration stages as pressure and temperature increase. As shown in Figure 5.26, progressive burial of phyto- plankton results in the early liberation of CO 2 and H 2 O and formation of kerogen (a collective name for sedimentary organic matter that is not soluble in organic solvents and has a polymer like struc- ture; Bjørlykke, 1989). As the kerogen “matures” the long-chain covalent bonds that typify organic molecules are progressively broken to form lower molecular weight compounds. At around 100– 120 °C and burial depths of 3–4 km (depending on the geothermal gradient), a liquid hydrocarbon fraction develops which can then migrate from the source rock. This interval is known as the “oil window.” With further burial and cracking of molecular bonds, signi?cant volumes of gas (mainly methane, CH 4 ) develop, which are also amenable to migration. The solid residue remain- ing in the sediment is referred to as kerogen, but with progressive burial it devolatilizes further and has a composition approaching pure carbon (graphite). By contrast, when humic, land-derived vegetation is buried, little or no liquid oil is formed, although signi?cant volumes of gas will, again, be generated. The solid residue that remains in this case is more voluminous and com- pacts to form coal seams. The nature of the solid coal residue changes with depth, from peat and lignite at shallow burial depths (less than about 500m), to bituminous coal and subsequently 288 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Mainly land- derived plants 1 5 Depth (km) 3 4 2 Gas 30 150 90 120 60 Kerogen Sub-bituminous coal Mainly phytoplankton T (°C) Oil Peat Lignite Bituminous coal Anthracite Diagenesis Catagenesis Metamorphism Figure 5.26 Simpli?ed scheme illustrating the formation of oil, gas, and coal by the progressive burial of different types of mainly vegetative matter (after Hunt, 1979). ITOC05 09/03/2009 14:34 Page 288SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 289 anthracitic coal (at depths of around 5000m). The calori?c value (i.e. the amount of energy pro- duced on combustion) of the coal increases with maturity or burial depth. 5.4.2 Oil and gas formation All living organisms are made up of relatively few molecular building blocks that have apparently changed little over geologic time. The main molecular building blocks involved in petroleum formation include: • Carbohydrates and related substances – these are mono- and polysaccharide polymers and in- clude the sugars, chitin and lignin, the latter being a major precursor to coal. • Proteins – these are high molecular weight amino acid polymers and represent one of the most important constituents of life processes. • Lipids, which are represented in animal fats and vegetable oils and are abundant in crude oils. • Other substances such as resins and pigments are of lesser importance but also occur variably in both plant and animal matter. Plants comprise mainly carbohydrates (40–70%), although the higher, woody forms also contain sub- stantial lignin for strength. Phytoplankton contains around 20% protein. Animals, by contrast, are made up mainly of proteins (55–70%) with lesser carbohydrates and lipids but no lignin (Hunt, 1979). The average chemical composition of these vari- ous organic building blocks is shown in Table 5.1. As the original organic constituents are buried, they are subjected to increases in pressure and temperature, resulting in systematic changes which are divided into three stages, termed diagenesis, catagenesis, and metamorphism. This progression, and its application to the burial of organic matter in sediments, together with the organic processes and changes that occur in each, are summarized in Figure 5.27. Diagenesis refers to the early biological and chemical changes that occur in organic-rich sediments prior to the pro- nounced temperature-dependent effects of later reactions (i.e. less than about 50 °C). During this stage the biopolymers of living organisms undergo a wide variety of complex low-temperature reac- tions. Put simply, they are either oxidized or attacked by microbes and converted into less complex molecules which may, in turn, react and condense to form more complex, high molecular weight geopolymers that are the precursors to kerogen. Biogenic reactions during diagenesis also produce signi?cant amounts of gas, termed biogenic gas, which typically escapes into the atmosphere or water column and is not retained in the sediment. Catagenesis occurs between about 50 and 150 °C and is the most import- ant stage as far as petroleum generation is con- cerned. During this stage temperature plays an important role in catalyzing reactions, the major- ity of which result in the formation of light hydro- carbons from high molecular weight kerogen by the breaking of carbon–carbon bonds. In this process, known as thermal cracking, a complex organic molecule such as a paraf?n or alkane will split and form two smaller molecules (a different paraf?n and an alkene or ole?n). Each of the pro- ducts contains a carbon atom, with an outer orbital electron from the original covalent pairing that is now shared. The shared electron means that the resultant alkane and alkene each contain a “free radical” that makes them reactive and amenable to further breakdown. Alternatively, reactions may occur by catalytic cracking, where a carbon atom in one of the resultant molecules takes both electrons, thereby becoming a “Lewis acid” or double electron acceptor. The molecule that loses the electron pair is called a carbonium ion and with its net positive charge is also amenable to further decomposition. For example, alkanes subjected to catalytic cracking could Table 5.1 Average chemical compositions (in wt%) of the main organic building blocks compared to those of petroleum and a typical coal CHONS Carbohydrates 44 6 50 – – Lignin 63 5 31.6 0.3 0.1 Proteins 53 7 22 17 1 Lipids 76 12 12 – – Petroleum 85 13 0.5 0.5 1 Coal 70 5 23 1 1 Source: after Hunt (1979). ITOC05 09/03/2009 14:34 Page 289yield gaseous products such as ethane or butane (Hunt, 1979). Catalytic cracking tends to be the dominant process in petroleum generation up to about 120 °C, but at higher temperatures thermal cracking becomes increasingly important. The integrated effects of time and temperature have important implications for petroleum gener- ation. Temperature and time are inversely related in terms of petroleum productivity so that, for example, a given quantity of oil formed at 110 °C over 25 million years would require 100 million years to form at 90 °C (Hunt, 1979). This relation- ship is quanti?ed on a plot of time versus temper- ature in Figure 5.28, where data for selected petroleum-bearing basins are also shown. The graph can be very useful in petroleum exploration as it predicts, for example, that oil would not be formed in a young basin with low geothermal gradient because the threshold for ef?cient hydro- carbon generation might not yet have been reached. Likewise, little oil could be preserved in an old, hot basin as it would all have been destroyed dur- ing metamorphism. The time–temperature evolu- tion of sedimentary basins is, therefore, a critical factor in petroleum exploration. As a general rule, oil resources are sought either in young, hot 290 PART 3 SEDIMENTARY/SURFICIAL PROCESSES PROCESSES BIOPOLYMERS Carbohydrates, proteins, lipids, lignin MICROBIAL ALTERATION, HYDROLYSIS Sugars, amino acids, fatty acids, phenols BIOMONOMERS CONDENSATION, REDUCTION, POLYMERIZATION Nitrogenous and humic complexes GEOPOLYMERS THERMOCATALYTIC CRACKING MATURATION RANGE DIAGENESIS 50 °C CATAGENESIS 150 °C GEOMONOMERS Petroleum hydrocarbons and low molecular weight organic compounds THERMAL CRACKING METAMORPHISM Gas and pyrobitumens ORGANIC MATTER Figure 5.27 Summary of the stages and processes involved in the transformation of organic matter during burial to form oil and gas (after Hunt, 1979). ITOC05 09/03/2009 14:34 Page 290SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 291 basins or in old, cold ones, as identi?ed by the rel- evant ?eld in Figure 5.28. Source considerations Most of the organic matter on Earth can be classi?ed into two major types, namely sapro- pelic, which refers to the decomposition products of microscopic plants such as phytoplankton, and humic, which refers essentially to the maturation products of macroscopic land plants. Sapropelic organic matter has H/C ratios in the range 1.3–1.7, whereas humic matter has a lower H/C ratio around 0.9. These compositional differences have led to a more rigorous classi?cation of kerogen types that has relevance to the generation of fossil fuels. The classi?cation was originally devised by Tissot et al. (1974) in terms of the so-called Van Krevelen diagram, a plot of H/C against O/C (see inset in Figure 5.29). A more recent version of the Tissot classi?cation is described in Cornford (1998) using a plot of hydro- gen index versus oxygen index (Figure 5.29), two parameters which are analogous to H/C and O/C, respectively, but can be obtained directly by spec- trometric analysis. The classi?cation scheme recognizes three main kerogen types, with a fourth type having little or no relevance to petroleum generation. Type I kerogen is derived from algal material and, on maturation, yields mainly oil. Type II kerogen is obtained from the breakdown of plant spores, pollens, exines, and resins, but it may also com- prise bacterially degraded algal (Type I) organic matter. The Jurassic source rocks of the North Sea oil deposits represent an example of hydrocarbon derivation from the Type II kerogen. Type III kerogen is derived from humic land plants rich in lignin and cellulosic tissue and on maturation yields only gas. As mentioned previously, this is the material from which coals are formed. Type IV kerogen is obtained from oxidized woody material and has no hydrocarbon potential. The arrows in Figure 5.29 indicate the compositional trends of different kerogens with progressive burial and maturation, and all coalesce toward the origin, represented by pure carbon (graphite). In considerations of source it should be noted that Time–temperature limits for petroleum prospective basins 600 40 Time (Myr) 20 Temperature (°C) 150 200 300 400 20 60 80 110 140 180 230 280 100 60 40 1 2 3 8 4 5 6 7 Main oil and gas zone Oil phase out Gas phase out Gas Oil Amazon Basin, Brazil 1 Paris Basin, France 2 Rio de Oro, West Africa 3 Douala Basin, Cameroon 4 Camargue Basin, France 5 Offshore Taranaki Basin, New Zealand 6 Los Angeles Basin, California 7 Aquitaine Basin, France 8 No oil Figure 5.28 Graph showing the relationships between temperature and time with respect to the generation of oil and gas in sedimentary basins. The shaded area refers to the optimum range of conditions for petroleum generation and the points are actual examples of petroleum-bearing basins from several locations around the world (after Connan, 1974). ITOC05 09/03/2009 14:34 Page 291oil-prone Type I and II kerogens will yield not only oil, but also both “wet” (i.e. with a high con- densable fraction) and “dry” gas (mainly methane). Gas-prone kerogen (usually Type III) will produce mainly dry gas. Modern analytical and modeling techniques allow the quantitative estimation of gas and oil yields to be calculated from a source rock undergoing progressive burial. A summary of hydrocarbon yields from two different source rocks is shown in Figure 5.30. Figure 5.30a shows the quantities of petroleum products derived from a source rock comprising Type II kerogen and dif- ferentiates between the condensate, comprising high (C15+) and low (C2–C8) molecular weight oils, and a dominantly methane gas component. Figure 5.30b is the equivalent scenario for a source rock sediment comprising kerogen that is only gas-prone, which on maturation initially yields only a minor amount of wet gas (ethane, propane, butane, etc.) but mainly dry gas (methane), leaving a coal residue. The latter process is relevant to the generation of coal-bed methane resources. Source rock considerations are important dur- ing petroleum exploration. The origin of organic matter is obviously a pointer to the depositional environment of the sedimentary rock and, there- 292 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 1000 0 0 Hydrogen index 400 200 600 50 100 150 Oxygen index 800 2.0 0 H/C 1.5 1.0 0.2 0.3 0.1 O/C 0.5 III II I Principal products of kerogen evolution CO 2 , H 2 O Oil Gas Type IV Land plants: cellulosic tissue, wood Type II Type I Algae (plankton) Bacterially degraded algae or mixtures Resin cuticle, spores Waxy oil and condensate prone Gas- prone “Dead carbon” No potential Oil- prone Type III Diagenesis Catagenesis Altered (oxidized) wood Figure 5.29 Classi?cation scheme for kerogen types based on the hydrogen index and oxygen index (after Cornford, 1998). Inset diagram shows the original classi?cation of kerogen on the basis of H/C and O/C atomic ratios (after Tissot and Welte, 1984). ITOC05 09/03/2009 14:34 Page 292SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 293 fore, also useful in placing constraints on the nature of oil/gas migration and entrapment. Petroleum migration and reservoir considerations One of the features of oil and gas formation is the fact that petroleum originates in a ?ne grained source rock and then migrates into more permeable, coarser grained, reservoir sediments. Knowledge of petroleum migration and entrap- ment processes is obviously important to both the exploration and exploitation stages of the indus- try. Although the processes are complex, they can now be accurately evaluated in terms of ?uid hydraulic principles and sophisticated computer modeling techniques. Petroleum engineers differentiate between primary migration, which accounts for the move- ment of oil and gas out of the source rock during compaction, and secondary migration, which describes ?ow within the permeable reservoir, as well as the segregation of oil and gas. As the organic source rock is lithi?ed during diagenesis, water (usually a low to moderate salinity brine) is expelled from the sediment to form a connate ?uid (see Chapter 3). In the early catagenic stages of hydrocarbon maturation, therefore, oil and gas 0 200 Temperature (°C) 100 150 50 4 Million m 3 oil/km 3 rock 6 8 10 12 4 682 (a) (b) Billion m 3 CH 4 /km 3 rock 300 200 100 Billion m 3 CH 4 /km 3 seam coal Early biogenic gas Early mature oils from well drained source Main phase of oil generation from thick source rock C 8 –C 15 CH 4 C 2 –C 8 Unexpelled oil cracks to gas C 15+ Early, wetter gas expulsion from thin coal seams Later expulsion of dry gas from thicker coal CH 4 Biogenic gas 0 1 2 3 4 5 Depth (km) Figure 5.30 Maturity curves showing (a) the cumulative generation of oil types and methane gas from sediment source rock originally containing a 1% total organic carbon (TOC) content of Type II oil-prone kerogen; and (b) the generation of wet and dry gas only from a 1 km 3 sediment volume of gas-prone kerogen which ultimately yields a coal deposit (after Cornford, 1998). ITOC05 09/03/2009 14:34 Page 293migrate in the presence of water. The actual mechanisms of hydrocarbon migration and the role of water in this process are not completely understood. Oil is unlikely to migrate in aqueous solution since hydrocarbons generally have low solubilities in water (e.g. 3ppm for pentane, 24 ppm for methane, 1800 ppm for benzene, all at room temperature; Bjørlykke, 1989). However, small hydrocarbon molecules dissolve more readily than large ones and methane (CH 4 ), for example, does become fairly soluble in water as pressure increases (about 7500 ppm at 6000 m depth; Hunt, 1979). Methane and other low molecular weight compounds can, therefore, be dissolved at depth and then released as the aqueous solution rises to the surface. Major amounts of hydrocarbon liquids and high molecular weight compounds, however, are unlikely to ?ow by this mechanism. At cata- genic pressures and temperatures, hydrocarbon gases themselves will increasingly dissolve oils and in some source rocks there is evidence that prim- ary migration is achieved by gas phase dissolution. Although dissolution mechanisms are likely to play a role in petroleum ?ow, they cannot account for the huge oil accumulations observed in the major oil producing regions of the world. Two other mechanisms are considered to be more important during the primary ?ow of hydrocar- bons and these are oil phase migration and diffu- sion. The movement of crude oil out of the source rock is initiated once compaction has driven off much of the bulk pore water in the sediment. Initially a source rock will usually contain a relat- ively high proportion of pore water relative to oil and in such a case the hydrocarbon will be unable to move. This is mainly because the ?ow of oil is impeded by the presence of water which, because of its lower surface tension, tends to “water-wet” the sediment pore spaces. In such a situation, oil globules ?oating in the water-dominated pore spaces cannot generate the capillary forces required to initiate migration (Figure 5.31a). As compaction progresses, however, the bulk of the pore water is driven off and the little remaining water is struc- turally bound on clay particles. Oil now occupies enough of the pore volume to be subjected to capillary forces and start ?owing through the rock (Figure 5.31b). Once “oil-saturated” pathways are established in the rock, oil migration becomes feasible because permeability is not impeded by the presence of water. In the less common situ- ation when pore spaces are initially “oil-wet” because of very high initial organic/hydrocarbon contents, then porosity reduction is not necessary to invoke the unimpeded ?ow of oil through the source rock. Another way in which hydrocarbon molecules could migrate in the primary environment is by diffusion along an activity or free energy gradi- ent (Hunt, 1979). Figure 5.31c illustrates, at a nanometer scale, the arrangement of structured water tetrahedra along the edge of a clay particle. Hydrocarbon molecules or aggregates in the sedi- ment pore space will typically be encased by a jacket of water molecules to form hydrate or clathrate compounds. Interaction of the hydro- carbon clathrate with structured water would, from a thermodynamic viewpoint, require energy to facilitate the breakdown of one or both molecules. A diffusional energy gradient is, there- fore, set up that promotes migration of the hydro- carbon clathrate towards lower free energy and, therefore, away from the structured water clay mineral edge. This type of diffusion translates, at a geological scale, to movement of hydrocarbons from ?ner grained shales towards coarser sands, which accords with natural observations. Although not directly involved as the transporting agent in primary migration, it is evident that water is implicated in all the mechanisms involved. In the case of low organic content source rocks generating mainly gas, hydrocarbon migration would occur by diffusion, in aqueous solution, or directly as the gas phase. In high organic content, oil-prone, source rocks, however, hydrocarbons will migrate predominantly as the oil phase. Intermediate situations might well exhibit the entire range of migration processes (Hunt, 1979). At a geological scale, where oil and gas are migrating over tens to hundreds of kilometers, the main factor that controls hydrocarbon migra- tion pathways is the existence of a pressure differ- ential in the host rock. Particularly relevant to petroleum migration is the concept of ?uid over- pressure, described in more detail in section 3.3.2 of Chapter 3. 294 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:34 Page 294SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 295 Bulk pore water (a) Water flows out of source rocks Oil globule prohibited from moving Sediment particle Oil Structured pore water Oil flows through to reservoir rocks (b) Structured pore water (c) Bulk pore water Migration of hydrocarbon clathrate along a diffusional gradient Water molecule Hydrocarbon complex Clathrate Sediment particle (clay) Figure 5.31 Explanation of oil-phase primary migration. (a) “Water saturated” pore spaces in a sedimentary source rock in which little or no oil migration takes place because oil capillarity is impeded by the presence of water-wet pores which block the pore throats. (b) “Oil saturated” pore spaces formed after compaction of the sediment has driven off much of the bulk pore water, which facilitates capillary-related oil migration by removing pore throat blocking free water (after Bjørlykke, 1989). (c) Simpli?ed arrangement, at a nanometer scale, of a hypothetical pore space and the setting in which diffusional migration of hydrocarbons takes place away from structural water-bound mineral grains (after Hunt, 1979). ITOC05 09/03/2009 14:34 Page 295If the removal of pore water is impeded by low permeability in, for example a ?ne grained shale, then compaction will be retarded and ?uid pres- sures increased to values above normal hydro- static pressures. Fine and coarse sediment will expel pore waters at different rates during burial and will, therefore, compact along different pres- sure gradients. Overpressured ?uids will tend to exist in the ?ner grained sediments. Figure 5.32 shows actual measurements of ?uid pressures in a series of drill wells from the Orinoco Basin and the hydrostatic overpressures that occur in the poorly drained shale units relative to the well drained sandstone. This information allows the migration pathways to be identi?ed and shows that the main sandstone aquifer can be fed by ?uids from both above and below. Several factors actually control the stresses that cause hydrostatic ?uid overpressures in rocks and these include rapid sediment deposition, thermal expansion of ?uids, tectonic compression, and the actual gen- eration of low density oil and gas in the source rock. Many of the factors discussed above also apply to secondary petroleum migration. In the advanced stages of catagenesis, however, temperatures and pressures are such that many hydrocarbons exist close to their critical points such that density con- trasts between liquid and vapor are minimized. The segregation of oil from gas, therefore, mainly re?ects the decline in temperatures that accom- panies migration of petroleum products into reservoirs that are well away from the sites of oil/gas generation. The progressive separation of liquid and vapor, into a low density, buoyant gas phase and a more dense, viscous oil, is a process that generally accompanies secondary migration. Entrapment of oil and gas Primary migration of hydrocarbons typically occurs over short distances (hundreds of meters or less) and is constrained by the proximity of the ?rst available aquifer to the source rock. Secondary migration, by contrast, occurs over tens, and possibly even hundreds, of kilometers and is only constrained by the presence of a trap which prevents the petroleum from further ?ow and allows it to accumulate. Hydrocarbon traps are critical to the formation of viable oil and 296 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 0 220 440 220 440 0 220 0 Excess hydrostatic pressure (KPa m –1 ) Sandstone Sandstone Shale Shale Orinoco basin Figure 5.32 Actual measurements of excess hydrostatic pressures (i.e. overpressure) in three drill wells in the Orinoco basin and the anticipated ?uid ?ow lines in the sequence of alternating shale and sandstone (after Hunt, 1979). ITOC05 09/03/2009 14:34 Page 296SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 297 gas deposits and take the form of any geological feature that either reduces permeability of the reservoir or provides a physical barrier that im- pedes the further migration of ?uid. Evaporite layers, for example, may be good hydrocarbon traps as they are laterally extensive and have vir- tually zero permeability because of the ability of halite to ?ow plastically at elevated temperat- ures. Many of the giant oil ?elds of the Arabian (Persian) Gulf are capped by evaporite sequences (Box 5.3), as are some of the North Sea gas deposits. In general, however, structural features such as faults and anticlines tend to be the dominant hydrocarbon traps. This section brie?y outlines some of the geological scenarios that represent potential traps for hydrocarbon deposits in vari- ous parts of the world. Figure 5.33 shows that there are basically three categories of trap site, namely stratigraphic, structural, and a less common miscellaneous class that includes hydrodynamic and asphalt traps. Stratigraphic traps arise entirely from sedi- mentological features and can be represented by features such as sediment pinch-out zones, where permeability is reduced as one sediment facies grades into another, or unconformities where the reservoir rock sub-outcrops against a unit of reduced ?ow capability (Figure 5.33a). The Athabasca tar sands (see section 5.4.4 below) are believed to have resulted from migration and entrapment of hydrocarbons along a major un- conformity in the early Paleogene period. In addition to sandstones, carbonate rocks represent important reservoirs because the high solubil- ity of minerals such as calcite promotes sec- ondary porosity and migration of hydrocarbons. Sedimentary features such as limestone “pinnacle” reefs (Figure 5.33a), are important trap sites in car- bonate sequences, especially in the giant Middle East deposits. Structural traps result from some form of sediment deformation and generally provide phys- ical barriers which prevent the continuation of ?uid ?ow along an aquifer. A fault, for example, might juxtapose a reservoir sediment against a shaley unit and, as long as the fault itself remains impervious, will act as a barrage behind which hydrocarbons will accumulate. Since most hydro- carbons ?ow up-dip, the folding of a reservoir rock into an anticline or dome-like feature also pro- vides a very ef?cient trap site. The huge oil?eld at Prudhoe Bay in Alaska represents an example where petroleum is trapped by a combination of both folding and faulting of the reservoir sedi- ments. The scale of hydrocarbon accumulation in this setting can be estimated by considering the closure volume of the structure and also the spill point, beyond which oil will continue its migra- tion into another aquifer system (Figure 5.33b). Other important structurally related trap sites are linked to the deformation that accompanies salt diapirism. When salt beds underlie more dense rock strata the resultant gravitational in- stability causes the diapir to pierce the overlying beds and terminate reservoir sediments creating ideal trap sites for hydrocarbons (Figure 5.33b). Such trap sites are of great importance in many parts of the Arabian Gulf. Other trap sites of lesser importance include sites where oil migration is diverted by a strong ?ow of groundwater which, as explained previously, will water-wet the sedi- ment pore spaces and impede the ?ow of hydro- carbons. These are known as hydrodynamic traps and result from the shear stresses set up at the oil–water contact in aquifers where strong water ?ow is occurring. Groundwater will also interact chemically with reservoir oil, resulting in oxida- tion, biodegradation, and the formation of a bitu- minous asphalt layer at the interface. This layer, known as an asphalt trap, will itself act as a barrier beneath which hydrocarbons may accu- mulate (Figure 5.33c). 5.4.3 Coali?cation processes The majority of the world’s energy requirements is still obtained from the burning of coal and lignite. The two largest producers of coal in the world are China and the USA, although Australia ranks top as an exporter of high quality coal. As mentioned previously, most coal is derived from Type III kerogen and generally represents the in situ accumulation of land-derived vegetation subjected to alteration and compaction. Figure 5.26 shows that coal is derived from a peat precursor and, with burial or coali?cation, is progressively ITOC05 09/03/2009 14:34 Page 297Shaley limestone Limestone (a) Stratigraphic traps Shale Pinch out Unconformity Sandstone (i) Oil (ii) Carbonate “pinnacle” reef Dolomite Calcareous shale (b) Structural traps (i) Fault Shale Sandstone (ii) Diapir Salt diapir Shale Anticline (iii) Oil Closure volume Spill point Hydrodynamic (i) Hydrostatic head Water Asphalt (ii) Biodegraded oil/asphalt Meteoric water (c) Other traps Reservoir rock Figure 5.33 Geological scenarios for hydrocarbon trap sites. (a) Stratigraphic traps represented by unconformities, pinch-outs, and carbonate “pinnacle” reefs. (b) Structural traps represented by faults, diapiric features, and anticlinal or dome like structures. (c) Other features, such as hydrodynamic and asphalt traps (after Hunt, 1979; Bjørlykke, 1989). 298 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:35 Page 298SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 299 The Arabian (or Persian) Gulf Basin contains well over half of the conventionally recoverable oil reserves of the world, and also huge reserves of natural gas. This extra- ordinarily rich basin extends over a length of more than 2000 km from Oman in the south to Syria and southeast Turkey in the north (Figure 1). Production has been derived from some 250 reservoirs in the region, of which more than 80% are in Jurassic–Cretaceous sediments. The Middle East is unique for the size of its individual deposits, with 14 ?elds having recoverable reserves in excess of 10 billion barrels (Shannon and Naylor, 1989). These include such giants as the Ghawar ?eld in Saudi Arabia and Burgan ?eld in Kuwait (both with more than 70 billion barrels of oil) and the Rumaila and Kirkuk ?elds in Iraq, each with around 15–20 billion barrels of oil (Tiratsoo, 1984). Many of the oil ?elds in the Persian Gulf are also associated with huge gas resources. In addition, however, some very large unassociated gas ?elds also occur in the region, such as the North Dome ?eld off Qatar. These gas deposits occur in the Permian Khuff limestone and are, therefore, older and underlie the main oil deposits. Unassociated gas deposits in many other parts of the world appear to be related to coal measures of Carboniferous age. There are several reasons why the Middle East is so well endowed in hydrocarbon resources and these are brie?y discussed below. The oil ?elds of the Arabian Gulf are located in late Mesozoic to early Paleogene reservoir sediments that were deposited on a continental shelf forming off the north- eastern margin of the Arabian portion of Gondwana. A signi?cant proportion of these sediments are limestones, indicating deposition at equatorial paleolatitudes in the Tethyan seaway (North, 1985). As Pangea fragmented the Arabian land mass became part of a downgoing slab involved in subduction to the north and east. Fortunately, subduction ceased in Pliocene times with the formation of the Zagros mountains in present day Iran, and this hydrocarbon-rich basin was preserved largely intact. It is likely that other parts of the Tethyan seaway, probably equally well endowed, were consumed by subduction in Cenozoic times, perhaps accounting for the fact that the western and eastern extremities of the seaway contain lesser oil and gas deposits. The tectonic and sedimentological regimes active in this environment produced at least three major source rock sediments in the Arabian foreland, along the axis of the present day basin and in the orogenic belt of Iran. Source rocks are all marine in origin and developed over the entire extent of the Arabian platform. Oil and gas is hosted in a variety of different lithologies and the principal reservoirs become progressively younger from southwest to north- east. The thick carbonate and sandstone reservoirs that Fossil fuels: oil and gas – the Arabian (Persian) Gulf, Middle East TURKEY Kirkuk IRAQ JORDAN ISRAEL IRAN QATAR IRAN KUWAIT OMAN Rumaila Burgan Kharg/Darius North dome Divided zone Ghawar UAE Maghrib 200 0 km 35° 35° LEBANON 25° 45° 55° 35° 45° 25° 55° N Ahwaz Dukhan Qatif SYRIA SAUDI ARABIA Figure 1 Location and distribution of some of the major oil and gas ?elds in the Arabian Gulf (after Shannon and Naylor, 1989). ITOC05 09/03/2009 14:35 Page 299300 PART 3 SEDIMENTARY/SURFICIAL PROCESSES (a) (b) Mixed carbonate and shale shelf Intrashelf shale basin Shallow carbonate shelf basin N Continental slope Zagros crush zone Deep marine 0 200 km 0 200 km Shallow carbonate shelf Zagros crush zone Mixed carbonate and clastic shelf Shallow clastic shelf Deep clastic shelf Deep carbonate shelf N Lower coastal plain Alluvial plain Mid-Jurassic Mid-Cretaceous Figure 2 Interpretations of the evolving sedimentary environment in the region of the Arabian Gulf during (a) mid-Jurassic and (b) mid-Cretaceous times (after Shannon and Naylor, 1989). formed in the Persian Gulf Basin are also characterized by a remarkable lack of intergranular cements and a high de- gree of primary porosity, making them very effective for the migration and ultimate concentration of hydrocarbons. This also contributes to very good recovery rates and the fact that very few wells need to be be drilled in order to exploit a given ?eld. Periodic uplift and subaerial expo- sure also resulted in the deposition of numerous evaporitic horizons that make good seals for oil and gas entrapment. Sedimentation in the Jurassic and Cretaceous periods was characterized in general by a marine transgression over the Arabian platform. Initially widespread carbonate sed- imentation occurred and this was followed by increasing aridity and evaporate formation. In detail, sea level ?uctu- ations and climate changes gave rise to complex inter?nger- ing of lithologies (Shannon and Naylor, 1989), as illustrated in the environmental interpretations for the region during mid-Jurassic and mid-Cretaceous times (Figure 2). The structural development of the region was also conducive to the formation of very large and effective trap sites. Three types of structural traps are recognized in the Arabian Gulf Basin. These include large, but gently dipping, anticlinal warps which caused oil entrapment mainly in Jurassic limestones (e.g. Ghawar in Saudi Arabia); diapiric salt dome structures arising from base- ment evaporitic sequences which deform Cretaceous and Jurassic limestones (e.g. Burgan in Kuwait); and steeply dipping anticlines in the foothills of the Pliocene moun- tain building episode in Iraq and Iran (e.g. Kirkuk in Iraq). The presence of thick evaporitic sequences is considered to be one of the most important features of the region because the salt horizons are believed to have absorbed much of the shortening that accompanied subduction to the north and east of the basin. This had the effect of preserving the large gently anticlinal structural traps that characterize the southwestern portions of the region and accounts for the huge sizes (up to 100 km long) of many of the oil ?elds (Shannon and Naylor, 1989). In addition to the presence of salt, some of the other geological features of the Middle East that have made it so productive for oil and gas include a long history of marine sedimentation at relatively low paleolatitudes, numerous transgressive and regressive sedimentary cycles resulting in a close association between source and reservoir rocks, and the occurrence of widespread seals, in the form of evaporitic sequences, that prevent leakage of the oil and gas. ITOC05 09/03/2009 14:35 Page 300transformed to a series of products ranging from lignite and sub-bituminous coal (known as “brown coals”), to bituminous coal and ?nally anthracite (the “hard coals”). The formation of peat is an essential initial step in the process and involves restrained biochemical degradation of plant mat- ter without rampant oxidative and bacterial decay. Coali?cation follows once peat is covered by overburden and subjected to an increase in tem- perature with progressive burial. Coal is a sedimentary rock that contains more than 50% by weight of carbonaceous material and can readily be burnt (McCabe, 1984). It forms by the compaction of peat in an environment rep- resented by a well vegetated land surface that is saturated with water. This environment is loosely termed a “swamp” but should not necessarily imply a tropical or equatorial climate as it is now well known that many coals have accumulated in cold climates at mid- to high latitudes. The great- est concentrations of peat are in fact accumulat- ing in Russia (with about 60% of the world’s resources; McCabe, 1984) at latitudes of around 50–70° N. High rainfall is not an environmental prerequisite, although abundant free-standing water promotes plant growth and also submerges dead vegetation, retarding the rate of decomposi- tion. Furthermore, swampy environments often contain waters that are both anaerobic and acidic, which promotes the preservation of organic ma- terial by minimizing oxidation and destroying bacteria. Tropical rain forests seldom form peat accumulations because the high temperatures promote rapid oxidation and decay of organic material. The best conditions for peat accumula- tion re?ect a balance between organic productiv- ity and decay, with the added prerequisite of low sediment input necessary to keep dilution of the peat by clastic material to a minimum. Coal characteristics Coal is a heterogeneous sedimentary rock that re?ects both the different sedimentological regimes within which peat forms and the varied vegeta- tion types from which it is derived. Peat deposits are known to have formed in many settings including lacustrine, ?uvial, deltaic, and beach- related (Galloway and Hobday, 1983; McCabe, 1984; Selley, 1988). Peat and coal deposits, there- fore, exhibit a wide range of shapes, and are also characterized by the facies variations expected of any sedimentary rock. Figure 5.34 illustrates the stratigraphic variations one might expect to see in peat, as a function of vegetation type and ?uctuat- ing water levels. SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 301 Marine Brackish Rhizophora Rhizophora/ Mariscus Fresh Mariscus Waterlily Waters Vegetation Marine trangression Peat stratigraphy Figure 5.34 Generalized scheme illustrating the nature of peat stratigraphy in terms of vegetation type and transgressive/regressive effects in a swamp environment (after Spackman et al., 1976). ITOC05 09/03/2009 14:35 Page 301The vegetation shown in Figure 5.34 applies speci?cally to the Everglades in Florida, where marine transgression results in lateral facies changes in swamp environment and deposition of different peats that originate from a variety of plant types. Coals that form by compaction of peat will also, therefore, exhibit marked variations in composition and vegetative make-up. This characteristic has led to a rigorous classi?cation of coals, the main elements of which are shown in Table 5.2. In the same way that a rock is made up of minerals, a coal is made up of “macerals,” which are components of plant tissue, or their degraded products, altered during compaction and diagenesis. The main maceral groups are vitrinite (also called huminite in brown coals), made up woody material (branches, roots, leaves, etc.), and exinite (also called liptinite), which comprises spores, cuticles, waxes, resins, and algae. A third grouping is also recognized, termed inertinite, which comprises mainly the remains of oxidized plant material and fungal remains. Each maceral group is further subdivided into individual macer- als, but listing of these is beyond the scope of this book and can be obtained in more detailed texts, such as Stach et al. (1982). Different combina- tions of maceral groups identify the coal litho- types (i.e. analogous to a rock type) and these are also listed in Table 5.2. The majority of coals are humic (i.e. made up of macroscopic plant parts), are peat derived, and can be subdivided into four lithotypes. These are vitrain, clarain, durain, and fusain, whose maceral compositions are shown in Table 5.2. The coal lithotypes are important in determining its prop- erties, in particular factors of economic import- ance such as calori?c value and liquefaction potential. In general, the highly re?ective, more combustive coals are made up of vitrain, whereas the duller, less combustive coals comprise essen- tially durains. It should be noted, however, that coal lithotypes should not be confused with coal rank, which separates low calori?c value brown coals from high heat capacity hard coals essen- tially on the basis of degree of compaction or burial depth. A small proportion of coals are sapropelic and formed largely from microscopic plant remains such as spores and algae (exinite). The sapropelic lithotypes, also referred to as cannel and boghead coals (Table 5.2), did not form from peat. They can be found overlying humic coals, forming in muddy environments as the swamp was ?ooded and buried, and occasionally are linked to the for- mation of oil shales (see section 5.4.4 below). In addition to the maceral-based classi?cation, coals can also be subdivided in terms of their chemical composition, speci?cally their carbon, hydrogen, and oxygen contents. Figure 5.35 is a ternary plot of normalized C, H, and O in which the compositional ?elds of peat and the coals are shown in relation to those for oil and oil shales. As peat is progressively compacted and coali?ca- tion commences, the carbon content increases noticeably, whereas hydrogen remains fairly con- stant and oxygen decreases. Hydrogen only starts to diminish noticeably once the hard coals start to develop. The chemical changes that accompany coali?cation are accompanied by the production of CO 2 and CH 4 gas. The source rocks for some of the gas ?elds in the North Sea are believed to be underlying Westphalian coal beds. 302 PART 3 SEDIMENTARY/SURFICIAL PROCESSES Table 5.2 Coal classi?cation in terms of maceral groups and lithotypes Maceral group Vitrinite (or huminite) Exinite (or liptinite) Inertinite Cell walls of vascular plants Spores, cuticles, waxes, resins and algae Oxidized and burnt (fusinite) plant material Vitrinite +< 20% exinite 5 Layers of all three maceral groups 4 6 Humic coals Mainly inertinite and exinite 4 Fusinite 7 Spore-dominated exinite # Sapropelic Algae-dominated exinite $ coals Lithotype Vitrain Clarain Durain Fusain Cannel coal Boghead coal ITOC05 09/03/2009 14:35 Page 302Another important property of coal, but one which is unrelated to its organic or maceral characteristics, is ash content. As peat forms it inevitably incorporates inorganic, clastic detritus from a variety of sources. During coali?cation this detritus is diagenetically altered to form a variety of authigenic mineral components such as quartz, carbonate, sul?de, and clay minerals. These components remain behind to make up its ash content once the coal is combusted or processed. Low ash content coals are obviously economically advantageous as this parameter translates favorably into higher calori?c values, better grindability, and less deleterious environ- mental effects such as sulfur emissions. A note concerning formation of economically viable coals The formation of economically viable coals is determined by processes active during both initial peat accumulation and later coali?cation. In the latter case bituminous coals and anthracite are the high rank products of thermal maturation and burial, and are characterized by the best calori?c values and lowest volatile contents. Progressive compaction will transform most humic coals to these products, which, if extensive and thick enough, will likely be economically viable. In addition simply to burial, some anthracites are also known to be products of tectonic deforma- tion or devolatilization by nearby igneous intru- sions, a good example of the former being the Pennsylvanian coals of the USA, which grade SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 303 H (30%) O (30%) Oil C (100%) Anthracite Bituminous Sub- bituminous Lignite Peat Coalification Oil shale Figure 5.35 Ternary C–H–O plot showing the compositional trend accompanying coali?cation and the comparison between coals and oil and oil shale compositions (after Forsman and Hunt, 1958). Figure 5.36 Bituminous coal seam from the Witbank area, South Africa, hosted in carbonaceous shales and deltaic sandstones of the Permian Ecca Group (photo is courtesy of Bruce Cairncross). ITOC05 09/03/2009 14:35 Page 303laterally into anthracites as they approach the Appalachian fold belt. Controls on the formation of economically viable coals during initial peat accumulation are, however, more dif?cult to constrain. Deltas have traditionally been regarded as the optimal environment for development of viable coal seams, although this concept is now questioned because the high clastic sediment input in this setting generally results in peat dilution and formation of less attractive, high ash coals (McCabe, 1984). Environmental models now favor the notion that optimal peat accumu- lation represents a discrete event that was temporally distinct from those of underlying and overlying clastic deposition. The notion of an environment where peat accumulation is optimized, but not diluted by clastic input, is best accommodated by the concept of raised and/or ?oating swamps. Floating swamps refer to buoy- ant peats that develop on the surface of shallow lakes to eventually accumulate and accrete on the lake margins (Figure 5.37). Raised swamps are seen as the ?nal product in a continuum of processes which extend from initial development of shallow lakes with ?oating peat, followed by formation of well vegetated, low-lying swamps, and eventually to raised swamps as accretion of organic material eventually sculpts a landscape that is elevated with respect to the sediment sub- strate. The latter environment is obviously one in which clastic sediment input is minimized. This evolutionary sequence, illustrated in Figure 5.37, serves as a model for the depositional environ- ment in which optimal formation of thick peat, and subsequently economic coal horizons, might form. Raised swamps can form in deltaic environ- ments, but the above model applies speci?cally to those settings in which sediment input is minimized and the resultant coals are low-ash in character. Examples of raised swamps include the Klang-Langat delta in Malaysia and areas of Sarawak where peat accumulation of more than 2m my r -1 is occurring (McCabe, 1984). 5.4.4 Oil shales and tar sands Although organic matter accumulates in most sediments, it is the shales or mud rocks which contain the highest hydrocarbon contents. Mud rocks that are rich in organic matter are called sapropelites, but those that yield free oil on heat- ing are speci?cally referred to as “oil shales.” Oil shales typically form in lacustrine settings where algae settle out of the aerated upper layers and accumulate in the anaerobic bottom muds where they are preserved from rapid oxidative decay. In this sense they are akin to the sapropelic cannel or boghead coals discussed above. Since algal (Type I) kerogen is oil prone, these shales represent a resource of crude oil that has not as yet matured, or passed through the oil window. A major chal- lenge for the future is to develop the technology to heat these rocks up to the temperatures required to ef?ciently extract oil, and to do so in an eco- nomically viable and environmentally respons- ible way. The best known oil shale is perhaps the Green River Formation which developed in an Eocene, shallow-lacustrine or sabkha setting, in parts of Utah, Colorado, and Wyoming, USA. Despite an enormous potential capacity to produce oil (grades of up to 240 liters of oil per ton of rock are reported), production from the Green River shale resource has, thus far, proven non-viable. Successful processing of oil shale has nevertheless been achieved in countries such as Estonia, Russia, and China. Present day oil shale forma- tion is taking place in Australian saline lakes where the alga Botryococcus braunii is accumu- lating to form a type of sapropelic peat known as coorongite. Likewise, the torbanites of Scotland formed in lacustrine settings from accumulation of a Carboniferous variety of Botryococcus, as did the Permian oil shales of Tasmania (tasmanites). Tar sands are often confused with oil shales, although they are in fact quite different. Tar sands are sandstones within which bitumen or asphalt occurs. Bitumen is a solid or semi-solid hydrocar- bon either derived by normal degradation of crude oils, or formed directly as a high molecular weight hydrocarbon without ever having been a light oil. In certain instances, tar sands that are near the surface can be extracted and processed to produce oil and a range of other useful hydrocarbon by- products. The best known of these is the Athabasca tar sand of northern Alberta, Canada. As with oil shales, though, technological and environmental 304 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:35 Page 304SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 305 dif?culties provide major obstacles to more wide- spread production from this resource at present. Both tar sands and oil shales, however, clearly represent signi?cant petroleum resources for the future. 5.4.5 Gas hydrates A fairly recent discovery in the permafrost regions of the world, as well as in ocean sediments sam- pled during the Deep Sea Drilling Program, is the (I) Floating swamp on lake margins (II) Low-lying swamp, diverse and luxuriant flora (IV) Extensive raised swamp, restricted and stunted flora Lakes Vertically zoned peat (III) Slightly raised swamp, restricted flora Figure 5.37 Environmental model for optimal peat accumulation, in terms of an evolutionary development of swamp types which leads ultimately to the formation of raised swamps in which extensive, thick peats, devoid of major sediment input, form (after Romanov, 1968; McCabe, 1984). ITOC05 09/03/2009 14:35 Page 305presence of vast resources of hydrocarbon locked up as frozen gas hydrates. Gas hydrates are crys- talline aqueous compounds that form at low temperatures when the ice lattice expands to accommodate a variety of gaseous molecules, the most important of which, for the purposes of this discussion, is methane (CH 4 ). A methane hydrate forms when suf?cient CH 4 is present in water, at the appropriate pressure and temperature, to form a solid compound with the ideal composition CH 4 .5 3 / 4 H 2 O (i.e. the unit cell comprises 46 water molecules with up to 8 methane molecules). Other gases, including CO 2 , H 2 S, and C 2 H 6 , also form gas hydrates with a similar structure. On a global scale the volumes of natural gas present in gas hydrate reservoirs are estimated to be a factor of two times greater than the total fossil fuel reserve to present (Kennicutt et al., 1993). Methane hydrates clearly, therefore, represent an enorm- ous energy resource for the future. They are also, however, an object of concern in that global warming might accelerate the natural release of CH 4 by melting gas hydrates and further accentu- ate the in?uences of greenhouse gas accumula- tion on climate change. It is of interest to consider the conditions under which methane hydrates form in the ocean basins of the world. Figure 5.38a shows a sim- pli?ed phase diagram with the H 2 O ice–water and CH 4 hydrate–gas boundaries marked. If suf?cient methane and pore water are available methane hydrates will stabilize in ocean ?oor sediment where temperatures are around 3–4 °C at pressures of about 50 atmospheres (equivalent to about 500 meters water depth). Gas hydrates will melt if either the temperature increases or the pressure decreases. The thickness of the methane hydrate stability zone in oceanic sediments will depend on the pressure (an increase in which has the effect of stabilizing hydrates to lower temperatures) and the geothermal gradient, the latter dictating the rate at which sediment is heated with progressive burial. For a constant geothermal gradient the thickness of the methane hydrate layer will increase directly as a function of water depth. Figure 5.38b shows this graphically for a typical ocean-sediment pro?le on the continental shelf. It illustrates the initiation of gas hydrate formation 306 PART 3 SEDIMENTARY/SURFICIAL PROCESSES 10 10 000 –10 Depth (m) 5000 01 02 0 4 0 Temperature (°C) 30 1000 50 100 Methane hydrate + ice + gas 500 1000 Ice–water phase boundary 1 5 10 50 100 500 Pressure (atmospheres) Methane hydrate + water + gas Methane gas + ice Methane gas + water Hydrate–gas phase boundary (a) Figure 5.38 (a) Phase diagram illustrating the regions of gas hydrate stability under most natural conditions in the near-surface (after Kvenvolden and McMenamin, 1980). (b) Pro?le across a typical ocean–sediment interface in a continental margin setting, showing the progressive increase in the width of the gas hydrate stability zone in the ocean sediment with increasing depth of sea water (after Kvenvolden, 1988). in sediment under about 500 m of water and the progressive increase in thickness of the hydrate stability zone with depth. Under 3 km of water a methane hydrate zone almost 1000 m thick could form if the ingredients were available. This par- ticular pro?le was calculated for a geothermal gradient of 27 °C km -1 but would be thinner if the sediment pro?le was subjected to higher geo- thermal gradients. These considerations, even if viewed conservatively, point to the enormous volumes of gaseous hydrocarbons that potentially could be locked up in the ocean sediments. ITOC05 09/03/2009 14:35 Page 306SEDIMENTARY ORE-FORMING PROCESSES CHAPTER 5 307 Sedimentation is a fundamental geological pro- cess that takes place over much of the Earth’s sur- face, in response to the pattern of global tectonic cycles. A number of very important mineral and fossil fuel commodities are concentrated during the formation of sedimentary rocks. Placer pro- cesses are important during clastic sedimentation and form by the sorting of light from heavy par- ticles. A number of hydrodynamic mechanisms, such as settling, entrainment, shear sorting, and transport sorting, are responsible for the concen- tration of different commodities in a variety of sedimentary micro- and meso-environments. At a larger scale, placer deposits form mainly in ?uvial and beach-related environments and include con- centrations of gold, diamonds, tin, zirconium, and titanium. Chemical sedimentation, where dis- solved components precipitate out of solution from brines, also gives rise to a wide range of important natural resources. Precipitation of fer- ric iron from sea water is one of the principal mechanisms for the formation of iron ore deposits. Banded iron-formations formed in a variety of oceanic settings mainly in the Paleoproterozoic and prior to the onset of signi?cant levels of oxy- gen in the atmosphere. Ironstone deposits tend to 4500 Depth (m) 3500 4000 3000 2°C 2500 2000 1500 1000 500 3°C Depth of gas hydrate stability zone 10° 20° 30° 40°C 1.5°C 10° 20° 30° 40°C 10° 20° 30° 40°C 3°C 10° 20° 30° 40°C 4°C 7°C 13°C 18°C Sediment– water interface Temperature of bottom waters Gas hydrate zone under the ocean Temperature profiles in sediment (geothermal gradient = 27°C km –1 ) 0 0 0 0 Sea level (b) Figure 5.38 (cont’d) ITOC05 09/03/2009 14:35 Page 307form mainly in Phanerozoic times. Other types of metal concentration linked to syn-sedimentary and early diagenetic chemical processes occur in carbonaceous black shales (Ni, Co, Cr, Cu, Mn, Zn, Ag, etc.) and in Mn nodules on the ocean ?oor. Accumulation of phosphorus and the development of phosphorites in ocean settings is a process that is linked both to direct chemical precipitation from sea water and to biological mediation. The upwelling of cold currents onto the continental shelf and biological productivity are processes implicated in the formation of most of the world’s important phosphate deposits, so important for fertilizer production and global food production. Chemical sediments formed by high rates of evaporation of sea water and lacustrine brines host most of the world’s Na, K, borate, nitrate, and sulfate resources, also important to the agri- cultural industry. Fossil fuels, in the form of oil, natural gas, and coal, are the most important and valuable of all A huge volume of literature is available in the ?elds of sedimentology and fossil fuels, but less with respect to chemical sedimentation and metallogeny. The follow- ing books and reviews are subdivided in accordance with the main topics covered in this chapter. Sedimentological and placer processes Allen, P.A. (1997) Earth Surface Processes. Oxford: Blackwell Science, 404 pp. Leeder, M. (1999) Sedimentology and Sedimentary Basins: From Turbulence to Tectonics. Oxford: Black- well Science, 592 pp. Slingerland, R. and Smith, N.D. (1986) Occurrence and formation of water-laid placers. Annual Reviews of Earth and Planetary Sciences, 14, 113–47. Chemical sedimentation and ore formation Mel’nik, Y.P. (1982) Precambrian Banded Iron Forma- tions. New York: Elsevier, 282 pp. Melvin, J.L. (1991) Evaporites, Petroleum and Mineral Resources. Developments in Sedimentology, 50. New York: Elsevier, Chapter 4. natural resources as they provide the world with most of its combustion-derived energy. Oil and natural gas are derived in the initial stages by biodegradation of largely planktonic organisms in marine environments to form oil-prone kerogen and biogenic gas. Further burial results in subse- quent chemical and thermal modi?cation of kero- gen to produce wet gas, and a liquid condensate in the “oil window,” typically at burial depths of 3000–4000 m and temperatures of 100–150 °C. Further burial results only in the production of more dry gas. Burial of vascular land plants gives rise initially to the formation of peat and the production of mainly dry gas. Further compaction of peat gives rise to coals, the rank of which is a function of progressive burial. Other forms of fossil fuel, important for the future but largely unexploited at present for technological and eco- nomic reasons, include oil shales and tar sands, as well as methane hydrates in permafrost regions and ocean ?oor sediments. Parnell, J., Ye, L. and Chen, C. (1990) Sediment-hosted Mineral Deposits. Special Publication, 11, International Association of Sedimentologists. Oxford: Blackwell Scienti?c Publications, 227 pp. Young, T.P. and Gordon Taylor, W.E. (1989) Phanerozoic Ironstones. Special Publication 46. London: The Geo- logical Society of London, 257 pp. Fossil fuels Engel, M.H. and Mack, S.A. (1993) Organic Geochem- istry: Principles and Applications. New York: Plenum Press, 861 pp. Glennie, K.W. (1998) Petroleum Geology of the North Sea: Basic Concepts and Recent Advances. Oxford: Blackwell Science, 636 pp. North, F.K. (1985) Petroleum Geology. London: Allen & Unwin, 607 pp. Selley, R.C. (1998) Elements of Petroleum Geology, 2nd edn. New York: Academic Press, 470 pp. Thomas, L. (1992) Handbook of Practical Coal Geology. New York: John Wiley. Ward, C.R. (1984) Coal Geology and Coal Technology. Oxford: Blackwell Science, 345 pp. 308 PART 3 SEDIMENTARY/SURFICIAL PROCESSES ITOC05 09/03/2009 14:35 Page 308Global Tectonics and Metallogeny ITOC06 09/03/2009 14:34 Page 309ITOC06 09/03/2009 14:34 Page 3106.1 INTRODUCTION Most of the world’s great mineral deposits are the products of a fortuitous superposition of geological processes that resulted in anomal- ous concentration of ores, often over an extended period of time. The preceding chapters have shown that the formation of ore deposits is, neverthe- less, related to much the same sort of processes that gave rise to the formation of normal igneous, sedimentary, and metamorphic rocks in the Earth’s crust. The fact that a close relationship exists between rock-forming and ore-forming pro- cesses means that metallogeny must be relevant to understanding the nature of crustal evolution through geological time (Barley and Groves, 1992; Windley, 1995). Conversely, crust-forming pro- cesses and the global plate tectonic paradigm have become indispensable to the broader understand- ing of how ore deposits form and no modern eco- nomic geologist can practice successfully without an appreciation of the history and evolution of the continents and oceans. Consequently, this chapter summarizes some of the recent thinking that relates the formation of ore deposits to global tectonics and the evolution of the continents. There are essentially two ways of doing this. One is to chart continental evolution and place ore deposits into a secular and tectonic framework (e.g. Meyer, 1981, 1988; Veizer et al., 1989; Barley and Groves, 1992). The other is to empirically describe ore deposits in the context of the tectonic environment and host rocks in which they occur (e.g. Mitchell and Garson, 1981; Hutchison, 1983; Sawkins, 1990). The latter approach has been covered in considerable detail in Sawkins’s (1990) Metal Deposits in Relation to Plate Tectonics, and there is little point in summarizing this ex- tensive compilation here. The former approach is a more dif?cult undertaking and requires a thor- ough knowledge of the evolution of continents with time, a topic that becomes progressively less well constrained – and more controversial – the further back in Earth history one goes. There are, however, signi?cant bene?ts to be gained by under- standing the intricate processes of ore formation in terms of the secular evolution of the Earth’s crust and atmosphere. This is the approach that has been adopted below, even though it is neces- sarily cursory and in places speculative. Ore deposits in a global tectonic context PATTERNS IN THE DISTRIBUTION OF MINERAL DEPOSITS CONTINENTAL GROWTH RATES CRUSTAL EVOLUTION AND METALLOGENESIS evolution of the hydrosphere and atmosphere global heat production mantle convection continental freeboard and eustatic sea-level changes crustal composition and the Archean–Proterozoic transition METALLOGENY THROUGH TIME the Archean Eon the Proterozoic Eon the Phanerozoic Eon PLATE TECTONIC SETTINGS AND ORE DEPOSITS – A SUMMARY ITOC06 09/03/2009 14:34 Page 3116.2 PATTERNS IN THE DISTRIBUTION OF MINERAL DEPOSITS It has long been known that the different styles of mineralization are not randomly distributed, either in time or in space, and that broad patterns exist when relating deposit types to crustal evolution and global tectonic setting. The original compilations of ore deposit type as a function of geological time by Meyer (1981, 1988) have been re-examined by Veizer et al. (1989) and Barley and Groves (1992) and reveal consistent patterns. Figure 6.1 distinguishes between the secular dis- tribution of metal deposits formed in orogenic settings and those formed in anorogenic environ- ments and in continental basins. Although this compilation is a “?rst-order” generalization, it is nevertheless clear that deposit types classi?ed as orogenic (including lode-gold or orogenic gold ores, volcanogenic massive base metal sul?des, and the porphyry-epithermal family of base and precious metal deposits) occur either in the late stages of the Archean Eon (between 3000 and 2500Ma) or in the Phanerozoic Eon (between 540 Ma and present). The Proterozoic Eon, between 2500 and 540 Ma, contains fewer of these types of deposits. By contrast, the majority of metal deposits that are associated with so-called “ano- rogenic” magmatism (such as the anorthosite- hosted Ti deposits and the Olympic Dam and Kiruna type Fe oxide–Cu–Au ores), as well as sedi- ment hosted ores (such as the SEDEX type Pb–Zn deposits and the stratiform “red-bed” type Cu deposits), are dominantly hosted in Proterozoic rocks (Figure 6.1). An explanation for this pattern, as discussed below (see sections 6.4.3 and 6.5 in particular), is that ore formation might be closely linked to the so-called “supercontinent cycle,” which describes the broad scale amalgamation and dispersal of the major continental fragments with time. It is, for example, evident that peaks in the production of anorogenic and continental sediment-hosted metal deposits (Figure 6.1, lower histogram) coincide with the periods of crustal stability and the existence of large, stable con- tinental amalgamations such as Nena in the Mesoproterozoic, Rodinia in the Neoproterozoic and Pangea in the early Mesozoic. Periods of large-scale continental fragmentation, by contrast, appear to be orogenically more active and give rise to a different suite of ore deposit types. In a study designed to examine the rates at which ore deposits are recycled or dispersed (either by erosive or tectonic dismembering) in the Earth’s crust, Veizer et al. (1989) have produced a very similar pattern of ore deposit distribution with time. In this compilation hydrothermal and volcano-sedimentary base–precious metal deposits formed mainly in late Archean and Phanerozoic times, whereas chemical–sedimentary and ultra- ma?c deposits re?ect concentration mechan- isms that took place in the mid-Proterozoic. The reasons behind this pattern are likely to be complex and multifaceted and are discussed in more detail below. It should be emphasized at this stage, however, that the relationships observed in Figure 6.1 could also be due, at least in part, to a preservation factor. The existence of numerous arc-related orogenic deposits in the late Archean, for example, re?ects the preservation of green- stone belts in stable shield areas of the world. In Proterozoic times, by contrast, the preponderance of collisional orogenies might have uplifted and eroded similar mineralized arcs so that many of the related near-surface ores were destroyed. The products of young Cenozoic orogenies are also preserved because many of them have not yet suffered collision and uplift and are, therefore, still preserved. 6.3 CONTINENTAL GROWTH RATES The relationships illustrated in Figure 6.1 indicate that estimates of the rates of continental growth with time are important to any consideration of the patterns of continental assembly and dispersal and, therefore, metallogeny. Such estimates are problematic because of the dif?culties in calculat- ing the rates of recycling (or destruction) of con- tinental material relative to new crust formation. Growth rate estimates vary enormously, and range from one scenario, where the majority of con- tinental crust formed rapidly in the early Arch- ean (curve A in Figure 6.2a), to another where continents grew exponentially with time, with the most substantial contributions coming after 312 PART 4 GLOBAL TECTONICS AND METALLOGENY ITOC06 09/03/2009 14:34 Page 312ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 313 Copper in clastic sediments Sediment-hosted Lead-zinc in carbonates (Mississippi Valley type) Lead-zinc in clastic sediments Ilmenite-anorthosite Olympic Dam-type Kiruna-type Anorogenic intrusions Uranium in weathered profiles 2000 1000 Present 3000 Ma Anorogenic and continental sediment-hosted Porphyry copper Porphyry deposits Gold and uranium in conglomerates Lode-gold veins Kuroko-type Abitibi-type Cyprus-type Volcanogenic massive sul?de (VMS) 2000 1000 Present 3000 Ma Porphyry molybdenum Orogenic Nena Rodinia Pangea Figure 6.1 Distribution of ore deposit styles, classi?ed according to orogenic (upper set of histograms) and anorogenic/continental sediment-hosted metal deposits (lower set of histograms) as a function of time. The length of each histogram bar is an estimate of the proportion of ore formed over a 50 million year interval relative to the global resource for that deposit type. Periods of supercontinent amalgamation are shown as Pangea, Rodinia, and Nena, with decreasing levels of con?dence further back in time. The diagram is originally after Meyer (1988) and modi?ed after Barley and Groves (1992). ITOC06 09/03/2009 14:34 Page 3131.5 Ga (curve B in Figure 6.2a). The most likely situation probably re?ects an intermediate case where approximately linear, but episodic, crustal growth took place. McCulloch and Bennett (1994) supported this scenario on the basis of Sm–Nd model and U–Pb zircon age distribution data for the continental crust, showing that the propor- tion of crustal recycling relative to growth is approximately constant. The model, illustrated in Figure 6.2b, suggests discrete episodes of enhanced crustal growth. It envisages that 50–60% of the Earth’s continental crust had been produced by the end of the Archean, but that this occurred in two major episodes at 3600–3500 Ma and at 314 PART 4 GLOBAL TECTONICS AND METALLOGENY 0.00 0 Fraction of crust produced per 200 Myr 0.15 0.10 0.05 1000 2000 3000 4000 Time before present (Ma) 1.0 0.8 0.6 0.4 0.2 0.0 Fraction of crust 0 1000 2000 3000 4000 Time before present (Ma) B A (a) (b) M & B Figure 6.2 (a) Different models for the rates of continental growth. Curves A and B contrast considerations of early Archean continental growth with an exponential growth rate model in which the bulk of the continents grew subsequent to about 1500 Ma. The curve M&B illustrates the broadly linear but episodic continental growth rate curve suggested by McCulloch and Bennett (1994). (b) Histogram showing the estimated rates of continental growth averaged over 200 million year intervals based on a global compilation of Sm–Nd model ages (after McCulloch and Bennett, 1994). ITOC06 09/03/2009 14:34 Page 314ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 315 2800–2600 Ma, with the latter being the most important period of crustal growth in Earth his- tory. Two important episodes of crust formation were also considered to have taken place in the Proterozoic, one at 2000–1800 Ma and the other at around 1100–1000 Ma. It is interesting to note that the McCulloch and Bennett model suggests that only a relatively small proportion of new crust was added to the Earth’s surface in the Phanerozoic Eon. This appears to be somewhat at odds with the observation that this period of Earth history was particularly pro- ductive in terms of ore deposit formation. This supports the notion of preferential preservation of younger ore-forming environments compared to older sequences which become either progress- ively more eroded, or buried and metamorphosed, with time. The episodic continental growth model described in Figure 6.2 is reasonably con- sistent with the pattern of metallogenic evolution described here and is used as the basis for a closer examination of the relationships between crustal evolution and ore deposit formation. 6.4 CRUSTAL EVOLUTION AND METALLOGENESIS The review by Barley and Groves (1992) suggested that there were three principal factors which affected the pattern of global metallogeny. These are the evolution of the hydrosphere–atmosphere, the secular decrease in global heat production, and long-term global tectonic trends. These major fac- tors, together with a number of related parameters, are discussed in more detail below, with an indica- tion of how and why they in?uence the temporal and geographic distribution of ore deposit types. 6.4.1 Evolution of the hydrosphere and atmosphere The changing budget of O 2 , and to a lesser degree CO 2 , in the Earth’s atmosphere with time had a signi?cant role to play in the formation of ore deposits, especially those related to redox processes and to the weathering and erosion of continental crust. During the Archean it is gen- erally held that the atmosphere contained very little free molecular oxygen (although the actual amounts are still strongly debated) and what little did exist was the result of inorganic photodissoci- ation of water vapor. A reduced atmosphere in the Archean helps to explain many of the features of ore formation at that time, including the widespread mobility of Fe 2+ and development of Algoma and Superior type banded iron-formations, and the preservation of detrital grains of uraninite and pyrite in sedimentary sequences such as those of the Witwatersrand and Huronian basins. The transition from the Archean to the Proterozoic Eons, at 2500 Ma, broadly coincides with the ?rst major increase in atmospheric oxygen (Figure 6.3). This event is possibly related to the evolution of primitive life and the development of cyanobac- teria capable of photosynthetically producing oxy- gen. The partial pressures of O 2 in the atmosphere can be estimated from the mobility of Fe 2+ in paleosols (the remnants of weathering horizons or soils preserved in the rock record) and suggest that a sharp increase in the partial pressure of oxy- gen (pO 2 ) occurred at around 2200 Ma (Rye and Holland, 1998). The interval between about 2500 and 2000 Ma is known as the period of “oxyat- moinversion” and is generally believed to coin- cide with a signi?cant rise in atmospheric oxygen levels. It is also the period after which develop- ment of Superior type banded iron-formations and related bedded manganese deposits ceased. An- other signi?cant increase in atmospheric oxygen levels coincided with the end of the Proterozoic Eon, when macroscopic, multicellular life forms (the Metazoan fauna which required oxygenated respiration) proliferated. In addition, the early Phanerozoic is also the time when vascular plants evolved and this, too, would have contributed to the increase in atmospheric oxygen levels (Figure 6.3). In contrast with the net addition of O 2 in the atmosphere, CO 2 has progressively decreased with time. The initially high levels of CO 2 in the atmosphere were a product of ex- tensive volcanism and outgassing during the Archean, but levels have dropped steadily, but episodically, as a result of carbon and carbonate deposition in sediments. The evolution of the Earth’s atmosphere and, in particular, the global O 2 and CO 2 trends are essentially non-recurrent ITOC06 09/03/2009 14:34 Page 315and irreversible and contribute to the time-speci?c occurrence of ore deposit types such as banded iron-formations and Witwatersrand type placers. 6.4.2 Secular decrease in global heat production The production of heat from the Earth, mainly from the decay of long-lived isotopes such as U, Th, and K, was signi?cantly greater (by a factor of two to three times) in the Archean than it is today. This pattern, like that for atmospheric evolution, is also non-recurrent and irreversible. Figure 6.4 shows the decline in global heat production with time and a corresponding crustal growth curve similar to that discussed above. There are several consequences for crust forma- tion related to the secular decrease in heat produc- tion. One of the most obvious is that the Archean mantle was hotter (by about 200–300 °C; Windley, 1995) than at present and this could have led to an increase in the volume of melting at this time. A higher degree of partial melting could account for production of the high-Mg komatiites that typify many Archean greenstone belts and which are relatively rare in younger oceanic settings (see Figure 6.4), although other explanations are also feasible. Higher heat production thus provides an explanation for the preferential occurrence, in the Archean, of komatiite-related magmatic Ni–Cu sul?de deposits, such as those at Kambalda, Western Australia and in Zimbabwe. Another consequence of higher heat ?ow in the Archean is the suggestion that convective overturn in the mantle would have been more rapid, leading to shorter ocean crust residence times and more rapid plate motion (Windley, 1995). Archean oceanic crust is also considered to have been thicker than in later periods, which is supported by suggestions that ophiolites greater than 1000 years old are more than double the thickness of their younger equivalents (Moores, 1993). As another consequence, Archean oceans may have been more shallow than those at present, result- ing in a reduced continental freeboard. 316 PART 4 GLOBAL TECTONICS AND METALLOGENY 1.0 0.000001 2.8 Atmospheric pO 2 (atm) 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 2.0 1.9 1.8 Time before present (Ga) 1.1 0.45 0.00001 0.0001 0.001 0.01 0.1 Metazoa Oxyatmoinversion Figure 6.3 Estimates of the episodic rise of pO 2 over the past 3 Gyr, constrained by data derived from the mobility of Fe in paleosols (fossilized soil horizons). Bars and arrows show actual estimates of the ages and pO 2 for a number of paleosol horizons worldwide, whereas the solid curve is the best estimate of the atmospheric pO 2 trajectory with time (after Rye and Holland, 1998). ITOC06 09/03/2009 14:34 Page 316ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 317 6.4.3 Long-term global tectonic trends The broad-scale amalgamation and dispersal of continental fragments with time, also referred to as the “supercontinent cycle,” is an all- encompassing feature of Earth history that has implications for virtually every aspect of global evolution, including metallogeny. As seen in Figure 6.1, broad temporal trends in ore formation re?ect the contrast between episodes of orogeny and periods of stability. In detail, however, it is the supercontinent cycle that controls the many other factors more directly related to secular met- allogenic trends. Unlike the patterns imposed by atmospheric evolution and heat production, the supercontinent cycle is recurrent and cyclical in that continental fragments have amalgamated and dispersed several times through Earth history. This cyclicity helps explain why some ore deposit types, or at least the processes that give rise to them, are recognizably repetitive despite the pre- servation bias of ore deposits in younger rocks. A detailed understanding of how the supercontinent cycle can be used in the study of ore deposits is a complex issue and likely to remain a pro?table area of future research. Some of the concepts that are relevant to both global tectonics and the pat- tern of crustal/metallogenic evolution are brie?y mentioned below. Rates of continental growth The growth of continental crust, whether it be episodic or linear, early or late, will affect global metallogeny simply because the majority of viable ore deposits are located in or on continental crust. Periods of rapid continental growth re?ect episodes of more intense magmatic activity and tectonism, and this, in turn, promotes the forma- tion of magmatic, magmatic-hydrothermal, and hydrothermal ores. The majority of the world’s gold deposits, for example, are related to events in either the late Archean or Mesozoic–Cenozoic (Figure 6.1), both periods of major orogenesis and crust formation (Figure 6.2b). The nature of mantle convection Models for mantle convection suggest two modes of occurrence (Figure 6.5), which may not be mutually exclusive and could operate at the same time. One model suggests overturn of material in two discrete layers with little or no mass transfer between them, and the other envisages that the entire mantle overturns due to the major down- welling of cold lithospheric slab material with concomitant up-rise of large magmatic plumes (Stein and Hofmann, 1994). The ?rst scheme (Figure 6.5a) is considered to represent the situation 5 1 4 Relative heat production (no units) 3 2 321 4 50 25 75 % Continental growth 0 Time before present (Ga) Heat Norites Komatiites Crustal growth Basalts Figure 6.4 Schematic diagram showing the exponential decline in global heat production with time and the matching decline in production of komatiite volcanism. Also shown is a crustal growth curve and the period of enhanced ma?c or “norite” magmatism in the late Archean and early Proterozoic (after Hall and Hughes, 1993). ITOC06 09/03/2009 14:34 Page 317during which continental growth occurs mainly by ocean subduction, arc construction, and accre- tion. In this model global tectonic patterns are re?ected in terms of the Wilson cycle, in which oceans open and close in ? 500 million year peri- ods. The other mode, called the MOMO (Mantle Overturn and Major Orogeny) model, involves signi?cant exchange of material between lower and upper mantle and, accordingly, higher rates of crust formation (Figure 6.5b). This mode is con- sidered to operate at longer periodicities than the Wilson cycle and produces magmatic activity that coincides with major orogenic events, such as those at circa 2.8, 2.0, 1.0, and 0.5 Ga (Figure 6.2b). From a metallogenic viewpoint, the proliferation of plume activity attributed to MOMO events might coincide with periods of alkaline magmat- ism and is, therefore, relevant to the ore deposits associated with such rocks. Eustatic sea-level changes and “continental freeboard” Continental freeboard, or the relative elevation of the continental land masses with respect to sea level or the geoid, is effectively a measure of the area of the exposed continents compared to that covered by the oceans. Continental freeboard is reduced when the average ocean depth is reduced, since this causes marine transgression and ?ood- ing of the continental shelf. This situation is linked to events of large-scale continental disper- sal, which in turn re?ects enhanced creation of oceanic crust, and active magmatism and uplift along mid-ocean ridges. By contrast, an increase in the area of the continents occurs when the oceans are deepened and the continental shelf exposed, a situation which tends to be associ- ated with periods of tectonic stability and muted 318 PART 4 GLOBAL TECTONICS AND METALLOGENY MOMO EPISODES (500–1000 Ma?) WILSON CYCLES (< 500 Ma) (a) Arc Upper mantle Mid-ocean ridge Lower mantle (b) Arc Plume heads Deep penetrating subducted slab Plumes 660 km 2900 km Figure 6.5 Two contrasting models for mantle circulation (after Stein and Hofmann, 1994). (a) Mantle convection in two discrete layers with little or no mass transfer from one to the other, corresponding to the conventional plate tectonic regime in which crustal growth occurs in accordance with the Wilson cycle. (b) Deeply penetrative mantle overturn involving the downwelling of cold lithosphere into the lower mantle and the associated upwelling of major plumes. These MOMO (mantle overturn and major orogeny) episodes are believed to replenish the depleted upper mantle chemically and also to coincide with the major periods of orogenic activity and enhanced crustal growth. ITOC06 09/03/2009 14:34 Page 318ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 319 mid-ocean ridge magmatic activity (Worsley et al., 1984; Nance et al., 1986). The cyclicity of continental freeboard and ocean depth in the Phanerozoic Eon is well documented (Figure 6.6) and shows that a period of maximum continental exposure and ocean low-stand (i.e. deep oceans) coincides with the amalgamation of the Pangean supercontinent in Permian–Triassic times. The nature and duration of Wilson cycles in the Phanerozoic, and their projection into the future, are also shown in Figure 6.6. A similar period of ocean lowstand may also have applied with respect to the amalgamation of an earlier super- continent, Rodinia, at approximately 1000Ma (see section 6.5.2 below). From a metallogenic viewpoint, continental freeboard has its biggest in?uence on the nature and preservation of sediments forming on the continental shelves. Marine transgressions ?ood the continental shelf, preserving the sediments within which deposits such as heavy mineral beach placers, SEDEX Pb–Zn ores, banded iron-formations, bedded man- ganese deposits, and phosphorites might have formed. Oceanic lowstand and maximal contin- ental area, on the other hand, would tend to result in shelf exposure and erosion, with possible de- struction of any ore deposits present in the sedi- mentary sequence. 6.5 METALLOGENY THROUGH TIME The pattern of crustal evolution in the Phanerozoic Eon, involving the progressive amalgamation of large continental fragments to form the super- continent of Pangea during Permian and Triassic times, and its subsequent dispersal to form the present day continental geography, is well 5 600 Water depth at shelf break (×100 m) 500 400 300 200 100 0 100 Fragmentation 200 Ma –3 –1 0 1 2 3 4 –2 100 0 200 300 400 440 Ma Wilson cycle 200 300 100 400 440 Wilson cycle Open Rift Close Dispersal Assem- bly Stasis Rift Pangea Fragmentation Open Rift Stasis Super- continent Assem- bly Close Dispersal Today SHALLOW OCEANS DEEP OCEANS The future Figure 6.6 Pattern of Wilson cycles, re?ecting Pangean assembly and fragmentation during the Phanerozoic Eon (and into the future), and showing water depth of the continental shelf (after Nance et al., 1986). ITOC06 09/03/2009 14:34 Page 319constrained. A relatively high degree of con?d- ence marks the reconstruction of continental con?guration during the Phanerozoic Eon and geoscientists generally agree on the paleogeo- graphy of Gondwana, Laurentia, and Pangea. As one goes back in time, however, even into the late Precambrian, the situation becomes progressively more uncertain and disagreement often accompan- ies the reconstruction of continental geometry. There are many reasons for this, including the dif?culties in acquiring and accurately dating apparent polar wander paths, the high rates of recycling (or destruction) of continental and oceanic crust, and progressive deformation and burial of crust with time. Despite these dif?culties, considerable progress has been made in reconstructing continental paleogeography during the latter stages of the Proterozoic Eon. There are indications that a Neoproterozoic supercontinent, generally referred to as Rodinia, was assembled from about 1000 Ma, and then dispersed again by about 700 Ma. There is also some evidence for the existence of a Mesoproterozoic supercontinent, although at this stage there is still considerable uncertainty as to the nature of its constituents, as well as its shape and position. It is also likely that substantial con- tinents existed in the Paleoproterozoic and even into the late Archean, although at this stage the con?guration of these continents is speculative. Windley (1995) has compiled a detailed account of the many ideas regarding the evolution of the crust over the entire span of Earth history, and this work provides an excellent platform upon which to base the following discussion. Rogers (1996) has also attempted to reconstruct a history of the continents and their con?guration over the past 3000 million years. He has also chosen to name the larger continental fragments over this time period and charted their evolution in terms of major periods of amalgamation and dispersal. In the sections that follow this scheme is sum- marized and complemented with more recent data pertaining to Neoproterozoic (Rodinia) and Phan- erozoic (Pangea) continental evolution. Although still in its early stages of development and, there- fore, speculative, the Rogers scheme nevertheless provides a useful framework upon which to base a discussion of metallogeny through time. It is also a working hypothesis that can be used to test ideas relating to crustal evolution and the link to global metallogeny. Figure 6.7 presents a schematic history of continental amalgamation and dispersal between 3000 and 300 Ma, together with names for suggested early continental fragments. In brief, the Rogers model commences with an inferred Archean continent known as Ur, com- prising the ancient Kaapvaal and Pilbara cratons of southern Africa and Western Australia, respect- ively, as well as parts of India and Antarctica. Ur is believed to have coalesced from about 3000 Ma and to have existed as a separate block throughout much of Earth history until it contributed to the assembly of a supercontinent at around 1000 Ma. 320 PART 4 GLOBAL TECTONICS AND METALLOGENY Arctica (also called Laurentia) 3.0 0.3 2.5 2.0 1.5 1.0 0.5 Age in Ga (not linear) Yilgarn N. India, central Aust. Nena E. Antarctica Baltica Atlantica Rodinia W. Gondwana E. Gondwana Gondwana Laurentia Kazakhstan, China, S. China, etc. Pangea Ur Present day configuration Figure 6.7 Schematic diagram showing the approximate timing of amalgamation and dispersal of the major continents through time (after Rogers, 1996). ITOC06 09/03/2009 14:34 Page 320ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 321 Another major continent, called Arctica, now largely preserved in parts of Canada, Greenland, and Siberia, is believed to have amalgamated toward the end of the Archean or in the early Proterozoic. At least two other substantial con- tinental fragments are believed to have formed during the Paleoproterozoic, namely Baltica and Atlantica. Arctica (also commonly referred to as Laurentia in earlier reconstructions; Hoffman, 1988) and Baltica are suggested to have amalgam- ated at around 1500 Ma to form the ?rst substan- tially consolidated mass of continental material (but not yet a supercontinent), called Nena (Figure 6.7). Nena, Ur, and Atlantica are considered to have been the founding blocks for what was arguably the ?rst supercontinent, Rodinia, which existed largely in the Neoproterozoic and com- prised an amalgamation of most continental mass at that time. Thereafter, the pattern of contin- ental evolution is far better constrained. Rodinia broke up into essentially three large fragments known as Laurentia (i.e. an enlarged version of the Paleoproterozoic Laurentia) and east and west Gondwana, although it is likely that other smaller continental entities also existed at this time. The two sections of Gondwana amalgam- ated into a single entity during the Panafrican and Brasiliano orogenies at around 500Ma. Laurentia collided with Baltica after the closure of the Iapetus Ocean in the early Paleozoic, at about 400 Ma, to form a transient entity known as Laurasia. It was the subsequent collision of Laurasia and Gondwana that gave rise to the sec- ond global supercontinent, Pangea, which formed at around 300 Ma. Since then continental frag- ments have dispersed to many parts of the globe, forming the present day geographic con?guration. Some areas are currently undergoing reamalgam- ation, with collision of India with Asia, and the partial amalgamation of Africa and Europe, having been particularly active over the past 20 million years. This pattern of continental evolution is used below as the framework around which to discuss global metallogenic trends. The discussion pro- gresses from oldest to youngest, even though much more is known about processes in younger periods of time. 6.5.1 The Archean Eon The Hadean (>4000 Ma) and Eoarchean (>3600 Ma) stages The Hadean Era refers to that period of Earth his- tory for which there is very little evidence in the rock record and which is nominally pre-Archean. It was a time of global differentiation and accre- tion, as well as intense meteorite bombardment. The Hadean was previously regarded as existing prior to 3800 Ma (Harland, 1989) although the growing evidence (from U–Pb zircon dating) for crustal remnants at close to 4000 Ma suggests that the latter date is perhaps a more accurate re?ection of the Hadean boundary (Windley, 1995). The lack of any meaningful preservation of Hadean crust anywhere on the face of the Earth is a feature generally attributed to widespread destruction of this ancient material, either by intense meteorite bombardment or by subduction associated with a turbulent, rapidly convecting mantle, or both. There is also evidence to suggest that the early atmosphere and ocean formed only at the end of the Hadean era, once the main period of accretion and meteorite bombardment had terminated (Kasting, 1993). De Wit and Hynes (1995) have suggested that the Hadean Earth was also characterized by loss of heat direct to the atmosphere, in contrast to later periods of time when heat loss is largely buffered by a liquid hydrosphere. The implications for metallogenesis are that sedimentation and hydrothermal process are likely to have been inconsequential in the Hadean, and any ore deposits that did form at that time were, therefore, probably igneous in charac- ter. It is conceivable, for example, that oxide and sul?de mineral segregations accumulated from anorthositic and basaltic magmas at this time. The only preserved record of such rocks within reach of humankind at present is, however, likely to be on the Moon. The Eoarchean refers to the dawn of Archean time and to rocks formed prior to 3600Ma, although for the purposes of this discussion it is considered to extend between 4000 and 3600 Ma. The best preserved section of Archean crust that falls into this time bracket is the 3800 Ma Isua ITOC06 09/03/2009 14:34 Page 321322 PART 4 GLOBAL TECTONICS AND METALLOGENY Ur at 3.0 Ga Bhandara Singhbhum Napier Vestfold Kaapvaal Dharwar Dronning Maud Land Pilbara Ur at ~1.5 Ga (a) UR boundary during collision with Nena at ~1 Ga Aravalli Bundelkhand Yilgarn Kimberley Gawler East Australia (b) Slave Nain Superior Wyoming Thelon Antarctica India AFRICA Madagascar ANTARCTICA AUST RALIA INDIA Zimbabwe Aldan Anabar Greenland Thelon-Akitkan trend Arctica Figure 6.8 Suggested con?guration and paleogeography of the two major continental blocks, Ur and Arctica, that may have been in existence by the end of the Archean Eon at 2500 Ma (after Rogers, 1996). (a) The con?guration of Ur at about 3.0 Ga relative to the continental outlines of southern Africa, India, Antarctica, and Australia in a Gondwana make-up and an enlarged con?guration for Ur at about 1.5 Ga that includes accretion of continental crust (such as the Zimbabwe craton) formed largely at 2800–2600 Ma. (b) The con?guration of Arctica at around 2500 Ma relative to continental outlines for North America and Greenland. ITOC06 09/03/2009 14:34 Page 322ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 323 supracrustal belt and associated Itsaq (previously called Amitsoq) gneisses of western Greenland. The Isua belt comprises ma?c and felsic metavol- canics, as well as metasediments, and resembles younger greenstone belts from elsewhere in the world. Although only 4 × 30 km in dimension, the Isua belt contains a major chert–magnetite banded iron-formation component as well as minor occurrences of copper–iron sul?des in banded amphibolites and in iron-formation (Appel, 1983). The largest iron-formation contains an estimated 2 billion tons of ore at a grade of 32% Fe. Scheelite mineralization has also been found in both amphibolite and calc–silicate rocks of the Isua belt, an association which suggests a submarine- exhalative origin. The coexistence of banded iron- formations and incipient volcanogenic massive sul?de style mineralization points to sea-?oor processes not unlike those active throughout much of subsequent Earth history. Although the zones of known mineralization in the Isua belt are sub-economic, at 3800 years old they clearly represent the oldest known ore deposits on Earth. The Paleo-, Meso-, and Neoarchean stages (3600 to 2500 Ma) The main stage of Archean crustal evolution took place over an extended duration of more than 1000 million years, during which time geo- logical processes were probably not too dissimil- iar to those of today – provided that allowances are made for features such as higher heat ?ow, thicker oceanic crust, and an anoxic atmosphere. De Wit et al. (1992) described the processes that took place during this period of time in terms of two principal stages, termed “intra-oceanic shield formation” (between about 3600 and 3100 Ma) and “intra- and inter-continental craton forma- tion” (between about 3100 and 2500 Ma). The latter stage terminated, in the Neoarchean, in what might have been the most intense period of crustal development in Earth history (Figure 6.2b). The Neoarchean was also characterized by extremely active ore-forming processes repre- senting igneous, hydrothermal, and sedimentary deposit types. The Rogers model (Figure 6.7) suggests that by the end of the Archean (2500 Ma) there might have been two major continental blocks in ex- istence, Ur and Arctica. The two blocks would have comprised segments of ancient crust now preserved in various parts of different present day continents and it is, therefore, doubtful that their con?guration in the Archean can be known with any real certainty. A suggested con?guration for Ur at around 3000 Ma is shown in Figure 6.8a and comprises the Kaapvaal and Pilbara cratons of southern Africa and Western Australia repect- ively, linked via a corridor made up of Archean segments from India and Antarctica. The existence of Ur receives some support from the similiarities that exist in the nature and ages of Archean greenstone belts and supracrustal sequences on the Pilbara and Kaapvaal cratons (Cheney, 1996; Martin et al., 1998a), a feature that is especially striking in the similarities between the huge Superior-type banded iron-formations of the two regions. The unique occurrence of the Archean- aged Witwatersrand basin on the Kaapvaal Craton, and its apparent absence in Western Australia, on the other hand, would tend to question the relevance of comparative geology that far back in Earth history. Likewise, paleomagnetic data contradict a close ?t of Pilbara and Kaapvaal (Wingate, 1998; Evans et al., 2000) so that the real paleogeography, and perhaps even the very existence of Ur at all, must remain question- able. Nevertheless, as a working hypothesis, a possible 1500 Ma con?guration for Ur is shown in Figure 6.8a, in which components of Neo- archean crustal growth, such as the highly miner- alized Zimbabwe craton, are also shown. Arctica (Figure 6.8b) is considered to have been made up of several ancient continental fragments (includ- ing west Greenland and the Slave province with their remnants of circa 4000–3800 million year old crust), although the actual amalgamation and of this continent is considered to have post- dated Ur, at around 2500 Ma. The ?t between the Siberian and Canadian portions of Arctica was originally suggested on the basis of colinearity in the trends of the Paleoproterozoic Akitkan and Thelon magmatic arcs. In a metallogenic context, Archean crustal evolution can be viewed in terms of a two stage ITOC06 09/03/2009 14:34 Page 323model which suggests that early “shield” forma- tion, in which amalgamation of oceanic basaltic terranes and emplacement of early tonalite– tronhjemite–granodiorite (TTG) magmas was followed by “cratonization,” where modern plate tectonic processes such as subduction and con- tinent collision occurred (De Wit et al., 1992). The characteristics and relevant ore-forming processes of these two stages are illustrated in Figure 6.9. Shield formation (circa 3600–3100 Ma) This period conforms approximately with the Paleoarchean era and is typi?ed by amalgama- tion of oceanic and arc-formed crust and the incipient stages of continental crust formation by emplacement of early, tonalite–trondhjemite– granodiorite (TTG) plutons (De Wit et al., 1992; Choukroune et al., 1997). This stage of Archean crustal evolution is characteized by the develop- ment of early continental shield areas comprising highly deformed (oceanic) greenstone remnants occurring as megaxenoliths within extensive TTG terranes (Figure 6.9a). The styles of mineralization that formed during this stage of Earth evolution are limited and best exempli?ed by the deposits previously described for the Isua belt of west Greenland. Algoma type banded iron-formations are a common compon- ent of early greenstone belt assemblages and re?ect the low oxygen levels of the atmosphere and the abundance of ferrous iron, sourced from exhalat- ive activity at the mid-ocean ridges. The exist- ence of oceans at this time and the likelihood of exhalative hydrothermal processes on the sea ?oor would have resulted in the formation of volcano- genic massive base metal sul?de deposits, although examples are rare. An exception is the well pre- served Big Stubby VMS deposit, in the 3460 Ma Warrawoona Group metavolcanics of the Pilbara Craton in Western Australia (Barley, 1992). Cratonization (circa 3100–2500 Ma) This stage of Archean crust formation coincides broadly with the Meso- and Neoarchean eras and is illustrated in Figure 6.9b. It is suggested to be the most proli?c period of crustal production in Earth history (Figure 6.2b) and is also a time of major global mineralization. The processes active at this time were not unlike those taking place later on in Earth history and involved widespread plate subduction, arc magmatism, continent col- lision and rifting, and cratonic sedimentation. In Figure 6.9b two sketches are presented. An earlier sub-stage illustrates consecutive accretion of island arcs onto a previously formed continental shield and stabilization of the latter by intrusion of large granite batholiths. A later sub-stage envis- ages the existence of Archean cratons consisting of numerous terranes, each bordered by major suture zones (possibly the fossilized sites of sub- duction or arc collision) and ?anked by both active and passive margins. Sites of intracratonic sedimentation are also envisaged in this stage of development. This scenario is consistent with the pattern of Archean crustal evolution envisaged, for example, in the Superior Province of Canada by Choukroune et al. (1997). From a metallogenic viewpoint this stage of Archean crustal evolution gave rise to a wide variety of ore-forming processes (Figure 6.9b). Arc-related volcanism associated with plate sub- duction contributed large volumes of magma to the accreting terranes of the time. Well mineral- ized examples of continental crust formed in the period 3100 – 2500 Ma are represented by the gran- ite–greenstone terranes of the Superior Province of Canada, as well as the Yilgarn and Zimbabwe cratons. Greenstone belts are hosts to numerous important volcanogenic massive sul?de (VMS) Cu–Zn ore bodies, such as those at Kidd Creek and Noranda in the Abitibi greenstone belt of the Superior Province. Off-shore, in more distal envir- onments, chemical sedimentation gave rise to Algoma type banded iron-formations, examples of which include the Adams and Sherman deposits, also in the Abitibi greenstone belt. Greenstone belts formed at this time also often contain komatiitic basalts that, under conditions favor- able for magma mixing and contamination, ex- solved immiscible Ni–Cu–Fe sul?de fractions to form deposits such as Kambalda in Western Aus- tralia and Trojan in Zimbabwe. During periods of compressive deformation, major suture zones became the focus of hydrothermal ?uid ?ow 324 PART 4 GLOBAL TECTONICS AND METALLOGENY ITOC06 09/03/2009 14:34 Page 324Intra-oceanic obduction Sea level Shallow oceans Thick crust Basaltic magmatism MOR VMS base–metal mineralization Algoma-style banded-iron formations Isua- type Calc–alkaline volcanism SHIELD FORMATION (3600–3100 Ma) (a) Diapiric gnelss domes Tonalite–trondhjemite magmatism MOR Sea level Archean VMS deposits Island arc Algoma-type banded-iron formation Island arc Witwatersrand type Au–U placer deposits Foreland basin Archean “lode-gold” or orogenic gold Major suture Craton Superior-type banded-iron formations Shield Greenstone belts Komatite-hosted Ni–Cu deposits Major granite emplacement Passive margin CRATONIZATION (3100–2500 Ma) (b) EARLY STAGES: 1 LATER STAGES: 2 Figure 6.9 (a) Early stage of Archean crustal evolution, termed intra-oceanic shield formation, occurring in the interval 3600–3100 Ma. Mineralization in this period seems to have been restricted mainly to ocean ?oor exhalative processes and formation of Algoma-type banded iron-formations. (b) Later stage of Archean crustal evolution, termed intra- and inter-continental craton formation, occurring between 3100 and 2500 Ma. Mineralization processes were varied and resulted in the formation of many important ore deposit types, including lode- or orogenic gold, VMS Cu–Zn, paleoplacer Au–U, Algoma- and Superior-type iron ores and komatiite-hosted Ni–Cu deposits (after De Wit et al., 1992). ITOC06 09/03/2009 14:34 Page 325derived from either metamorphic devolatilization or late-orogenic magmatism. This resulted in the formation of the voluminous and characteristic styles of orogenic gold mineralization which are typical of most late Archean granite–greenstone terranes worldwide. Examples include important deposits such as the Golden Mile, Kalgoorlie district of Western Australia, the Hollinger– McIntyre deposits in the Abitibi greenstone belt, the Sheba–Fairview deposits of the Barberton greenstone belt, and the Freda–Rebecca mine in Zimbabwe. Early intracratonic styles of sedimen- tation, often in a foreland basinal setting, gave rise to concentrations of gold and uraninite repres- ented by the Witwatersrand basin in South Africa. At least some of this mineralization is placer in origin and was derived by eroding a fertile Archean hinterland. The passive margins to these early continents would have developed stable platformal settings onto which laterally extensive Superior type banded iron-formations could have been deposited. A very signi?cant period for depo- sition of iron ores such as those of the Hamersley and Transvaal basins of Western Australia and South Africa respectively, as well as the Mesabi range of Minnesota, seems to have been around the Archean–Proterozoic boundary at 2500 Ma. 6.5.2 The Proterozoic Eon The period of time around 2500 Ma represents a major transition in the nature of crustal evolu- tion, involving changes in the volume and com- position of the continents, tectonic regimes, and atmospheric make-up. It is also clear from secular metallogenic patterns (Figure 6.1) that these evo- lutionary changes affected ore-forming processes and characteristics. The Proterozoic Eon spans a vast period of geological time, from 2500 to 540 Ma, including the period between 2000 and 1000 Ma that was marked by a relative paucity of orogenic and/or magmatic-hydrothermal deposit types, but abundant ores hosted in intracontin- ental sedimentary basins and anorogenic igneous complexes (Figure 6.1). The reasons for this pat- tern are multifaceted and complex, but, as a ?rst order approximation, are related to a higher degree of continental stability and the existence of major land masses which amalgamated and dispersed in relatively long-lived Wilson cycles (Windley, 1995). A substantial volume of contin- ental crust must have been in existence by the beginning of the Proterozoic Eon (Figure 6.2) and it is widely held that the ?rst supercontinent came into existence during this time, although its form and evolution are still largely speculative (Piper, 1976; Hoffman, 1988; Park, 1995). The Rogers model (Figure 6.7) envisages that, in addition to Ur and Arctica, the Paleoprotero- zoic witnessed continuing amalgamation of land masses to form Atlantica (from about 2000 Ma and comprising parts of west Africa and South America; Figure 6.10a) and Baltica (the Paleopro- terozoic basement to what is now western Eur- ope). Further amalgamation of two (i.e. Arctica and Baltica) of the four continental land masses in existence at this time is believed to have occurred at about 1500 Ma, to form a new continent that Gower (1990) and Rogers (1996) called Nena (an acronym for Northern Europe and North America). Figure 6.10b shows a possible reconstruction of Nena, at about 1500 Ma, and is very similar to a scheme proposed by Park (1995) in which Laurentia and Baltica were joined together between 1900 and 1500 Ma. The process of con- tinental amalgamation continued through the Mesoproterozoic, until approximately 1000Ma, when virtually complete consolidation gave rise to the formation of a single supercontinent, now widely referred to as Rodinia (McMenamin and McMenamin, 1990). Rodinia formed largely by accretion of Ur and Atlantica to Nena (Figure 6.7) along a major, almost continuous (in some recon- structions) suture zone known, at least in North America, as the Grenville orogeny. In the Neopro- terozoic, and certainly by about 700 Ma, Rodinia was broken apart again, mainly along two major rifts. One of these separated Atlantica from Nena, along a previous suture, to form two fragments known thereafter as Laurentia and west Gond- wana. The other split Nena internally to separate Laurentia from east Gondwana. Figure 6.11a shows a con?guration for Rodinia at 1000–700 Ma and Figure 6.11b illustrates a possible break-up scheme in which Laurentia has extricated itself, leaving a recombined west and east Gondwana at the end of 326 PART 4 GLOBAL TECTONICS AND METALLOGENY ITOC06 09/03/2009 14:34 Page 326ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 327 the Proterozoic Eon. This con?guration, although still largely conjectural and the subject of con- tinual modi?cation and further research (Dalziel et al., 2000), sets the scene for the subsequent pattern of crustal evolution and global tectonics in the Phanerozoic Eon. In summary, crustal evolution through the Proterozoic Eon was characterized by intermit- tent continental growth (primarily at around 2000 and 1000 Ma; Figure 6.2) and amalgamation which ultimately gave rise to the existence of a stable and long-lived supercontinent in the Neoprotero- zoic Era. Although geological processes were not signi?cantly different from those of other periods of Earth history, the interval between 2000 and 1000 Ma is typi?ed by ore deposits related to anorogenic magmatism and intracontinental basin deposition (Figure 6.1). These ore-forming pro- cesses point to a pattern in which a substantial part of Proterozoic crustal evolution was charac- terized by relatively long-lived periods of contin- ental stability. Continental stability at this time may have been promoted by an asymmetry in the distribution of crustal type, where a largely West Africa (a) (b) West Nile Congo- Kasai Guyana Brazil Tanzania Rio de la Plata Enlarged Atlantica Siberia Greenland Baltica North America East Antarctica Nena: 1.5 Ga Atlantica: 2.0 Ga Sao Francisco ~ Figure 6.10 (a) Map showing a possible con?guration for Atlantica at 2000 Ma, comprising mainly remnants of Archean crust in Brazil and west-central Africa. (b) Map showing the con?guration envisaged for the possible amalgamation of Arctica, Baltica, and east Antarctica to form Nena at about 1500 Ma. Both maps are after Rogers (1996). ITOC06 09/03/2009 14:34 Page 327Baltica Phanerozoic belts 800–500 Ma belts 1000 Ma belts (Grenville) pre-1000 Ma cratons West Africa Amazonia N NENA ATLANTICA Siberia Laurentia Congo 0° East Antarctica Australia ARCTICA Rift Rift UR (a) India Kalahari (b) N Iapetus Ocean Baltica Laurentia 0° Proto Pacific Ocean Amazonia Congo West Africa WEST GONDWANA Kalahari EAST GONDWANA East Antarctica Plate India ~750 Ma ~500 Ma ITOC06 09/03/2009 14:34 Page 328ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 329 oceanic hemisphere was antipodal to a static con- tinental domain. This feature might have been further accentuated, at around 1000–700 Ma, by the presence of a Rodinia. A similar situation is suggested to have applied during the existence of the Pangean supercontinent in Permian–Triassic times (Nance et al., 1986). The metallogenic con- texts of the three Proterozoic eras are discussed in more detail below. The Paleoproterozoic Era (2500–1600 Ma) From a metallogenic viewpoint, the period of earth history between 2500 and 1600 Ma is very signi?cant because of the major changes that occurred to the atmosphere, especially the rise in atmospheric oxygen levels at around 2200 Ma (Figure 6.3). Prior to this time, the major oxygen sink was the reduced deep ocean where any pho- tosynthetically produced free oxygen was con- sumed by the oxidation of volcanic gases, carbon, and ferrous iron. In this environment banded iron-formations, as well as bedded manganese ores, developed, as is evident from the widespread preservation of both Algoma and Superior type iron deposits. The increase in ferric/ferrous iron ratio in the surface environment that accompanied oxyatmoinversion at 2200 Ma, and the accompa- nying depletion in the soluble iron content of the oceans at around this time, resulted in few BIFs forming after this time. The stability of easily oxid- izable minerals such as uraninite and pyrite is also to a certain extent dependent on atmospheric oxygen levels and it is, therefore, relevant that major Witwatersrand-type placer deposits did not form after about 2000 Ma. Besides these non- recurrent changes, which only affected oxygen- sensitive ore-forming processes, the pattern of metallogeny in the Paleoproterozoic followed normal global tectonic constraints. The continent of Ur remained tectonically quiescent throughout much of the Proterozoic period although it witnessed episodic growth at around 2000 Ma (possible amalgamation with the Yilgarn and Zimbabwe cratons) and again at about 1500 Ma (amalgamation with parts of northern India and central Australia; Figure 6.7). This early stability is re?ected in the widespread deposition and preservation of banded iron-formations along shallow continental platforms at the Archean– Proterozoic boundary, already mentioned. On the Zimbabwe craton, rifting at around 2500 Ma gave rise to intrusion of the Great Dyke, with its signi- ?cant Cr and PGE reserves. At 2060 Ma on the Kaapvaal craton, the enormous Bushveld com- plex with its world-class PGE, Cr, and Fe–Ti–V reserves was emplaced, as was the Phalaborwa alkaline complex with its contained Cu–P–Fe–REE mineralization. The period between 2000 and 1800 Ma, however, was characterized by a global orogeny, largely accretionary in nature (Windley, 1995), which also affected Ur. The Australian Barramundi and southern African Kheis orogenies are events which contributed to the growth of Ur and, in the latter region, for example, gave rise to the Haib porphyry Cu deposit and small MVT- type Pb–Zn deposits along the western edge of the Kaapvaal craton. A further long period of cratonic stability ensued and Ur was subjected to rifting and intracontinental sedimentation. The period between 1700 and 1600 Ma saw the deposition of large dominantly clastic sedimentary basins that host the world-class SEDEX Pb–Zn ores of eastern Australia (Mount Isa, Broken Hill, and McArthur River) and South Africa (Aggeneys and Gamsberg). Elsewhere on Ur at this time, anoro- genic magmatism was also occurring, importantly in the form of the 1600 Ma Roxby Downs granite– rhyolite complex, host to the enormous magmatic- hydrothermal Olympic Dam iron oxide–Cu–Au deposit in South Australia. A slightly different pattern of Paleoproterozoic metallogeny is evident with respect to the con- tinental fragments of Arctica and Baltica, which Figure 6.11 (Opposite) (a) Simpli?ed map showing a possible con?guration for Rodinia, comprising the amalgamated continental land-masses of Nena (previously Arctica and Baltica), Atlantica, and Ur, from about 1000 Ma. Rift lines and arrows show the way in which Rodinia might have started to break apart, from about 750 Ma onwards (details from Hoffman, 1991). (b) Simpli?ed map showing how fragments of Rodinia might have reconsolidated by the end of the Proterozoic and early Phanerozoic (at about 500 Ma) to form an early Gondwana con?guration. Both maps are after Rogers (1996). ITOC06 09/03/2009 14:34 Page 329had combined by about 1500 Ma to form Nena (Figure 6.7). Early cratonic stability of Arctica is evident in the deposition of the Huronian Super- group at 2450 Ma which contains the paleoplacer uranium ores of the Eliot Lake–Blind River regions of the Superior province, Canada. Both Arctica and Baltica were, however, subjected to extens- ive accretionary orogenies between 2000 and 1700 Ma. The Trans-Hudson, Yavapai–Mazatzal and Svecofennian orogenies, for example, produced signi?cant new crust within which volcanogenic massive sul?de Cu–Zn deposits such as Flin Flon in Canada, Jerome, Arizona, and the Skellefte (Sweden)–Lokken (Norway) ores of Scandanavia are preserved. Relative stability followed this period of orogenesis, during which time large intracontinental sedimentary sequences such as the Athabasca basin formed at around 1700 Ma. It should be noted that the very rich uranium ores in the latter are epigenetic and probably formed during several later episodes of ?uid ?ow between 1500 and 1000 Ma (Hecht and Cuney, 2000). The continent of Atlantica (Figure 6.10a) was consolidated only after 2000 Ma, subsequent to a major compressional event re?ected in west Africa as the Birimian orogeny at 2100–2000 Ma. This major crust-forming event gave rise to the important orogenic or lode-gold deposits of Ghana, such as Ashanti. Atlantica was relatively stable for the remainder of the Paleoproterozoic and saw the emplacement of anorogenic type granite mag- matism at around 1900–1800 Ma, with which the large iron oxide–Cu–Au deposits of the Carajas region, Brazil, are associated. The Mesoproterozoic Era (1600–1000 Ma) The period of geological time after the formation of Nena (i.e. the amalgamation of Arctica and Baltica) at around 1500 Ma appears to have been one of tectonic quiescence and continental stability which lasted for several hundred million years. It culminated at around 1000 Ma in an episode of widespread orogenic activity (the Grenville orogeny and its many analogues worldwide) which resulted in the ?nal amalgamation of the Rodinia supercontinent. Although this period of orogenesis affected the continental margins of Ur it appears to have contributed little to the forma- tion of mineral deposits. An exception is provided in South Africa by the magmatic Cu–sul?de ores associated with ma?c intrusions of the Okiep copper district in the 1060–1030 Ma Namaqua- land belt. In Nena, by contrast, enormous vol- umes of anorogenic magmatism, especially in the period 1500–1300 Ma, provided the host rocks to a number of very important deposits. In a belt stretching from southern California through Labrador into Scandinavia, numerous intrusions of gabbro–anorthosite host the large magmatic Fe–Ti (ilmenite) ore bodies of the Marcy massif in the Adirondacks and Lac Allard in Quebec. The same belt also contains alkali granite– rhyolite complexes which give rise to Fe–Au–REE resources such as those of the St Francois moun- tains of Missouri. The granite–rhyolite magmatic complexes were also eroded to form the sedi- ments of the 1440 Ma Belt basin in the northwest USA, host to the Sullivan Pb–Zn SEDEX deposit in British Columbia in Canada. In addition, intra- continental rifting at around 1100 Ma in Nena gave rise to the formation of the 2000 km long Keweenawan mid-continental rift, stretching from Michigan to Kansas and ?lled with a thick sequence of bimodal basalt–rhyolite volcanics overlain by rift sediments. The latter form the host rocks to the stratiform Cu–Ag White Pine deposit in Michigan. The Neoproterozoic Era (1000–540 Ma) The Neoproterozoic commenced with the forma- tion, at around 1000 Ma, of the supercontinent Rodinia, arguably the ?rst substantially consol- idated land-mass in Earth history. Rodinia was long-lived and only started to partially fragment after a static period of more than 250 million years. Substantial parts of what had been Rodinia then reconvened toward the end of the Protero- zoic (at 540Ma) to form the very substantial Gondwanan land-mass during the Pan-African orogeny (Figure 6.11b). Break-up of Rodinia started at around 750 Ma when east Gondwana (previously referred to as Ur) rifted away from the western edge of Laurentia to initiate the opening of the proto-Paci?c ocean. The remaining portion 330 PART 4 GLOBAL TECTONICS AND METALLOGENY ITOC06 09/03/2009 14:34 Page 330of Rodinia (i.e. Laurentia, west Gondwana, Siberia, and Baltica) drifted southwards, remaining intact while at the same time rotating clockwise (Torsvik et al., 1996). At about 650–600 Ma both Baltica and Siberia started to break away from Laurentia and the Iapetus Ocean formed. By the end of the Proterozoic Eon, the amalgamation of east and west Gondwana had taken place and this large continental mass was situated at polar latitudes and across the Iapetus seaway from an equatorially located Laurentia (Figure 6.11b). In detail, the assembly of Gondwana was long-lived and polyphase and occurred progressively from about 750 to 550Ma. The Neoproterozoic is, therefore, sometimes referred to as the period of two supercontinents and was notable for its protracted periods of continental amalgamation, enhanced freeboard, and tectonic stability. The time between about 750 and 550 Ma was also characterized by the development of at least two, and in places possibly four, major ice ages, one or more of which was near global in cover and extended to equatorial latitudes. The concept of the Neoproterozoic “Snowball Earth” (Harland, 1965; Hoffman et al., 1998) has important impli- cations for understanding climate change and, especially, for the proliferation and diversi?ca- tion of organic life at the Precambrian–Cambrian boundary. Global glaciations also have implica- tions for the nature and formation of ore deposits in the Neoproterozoic Era. The major ore deposits of the Neoproterozoic re?ect the conditions of continental stability, as well as the periods of near global ice cover and attendant anoxia, that prevailed at this time. The extensively developed ironstone ores of north- west Canada and South Australia, associated with the 750–725 Ma Rapitan and Sturtian glaciogenic rocks respectively, are considered to be the result of the build-up of ferrous iron derived from off- shore hydrothermal vents in the reduced ocean waters that accompanied the development of vast continental and oceanic ice sheets at this time. Receding glaciers and a return to more oxidizing conditions would have resulted in conversion of labile ferrous iron to insoluble ferric iron and precipitation of the latter from the ocean water column, together with clastic and glaciomarine detritus to form the ironstones (Lottermoser and Ashley, 2000). Even more oxidizing conditions would also have resulted in the precipitation of manganese oxides or carbonates in the succession. In a similar vein, the Precambrian–Cambrian boundary at around 540 Ma is also characterized by the ?rst major global phosphogenic event that resulted in the development of vast deposits of phosphatic sedimentary rock (phosphorites) in several parts of the world (Cook and Shergold, 1984). As with sedimentary iron ores, phosphorites re?ect the upwelling of deep, nutrient-rich ocean waters onto shallow continental shelves with the syn-sedimentary precipitation of carbonate– apatite (or collophane) onto the shelf ?oor. Although the actual formation of phosphate ores is a complex process (see Chapter 5), there is com- pelling evidence to suggest that the onset of phos- phogenesis at 540 Ma was related to conditions prevailing toward the end of the Neoproterozoic. Many phosphorite deposits worldwide immedi- ately overlie glaciogenic sediments, suggesting that upwelling of phosphorus-rich ocean waters was promoted by overturn of a stagnant ocean during the widespread blanketing of sea-ice associated with a Snowball Earth scenario. The Precambrian–Cambrian phosphogenic event also coincides with the proliferation and diversi?ca- tion of organic life and it is pertinent that a signi?cant proportion of organisms that evolved at this time developed calcium phosphate skeletal structures. Phosphorus is, in addition, a universal nutrient (unlike oxygen) and its concentration in the oceans at the end of the Proterozoic, and in the Cambrian, may also be linked to the evolution of organic life. The formation of the vast, stratiform Cu–Co clastic sediment hosted ores of the Central African Copperbelt is also considered to have formed in an environment in?uenced by the Snowball Earth. The host Katangan sediments were deposited on a fertile Paleoproterozoic basement in an intra- continental rift, the development of which over- lapped with both the Sturtian and Marinoan glacial events. The Grand and Petit Conglomerats of the Katangan sequence, for example, represent glaciogenic sediments capped by carbonates which are correlated, respectively, with the Sturtian ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 331 ITOC06 09/03/2009 14:34 Page 331and Marinoan events (Windley, 1995). In?ux of oxide-soluble Cu and Co, perhaps derived from the local basement, might have occurred as dia- genetic ?uids migrated along growth faults and through the basin during the postglacial stages of deposition. Precipitation of ore sul?des would have occurred when the metal-charged oxidized ?uids encountered reduced sediments or ?uids. The enormous proliferation and diversi?cation of organic life in the late Neoproterozoic also resulted in the ?rst substantial development of highly carbonaceous sediments. A good example of these are the Hormuz sediments that ?ank the eastern Arabian shield and which could represent some of the source rocks for the vast Mesozoic oil and gas ?elds of the Arabian Gulf (Chapter 5, Box 5.3; Windley, 1995). The Neoproterozoic is notable for the paucity of orogenic type deposits (Figure 6.1). Even the extensive Pan-African oro- genic belts representing the suturing of contin- ental fragments during Gondwanan assembly are strangely devoid of, for example, world-class volcanogenic massive sul?de deposits, the latter being relatively abundant in the Paleoproterozoic orogenic belts of the world. A few small examples do nevertheless occur, such as the deposits of the Matchless amphibolite belt in Namibia, Bleïda in Morrocco, and the Ducktown, Tennessee deposit (Titley, 1993). The global shortage of these ores in the Neoproterozoic perhaps re?ects either a preser- vation factor or simply lack of exploration success. 6.5.3 The Phanerozoic Eon By comparison with earlier eons, crustal evolu- tion during the Phanerozoic is well understood and there is a large measure of con?dence that accompanies the interpretation of tectonic, chemical, and biological processes over the past 540 million years. The latter half of this period, the Mesozoic and Cenozoic eras, has been accom- panied by a proli?c development of mineraliza- tion, the likes of which has probably not been seen since the late Archean bonanza, despite the in?uence of a preservation factor. Ore-forming processes and their relationships to the pattern of Earth evolution in the Phanerozoic are also reasonably well understood and additional details regarding some of the topics discussed below can be found in Nance et al. (1986), Larson (1991), Barley and Groves (1992), Titley (1993), Kerrich and Cassidy (1994), Windley (1995), and Barley et al. (1998). A signi?cant proportion of the Phanerozoic was characterized by geological processes that re?ect a Wilson cycle, namely the sequence of events that saw the dispersal of Gondwana in the early Paleozoic, followed by reamalgamation of contin- ental material to form the Pangean supercontin- ent, by the early Mesozoic. The remainder of the Phanerozoic has witnessed the start of another Wilson cycle, involving the dispersal of Pangea to form the present day continental geography. It seems likely that we are presently about half way through this Wilson cycle (Nance et al., 1986) and that it will be responsible for reassembling a signi?cant proportion of the continents over the next 100–200 million years (Figure 6.6). Con- tinued continental amalgamation in the future will extend processes currently taking place, such as the collision of the Indo-Australian plate (previously part of east Gondwana) with the com- bined Baltica–Siberia (now called the Eurasian plate) and closure of the Mediterranean Sea. As previously mentioned, break-out of Laurentia from Rodinia and rotation of Ur at around 725 Ma (Figure 6.11) led to the formation of a Gondwanan land-mass by the end of the Proterozoic. Dispersal of Gondwana in the Cambrian–Ordovician may have been facilitated by the earlier development of a superplume (Larson, 1991), evidence for which is seen in the formation of major dyke swarms in different parts of Gondwana at 650–580 Ma (Torsvik et al., 1996). The Iapetus Ocean formed as Laurentia drifted away from Baltica and Gondwana from the late Precambrian onwards. The pattern of continental fragmentation is illus- trated in Figure 6.12a and was accompanied by a 332 PART 4 GLOBAL TECTONICS AND METALLOGENY Figure 6.12 (Opposite) The paleogeographic patterns of Gondwanan and Pangean dispersal in the Phanerozoic Eon. (a) and (b) Two reconstructions in the early Ordovician and Silurian (modi?ed after Torsvik et al., 1996). (c) A reconstruction of Pangea at its peak amalgamation in the Permian–Triassic. (d) and (e) Two instants of Pangean dispersal in the mid-Cretaceous and Eocene (after Windley, 1995). ITOC06 09/03/2009 14:34 Page 332Ordovician (490 Ma) Siberia Equator Baltica 30S 60S SCB Silurian (420 Ma) (a) (b) Siberia Gondwana Baltica Laurentia 30°S Armorica-Iberian Massifs Bohemian Massif SCB (c) Permian–Triassic (250 Ma) Panthalassan Ocean Tethys Ocean 0 (d) Mid-Cretaceous (100 Ma) 0 0 (e) Eocene (50 Ma) Iapetus Gondwana Avalonia GONDWANAN DISPERSAL PANGEAN DISPERSAL PANGEA ITOC06 09/03/2009 14:34 Page 333decrease in continental freeboard, sea-level high- stand and marine transgression. This was followed by progressive basin closure, terrane accretion, and granitoid magmatism during the period extending from the Silurian through to the early Carboniferous (about 420–300Ma). This colli- sional phase commenced with the rapid con- sumption of Iapetus as Baltica and Avalonia (i.e. a small terrane comprising England, Wales, southern Ireland, and eastern Newfoundland) moved toward lower latitudes and collided with Laurentia in Silurian times (Figure 6.12b). The orogenies that re?ect this collision phase are referred to as the Caledonian in Scandanavia and Scotland/Ireland and Appalachian in the eastern USA. Continental amalgamation continued as Gondwana drifted northwards toward lower lat- itudes, consuming the Rheic Ocean during the Devonian and Carboniferous, and eventually colliding with Laurentia–Avalonia–Baltica in the late Carboniferous–Permian. These complex and polyphase collisions are referred to as the Variscan and Hercynian orogenies (the terms are essen- tially synonomous) in present day Europe and the Alleghanian orogeny in the eastern USA. Contin- ental amalgamation continued into the Permian with the accretion of Siberia to Baltica along the Urals suture, after which the Pangean supercon- tinent had essentially formed (Figure 6.12c). The break-up of Pangea commenced soon after the time of its maximum coalescence in the Triassic at around 230 ± 5 Ma (Veevers, 1989). It was, therefore, a very short-lived supercontinent compared to Rodinia, which appears to have endured during much of the early Neoproterozoic. One reason for this might have been the develop- ment of a superplume, or plumes, that existed for some 70 million years, in Carboniferous and Permian times (320–250 Ma), an interval that also coincided with a protracted period of constant reversed magnetic polarity (Larson, 1991). The effects are seen geologically by increased produc- tion of oceanic crust and global volcanism (e.g. the outpouring of the Siberian continental ?ood basalt province and also the central Atlantic mag- matic province), eustatic sea-level rise, increased atmospheric greenhouse gas production (with associated warming and organic proliferation), and enhanced deposition of organic-rich black shales. The ?rst land-masses to actually drift apart, more than 70 million years after initial plume- related rifting, were Laurentia from Gondwana in the Jurassic at around 180 Ma. West and east Gondwana also started to split at this time, with the development of the proto-Indian Ocean and outpouring of the Karoo igneous province. The opening of the Atlantic Ocean was particularly prevalent in the early Cretaceous (around 130 Ma) and this extensional phase was also marked by continental ?ood basalt volcanism of the Etendeka–Parana provinces in Namibia and Brazil respectively. Continued dispersal of Pangea was stimulated by a second superplume event in the mid-Cretaceous between 120 and 80Ma (Figure 6.12d). The evidence for this event is again seen in a 40 million year period of constant, normal magnetic polarity, as well as the expected increases in oceanic crust production, eustatic sea level, atmospheric temperatures, organic pro- ductivity, and black shale deposition. The very large oceanic plateaux of the western Paci?c (e.g. Ontong–Java) are a magmatic re?ection of this mid-Cretaceous event, as is the development of very signi?cant global oil reserves related to the surges in nutrient supply and organic productivity on the voluminous continental shelves formed by the rise in sea level (Larson, 1991). In post-plume times the late Cretaceous saw the ?nal separation of Indo-Australia from Antarctica (Figure 6.13e), propelling the former northwards and resulting in continent–continent collision with Eurasia and the formation of the Himalayan orogeny in the Cenozoic. The Alpine orogeny of southern Europe was more or less coeval with its Himalayan counterpart and resulted from the collision of the African and Arabian plates with Eurasia after consumption of the Tethyan Ocean. A very important component of Pangean break- up in the Mesozoic–Cenozoic is re?ected in the development of new crust that accompanied the Cordilleran and Andean orogenies along the western margins of North and South America (Windley, 1995). This huge and complex orogeny, extending along the ocean–continent interface of western Pangea (Figure 6.12c), occurred in response to subduction and translocation of the 334 PART 4 GLOBAL TECTONICS AND METALLOGENY ITOC06 09/03/2009 14:34 Page 334original Panthalassan Ocean (now the eastern Paci?c) beneath Laurentia (now North America), the Cocos plate (now central America), and seg- mented west Gondwana (now South America). An island arc lay to the west of North America during Pangean times and this was accreted onto the continental margin, together with numerous other exotic terranes, during the consumption of several thousand kilometers of the Paci?c beneath North America during the Mesozoic Era. The enormous continent-?oored magmatic arc that resulted from this subduction gave rise to the 130–80 Ma granite batholiths and felsic volcanic rocks (such as the Sierra Nevada batholith) of the Cordillera. Continued subduction, albeit at a shallower angle than previously, and crustal thickening during the early Cenozoic, gave rise to in-board extension and exhumation of metamor- phic core complexes. Ongoing subduction during the Laramide orogeny (80–40 Ma) continued to feed the magmatic arc and numerous, metallo- genically important I-type granite batholiths were emplaced at this time. As subduction waned in the late Eocene–Oligocene the magmatic arc migrated westwards, resulting in continued calc–alkaline magmatism. Thermal collapse com- menced in the Neogene and the resulting crustal extension ultimately led to the formation of the Basin and Range province. In South America, Andean evolution was analogous, but less complex, than further north, and was also characterized by a lesser degree of accretionary and extensional processes. Subduction of the Nazca plate beneath a Paleozoic sedimentary margin, in the late Cretaceous, gave rise to a volcanic arc, behind which back-arc sedimentation took place (Lamb et al., 1997). This arc was the site of protracted magmatism that built the Western Cordillera and gave rise to the volcanic edi?ces which host the very important porphyry and epithermal styles of polymetallic mineralization in Peru, Bolivia, and Chile. Continued subduction transferred com- pressional stresses in-board and created a thin- skin fold and thrust belt in what is now referred to as the Eastern Cordillera. In the late Oligocene, crust was shortened even more and uplift gave rise to rapid erosion of the mountain belt to form a thick sedimentary plateau of dominantly Miocene-aged gravel and red-bed sequences known as the Altiplano. Subduction-related volcanism continued throughout this period in the Western Cordillera, which was also elevated to form the high Andes and then eroded to contribute sedi- ment into the Altiplano basin as well as west- wards onto the coastal plain. Further to the east, magmatism commenced in the early Miocene as a result of convective removal of the basal litho- sphere and crustal melting beneath the Altiplano (Lamb et al., 1997). The buoyancy and elevated pro?le of the Andes is maintained by present day underthrusting of the Brazilian shield beneath the Eastern Cordillera. Tectonic cycles and metallogeny The relatively well de?ned global tectonic cycles of the Phanerozoic Eon, summarized above, are also clearly re?ected in secular metallogenic trends. Titley (1993), for example, noted that the distribution of stratabound ores (i.e. volcanogenic massive sul?de, clastic sediment hosted Pb–Zn (SEDEX), and Cu (red-bed) deposit types) could be related to the tectonic cycles of Pangean amal- gamation and break-up (Figure 6.13). Preferential development of VMS deposits appears to be asso- ciated with periods in the Wilson cycle of elevated sea levels (highstand) associated with continental dispersal, namely, after Gondwana break-up in the early Paleozoic and in post-Pangean Mesozoic times. This association was considered to re?ect the processes of rifting, enhanced ocean crust production, and hydrothermal exhalation that accompany continental dispersal. Similar patterns are also evident in the accumulation of organic- rich shales and oolitic ironstone ores, which pref- erentially occur in the same two intervals, namely after Gondwana break-up in the Ordovician– Devonian and in post-Pangean Jurassic–Cretaceous times (Figure 6.13). In these cases, increased exhalative activity, carbon dioxide production, global warming, and organic productivity are interrelated processes which result in suitable conditions for black shale and ironstone precipita- tion in the oceans. In contrast, clastic sediment hosted base metal ores of the SEDEX Pb–Zn–Ag and red-bed Cu types tend to form at different ORE DEPOSITS IN A GLOBAL TECTONIC CONTEXT CHAPTER 6 335 ITOC06 09/03/2009 14:34 Page 335400 0 600 (meters) 500 400 300 200 100 WILSON CYCLE SEA LEVEL Wilson cycle FRAGMENTATION ASSEM- BLY DISPERSAL STASIS FRAGMENTATION P A N G E A Fossil fuels Petroleum Coal (Red-bed) Cu(Ag) (SEDEX) Pb-Zn-Ag Volcanogenic massive sul?des Clastic- hosted Marine black shale Oolitic Fe Superplumes 600 500 400 SIL ORD DEV 300 CARB 200 PER TR JUR CRET 100 CEN 0 Ma CAMB Ma Figure 6.13 Occurrences (in t