FRONTIERS IN GEOFLUIDS Edited by
Bruce Yardley, Craig Manning and Grant Garven
A John Wiley & Sons, Ltd., Publication
Frontiers in Geofluids
FRONTIERS IN GEOFLUIDS Edited by
Bruce Yardley, Craig Manning and Grant Garven
A John Wiley & Sons, Ltd., Publication
This edition first published 2011 © 2011 by Blackwell Publishing Ltd Originally published as Volume 10, Numbers 1–2 of Geofluids Blackwell Publishing was acquired by John Wiley & Sons in February 2007. Blackwell’s publishing program has been merged with Wiley’s global Scientific, Technical and Medical business to form Wiley-Blackwell. Registered Office John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK Editorial Offices 9600 Garsington Road, Oxford, OX4 2DQ, UK The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK 111 River Street, Hoboken, NJ 07030-5774, USA For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell. The right of the author to be identified 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. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Library of Congress Cataloguing-in-Publication Data has been applied for. ISBN 978-1-4443-3330-5 A catalogue record for this book is available from the British Library. This book is published in the following electronic formats: ePDF 978-1-4443-9488-7; Wiley Online Library 978-1-4443-9490-0; ePub 978-1-4443-9489-4 Set in 9/12pt ITC Galliard by SPi Publisher Services, Pondicherry, India
1
2011
CONTENTS
List of Contributors
vii
Frontiers in geofluids: introduction G. Garven, C. E. Manning and B. W. D. Yardley
1
Aqueous fluids at elevated pressure and temperature A. Liebscher
3
Thermodynamic model for mineral solubility in aqueous fluids: theory, calibration and application to model fluid-flow systems D. Dolejš and C. E. Manning
20
Metal complexation and ion association in hydrothermal fluids: insights from quantum chemistry and molecular dynamics D. M. Sherman
41
Role of saline fluids in deep-crustal and upper-mantle metasomatism: insights from experimental studies R. C. Newton and C. E. Manning
58
Potential of palaeofluid analysis for understanding oil charge history J. Parnell Spatial variations in the salinity of pore waters in northern deep water Gulf of Mexico sediments: implications for pathways and mechanisms of solute transport J. S. Hanor and J. A. Mercer Faults and fault properties in hydrocarbon flow models T. Manzocchi, C. Childs and J. J. Walsh
73
83
94
Hydrostratigraphy as a control on subduction zone mechanics through its effects on drainage: an example from the Nankai Margin, SW Japan D. M. Saffer
114
The interplay of permeability and fluid properties as a first order control of heat transport, venting temperatures and venting salinities at mid-ocean ridge hydrothermal systems T. Driesner
132
Using seafloor heat flow as a tracer to map subseafloor fluid flow in the ocean crust A. T. Fisher and R. N. Harris
142
The potential for abiotic organic synthesis and biosynthesis at seafloor hydrothermal systems E. Shock and P. Canovas
161
vi
Contents
Permeability of the continental crust: dynamic variations inferred from seismicity and metamorphism S. E. Ingebritsen and C. E. Manning
193
Hydrologic responses to earthquakes and a general metric Chi-Yuen Wang and Michael Manga
206
The application of failure mode diagrams for exploring the roles of fluid pressure and stress states in controlling styles of fracture-controlled permeability enhancement in faults and shear zones S. F. Cox
217
Rates of retrograde metamorphism and their implications for crustal rheology B. W. D. Yardley, D. E. Harlov and W. Heinrich
234
Fluids in the upper continental crust Kurt Bucher and Ingrid Stober
241
Fluid-induced processes: metasomatism and metamorphism A. Putnis and H. Austrheim
254
Fluid flows and metal deposition near basement ⁄cover unconformity: lessons and analogies from Pb–Zn–F–Ba systems for the understanding of Proterozoic U deposits M.-C. Boiron, M. Cathelineau and A. Richard
270
Magmatic fluids immiscible with silicate melts: examples from inclusions in phenocrysts and glasses, and implications for magma evolution and metal transport Vadim S. Kamenetsky and Maya B. Kamenetsky
293
Index
312
CONTRIBUTORS
H. Austrheim Physics of Geological Processes, University of Oslo, Oslo, Norway
J. S. Hanor Department of Geology and Geophysics, Louisiana State University, Baton Rouge, LA, USA
M.-C. Boiron G2R, Nancy Université, CNRS, CREGU, Vandoeuvre lés Nancy, France
D. E. Harlov Section 3.3, Chemistry and Physics of Earth Materials, Deutsches GeoForschungsZentrum, Telegrafenberg, Potsdam, Germany
Kurt Bucher Institute of Geosciences, Geochemistry, University of Freiburg, Freiburg, Germany P. Canovas GEOPIG, School of Earth & Space Exploration Arizona State University, Tempe, AZ, USA M. Cathelineau G2R, Nancy Université, CNRS, CREGU, Vandoeuvre lés Nancy, France C. Childs Fault Analysis Group, UCD School of Geological Sciences, University College Dublin, Dublin, Ireland S. F. Cox Research School of Earth Sciences, The Australian National University, Canberra, ACT, Australia D. Dolejš Bayerisches Geoinstitut, University of Bayreuth, Bayreuth, Germany and Institute of Petrology and Structural Geology, Charles University, Praha, Czech Republic
R. N. Harris College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR, USA W. Heinrich Section 3.3, Chemistry and Physics of Earth Materials, Deutsches GeoForschungsZentrum, Telegrafenberg, Potsdam, Germany S. E. Ingebritsen US Geological Survey, Menlo Park, CA, USA Maya B. Kamenetsky ARC Centre of Excellence on Ore Deposits and School of Earth Sciences, University of Tasmania, Hobart, Tas., Australia Vadim S. Kamenetsky ARC Centre of Excellence on Ore Deposits and School of Earth Sciences, University of Tasmania, Hobart, Tas., Australia
T. Driesner Department of Earth Sciences, ETH Zurich, Switzerland
A. Liebscher Centre for CO2 Storage, Helmholtz Centre Potsdam, German Research Centre for Geosciences GFZ, Telegrafenberg, Potsdam,Germany
A. T. Fisher Earth and Planetary Sciences Department and Institute for Geophysics and Planetary Physics, University of California, Santa Cruz, CA, USA
Michael Manga Department of Earth and Planetary Science, University of California, Berkeley, CA, USA
G. Garven Department of Geology, Tufts University, Medford, MA, USA
C. E. Manning Department of Earth and Space Sciences, University of California, Los Angeles, CA, USA
viii Contributors T. Manzocchi Fault Analysis Group, UCD School of Geological Sciences, University College Dublin, Dublin, Ireland J. A. Mercer Department of Geology and Geophysics, Louisiana State University, Baton Rouge, LA, USA R. C. Newton Department of Earth and Space Sciences, University of California, Los Angeles, CA, USA J. Parnell Department of Geology and Petroleum Geology, University of Aberdeen, Aberdeen, UK A. Putnis Institut für Mineralogie, University of Münster, Münster, Germany A. Richard G2R, Nancy Université, CNRS, CREGU, Vandoeuvre lés Nancy, France D. M. Saffer Department of Geosciences, The Pennsylvania State University, University Park, PA, USA
D. M. Sherman Department of Earth Sciences, University of Bristol, Bristol UK. E. Shock GEOPIG, School of Earth & Space Exploration and Department of Chemistry & Biochemistry, Arizona State University, Tempe, AZ, USA Ingrid Stober Institute of Geosciences, Geochemistry, University of Freiburg, Freiburg, Germany J. J. Walsh Fault Analysis Group, UCD School of Geological Sciences, University College Dublin, Dublin, Ireland Chi-Yuen Wang Department of Earth and Planetary Science, University of California, Berkeley, CA, USA B. W. D. Yardley School of Earth and Environment, University of Leeds, Leeds, UK
INTRODU CTION Frontiers in geofluids: introduc tion This set of papers was originally published electronically as a special double issue of Geofluids to mark the tenth anniversary of the launch of the journal. For this volume, we sought to bring together a collection of papers spanning a range of topics to which the role of fluids in the Earth is central. Geofluids was founded to help emphasise the common ground between fluid processes that take place in different geological settings, and to provide an outlet for research that considers the interactions of chemical and physical processes. While we cannot pretend to provide a comprehensive coverage of all the important recent advances, we are delighted to have been able to bring together some excellent and wide-ranging new science that continues in this tradition. The first four articles all concern our fundamental theoretical and experimental understanding of essentially aqueous fluids. Liebscher provides an overview of the properties of water-rich fluid systems and how these are affected by solutes, while noting the remaining limitations in the experimental database. Dolejs and Manning present the first comprehensive study to produce a more flexible alternative to the HKF model for aqueous electrolytes, better suited to the range of compositions and conditions encountered in nature, while Sherman shows how modern computational power means that some fundamental problems in natural fluid chemistry can be addressed from first principles using quantum chemistry and molecular dynamics. In the final article in this section, Newton and Manning review recent experimental results for lower crustal conditions and present new data to quantify the importance of dissolved salts for the solubility of the major rock –forming elements, Si and Al, and for a range of important Ca-minerals. The second group of articles relate to a specific geological setting where fluid processes are of the highest importance: sedimentary basins. Parnell provides a concise review of the use of hydrocarbon fluid inclusions to understand the evolution of reservoirs through time and the relationships between fluid stages and mineral cements. He shows in particular how this approach has contributed to understanding the oil charge history of the North Sea and UK Atlantic margin. Hanor and Mercer describe the behaviour of saline waters and their distribution in the Gulf of Mexico, and show how salinity differences arising through salt dissolution can dictate flow patterns. They
also explore the likely impacts of salt on the potential of the region as a source of methane hydrates. The article by Manzocchi, Childs and Walsh reviews how faults affect the flow of fluids, in particular hydrocarbons, in siliciclastic basins, and also comment on the extent to which current industry practice for evaluating the effects of faults is actually grounded in science. A third group of article deals with fluid processes in oceanic settings. Saffer has modelled the lateral variations along the Nankai margin of Japan and shown that large scale variations along strike in the taper angle of the accretionary wedge can be linked back to lithological variations from more turbidite-rich sequences to mudrocks. The lithology affects the development of fluid overpressure and the draining of the subduction zone fault, which in turn influences the overall geometry of the wedge. The interplay between permeability, heat flow and discharge characteristics at mid-ocean ridges is explored by Driesner. His results support some findings from terrestrial geothermal systems: high temperature discharges, and the highest fluid salinities, may be associated with low fluid fluxes, while large discharges at relatively low temperatures may in fact dominate the removal of heat. Fisher and Harris take three specific examples of mid-ocean ridge settings to explore the controls on heat loss. The relative importance of conductive heat loss is variable, and specific features of the basement geology can serve to target fluid flow and hence heat loss. Hydrothermal vents are also of likely significance for both abiotic and metabolic organosynthesis and this is explored by Shock and Canovas. Different patterns of mixing of seawater with different hydrothermal fluids can lead to different evolutionary paths, but in general, the mixing favours formation of organic compounds from inorganic reactants. Hence, microbes could produce components of biomolecules simply by catalysis of reactions that are already energetically favoured. A fourth group of articles deals with the continental crust. Ingebritsen and Manning present a crustal-scale overview of permeability and argue that while there is a power–law relation between permeability and depth in tectonically active continental crust, some regions exhibit markedly higher permeabilities, probably as transients, while stable crust may decay to lower permeability. The specific issue of the relationship of hydrologic response to earthquake activity is discussed by Wang and Manga. They
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
2 C. Garven et al. demonstrate that, in the intermediate and far-field, changes in groundwater flow are linked to changes in permeability which arise in response to cyclic deformation and oscillatory flow. The relationships between faulting and flow at depth is explored by Cox, who shows how fluid pressure and stress influence failure modes and hence the styles of permeability enhancement and vein development, in both mineralized and unmineralized systems. Fluid flow coupled to deformation often introduces water into high grade crystalline basement rocks which undergo retrogression. Yardley, Harlov and Heinrich present the results of experiments designed to measure the rate at which high grade rocks undergo retrogression under mid- to lower-crustal conditions, and conclude that water infiltrated along fine cracks is likely to be rapidly consumed. The article by Bucher and Stober addresses deep groundwaters found in crystalline basement rocks today by tunnelling and drilling. They argue that in areas of high relief such as the Alps, such waters are of relatively low TDS because of flushing by meteoric water, whereas much more saline brines may evolve where the hydraulic gradients are less. Migration of fluids can lead to mineralogical and chemical changes (metasomatism) in a wide variety of crustal settings, and Putnis and Austreheim explore some diverse examples of metasomatism on a range of scales. They are able to demonstrate that, while aqueous fluids partly act as a catalyst to permit minerals to react, they can also influence the course of the reaction through a thermodynamic role. The final section comprises two interdisciplinary articles that deal with ore deposits and draw on a range of aspects of fluids. Boiron, Cathelineau and Richard review the fluid systems that give rise to ore deposits near the unconformity between sedimentary basins and their underlying crystalline basement. They contrast the Proterozoic uncon-
formity uranium deposits with younger base metal deposits that develop in similar settings, and conclude that there are many similarities in both the nature of the fluids and the flow patterns that give rise to mineralization. Kamenetsky and Kamenetsky evaluate fluid processes at the other temperature extreme of ore formation, associated with magmatism. They present evidence from inclusions to document the development of immiscibility as magmas cool, and evaluate the importance of immiscibility for magma chamber processes, including degassing and the partitioning of metals. Although we have grouped the articles for convenience, we believe that the true value of the collection arises from the basic new data presented, from the insight into the interactions between physical and chemical processes, and from the opportunity they provide to take ideas developed in a field where particular types of observation or measurement may be possible to understand processes in different settings or at different times, where different types of data may be available. G. GARVEN1, C. E. MANNING 2 and B. W. D. YARDLEY3 1 Department of Geology, Tufts University, Medford, MA, USA; 2Department of Earth and Space Sciences, University of California, Los Angeles, CA, USA; 3 School of Earth and Environment, University of Leeds, Leeds, UK Corresponding author: B. W. D. Yardley School of Earth and Environment, University of Leeds, Leeds LS2 9JT, UK. Email: B. W. D.
[email protected]. Tel: +44 113 343 5227. Fax: +44 113 343 5259.
REVIEW Aqueous fluids at elevated pressure and temperature A. LIEBSCHER Centre for CO2 Storage, Helmholtz Centre Potsdam, German Research Centre for Geosciences GFZ, Telegrafenberg, Potsdam, Germany
ABSTRACT The general major component composition of aqueous fluids at elevated pressure and temperature conditions can be represented by H2O, different non-polar gases like CO2 and different dissolved metal halides like NaCl or CaCl2. At high pressure, the mutual solubility of H2O and silicate melts increases and also silicates may form essential components of aqueous fluids. Given the huge range of P–T–x regimes in crust and mantle, aqueous fluids at elevated pressure and temperature are highly variable in composition and exhibit specific physicochemical properties. This paper reviews principal phase relations in one- and two-component fluid systems, phase relations and properties of binary and ternary fluid systems, properties of pure H2O at elevated P–T conditions, and aqueous fluids in H2O–silicate systems at high pressure and temperature. At metamorphic conditions, even the physicochemical properties of pure water substantially differ from those at ambient conditions. Under typical mid- to lower-crustal metamorphic conditions, the density of pure H2O is qH2 O ¼ 0:61:0 g cm3 , the ion product Kw = 10)7.5 to approximately 10)12.5, the dielectric constant e = 8–25, and the viscosity g = 0.0001– 0.0002 Pa sec compared to qH2 O ¼ 1:0 g cm3 , Kw = 10)14, e = 78 and g = 0.001 Pa sec at ambient conditions. Adding dissolved metal halides and non-polar gases to H2O significantly enlarges the pressure–temperature range, where different aqueous fluids may co-exist and leads to potential two-phase fluid conditions under must mid- to lower-crustal P–T conditions. As a result of the increased mutual solubility between aqueous fluids and silicate melts at high pressure, the differences between fluid and melt vanishes and the distinction between fluid and melt becomes obsolete. Both are completely miscible at pressures above the respective critical curve giving rise to so-called supercritical fluids. These supercritical fluids combine comparably low viscosity with high solute contents and are very effective metasomatising agents within the mantle wedge above subduction zones. Key words: fluid–fluid interactions, fluid phase relations, fluid properties, fluid systems, metamorphic fluids, supercritical melts ⁄ fluids Received 23 July 2009; accepted 14 March 2010 Corresponding author: Axel Liebscher, Centre for CO2 Storage, Helmholtz Centre Potsdam, German Research Centre for Geosciences GFZ, Telegrafenberg, D-14473 Potsdam, Germany. Email:
[email protected]. Tel: +49 (0)331 288 1553. Fax: +49 (0)331 288 1502. Geofluids (2010) 10, 3–19
INTRODUCTION Aqueous fluids play a fundamental role in the geochemical evolution of the Earth. By transport and recycling of volatile components and different solutes from the atmosphere and hydrosphere through the solid interior of the Earth and back to the surface they chemically link the different spheres of the Earth on the local, regional and also global scale. As a mobile phase, they can transport heat very effi-
ciently by convection and contribute to the local and regional heat budget and heat distribution. The highly variable P–T–x regimes at the surface, in the crust and in the mantle generate equally variable, characteristic fluids with specific compositions and specific physical properties like density and viscosity. Aqueous fluids at elevated pressure and temperature conditions may form by infiltration of meteoric waters in geothermal and basinal systems (e.g., Hanor 1994; Arno´rsson et al. 2007), by diagenetic reac-
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
4 A. LIEBSCHER tions with connate fluids (Hanor 1994), by infiltration of seawater in oceanic hydrothermal systems (e.g., German & Von Damm 2003; Foustoukos & Seyfried 2007), by prograde metamorphic dehydration and decarbonation reactions as response to changes in P–T–x conditions (e.g., Yardley & Graham 2002; W. Heinrich 2007), and by liberation from crystallizing magmas (e.g., Cline & Bodnar 1994; C.A. Heinrich 2007). Due to the different fluid sources and different geologic environments with notably different P–T–x regimes, the composition of these aqueous fluids is highly variable. Meteoric fluids are almost pure water, which may gain considerable amounts of dissolved salts during diagenetic reactions. The salinity of aqueous basinal fluids ranges over five orders of magnitude from a few ppm in shallow meteoric regimes to more than approximately 40 wt% in evaporite-rich basins (Hanor 1994). In oceanic hydrothermal systems, the dominant fluid source is seawater, which can be modelled as a 3.3 wt% aqueous NaCl equivalent solution (NaCleq; Bischoff & Rosenbauer 1984). However, by interaction with the oceanic crust, the salinity of the generated hydrothermal fluids may range from only approximately 0.2 up to approximately 7.3 wt% NaCl (e.g., Oosting & Von Damm 1996; Lu¨ders et al. 2002). Prograde metamorphic dehydration reactions form highly variable saline fluids that may even reach salt saturation (Fig. 1B; Yardley & Graham 2002). However, the salinity of metamorphic fluids depends only little on metamorphic grade but is strongly linked to the protolith’s original setting. Metamorphism of oceanic or accretionary prism protoliths generally forms fluids with salinity below approximately 6 wt% NaCleq, whereas metamorphism of rocks from shallow marine or continental margin origin forms fluids that span the complete range from almost salt free up to approximately 60 wt% NaCleq (Fig. 1B; most (A)
salinity data for metamorphic fluids come from fluid inclusions and are estimated from the melting temperature of ice. They are usually given as ‘equivalent’ concentration of NaCl or NaCleq, referring to the salinity of a NaCl solution that would yield the same melting temperature of ice as the measured fluid inclusion). Metamorphic decarbonation reactions, triggerd by changes in pressure and temperature or by an infiltrating fluid itself, potentially supply CO2 to the system, force the fluids to un-mix and generate aquocarbonic fluids co-existing with variable saline H2O–salt fluids (e.g., Trommsdorff et al. 1985; Skippen & Trommsdorff 1986; Trommsdorff & Skippen 1986; W. Heinrich 1993, 2007). H2O–salt fluids are also reported from fluid inclusions in diamonds from the upper mantle (Navon et al. 1988; Izraeli et al. 2001). But at high pressure, also the solubility of silicates in aqueous fluids notably increases and silicates may form important solutes at these conditions. This may even lead to complete miscibility between aqueous fluids and silicate melts (e.g., Shen & Keppler 1997; Hack et al. 2007) and individual components continuously change their character from low, typical solute-like concentrations to high, typical solvent-like concentrations; finally, any major element may become an essential phase component of the fluid. This review aims at presenting some of the fundamental topics of aqueous fluids at elevated pressure and temperature. Given the vast number of different geological environments in which fluids evolve, each of which is unique in space, time and physicochemical properties, such a review is necessarily incomplete and the reader is kindly referred to the references given for more and detailed information on specific aspects of aqueous fluids. The review first describes the principal phase relations in one- and two-component fluid systems. Then, phase (B) 700
Geothermal systems
Magmatic
600
Magmatic hydrothermal systems Metamorphic
Basinal systems systems
on
Hydrated oceanic crust Mid ocean C ridge
m
is ph or am et m of n -P io gh at Hi b dr hy sla de ng us cti uo du tin sub
Mantle
ting
500 400 300 Oceanic
200
erpla
Und
Mantle melting
Basinal
100 Mantle
0 0
sa Ap lt sa pro tu x. ra tio n
Meteoric systems
Oceanic systems
Temperature (°C)
Volcanic systems
Metamorphic
10
20 30 40 50 Salinity (eq. wt% NaCl)
60
Fig. 1. (A) Schematic drawing showing the different geological fluid systems in crust and mantle. Modified from Liebscher & Heinrich (2007). (B) Compilation of salinity data for metamorphic fluids. Dark grey fields represent protoliths of shallow marine or continental margin origin, white boxes those from oceanic or accretionary prism origin. Hatched boxes indicate high-pressure metamorphic rocks. Redrawn and modified from Yardley & Graham (2002). Lighter grey fields indicate compositional ranges for fluids from magmatic, metamorphic, oceanic and basinal systems as compiled by Kesler (2005).
Aqueous fluids at elevated pressure and temperature 5 relations and properties of binary and ternary fluid systems, mostly based on experimental studies, are summarized followed by a review of the properties of pure H2O at elevated P–T conditions. Finally, aqueous fluids in H2O–silicate systems at high pressure and temperature are discussed.
PRINCIPAL PHASE RELATIONS IN ONE- AND TWO-COMPONENT FLUID SYSTEMS Aqueous geological fluids are rarely pure water at elevated P–T conditions and represent mixtures with several additional major components like salts (here used in the restricted sense of dissolved metal halides), non-polar gases and rock components like silica. In these multicomponent systems, phase relations become more complicated, fluid miscibility and immiscibility play an important role and concentrations of individual components may continuously change from solute-like to solvent-like or vice versa. In the following, the principal phase relations in two-component model fluid systems are described starting from the onecomponent case as exemplified for H2O (Fig. 2A). In any one-component system, the three phase states solid (in case of H2O ‘ice’), liquid and vapour (in case of H2O ‘steam’) have identical and fixed composition. All three co-exist at the invariant triple point from which the three univariant solid–liquid, solid–vapour and liquid– vapour equilibria emanate (Fig. 2A). In case of H2O, the triple point is at 0.00061 MPa ⁄ 0.01C. In one-component systems, co-existence of different fluid phases is exclusively restricted to the liquid–vapour equilibrium. However, although identical in composition, liquid and vapour differ in their physical properties like density, viscosity and electric permittivity. These properties show abrupt, discontinuous changes at the first-order liquid–vapour phase transition along the liquid–vapour equilibrium. The differences in physical properties, however, diminish along the Critical isochore (0.322 g cm–3)
(B)
100
H2O–
22.1 Liquid
Critical point
0.00061 s+
Pressure (MPa)
CO2
s+l
Pressure (MPa)
(A)
liquid–vapour equilibrium towards higher temperature and pressure and finally disappear at the critical point. The critical point of H2O is at 373.95C ⁄ 22.06 MPa (Wagner & Pruß 2002). Above critical temperature and pressure, neither changes in temperature nor pressure induce any phase transition and the physical properties of the homogeneous single phase fluid continuously change in response to changes in P and T. Addition of a second component to a one-component system adds composition as an additional degree of freedom and co-existing fluids not only differ in their physical properties but also in composition. Consequently, the invariant triple point, the univariant liquid–vapour equilibrium and the critical point of the one-component system turn into a univariant solid–liquid–vapour equilibrium, a divariant liquid–vapour field and a critical curve in the twocomponent system. The P–T range in which two fluids coexist may thereby greatly expand (Fig. 2B). The principal phase relations of two-component fluid systems and how they apply to different H2O–salt systems are shown in Fig. 3. For an in-depth presentation and discussion of the different phase topologies the reader is referred to Ravich (1974) and Valyashko (1990, 2004) and references given therein. A discussion of the principal phase topologies in the context of liquid–liquid immiscibility and the magmatic-hydrothermal transition is given by Veksler (2004). The overall phase topology of H2O–salt systems is determined by the pressure and temperature conditions of the univariant salt–liquid–vapour equilibrium, which emanates from the triple point of the salt endmember systems, relative to the pressure–temperature conditions of the critical curve. In so-called Type 1-systems, with a high and prograde solubility of the salt in H2O, the univariant salt– liquid–vapour equilibrium does not intersect the critical curve, which is continuous over the entire P–T–x space and connects the critical points of the pure systems (Fig. 3A, B). Between the univariant salt–liquid–vapour
H2O– CaCl2
50
NaCl Critical point H2O
v
Triple point
0.01 374 Temperature (°C)
CH4
apo Liquid-v
0 200
2O ur H
300 400 Temperature (°C)
500
Fig. 2. (A) Phase diagram for pure H2O showing the stability fields of the different phase states of H2O. Co-existing fluid phases are restricted to the liquid–vapour (‘l + v’) curve. Above the critical point, only one homogeneous single phase fluid exists. (B) P–T projection of the critical curves for the systems H2O–CO2, H2O–CH4, H2O–CaCl2 and H2O–NaCl. Hatched sides for of the critical curves mark direction of fluid un-mixing. Based on data by Takenouchi & Kennedy (1964) for H2O–CO2, Welsch (1973) for H2O–CH4, Shmulovich et al. (1995), Bischoff et al. (1996) and Zhang & Frantz (1989) for H2O–CaCl2, and Driesner & Heinrich (2007) for H2O–NaCl. Liquid–vapour curve for H2O from Wagner & Pruß (2002).
6 A. LIEBSCHER
(A)
Type 1a
Liquid c.p. (H2O)
t.p. (H2O)
v+l Water Steam
Ice + v+l
Steam
I
t.p. (salt)
lt
Temperature
Salt
T1
T3
T2
Tem
salt = NaCl, KCl, LiCl, MgCl2, CaCl2, MnCl2, NaOH, K2CO3
TI < T1 < Tt.p. (H2O)
Tt.p. (H2O) < T2 < Tc.p. (H2O)
Tc.p. (H2O) < T3 < Tt.p. (salt)
Water Ice
Pressure
Pressure
Ice + l l + Salt
Ice Steam
Pressure
Single phase fluid
l
Ice +v
t.p. Vapour
ture pera
Compo
sition
c.p. Liquid
Sa l t + l + v
Solid salt
l + Salt
v + Sa
H2O
t.p.
Ice
salt + v+l
v+l
Ice Steam v
c.p. (salt)
v+l
urve v = l lc
c.p. C
Pressure
Pressure
Water Ice
rit ic a
Critical cur ve
l
water steam
l + Salt v+l
v
l + Salt
l v+l
v
v+l
v + Salt
v
v + Salt
v + Salt
H2O
Salt
H2O
Salt
(B)
H2O
Salt
Type 1d Critical cur ve
c.p. (H2O)
Ice
R
l1 + l 2
c.p. (salt)
t.p. (H2O)
Steam v
H2O
N Steam
v + Sa
lt
Compo
l + Salt
sition
p Tem
t.p. Vapour
Temperature
N
v+l
Liquid
Solid salt
ure erat Salt = KH2PO4, Na2B4O7, Na2HPO4, UO2SO4
Salt
(C)
Type 2a Upper critical endpoint
Critical
Upper
C
Lower c.p.
t.p.
v
Ice Steam
c.p. Liquid
+
Salt + v+l
Liquid
=l
c.p. (salt)
l+
v+l
v ve
Lower critical endpoint
v
t.p. (H2O)
c.p. (H2O)
Critical endpoints
ur
Ice
Critical cur ve
salt + l +
Water
Pressure
Ice
t.p. (salt)
vv ++ l l
Ice
c.p.
l2 l1 + v+ Sa l t + l + v
t.p.
v + l1
Steam
cur ve l1 = l2
c.p. v = l1 R
Liquid
l1 + l2
Water
Pressure
Pressure
Water
Pressure
al itic Cr
lt Sa
Water
Solid salt
t.p.
Vapour
Steam t.p. (salt)
Temperature
Ice
v+l Steam v
H2O
v + Sa
lt
sition
ure
erat
l + Salt
p Tem
Compo
Salt = Al2O3, SiO2, BaCl2, BaSO4, K2SO4, Na2SO4, Na2CO3
Salt
Fig. 3. Schematic drawings of phase relations in water–salt systems of (A) Type 1a, (B) Type 1d and (C) Type 2a as P–T–x presentation and projections onto the P–T and P–x planes. Assignment of different salts to the different systems after Valyashko (2004). Drawn and modified according to Valyashko (1990, 2004).
Aqueous fluids at elevated pressure and temperature 7 equilibrium and the critical curve a two-phase fluid volume forms in P–T–x space. The univariant salt–liquid– vapour equilibrium may intersect the univariant ice–liquid– vapour equilibrium, which emanates from the triple point of pure H2O, forming an invariant point where liquid, vapour, ice, and salt co-exist (Fig. 3A). In so-called Type 2-systems, with a low solubility of the salt in H2O, the univariant salt–liquid–vapour equilibrium intersects the critical curve at the lower and upper critical endpoints (Fig. 3C). These general Type 1 and 2 phase topologies may get more complicated due to liquid–liquid immiscibility. Such liquid–liquid immiscibility gives rise to a number of principal phase topologies, which are presented and discussed in detail by Valyashko (1990, 2004). Not all of these principal phase topologies, however, have yet been found or are relevant for geological fluid systems. The most simple Type 1a system represents the binaries between H2O and the highly soluble salts like NaCl (e.g., Sourirajan & Kennedy 1962), KCl (Tkachenko & Shmulovich 1992; Dubois et al. 1994), CaCl2 (Tkachenko & Shmulovich 1992; Bischoff et al. 1996) but also NaOH (Urusova 1974) (Fig. 3A). Here, the critical curve is continuous over the entire P–T–x space and no liquid–liquid immiscibility occurs. The effects of liquid–liquid immiscibility are exemplified by Type 1d (Fig. 3B). Here, two separate critical curves form. One critical curve emanates from the critical point of the more volatile component (H2O in Fig. 3B) and shows critical behaviour between vapour and liquid 1, the other critical curve emanates from the critical point of the less volatile component (‘salt’ in Fig. 3B) and shows critical behaviour between liquid 1 and 2. H2O–salt systems of Type 1d include exotic salts like KH2PO4 (Marshall et al. 1981) and are not relevant for geological fluids. The simple Type 2a topology without liquid–liquid immiscibility is representative for binaries between H2O and sparingly soluble second components like SiO2, Al2O3 and other silicates, different sulphates and carbonates (Fig. 3C). Owing to the intersection of the univariant liquid–vapour– salt equilibrium with the critical curve at the lower and upper critical endpoints, lower and upper segments of the two-phase fluid volume form. In systems in which silicates are the additional second component, the high temperature part of the liquid–vapour–salt equilibrium is identical to the wet solidus and describes the silicate’s water saturated melting behaviour. As the upper liquid–vapour–salt equilibrium or wet solidus terminates at the upper critical endpoint, no discrete melting reaction is possible at pressure above the upper critical endpoint (see below).
BINARY AND TERNARY FLUID SYSTEMS The phase relations in binary and ternary fluid systems are simplified representations of natural fluids but form the framework for the more realistic higher component sys-
tems. The most important binary systems for aqueous fluids are the H2O–non-polar gas systems H2O–CO2 and H2O–CH4 and the H2O–salt systems H2O–NaCl, H2O–CaCl2 and H2O–KCl. Other components like the non-polar gas N2 and salts like LiCl, SrCl2 or MgCl2 are normally present only at the minor or even trace element level. Combining these binary systems yields the important ternary fluid systems H2O–NaCl–CO2, H2O–CaCl2–CO2 and H2O–NaCl–CH4. The principal phase relations in the systems H2O–CO2 and H2O–CH4 resemble each other and can be taken as proxies for other H2O–non-polar gas systems (Fig. 4A, B). Starting from the critical point of pure water, their critical curves initially extend towards lower temperature with only minor pressure dependence, then pass through temperature minima and extend to high pressure with only minor temperature increase. The temperature minima occur at 155–190 MPa and approximately 265C in the H2O–CO2 system (Takenouchi & Kennedy 1964) and approximately 100 MPa and 353C in the H2O–CH4 system (Welsch 1973). At temperatures below the critical curves, both systems show pronounced opening of the two-phase fluid region towards lower temperature, reflecting the very low solubility of non-polar solutes in water at low temperatures. A detailed and thorough discussion of phase relations within the H2O–CO2 system that also treats the important formation of clathrates at low temperatures is given by Diamond (2001). Most aqueous fluids contain important amounts of dissolved salts (see above), and water–salt systems therefore have attracted much interest. The H2O–NaCl system has been studied among others by Keevil (1942), Sourirajan & Kennedy (1962), Khaibullin & Borisov (1966), Urusova (1975), Bischoff & Rosenbauer (1984), Bodnar et al. (1985), Chou (1987), Bischoff et al. (1986), Rosenbauer & Bischoff (1987), Bischoff & Rosenbauer (1988), Bischoff & Pitzer (1989), Bischoff (1991) and Shmulovich et al. (1995). Based on the available experimental data, Driesner & Heinrich (2007) derived the most recent representation of the phase topology in the H2O–NaCl system up to 1000C ⁄ 220 MPa, and xNaCl = 0–1.0, where x is mole fraction. Their data form the basis for the H2O–NaCl phase relations shown in Fig. 4C. The critical curve in the H2O–NaCl system monotonously extends from the critical point of pure H2O towards higher pressure and salinity with increasing temperature, at least within reasonable crustal P–T conditions. Beside salinity also the density increases along the critical curve with increasing pressure and temperature (Fig. 4D; Urusova 1975; Chou 1987; Bischoff 1991). The data show that at elevated P–T conditions even the vapour in H2O–salt systems is considerably dense and notably differ from vapour at ambient conditions. For the role of dense vapour as an ore-forming fluid of its own the reader is referred to C.A. Heinrich (2007;
8 A. LIEBSCHER
50 c.p. H2O 0
50
0
0.4 0.6 XCO2 H2O–NaCl
140
0.8
0
1
0
Pressure (MPa)
450
40 c.p. H O 2 20
400 350
55
60
500
50
475
40
0
430 400
30 20
350 310
c.p. H2O
10 0
0.0001 0.01 XNaCl
1
0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Density (g cm–3)
(E)
(F)
H2O–CaCl2
CaCl2
200
90
500°C
CaCl2
0 200
NaCl KCl
300 400 500 600 700 Temperature (°C)
800
ur po Va
Onset of hydrolysis
40 30
a 2+
50
C
Pressure (MPa)
c.p. H2O
60
Cl–
Pressure (MPa)
100
70
id
NaCl KCl
150
Liqu
80
50
1
70
550 500
60
0.8
Critical curve
80
600
80
0.4 0.6 XCH4 H2O–NaCl
90
650
100
0.2
(D) Critical curve 70
120
0 0.000001
330
0
0.2
(C)
0
36
c.p. H2O
300
325
0
100
30
250 200
0
270 275
150
28
100 200
36
0
265
250
200
35 35 0 3
150
264
3
Pressure (MPa)
265 264
350 330
Pressure (MPa)
250 280
200
H2O–CH4 Critical curve
300
Critical curve
250
Pressure (MPa)
(B)
H2O–CO2
300
35
(A)
20 10
CaCl2 Ion imbalance 0 0.000001 0.0001 0.01 1 100 CaCl2, Ca2+, Cl– (eq per kg)
and references given therein). The system H2O–KCl has been studied by Keevil (1942), Hovey et al. (1990), Dubois et al. (1994) and Shmulovich et al. (1995). The critical curve in the H2O–KCl system as well as the corresponding phase relations closely resemble those of the H2O–NaCl system and the phase relations of natural KCl dominated fluids may be approximated by the H2O–NaCl system (Fig. 4E). However, the liquid–vapour–salt equilibrium is at slightly lower pressure in the H2O–KCl than in the H2O–NaCl system. H2O–alkaline earth salt systems were studied among others by Zhang & Frantz (1989), Tkachenko & Shmulovich (1992), Shmulovich et al. (1995), and Bischoff et al. (1996) for H2O–CaCl2 and Urusova & Valyashko (1983, 1984) and Shmulovich et al. (1995) for H2O–MgCl2. The data indicate that the critical curves in H2O-alkaline earth salt systems are at higher pressure than in the H2O-alkaline salt systems, whereas the
Fig. 4. Phase relations in the binary systems (A) H2O–CO2, (B) H2O–CH4, and (C) H2O– NaCl; (D) densities in the system H2O–NaCl; (E) P–T projections of the critical curves and liquid + vapour + salt equilibria in the systems H2O–NaCl, H2O–CaCl2 and H2O–KCl showing the limits for co-existing fluids; and (F) hydrolysis reaction in the system H2O–CaCl2. Thin lines in (A), (B), (C), and (E) are isotherms with temperature given in C. Based on data in (A) from Takenouchi & Kennedy (1964), in (B) from Welsch (1973), in (C) from Shmulovich et al. (1995) and Driesner & Heinrich (2007), in (D) from Urusova (1975), Chou (1987) and Bischoff (1991), in (E) from Keevil (1942), Sourirajan & Kennedy (1962), Ketsko et al. (1984), Chou (1987), Zhang & Frantz (1989), Hovey et al. (1990), Shmulovich et al. (1995), Bischoff et al. (1996), and in (F) from Bischoff et al. (1996).
liquid–vapour–salt equilibrium is at lower pressure. In H2O-alkaline earth salt systems the P–T range where a low salinity vapour co-exists with high salinity liquid therefore expands compared to H2O-alkaline salt systems. Fluid un-mixing into low salinity vapour and high salinity liquid in water–salt systems not only influences salinity and density of the co-existing fluid phases but may also change the availability of ligands and the acid or basic character of the co-existing fluids. In H2O–salt systems hydrolysis occurs according to the equilibrium xþ xH2 O þ Mxþ Cl x ¼ xHCl þ M ðOHÞx :
At ambient conditions, the equilibrium constant is such that reactant activities are significantly greater than those of the products and the amounts of HCl and Mxþ ðOHÞ x in the fluid are negligible. At elevated P–T conditions, however, the equilibrium constant may change in such a
Aqueous fluids at elevated pressure and temperature 9
(A)
160
Increasing salt
140 Pressure (MPa)
120 100 80 60 nNaCl/(nNaCl + nH O) 2 = 0.0 = 0.0715 250°C 150°C
40 20 0 0
0.04
(B)
0.08 XCO2
0.12
0.16
500°C/50 MPa XNaCl 0.6 Na C +N l aC l l+v
l
0.2
Single-phase Liquid or vapour
l+
0.4
0.2
v + NaCl l+v v
0
H2 O
0
0.2
0.4
0.6
0 1.0
0.8
XCO2
(C)
CO2
500°C/100 MPa NaCl, CaCl2
v + Salt
id lim
xC l;
0.2
Liqu
aC
l + v + Salt
b
aC
l2
l+
sal
t
Salt saturation in H2O–NaCl–CO2
xN
Two-phase fluid
Vapour limb
0 H2 O 0
0.2 xCO2
CO2
0.4
Single-phase fluid H2O–NaCl–CO2
(D)
H2O–CaCl2–CO2
50 MPa
550°C
l+v 500°C Critical point H2O-NaCl
450°C
400°C
Temperature
way that product activities are significantly greater than those of the reactants leading to notable amounts of HCl and Mxþ ðOHÞ x . In case of co-existing fluids, HCl preferentially fractionates into the vapour whereas Mxþ ðOHÞ x fractionates into the liquid. This gives rise to an HCl-enriched, potentially acidic vapour and a co-existing Mxþ ðOHÞ x -enriched, potentially basic liquid. This effect of fluid un-mixing has been experimentally studied by Bischoff et al. (1996), Vakulenko et al. (1989) and Shmulovich et al. (2002). Bischoff et al. (1996) analysed experimentally phase separated fluids in the system H2O– CaCl2 for Ca and Cl. Below 25.0 MPa at 400C and 58.0 MPa at 500C they observed an increasing ion imbalance between Ca2+ and Cl) in the vapour with Cl) concentrations being notably higher than Ca2+ concentrations (Fig. 4F). This observation reflects formation and preferential fractionation of HCl into the vapour. The amount of HCl produced by this reaction is remarkable and may reach 0.1 mol kg)1 in the vapour. The phase relations of the two binary H2O–non-polar gas and H2O–salt subsystems form the framework for the phase relations within the ternary H2O–salt–non-polar gas systems. The third binary subsystem salt–non-polar gas can safely be assumed as immiscible at all geologic relevant P–T conditions. All ternary H2O–salt–non-polar gas systems share some common features. Increasing salt concentrations decrease the solubility of the non-polar gas in H2O–salt mixtures (‘salting out’; Fig. 5A) and salt concentrations in H2O–non-polar gas mixtures are generally low. The P–T–x range of fluid immiscibility therefore greatly expands and extends to conditions where the binary subsystems may already be completely miscible. Fluid immiscibility may therefore prevail over the entire crustal P–T range (Heinrich et al. 2004). The H2O–NaCl–CO2 system is the most relevant ternary system and has been studied, among others, by Takenouchi & Kennedy (1965), Gehrig et al. (1986), Kotel’nikov & Kotel’nikova (1991), Joyce & Holloway (1993), Shmulovich & Plyasunova (1993), Gibert et al. (1998), Shmulovich & Graham (1999, 2004), Schmidt & Bodnar (2000) and Anovitz et al. (2004). The H2O–CaCl2–CO2 ternary has been experimentally determined by Zhang & Frantz (1989), Plyasunova & Shmulovich (1991), Shmulovich & Plyasunova (1993), and Shmulovich & Graham (2004) and the H2O–NaCl–CH4 ternary by Krader (1985) and Lamb et al. (1996, 2002).
Critical point H2O-CH4
Fig. 5. (A) Salting-out effect in the system H2O–NaCl–CO2 at 150 and 250C, and phase relations in the ternary systems (B) H2O–NaCl–CO2, (C) H2O–NaCl–CO2 and H2O–CaCl2–CO2 and (D) H2O–NaCl–CH4. Thin lines in (B) and dashed lines in (C) are only schematic to indicate principal phase relations. Data in (A) from Takenouchi & Kennedy (1964, 1965), in (B) from Anovitz et al. (2004), and in (C) from Kotel’nikov & Kotel’nikova (1991) and Zhang & Frantz (1989). (D) Redrawn and modified after Krader (1985).
350°C
NaCl l+v
CH4
X
Na
Cl
H2O
X CH
4
Liquid + vapour Single-phase fluid
10 A. LIEBSCHER The principal phase relations that occur in the H2O–NaCl– CO2 but also in the other H2O–salt–non-polar gas systems are exemplified in Fig. 5B based on the data by Anovitz et al. (2004) for 500C ⁄ 50 MPa. At these conditions, the H2O–NaCl binary shows liquid–vapour immiscibility and six principal stability fields can be distinguished: (i) At low NaCl concentrations, single phase H2O–CO2 vapour is stable; (ii) at H2O rich compositions, the NaCl poor H2O– CO2 vapour co-exists with a CO2 poor H2O–NaCl liquid, defining a vapour + liquid two-phase field; (iii) at higher NaCl concentrations, a single phase field of H2O–NaCl liquid with only minor amounts of CO2 forms; (iv)–(vi) at high NaCl concentrations the fluids are salt saturated and, depending on the H2O ⁄ CO2 ratio, liquid + salt, liquid + vapour + salt or vapour + salt co-exist. With increasing pressure, the miscibility gap within the H2O–NaCl binary closes and the single phase fluid field extend from salt-saturated CO2–poor H2O–NaCl fluids through ternary fluids at H2O-dominated conditions to NaCl poor H2O–CO2 fluids (Fig. 5C). The available data clearly show that the solubility of NaCl in H2O–CO2 vapour is very low and substantially decreases with increasing pressure and temperature. For instance, at 500C, e.g., xNaCl in H2O–CO2 vapour decreases from about 0.074 at 50 MPa to <0.025 at 500 MPa (Shmulovich & Plyasunova 1993; Anovitz et al. 2004), whereas at 200 MPa a temperature increase from 600 to 800C decreases xNaCl from about 0.052 to <0.032 (Kotel’nikov & Kotel’nikova 1991). In the ternary systems H2O–NaCl–CH4 and H2O–CaCl2–CO2 the solubility of the salt in H2O–non-polar gas vapour is even less and the field of co-existing fluids greatly expands (Fig. 5C; see also Liebscher 2007). The progression of the phase relations in ternary systems as function of T and x and the effects of opening and closure of two-phase fluid fields are shown for the system H2O–NaCl–CH4 at 50 MPa in Fig. 5D based on the data by Krader (1985). Remembering that increasing pressure favours fluid immiscibility in H2O–salt–non-polar gas systems, it is clear, that single phase fluids are restricted to very narrow P–T–x regimes and that most parts of the geological relevant P–T–x space are characterized by co-existing fluids with different physicochemical properties. Numerous equations of state (EOS) exist to describe the pressure–temperature–volume relations in the different fluid systems at elevated pressure–temperature conditions. For a thorough review of the different EOS and the underlying fundamentals, the reader is referred to the reviews by Ferry & Baumgartner (1987) and Gottschalk (2007) and to Prausnitz et al. (1999) and Sengers et al. (2000). EOS for H2O–non-polar gas systems are given among others by Kerrick & Jacobs (1981) and Duan & Zhang (2006) for H2O–CO2, Jacobs & Kerrick (1981) and Duan et al. (1992a,b) for H2O–CO2–CH4, and Saxena & Fei (1987, 1988), Belonoshko & Saxena (1992), Duan et al. (1992c,
1996), Shi & Saxena (1992) and Churakov & Gottschalk (2003a,b) for H2O–CO2–CH4 plus additional phase components. H2O–salt systems are covered by Anderko & Pitzer (1993a,b) for H2O–NaCl–KCl, Jiang & Pitzer (1996) for H2O–CaCl2 and Duan et al. (2006) for H2O–CaCl2 and H2O–MgCl2. EOS for ternary H2O–salt–non-polar gas systems are restricted to those by Duan et al. (1995, 2003) for H2O–NaCl–CO2 and H2O–NaCl–CO2–CH4. According to the authors, some of the EOS for H2O–nonpolar gas systems are valid to rather high pressure and temperature conditions, e.g., Duan & Zhang (2006) up to 2300C ⁄ 10 GPa and Churakov & Gottschalk (2003a,b) up to 1730C ⁄ 10 GPa. Unfortunately, those that include salts are applicable only up to mid-crustal pressure conditions of maximum 500 MPa (e.g., Duan et al. 1995, 2003).
PROPERTIES OF PURE H 2 O AT ELEVATED P–T CONDITIONS For the evaluation of the properties of aqueous fluids at elevated pressure and temperature conditions, knowledge of the fundamental properties density, dissociation behaviour, static dielectric constant and dynamic viscosity of pure water at elevated pressure and temperature conditions is of primary importance. The available experimental data as well as formulations for the calculation of these properties at elevated pressure and temperature are summarized below and shown in Fig. 6. Emphasis is given to metamorphic P–T conditions (grey fields in Fig. 6), which are here somewhat arbitrarily defined as P > 100 MPa and T > 150C. Metamorphic conditions are limited towards low temperature and high pressure by the lowest possible geotherm, which was assumed to 5 MPa ⁄ C. The shown boundaries of the different metamorphic facies are from Spear (1993). Some of the reviewed studies report the respective properties as a function of temperature and density while others use temperature and pressure. In Fig. 6, all properties are plotted as a function of temperature with isobars showing the pressure dependence. For data that were presented as function of temperature and density in the original papers density was converted to pressure by applying the IAPWS formulation for H2O by Wagner & Pruß (2002). For near-critical behaviour of H2O and aqueous systems, see Anisimov et al. (2004). Density At ambient conditions, liquid water has a density of approximately 1 g cm)3. With increasing pressure and temperature along the liquid–vapour equilibrium, the density contrast between liquid water and steam diminishes and at the critical point both approach the critical density of 0.322 g cm)3. Numerous EOS and molecular dynamics simulations describe the PVT properties of pure H2O.
Aqueous fluids at elevated pressure and temperature 11
(A)
(B)
1100
Isobars (MPa) 100
1000
250
900
500
1000
2000
Temperature (°C)
th
700 600
Not realiz ed on Ear
Temperature (°C)
800 A EA
500 Gr
400 10
Critical point
300 200
B
0.4 0.6 0.8 1.0 Density (g cm–3)
Temperature (°C)
800 700
Fernández et al. (1997)
E
400
B
A
600
Franck et al. (1990a,b) 255 MPa ~ 500 MPa 1.55 GPa
EA E
50 10
Gr
B
229 MPa 885 MPa
SGr Liquid + Vapour
100 0 0
10 20
tr ea
5
30 40 50 60 70 80 Static dielectric constant
50
10
Quist (1970) 100 200 MPa
Tanger & Pitzer (1989) Marshall & Franck (1981)
15 20 25 -log ion product KW
30
1100 Isobars (MPa) 100
900 800
200 300
700
Sengers & Kamgar-Parsi (1984)
G
600 A
500
Abramson (2007)
350
10 50
400
500 1000 1500
300
Critical point
200 liz ed on Ea r th
Liquid + Vapour
SGr
Gr B SGr
100
ze ali
No
200
Franck (1956) 350 670 970 MPa
t re No
Critical point
300
Pitzer (1983)
Critical point
Gr
1000
Shock et al. (1992) Heger (1969) Deul (1984) 50 MPa 100 MPa 250 MPa 500 MPa
G
400
A
500
0
1.4
(D) Isobars (MPa) 100 250 500 1000
900
500
1.2
Temperature (°C)
1000
EA
600
200
SGr
0
1100
700
100 0.2
25
G
300
100 0
50
ed liz ea rth t r Ea No on
Liquid + Vapour
100
900
E
800
(C)
Isobars (MPa) 200
1000 500
1000
G
50
1100
d
on E
Liquid + Vapour
ar th
0 90
0
0.0001 0.0002 0.0003 0.0004 0.0005 Dynamic viscosity (Pa s)
Fig. 6. Density (A), ion product (B), static dielectric constant (C) and dynamic viscosity (D) of pure water as function of temperature. Thin lines are isobars with pressure given in MPa. Shadowed fields represent range of metamorphic conditions with 100 MPa arbitrarily chosen as lower P-limit. Thick grey lines define the different metamorphic facies where SGr = sub-greenschist, Gr = greenschist, B = blueschist, EA = epidote-amphibolite, A = amphibolites, G = granulite and E = eclogite facies (after Spear 1993). Data in (A) from Wagner & Pruß (2002), other data sources given in the Figure. Stippled lines in (D) are eye-drawn through the data points of Abramson (2007).
Among others, these include those by Kerrick & Jacobs (1981), Halbach & Chatterjee (1982), Saul & Wagner (1989), Belonoshko & Saxena (1991) and Wagner & Pruß (2002). At elevated P–T conditions, the density of H2O ranges from approximately 0.3 g cm)3 up to >1.2 g cm)3 (Fig. 6A; calculated with the IAPWS formulation by Wagner & Pruß 2002). Only at low pressure the density may be < 0.3 g cm)3. Increasing temperature generally decreases density whereas increasing pressure has the opposite effect. In low-P ⁄ high-T settings as they occur in oceanic hydrothermal systems, volcanic systems and meta-
morphic systems with very high geothermal gradients H2O densities are below approximately 0.6 g cm)3. Under typical crustal metamorphic conditions, however, the density of H2O ranges between approximately 0.6 and 1.0 g cm)3. Only at rather high pressures, as in subduction zones or within the upper mantle, are conditions such that H2O densities exceed 1.2 g cm)3. Self-dissociation The dissociation equilibrium of water
12 A. LIEBSCHER H2 O ¼ Hþ þOH
Dielectric constant
is described by the equilibrium constant
The dielectric constant is of primary importance for the evaluation of the solvent properties of water. The attractive or repulsive force between any point charges q1 and q2 in a medium, e.g., ions in aqueous solution, that are separated by distance r is given by
K¼
aHþ aOH aH2 O
or by the ion product Kw ¼ aHþ aOH : At ambient conditions, the dissociation equilibrium of liquid water is strongly shifted to the left hand side, concentrations and activities of H+ and OH) are low and Kw = 1 · 10)14. With pH ¼ log aHþ this turns into the well known value of pH = 7 for neutral water. At elevated P–T conditions, however, Kw notably changes. The ion product along the liquid–vapour equilibrium has been experimentally determined among others by Bignold et al. (1971), Fischer & Barnes (1972), MacDonald et al. (1973) and Sweeton et al. (1974). Measurements at elevated pressure and temperature conditions above the critical point of water were performed by Franck (1956) at 350–970 MPa ⁄ 500–1000C and by Quist (1970) up to 555 MPa and 800C. Extreme conditions have been studied by Holzapfel & Franck (1966) at 4.5– 10.0 GPa ⁄ 600–1000C. Formulations for the change of Kw with pressure and temperature are given by Marshall & Franck (1981) up to 1000C ⁄ 1000 MPa, Pitzer (1983) for low densities up to approximately 0.1 g cm)3, Tanger & Pitzer (1989) up to 500 MPa ⁄ 2000C, and Bandura & Lvov (2006) up to 1.0 GPa ⁄ 1000C. Kw as function of pressure and temperature calculated by the formulations of Marshall & Franck (1981) and Tanger & Pitzer (1989) together with selected data by Franck (1956) and Quist (1970) are shown in Fig. 6B. The Marshall & Franck (1981) and Tanger & Pitzer (1989) data agree well for the pressure range between 100 and 500 MPa and reproduce the available experimental data by Quist (1970). However, at pressures below approximately 100 MPa, both formulations diverge and the formulation by Marshall & Franck (1981) predicts notably higher values for Kw than the Pitzer (1983) and Tanger & Pitzer (1989) formulations. In this low pressure, low density region the formulations of Pitzer (1983) and Tanger & Pitzer (1989) are probably more reliable (see discussion in Tanger & Pitzer 1989). Within the pressure range 200–1000 MPa, temperature has only a minor effect on Kw, whereas Kw generally increases with increasing pressure. Under normal mid- to lower-crustal P–T conditions, Kw ranges between approximately 10)7.5 and approximately 10)12.5. However, at very high pressures of the eclogite facies, Kw may even be substantially higher than 10)7.5. In line with this, Holzapfel & Franck (1966) determined Kw = 10)2.8 to 10)1.2 at approximately 7–9 GPa ⁄ 500–1000C.
F¼
q1 q2 4p"r 2
with e being the electric permittivity of the medium. The electric permittivity of the vacuum is typically designated e0 and the dimensionless ratio e ⁄ e0 then yields the static relative permittivity or static dielectric constant for any medium. A high dielectric constant corresponds to a high resistance of the medium to the transmission of an electric field. At ambient pressure and 25C liquid water has a notably high static dielectric constant of 78.4 (Ferna´ndez et al. 1995). This high static dielectric constant makes liquid water a good solvent for charged species at ambient conditions as it minimizes the electrostatic forces between the dissolved ions and prevents them to combine to crystals. However, with increasing pressure and temperature, the static dielectric constant of liquid water substantially decreases. Ferna´ndez et al. (1995) presented a database for the static dielectric constant of water and steam. It extends up to a temperature of 873 K and a pressure of 1189 MPa and covers the data available at that time. The static dielectric constant along the liquid–vapour equilibrium has been studied by Oshry (1949), Svistunov (1975), Lukashov (1981), Muchailov (1988) and Mulev et al. (1994). Along the liquid–vapour equilibrium, e of liquid water continuously decreases from approximately 55 at 100C to only approximately 9.7 at 370C, whereas e of co-existing steam only slightly increases from approximately 1.03 at 150C to approximately 3.02 at 370C. Above the critical point, e has been measured by Heger (1969) and Heger et al. (1980) at 25–500 MPa ⁄ 400–550C, Lukashov et al. (1975) at 23–58 MPa ⁄ 400–600C, Golubev (1978) at 23–39 MPa ⁄ 420–510C, Lukashov (1981) at 24–580 MPa ⁄ 400–600C, and Deul (1984) at 30–300 MPa ⁄ 400C. Based on the available experimental data, formulations and calculated values for e up to high P–T conditions are given by Quist & Marshall (1965) up to 1.55 GPa ⁄ 800C, Bradley & Pitzer (1979) up to 100 MPa ⁄ 350C, Pitzer (1983) up to 880 MPa ⁄ 927C, Franck et al. (1990a,b) up to 1.55 GPa ⁄ 1000C, Shock et al. (1992) up to 500 MPa ⁄ 1000C, Wasserman et al. (1995), and Ferna´ndez et al. (1997) up to 1.2 GPa ⁄ 600C. Fig. 6C reviews the experimental data by Heger (1969) and Deul (1984) together with isobars calculated by formulations of Shock et al. (1992) and Ferna´ndez et al. (1997). Also shown are values for the static dielectric constant for selected P–T conditions as calculated by formulations of Pitzer (1983) and
Aqueous fluids at elevated pressure and temperature 13 Franck et al. (1990a,b). For temperatures up to 1000C and pressures up to 500 MPa, i.e. the P-range largely covered by experimental data, the different formulations agree quite well. However, at pressures above 500 MPa, the different formulations diverge. At 800C, e.g., Ferna´ndez et al. (1997) predict a static dielectric constant of approximately 14 at 1.0 GPa, whereas Franck et al. (1990b) predict e 14 only at notably higher pressure of 1.55 GPa. At slightly higher temperature of 930C, Pitzer (1983) predicts e 11 at 885 MPa, roughly 120 MPa below the prediction of Ferna´ndez et al. (1997). This indicates a higher pressure dependence of e in Pitzer (1983) compared to Ferna´ndez et al. (1997) and Franck et al. (1990a,b). Despite these differences at high P conditions, the static dielectric constant of H2O generally decreases with increasing temperature but increases only slightly with increasing pressure. Because the effect of temperature is more pronounced than that of pressure, e decreases with increasing P–T conditions and range between approximately 8 and 25 at normal crustal P–T conditions. Only at low temperature of greenschist to sub-greenschist facies conditions or notably high pressure of blueschist to eclogite facies conditions e may exceed 25. But given the restricted P–T range covered by experimental data, any calculation of the static dielectric constant at pressures notably in excess of 500 MPa has to be done with great caution. Viscosity The transport properties of water strongly depend on its dynamic viscosity g, which also plays an important role for diffusion. Several studies have addressed the dynamic viscosity of water at low to moderate temperature and pressure (see Watson et al. (1980) for a review of available experiments at that time). The dynamic viscosity of water at elevated P–T conditions has been studied by Dudziak & Franck (1966) up to 350 MPa and 560C and by Abramson (2007) up to 6 GPa and 300C. Based on the available data, Watson et al. (1980) and Sengers & Kamgar-Parsi (1984) derived representative equations for g (Fig. 6D). At ambient conditions, liquid water has a dynamic viscosity of g = 0.001 Pa sec (Sengers & Kamgar-Parsi 1984). Along the liquid–vapour equilibrium, the viscosity of liquid water substantially decreases whereas that of steam only slightly increases, so that the critical dynamic viscosity of H2O is approximately 3.9 · 10)5 Pa sec. At temperature conditions above approximately 400C, the data suggest that at isobaric conditions the temperature effect on the dynamic viscosity is only minor whereas increasing pressure increases g. Except for low temperatures of sub-greenschist facies conditions, H2O has dynamic viscosities at normal crustal P–T conditions between 0.0001 and 0.0002 Pa sec. Although no data are available for upper blueschist and
eclogite facies, the data suggest that at these high pressure conditions the dynamic viscosity of H2O may range up to 0.0003 Pa sec (Fig. 6D).
H 2 O–SILICATE SYSTEMS AT HIGH P–T CONDITIONS Aqueous fluids and silicate melts are the two most important mobile phases in the Earth crust and mantle. H2O– silicate systems typically belong to Type 2 systems (see above), which are characterized by a discontinuous critical curve and lower and upper segments of the two-phase fluid volume that terminate at the lower and upper critical endpoints (see Fig. 7C). Up to moderate pressure and temperature conditions the solubility of silicates in aqueous fluids and of H2O in silicate melts is typically low and both phases fundamentally differ in their physicochemical properties. With increasing pressure, however, the solubility of H2O in silicate melts and of silicates in aqueous fluids increases and the differences between both phases finally vanish at the critical curve above which aqueous fluids and silicate melts are completely miscible giving rise to socalled ‘supercritical’ fluids. Here, only some key aspects of H2O–silicate systems at high pressure and temperature conditions are reviewed. A thorough review of this topic is given by Hack et al. (2007). Kennedy et al. (1962) provided the first experimental evidence for complete miscibility between aqueous fluids and silicate melts at high pressure and temperature within the system H2O–SiO2 and proposed an upper critical endpoint at 970 MPa ⁄ 1080C with a composition of approximately 25 wt% H2O ⁄ 75 wt% SiO2. This has been confirmed by Newton & Manning (2008). Shen & Keppler (1997) directly observed complete miscibility between aqueous fluids and silicate melts in the system H2O–albite and determined the location of the critical curve (Fig. 7A). Further experiments by Stalder et al. (2000) at near wet solidus conditions then located the upper critical endpoint in the system H2O– albite at approximately 1.5 GPa ⁄ 700C (Fig. 7A). The resulting phase relations within the system H2O–albite up to 1.7 GPa and T > 500C based on the data by Paillat et al. (1992), Shen & Keppler (1997) and Stalder et al. (2000) are shown in Fig. 7B. At pressure below approximately 1.5 GPa, the critical curve is at higher temperature than the wet solidus, which is defined as onset of melting by the reaction albite + vapour = liquid. However, with increasing pressure, the critical curve shifts to lower temperature, the liquid + vapour two-phase field narrows and at approximately 1.5 GPa the critical curve intersects the wet solidus at about 700C. Above approximately 1.5 GPa no discrete melting occurs but albite continuously dissolves in the fluid with increasing temperature giving rise to a continuous range from solute poor, almost pure H2O at low temperature to H2O poor silicate dominated liquids at
14 A. LIEBSCHER
(A)
(B) T (°C)
Upper critical endpoint
1.5
2500
2500
2000
2000 C
rit ur lc ica
Albite Liquid
1000
1000
5
0.
75
P
0.
) Pa
(G
900 Temperature (°C)
1300
us lid us so lid so
id us
ol
0 500
ab + l
y
1 0. 5 2 0.
Dr
ts We
Dr y s olidus
500
et W
ur vapo
Liquid
0.5
1500
ve
1.0
Albite +
Pressure (GPa)
1500
ab + v
Upper critical endpoint
1500 1000
0
1.
25
1.
500
ab + scf
5
1.
.7
25 1 NaAlSi3O8
(C)
l+v
50 75 mol % H2O
H2O
(D)
T (°C) 1600 1400
Critical curve
4.0
2
1200 l+v
t us We solid
800
1400
33 mol % H2O
Rock + l
us solid
1000
3
1000 800
Rock +v
Upper critical endpoint
Rock + scf
Pa)
P (G
1000
5.0
6.0 Eclogite
Log viscosity (Pa s)
Dr y
1200
1600
1 0
57 mol % H2O
–1
72 mol % H2O
–2
81 mol % H2O
–3 –4 6
20
40 60 mol % H2O
93 mol % H2O
800
80
H2O
8
10
12
14
Temperature (10 000/K)
Fig. 7. Water–silicate systems at high pressure and temperature. (A) P–T diagram showing location of the wet and dry solidus, critical curve and upper critical endpoint in the system H2O–albite. Wet solidus, filled and empty circles from Stalder et al. (2000), critical curve and diamonds from Shen & Keppler (1997) and dry solidus from Boyd & England (1963). (B) Phase relations within the system albite–H2O up to 1.7 GPa and T > 500C showing the termination of the water saturated solidus and the critical curve at the upper critical endpoint. Drawn and modified after Paillat et al. (1992), Shen & Keppler (1997) and Stalder et al. (2000) (ab = albite, v = vapour, l = liquid, scf = supercritical fluid). (C) Phase relations within the system eclogite–H2O at 4–6 GPa and T > 600C based on the data of Kessel et al. (2005a). Critical curve, dashed and stippled lines are hypothetical additions to the data given in Kessel et al. (2005a) to clarify phase relations (v = vapour, l = liquid, scf = supercritical fluid). (D) Arrhenius plot showing the viscosity of H2O–albite solutions as function of temperature and water content at high pressures >1.0 GPa. Data from Aude´tat & Keppler (2004), stippled lines calculated according to their Eqn (1).
high temperature. Addition of fluorine, boron and sodium to the H2O–albite system shifts the location of the critical curve to even lower pressure and temperature (Sowerby & Keppler 2002). Critical curves for the systems H2O–nepheline and H2O–jadeite have then been determined by Bureau & Keppler (1999). Termination of the wet solidus at the upper critical endpoint, however, not only occurs in simple H2O–mineral systems but is also observed in H2O–rock systems. Kessel et al. (2005a,b) studied the system H2O– potassium-free basalt at 4–6 GPa ⁄ 700–1400C. They observed eutectic and peritectic melting at pressures of 4 and 5 GPa, respectively, but no discrete melting at
6 GPa (Fig. 7C). At 6 GPa, the aqueous fluid continuously increases its solute content with increasing temperature suggesting a critical endpoint in this system between 5 and 6 GPa ⁄ 1050C. Complete miscibility between aqueous fluids and silicate melts in the systems H2O–Ca-bearing granite and H2O–haplogranite was observed by Bureau & Keppler (1999). The corresponding critical curves are located at 1.34 GPa ⁄ 1003C to 2.04 GPa ⁄ 735C in the H2O–haplogranite system and at slightly higher pressure and temperature conditions of 1.61 GPa ⁄ 898C to 2.08 GPa ⁄ 820C in the H2O–Ca-bearing granite system. However, no critical endpoints, which determine the ter-
Aqueous fluids at elevated pressure and temperature 15 mination of the wet solidus, are given for the systems H2O–Ca-bearing granite and H2O–haplogranite in Bureau & Keppler (1999). One important aspect of H2O–silicate systems at pressure conditions above the critical endpoint is the combination of high solute contents in the fluids with comparably low viscosities. Aude´tat & Keppler (2004) performed a seminal experimental study on the viscosity of high-pressure silicate rich aqueous fluids in the systems H2O–albite, H2O–nepheline and H2O–pectolite. In the H2O–albite system, fluid viscosity linearly increases with increasing solute content from approximately 3.5 · 10)4 Pa sec for 7 mol% solutes to approximately 4 · 100 Pa sec for 67 mol% solutes (Fig. 7D). The viscosity of dry and hydrous albite melts with up to 3 wt% H2O is notably higher and ranges between approximately 109 and 1011.5 Pa sec (Romano et al. 2001). The data therefore indicate an exponentially decrease of viscosity with increasing H2O content for <20 wt% H2O (Aude´tat & Keppler 2004). Given their comparably low viscosity with high solute content, aqueous fluids above the upper critical endpoint potentially form very effective metasomatising agents within the mantle wedge.
SUMMARY This paper reviews some of the fundamental physical and chemical properties of aqueous fluids at elevated pressure and temperature conditions. For certain geological environments these physicochemical fluid properties are fairly well known. This is, at least partly, true for low pressure and temperature systems like geothermal systems, basinal systems, and oceanic hydrothermal systems, in which fluids can be sampled and studied directly or which are easily investigated by experimental methods. At higher, typical metamorphic pressure and temperature conditions, the principal phase relations in model fluid systems are known. However, there are important gaps in knowledge of aqueous fluids at elevated pressure and temperature. First, even for pure H2O, most data for important properties like ion product, static dielectric constant, and viscosity are restricted to pressures below 1.0 GPa. Also, some of the available formulations for these properties substantially diverge outside the pressure–temperature range covered by the experimental data. Second, while the principal phase relations in binary and ternary aqueous fluid systems are well known, experimental data precisely addressing the composition of co-existing fluid phases are rare. In addition, data on trace element and isotope fractionation behaviour between co-existing aqueous fluids are rare. Knowledge of these fractionation behaviours are a prerequisite for modelling geochemical cycles in fluid mediated or fluid dominated systems. Finally, for high pressure and temperature of the lower crust and upper mantle, the knowledge of physical fluid properties like viscosity and density in ‘supercritical’
H2O–silicate systems is rather limited. However, for modelling the role of these ‘supercritical’ H2O–silicate fluids as metasomatizing agent for crust–mantle interactions within subduction zones, these data are essential.
ACKNOWLEDGMENTS I thank the editors for inviting me to contribute to this Geofluids anniversary volume. This paper greatly benefitted from careful and critical reviews by two anonymous reviewers. Helpful editorial handling by C. Manning is gratefully acknowledged.
REFERENCES Abramson EH (2007) Viscosity of water measured to pressures of 6 GPa and temperature of 300C. Physical Review E, 76, 051203. Anderko A, Pitzer KS (1993a) Equation-of-state representation of phase equilibria and volumetric properties of the system NaCl– H2O above 573 K. Geochimica et Cosmochimica Acta, 57, 1657–80. Anderko A, Pitzer KS (1993b) Phase equilibria and volumetric properties of the systems KCl–H2O and NaCl–KCl–H2O above 573 K: equation of state representation. Geochimica et Cosmochimica Acta, 57, 4885–97. Anisimov MA, Sengers JV, Levelt Sengers JMH (2004) Nearcritical behaviour of aqueous systems. In: Aqueous Systems at Elevated Temperatures and Pressures (eds Palmer DA, Ferna´ndez-Prini R, Harvey AH), pp. 29–71. Elsevier, London. Anovitz LM, Labotka TC, Blencoe JG, Horita J (2004) Experimental determination of the activity-composition relations and phase equilibria of H2O–CO2–NaCl fluids at 500C, 500 bars. Geochimica et Cosmochimica Acta, 68, 3557–67. ¨ (2007) Fluid–fluid interArno´rsson S, Stefa´nsson A, Bjarnason JO actions in geothermal systems. Reviews in Mineralogy and Geochemistry, 65, 259–312. Aude´tat A, Keppler H (2004) Viscosity of fluids in subduction zones. Science, 303, 513–6. Bandura AV, Lvov SN (2006) The ionization constant of water over wide ranges of temperature and density. Journal of Physical and Chemical Reference Data, 35, 15–30. Belonoshko AB, Saxena SK (1991) A molecular dynamics study of the pressure–volume–temperature properties of supercritical fluids: I. H2O. Geochimica et Cosmochimica Acta, 55, 381–8. Belonoshko AB, Saxena SK (1992) A unified equation of state for fluids of C–H–O–N–S–Ar composition and their mixtures up to high temperatures and pressures. Geochimica et Cosmochimica Acta, 56, 3611–36. Bignold GJ, Brewer AD, Hearn B (1971) Specific conductivity and ionic product of water between 50 and 271C. Transactions of the Faraday Society, 67, 2419–30. Bischoff JL (1991) Densities of liquids and vapors in boiling NaCl-H2O solutions: a PVTx summary from 300 to 500C. American Journal of Science, 291, 309–38. Bischoff JL, Pitzer KS (1989) Liquid–vapor relations for the system NaCl–H2O: summary of the P–T–x surface from 300 to 500C. American Journal of Science, 289, 217–48. Bischoff JL, Rosenbauer RJ (1984) The critical point and twophase boundary of seawater, 200–500C. Earth and Planetary Science Letters, 68, 172–80.
16 A. LIEBSCHER Bischoff JL, Rosenbauer RJ (1988) Liquid–vapor relations in the critical region of the system NaCl–H2O from 380 to 415C: a refined determination of the critical point and two-phase boundary of seawater. Geochimica et Cosmochimica Acta, 52, 2121–6. Bischoff JL, Rosenbauer RJ, Pitzer KS (1986) The system NaCl– H2O: relations of vapour–liquid near the critical temperature of water and of vapour–liquid–halite from 300 to 500C. Geochimica et Cosmochimica Acta, 50, 1437–44. Bischoff JL, Rosenbauer RJ, Fournier RO (1996) The generation of HCl in the system CaCl2–H2O: vapor–liquid relations from 380-500C. Geochimica et Cosmochimica Acta, 60, 7–16. Bodnar RJ, Burnham CW, Sterner SM (1985) Synthetic fluid inclusions in natural quartz. III. Determination of phase equilibrium properties in the System H2O–NaCl to 1000C and 1500 bars. Geochimica et Cosmochimica Acta, 49, 1861–73. Boyd FR, England JL (1963) Effect of pressure on melting of diopside, CaMgSi2O6, and albite, NaAlSi3O8, in the range up to 50 kbar. Journal of Geophysical Research, 68, 311–23. Bradley DJ, Pitzer KS (1979) Thermodynamics of electrolytes. 12. Dielectric properties of water and Debye-Hu¨ckel parameters to 350C and 1 kbar. Journal of Physical Chemistry, 83, 1599– 603. Bureau H, Keppler H (1999) Complete miscibility between silicate melts and hydrous fluids in the upper mantle: experimental evidence and geochemical implications. Earth and Planetary Science Letters, 165, 187–96. Chou I-M (1987) Phase relations in the system NaCl–KCl–H2O. III: solubilities of halite in vapor-saturated liquids above 445C and redetermination of phase equilibrium properties in the system NaCl–H2O to 1000C and 1500 bars. Geochimica et Cosmochimica Acta, 51, 1965–75. Churakov SS, Gottschalk M (2003a) Pertubation theory based equation of state for polar molecular fluids: I. Pure fluids. Geochimica et Cosmochimica Acta, 67, 2397–414. Churakov SS, Gottschalk M (2003b) Pertubation theory based equation of state for polar molecular fluids: II. Fluid mixtures. Geochimica et Cosmochimica Acta, 67, 2415–25. Cline JS, Bodnar RJ (1994) Direct evolution of brine from a crystallizing silicic melt at the Questa, New Mexico, molybdenum deposit. Economic Geology, 89, 1780–802. Deul R (1984) Dielektrizita¨tskonstante und Dichte von Wasser-Benzol-Mischungen bis 450C und 3000 bar. PhD Dissertation, University of Karlsruhe, Germany. Diamond LW (2001) Review of the systematic of CO2–H2O fluid inclusions. Lithos, 55, 69–99. Driesner T, Heinrich CA (2007) The System H2O–NaCl. I. Correlation formulae for phase relations in pressure-temperaturecomposition space from 0 to 1000C, 0 to 5000 bar, and 0 to 1 XNaCl. Geochimica et Cosmochimica Acta, 71, 4880–901. Duan Z, Zhang Z (2006) Equation of state of the H2O, CO2, and H2O–CO2 systems up to 10 GPa and 2573,15 K: molecular dynamics simulations with ab initio potential surface. Geochimica et Cosmochimica Acta, 70, 2311–24. Duan Z, Moller N, Weare JH (1992a) An equation of state (EOS) for CH4–CO2–H2O system: I. Pure systems from 0 to 1000C and 0 to 8000 bar. Geochimica et Cosmochimica Acta, 56, 2605–17. Duan Z, Moller N, Weare JH (1992b) An equation of state for the CH4–CO2–H2O system: II. Mixtures from 50 to 1000C and 0 to 1000 bar. Geochimica et Cosmochimica Acta, 56, 2619–31. Duan Z, Moller N, Weare JH (1992c) Molecular dynamics simulations of PVT properties of geological fluids and a equation of state of non-polar and weakly polar gases up to 2000 K and 20,000 bar. Geochimica et Cosmochimica Acta, 56, 3839–45.
Duan Z, Moller N, Weare JH (1995) Equation of state for the NaCl– H2O–CO2 system: prediction of phase equilibria and volumetric properties. Geochimica et Cosmochimica Acta, 59, 2869–82. Duan Z, Moller N, Weare JH (1996) A general equation of state for supercritical fluid mixtures and molecular dynamics simulation of mixture PVTX properties. Geochimica et Cosmochimica Acta, 60, 1209–16. Duan Z, Moller N, Weare JH (2003) Equations of state for the NaCl–H2O–CH4 system and the NaCl–H2O–CO2–CH4 system: phase equilibria and volumetric properties above 573 K. Geochimica et Cosmochimica Acta, 67, 671–80. Duan Z, Moller N, Weare JH (2006) A high temperature equation of state for the H2O–CaCl2 and H2O–MgCl2 systems. Geochimica et Cosmochimica Acta, 70, 3765–77. Dubois M, Weisbrod A, Shtuka A (1994) Experimental determination of the two-phase (liquid and vapour) region in water-alkali chloride binary mixtures at 500 and 600C using synthetic fluid inclusions. Chemical Geology, 115, 227–38. Dudziak KH, Franck EU (1966) Measurements of water viscosity up to 560C and 3500 bars. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 70, 1120–8. Ferna´ndez DP, Mulev Y, Goodwin ARH, Levelt Sengers JMH (1995) A database for the static dielectric constant of water and steam. Journal of Physical and Chemical Reference Data, 24, 33–69. Ferna´ndez DP, Goodwin ARH, Lemmon EW, Levelt Sengers JMH, Williams RC (1997) A formulation for the static permittivity of water and steam at temperatures from 238 K to 873 K at pressures up to 1200 MPa, including derivatives and DebyeHu¨ckel coefficients. Journal of Physical and Chemical Reference Data, 26, 1125–66. Ferry JM, Baumgartner L (1987) Thermodynamic models of molecular fluids at the elevated pressures and temperatures of crustal metamorphism. Reviews in Mineralogy, 17, 323–66. Fischer JR, Barnes HL (1972) The ion-product constant of water to 350C. Journal of Physical Chemistry, 76, 90–9. Foustoukos DI, Seyfried WE Jr (2007) Fluid phase separation processes in submarine hydrothermal systems. Reviews in Mineralogy and Geochemistry, 65, 213–39. Franck EU (1956) Highly compressed steam. III. Ion dissociation of hydrochloric acid, potassium hydroxide, and water in water above the critical point. Zeitschrift fu¨r physikalische Chemie, 8, 192–206. Franck EU, Rosenzweig S, Christoforakos M (1990a) Calculation of the dielectric constant of water to 1000C and very high pressures. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 94, 199–203. Franck EU, Rosenzweig S, Christoforakos M (1990b) Erratum: calculation of the dielectric constant of water to 1000C and very high pressures. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 94, 1165. Gehrig M, Lentz H, Franck EU (1986) The system water–carbon dioxide–sodium chloride to 773 K and 300 MPa. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 90, 525–33. German CR, Von Damm KL (2003) Hydrothermal processes. Treatise in Geochemistry, 6, 181–222. Gibert F, Guillaume D, Laporte D (1998) Importance of fluid immiscibility in the H2O–NaCl–CO2 system and selective CO2 entrapment in granulites: experimental phase diagram at 5–7 kbar, 900C and wetting textures. European Journal of Mineralogy, 10, 1097–123. Golubev BP (1978) Dissertation (unpublished), Moscow Power Institute, Russia [values for the dielectric constant listed in Ferna´ndez et al. (1995)].
Aqueous fluids at elevated pressure and temperature 17 Gottschalk M (2007) Equations of state for complex fluids. Reviews in Mineralogy and Geochemistry, 65, 49–97. Hack A, Thompson AB, Aerts M (2007) Phase relations involving hydrous silicate melts, aqueous fluids, and minerals. Reviews in Mineralogy and Geochemistry, 65, 129–85. Halbach H, Chatterjee ND (1982) An empirical Redlich-Kwongtype equation of state for water to 1000C and 200 kbar. Contributions to Mineralogy and Petrology, 79, 337–45. Hanor JS (1994) Origin of saline fluids in sedimentary basins. In: Geofluids: Origin, Migration and Evolution of Fluids in Sedimentary Basins (ed Parnell J), pp. 151–74. Geological Society Special Publication, London, 78. Heger K (1969) Die Statische Dielektrizita¨tskonstante von Wasser ¨ berkritischen Temperatur- und Druckberund Methylalkohol im U eich. Dissertation, University of Karlsruhe, Germany. Heger K, Uematsu M, Franck EU (1980) The static dielectric constant of water at high pressures and temperatures to 500 MPa and 550C. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 84, 758–62. Heinrich W (1993) Fluid infiltration through metachert layers at the contact aureole of the Bufa del Diente intrusion, northeast Mexico: implications for wollastonite formation and fluid immiscibility. American Mineralogist, 78, 804–18. Heinrich CA (2007) Fluid-fluid interactions in magmatic-hydrothermal ore formation. Reviews in Mineralogy and Geochemistry, 65, 363–87. Heinrich W (2007) Fluid immiscibility in metamorphic rocks. Reviews in Mineralogy and Geochemistry, 65, 389–430. Heinrich W, Churakov SS, Gottschalk M (2004) Mineral-fluid equilibria in the system CaO–MgO–SiO2–H2O–CO2–NaCl and the record of reactive fluid flow in contact metamorphic aureoles. Contributions to Mineralogy and Petrology, 148, 131– 49. Holzapfel WB, Franck EU (1966) Conductance and ion dissociation of water up to 1000C and 100 kilobars. Berichte der Bunsengesellschaft fu¨r Physikalische Chemie, 70, 1105–12. Hovey JK, Pitzer KS, Tanger IVJC, Bischoff JL, Rosenbauer RJ (1990) Vapor–liquid phase equilibria of potassium chloridewater mixtures: equation-of-state representation for KCl–H2O and NaCl–H2O. Journal of Physical Chemistry, 94, 1175–9. Izraeli ES, Harris JW, Navon O (2001) Brine inclusions in diamonds: a new upper mantle fluid. Earth and Planetary Science Letters, 187, 323–32. Jacobs GK, Kerrick DM (1981) Methane: an equation of state with application to the ternary system H2O–CO2–CH4. Geochimica et Cosmochimica Acta, 45, 607–14. Jiang S, Pitzer KS (1996) Phase equilibria and volumetric properties of aqueous CaCl2 by an equation of state. AIChE Journal, 42, 585–94. Joyce DB, Holloway JR (1993) An experimental determination of the thermodynamic properties of H2O–CO2–NaCl fluids at high pressures and temperatures. Geochimica et Cosmochimica Acta, 57, 733–46. Keevil NB (1942) Vapor pressures of aqueous solutions at high temperatures. Journal of the American Chemical Society, 64, 841–50. Kennedy GC, Wasserburg GJ, Heard HC, Newton RC (1962) The upper three-phase region in the system SiO2–H2O. American Journal of Science, 260, 501–21. Kerrick DM, Jacobs GK (1981) A modified Redlich-Kwong equation for H2O, CO2, and H2O–CO2 mixtures at elevated pressures and temperatures. American Journal of Science, 281, 735–67. Kesler SE (2005) Ore-forming fluids. Elements, 1, 13–8.
Kessel R, Ulmer P, Pettke T, Schmidt MW, Thompson AB (2005a) The water-basalt system at 4 to 6 GPa: phase relations and second critical endpoint in a K-free eclogite at 700 to 1400C. Earth and Planetary Science Letters, 237, 873–92. Kessel R, Schmidt MW, Ulmer P, Pettke T (2005b) Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120-180 km depth. Nature, 437, 724–7. Ketsko VA, Urusova MA, Valyashko VM (1984) Solubility and vapour pressure of solutions in the CaCl2–H2O system at 250-400C. Russian Journal of Inorganic Chemistry, 29, 1398–9. Khaibullin K, Borisov NM (1966) Experimental investigation of the thermal properties of aqueous and vapour solutions of sodium and potassium chloride at phase equilibrium. Teplofizika Vysokikh Temperatur, 4, 518–23[English translation 489–494]. Kotel’nikov AR, Kotel’nikova ZA (1991) The phase state of the H2O–CO2–NaCl system examined from synthetic fluid inclusions in quartz. Geochemistry International, 29, 55–66. Krader T (1985) Phasengleichgewichte und Kritische Kurven des Terna¨ren Systems H2O–CH4–NaCl bis 250 MPa und 800 K. Dissertation, University of Karlsruhe, Germany. Lamb WM, Popp RK, Boockoff LA (1996) The determination of phase relations in the CH4–H2O–NaCl system at 1 kbar, 400 to 600C using synthetic fluid inclusions. Geochimica et Cosmochimica Acta, 60, 1885–97. Lamb WM, McShane CJ, Popp RK (2002) Phase relations in the CH4–H2O–NaCl system at 2 kbar, 300 to 600C as determined using synthetic fluid inclusions. Geochimica et Cosmochimica Acta, 66, 3971–86. Liebscher A (2007) Experimental studies in model fluid systems. Fluid–fluid interactions. Reviews in Mineralogy and Geochemistry, 65, 15–47. Liebscher A, Heinrich CA (2007) Fluid-fluid interactions in the Earth’s lithosphere. Reviews in Mineralogy and Geochemistry, 65, 1–13. Lu¨ders V, Banks DA, Halbach P (2002) Extreme Cl ⁄ Br and d37Cl isotope fractionation in fluids of modern submarine hydrothermal systems. Mineralium Deposita, 37, 765–71. Lukashov YM (1981) Dissertation (unpublished), Moscow Power Institute, Russia [values for the dielectric constant listed in Ferna´ndez et al. (1995)]. Lukashov YM, Golubev BP, Ripol-Saragosi FB (1975) Teploenergetika, 22, 79–82 [values for the dielectric constant listed in Ferna´ndez et al. (1995)]. MacDonald DD, Butler P, Owen D (1973) High temperature aqueous electrolyte concentration cells and the ionization of liquid water to 200C. Canadian Journal of Chemistry, 51, 2590–5. Marshall WL, Franck EU (1981) Ion product of water substance, 0-1000C, 1-10,000 bars, new international formulation and its background. Journal of Physical and Chemical Reference Data, 10, 295–304. Marshall WL, Hall CE, Mesmer RE (1981) The system dipotassium hydrogen phosphate-water at high temperatures (100400C); liquid-liquid immiscibility and concentrated solutions. Journal of inorganic and nuclear Chemistry, 43, 449–4551. Muchailov MR (1988) Dissertation (unpublished), Moscow Power Institute, Russia [values for the dielectric constant listed in Ferna´ndez et al. (1995)]. Mulev YV, Smirnov SN, Muchailov MR (1994) Teploenergetika [values for the dielectric constant listed in Ferna´ndez et al. (1995)] Navon O, Hutcheon ID, Rossman GR, Wasserburg GJ (1988) Mantle-derived fluids in diamond micro-inclusions. Nature, 335, 784–9.
18 A. LIEBSCHER Newton RC, Manning CE (2008) Thermodynamics of SiO2–H2O fluid near the upper critical end point from quartz solubility measurements at 10 kbar. Earth and Planetary Science Letters, 274, 241–9. Oosting SE, Von Damm KL (1996) Bromide ⁄ chloride fractionation in seafloor hydrothermal fluids from 9-10N East Pacific Rise. Earth and Planetary Science Letters, 144, 133–45. Oshry HI (1949) Dissertation (unpublished), University of Pittsburgh, USA [values for the dielectric constant listed in Ferna´ndez et al. (1995)]. Paillat E, Elphick SC, Brown WL (1992) The solubility of water in NaAlSi3O8 melts: a re-examination of Ab–H2O phase relationships and critical behaviour at high pressures. Contributions to Mineralogy and Petrology, 112, 490–500. Pitzer KS (1983) Dielectric constant of water at very high temperature and pressure. Proceedings of the National Academy of Science of the United States of America, 80, 4575–6. Plyasunova NV, Shmulovich KI (1991) Phase equilibria in the system H2O–CO2–CaCl2 at 500C. Transactions of the USSR Academy of Sciences, 320, 221–5. Prausnitz JM, Lichtenthaler RN, Comez de Azevedo E (1999) Molecular Thermodynamics of Fluid-Phase Equilibria. PrenticeHall, New Jersey. Quist AS (1970) The ionization constant of water to 800C and 4000 bars. Journal of Physical Chemistry, 74, 3396–402. Quist AS, Marshall WL (1965) Estimation of the dielectric constant of water to 800C. Journal of Physical Chemistry, 69, 3165–7. Ravich MI (1974) Water-Salt Systems at Elevated Temperatures and Pressure [in Russian]. Nauka, Moscow. p. 150. Romano C, Poe B, Mincione V, Hess KU, Dingwell DB (2001) The viscosities of dry and hydrous XAlSi3O8 (X = Li, Na, K, Ca0.5, Mg0.5) melts. Chemical Geology, 174, 115–32. Rosenbauer RJ, Bischoff JL (1987) Pressure-composition relations for coexisting gases and liquids and the critical points in the system NaCl–H2O at 450, 475, and 500C. Geochimica et Cosmochimica Acta, 51, 2349–54. Saul A, Wagner W (1989) A fundamental equation for water covering the range from the melting line to 1273 K at pressures up to 25000 MPa. Journal of Physical and Chemical Reference Data, 18, 1537–64. Saxena SK, Fei Y (1987) Fluids at crustal pressures and temperatures. I. Pure species. Contributions to Mineralogy and Petrology, 95, 370–5. Saxena SK, Fei Y (1988) Fluid mixtures in the C–H–O system at high pressure and temperature. Geochimica et Cosmochimica Acta, 52, 505–12. Schmidt C, Bodnar RJ (2000) Synthetic fluid inclusions: XVI. PVTX properties in the system H2O–NaCl–CO2 at elevated temperatures, pressures, and salinities. Geochimica et Cosmochimica Acta, 64, 3853–69. Sengers JV, Kamgar-Parsi B (1984) Representative equations for the viscosity of water substance. Journal of Physical and Chemical Reference Data, 13, 185–205. Sengers JV, Kayser RF, Peters CJ, White HJ Jr (eds.) (2000) Equations of State for Fluids and Fluid Mixtures. Experimental Thermodynamics. Elsevier, Amsterdam. Shen AH, Keppler H (1997) Direct observation of complete miscibility in the albite–H2O system. Nature, 385, 710–2. Shi P, Saxena SK (1992) Thermodynamic modelling of the C–H– O–S fluid system. American Mineralogist, 77, 1038–49. Shmulovich KI, Graham CM (1999) An experimental study of phase equilibria in the system H2O–CO2–NaCl at 800C and 9 kbar. Contributions to Mineralogy and Petrology, 136, 247–57.
Shmulovich KI, Graham CM (2004) An experimental study of phase equilibria in the systems H2O–CO2–CaCl2 and H2O– CO2–NaCl at high pressures and temperatures (500-800C, 0.5-0.9 GPa): geological and geophysical applications. Contributions to Mineralogy and Petrology, 146, 450–62. Shmulovich KI, Plyasunova NV (1993) Phase equilibria in ternary systems formed by H2O and CO2 with CaCl2 or NaCl at high T and P. Geochemistry International, 30, 53–71. Shmulovich KI, Tkachenko SI, Plyasunova NV (1995) Phase equilibria in fluid systems at high pressures and temperatures. In: Fluids in the Crust: Equilibrium and Transport Properties (eds Shmulovich KI, Yardley BWD, Gonchar GG). pp. 193–214. Chapman & Hall, London. Shmulovich K, Heinrich W, Mo¨ller P, Dulski P (2002) Experimental determination of REE fractionation between liquid and vapour in the systems NaCl–H2O and CaCl2–H2O up to 450C. Contributions to Mineralogy and Petrology, 144, 257–73. Shock EL, Oelkers EH, James S, Johnson W, Sverjensky DA, Helgeson HC (1992) Calculation of the thermodynamic properties of aqueous species at high pressures and temperatures – effective electrostatic radii, dissociation constants and standard partial molal properties to 1000C and 5 kbar. Journal of the Chemical Society, Faraday Transactions, 88, 803–26. Skippen G, Trommsdorff V (1986) The influence of NaCl and KCl on phase relations in metamorphosed carbonate rocks. American Journal of Science, 286, 81–204. Sourirajan S, Kennedy GC (1962) The system H2O–NaCl at elevated temperatures and pressures. American Journal of Science, 260, 115–41. Sowerby JR, Keppler H (2002) The effect of fluorine, boron and excess sodium on the critical curve in the albite–H2O system. Contributions to Mineralogy and Petrology, 143, 32–7. Spear FS (1993) Metamorphic Phase Equilibria and Pressure-Temperature-Time Paths. Mineralogical Society of America Monograph, Washington, p. 799. Stalder R, Ulmer P, Thompson AB, Gu¨nther D (2000) Experimental approach to constrain second critical endpoints in fluid ⁄ silicate systems: near-solidus fluids and melts in the system albite-H2O. American Mineralogist, 85, 68–77. Svistunov EP (1975) Dissertation (unpublished), Moscow Power Institute, Russia [values for the dielectric constant listed in Ferna´ndez et al. (1995)].. Sweeton FH, Mesmer RE, Baes CF Jr (1974) Acidity measurements at elevated temperatures. VII. Dissociation of water. Journal of Solution Chemistry, 3, 191–214. Takenouchi S, Kennedy GC (1964) The binary system H2O–CO2 at high temperatures and pressures. American Journal of Science, 262, 1055–74. Takenouchi S, Kennedy GC (1965) The solubility of carbon dioxide in NaCl solutions at high temperatures and pressures. American Journal of Science, 263, 445–54. Tanger IVJC, Pitzer KS (1989) Calculation of the ionization constant of H2O to 2,273 K and 500 MPa. AIChE Journal, 35, 1631–8. Tkachenko SI, Shmulovich K (1992) Liquid–vapour equilibria in water-salt systems (NaCl, KCl, CaCl2, MgCl2) at 400-600C. Doklady Akademii Nauk SSSR, 326, 1055–9. Trommsdorff V, Skippen G (1986) Vapour loss (‘boiling’) as a mechanism for fluid evolution in metamorphic rocks. Contributions to Mineralogy and Petrology, 94, 317–22. Trommsdorff V, Skippen G, Ulmer P (1985) Halite and sylvite as solid inclusions in high-grade metamorphic rocks. Contributions to Mineralogy and Petrology, 89, 24–9.
Aqueous fluids at elevated pressure and temperature 19 Urusova MA (1974) Phase equilibria in the sodium hydroxidewater and sodium chloride-water systems at 350-550C. Russian Journal of Inorganic Chemistry, 19, 450–4. Urusova MA (1975) Volume properties of aqueous solutions of sodium chloride at elevated temperatures and pressures. Russian Journal of Inorganic Chemistry, 20, 1717–21. Urusova MA, Valyashko VM (1983) Solubility, vapour pressure, and thermodynamic properties of solutions in the MgCl2–H2O system at 300-350C. Russian Journal of Inorganic Chemistry, 28, 1045–8. Urusova MA, Valyashko VM (1984) Vapour pressure and thermodynamic properties of aqueous solutions of magnesium chloride at 250C. Russian Journal of Inorganic Chemistry, 29, 1395–6. Vakulenko AG, Alekhin YV, Rasina MV (1989) Solubility and thermodynamic properties of alkali chlorides in steam. Proceedings of the II International Symposium ‘Properties of water and steam’, Prague, 395–401. Valyashko VM (1990) Phase Equilibria and Properties of Hydrothermal Systems (in Russian). Nauka, Moscow. Valyashko VM (2004) Phase equilibria in water-salt systems at high temperatures and pressures. In: Aqueous Systems at Elevated Temperatures and Pressures (eds Palmer DA, Ferna´ndez-Prini R, Harvey AH), pp. 597–641. Elsevier, London.
Veksler IV (2004) Liquid immiscibility and its role at the magmatic-hydrothermal transition: a summary of experimental results. Chemical Geology, 210, 7–31. Wagner W, Pruß A (2002) The IAPWS formulation 1995 for the thermodynamic properties of ordinary water substance for general and scientific use. Journal of Physical and Chemical Reference Data, 31, 378–535. Wasserman E, Wood B, Brodholt J (1995) The static dielectric constant of water at pressures up to 20 kbar and temperatures to 1273 K – experiment, simulations, and empirical equations. Geochimica et Cosmochimica Acta, 59, 1–6. Watson JTR, Basu RS, Sengers JV (1980) An improved representative equation for the dynamic viscosity of water substance. Journal of Physical and Chemical Reference Data, 9, 1255–90. Welsch H (1973) Die Systeme Xenon-Wasser und Methan-Wasser bei Hohen Dru¨cken und Temperaturen. Dissertation, University of Karlsruhe, Germany. Yardley BWD, Graham JT (2002) The origins of salinity in metamorphic fluids. Geofluids, 2, 249–56. Zhang Y-G, Frantz JD (1989) Experimental determination of the compositional limits of immiscibility in the system CaCl2– H2O–CO2 at high temperatures and pressures using synthetic fluid inclusions. Chemical Geology, 74, 289–308.
Thermodynamic model for mineral solubility in aqueous fluids: theory, calibration and application to model fluid-flow systems D . D O L E J Sˇ 1 , 2 A N D C . E . M A N N I N G 1 , 3 Bayerisches Geoinstitut, University of Bayreuth, Bayreuth, Germany; 2Institute of Petrology and Structural Geology, Charles University, Praha, Czech Republic; 3Department of Earth and Space Sciences, University of California Los Angeles, Los Angeles, CA, USA
1
ABSTRACT We present a thermodynamic model for mineral dissolution in aqueous fluids at elevated temperatures and pressures, based on intrinsic thermal properties and variations of volumetric properties of the aqueous solvent. The standard thermodynamic properties of mineral dissolution into aqueous fluid consist of two contributions: one from the energy of transformation from the solid to the hydrated-species state and the other from the compression of solvent molecules during the formation of a hydration shell. The latter contribution has the dimension of the generalized Krichevskii parameter. This approach describes the energetics of solvation more accurately than does the Born electrostatic theory and can be extended beyond the limits of experimental measurements of the dielectric constant of H2O. The new model has been calibrated by experimental solubilities of quartz, corundum, rutile, calcite, apatite, fluorite and portlandite in pure H2O at temperatures up to 1100C and pressures up to 20 kbar. All minerals show a steady increase in solubility along constant geothermal gradients or water isochores. By contrast, isobaric solubilities initially increase with rising temperature but then decline above 200–400C. This retrograde behavior is caused by variations in the isobaric expansivity of the aqueous solvent, which approaches infinity at its critical point. Oxide minerals predominantly dissolve to neutral species; so, their dissolution energetics involve a relatively small contribution from the solvent volumetric properties and their retrograde solubilities are restricted to a relatively narrow window of temperature and pressure near the critical point of water. By contrast, Ca-bearing minerals dissolve to a variety of charged species; so, the energetics of their dissolution reactions involve a comparatively large contribution from volume changes of the aqueous solvent and their isobaric retrograde solubility spans nearly all metamorphic and magmatic conditions. These features correlate with and can be predicted from the standard partial molar volumes of aqueous species. The thermodynamic model can be used over much wider range of settings for terrestrial fluid–rock interaction than has previously been possible. To illustrate, it is integrated with transport theory to show quantitatively that integrated fluid fluxes characteristic of crustal shear zones are capable of precipitating quartz or calcite veins from low- and medium-grade metamorphic conditions, at a geothermal gradient of 20C km)1. For subduction zones, modeled by a geotherm of 7C km)1, the required fluid fluxes are one to two orders of magnitude lower and predict enhanced efficiency of mass transfer and metasomatic precipitation in comparison with orogenic settings. The new model thus can be applied to shallow hydrothermal, metamorphic, magmatic and subduction fluids, and for retrieval of dependent thermodynamic properties for mass transfer or geodynamic modeling. Key words: density model, hydration, mass transfer, mineral solubility, thermodynamics Received 24 August 2009; accepted 30 January 2010 Corresponding author: D. Dolejsˇ, Institute of Petrology and Structural Geology, Charles University, 12843 Praha 2, Czech Republic. Email:
[email protected]. Tel.: +420-221 951 532. Fax: +420-221 951 533. Geofluids (2010) 10, 20–40
INTRODUCTION Aqueous fluids play a fundamental role in mass and energy transfer in the Earth’s interior. Fluids are produced by
diagenetic and metamorphic devolatilization, and magmatic degassing in geodynamic settings ranging from mid-ocean ridges to convergent plate boundaries and collisional orogens (Fig. 1). In turn, they are responsible for mobility
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Thermodynamic model for mineral solubility in aqueous fluids 21
30
ρ = 1.3 g cm–3
1.1
ECL 1.0
20
tle
15
M
BSC
an
GRA
10 PEP
0 200
0.9
0.8 0.7
AMP
0.6 0.5 0.4
ZEO
5
GSC
Pressure (kbar)
No
t at
tain
25
Cr ato nic Sla b
abl
e
1.2
crit
SAN
400
600
800
1000
1200
Temperature (°C) Fig. 1. Pressure–temperature diagram relating the metamorphic facies, beginning of the hydrous haplogranite melting (dot-dashed curve), representative geotherms (Vernon & Clarke 2008, their fig. 1.15), and isochores of H2O calculated using the IAPWS 95 equation of state (Wagner & Pruß 2002). Abbreviations for metamorphic facies: AMP, amphibolite; BSC, blueschist; ECL, eclogite; GSC, greenschist; PEP, prehnite-pumpellyite; SAN, sanidinite; ZEO, zeolite facies.
and transport of inorganic and organic solid and liquid phases and gaseous non-electrolytes (Ague 1994, 2003; Jamtveit & Yardley 1997; Ferry & Gerdes 1998). The mass transfer is manifested by distinct element depletion– enrichment patterns in subduction-zone metamorphism and arc magmas (Scambelluri et al. 2004; Zack & John 2007), in metamorphic rocks and veins in convergent orogens (Widmer & Thompson 2001; Spandler & Hermann 2006), in alteration patterns and in ore-forming and contact metasomatic environments (Reed 1997; Mehnert et al. 2005) as well as in small-scale fluid-mediated dissolution–precipitation phenomena (Pina et al. 2004; Putnis & Putnis 2007). Many of these alteration styles are strongly non-isochemical, and this is corroborated by high time-integrated fluid fluxes, 101– 106 m3 m)2, inferred for diffuse and focused flow and alteration (Ferry 1994; Thompson 1997; Ferry & Gerdes 1998; Ague 2003). The thermodynamic properties of dissolved aqueous species at elevated temperatures and pressures are most frequently described by the Helgeson–Kirkham–Flowers model (Helgeson et al. 1981; Tanger & Helgeson 1988; Shock et al. 1992). In this formulation, the thermodynamic state functions of an aqueous species consist of a solvation contribution, defined by the Born (1920) theory, and empirical heat capacity and volumetric contributions, which also include anomalous solute–solvent interactions at low temperatures (Angell 1983). This model provides a
versatile tool for modeling mass transfer up to 5 kbar (Johnson et al. 1992; Oelkers et al. 1995); however, it suffers from several deficiencies: (i) it is only applicable at liquid-like density of H2O, q > 0.322 g cm)3 to 1000C and 5 kbar; (ii) it cannot accurately reproduce the derivative properties in the vicinity of the critical point of H2O; (iii) it becomes particularly inaccurate for gaseous non-electrolytes at their low-density limit (Plyasunov et al. 2000; Akinfiev & Diamond 2003); (iv) the static permittivity of water, which is required for the Helgeson–Kirkham–Flowers equation of state, has not been determined experimentally at T > 600C or P > 20 kbar, and current models strongly diverge in these regions (Franck et al. 1990; Shock et al. 1992; Wasserman et al. 1995, Fernande´z et al. 1997; Marshall 2008a). In contrast to the static permittivity, the volumetric properties of H2O are known with reasonable accuracy at nearly all conditions of terrestrial water–rock interaction. Simple extrapolation schemes and more advanced models based on H2O volumetric properties have been successful in representing association–dissociation equilibria (Franck 1956a; Marshall & Quist 1967; Mesmer et al. 1988; Anderson et al. 1991; Plyasunov 1993), element partitioning in vapor–liquid systems (Pokrovski et al. 2005a, 2008), dissolution of gaseous species in aqueous fluids (Plyasunov et al. 2000; Akinfiev & Diamond 2003), mineral solubilities over a wide range of temperatures and pressures in aqueous solvents (Fournier & Potter 1982; Manning 1994; Pokrovski et al. 2005b), extrapolation to mixed solvents (Marshall 2008b; Akinfiev & Diamond 2009) and describing near-critical thermodynamic properties of inorganic and organic solutes (Clarke et al. 2000; Sedlbauer & Wood 2004; Majer et al. 2008). In this paper, we investigate the thermodynamic relationships between mineral solubility, solvent density and other intensive variables, with the main focus on metamorphic and magmatic temperatures and pressures. The ‘density model’ will be derived from the thermodynamics of solute hydration, emphasizing consistency with conventional caloric properties and thermodynamic parameters resulting from statistical mechanics of solute–solvent interactions. This optimized functional form is shown to serve for interpolation and extrapolation of mineral solubilities in aqueous fluids at high temperatures and pressures, as well as for retrieving the thermodynamic properties of dissolution that are necessary for mass transfer and transport modeling.
THERMODYNAMIC MODEL Thermodynamic properties of hydration, which result from electrostatic interactions between solute species and aqueous solvent, have traditionally been described by the Born theory (Born 1920; Helgeson & Kirkham 1974; Wood
22 D. DOLEJSˇ & C. E. MANNING et al. 1981, 1994; Atkins & MacDermott 1982; Pitzer 1983b; Tanger & Helgeson 1988; Tremaine et al. 1997), which states that 1 DB G ¼ ! ; "
ð1Þ
where DBG is the Born energy, x is the species-specific Born parameter and e is static permittivity (dielectric constant) of the aqueous solvent. However, the reciprocal dielectric constant varies nearly linearly with temperature at constant water density at 200–1100C and up to 10 kbar (Fig. 2). The linear trend corresponds to 1 ¼ 4:928 104 T log w þ 3:497 102 "
ð2Þ
where T is absolute temperature (K) and qw is water density (g cm)3). Consequently, the electrostatic contribution can be, to a high degree of accuracy, implicitly accounted for via correlations with solvent density. Such an approach would alleviate a number of drawbacks due to: (i) lack of calibration of the static permittivity at high temperatures and pressures (cf. Pitzer 1983b; McKenzie & Helgeson 1984; Franck et al. 1990; Shock et al. 1992; Wasserman et al. 1995; Fernande´z et al. 1997) and (ii) inadequate representation by Born theory of near-critical and low-pressure thermodynamic properties (Tanger & Pitzer 1989a; Shock et al. 1992; Plyasunov et al. 2000; Sue et al. 2002; Akinfiev & Diamond 2003, 2004), where formulations based on solvent density are more accurate (Fournier 1983, Manning 1994; Plyasunov et al. 2000; Sedlbauer et al. 2001; Sedlbauer & Wood 2004; Majer et al. 2008).
An alternative thermodynamic model can be formulated by accounting for three thermodynamic contributions to mineral dissolution into an aqueous fluid. At a given temperature and pressure, these are: (i) disruption of local structure of the crystal lattice, (ii) hydration of the solute species and (iii) volumetric solute–solvent interactions resulting from electrostriction. These contributions are added in a Born–Haber thermodynamic cycle to derive the standard thermodynamic properties of dissolution. Lattice breakdown and hydration The first two steps in the dissolution process – disruption of solid phase MX and formation of hydrated aqueous species – are represented by a series of chemical equilibria (cf. Kebarle 1977; Pitzer 1983a; Borg & Dienes 1992, pp. 126–137), e.g. for ion MðH2 O)zþ n : MXðsÞ ! Mzþ þ Xz zþ
M
ð3Þ zþ
þ H2 O ! MðH2 OÞ ; . . . ;
zþ MðH2 OÞzþ n1 þ H2 O ! MðH2 OÞn
0.5 0.4
1/
2 0.3 0.2 5
Dcl G ¼ a þ bT þ cT ln T þ dT 2
10
0.1
–800
Tref
where T and Tref represent the temperature of interest and reference temperature (e.g. 298.15 K) respectively. An explicit PV term is not required here because the standard state is chosen to be that at the pressure of interest and the volumetric contribution will be included in a density term, as shown below. When the heat capacity, cP, is a polynomial linear in temperature, which appears to be sufficient to represent the properties of aqueous species up to high temperatures (Pitzer 1982; Tanger & Pitzer 1989a,b; Holland & Powell 1998), Eq. 6 leads upon substitution and integration to the following Gibbs function:
1
0 –1200
ð5Þ
where n represents hydration number. Equations similar to 3–5 also apply to negatively charged species, ion pairs and neutral complexes. The standard Gibbs energy change of the two combined steps is described at the pressure of interest by a caloric (cl) equation of state: 1 0 ZT ZT cP C B ð6Þ dTA cP dT T @DS þ Dcl G ¼ DH þ T Tref
0.6
ð4Þ
–400
T log
w
0
200
(K)
Fig. 2. Correlation between the reciprocal dielectric constant of H2O, 1 ⁄ e, and the T log q term illustrated for isobars of 1, 2, 5 and 10 kbar (labels of dotted curves) at a temperature range of 200–1100C (point symbols in 100C steps). Dashed line is a linear fit of the set (Eq. 2).
ð7Þ
where the coefficients a to d are related to enthalpy, entropy and the heat capacity polynomial (Hillert 2008, pp. 407–408). Volumetric solute–solvent interactions Additional thermochemical contributions arise from volumetric collapse (electrostriction) of the hydration shell. The contribution to the Gibbs energy corresponds to the
Thermodynamic model for mineral solubility in aqueous fluids 23 PV work required to compress H2O molecules from the bulk solvent (w) density to that in the hydration shell (hs) (Pierotti 1963, 1976; Ben-Naim 2006, pp. 12), Dco G ¼ Dw!hs G ¼ RT ln
Vw ¼ RT ln hs Vhs w
ð8Þ
where bw is the compressibility of H2O (cf. Mesmer et al. 1988; Anderson et al. 1991; Ben-Amotz et al. 2005). This form has an independent origin in fluctuation solution theory, which defines linear scaling between the standard partial molar volume of aqueous species, Vaq, and the solvent compressibility as the generalized Krichevskii parameter, A (Levelt Sengers 1991; O’Connell et al. 1996; Plyasunov et al. 2000): Vaq : bRT
Dds G ¼ Dcl G þ Dco G
ð11Þ
leading to
where the subscript ‘co’ stands for compression, and V and q are the molar volume and density at each state respectively. Assuming negligible compressibility of the hydration shell, that is, qhs is constant (Mesmer et al. 1988; Tanger & Pitzer 1989a), Eq. 8 introduces two terms to the overall Gibbs energy of dissolution – an RT ln qhs term, where R ln qhs has dimension of entropy, and an RT ln qw term, which varies with temperature and pressure. Note that the T ln qw product has the same form as the term in the proposed correlation for the reciprocal dielectric constant (Eq. 2). Differentiation of DcoG (Eq. 8) with respect to pressure provides the standard molar volume change during the formation of the hydration shell, oDco G Dco V ¼ ¼ bw RT ; ð9Þ oP T
A¼
solvent compression (DcoG, Eq. 8) including solute–solvent interactions, are summed to give the standard Gibbs energy of dissolution, DdsG:
ð10Þ
The function 1 ) A is complementary to the generalized Krichevskii parameter and represents the dimensionless spatial integral of the infinite-dilution solute–solvent direct-correlation function (Kirkwood & Buff 1951; O’Connell 1971, 1990). It has the advantage of behaving as a finite smooth function in the vicinity of the critical point and it is nearly independent of temperature when applied to both electrolyte and non-electrolyte systems (Cooney & O’Connell 1987; Crovetto et al. 1991). The constant relationship between the solute partial molar volume and the solvent compressibility has, in addition, been confirmed experimentally up to elevated temperatures for inorganic solutes in aqueous and organic solvents (Hamann & Lim 1954; Ellis 1966).
MODEL FOR MINERAL DISSOLUTION The energetic terms associated with lattice breakdown (dissociation) and solute hydration (DclG, Eq. 7), and with
Dds G ¼ a þ bT þ cT ln T þ dT 2 þ eT ln ;
ð12Þ
which represents the standard reaction Gibbs energy of equilibria such as SiO2 (quartz) = SiO2 (aq) or CaF2 (fluorite) = Ca2+ (aq) + 2F) (aq). In Eq. 12, a to e are parameters of the model. Parameters c and d represent the constant and the linear terms in the heat capacity of dissolution respectively; where experimental data are sparse, these terms may be set to zero. As discussed below, the reduced three-parameter form (a, b and e) is sufficient to represent data that extend over at least 400C. An explicit term representing the intrinsic volume of species is not required for the properties of dissolution when the hardcore volumes of ions in the lattice and in the hydration sphere are assumed to be comparable. Using the relationship between the standard reaction Gibbs energy and equilibrium constant, K, 0 ¼ Dds G þ RT ln K ;
ð13Þ
Eq. 12 leads to ln K ¼
o 1 na þ b þ c ln T þ dT þ e ln w : R T
ð14Þ
Equation 14 indicates that there is an isothermal linear dependence of the logarithm of the equilibrium constant on the logarithm of solvent density, and that this linear slope is independent of temperature. Dissolution and association–dissociation equilibria at elevated pressures indeed appear to conform to a simple linear relationship including a logarithmic term of the solvent density, log K ¼ m þ n log w ;
ð15Þ
where m and n are fit parameters, which are often empirical polynomial expansions in temperature (Mosebach 1955; Franck 1956a; Marshall & Quist 1967; Mesmer et al. 1988; Anderson et al. 1991). Such a behavior is characteristic of the dissociation of water up to 1000C and 10 kbar (Sweeton et al. 1974, Marshall & Franck 1981; Fig. 3A), the dissociation of alkali halides up to 800C and 4 kbar (Franck 1956b; Quist & Marshall 1968; Frantz & Marshall 1982; Fig. 3B), the dissociation of acids and bases (Eugster & Baumgartner 1987; Mesmer et al. 1988, 1991; Tremaine et al. 2004) and the dissolution of quartz, calcite, apatite, halite and other minerals up to 900C and 20 kbar (Mosebach 1955; Franck 1956b; Martynova 1964; Fournier & Potter 1982; Manning 1994;
24 D. DOLEJSˇ & C. E. MANNING
–7
(A) H2O dissociation
800 600 400
–8
Log Kw
–9 –10 –11 –12 –13 –0.4
–0.3
–0.2
–0.1
0
0
0.1
400 600 800
(B) NaCl dissociation –1
Log K
–2 –3 –4 –5
–6 –0.6
–0.5
–0.4 –0.3
–0.2
–0.1
0
–1
(C) NaCl solubility
550 450 350
Log molality NaCl
–2
Caciagli & Manning 2003; Antignano & Manning 2008a; Fig. 3C). The solubility model represented by Eqs 12 and 14 differs from previous empirical density models. The first part (a–d terms) has frequently been used with inverse temperature factors (e.g. Fournier & Potter 1982; Mesmer et al. 1988, 1991; Anderson et al. 1991). Such a Gibbs energy function is, however, inconsistent with a constant heat capacity term (cf. Hillert 2008, pp. 407–408) and we prefer a rigorous combination of enthalpy, entropy and heat capacity contributions. Similarly, the ln q term was used with either no or inverse temperature dependence (e.g. Fournier & Potter 1982; Mesmer et al. 1988, 1991; Holland & Powell 1998), which may, in part, simplify the macroscopic relationship between the solute volume and heat capacity (Anderson et al. 1991; Anderson 2005, pp. 260–263). However, these forms prevent extrapolation of thermodynamic properties above approximately 300C (Anderson et al. 1991). Therefore, Anderson et al. (1991) and Holland & Powell (1998) proposed a modification, essentially a scaled T ln qw term, in order to improve and extend the applicability to 800C. This modification is fortuitously identical in form to the model that we derived using the compression of H2O in the hydration sphere. Thermodynamic identities lead to the standard thermodynamic properties of dissolution, as follows: oDds G Dds S ¼ ¼ b cð1 þ ln T Þ 2dT oT P e ln w þ eT w ; Dds H ¼ Dds G þ T Dds S ¼ a cT dT 2 þ eT 2 w ;
–3
–4
Dds cP ¼ T
–5
–6
o2 Dds G oT 2
oDds G Dds V ¼ oP
–7 –8 –3.0
ð16Þ
–2.5
–2.0
–1.5
–1.0
–0.5
Log H2O density (g cm–3) Fig. 3. Linear relationships between logarithmic density of aqueous solvent and logarithmic equilibrium constants for homogeneous solvent and salt dissociation, and salt solubility: (A) Self-dissociation of H2O with experimental data at liquidlike densities. Solid lines represent the fit of Marshall & Franck (1981), dotted curves are the model of Bandura & Lvov (2006), and point symbols are experimental data from Quist (1970), at 400, 600 and 800C. (B) Dissociation of NaCl (aq). Experimental data by Quist & Marshall (1968) are shown by point symbols at 400, 500, 600, 700 and 800C, with isotherms fitted to the data set. (C) Halite solubility in aqueous vapor. Experimental measurements are indicated by the following symbols – circles: Galobardes et al. (1981); diamonds: Armellini & Tester (1993); triangles: Higashi et al. (2005). Dashed lines are isotherms at 350, 450 and 550C fitted to the data set.
Dds b ¼
¼ c 2dT þ2eT w þeT 2
P
ð17Þ
ow ; oT P ð18Þ
¼ eT bw ;
ð19Þ
1 oDds V obw ¼ eT ; oP T Dds V oP T
ð20Þ
T
where qw, aw and bw represent the density, isobaric expansivity and isothermal compressibility of the solvent respectively. These thermodynamic relationships can also be used for deriving model parameters for a solubility equilibrium of interest by using the DdsH, DdsS and DdsV or DdscP at reference temperature and pressure (e.g. 25C and 1 bar) from existing thermodynamic databases (e.g. Wagman et al. 1982; Johnson et al. 1992; Oelkers et al. 1995; Parkhurst & Appelo 1999; Hummel et al. 2002). The calculation procedure for a three-parameter version of the model is given in Appendix A.
Thermodynamic model for mineral solubility in aqueous fluids 25 The new thermodynamic model (Eq. 12) suggests a functional form for a stand-alone equation of state for aqueous species. Any arbitrary standard thermodynamic property, Y, of bulk aqueous solute (aq) is a combination of corresponding property of dissolving solid phase (s) and its change during dissolution: ð21Þ
Y ðaqÞ ¼ Y ðsÞ þ Dds Y :
An equation of state for a solid phase and the standard thermodynamic properties of dissolution (Eq. 12, 16–18 or 19) can thus be added to obtain an equation of state for aqueous solutes applicable at high temperatures and pressures, which will consist of a caloric term (enthalpy, entropy and a heat capacity polynomial), a volumetric term (intrinsic volume of the species) and a density term.
CALIBRATION OF THE MODEL Solubilities of seven simple minerals – three oxides (quartz, corundum and rutile) and four calcium-bearing phases (anhydrite, apatite, calcite and fluorite) – have been determined experimentally at high temperatures and pressures and were used to explore the model and evaluate its performance. These minerals dissolve both as neutral or charged aqueous species, and their solubilities differ by seven orders of magnitude. For the solvent properties, we
have employed the equation of state of H2O for scientific use (Harvey 1998; Wagner & Prub 2002), which is calibrated by experiments to 1 GPa but extrapolates reasonably to very high pressures. Experimental solubility data at elevated pressure and temperature were first converted to molal concentration of the solute, temperature and density of H2O; multiple experiments at the same pressure and temperature were averaged to a single value in order to avoid artificial weighting of the fit. The reduced data sets were fitted by linear least squares to the equation of state for dissolution (Eq. 14). The resulting model parameters are listed in Table 1, standard thermodynamic properties at 25C and 1 bar are presented in Table 2 and the results plotted in Figs 4 and 5. The solubility of quartz in aqueous fluids and fluid mixtures has been extensively studied from ambient conditions up to 1100C and 20 kbar (e.g. Manning 1994 and references therein, Newton & Manning 2008a) and evaluated by Walther & Helgeson (1977), Fournier & Potter (1982), Manning (1994), Akinfiev (2001) and Gerya et al. (2005). We have calibrated our thermodynamic model using a subset of direct-sampling or rapid-quench experimental data (Hemley et al. 1980; Walther & Orville 1983; Manning 1994) augmented by calculated values from Fournier & Marshall (1983). The five-parameter fit (Table 1) reproduces experimental data set from 25 to 900C and from saturation vapor pressure up to 20 kbar
Table 1 Parameters of the thermodynamic model (Eqs 12, 14 and 16–20).
Apatite-F Calcite Corundum Fluorite Portlandite Quartz Rutile
a (kJ mol)1)
b (J K)1 mol)1)
63.4 57.4 80.3 56.4 13.1 23.6 104.0
3.90 )35.71 )29.31 )24.89 10.05 )52.92 )33.72
(11.6) (7.6) (5.9) (5.0) (2.8) (1.5) (5.1)
(10.83) (8.55) (5.72) (4.92) (4.99) (34.9) (4.88)
c (J K)1 mol)1)
10.93 (5.46)
d (J K)2 mol)1)
)0.0463 (0.0044)
e (J K)1 mol)1)
Data sources
)89.32 )72.98 )37.01 )59.73 )94.74 )18.52 )13.32
1 2, 3 4, 5 6 7 8–11 12
(8.95) (6.83) (2.82) (3.89) (3.47) (0.27) (3.77)
The model parameters are consistent with equilibrium constant defined on the molal concentration scale (Eq. 14). Values in parentheses are 1 standard error. Sources of experimental data – 1: Antignano & Manning (2008a); 2: Fein & Walther (1989); 3: Caciagli & Manning (2003); 4: Becker et al. (1983); 5: Tropper & Manning (2007a); 6: Tropper & Manning (2007b); 7: Walther (1986); 8: Hemley et al. (1980); 9: Fournier & Marshall (1983); 10: Walther & Orville (1983); 11: Manning (1994); 12: Antignano & Manning (2008b).
Table 2 Standard thermodynamic properties of dissolution at T = 25C and P = 1 bar. log K Apatite-F Calcite Corundum* Fluorite Portlandite Quartz Rutile
)11.31 )8.20 )12.54 )8.59 )2.82 )3.91 )16.45
(0.09) (0.22) (0.15) (0.11) (0.09) (0.02) (0.05)
DdsG (kJ mol)1)
DdsH (kJ mol)1)
DdsS (J K)1 mol)1)
DdscP (J K)1 mol)1)
DdsV (cm3 mol)1)
64.53 46.78 71.60 49.01 16.08 22.32 93.93
61.24 55.69 79.46 55.01 10.84 24.05 103.98
)11.02 29.90 26.37 20.13 )17.60 5.80 33.72
)90.13 )73.65 )37.35 )60.27 )95.60 )2.04 )13.43
)12.08 )9.87 )5.00 )8.07 )12.81 )2.50 )1.80
(0.49) (1.27) (0.83) (0.60) (0.53) (0.13) (0.27)
*Properties metastable with respect to gibbsite. Thermodynamic properties are referred to the molality concentration scale. Values in parentheses represent 1 standard error on the fit over the whole range of temperature and pressure. Comparison with experimental solubilities not used in regression reveals errors of 1 kJ for DdsG and 3 kJ for DdsH, respectively, at 25C and 1 bar for quartz (Rimstidt 1997).
26 D. DOLEJSˇ & C. E. MANNING
–1
2
(A) Quartz 1
1000
0 –1
400 300
–2
200
200
–3
100
Log molality AIO1.5
800
Log molality SiO2
(B) Corundum
1000
V+L
–2
800
–3
600
–4 400
–4 –5 –0.4
–0.3
–0.2
0
–0.1
–5 –0.4
0.1
–0.2
–0.3
–2
0 1000
(C) Calcite
800
0.1
0
–0.1
1000
(D) Apatite
Log molality CaCO3
400
–2
–3 200
–4
–5 –0.3
Log molality Ca5 (PO4)3F
800 600
–1
V+L
–0.2
–0.1
0
–3
600
–4 400
–5 –0.2
0.1
–1
–0.1
0
0.1
0
(E) Rutile
(F) Fluorite 1000
–2 –3
800
–4 600
–5 –6
Log molality CaF2
Log molality TiO2
1000
800
–1
600
–2 400
–3
400
–7 –0.4
–0.3
–0.2
–0.1
0
0.1
–4 –0.4
–0.3
–0.2
–0.1
0
Log H2O density (g cm–3) –1
Log molality Ca(OH)2
(G) Portlandite
1000 400
–2 200
–3
V+L
–4
–5 –0.4
–0.3
–0.2
–0.1
0
Log H2O density (g cm–3)
0.1
0.1
Fig. 4. Solubilities of rock-forming minerals in pure H2O, expressed on the molality scale, and their changes with the H2O density. Isotherms of solubility are calculated using the thermodynamic model (Eq. 14, Table 1), fitted to experimental data shown by filled symbols at selected temperatures and the liquid–vapor coexistence curve (experimental studies illustrated in open symbols were not used in fitting). Sources of experimental data – (A) solid upright triangles: Anderson & Burnham (1965); solid inverted triangles: Hemley et al. (1980); solid diamonds: Walther & Orville (1983); solid circles: Manning (1994); open circles: Kennedy (1950), Morey & Hesselgesser (1951), Wyart & Sabatier (1955), Kitahara (1960), Morey et al. (1962), Weill & Fyfe (1964), Crerar & Anderson (1971), at isotherms of 25, 100–900C; (B) solid diamonds: Becker et al. (1983) at 666–700C; solid circles: Tropper & Manning (2007a) at 700, 800, 900, 1000 and 1100C; open circles: Walther (1997) at 400, 500 and 600C; (C) solid diamonds: Fein & Walther (1989) at 400, 500 and 600C; solid circles: Caciagli & Manning (2003) at 500, 600 and 700C; (D) solid circles: Antignano & Manning (2008a) at 700, 800 and 900C; (E) solid circles: Antignano & Manning (2008b) at 800, 900 and 1000C; open circles: Tropper & Manning (2005) at 1000 and 1100C; open diamonds: Aude´tat & Keppler (2005) at 821–1025C; open triangle: Ryzhenko et al. (2006) at 500C; (F) solid circles: Tropper & Manning (2007b) at 600, 700, 800, 900 and 1000C; open diamonds: Stru¨bel (1965) at 200, 300, 400, 500 and 600C; (G) solid circles: Walther (1986) at 300, 400, 500 and 600C. The V + L represents subcritical coexistence of aqueous liquid and vapor. Calculated solubilities in this and subsequent figures are extended into metastable regions where some phases may either transform (e.g. portlandite to portlandite II), hydrate (e.g. corundum to boehmite or gibbsite) or melt (e.g. calcite at T > 750C).
Thermodynamic model for mineral solubility in aqueous fluids 27
0
2
(A) Quartz
(B) Corundum –1 1.2
Log molality AIO1.5
Log molality SiO2
1.2 1 20 5
0
0.4 2 1
–1 0.7 0.5 0.4
–2
20
0.6 5
–3 0.4
–5
–3 0.6
1.0
1.8
1.4
0.7
–1
(C) Calcite 1.0
20
0.8
1.5
1.2
1.8
(D) Apatite 1.2
1.2
Log molality Ca5(PO4)3F
Log molality CaCO3
1
–6 0.6
2.2
–1 10 0.8
–2 5 0.6
–3 –4
2
–4
V+L
0
10
–2
0.4 2
–5
–2 1.0 20
–3 0.8 10
–4
0.6
–5
5 0.7
1
–6 0.8
Log molality TiO2
–3 –4
1.6
–1
(E) Rutile 1.2 20 5 0.4 2
–2
1 0.7 0.4
–5
0.8
1.0
1.2
(F) Fluorite 1.0 20
1.2
0.8 10
–3 0.6 5
–4 0.4
–5
–6
2
–7 0.6
0.8
1.0
1.2
1.4
–6 0.6
1
0.8
1.0
1.2
0.7
1.4
0.4
1.6
1.8
2.0
Inverse temperature (10–3 K–1) –1
(G) Portlandite Log molality Ca(OH)2
Fig. 5. Solubilities of rock-forming minerals in pure H2O as a function of inverse absolute temperature. Isobars of solubility are indicated by solid and dotted curves, respectively, and labeled in upright numerals (pressure in kbar), whereas isochores are drawn by dashed and dot-dashed curves, respectively, and labeled in italic numerals (H2O density in g cm)3). Experimental solubility measurements are plotted along selected isobars and the liquid–vapor coexistence curve – (A) open circles: liquid–vapor coexistence curve; solid circles: 0.5 kbar; upright triangles: 1 kbar; inverted triangles: 2 kbar; diamonds: 5 kbar; squares: 10 kbar; hexagons: 20 kbar (Kennedy 1950; Morey & Hesselgesser 1951; Kitahara 1960; Morey et al. 1962; Weill & Fyfe 1964; Anderson & Burnham 1965; Crerar & Anderson 1971; Hemley et al. 1980; Walther & Orville 1983; Manning 1994); (B) diamonds: Becker et al. (1983) at 5, 10 and 20 kbar; circles: Tropper & Manning (2007a) at 10 and 20 kbar; (C) diamonds: Fein & Walther (1989) at 2 kbar; circles: Caciagli & Manning (2003) at 10 kbar; (D) circles: Antignano & Manning (2008a) at 10, 15 and 20 kbar; (E) diamond: Ryzhenko et al. (2006) at 1 kbar; circles: Antignano & Manning (2008b) at 10 and 20 kbar.
–2
1.4
–6 0.6
Log molality CaF2
–1
1.2
1.0
0.4
1.0
–2 0.8 10
–3 0.6
–4
–5 0.6
5
V+L
1 0.7
2
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Inverse temperature (10–3 K–1)
with remarkable accuracy (Figs 4A and 5A). In most cases, the scatter of multiple experiments exceeds the deviation between the experimental and calculated solubilities.
Despite the small data subset (n = 84) used in fitting the model, other experimental data are reproduced remarkably well (Fig. 5), with average deviation r = 0.017, 5.23%.
28 D. DOLEJSˇ & C. E. MANNING Corundum solubilities in pure H2O were experimentally determined by Becker et al. (1983), Ragnarsdo´ttir & Walther (1985), Walther (1997) and Tropper & Manning (2007a). The measurements of Becker et al. (1983) and Tropper & Manning (2007a) at 660–1100C and 2.5– 20 kbar are very consistent mutually and in their temperature and pressure dependencies (Figs 4B and 5B). The two studies of Walther and coworkers differ by log molality unit at H2 O > 0.5 g cm)3. Careful examination of the data reveals that isothermal solubilities of Walther (1997) have a very similar dependence on the solvent density as those of Becker et al. (1983) and Tropper & Manning (2007a), but his reported solubilities are systematically greater by 1.2–1.5 log molality units. This suggests formation of additional Al-bearing complexes in Walther’s (1997) experiments. Consequently, our thermodynamic model was fitted with experimental solubilities of Becker et al. (1983) and Tropper & Manning (2007a). The set of experiments (n = 24) is satisfactorily reproduced (r = 0.092, 3.12%) with the reduced, three-parameter form of the model, ln K ¼ 1=Rða=T þ b þ e ln Þ; of the model (Table 1). Rutile dissolves in pure H2O to form neutral hydroxyspecies but is characterized by extremely low solubility. Neglecting an early study of Ayers & Watson (1993), which reported anomalously high solubilities due to mass transport in experiments, the rutile solubility has been experimentally determined by Tropper & Manning (2005), Aude´tat & Keppler (2005), Ryzhenko et al. (2006) and Antignano & Manning (2008b). These studies cover a temperature range of 500–1100C at pressures of 1–20 kbar, and are complemented by low-temperature investigations of Vasilev et al. (1974), Zemniak et al. (1993) and Knauss et al. (2001) between 100 and 300C and up to 200 bar. Earlier measurements of Tropper & Manning (2005) are associated with high analytical uncertainties and limited to temperatures of 1000–1100C. By contrast, the diamond-anvil cell determinations by Aude´tat & Keppler (2005) gave consistently lower titanium concentrations and much lower dependence on temperature. A three-parameter fit to the data at 700–1000C and 7–20 kbar of Antignano & Manning (2008b) (n = 8, r = 0.028, 0.85%) reproduces rutile solubilities determined by fluid extraction at 500C and 1 kbar (Ryzhenko et al. 2006) and 300C and 200 bar (Knauss et al. 2001), respectively, to 0.2 log molality units (Fig. 4E). Calcite dissolves congruently in pure water, and its solubility has been experimentally determined at high temperatures and pressures by Walther & Long (1986), Fein & Walther (1989) and Caciagli & Manning (2003), in addition to exploratory studies of Schloemer (1952) and Morey (1962). The solubilities reported by Walther & Long (1986) are 0.3–0.5 log molality units lower than subsequent measurements of Fein & Walther (1989) and were, therefore, discarded. The experimental results of Fein & Walther (1989) and Caciagli & Manning (2003) show
consistent and continuous behavior from 400 to 750C and from 2 to 16 kbar, and were satisfactorily described by a three-parameter model (n = 21, r = 0.159, 9.28%) to the thermodynamic model (Figs 4C and 5C). Apatite solubility, using nearly pure fluorapatite (Durango, Mexico) comes from a single study of Antignano & Manning (2008a) at 700–900C and 10–20 kbar (n = 6). Individual experimental measurements show consistent isothermal linearity in the log q versus log m space (Fig. 4D) and were fitted to a three-parameter model with r = 0.050, 1.34%. Fluorite solubility in pure water has been investigated by Stru¨bel (1965) at 150–620C up to 2.1 kbar and by Tropper & Manning (2007b) at 600–1000C and 5–20 kbar. The earlier experiments are associated with very large analytical uncertainties and show no clear temperature dependence. By contrast, the recent measurements of Tropper & Manning (2007b) have high precision and display systematic variations with temperature and solvent density. This data set (n = 12) has been used for calibrating a three-parameter model fit (r = 0.086, 3.84%). Walther (1986) determined the solubility of portlandite in pure H2O. Although the pressure–temperature range was smaller than for the other minerals considered here, the data were included in this study for several comparative purposes: (i) portlandite produces basic solution with high ionic strength, (ii) the data set covers a temperature range of 300–600C, where significant changes in speciation are expected to occur and (iii) solubilities show a strong negative temperature dependence. As shown in Figs 4G and 5G, the experimental data (n = 34) are very well reproduced by a three-parameter fit to the thermodynamic model (r = 0.087, 1.82%), which extends and confirms its application to systems with substantial changes in dissociation and variable ionic strength.
DISCUSSION We fitted the experimental solubilities of seven minerals in pure water to a thermodynamic model (Eq. 14). The fitted experimental data cover a temperature range of 100– 1100C at pressures up to 20 kbar. The solubilities vary by up to six orders of magnitude on the molality concentration scale and are reproduced by the model to better than 5%, or 0.1 log molality units. Larger deviations, such as for calcite, are related not to the functional form of the thermodynamic model but rather to the scatter of repeated experiments at the same temperature and pressure or to poor consistency of different experimental or analytical methods. The solubilities of six minerals were reproduced satisfactorily by a three-parameter function only, whereas quartz solubility required two additional terms to describe the heat capacity of the dissolution reaction and its linear dependence on temperature. It is noteworthy that quartz is
Thermodynamic model for mineral solubility in aqueous fluids 29 the most soluble mineral in our data set and that solute silica is known to polymerize (Zotov & Keppler 2000, 2002; Newton & Manning 2002, 2003, 2008a). In our model, aqueous silica reflects total dissolved silica in an unspecified but implicit mixture of monomeric and polymerized species. As the equilibrium constant for the homogeneous polymerization equilibrium is temperature dependent (Newton & Manning 2003; Manning et al. 2010), the heat capacity term – required to fit experimental data over an extended temperature range – presumably incorporates differences in enthalpies and entropies of the monomer and dimer, respectively, as their proportions in the bulk solute change with increasing temperature and pressure. The isothermal, isobaric, isochoric and vapor-saturation solubilities presented in Figs 4 and 5 provide insights into interpolation and predictive capabilities of the new thermodynamic model, and illustrate the controls of aqueous speciation on these trends. In all cases, the mineral solubilities increase with the H2O density at constant temperature (Fig. 4). The magnitude of this increase, (¶ log m ⁄ ¶ log q)S, depends on parameter e only, i.e. it is dictated by the combination of the generalized Krichevskii and Born parameter of the solute species. Quartz, corundum and rutile, which congruently dissolve as neutral hydroxyspecies at neutral conditions, e.g. Si(OH)4, Al(OH)3 and Ti(OH)4 (Walther & Helgeson 1977; Bourcier et al. 1993; Knauss et al. 2001), show a relatively small increase in solubility with increasing water density at constant temperature, (¶ log m ⁄ ¶ log q)S = 1.6–4.5. By contrast, the solubilities of Ca-bearing minerals – calcite, apatite, fluorite and portlandite – exhibit much stronger dependence on the water density, (¶ log m ⁄ ¶ log q)S = 7.2–11.4. These phases involve congruent dissolution to a variety of charged species, ion pairs and neutral complexes (e.g. Ca2+, CO2 3 , ) ) + HCO 3 , OH , F , CaF , H2 PO4 , etc.; Walther 1986; Shock & Helgeson 1988; Fein & Walther 1989). The charged species cause much stronger electrostriction effects in the hydration shell as illustrated by the dependence of the species Helgeson–Kirkham–Flowers Born parameter on species charge (Shock & Helgeson 1988; Sverjensky et al. 1997). This translates to partial molar volumes of dissolution that grow more negative with increasing charge of the solute species. Using Eq. 19 at 25C and 1 bar, the partial molar volumes of dissolution for quartz, corundum and rutile are )1.8 to )5.0 cm)3 mol)1, whereas those for Ca-bearing phases are )8.1 to )12.8 cm)3 mol)1. Consequently, electrostriction in the vicinity of the charged species promotes the solubility increase with pressure by a factor of three to four. The spacing of isotherms in Fig. 4, (¶ log m ⁄ ¶T)q, varies between 7.7 · 10)3 and 6.1 · 10)2 and 5.9 · 10)4 and 4.7 · 10)3 K)1 at 25 and 800C, respectively, and q = 1 g cm)3. These variations are related to caloric properties in our model (Eq. 14) as follows:
o o ln K 1na c ¼ d : oT R T 2 T
ð22Þ
Equation 22 demonstrates that an increase in mineral solubility with temperature at constant density is free of hydration–compression effects. The caloric terms of all minerals broadly overlap at 25C, but quartz, corundum and rutile are more endothermic at 800C. Consequently, solubilities of these three minerals increase by a greater degree with temperature than that of the Ca-bearing minerals along isochores (cf. Fig. 1). This behavior is related either to distinct lattice enthalpies of oxides versus other minerals considered in this study, or to variably endothermic nature of chemical hydration of neutral versus charged aqueous species. The changes in mineral solubilities with temperature at constant pressure are depicted in Fig. 5. The solubility isobars are highly nonlinear and often show retrograde solubility effects. The slope
o ln K oð1=T Þ
¼ P
Dds H R
ð23Þ
corresponds to the standard enthalpy of dissolution, which consists of caloric and volumetric contributions (Eq. 17). The caloric term is quasilinear or, more often, constant (when c = d = 0). By contrast, the volumetric term scales by parameter e in the model and is responsible for the reversals of solubility isobars as temperature changes. The solvent contribution to the enthalpy (Eq. 17) depends on the isobaric expansivity of H2O, a, which is low, 0.5– 2.0 · 10)3 K)1 at low and high temperatures, respectively, but it diverges to infinity at the critical point of H2O. This feature is inherited as a maximum in the expansivity versus temperature at supercritical pressures until it disappears at approximately 8 kbar (Helgeson & Kirkham 1974; Fletcher 1993, p. 221). As a consequence, the eawT2 term in Eq. 17 becomes very negative as the expansivity approaches its maximum. Depending on the magnitude of e, the expansivity term may counteract the positive caloric contribution (a – cT – dT2), leading to negative enthalpy of dissolution and hence mineral solubility that decreases with temperature at the highest values of a. The magnitude of this effect increases towards the critical point of H2O. The onset and extent of the isobaric retrograde solubility is directly related to the parameter e of the density term, i.e. it depends on the generalized Krichevskii and Born parameters of aqueous solute (Fig. 6A). For Ca-bearing minerals, the isobaric retrograde solubility appears from low- to medium-grade metamorphic temperatures and covers the high-temperature space through the granulite facies (Figs 6 and 7). The remarkable offset of the
30 D. DOLEJSˇ & C. E. MANNING
(A) 5
Mineral solubilities along geothermal gradients
Po (–94.7) Ap (–89.3)
retrograde solubility of portlandite is due to its unusually low-standard enthalpy of dissolution (Table 2). By contrast, retrograde solubility behavior is suppressed where mineral dissolution produces neutral solutes, due to their very low electrostriction volumes (Table 2). Thus, for oxide minerals, retrograde solubility is more restricted in the pressure–temperature space (Fig. 7), and for the oxides considered in this study is limited to low pressures (rutile: 403 bar, quartz: 972 bar, corundum: 1231 bar).
The calibrated thermodynamic model (Table 1, Eq. 14) allows the calculation of mineral solubilities at geothermal gradients of interest, chosen here as 20C km)1 approximating the Barrovian continental gradient and 7C km)1, which corresponds to a representative subducting slab geotherm (cf. Fig. 1). In order to express the results in comparable quantities (ppm, wt% solute or vol% mineral), mass amounts of minerals have been recalculated to volumes using volumetric properties summarized in Table 3. The density of water has been calculated using the equation of state of Wagner & Pruß (2002), consistent with the model. Figure 8 shows that mineral solubilities in aqueous fluids vary by 10 orders of magnitude, from sub-ppb to greater than 10 wt%. For all minerals except portlandite, the solubilities regularly increase with increasing depth along both gradients. The solubility increases by three to seven orders of magnitude between 200 and 1100C. Although the portrayal using temperature as the independent variable (Fig. 8) shows little difference between the continental and the slab gradients, respectively, the limits of the diagram correspond to different pressures and depths. The observed trends are in agreement with general rule that mineral solubilities are mainly dependent on temperature. The calculations also imply that substantial changes in the solubilities are expected around thermal disturbances, e.g. magma chambers, emplaced at a given depth. The sequence of minerals in Fig. 8 starts with quartz and calcite, followed by portlandite, fluorite, and apatite, whereas corundum and rutile are the least soluble and mobile phases in pure H2O. These predictions are in good agreement with observed or inferred element mobilities during crustal fluid flow (e.g. Ague 2003). Titanium and aluminum solubilities are quite low in pure H2O at all conditions. While this nominally supports their use as immobile elements in mass transfer, hydrothermal alteration and ore deposit studies (Maclean 1990; Maclean & Barrett 1993; Ague 1994; Dolejsˇ & Wagner 2008), it is important to note that both elements may form soluble complexes with other elements (alkalies, halogens, silica, etc.) which can cause them to be mobilized in certain metasomatic environments (e.g. Tagirov & Schott 2001; Manning 2007; Manning et al. 2008; Newton & Manning 2008b).
GEOLOGICAL APPLICATIONS
Mineral solubilities at constant pressure
We illustrate the performance and application of the thermodynamic model by evaluating the mineral solubilities in aqueous fluids along representative geotherms and by applying transport theory to assess mass transfer and fluid flux associated with model metasomatic scenarios.
Mineral solubilities at constant pressure allow investigation of the potential for element mobility along flow paths associated with thermally induced gradients at a fixed crustal level, such as out- and inflow accompanying cooling of intrusions (Norton & Knight 1977; Hanson 1992;
4
Cc (–73.0)
Pressure (kbar)
Fr (–59.7)
3
2 Co (–37.0) Qz (–18.5)
1
Ru (–13.3)
0
0
200
400
600
800
1000
Temperature (°C) 0
(B)
Ru Qz
(cm3 mol–1)
–6
dsV
–3
–9
Co Fr Cc
–12
–15
Ap
Po
100
200
300
400
Temperature (°C) Fig. 6. Correlations between the isobaric retrograde solubility and its onset, and (A) model parameter e (numerical labels), which scales the solvent density, and (B) partial volume of dissolution (per mole of solute at T = 25C and P = 1 bar) versus the onset of isobaric retrograde solubility (at P = 300 bar).
Thermodynamic model for mineral solubility in aqueous fluids 31
Quartz
(A) 105
20
Corundum
(B) 103
5
20
10
104
1 0.7 0.5 0.4
103
102 200
500
800
Solubility (ppm Al2O3)
Solubility (ppm SiO2)
2
101
1
10–1
Calcite
Apatite-F Calcite Corundum Fluorite Portlandite Quartz Rutile
800
1100
Rutile 20 5
102 5
102 101 2
100
2 1 0.7
101
0.4
100
10–1
10–1 0.4
10–2 200
500
0.7
1
800
Temperature (°C)
6
500
10
103
Table 3 Volumetric properties of minerals. 5
0.4
(D) 103
20
Solubility (ppm TiO2)
Solubility (ppm CaCO3)
Fig. 7. Retrograde solubility behavior illustrated for (A) quartz, (B) corundum, (C) calcite and (D) rutile. Gray shading indicates regions where the isobaric mineral solubility decreases with increasing temperature. Experimental data and symbols are as in Fig. 5.
0.7
Temperature (°C)
Temperature (°C)
(C) 104
2
100
10–2 200
1100
5
102
V (cm3 mol)1)
¶V ⁄ ¶T · 10 (cm3 K)1 mol)1)
¶V ⁄ ¶P · 10 (cm3 bar)1 mol)1)
Data sources
157.14 36.59 25.37 23.30 32.25 22.55 18.65
12.24* 12.24 6.70 41.97 19.14* 11.03 5.33
)5.38* )5.38 )1.21 )10.15 )5.11* )5.21 )1.00
1 2 2 3 4 2 2
*Estimate obtained by correlation with similar substance. Molar volume of mineral at temperature and pressure of interest is defined by: VP,T = V + (¶V ⁄ ¶T)T + (¶V ⁄ ¶P)P where temperature, T, is in kelvin and pressure, P, in bar. Sources of data – 1: Mackie & Young (1974); 2: Holland & Powell (1998), updated in 2004 and available from the Thermocalc website (Richard White, University of Mainz); 3: Speziale & Duffy (2002); 4: Megaw (1933).
Norton & Dutrow 2001). Figure 9 compares predicted isobaric temperature dependence of mineral solubilities at 1 and 10 kbar respectively. With the exception of portlandite, the 10 kbar mineral solubilities and their temperature dependencies are very similar to those along geothermal gradients (Fig. 8); however, in detail, the magnitude of solubility increase above approximately 700C is lower. Portlandite solubility declines with rising temperature. These features are a consequence of the larger decrease in H2O density with temperature than along the geotherms.
1100
10–2 200
500
800
1100
Temperature (°C)
The solubility behavior changes dramatically at low pressures, e.g. 1 kbar (Fig. 9B), as is expected from the variations in isobaric expansivity of H2O. Rutile, which has the smallest parameter e is above its retrograde maximum, and its solubility increases monotonously with temperature. Quartz and corundum show a solubility plateau above 500C; hence, mass transfer will be minimized at these conditions. The solubilities of Ca-bearing phases substantially decrease above 350–450C; that of portlandite shows a steady decline over the whole temperature range. These results imply that temperature gradients in isobaric aquifers may alone be responsible for substantial decoupling of mineral precipitation and dissolution. The Ca-bearing minerals all dissolve as the fluid cools down to approximately 400C, and the precipitation of, for instance, calcite or fluorite will be suppressed until temperature declines to below 350C. This may explain a generally late nature of carbonate precipitation in hydrothermal ore veins (Mangas & Arribas 1987) or low temperatures of fluorite–barite mineralization (Baatartsogt et al. 2007). By contrast, quartz and aluminosilicate alteration reactions are predicted to experience a reversal (cf. Fournier & Potter 1982; Fournier 1999), i.e. mineral precipitation at magmatic and low temperatures, separated by a dissolution gap at near-critical conditions (cf. Fig. 7A, B). This mechanism has been postulated for the formation of quartz veins and desilicification in plutonic environments (Nichols & Wiebe 1998).
32 D. DOLEJSˇ & C. E. MANNING
(A)
6
(B) dT/dz = 20°C km–1
6
dT/dz = 7°C km–1
Qz
Qz
4
Cc Po
Fr
Log solubility (ppm)
Log solubility (ppm)
4
1 wt.%
Co Ap
2 Ru
0
–2
Ap
500
800
Ru
0
–4 200
1100
500
800
(B)
Fig. 8. Solubilities of rock-forming minerals in pure H2O along geothermal gradients of (A) 20 and (B) 7C km)1. Solubilities of oxides are shown by solid curves whereas Ca-bearing phases are in dashed style.
6
P = 10 kbar
P = 1 kbar Qz
1 wt.%
4
Log solubility (ppm)
Cc
Po Fr
Co
2 Ap
Ru
0
–2
1 wt.% Qz
2
0 Co Ru Cc
–2
500
800
–4 200
1100
Temperature (°C)
Fr
Po
Ap
–4 200
1100
Temperature (°C)
(A) 6
500
800
1100
Temperature (°C)
(A)
Fig. 9. Solubilities of rock-forming minerals in pure H2O at constant pressure of (A) 10 and (B) 1 kbar. Line styles are as in Fig. 8.
(B)
14
14
dT/dz = 20°C km–1
Log integrated fluid flux (m3 m–2)
Log integrated fluid flux (m3 m–2)
Co
2
Temperature (°C)
Log solubility (ppm)
Fr
Cc
–2
–4 200
4
Po
1 wt.%
12 Ru
10
Ap Po
8
Fr 1%
Co
Cc
6 Shear zone veins 100%
Qz
4 Pervasive metamorphic flow
2 200
500
8 00
Temperature (°C)
1100
dT/dz = 7°C km–1
12 10 Ru Co
8
Fr
Ap 1%
Po
6 Shear zone veins 100%
Cc Qz
4 Pervasive metamorphic flow
2 200
500
800
Temperature (°C)
Mass transfer and mineral mobility in the Earth’s interior The present thermodynamic model provides a self-consistent formulation of all thermodynamic properties of mineral dissolution equilibria and their pressure and temperature dependence. We illustrate its utility by calculating the intensity of mass transfer and metasomatism and the related integrated fluid fluxes in several representative settings. Transport theory (see Appendix B) was applied to the calculation of the time-integrated fluid fluxes necessary to precipitate a unit volume of vein material (1 m3). The integrated fluid fluxes for mineral precipitation along the
1100
Fig. 10. The time-integrated fluid flux necessary to precipitate mineral vein filling (1 m3) as a function of temperature along geothermal gradients of (A) 20 and (B) 7C km)1. Typical integrated fluid fluxes for diffuse metamorphic fluid flow and focused shear zone flow (Ague 2003) are shown by dotted lines. The range of partial vein filling (100 to 1 vol% of precipitating phase) is indicated by gray areas. Line styles are as in Fig. 8.
geothermal gradients of 20 and 7C km)1 are very similar, but vary significantly with temperature. They range from 107–1015 m3 m)2 at 200C to 104–108 m3 m)2 at 1100C (Fig. 10). When compared with characteristic integrated fluid fluxes during diffusive metamorphic flow and in crustal shear zones (Ferry 1994; Ague 2003), medium- to high-grade metamorphic temperatures are sufficient to cause substantial mobility of quartz, calcite and fluorite (in decreasing order; Fig. 10). Conversely, a predefined integrated fluid flux may be used to calculate and compare the magnitudes of mass transfer (in volume fraction precipitated or dissolved; Fig. 11). Along the geotherms and at a
Thermodynamic model for mineral solubility in aqueous fluids 33
(B)
Log amount of mineral precipitated (vol. fraction)
0
dT/dz = 20°C
km–1
Log amount of mineral precipitated (vol. fraction)
(A) Qz Cc
–2
Fr Po
–4
Co
Ap
Ru
–6
–8
–10 200
500
800
0
–2
Fr Ap Co
–4
Ru
–6
–8
500
Temperature (°C)
1100
(D) 0
Log amount of mineral precipitated or dissolved (vol. fraction)
0
Log amount of mineral precipitated or dissolved (vol. fraction)
800
Temperature (°C)
(C)
Fig. 11. Amount of precipitated or dissolved minerals produced by the integrated fluid flux, q = 105 m3 m)2 (Ague 2003) at geothermal gradients of (A) 20 and (B) 7C km)1, and at constant pressure of (C) 10 and (D) 1 kbar. Mineral precipitation (as temperature decreases) is indicated by black curves, whereas dissolution is shown by gray dashed and dotted patterns. Gray area indicates a range between 1 and 100 vol% in the rock. Line styles are as in Fig. 8.
Cc Po
–10 200
1100
Qz
dT/dz = 7°C km–1
Qz Cc
Po Fr
–2
Co Ap
–4
Ru
–6
–8
P = 10 kbar & dT/dz = 200°C km–1 –10 200
500
800
Temperature (°C)
constant pressure of 10 kbar, the integrated fluid flux characteristic for metamorphic shear zones (105 m3 m)2; Ague 2003) leads to mobilities of SiO2 from 1 vol% at 300–350C to 10 vol% at 600–620C and a complete silicification above 900C. This is in good agreement with macroscopic observation of quartz segregations and veining from greenschist facies conditions. The fluid-mediated mobilities of calcite are one-half to one order of magnitude lower, but metasomatism of fluorite and apatite components may reach vol% levels at the highest subduction temperatures (Fig. 11b). Amounts of precipitation of rutile and corundum are low, from sub-ppm (at 200C) to a few tenths of vol% (at 1100C) but, importantly, these values are probably still significant for trace element redistribution during metamorphic events, focused fluid flow, or hydrothermal alteration. During lateral fluid flow at constant pressure the extent of the mineral precipitation is expected to vary along the cooling path, or precipitation will alternate with mineral dissolution and loss in certain temperature segments (Fig. 11C, D). At 10 kbar, retrograde solubility occurs for portlandite above 290C and above 1000C for apatite. Other Ca-bearing phases such as calcite or fluorite variably precipitate during cooling (2–7 vol%), whereas the capability of oxide precipitation (quartz, corundum and rutile) strongly decreases with decreasing temperature. At 1 kbar, all minerals except rutile exhibit rapid changes in
1100
Qz
–2 Po
–4
Cc Fr Ru Co
–6
Ap
–8
P = 1 kbar & dT/dz = 200°C km–1 –10 200
500
800
1100
Temperature (°C)
precipitation and ⁄ or dissolution as a function of temperature (Fig. 11D). The intensity of silicification drops to below 1 vol% between 700 and 400C, whereas corundum will dissolve between 720 and 530C bracketed by precipitation at high or low temperatures. No Ca-bearing minerals precipitate above 390C. These results are applicable to spatial hydrothermal zoning in the vicinity of upper crustal intrusions. The intensity of silicification varies by nearly two orders of magnitude and quartz veining is expected to be most intense at 300– 400C. The behavior of aluminum, as judged from the corundum solubility, reverses from precipitation and dissolution, and back, which will probably enhance its local redistribution and hence mobility. Carbonates will not precipitate in high-temperature contact aureoles even if the fluids are CO2 bearing, which promotes, by extrapolation, the formation of skarns replacing carbonate host rocks. Limitations and extensions of the modeling Our mass transport calculations illustrate the applicability and versatility of a density-based model to investigate fluid-mediated mass transfer in the Earth’s crust and upper mantle. The results also highlight some limitations that stem from focusing only on congruent dissolution equilibria in pure water in the modeling. For example, for phases where the fluid–mineral interaction is controlled by the
34 D. DOLEJSˇ & C. E. MANNING formation of additional species or complexing, our calculations provide a minimum estimate of the mass transfer (that is, the amount of mineral precipitated or dissolved) or a maximum estimate for the integrated fluid flux. These observations indicate that, if formed under local equilibrium, metamorphic veins containing sparingly soluble minerals such as kyanite or rutile require either enormously large fluid fluxes or the presence of complexing agents, such as, halide, carbonate or aluminosilicate ligands and their polymeric successors (Tagirov & Schott 2001; Manning 2007; Antignano & Manning 2008b). The magnitude of this effect can be quantified by considering that ¶ log m ⁄ ¶z is approximately independent of the nature of the mineral (Fig. 8); hence, the integrated fluid flux scales in direct proportion to the solubility increase in the presence of other aqueous complexes. The presence of 10 wt% dissolved silicates lowers the necessary fluid flux for rutile crystallization by a factor of 13 at 900C and 10 kbar, or the quartz saturation enhance the kyanite precipitation by a factor of 6 at 700C and 10 kbar. This brings the mobility of Ti and Al to wt% level at the highest temperatures modeled in this study (Fig. 11). Increasing self-dissociation of H2O at high pressures and the presence of other complexing ions also play important roles.
CONCLUSIONS (1) We have developed a new thermodynamic model for dissolution of minerals in aqueous fluids at high temperatures and pressures. The model incorporates thermodynamic contributions from lattice breakdown, ionization and hydration, which depend on temperature, and the effects associated with the compression in the hydration sphere and electrostatic solute–solvent interactions, which are formulated as a function of solvent volumetric properties. The solvent density term has the form of the generalized Krichevskii parameter, which is a finite and smooth function near the critical point of water, and has a simple linear scaling with the reciprocal dielectric constant. (2) Experimental solubilities of seven rock-forming minerals were used to calibrate the model and demonstrate its performance. With the exception of quartz, solubilities in aqueous fluids at metamorphic and magmatic conditions can be described by three parameters to within the experimental scatter or accuracy. The solubility of quartz required a five-parameter formulation, which includes heat capacity and its temperature dependence. We propose that these terms probably assimilate the distinct enthalpy and entropy of the silica monomer and dimer, respectively, and ⁄ or deviations from the infinite-dilution limiting behavior as fluids become solute rich. (3) Solubilities of all seven rock-forming minerals increase smoothly with temperature along metamorphic geotherms or water isochores. Temperature dependence is
fairly similar for all solid phases but portlandite. The solubility of a given phase increases for a given phase by three to seven orders of magnitudes as temperature rises from 200 to 1100C along typical geotherms. At constant pressure, however, mineral solubilities initially increase with rising temperature but subsequently drop. This effect is caused by a reversal in isobaric expansivity of the aqueous solvent, which propagates into the enthalpy of dissolution. The onset of retrograde solubility typically occurs at 300– 400C and it is a characteristic of all minerals. (4) Solute speciation, volume of dissolution and the prograde–retrograde solubility behavior are inter-related. Oxide minerals such as quartz, corundum and rutile, which dissolve as predominantly neutral species, exhibit very small dependence on the solvent properties. Consequently, their isobaric retrograde solubility is limited to pressures below 1.3 kbar. The Ca-bearing phases, by contrast, produce variably charged species upon dissolution, where the electrostriction effects are more significant. Calcite, fluorite, apatite and portlandite show a decrease in mineral solubility at medium to high metamorphic grades over a wide range of pressures. (5) Application of solute transport theory to our thermodynamic model permits calculation of time-integrated fluid fluxes, which are necessary to precipitate mineral veins during metamorphic events. The integrated fluid fluxes along geotherms of 20 and 7C km)1 vary from 104 to 1015 m3 m)2 in the following sequence: quartz, calcite, fluorite, corundum and apatite, and rutile. This is in broad agreement with observations of high mobility and veining of quartz and calcite as the most mobile-predicted phases in many metamorphic environments. Conversely, typical integrated fluid fluxes in crustal shear zones produce a transfer of quartz and calcite in quantities of several tens of vol%, whereas the solubility of apatite or rutile lies below 1000 ppm, which may still be important for the trace element budget in metasomatized rocks. (6) The small number of parameters in the thermodynamic model allows correlation with the standard thermodynamic properties at arbitrary reference conditions (e.g. 25C and 1 bar), which are readily available. In the appendix A we provide relationships for transformation of standard enthalpies, entropies, volumes and heat capacities into the model parameters. This approach enables utilization of existing thermodynamic data (e.g. the Helgeson– Kirkham–Flowers equation of state) in the framework of our model and allows extrapolation of standard thermodynamic properties over a wide range of metamorphic and magmatic temperatures and pressures.
ACKNOWLEDGEMENTS This study was supported by a postdoctoral fellowship from the Elite Network of Bavaria, the Ministry of Education of the Czech Republic research project MSM002162085, by
Thermodynamic model for mineral solubility in aqueous fluids 35 the Czech Science Foundation project 205 ⁄ 09 ⁄ P135 (to D.D.) and by a research award from the Alexander von Humboldt Foundation (to C.E.M.). We appreciate stimulating discussions with T. Wagner (ETH Zu¨rich) as well as critical comments by the journal reviewers: G. M. Anderson and G. Pokrovski that helped us to improve the manuscript.
REFERENCES Ague JJ (1994) Mass transfer during Barrovian metamorphism of pelites, south-central Connecticut. I. Evidence for changes in composition and volume. American Journal of Science, 294, 989–1057. Ague JJ (1998) Simple models of coupled fluid infiltration and redox reactions in the crust. Contributions to Mineralogy and Petrology, 132, 180–97. Ague JJ (2003) Fluid flow in the deep crust. In: Treatise of Geochemistry, v. 3: The crust (ed. Rudnick RL), pp. 195–228. Elsevier, Amsterdam. Akinfiev NN (2001) Equation of state of SiO2aq for description of silica dissolution at temperatures of 0–600C and pressures of 1–1000 bar. Geochemistry International, 39, 1242–4. Akinfiev NN, Diamond LW (2003) Thermodynamic description of aqueous nonelectrolytes at infinite dilution over a wide range of state parameters. Geochimica et Cosmochimica Acta, 67, 613–27. Akinfiev NN, Diamond LW (2004) A three-parameter EOS to describe aqueous non-electrolytes at infinite dilution over a wide range of state parameters, with preliminary application to 1:1 electrolytes. Fluid Phase Equilibria, 222 ⁄ 223, 31–37. Akinfiev NN, Diamond LW (2009) A simple predictive model of quartz solubility in water-salt-CO2 systems at temperatures up to 1000C and pressures up to 1000 MPa. Geochimica et Cosmochimica Acta, 73, 1597–608. Anderson GM (2005) Thermodynamics of Natural Systems, 2nd edn. Cambridge University Press, Cambridge. Anderson GM, Burnham CW (1965) The solubility if quartz in supercritical water. American Journal of Science, 263, 494–511. Anderson GM, Castet S, Schott J, Mesmer RE (1991) The density model for estimation of thermodynamic parameters of reactions at high temperatures and pressures. Geochimica et Cosmochimica Acta, 55, 1769–79. Angell CA (1983) Supercooled water. Annual Review in Physical Chemistry, 34, 593–630. Antignano A, Manning CE (2008a) Fluorpatite solubility in H2O and H2O–NaCl at 700 to 900C and 0.7 to 2.0 GPa. Chemical Geology, 251, 112–9 & 255, 294. Antignano A, Manning CE (2008b) Rutile solubility in H2O, H2O–SiO2, and H2O–NaAlSi3O8 fluids at 0.7–2.0 GPa and 700–1000C: implications for mobility of nominally insoluble elements. Chemical Geology, 255, 283–93. Armellini FJ, Tester JW (1993) Solubility of sodium chloride and sulfate in sub- and supercritical water vapor from 450–500C and 100–250 bar. Fluid Phase Equilibria, 84, 123–42. Atkins PW, MacDermott AJ (1982) The Born equation and ionic solvation. Journal of Chemical Education, 59, 359–60. Aude´tat A, Keppler H (2005) Solubility of rutile in subduction zones fluids, as determined by experiments in the hydrothermal diamond anvil cell. Earth and Planetary Science Letters, 232, 393–402. Ayers JC, Watson EB (1993) Rutile solubility and mobility in supercritical aqueous fluids. Contributions to Mineralogy and Petrology, 114, 321–30.
Baatartsogt B, Schwinn G, Wagner T, Taubald H, Beitter T, Markl G (2007) Contrasting paleofluid systems in the continental basement: a fluid inclusion and stable isotope study of hydrothermal vein mineralization, Schwarzwald district, Germany. Geochimica et Cosmochimica Acta, 64, 123–47. Bandura AV, Lvov SN (2006) The ionization constant of water over wide ranges of temperature and density. Journal of Physical and Chemical Reference Data, 35, 15–30. Baumgartner LP, Ferry JM (1991) A model for coupled fluid-flow and mixed-volatile mineral reactions with applications to regional metamorphism. Contributions to Mineralogy and Petrology, 106, 273–85. Bear J (1972) Dynamics of Fluids in Porous Media. Elsevier, New York. Becker KH, Cemicˇ L, Langer KEOE (1983) Solubility of corundum in supercritical water. Geochimica et Cosmochimica Acta, 47, 1573–8. Ben-Amotz D, Raineri FO, Stell G (2005) Solvation thermodynamics: theory and applications. Journal of Physical Chemistry B, 109, 6866–78. Ben-Naim A (2006) Molecular Theory of Solutions. Oxford University Press, Oxford. Borg RJ, Dienes GJ (1992) The Physical Chemistry of Solids. Academic Press Inc., Boston, MA. Born M (1920) Volumen und Hydratationswa¨rme der Ionen. Zeitschrift fu¨r Physik, 1, 45–8. Bourcier WL, Knauss KG, Jackson KJ (1993) Aluminum hydrolysis constants to 250C from boehmite solubility measurements. Geochimica et Cosmochimica Acta, 57, 747–62. Caciagli NC, Manning CE (2003) The solubility of calcite in water at 5–16 kar and 500–800C. Contributions to Mineralogy and Petrology, 146, 275–85. Clarke R, Hneˇdkovsky´ L, Tremaine PR, Majer V (2000) Amino acids under hydrothermal conditions: apparent molar heat capacities of aqueous a-alanine, b-alanine, glycine, and proline at temperature from 298 to 500 K and pressures up to 30.0 MPa. Journal of Physical Chemistry B, 104, 11781–93. Cooney WJ, O’Connell JP (1987) Correlation of partial molar volumes at infinite dilution of salts in water. Chemical Engineering Communications, 56, 341–7. Crerar DA, Anderson GM (1971) Solubility and solvation reactions of quartz in dilute hydrothermal solutions. Chemical Geology, 8, 107–22. Crovetto R, Wood RH, Majer V (1991) Revised densities of (x CO2 + (1 ) x)H2O) with x < 0.014 at supercritical conditions: molar volumes, partial molar volumes of CO2 at infinite dilution, and excess molar volumes. Journal of Chemical Thermodynamics, 23, 1139–46. Dolejsˇ D, Wagner T (2008) Thermodynamic modeling of non-ideal mineral–fluid equilibria in the system Si–Al–Fe–Mg– Ca–Na–K–H–O–Cl at elevated temperatures and pressures: implications for hydrothermal mass transfer in granitic rocks. Geochimica et Cosmochimica Acta, 72, 526–53. Ellis AJ (1966) Partial molal volumes of alkali chlorides in aqueous solution to 200. Journal of Chemical Society (A), 1966, 1579– 84. Eugster HP, Baumgartner LP (1987) Mineral solubilities and speciation in supercritical metamorphic fluids. Reviews in Mineralogy, 17, 367–403. Fein J, Walther JV (1989) Calcite solubility and speciation in supercritical NaCl–HCl aqueous fluids. Contributions to Mineralogy and Petrology, 103, 317–24. Fernande´z DP, Goodwin AR, Lemmon EW, Levelt Sengers JMH, Williams RC (1997) A formulation for the static permittivity of
36 D. DOLEJSˇ & C. E. MANNING water and steam at temperatures from 238 K to 873 K at pressures up to 1200 MPa, including derivatives and Debye–Hu¨ckel coefficients. Journal of Physical and Chemical Reference Data, 26, 1125–66. Ferry JM (1994) Overview of the petrological record of fluid flow during regional metamorphism in northern New England. American Journal of Science, 294, 905–88. Ferry JM, Burt DM (1982) Characterization of metamorphic fluid composition through mineral equilibria. Reviews in Mineralogy, 10, 207–62. Ferry JM, Dipple GM (1991) Fluid flow, mineral reactions, and metasomatism. Geology, 19, 211–4. Ferry JM, Gerdes ML (1998) Chemically reactive fluid flow during metamorphism. Annual Review of Earth and Planetary Sciences, 26, 255–87. Fletcher P (1993) Chemical Thermodynamics for Earth Scientists. Longman Scientific & Technical, Harlow. Fournier RO (1983) A method for calculating quartz solubilities in aqueous sodium chloride solutions. Geochimica et Cosmochimica Acta, 47, 579–86. Fournier RO (1999) Hydrothermal processes related to movement of fluid from plastic into brittle rock in the magmatic-epithermal environment. Economic Geology, 94, 1193–212. Fournier RO, Marshall WL (1983) Calculation of amorphous silica solubilities at 25 to 300C and apparent cation hydration numbers in aqueous salt solutions using the concept of effective density of water. Geochimica et Cosmochimica Acta, 47, 587–96. Fournier RO, Potter RW, II (1982) An equation correlating the solubility of quartz in water from 25 to 900C at pressures up to 10,000 bars. Geochimica et Cosmochimica Acta, 46, 1969– 73. Franck EU (1956a) Hochverdichteter Wasserdampf III. Ionendissoziation von HCl, KOH und H2O im u¨berkritischem Wasser. Zeitschrift fu¨r Physikalische Chemie, Neue Folge, 8, 192–206. Franck EU (1956b) Zur Lo¨slichkeit fester Stoffe in verdichteten Gasen. Zeitschrift fu¨r Physikalische Chemie, Neue Folge, 6, 345– 55. Franck EU, Rosenzweig S, Christoforakos M (1990) Calculation of the dielectric constant of water to 1000C and very high pressures. Berichte der Bunsen-Gesellschaft, 94, 199–203. Frantz JD, Marshall WL (1982) Electrical conductances and ionization constants of CaCl2 and MgCl2 in aqueous solutions at temperatures to 600C and pressures to 4000 bars. American Journal of Science, 282, 1666–93. Galobardes JV, Van Hare DR, Rogers LB (1981) Solubility of sodium chloride in dry steam. Journal of Chemical Engineering Data, 26, 363–5. Gerya TV, Maresch WV, Burchard M, Zakhartchouk V, Doltsinis NL, Fockenberg T (2005) Thermodynamic modeling of solubility and speciation of silica in H2O–SiO2 fluid up to 1300C and 20 kbar based on the chain reaction formalism. European Journal of Mineralogy, 17, 269–83. Hamann SD, Lim SC (1954) The volume changes on ionization of weak electrolytes. Australian Journal of Chemistry, 7, 329– 34. Hanson RB (1992) Effects of fluid production on fluid flow during regional and contact metamorphism. Journal of Metamorphic Geology, 10, 87–97. Harvey AH (1998) Thermodynamic Properties of Water: Tabulation from the IAPWS Formulation 1995 for the Thermodynamic Properties of Ordinary Water Substance for General and Scientific Use. NIST Internal Report 5078.
Helgeson HC, Kirkham DH (1974) Theoretical prediction of the thermodynamic behavior of aqueous electrolytes at high pressures and temperatures. I. Summary of the thermodynamic ⁄ electrostatic properties of the solvent. American Journal of Science, 274, 1089–198. Helgeson HC, Kirkham D, Flowers GC (1981) Thermodynamic behavior of aqueous electrolytes at high pressures and temperatures. IV. Calculation of activity coefficients, osmotic coefficients, and apparent molal and standard and relative partial molal properties to 600C and 5 kb. American Journal of Science, 281, 1249–516. Hemley JJ, Montoya JW, Marinenko JW, Luce RW (1980) Equilibria in the system Al2O3–SiO2–H2O and some general implications for alteration ⁄ mineralization processes. Economic Geology, 75, 210–28. Higashi H, Iwai Y, Matsumoto K, Kitani Y, Okazaki F, Shimoyama Y, Arai Y (2005) Measurement and correlation for solubilities of alkali metal chlorides in water vapor at high temperature and pressure. Fluid Phase Equilibria, 228 ⁄ 229, 547–551. Hillert M (2008) Phase Equilibria, Phase Diagrams and Phase Transformations. Their Thermodynamic Basis, 2nd edn. Cambridge University Press, Cambridge. Holland TJB, Powell R (1998) An internally consistent thermodynamic data set for phases of petrological interest. Journal of Metamorphic Geology, 16, 309–43. Hummel W, Berner U, Curti E, Pearson FJ, Thoenen T (2002) Nagra ⁄ PSI Chemical Thermodynamic Database 01 ⁄ 01. Universal Publishers, Parkland, FL. Jamtveit B, Yardley B (Eds) (1997) Fluid Flow and Transport in Rocks. Chapman & Hall, London. Johnson JW, Oelkers EH, Helgeson HC (1992) SUPCRT92: a software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000C. Computers & Geosciences, 18, 899–947. Kebarle P (1977) Ion thermochemistry and solvation from gas phase ion equilibria. Annual Review of Physical Chemistry, 28, 445–76. Kennedy GC (1950) A portion of the system silica-water. Economic Geology, 45, 629–53. Kirkwood JG, Buff FP (1951) The statistical mechanical theory of liquids. Journal of Chemical Physics, 19, 774–82. Kitahara S (1960) The solubility equilibrium and the rate of solution of quartz in water at high temperatures and high pressures. Review of Physical Chemistry of Japan, 30, 122–30. Knauss KG, Dibley MJ, Bourcier WL, Shaw HF (2001) Ti(IV) hydrolysis constants derived from rutile solubility measurements made from 100 to 300C. Applied Geochemistry, 16, 1115–28. Levelt Sengers JMH (1991) Solubility near the solvent’s critical point. Journal of Supercritical Fluids, 4, 215–22. Mackie PE, Young RA (1974) Fluorine-chlorine interaction in fluor-chlorapatite. Journal of Solid State Chemistry, 11, 319– 29. Maclean WH (1990) Mass change calculations in altered rock series. Mineralium Deposita, 25, 44–9. Maclean WH, Barrett TJ (1993) Lithogeochemical technique using immobile elements. Journal of Geochemical Exploration, 48, 109–33. Majer V, Sedlbauer J, Bergin G (2008) Henry’s law constant and related coefficients for aqueous hydrocarbons, CO2 and H2S over a wide range of temperature and pressure. Fluid Phase Equilibria, 272, 65–74.
Thermodynamic model for mineral solubility in aqueous fluids 37 Mangas J, Arribas A (1987) Fluid inclusion study in different types of tin deposits associated with the Hercynian granites of western Spain. Chemical Geology, 61, 193–208. Manning CE (1994) The solubility of quartz in H2O in the lower crust and upper mantle. Geochimica et Cosmochimica Acta, 58, 4831–9. Manning CE (2007) Solubility of corundum + kyanite in H2O at 700C and 10 kbar: evidence for Al–Si complexing at high pressure and temperature. Geofluids, 7, 258–69. Manning CE, Wilke M, Schmidt C, Cauzid J (2008) Rutile solubility in albite–H2O and Na2Si3O7–H2O at high temperatures and pressures by in-situ synchrotron radiation micro-XRF. Earth and Planetary Science Letters, 272, 730–7. Manning CE, Antignano A, Lin HA (2010) Premelting polymerization of crustal and mantle fluids, as indicated by solubility of albite + paragonite + quartz in H2O at 1 GPa and 350–620C. Earth and Planetary Science Letters, 292, 325–36. Marshall WL (2008a) Dielectric constant of water discovered to be simple function of density over extreme ranges from )35 to +600C and to 1200 MPa (12000 Atm.), believed universal. Nature Precedings, doi: 10.1038/npre.2008.2472.1, 1–40. Marshall WL (2008b) Aqueous electrolyte ionization over extreme ranges as simple fundamental relation with density and believed universal; sodium chloride ionization from 0 to 1000C and to 1000 MPa (10000 Atm.). Nature Precedings, hdl:10101 ⁄ npre. 2008.2476.1, 1–40. Marshall WL, Franck EU (1981) Ion product of water substance, 0–1000 C, 1–10,000 bars. New international formulation and its background. Journal of Physical and Chemical Reference Data, 10, 295–304. Marshall WL, Quist AS (1967) A representation of isothermal ion–ion-pair-solvent equilibria independent of changes in dielectric constant. Proceedings of the National Academy of Sciences, 58, 901–6. Martynova OI (1964) Some problems of the solubility of involatile inorganic compounds in water vapour at high temperatures and pressures. Russian Journal of Physical Chemistry, 38, 587–92. McKenzie WF, Helgeson HC (1984) Estimation of the dielectric constant of H2O from experimental solubilities of quartz, and calculation of the thermodynamic properties of aqueous species to 900C at 2 kbar. Geochimica et Cosmochimica Acta, 48, 2167–77. Megaw HD (1933) The thermal expansions of certain crystals with layer lattices. Proceedings of the Royal Society London A, 142, 198–214. Mehnert LD, Dipple GM, Nicolescu S (2005) World skarn deposits. Economic Geology, 100th Anniversary Volume, 299–336. Mesmer RE, Marshall WL, Palmer DA, Simonson JM, Holmes HF (1988) Thermodynamics of aqueous association at high temperatures and pressures. Journal of Solution Chemistry, 17, 699–718. Mesmer RE, Palmer DA, Simonson JM (1991) Ion association at high temperatures and pressures. In: Activity Coefficients in Electrolyte Solutions, 2nd edn (ed. Pitzer KS), pp. 491–529. CRC Press, Boca Raton, FL. Morey GW (1962) The action of water on calcite, magnesite and dolomite. American Mineralogist, 47, 1456–60. Morey GW, Hesselgesser JM (1951) The solubility of some minerals in superheated steam at high pressures. Economic Geology, 46, 821–35. Morey GW, Fournier RO, Rowe JJ (1962) The solubility of quartz in water in the temperature interval from 25 to 300C. Geochimica et Cosmochimica Acta, 26, 1029–43.
Mosebach R (1955) Die hydrothermale Lo¨slichkeit des Quarzes als heterogenes Gasgleichgewicht. Neues Jahrbuch fu¨r Mineralogie Abhandlungen, 87, 351–88. Newton RC, Manning CE (2002) Solubility of silica in equilibrium with enstatite, forsterite, and H2O at deep crust ⁄ upper mantle pressures and temperatures and an activity-coefficient model for polymerization of aqueous silica. Geochimica et Cosmochimica Acta, 66, 4165–76. Newton RC, Manning CE (2003) Activity coefficient and polymerization of aqueous silica at 800C, 12 kbar, from solubility measurements on SiO2-buffering mineral assemblages. Contributions to Mineralogy and Petrology, 146, 135–43. Newton RC, Manning CE (2008a) Thermodynamics of SiO2– H2O fluid near the upper critical end point from quartz solubility measurements at 10 kbar. Earth and Planetary Science Letters, 274, 241–9. Newton RC, Manning CE (2008b) Solubility of corundum in the system Al2O3–SiO2–H2O–NaCl at 800C and 10 kbar. Chemical Geology, 249, 250–61. Nichols GT, Wiebe RA (1998) Desilication veins in the Cadillac Mountain granite (Maine, USA): a record of reversals in the SiO2 solubility of H2O-rich vapor released during subsolidus cooling. Journal of Metamorphic Geology, 16, 795–808. Norton DL, Dutrow BL (2001) Complex behavior of magmahydrothermal processes: role of supercritical fluid. Geochimica et Cosmochimica Acta, 65, 4009–17. Norton DL, Knight JE (1977) Transport phenomena in hydrothermal systems: cooling plutons. American Journal of Science, 277, 937–81. O’Connell JP (1971) Thermodynamic properties of solutions based on correlation functions. Molecular Physics, 20, 27–33. O’Connell JP (1990) Thermodynamic properties of mixtures from fluctuation solution theory. In: Fluctuation Theory of Mixtures (eds Matteoli E, Mansoori GA), pp. 45–67. Taylor & Francis, New York. O’Connell JP, Sharygin AV, Wood RH (1996) Infinite dilution partial molar volumes of aqueous solutes over wide ranges of conditions. Industrial Engineering Chemistry Research, 35, 2808–12. Oelkers EH, Helgeson HC, Shock EL, Sverjensky DA, Johnson JW, Pokrovskii VA (1995) Summary of the apparent standard partial molal Gibbs free energies of formation of aqueous species, minerals, and gases at pressures 1 to 5000 bars and temperatures 25 to 1000C. Journal of Physical and Chemical Reference Data, 24, 1401–560. Parkhurst DL, Appelo CAJ (1999) User’s guide to PHREEQC (version 2) – a computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. Water-Resources Investigations Report 99-4259. Pierotti RA (1963) The solubility of gases in liquids. Journal of Physical Chemistry, 67, 1840–5. Pierotti RA (1976) A scaled particle theory of aqueous and nonaqueous solutions. Chemical Reviews, 76, 717–26. Pina CM, Putnis A, Astilleros JM (2004) The growth mechanisms of solid solutions crystallising from aqueous solutions. Chemical Geology, 204, 145–61. Pitzer KS (1982) Self-ionization of water at high-temperature and the thermodynamic properties of the ions. Journal of Physical Chemistry, 86, 4704–8. Pitzer KS (1983a) Thermodynamics of sodium chloride solutions in steam. Journal of Physical Chemistry, 87, 1120–5. Pitzer KS (1983b) Dielectric constant of water at very high temperature and pressure. Proceedings of the National Academy of Sciences, 80, 4575–6.
38 D. DOLEJSˇ & C. E. MANNING Plyasunov AV (1993) Description of the standard thermodynamic properties of dissolved electrolytes from the total equilibrium constant. Geochemistry International, 30, 52–65. Plyasunov AV, O’Connell JP, Wood RH (2000) Infinite dilution partial molar properties of aqueous solutions of nonelectrolytes. I. Equations for partial molar volumes at infinite dilution and standard thermodynamic functions of hydration of volatile nonelectrolytes over wide ranges of conditions. Geochimica et Cosmochimica Acta, 64, 495–512. Pokrovski GS, Roux J, Harrichoury JC (2005a) Fluid density control on vapor-liquid partitioning of metals in hydrothermal systems. Geology, 33, 657–60. Pokrovski GS, Roux J, Hazemann JL, Testemale D (2005b) An X-ray absorption spectroscopy study of argutite solubility and aqueous Ge(IV) speciation in hydrothermal fluids to 500 C and 400 bar. Chemical Geology, 217, 127–45. Pokrovski GS, Borisova AYu, Harrichoury JC (2008) The effect of sulfur on vapor–liquid fractionation of metals in hydrothermal systems. Earth and Planetary Science Letters, 266, 345–62. Putnis A, Putnis CV (2007) The mechanism of reequilibration of solids in the presence of a fluid phase. Journal of Solid State Chemistry, 180, 1783–6. Quist AS (1970) The ionization constant of water to 800C and 4000 bars. Journal of Physical Chemistry, 74, 3393–402. Quist AS, Marshall WL (1968) Electrical conductances of aqueous sodium chloride solutions from 0 to 800 and at pressures to 4000 bars. Journal of Physical Chemistry, 72, 684–703. Ragnarsdo´ttir KV, Walther JV (1985) Experimental determination of corundum solubilities in pure water between 400– 700C and 1–3 kbar. Geochimica et Cosmochimica Acta, 49, 2109–15. Reed M (1997) Hydrothermal alteration and its relationship to ore fluid composition. In: Geochemistry of Hydrothermal Ore Deposits, 3rd edn (ed. Barnes HL), pp. 303–65. John Wiley & Sons, New York. Rimstidt JD (1997) Quartz solubility at low temperatures. Geochimica et Cosmochimica Acta, 61, 2553–8. Ryzhenko BN, Kovalenko NI, Prisyagina NI (2006) Titanium complexation in hydrothermal systems. Geochemistry International, 44, 879–95. Scambelluri M, Fiebig J, Malaspina N, Mu¨ntener O, Pettke T (2004) Serpentinite subduction; implications for fluid processes and trace-element recycling. International Geology Review, 46, 595–613. Schloemer VH (1952) Hydrothermale untersuchungen u¨ber das system CaO–MgO–CO2–H2O. Neues Jahrbuch fu¨r Mineralogie Monatshefte, 1952, 129–35. Sedlbauer J, Wood RH (2004) Thermodynamic properties of dilute NaCl(aq) solutions near the critical point of water. Journal of Physical Chemistry B, 108, 11838–49. Sedlbauer J, O’Connell JP, Wood RH (2001) A new equation of state for correlation and prediction of standard molal thermodynamic properties of aqueous species at high temperatures and pressures. Chemical Geology, 163, 43–63. Shock EL, Helgeson HC (1988) Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: Correlation algorithms for ionic species and equation of state predictions to 5 kb and 1000C. Geochimica et Cosmochimica Acta, 52, 2009–36. Shock EL, Oelkers EH, Johnson JW, Sverjensky DA, Helgeson HC (1992) Calculation of the thermodynamic properties of aqueous species at high pressures and temperatures: effective electrostatic radii, dissociation constants and standard partial
molal properties to 1000C and 5 kbar. Journal of the Chemical Society, Faraday Transactions, 88, 803–26. Spandler C, Hermann J (2006) High-pressure veins in eclogite from New Caledonia and their significance for fluid migration in subduction zones. Lithos, 89, 135–53. Speziale S, Duffy TS (2002) Single-crystal elastic constants of fluorite (CaF2) to 9.3 GPa. Physics and Chemistry of Minerals, 29, 465–72. Stru¨bel G (1965) Quantitative Untersuchungen u¨ber die hydrothermale Lo¨slichkeit von Flußspat (CaF2). Neues Jahrbuch fu¨r Mineralogie, Monatshefte, 1965, 83–95. Sue K, Adschiri T, Arai K (2002) Predictive model for equilibrium constants of aqueous inorganic species at subcritical and supercritical conditions. Industrial Engineering and Chemistry Research, 41, 3298–306. Sverjensky DA, Shock EL, Helgeson HC (1997) Prediction of the thermodynamic properties of aqueous metal complexes to 1000C and 5 kb. Geochimica et Cosmochimica Acta, 61, 1359– 412. Sweeton FH, Mesmer RE, Baes CF Jr (1974) Acidity measurements at elevated temperatures. VII. Dissociation of water. Journal of Solution Chemistry, 3, 191–214. Tagirov B, Schott J (2001) Aluminum speciation in crustal fluids revisited. Geochimica et Cosmochimica Acta, 65, 3965–92. Tanger JC IV, Helgeson HC (1988) Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: revised equations of state for the standard partial molal properties of ions and electrolytes. American Journal of Science, 288, 19–98. Tanger JC IV, Pitzer KS (1989a) Calculation of the ionization constant of H2O to 2,273 K and 500 MPa. AIChE Journal, 35, 1631–8. Tanger JC IV, Pitzer KS (1989b) Calculation of the thermodynamic properties of aqueous electrolytes to 1000C and 5000 bar from a semicontinuum model for ion hydration. Journal of Physical Chemistry, 93, 4941–51. Thompson AB (1997) Flow and focusing of metamorphic fluids. In: Fluid Flow and Transport in Rocks (eds Jamtveit B, Yardley BWD), pp. 297–314. Chapman & Hall, London. Tremaine PR, Shvedov D, Xiao C (1997) Thermodynamic properties of aqueous morpholine and morpholinium chloride at temperatures from 10 to 300C: apparent molar volumes, heat capacities, and temperature dependence of ionization. Journal of Physical Chemistry B, 101, 409–19. Tremaine P, Zhang K, Benezeth P, Xiao C (2004) Ionization equilibria of acids and basis under hydrothermal conditions. In: Aqueous Systems at Elevated Temperatures and Pressures. Physical Chemistry in Water, Steam and Hydrothermal Solutions (eds Palmer DA, Ferna´ndez-Prini R, Harvey AH), pp. 441–92. Elsevier, Amsterdam. Tropper P, Manning CE (2005) Very low solubility of rutile in H2O at high pressure and temperature, and its implications for Ti mobility in subduction zones. American Mineralogist, 90, 502–5. Tropper P, Manning CE (2007a) The solubility of corundum in H2O at high pressure and temperature and its implications for Al mobility in the deep crust and upper mantle. Chemical Geology, 240, 54–60 & 248, 114. Tropper P, Manning CE (2007b) The solubility of fluorite in H2O and H2O–NaCl at high pressure and temperature. Chemical Geology, 242, 299–306. Vasilev VP, Vorobev PN, Khodakovskii IL (1974) Standard free energies of formation of titanium hydroxocomplexes and of the
Thermodynamic model for mineral solubility in aqueous fluids 39 Ti4+ ion in aqueous solution. Russian Journal of Inorganic Chemistry, 19, 1481–3. Vernon RH, Clarke GL (2008) Principles of Metamorphic Petrology. Cambridge University Press, Cambridge. Wagman DD, Evans WH, Parker VB, Schumm RH, Halow I, Bailey SM, Churney KL, Nuttall RL (1982) The NBS tables of chemical thermodynamic properties. Selected values for inorganic and C1 and C2 organic substances in SI units. Journal of Physical and Chemical Reference Data, 11 (Suppl. 2), 1–392. Wagner W, Pruß A (2002) The IAPWS formulation 1995 for the thermodynamic properties of ordinary water substance for general and scientific use. Journal of Physical and Chemical Reference Data, 31, 387–535. Walther JV (1986) Experimental determination of portlandite and brucite solubilities in supercritical H2O. Geochimica et Cosmochimica Acta, 50, 733–9. Walther JV (1997) Experimental determination and interpretation of the solubility of corundum in H2O between 350 and 600C from 0.5 to 2.2 kbar. Geochimica et Cosmochimica Acta, 61, 4955–64. Walther JV, Helgeson HC (1977) Calculation of the thermodynamic properties of aqueous silica and the solubility of quartz and its polymorphs at high pressures and temperatures. American Journal of Science, 277, 1315–51. Walther JV, Long MI (1986) Experimental determination of calcite solubilities in supercritical H2O. In: Proceedings of the Fifth International Symposium on Water–Rock Interaction. (ed Hitchon RB), pp. 609–11. National Energy Authority, Reykjavik, Iceland. Walther JV, Orville P (1983) The extraction-quench technique for determination of the thermodynamic properties of solute complexes: application to quartz solubility in fluid mixtures. American Mineralogist, 68, 731–41. Wasserman E, Wood B, Brodholt J (1995) The static dielectric constant of water at pressures up to 20 kbar and temperatures to 1273 K: experiment, simulations, and empirical equations. Geochimica et Cosmochimica Acta, 59, 1–6. Weill DF, Fyfe WS (1964) The solubility of quartz in H2O in the range 1000–4000 bars and 400–550C. Geochimica et Cosmochimica Acta, 28, 1243–55. Widmer T, Thompson AB (2001) Local origin of high pressure vein material in eclogite facies rocks of the Zermatt–Saas Zone, Switzerland. American Journal of Science, 301, 627– 56. Wood RH, Quint JR, Groller J-PE (1981) Thermodynamics of a charged hard sphere in a compressible dielectric fluid. A modification of the Born equation to include the compressibility of the solvent. Journal of Physical Chemistry, 85, 3944– 9. Wood RH, Carter RW, Quint JR, Majer V, Thompson PT, Boccio JR (1994) Aqueous electrolytes at high temperatures: comparison of experiment with simulation and continuum models. Journal of Chemical Thermodynamics, 26, 225–49. Wyart J, Sabatier G (1955) Nouvelles mesures de la solubilite´ du quartz, de la tridymite et de la cristobalite dans l’eau sous pression au-dessus de la tempe´rature critique. Comptes Rendus de l’Acade´mie des Sciences, 240, 1905–7. Zack T, John T (2007) An evaluation of reactive fluid flow and trace element mobility in subducting slabs. Chemical Geology, 239, 199–216. Zemniak SE, Jones ME, Combs KE (1993) Solubility behavior of titanium(IV) oxide in alkaline media at elevated temperatures. Journal of Solution Chemistry, 22, 601–23.
Zotov N, Keppler H (2000) In-situ Raman spectra of dissolved silica species in aqueous fluids to 900C and 14 kbar. American Mineralogist, 85, 600–3. Zotov N, Keppler H (2002) Silica speciation in aqueous fluids at high pressures and high temperatures. Chemical Geology, 184, 71–82.
APPENDIX. A Thermodynamic equivalences The new thermodynamic model in reduced form with three independent parameters can be calibrated using the standard thermodynamic properties of minerals and aqueous species at reference temperature and pressure, e.g. 25C and 1 bar (e.g. Johnson et al. 1992; Oelkers et al. 1995). The three-parameter form, Dds G ¼ a þ bT þ eT ln ; o 1 na ln K ¼ þ b þ e ln ; R T
(A1) (A2)
requires that three independent thermodynamic properties must be known – DdsG or DdsH, DdsS, and DdsV or DdscP; note that volume and heat capacity are not independent (cf. Anderson et al. 1991). Using Eq. A1 and rearranging Eqs 16–19, the model parameters are obtained as follows: e¼
Dds V Dds cP ; ¼ 2T w þ T 2 ðow =oT ÞP T
(A3)
b ¼ Dds S þ eðT w ln w Þ; 2
(A4) 2
a ¼ Dds H eT w ¼ Dds G þ T Dds S eT w :
(A5)
The above relationships can be used with the standard thermodynamic properties at any temperature and pressure by employing T and q, b, a and (da ⁄ dT)P of water at the preferred reference conditions (e.g. 25C and 1 bar). The calibration provides means of predicting mineral solubilities (Eq. 14) at any temperature and pressure of interest.
APPENDIX. B Transport theory The transport theory is used to calculate amounts of solids precipitating from ascending aqueous fluids (Baumgartner & Ferry 1991; Ferry & Dipple 1991). In advectionreaction equation the mass conservation during porous isotropic flow is expressed as follows (e.g. Bear 1972, pp. 77–78; Ague 1998): oci oci ¼ v þ Ri ot ox
(B1)
where c is the molar concentration of solute i, t is time, x and v are the distance and flow velocity, respectively, and
40 D. DOLEJSˇ & C. E. MANNING R denotes the reaction rate. At steady state, which is closely approached during mineral-fluid buffering (Ferry & Burt 1982), the solute concentration is invariant with time, ¶ci ⁄ ¶t = 0; hence, Ri ¼
oci v: ox
(B2)
As mineralogical record in rocks reflects the time-integrated result of fluid–rock interaction, integrating Eq. B2 over time, Z t Z t oci v dt; (B3) Ri dt ¼ 0 0 ox
¶mi ⁄ ¶z, is recast into temperature and pressure dependence: omi omi oT omi oP ¼ þ oz oT oz oP oz
(B5)
where ¶T ⁄ ¶z and ¶T ⁄ ¶P represent the geothermal and pressure gradient of interest, whereas the molality changes with temperature and pressure are obtained from the standard thermodynamic properties of dissolution (cf. Eqs 17 and 19) as follows: omi o ln K Dds H ¼K ; ¼K oT oT RT 2
(B6)
omi o ln K Dds V ¼K ¼ K : oP oP RT
(B7)
leads to qV omi ; ni ¼ V oz
(B4)
where n is the number of moles of i precipitated per rock volume, qV is the time-integrated fluid, and V is the molar volume of aqueous fluid. The gradient of molality with vertical distance during one-dimensional flow,
These relationships illustrate that the amount of substance precipitated from 1 kg of aqueous fluid per temperature or pressure increment depends on both the solute concentration (expressed by K = m) and its gradient.
Metal complexation and ion association in hydrothermal fluids: insights from quantum chemistry and molecular dynamics D. M. SHERMAN Department of Earth Sciences, University of Bristol, Bristol, UK
ABSTRACT Complexation by ligands in hydrothermal brines is a fundamental step in the transport of metals in the Earth’s crust and the formation of ore deposits. Thermodynamic models of mineral solubility require an understanding of metal complexation as a function of pressure, temperature and composition. Over the past 40 years, mineral solubilities and complexation equilibria under hydrothermal conditions have been predicted by extrapolating thermodynamic quantities using equations of state based on the Born model of solvation. However, advances in theoretical algorithms and computational facilities mean that we can now explore hydrothermal fluids at the molecular level. Molecular or atomistic models of hydrothermal fluids avoid the approximations of the Born model and are necessary for any reliable prediction of metal complexation. First principles (quantum mechanical) calculations based on density functional theory can be easily used to predict the structures and relative energies of metal complexes in the ideal gas phase. However, calculations of metal complexation in condensed fluids as a function of temperature and pressure require sampling the configuration degrees of freedom using molecular dynamics (MD). Simulations of dilute solutions require very large systems (thousands of atoms) and very long simulation times; such calculations are only practical by treating the interatomic interactions using classical two- or threebody interatomic potentials. Although such calculations provide some fundamental insights into the nature of crustal fluids, simple two- or three-body classical potentials appear to be inadequate for reliably predicting metal complexation, especially in covalent systems such as Sn2+, Au3+ and Cu+. Ab initio MD (i.e. where the bonding is treated quantum mechanically, but the molecular motions are treated classically) avoids the use of interatomic potentials. These calculations are practical for systems with hundreds of atoms over short times (<10 psec) but enable us to predict complexation as a function of pressure, temperature and composition. In this paper, I provide an introductory outline of the computational methods and illustrations of their application to NaCl brines and the complexation of Cu, Au, Sn and Zn. Key words: ab initio, Au, complexation, Cu, hydrothermal, molecular dynamics, NaCl, Sn, Zn Received 16 July 2009; accepted 25 October 2009 Corresponding author: David M. Sherman, Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK. Email:
[email protected]. Tel: +44 (0) 1179545446. Fax: +44 (0) 1179253385 Geofluids (2010) 10, 41–57
INTRODUCTION One of the first challenges of geochemistry was to explain how metals are extracted from the crust and mantle, transported to the near surface and precipitated to form ore deposits. Most sulfide and oxide ore minerals are highly insoluble under reducing conditions even at elevated temperatures. In the 1960s, it was demonstrated that metals can be extracted from primary rocks and transported by acidic hydrothermal brines in which metals form chloride
complexes (e.g. Barnes 1979; Eugster 1986; Seward & Barnes 1997). Low-temperature hydrothermal solutions may also have HS) ligands which complex metals such as As, Hg and Sb to form epithermal deposits. Before we can quantitatively model the formation of hydrothermal ore deposits using reactive transport simulations, we need thermodynamic models of mineral solubilities as a function of pressure, temperature and fluid composition. In practice, these models are developed by fitting high-temperature solubility experiments to a set of complexation equilibria
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
42 D. M. SHERMAN using an equation of state that describes the change in free energy of each species with pressure and temperature. In many systems, a molecular understanding of metal complexation can be obtained from spectroscopy (e.g. Seward 1981; Liu et al. 2001; Muller & Seward 2001). Thermodynamic parameters (e.g. standard entropies, heat capacities and molar volumes) for many minerals and aqueous species are still unavailable; however, Shock & Helgeson (1988) developed a number of empirical correlations to estimate thermodynamic quantities. To extrapolate thermodynamic data to hydrothermal conditions, Helgeson et al. (1981) developed the HKF equation of state based on the dielectric continuum theory of Born (1920). In the Born theory, the solvation free energy of a metal ion with radius R and charge q is q 2 e 2 1 DG ¼ 1 ð1Þ 2R " Here, " is the dielectric constant of the solvent. We can predict how the solvation energy changes with pressure and temperature if we know how the dielectric constant of water changes as a function of pressure and temperature (e.g. Fig. 1). Using estimated thermodynamic quantities and the HKF equation of state, Helgeson and coworkers provided the geochemical community with a ‘database’ (SUPCRT) of estimated equilibrium constants for the formation of minerals and aqueous species as a function of temperature. (Johnson et al. 1992). This database now underlies much work in geochemistry. However, two major concerns must be pointed out (e.g. Oelkers et al. 2009): first, many of the equilibrium constants quantities in SUPCRT are based on estimated thermodynamic quantities from empirical correlations between limited data (e.g. Sverjensky et al. 1997); second, the extrapolation to high temperatures is based on the Born model. The primary
80 4
70
Pressure (kbar)
60
50 30
2
Quantum chemistry of metal complexation The most fundamental approach towards understanding metal complexation in hydrothermal fluids would be to explicitly evaluate the interactions between cations, ligands and the solvent at a molecular or atomistic level. With advances in computational algorithms and computing hardware, we can do this using quantum mechanical calculations. Nearly all applications of quantum chemistry to systems that are geochemically relevant are based on density functional theory. It is worthwhile, here, to give a brief outline of density functional theory to point out some of the limitations we face when trying to understand metal complexation in crustal fluids. Density functional theory The basis of density functional theory (e.g. Parr & Yang 1994) is that the ground state total energy E of a system can be exactly determined in terms of functionals (¼a function of a function) of the electronic charge density [q(r), where r is the electronic coordinate] Z E½ðrÞ ¼ T ½ðrÞ þ U ½ðrÞ þ Vext ðrÞðrÞdr ð2Þ Here T is the kinetic energy functional, U is the electron– electron interaction and Vext contains the electron–nuclear and (for a molecule) nuclear–nuclear potentials. The problem is that we do not know the T and U functionals. However, we know some aspects of them. For example, the classical part of the U functional is the Hartree term describing the classical coulombic repulsion between electrons: Z Z 1 ðrÞðr 0 Þ ð3Þ dr dr0 2 jr r 0 j Part of the kinetic energy functional describes non-interacting electrons. In the Kohn–Sham formalism, we express the wave function of the system in terms of simple oneparticle orbitals wj such that the charge density can be written as:
40
3
assumption of the Born model is that the response of the solvent molecules to the solute charge is linear. However, solvent molecules form structured complexes with solute cations; consequently, the dielectric constant of the solvent near the solvent molecules will be different from that of the bulk solution.
20
ðrÞ ¼
1
10
100
200 300 400 Temperature (°C)
Fig. 1. Dielectric constant of water as a function of T and P.
500
X
j j ðrÞj2
ð4Þ
j
A fundamental requirement of our single-particle wavefunctions is that they be orthonormal so that Z 0; if i 6¼ i 0 ; 0 dr ¼ ¼ ð5Þ 0 i;i i i 1; if i ¼ i 0 :
Complexation and ion association 43 The kinetic energy of the non-interacting quasi-particles is Z 1X 2 T0 ¼ ð6Þ j ðrÞdr j ðrÞr 2 j We can then write Eqn (2) as Z Z 1 ðrÞðr 0 Þ dr dr 0 E½ðrÞ ¼ T0 ½ðrÞ þ 2 jr r 0 j Z þ Vext ðrÞðrÞdr þ Exc ½ðrÞ
ð7Þ
where we have defined a new functional (Exc), called the exchange-correlation functional, that contains all the aspects of the electronic kinetic and potential energy that results from electronic interactions other than that described by Eqns (3) and (6). Equation (7) is valid for any groundstate charge density. This means that we can express any system in terms of single-particle orbitals wj. The singleparticle orbitals are solutions of the Kohn–Sham equations which take the form of the one-electron Schrodinger equation for the hydrogen atom: 1 2 ð8Þ r þ veff ðrÞ j ¼ "j j ðrÞ 2m veff ¼ Vext ðrÞ þ
Z
ðrÞ Exc dr0 þ ðrÞ jr r 0 j
ð9Þ
where "j is the single particle orbital energy. The problem, of course, is that we still do not know the exchange-correlation functional Exc. Advances in computational quantum chemistry and density functional theory are largely concerned with the development of improved exchange-correlation functionals. The work described below is based on the Perdew, Burke and Ernzerhof (PBE) functional (Perdew et al. 1996). The single-particle molecular orbitals in Eqn (8) must be constructed from a basis set {vk} X cki k ð10Þ i ðrÞ ¼ k
where ck are the coefficients used to describe the contribution of each basis function to the molecular orbital. In most quantum chemistry codes, the molecular orbitals and expressed in terms of a basis set of atomic orbitals; these, in turn, are expressed as either Slater functions (as in ADF; te Velde et al. 2001) or Gaussian functions (nearly all other codes). Such basis sets are very convenient to describe the localized electrons in non-metallic systems. For computational reasons, the basis sets used in ab initio molecular dynamics (MD) simulations (discussed below) are plane waves. The advantage of plane waves is that they are already orthonormal so that Eqn (5) becomes: X 0 cki cki ¼ ii 0 ð11Þ k
Plane waves, however, are completely delocalized in space and cannot be used to describe the localized orbitals corresponding to the core electrons. To circumvent this problem, the core electrons are not included in the calculation; instead, the valence electrons are subjected to a pseudopotential which simulates the effect of the core electrons. In the work described here, the Vanderbilt ultrasoft pseudopotentials (Laasonen et al. 1993) were used. In summary, the reliability of any quantum chemical calculation will be ultimately limited by the exchange-correlation functional, the basis set used and the pseudopotentials, if any, describing the core electrons. Application of quantum chemistry to metal complexation Given that we can calculate the total internal energy (E) of a complex as a function of its internal nuclear coordinates xj, we can start to understand the nature of metal complexes in hydrothermal fluids. First, we can calculate the optimized structure of metal complexes and also their vibrational spectra. From the total internal energies and vibrational spectra we can calculate the enthalpy and entropy of each complex from the vibrational and ideal gas translational partition functions. In quantum chemistry codes such as ADF (te Velde et al. 2001), this is done automatically. The thermodynamic information we obtain would be applicable to the complexes in the ideal gas phase (i.e. where no inter-molecular interactions are present). Such thermodynamics might be a good approximation for metal complexation in low-density supercritical fluids. In denser fluids, where the ideal gas approximation is unrealistic, an atomistic evaluation of thermodynamic quantities is much more complicated. First, we need to include the solvation effects on the energetics of the metal complexes. Short-range solvation effects can be treated by explicitly including the local solvent water molecules in the complex. Furthermore, the long-range solvation effects can be approximated using one of several dielectric continuum models (e.g. the COSMO model, Klamt & Schu´rmann 1993; Klamt 1995; Klamt & Jones 1996) which can be included self-consistently in the quantum mechanical calculations. Although this means we are returning to the Born model, we are doing so for only the long-range solvation where it will not be as drastic an approximation. However, an important effect of intermolecular interactions is to greatly complicate the evaluation of the translational entropy. To evaluate the translational entropy in a real fluid, we need to determine the energies of the systems in all of the microstates determined by the range of atomic positions and velocities. In a real gas or fluid, where intermolecular interactions cannot be neglected, this is a formidable task. MD and Monte Carlo methods, however, allow us to sample the configurations of the system that are most significant.
44 D. M. SHERMAN
Molecular dynamics In a MD simulation, we sample the configurations of the system dynamically (i.e. by letting all of the atoms move and exchange kinetic and potential energy with each other). The hope is that the time average sampling of the configurations is equivalent to the ensemble averaging of the configurations (this is the Ergodic hypothesis). In a Monte Carlo simulation, we do the ensemble configuration sampling directly. In practice, we can only hope to approach ergodicity for large simulations run over long times. For understanding metal speciation in hydrothermal fluids, this might require simulations with thousands of atoms run for several nanoseconds. In computationally demanding ab initio MD simulations (discussed below), this is impractical and techniques such as metadynamics (Liao & Gervasio 2008) are needed to bias the system to explore the free energy surfaces of interest. The sampling of the positions and momenta of the atoms in a MD simulations is done by solving the equations of motion for the system. Typically, we neglect the quantization of the vibrational and rotational energies of the molecules and use Newton’s equation rather than the Schrodinger equation for the nuclear motions. For heavy atoms, this is not too drastic an approximation. However, for hydrogen, we must be careful as quantum effects on nuclear motion are appreciable. The general formalism of MD is best described in terms of Lagrangian mechanics. This approach enables us to see how various constraint techniques (e.g. thermostats and metadynamics) are implemented. It is also the basis for discussing the Car–Parrinello method of ab initio MD. The Lagrangian (L) of a system is _ U ðx; xÞ _ L ¼ T ðx; xÞ
ð12Þ
where T is the kinetic energy and U is the potential energy, x is the positional coordinates of the particles and x_ is their time derivatives (velocities). The equations of motion in Lagrange mechanics are: d oL oL ¼ dt ox_ ox
ð13Þ
We start with an initial set of positions xj(0) and velocities vj (0) at time t ¼ 0. We evaluate the forces on each particle to give the particle accelerations aj(0). From our initial conditions, we can integrate the equation of motion using the Verlet algorithm with a suitably small time step Dt tnþ1 ¼ tn þ Dt 1 vj ðtnþ1 Þ ¼ vðtn Þ þ ðaj ðtn Þ þ aj ðtnþ1 ÞÞDt 2 1 xj ðtnþ1 Þ ¼ xj ðtn Þ þ vj ðtn ÞDt þ aj ðtn ÞDt 2 2 where n is the index over time steps and j is the index over atoms. In the spirit of the Ergodic hypothesis, simulations are run for an adequate time to sample configuration space (i.e. the most important particle positions and velocities). The time needed for a simulation will vary depending on the lifetime of the complexes of interest, their probability of formation and the number of atoms in the simulation. As we wish to model a bulk fluid using a finite number of atoms, we usually impose periodic boundary conditions by placing the atoms in a periodic box of length L and imposing the condition that x + nL ¼ x. The periodic boundary conditions remove effects resulting from the fluid–vacuum interface that would be present in a finite cluster. Even if we neglect the quantization of the nuclear motion, we still prefer to calculate the interatomic potential and the forces acting on each atom quantum mechanically. This still makes any reasonable system a serious computational challenge. One approach to make calculations practical is the Car & Parrinello (1985) method discussed below. This allows us only to calculate systems containing several hundreds of atoms for simulation times in the order of 10 psec. Car–Parrinello method In the Car–Parrinello method, we incorporate the electronic wavefunctions into the MD scheme by defining the extended Lagrangian:
For a simple system where T ¼
N X 1 j ¼1
2
mj ðx_ j Þ2
ð14Þ
L¼
N X 1 j
þ U ¼ U ðx1; x2 ; . . . ; xN Þ
ð15Þ
the equation of motion reduces to Newton’s second law: mj
d2 x j dt 2
¼ Fj ¼ rj U ðx1; x2; x3 ; . . . ; xN Þ ¼ mj aj
ð16Þ
where Fj is the force and aj is the acceleration of particle j.
ð17Þ
2
mj x_ j
X i;i 0
i;i 0
2
2 X 1 dc i þ k E½ðrÞ; x1; x2 ; . . . ; xN dt 2 i;k ! X i i0 ð18Þ ck ck 0 i;i 0 k;k 0
That is, the coefficients for the wave functions are treated just like the other dynamical coordinates. The first term is simply the classical kinetic energy of the nuclei; the second term is a fictitious kinetic energy of the wavefunctions with fictitious mass l; the third term is the Kohn–Sham total
Complexation and ion association 45 energy that is a function of the electronic charge density and the nuclear positions; fourth term is the constraint that the one-electron Kohn–Sham orbitals are orthonormal with ki,i¢ being the Lagrange multipliers. As the fictitious mass of the electrons goes to zero, we approach Born– Oppenheimer MD. This is much slower as the wavefunctions must be explicitly re-evaluated at each time; however, it allows a longer time step to be used. Classical molecular dynamics simulations based on interatomic potentials Ab initio MD simulations are very computationally demanding and are limited to simulations with approximately hundreds of atoms over approximately tens of picoseconds. Problems such as equations of state, phase separations, polynuclear clusters and time-dependent phenomena that require simulations of approximately thousands of atoms over approximately hundreds of picoseconds or nanoseconds cannot be dealt with. Large simulations, however, are feasible if we approximate the potential energy and forces on each atom in terms of classical two- and three-body potentials that are derived either empirically or by fitting to quantum mechanically calculated potentials in finite clusters. In a simple pairwise interaction simulation (the most common approach), we assume that the potential between two atoms is independent of the positions of all the other atoms in the system. In reality, the most successful interatomic pair potentials are usually derived to implicitly take some aspects of the many-body effects into account. The potential energy between two atomic sites is usually viewed as consisting of a long-range coulombic part and a short-range repulsive/attractive part. A variety of analytic forms can be used for the short-range potential. A common approach is to augment a simple Coulombic term (long-range interaction) with a Lennard–Jones-type function (short-range interaction) so that the total potential between two atoms i and j is " # ij 12 ij 6 qi qj Uij ¼ 4"ij þ rij rij rij
ð19Þ
where qi and qj are the charges of ion i and j; "ij and rij are the potential parameters between ions i and j and rij is the distance between ion i and j. For monatomic ions, it is usually the easiest to use the formal charges in the coulombic term. For molecules, effective charges must be used. Information from molecular dynamics simulations To get anything useful from an MD simulation, we usually need to take a time average of the atomic positions and momenta. From the time average of the atomic positions,
we can understand metal complexation and the molecular structure of hydrothermal fluids. A very useful quantity is the pair distribution function: gij ðrÞ ¼
hNij ðr Dr; r þ DrÞi N V ðr Dr; r þ DrÞ
ð20Þ
where ÆNij(r)Dr,r+Dr)æ is the average (over time) of the number of particles of type j at a distance from particle i between r)Dr and r+Dr; qN is the number of particles per unit volume and V(r)Dr,r+Dr) is the volume of the shell between r)Dr and r+Dr. The pair distribution function gives us a measure of the probability of finding an atom at a given distance relative to that in a homogeneous liquid or an ideal gas of point particles. If we integrate the ij pair distribution function, we get the coordination number of i by j: Z 4Nj CN ¼ gij ðrÞr 2 dr ð21Þ V where Nj/V is the number of atoms of type j in volume V. As we run the MD simulation, the atoms will acquire a distribution of kinetic energies. Temperature is a measure of the average kinetic energy of the particles in a system NkBT/2. However, in a finite simulation, temperature becomes illdefined. To correctly define a temperature in a finite MD simulation, we cannot simply scale the velocities as we would yield an ensemble with any statistical mechanical significance. Instead we use a Nose–Hoover thermostat which keeps the system in a canonical (NVT ¼ constant mass, volume and temperature) ensemble. The Nose Hoover thermostat is implemented by defining a fictitious coordinate (s) associated with a heat bath to give the extended Lagrangian: L¼
N X 1 j ¼1
2
mj s 2 ðx_ j Þ2 U ðx1; x2 ; . . . ; xN Þ
Q 2 ð22Þ ð_s Þ gkB T lnðsÞ 2 where g is a parameter that scales the heat bath and Q is the effective mass associated with our fictitious coordinate (s) and kB is Boltzmann’s constant. Analogous thermostats are available for pressure and one can perform NPT simulations as well. þ
APPLICATIONS Nature of water at ambient and supercritical conditions A molecular-level understanding of the structure of water has been developed by the pioneering work of, e.g. Soper et al. (1997) and Soper (2000) using neutron diffraction. The structure of water can be understood in terms of the O–O and OH pair distribution functions (Fig. 2) which show the tetrahedral framework resulting from hydrogen
46 D. M. SHERMAN
3
2.9 Å
25°C Expt. (Soper, 2000) 2
2.9 Å
g(r)O–O
SPC/E
4.5 Å 1
0
0
1
2
3
4 5 6 O–O distance
7
8
9
10
3
Fig. 3. Interpretation of O–O pair distribution to yield the tetrahedral structure of water caused by hydrogen bonding.
400°C, 0.34 kbar
g(r)O–O
2
1
0
0
1
2
3
6.9 Å
4 5 6 O–O distance
7
8
9
10
Fig. 2. O–O pair distributions derived from neutron diffraction (Soper 2000) compared with simulations based on the SPC/E model (e.g. Sherman & Collings 2002).
bonding (Fig. 3). Changes in the O–O pair distribution function with temperature shows that the hydrogen bonding framework is substantially broken down under supercritical conditions. These data provide a useful benchmark for theoretical simulations. Simulations of water using classical potentials A number of rigid models for water have been developed. These models differ primarily in how the charge is spatially partitioned among the atoms. In the rigid models, the O–H bond length is fixed as is the H–O–H angle. Examples include the simple point charge (SPC, Berendsen et al. 1987), TIPS2 (Jorgensen 1982), TIP4P (Jorgensen et al. 1983), MCY (Matsuoka et al. 1976) and extended simple point charge (SPC/E, Berendsen et al. 1987) models. The SPC/E water model of Berendsen et al. (1987) gives a very simple parameterization for the effective charges and short-range potential for water. In the SPC/E model we represent water molecules by three point charges in a rigid geometry with an HOH bond angle slightly different
(109.5) from that in the gas phase H2O molecule (104.5). Although the model does not allow any polarization or dissociation of the water molecules, it accurately reproduces the structural, thermodynamic and dielectric properties of water, at least along the liquid–vapor coexistence curve (Guissani & Guillot 1993). In particular, the SPC/E model predicts the critical point of water to be at 640–652 K with a density of 0.29–0.326 g cm)3. For real water, the critical point is at 647 K with a density of 0.322 g cm)3 (Haar et al., 1984). The SPC/E model also gives a good prediction of the dielectric constant of water and its temperature dependence: the predicted values of e(SPC/E) ¼ 81.0 at 300 K and e(SPC/E) ¼ 6 at T(c) versus 78.0 and 5.3, respectively, in real water. In the light of the Born model of solvation (Eqn 1) an accurate prediction of the dielectric constant of water is crucial to predicting metal solvation and complex speciation in aqueous solutions. Ab initio molecular dynamics simulations of water Ab initio simulations of water using Born Oppenheimer and Car–Parrinello algorithms have been done on systems typically containing approximately 64 H2O atoms for runs lasting up to 60 psec (e.g. Lee & Tuckerman 2006 and references therein). Most of the simulations tend to ‘overstructure’ the water and yield gO)O(r) that is between 3.1 and 3.8 compared with the experimental value near 2.93 (Soper 2000). There are several possible reasons for the discrepancy between experiment and ab initio MD simulations: first, the exchange-correlation functional may be inadequate to describe hydrogen bonding; the simulations may simply have been performed on systems that are too small for times that are too short (non-ergodicity); the basis set may be inadequate. In the latter regard, Lee & Tuckerman (2006) have obtained good agreement with experiment from ab initio MD simulations on a small (32 waters) system for 30 psec but using a nearly complete
Complexation and ion association 47
The compositions of fluid inclusions (e.g. Roedder 1967; Eugster 1986) indicate that extremely concentrated NaCl brines can be involved in the transport of metals and the formation of hydrothermal ore deposits. There is a variety of evidence that Na and Cl ions become highly associated into NaCl ion pairs and polynuclear clusters in high-temperature aqueous solutions at pressures below 2 kbar. Under such conditions, water and NaCl appear to behave like ideal mixtures (Quist & Marshall 1968; Franz 1982) so that the activity of water, aH2O equals its mole fraction (XH2O). Above 4 kbar, however, there is evidence that NaCl ion pairs become dissociated in NaCl–water mixtures. For example, Quist & Marshall (1968) found an increase in the electrical conductance of NaCl–H2O mixtures between 2 and 4 kbar at 800C. Subsequently, Aranovich & Newton (1996) measured the activity of water in concentrated NaCl solutions at 600–900C and 2–15 kbar using the brucite–periclase equilibrium. At high pressures, the activity of water is well approximated by aH2 O ¼
XH2 O 2 XH2 O
ð23Þ
Asargued by Aranovich & Newton (1996), this is the athermal activity expected if the NaCl–water system behaved as a hydrous salt melt consisting of water and completely dissociated NaCl. The much smaller activity of water under these conditions means that a free solution phase could coexist with an H2O-saturated granitic melt in the deep crust. Such a chloride-rich solution phase must play a fundamental role in scavenging metals prior to the formation of hydrothermal ore deposits. Understanding ion hydration, ion association and phase separation in the NaCl–water system is an obvious problem for MD. Ion pairing and phase separation in NaCl–H2O brines cannot be treated using ab initio MD because the systems required to define the problem are too large. Smith & Dang (1994) derived ion–water interaction parameters for Na and Cl by fitting potential functions to the gas phase ion hydration energies. The parameters were used in MD simulations of an NaCl ion pair in 216 water molecules. This is a fairly small simulation but it gave very accurate predictions of experimental hydration energies and complex geometries.
14
8 M NaCl at 25°C, 1 bar
12
Cl--Na RDF
10
O–Na RDF
10
Na–O CN
8
8
Na–Cl CN
6
6
4
4
2
2
0
0 8 M NaCl at 325°C, 1 kbar
12 10
10
8
8
6
6
4
4
2
2
0
0
Coordination number
Concentrated NaCl brines
Smith & Dang (1994) predict that at 25C, Na+ cations will be surrounded by 5.8 water molecules with a Na–O bond length of 2.33 A˚. The Cl) anions will be solvated by 6.9 waters with a Cl–H distance of 2.22 A˚. These are in good agreement with those observed experimentally (Powell et al. 1993). Subsequently, a number of simulations on the NaCl–H2O system have been done, especially with the goal of understanding the effects of temperature and density on ion hydration and clustering (e.g. Chialvo et al. 1996, 1997; Lyubartsev & Laaksonen 1996; Driesner et al. 1998; DeGreve and da Silva 2000; Koneshan and Rasaiah 2000; Sharygin et al. 2002). Sherman & Collings (2002) used the Smith & Dang (1994) potentials, together with the SPC/E model, to investigate ion hydration and clustering in highly concentrated (0.5–8.0 molal) NaCl brines at 0–15 kbar and 25–650C. Figure 4 shows the Na–O and Na–Cl pair distribution
gNa–O(r) or gNa–Cl(r)
basis set and the Becke, Lee, Yang, Parr (BLYP) exchangecorrelation functional. Todorova et al. (2006) explored the effect of different exchange-correlation functionals on water structure. Most of the commonly used functionals perform similarly and slightly overstructure water. They are all a major improvement over the local density approximation and Hartree–Fock exchange only. Hence, the structure of water must be strongly controlled by the electronic correlation energy.
8 M NaCl at 625°C, 15 kbar
12 10
10
8
8
6
6
4
4
2
2 0
0 0
1
2
3 4 5 Distance (Å)
6
7
8
Fig. 4. Na–Cl and Na–O pair distribution functions showing the presence ˚ of both NaðH2 OÞþ 6 and NaCl ion pairs. The Na–Cl peak at 5 A corresponds to an outer sphere ion pair (from Sherman & Collings 2002).
48 D. M. SHERMAN (2002) found that, at supersaturation (8 molal NaCl) the local water structure is broken down to force water molecules to occupy the interstitial sites in the hydrogen bonding framework. The change in water structure must explain the saturation concentration of NaCl at these conditions. (Presumably, this simulation would eventually crystallize out NaCl; however, it would take simulation times so long as to be computationally infeasible). At 325C and 1 kbar, however, the rotational motion of the water molecules breaks down the long-range structure of pure water; only ˚ shows up in the pair distribution the O–O peak at 2.8 A function of pure water. At the same time, the structure of water in concentrated brines is the same as in pure water. The O–O pair distribution functions do not change with NaCl concentration under P–T conditions corresponding to those of the lower crust (Fig. 5). Calculated densities of Na–Cl brines are in good agreement with experiment, although there is an inexplicable discontinuity in density at high concentrations (Fig. 6). 3 25°C, 1 bar Pure water
2 gO–O(r)
8 M NaCl
1
0
0
1
2
3 4 5 Distance (Å)
6
7
8
325°C, 1 kbar
gO–O(r)
2
1
0
0
1
2
3 4 5 Distance (Å)
6
7
8
7
8
625°C, 15 kbar
2 gO–O(r)
functions at several temperatures and pressures. With increasing temperature there is less solvation of Na+ and Cl) ions by water molecules and more NaCl ion pairs and polynuclear clusters form. Not shown is that, with increasing Cl concentration, more ion-pairing occurs so that, in the very concentrated solutions found in fluid inclusions, most Na and Cl ions exist in polynuclear clusters. The coordination numbers and bond lengths for dilute NaCl solutions at 25C, 1 bar are in good agreement with those from previous calculations (Smith & Dang 1994) and experiment (Powell et al. 1993). For dilute NaCl, Sherman & Collings (2002) predicted a coordination number of 5.3 with only waters in the solvation sphere for Na+ at 25C, 1 bar. With increasing NaCl molality, the Na–Cl association increases: at 25C and 1 bar the Na+ ions in an 8 molal solution of NaCl have 4.3 waters and 1.0 Cl ligands in their first coordination shell. There is some evidence of solvent-separated Na–Cl ion pairs from the peak ˚ . However, in the Na–Cl pair distribution function at 5 A this disappears with increasing temperature. For a given NaCl concentration, the degree of Na–Cl association increases at the expense of Na–O association as we approach the critical point (Fig. 4). For example, the Na+ ions in an 8 molal NaCl solution have, on average, 3.0 H2O and 1.8 Cl) in their first neighbor shell at 325C, 1 kbar. What is especially significant is that there is still considerable association of Na and Cl in concentrated brines under deep-crustal conditions (e.g. at 625C and 15 kbar). In the 8 molal solution, we have 3.0 waters and 2.0 Cl) ions in the first coordination shell. A completely random coordination environment would have 0.6 Cl) and 4.4 waters for this composition if the Na+ coordination number is 5. The ion association persists even at very high concentrations: in the 16 molal NaCl solution at 625C, 15 kbar, each Na ion is coordinated to, on average, 2.6 Cl) ions and 2.7 water molecules. A completely random coordination environment would have 1.2 Cl) and 4.1 waters if the total coordination number of Na is 5.3. As discussed by Sherman & Collings (2002), the Na and Cl ions form complex polynuclear clusters that are not readily apparent from the pair distribution functions. Simple Na–Cl ion pairs are not very important; most of the Na+ and Cl) start to form NamCln (m, n > 2) clusters with increasing salinity and temperature. These clusters persist at high pressures. However, relatively dilute NaCl solutions are still mostly dissociated even at extreme conditions (625C, 15 kbar). As discussed by Manning (2004) fluids coming off a subducted slab near these conditions may be fairly dilute. Aside from the Na–Cl association, MD simulations give insight into the effect of electrolytes on the structure of water. Zhu & Robinson (1992) found, in a simulation of a 1.79 molal NaCl solution, that the water structure resulting from hydrogen bonding begins to be broken down by the solvation of the Na+ and Cl) ions. Sherman & Collings
1
0 0
1
2
3 4 5 Distance (Å)
6
Fig. 5. O–O pair distribution in 8 molal NaCl solution as a function of temperature and pressure (from Sherman & Collings 2002).
Complexation and ion association 49
1600
1.3
P = 1 kbar
1-phase
1400
1.2
2-phase
Fluid
1200 Temperature (°C)
Density (g cm–3)
1.1
1
0.9 25°C, 1 bar
1000 Liquid + vapor 800
600
400
0.8 325°C, 1 kbar
200 0.7
0 0
1
2
3
4 5 6 NaCl molality
7
8
0
2
4
6 8 NaCl molality
10
12
9
Fig. 6. Density of NaCl–water mixtures (Sherman & Collings 2002) compared with those given by the Anderko and Pitzer (1993) equation of state.
The simulations also seem to give a good prediction of the NaCl–H2O phase diagram (Brodholt 1998; Collings 2000). At 1 kbar, when the temperature is increased to 730C, a strong Na–Na peak occurs in the pair distribution function and the Na and Cl ions cluster together. Collings (2000) interpreted this as a boiling and a separation of the NaCl–H2O mixture into a vapor and a highly concentrated NaCl–H2O liquid. Based on the presence of the strong Na–Na peak in the pair distribution function, Collings (2000) used a series of simulation runs that allows us to approximately map out the NaCl–H2O phase diagram (Fig. 7); the results are in good agreement with the model of Bowers and Helgeson (1983) and the equation of state of Archer (1992). It is remarkable how well the Smith & Dang (1994) potentials, together with the SPC/E model for water are remarkably successful at predicting densities of NaCl–H2O mixtures, NaCl speciation and the NaCl–H2O phase diagram. It is not obvious that the SPC/E model and the interatomic potentials derived by Smith & Dang (1994) should be reliable for supercritical fluids. What is not clear is whether such simple pair potentials could describe complexation in more covalent systems such as transition metal chlorides and aquo complexes.
Fig. 7. Calculated NaCl–H2O phase diagram (Collings 2000). Dashed lines indicate two-phase regions from Bowers and Helgeson (1983).
shown in Fig. 8 for a 0.1 M Zn solution. The range of complexes makes the Zn–Cl system of particular interest and Tossell (1991) used gas phase ab initio cluster calculations to predict the structures, vibrational spectra and relative stabilities of different complexes. Simulations based on classical interatomic potentials Yongyai et al. (1992) developed Zn–H2O and Zn–Cl potentials and modeled the complexation of Zn2+ in 300
250
Temperature (°C)
0.6
Liquid + solid
625°C 15 kbar
ZnCl4–
200 ZnCl3– 150
ZnCl+
100
ZnCl2
50 Zn+ + 0
Zinc chloride solutions Stability constants for Zn–Cl complexes in the SUPCRT database (Johnson et al. 1992) predict the complexation
–1
0 –0.5 log Cl– activity
0.5
Fig. 8. Speciation of Zn in NaCl brine as calculated from the SUPCRT (Johnson et al. 1992) model.
50 D. M. SHERMAN
Zn2þ þ 4Cl ¼ ZnCl2 4 is really ZnðH2 OÞ2þ 6 þ 4ClðH2 OÞ6 ¼ ZnCl4 ðH2 OÞ20 þ 10H2 O with the translational entropy provided by the liberated solvation waters. Copper(II) complexation The copper(II)–NaCl–H2O system is fundamental to supergene ore-forming fluids. A number of Cu–Cl complexes
20
9 8
15
7 6
10 5
2 1
0
0 1 kbar, 325° C
ZnCl2 (H2O)2,
8 7 6
ZnCl3 (H2O)–
150 gZnCl(r)
5 4 3
Zn (H2O)62+
5 4 3 2
100
50
1 0
0 15 kbar, 625° C
140 120
9 8 7
ZnCl42–
100
6 5
80
4 3
60 40
2 1
20 0
Zn–O coordination number
gZn-O(r)
1 bar, 25° C
Zn–Cl coordination number
Simulations based on ab initio molecular dynamics Harris et al. (2003) performed Born–Oppenheimer ab initio MD simulations on Zn–Cl complexation. They found that the ZnCl+ and ZnCl2 complexes exist as pseudo-octahedral ZnClm(H2O)6)m clusters at 25C but occur as pseudo-tetrahedral ZnClm(H2O)4)m clusters at 300C. The ZnCl 3 complex occurs as the pseudo-tetrahedral ZnCl3(H2O)) cluster at 25 and 300C. The tetrahedral ZnCl2 4 complex, however, is the dominant Zn–Cl complex at 25C, at least in highly concentrated (7.4 molal) Cl) brines. It is encouraging that structures predicted by the pair potential simulations are consistent with the ab initio results. However, much more work in this system can be done. The Zn–Cl–H2O systems seems especially rich and could be a very useful testing ground for theory and experiment. One immediate result comes out of the simulations: the change in coordination of Zn with temperature is driven by the increase in translational entropy. However, to see this, we need to know the actual solvation structure. The reaction
25
gZn-Cl(r)
1–5 molal ZnCl2 solutions. Their results indicate that ZnCl2 complexes form but not higher order complexes such as ZnCl2 4 . Interestingly, their simulations predict the trans-ZnCl4(H2O)4 complex. Marini et al. (1996) have developed a potential set for the Zn–H2O system that uses three-body terms. This predicts the sixfold coordination of Zn by H2O. Harris et al. (2001) derived a new Zn–Cl potential from ab initio calculations on simple gas phase molecules. This potential, however, gave poor agreement with experiment. This suggests that many-body effects are important and must be indirectly incorporated into any pair potential scheme. A new potential set was derived by fitting to the observed Zn–Cl bond lengths in a solution of SPC/E water molecules. Simulations with this potential set predict an increase in Cl) complexation with temperature and Cl concentration (Fig. 9). Below 300C, Zn occurs as sixfold coordinated ZnðH2 OÞ2þ (3 molal Cl) and ZnClðH2 OÞþ 6 5 (6 molal Cl). Above 300C, Zn adopts fourfold coordination to give ZnCl2(H2O)2, ZnCl3(H2O) and ZnCl 4 depending upon the Cl concentration.
0 0
1
2 3 4 Distance (Å)
5
6
Fig. 9. Complexation of Zn in 3m NaCl brines as a function of temperature based on the interatomic potentials of Harris et al. (2001).
appear to form and, using extended X-ray absorption fine structure (EXAFS) spectroscopy, Collings et al. (2000) found a change in Cu2+ complexation from CuðH2 OÞ2þ n to CuCl2 with increasing [Cl)] and temperature 4 (25–125C). Existing stability constants found in the SUPCRT model (Johnson et al. 1992) predict the Cu2+ speciation shown in Fig. 10. Classical molecular dynamics simulations Simulating the complexation of Cu2+ using classical interatomic potentials is complicated by the d 9 configuration of Cu2+ which induces a Jahn–Teller distortion of the Cu
Complexation and ion association 51
14
CuCl42–
200
CuCl2 150
CuCl+
100
10 8
8
6
6
4
4
2
2
0
0
1
2 3 4 Cu–O distance (Å)
5
Coordination number
12
CuCl3– gCu–O(r)
Temperature (°C)
250
6
50 Fig. 11. Cu–O pair distribution function in 3 molal Cl solution (1 CuCl2 and 36 waters) showing Cu in fivefold coordination with water; no Cu–Cl complexation is found.
Cu2+ –0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
log aCl Fig. 10. Speciation of Cu2+ as predicted from the SUPCRT (Johnson et al. 1992) model.
coordination environment. This is a breakdown of the Born–Oppenheimer approximation and means that the ground electronic state is a function of the coordination geometry. Early simulations based on classical interatomic potentials (Rode & Islam 1992) predicted that at 25C in 1–5 molal CuCl2 solutions, the dominant species will be CuCl2 4 . Based on the thermodynamic data (Fig. 10) and EXAFS data (Collings et al. 2000), the Rode and Islam potential appears to overestimate the Cu–Cl coordination. To deal with the complication of the Jahn–Teller effect using classical potentials, Texler and Rode (1997) developed a three-body potential to describe the Cu–H2O– H2O and Cu–Cl–H2O interactions. This was used in simulations (Texler et al., 1998) which predict that [Cu(H2O)5Cl]+ will be the dominant species in 0.5 molal CuCl2 and 5 molal NaCl. Ab initio molecular dynamics of Cu(II) complexation Ab initio MD simulations of Cu2+ complexation by water have been done by Pasquarello et al. (2001) and more recently by Amira et al. (2005). These simulations show that Cu2+ actually adopts a fivefold coordination with water. Thus far, no ab initio simulations have been published that explore Cu(II) complexation with Cl. A short (2 psec) Car–Parrinello simulation on a system with 1 Cu, 2 Cl) and 36 waters fails to show any Cu–Cl complexation. It does show that the fivefold coordination of Cu is a dynamical average between square planer CuðH2 OÞ2þ 4 and the Jahn–Teller distorted CuðH2 OÞ4 ðH2 OÞ2þ complex 2 (Fig. 11). The dramatic discrepancy between the classical potentials and the ab initio simulations suggests that even three body potentials may not be adequate.
Copper(I) complexation Under the reducing conditions of ore-forming hydrothermal solutions, copper exists in the +1 oxidation state. However, the Cu+ ion is unstable in aqueous solutions unless complexing ligands such as Cl) or HS) are present. Fritz (1980) suggested that Cu+ forms CuCl, CuCl2 and CuCl2 3 complexes in Cl brines at 25C. Until recently, our understanding of copper(I) speciation at high temperatures has been derived from fitting solubility measurements, as a function of [Cl)], to a set of complexation equilibria. From the solubilities of chalcocite and chalcopyrite, Crerar & Barnes (1976) concluded that the CuCl0 complex is dominant at temperatures from 200 to 350C. Var’yash (1992) found CuCl 2 to be the dominant species in 1 molal NaCl solutions at 300–350C. Xiao et al. (1998) estimated stability constants for Cu+–Cl) complexes from 40 to 300C from measurements of the solubility of Cu phases in vapor-saturated HCl/NaCl solutions (up to 1 molal Cl)) 2 and find CuCl 2 and CuCl3 to be the dominant species. Liu et al. (2001) measured the solubility of cuprite (Cu2O) from 50 to 250C at vapor saturation and fit the data to a set of stability constants for the CuCl, CuCl 2 and CuCl2 complexes. They argue that the CuCl2 complex 3 3 is important at 250C but was not resolvable in the fits to solubilities at higher temperatures. Spectroscopic measurements are a useful adjunct to solubility measurements as they allow us to directly determine the aqueous speciation of metal complexes in aqueous solutions. Fulton et al. (2000) have measured EXAFS spectra of 0.2–0.4 molal Cu+ solutions with 0.43–2.37 molal Cl from 100 to 325C. No evidence for CuCl2 complexes was 3 when chloride was found; the dominant species was CuCl 2 in excess and CuCl when [Cu] ¼ [Cl] ¼ 0.4 molal. Liu et al. (2002a) measured UV optical spectra of Cu+ in concentrated Cl) brines up to 250C. They argue that not only 3 does the CuCl2 3 complex form but also a CuCl4 complex
52 D. M. SHERMAN
10 25°C, 4 M Cl– 8 6 4 2 0 325°C, 4 M Cl– Cu–Cl distance (Å)
becomes important in concentrated (up to 9 molal) Cl) solutions. The existence of the CuCl3 complex at high 4 temperatures would imply that the solubilities of CuFeS2 and Cu2S in concentrated brines are even higher than would be predicted from extrapolating solubilities based on 2 the CuCl 2 and CuCl3 complexes. Sherman (2007) approached the complexation of Cu+ using Car–Parrinello-based ab initio MD simulations on a periodic system consisting of 1 Cu, 55 water molecules, 4 Cl and 3 Na atoms. Those simulations used the PBE exchange-correlation functional and Vanderbilt ultrasoft pseudopotentials. As the simulations can only be carried out over short times (0.1 psec required about 330 CPU hours with the computational resources available at the time) it was possible that the simulations would fail to form the most stable clusters starting from a random molecular configuration. In order to test the stability of the CuCl 2, 3 CuCl2 3 and CuCl4 complexes, simulation runs were set up with biased initial starting points. Hence, initial atomic positions started out with CuCl2 3 complexes surrounded by 55 (or 32) water molecules and the remaining ions. These were then allowed the system to equilibrate for 1 psec. These simulations were also used to determine the inner- versus outer-sphere hydration of the CuCln complexes. Figure 12 shows the Cu–Cl bond lengths as a function of time for simulations at 25C (18.13 cm3 mol)1 and approximately 1 bar pressure) and 325C (21.96 cm3 mol)1 and approximately 1 kbar pressure). At 25C, the CuCl2 3 complex is found to be stable over 5 psec. The Cu–Cl pair distribution function (Fig. 13) has a maximum at a Cu–Cl bond length of approximately 2.21 A˚. However, the shape of the Cu–Cl peak shows an asymmetry with a second peak (corresponding to the third Cl)) near 2.3–2.4 A˚. When the solution is heated to 125C along the liquid–vapor curve, the CuCl2 3 complex dissociates within 0.5 psec into CuCl2 ) and Cl . At this temperature, the CuCl2 complex is actually 2 CuCl2(H2O)) and CuCl 2 . At 325C the CuCl3 complex is unstable; within 0.5 psec of the simulation, the starting ) CuCl2 3 complex dissociates into CuCl2 and Cl and fails to reform after 4 psec. The oscillations of the Cu–Cl bond lengths in the CuCl 2 complex shown in Fig. 12 correspond to the Cu–Cl vibrations with frequency near 9.5 · 1012 Hz or 318 cm)1. However, this really is the superposition of two modes (symmetric and asymmetric stretch) which are calculated to occur at 280 and 373 cm)1 in a gas phase CuCl 2 cluster. The Cu–Cl pair distribution functions (Fig. 13) show that the average Cu–Cl bond length is 2.15 A˚ in the CuCl 2 complex. The results are in close agreement with the EXAFS results of Fulton et al. (2000): they found Cu–Cl bond lengths of 2.12–2.13 A˚ with coordination numbers of 1.8–2.0 (±0.2) from 100 to 325C. To test the possible formation of a CuCl3 4 complex, as proposed by Liu et al. (2002a,b), a simulation was done for Cu in a 13.9 molal Cl solution (1 Cu, 7 Na, 8 Cl and
8 6 4 2 0 450°C, 13 M Cl– 8 6 4 2 0
0
1
2
3
4
5
Time (ps) Fig. 12. Cu–Cl distances as a function of time showing relaxation to CuCl2 3 at 25C and CuCl2 at elevated temperatures (from Sherman, 2007).
32 H2O molecules) at 450C and a volume (23.99 cm3 mol)1) corresponding to approximately 1 kbar pressure (Anderko & Pitzer 1993). Only the CuCl 2 complex forms in this system over a 4 psec simulation time. The results are consistent with static cluster calculations in the COSMO solvation field that show the CuCl3 4 complex to ) be unstable to dissociation into CuCl 2 and Cl . Subsequent X-ray absorption spectra of high-temperature solutions (Brugger et al. 2007) confirmed the absence of 2 stable CuCl3 4 complex, the decreasing stability of CuCl3 with temperature and the dominant CuCl2 complex in excess Cl. Liu et al. (2008) further explored the complexation of Cu in solutions where [Cu] [Cl] and showed that the dominant complexes were CuCl(H2O) and CuCl 2. Tin(II) complexation Tin is found primarily as the ore mineral cassiterite (SnO)2 which is deposited by hot, acidic hydrothermal fluids
Complexation and ion association 53
4
15
0.7
25°C (4.0 M Cl–)
3
SnCl3–
0.6 Fraction of Sn
10
0.8
2 5 1
1.0 M NaCl at Psat
0.5 0.4 SnCl2
0.3 0.2
0
gCu–Cl(r)
325°C
3
(4.0 M Cl–)
10
2 5 1 0
Coordination number
0
0
425°C (13.9 M Cl–)
4
3 2
2 1 0
0
1
2 Distance (Å)
3
4
0
Fig. 13. Cu–Cl pair distribution functions (solid lines) and coordination numbers (dashed lines) from simulations in Fig. 12 (from Sherman, 2007).
associated with granitic intrusions (e.g. Heinrich 1990). The solubility of cassiterite is believed to result from the formation of Sn–Cl complexes: SnO2 þ 2Hþ þ nCl ¼ ðSnCln Þ2n þ H2 O þ 12 O2 Fluid inclusion homogenization temperatures and oxygen isotope studies indicate that cassiterite forms at temperatures ranging from 200–400C (e.g., Chesley et al. 1993; Linnen and Williams-Jones 1994). In the thermodynamic model provided by Jackson & Helgeson (1985), the dominant Sn2+ complexes under acidic conditions are SnCl+, SnCl2 and SnCl 3 depending on Cl concentration and temperature (Fig. 14). Several attempts have been made to directly observe Sn speciation using spectroscopy. Sherman et al. (2000) measured the EXAFS spectra of 0.1 molal Sn2+ solutions in NaCl–HCl brines from 25 to 350C along the liquid vapor
0.1
Sn++
0.0 0
50
SnCl+
100 150 200 Temperature (°C)
250
300
Fig. 14. Speciation of Sn–Cl based on the thermodynamic model of Jackson and Helgeson (1985).
curve. Fits to the EXAFS spectra suggested the formation of a SnCl2 complex at high temperatures. Subsequently, 4 Muller & Seward (2001) investigated Sn–Cl solutions using visible to near-UV optical absorption spectra to 300C. They found no evidence for SnCl2 4 at T > 150C and suggested that the solutions in Sherman et al. (2000) had oxidized at high temperatures. Atomistic simulations on the Sn–Cl–H2O system would be useful. However, classical two-body potentials would be unable to model Sn2+ complexation as Sn2+ ion has a localized lone pair that affects its stereochemistry. Consequently, there are no two- or three-body potentials to simulate Sn2+ coordination. The Sn–Cl system needs to be treated quantum mechanically and provides a good problem for the application of Car–Parrinello MD. Similar to the previous example on Cu–Cl complexation, simulation runs were performed on systems containing 55 H2O, 1 Sn, 4 Cl and 2 Na atoms corresponding to a 1 molal SnCl2 solution in 2 molal NaCl. The simulation volumes were chosen to correspond to the densities of a 4 molal NaCl solution along the liquid–vapor curve at 25 and 125C (based on the equation of state of Pitzer et al. 1984) and at 325C, 1 kbar (based on volumes from Anderko & Pitzer 1993). Sn–Cl radial distribution functions for these simulations at 25 and 325C are shown in Fig. 15. At 25C, the average Sn–Cl coordination number is 2.4 corresponding to a mixture of 60% SnCl2 and 40% SnCl 3 . At 325C, the coordination number is 2.98 corresponding to approximately 100% SnCl 3 . As expected, the geometry of the SnCl complex is not trigonal planar but trigonal pyrami3 dal. This geometry results from the presence of a localized lone pair of electrons on the Sn atom. To test the stability of the SnCl2 4 complex, simulations were run with a biased starting configuration having an SnCl2 cluster. However, at 25C, the SnCl2 cluster 4 4 quickly dissociates into Sn(H2O)Cl2 and Cl). At 325C, 2 the stable complex is SnCl 3 . The apparent SnCl4 complex
54 D. M. SHERMAN
4
4
3
3
2
2
1
1
0
0
50 40
4
30
3
20
2
10
1
0
0
Coordination number
325°C
SUMMARY AND CONCLUSIONS Classical MD simulations using simple two- and threebody potentials are easy to do and take on very complex
30
3
20
2
10
1
0
0
1
2 3 Sn–Cl distance (Å)
4
5
140
Gold(III) complexation Under oxidizing conditions, Au dissolves as Au(III) which forms a series of hydroxy and chloride complexes. Recently, Usher et al. (2009) have measured the speciation of Au(III) in halide solutions using spectrophotometry. We can test some aspects of the model using ab intio MD
6 5
100
Fig. 15. Sn–Cl pair distribution showing change from SnCl2 to SnCl 3 with increasing temperature.
80
4
60
3
40
2
20
1
0
0
1
2 3 4 Au–Cl distance (Å)
5
6
0
6 5 gAu–O(r)
in the EXAFS study of Sherman et al. (2000) appears to be an artifact of the EXAFS fitting. Coordination numbers derived from EXAFS can have a ±20% uncertainty because they strongly correlated with the Debye–Waller factors. Moreover, the Sn EXAFS spectra had a strong atomic selfabsorption peak which complicated the determination of the Sn coordination number.
25°C
120
Coordination number
4
4
4
3
3
2
2
1
1
0
0
1
2 3 4 Au–O distance (Å)
5
6
Coordination number
40
Coordination number
25°C
0
simulations on AuCl3–NaCl solutions. Here, a simulation was done with 1 Au, 55 H2O, 1 Na and 4 Cl at 25 and 325C. At both temperatures, only the AuCl3(OH)) complex forms (Fig. 16) and this seems to be stable over the short (several picoseconds) run times. The predicted AuCl3(OH)) complex is in agreement with what should be the dominant Au complex in a 4 molal Cl solution at pH 7 based on the thermodynamic model of Usher et al. (2009). However, the Au–OH–Cl speciation is a strong function of pH and it is impossible to simulate deviations from neutral pH that are less than 5 pH units in a system this small (i.e. we could simulate a pH of 0, 7 or 14 by adding 1, 0 or )1 protons to the simulation box respectively; we would still need to perform the run long enough to give the dominant species time to form). This is a limitation of ab initio MD simulations. Another issue with proton-transfer equilibria is that we are using classical dynamics to model the motion of the H atoms; however, they are small enough so that quantization of the nuclear motion is significant.
gAu–Cl(r)
60
gSn–Cl(r)
Coordination number
625°C
0
Fig. 16. Au–O and Au–Cl pair distributions at 25C showing the only species over the time of the simulation is AuCl3(OH)). The pair distributions at 325C are identical.
Complexation and ion association 55 systems. The very simple SPC/E model for water is remarkably successful and simulations with two-body potentials appear to successfully describe alkaline metal and alkaline earth cations such as NaCl. However, using this approach for transition metals and metalloids appears to be much less reliable. For example, complexes of Cu2+ are subject to the Jahn–Teller effect which is difficult to model using classical two- and three-body potentials. Using ab initio MD, we need not worry about interatomic potentials (just the atomic pseudopotentials). However, ab initio simulations are limited to small systems. Phenomena such as phase separation and polynuclear complexes seen in the NaCl simulations cannot be treated at this level. It is also very difficult to run ab initio simulations long enough (and on large enough systems) to reliably sample the speciation that a metal cation might have in an NaCl brine. However, the application of metadynamics (e.g. Liao & Gervasio 2008) to MD simulations will enable us to predict relative free energies of complexes in hydrothermal solutions. Metadynamics allows us to constrain coordinates to explore free energy surfaces and we can incorporate coordination numbers into the extended Lagrangian. Recent applications of metadynamics to systems of geochemical interest include the hydration of Y3+ and La3+ by Ikeda et al. (2005). It is anticipated that the next several years will see exciting new insights into the molecular nature not only of shallow crustal ore-forming fluids but also of deep crustal fluids associated with slab dehydration in the upper mantle.
ACKNOWLEDGEMENTS Many of the calculations described here were performed using the Laboratory of Advanced Computation, Department of Mathematics, University of Bristol. I am very grateful for past collaborations and discussions with Terry Seward, Duncan Harris, Matt Collings and John Brodholt.
REFERENCES Amira S, Spangberg D, Hermansson K (2005) Distorted five-fold coordination of Cu2+ (aq) from a Car–Parrinello molecular dynamics simulation. Physical Chemistry Chemical Physics, 7, 2874–80. Anderko A, Pitzer KS (1993) Equation-of-state representation of phase equilibria and volumetric properties of the system NaCl– H2O above 573 K. Geochimica et Cosmochimica Acta, 57, 1657–80. Aranovich LYa, Newton RC (1996) H2O activity in concentrated NaCl solutions at high pressures and temperatures measured by the brucite–periclase equilibrium. Contributions to Mineralogy and Petrology, 125, 200–12. Archer DG (1992) Thermodynamic properties of the NaCl+H2O system. 2. Thermodynamic properties of NaCl(aq), NaClÆ2H2O(cr), and phase equilibria. Journal of Physical and Chemical Reference Data, 21, 793–829.
Barnes HL (1979) Solubilities of ore minerals. In: Geochemistry of Hydrothermal Ore Deposits, 2nd edn (ed. Barnes HL), pp. 404– 60. John Wiley, New York. Berendsen HJC, Grigera JR, Straatsma TP (1987) The missing term in effective pair potentials. Journal of Physical Chemistry, 91, 6269–71. Born M (1920) The hydration heat of ions. Zeitschrift fur Physik, 1, 45–8. Bowers TS, Helgeson HC (1983) Calculation of the thermodynamic and geochemical consequences of non-ideal mixing in the system H2O-CO2-NaCl on phase relations in geologic systems: equation of state for H2O-CO2-NaCl fluids at high temperatures and pressures. Geochimica et Cosmochimica Acta, 47, 1247–75. Brodholt JP (1998) Molecular dynamics simulations of aqueous NaCl solutions at high pressures and temperatures. Chemical Geology, 151, 11–9. Brugger J, Etschmann B, Liu W, Testemale D, Hazemann JL, Emerich H, van Beek WH, Proux O (2007) An XAS study of the structure and thermodynamics of Cu(I) chloride complexes in brines up to high temperature (400C, 600 bar). Geochimica et Cosmochimica Acta, 71, 4920–41. Car R, Parrinello M (1985) Unified approach for molecular dynamics and density functional theory. Physical Review Letters, 55, 2471. Chesley JT, Halliday AN, Snee LW, Mezger KS, Shepherd TJ, Scrivener RC (1993) Thermochronology of the Cornubian batholith in southwest England: implications for pluton emplacement and protracted hydrothermal mineralization. Geochimica et Cosmochimica Acta, 57, 1817–35. Chialvo AA, Cummings PT, Simonson JE, Mesmer RE (1996) Temperature and density effects on the high temperature ionic speciation in dilute Na+/Cl) aqueous solutions. Journal of Chemical Physics, 105, 9248–57. Chialvo AA, Cummings PT, Simonson JE, Mesmer RE (1997) Molecular simulation study of speciation in supercritical aqueous NaCl solutions. Journal of Molecular Liquids, 73–4, 361–72. Collings MD (2000) Complexation of Ba and Cu in hydrothermal NaCl brines: insights from EXAFS spectroscopy and molecular dynamics. PhD Dissertation, 103 pp. University of Bristol, Bristol UK. Collings MD, Sherman DM, Ragnarsdottir KV (2000) Complexation of Cu2+ in oxidized NaCl brines from 25C to 175C: results from in situ EXAFS spectroscopy. Chemical Geology, 167, 65–73. Crerar DA, Barnes HL (1976) Ore solution chemistry. V. Solubilities of chalcopyrite and chalcocite assemblages in hydrothermal solution at 200 to 350C. Economic Geology and the Bulletin of the Society of Economic Geologists, 71, 772–94. DeGreve L, da Silva FLB (2000) Detailed microscopic study of 1 M aqueous NaCl solution by computer simulations. Journal of Molecular Liquids, 87, 217–32. Driesner T, Seward TM, Tironi IG (1998) Molecular dynamics simulation study of ionic hydration and ion association in dilute and 1 molal aqueous sodium chloride solutions from ambient to supercritical conditions. Geochimica et Cosmochimica Acta, 62, 3095–107. Eugster HP (1986) Minerals in hot water. American Mineralogist, 71, 655–73. Franz G (1982) The brucite–periclase equilibrium at reduced H2O activities: some information about the system H2O–NaCl. American Journal of Science, 282, 1325–39. Fritz JJ (1980) Chloride complexes of CuCl in aqueous solution. Journal of Physical Chemistry, 84, 2241–6.
56 D. M. SHERMAN Fulton JL, Hoffman MM, Darab JG (2000) An X-ray absorption fine structure study of copper(I) chloride coordination structure in water up to 325C. Chemical Physics Letters, 330, 300–8. Guissani Y, Guillot B (1993) A computer simulation study of the liquid–vapour coexistence curve of water. Journal of Chemical Physics, 98, 8221–35. Haar L, Gallagher JS, Kell GS (1984) NBS/NRC Steam Tables. Hemisphere, Bristol, PA. Harris DJ, Brodholt JP, Harding JH, Sherman DM (2001) Molecular dynamics simulation of aqueous ZnCl2 solutions. Molecular Physics, 99, 825–33. Harris DJ, Brodholt JP, Sherman DM (2003) Zinc complexation in hydrothermal chloride brines: results from ab initio molecular dynamics calculations. Journal of Physical Chemistry A, 107, 1050–4. Heinrich CA (1990) The chemistry of hydrothermal tin–tungsten ore deposition. Economic Geology and the Bulletin of the Society of Economic Geologists, 90, 705–29. Helgeson HC, Kirkham DH, Flowers GC (1981) Theoretical prediction of the thermodynamics behavior of aqueous electrolytes at high pressures and temperatures IV. Calculation of activity coefficients, osmotic coefficients, and apparent molal and standard and relative partial molal properties to 600C and 5 kbar. American Journal of Science, 281, 1249–516. Ikeda T, Hirata M, Kimura T (2005) Hydration structure of Y3+ and La3+ compared: an application of metadynamics. Journal of Chemical Physics, 122, 244507–12. Jackson KJ, Helgeson HC (1985) Chemical and thermodynamic constraints on the hydrothermal transport and deposition of tin. 1. Calculation of the solubility of cassiterite at high-pressures and temperatures. Geochimica et Cosmochimica Acta, 49, 1–22. Johnson JW, Oelkers EH, Helgeson HC (1992) SUPCRT92: a software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000 degrees C. Computers and Geosciences, 18, 899–947. Jorgensen WL (1982) Revised TIPS model for simulations of liquid water and aqueous solutions. Journal of Chemical Physics, 77, 4156–63. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML (1983) Comparison of simple functions for simulating water. Journal of Chemical Physics, 79, 926–35. Klamt A (1995) Conductor-like screening model for real solvents: a new approach to the quantitative calculation of solvation phenomena. Journal of Physical Chemistry, 99, 2224–35. Klamt A, Jones V (1996) Treatment of the outlying charge in continuum solvation models. Journal of Chemical Physics, 105, 9972–81. Klamt A, Schu´rmann G (1993) COSMO: a new approach to dielectric screening in solvents with explicit expressions for the screening energy and its gradient. Journal of the Chemical Society, Perkin Transactions 2, 799–805. Koneshan S, Rasaiah JC (2000) Computer simulation studies of aqueous sodium chloride solutions at 298 K and 683 K. Journal of Chemical Physics, 113, 8125–37. Laasonen K, Pasquarello A, Lee C, Car R, Vanderbilt D (1993) Car–Parrinello molecular dynamics with Vanderbilt’s ultrasoft pseudopotentials. Physical Review B, 47, 10142–53. Lee HS, Tuckerman ME (2006) Structure of water at ambient temperature from ab initio molecular dynamics performed in the complete basis set limit. Journal of Chemical Physics, 125, 154507. Liao A, Gervasio FL (2008) Metadynamics: a method to simulate rare events and reconstruct the free energy in biophysics, chem-
istry and material science. Reports on Progress in Physics, 71, 126601. Linnen RL, Williams-Jones AE (1994) The evolution of pegmatite hosted Sn–W mineralization at Nong Sua, Thailand: evidence from fluid inclusions and stable isotopes. Geochimica et Cosmochimica Acta, 58, 735–47. Liu W, McPhail DC, Brugger J (2001) An experimental study of copper(I)-chloride and copper(I)-acetate complexing in hydrothermal solutions between 50 degrees C and 250 degrees C and vapor-saturated pressure. Geochimica et Cosmochimica Acta, 65, 2937–48. Liu W, Brugger J, McPhail DC, Spiccia L (2002a) A spectrophotometric study of aqueous copper(I)-chloride complexes in LiCl solutions between 100 degrees C and 250 degrees C. Geochimica et Cosmochimica Acta, 66, 3615–33. Liu W, Sakane S, Wood RH, Doren DJ (2002b) The hydration free energy of aqueous Na+ and Cl) at high temperatures predicted by ab initio/classical free energy perturbation: 973 K with 0.535 g/cm3 and 573 K with 0.725 g/cm3. J. Phys. Chem. A, 106, 1409–18. Liu W, Joe¨l Brugger J, Etschmann B, Testemale D, Hazemann JL (2008) The solubility of nantokite (CuCl(s)) and Cu speciation in low-density fluids near the critical isochore: An in-situ XAS study. Geochimica et Cosmochimica Acta, 72, 4094–106. Lyubartsev AP, Laaksonen A (1996) Concentration effects in aqueous NaCl solutions. A molecular dynamics simulation. Journal of Physical Chemistry, 100, 16410–8. Manning CE (2004) The chemistry of subduction zone fluid. Earth and Planetary Science Letters, 223, 1–16. Marini GW, Texler NR, Rode BM (1996) Monte Carlo simulations of Zn(II) in water including three-body effects. Journal of Physical Chemistry-US, 100, 6808–13. Matsuoka O, Clementi E, Yoshimine M (1976) CI study of the water dimer potential surface. Journal of Chemical Physics, 64, 1351–61. Muller B, Seward TM (2001) Spectrophotometric determination of the stability of tin(II) chloride complexes in aqueous solution up to 300 degrees C. Geochimica et Cosmochimica Acta, 65, 4187–99. Oelkers EH, Benezeth P, Pokrovski GS (2009) Thermodynamic databases for water–rock interaction. In: Thermodynamics and Kinetics of Water–Rock Interaction (eds Oelkers EH, Schott J), Reviews in Mineralogy and Geochemistry, 70, 1–46. Parr RG, Yang W (1994) Density-Functional Theory of Atoms and Molecules. Oxford University Press, New York. Pasquarello A, Petri I, Salmon PS, Parisel O, Car R, Toth E, Powell DH, Fischer HE, Helm L, Merbach AE (2001) First solvation shell of the Cu(II) aqua ion: evidence for fivefold coordination. Science, 291, 856–9. Perdew JP, Burke K, Ernzerhof M (1996) Generalized gradient approximation made simple. Physical Review Letters, 77, 3865–8. Pitzer KS, Peiper JC, Busey RH (1984) Thermodynamic properties of aqueous sodium-chloride solutions. Journal of Physical and Chemical Reference Data, 13, 1–102. Powell DH, Neilson GW, Enderby GW (1993) The structure of Cl) in aqueous solution – an experimental determination of G(ClH)(R) and G(ClO)(R). Journal of Physics: Condensed Matter, 5, 5723–30. Quist AS, Marshall WL (1968) Electrical conductances of aqueous sodium chloride solutions from 0 to 800C at pressures up to 4000 bars. Journal of Physical Chemistry, 72, 684–703. Rode BM, Islam SM (1992) Structure of aqueous copper chloride solutions – results from Monte-Carlo simulations at various concentrations. Journal of the Chemical Society, Faraday Transactions, 88, 417–22.
Complexation and ion association 57 Roedder E (1967) Fluid inclusions as samples of ore fluids. In: Geochemistry of Hydrothermal Ore Deposits (ed. Barnes HL), pp. 515–74. Holt, Rinehart, and Winston, Inc., New York. Seward TS (1981) Metal complex formation in aqueous solutions at elevated temperatures and pressures. In: Chemistry and Geochemistry of Solutions at High Temperatures and Pressures (eds Rickard D, Wickman F), Physics and Chemistry of the Earth, 13/14, 113–32. Pergamon Press, New York. Seward TM, Barnes HL (1997) Metal transport by hydrothermal ore fluids. In: Geochemistry of Hydrothermal Ore deposits, 3rd edn (ed. Barnes HL), pp. 435–86. John-Wiley and Sons, Hoboken, NJ. Sharygin AV, Wood RH, Zimmerman GH, Balashov VN (2002) Multiple ion association versus redissociation in aqueous NaCl and KCl at high temperatures. Journal of Physical Chemistry B, 106, 7121–34. Sherman DM (2007) Complexation of Cu+ in hydrothermal NaCl brines: ab initio molecular dynamics and energetics. Geochimica et Cosmochimica Acta, 71, 714–22. Sherman DM, Collings MD (2002) Ion association in concentrated NaCl brines under ambient to supercritical conditions: results from classical molecular dynamics simulations. Geochemical Transactions, 3, 102–7. Sherman DM, Ragnarsdottir KV, Oelkers EH, Collins CR (2000) Speciation of tin (Sn2+ and (Sn4+) in aqueous Cl solutions from 25 degrees C to 350 degrees C: an in situ EXAFS study. Chemical Geology, 167, 169–76. Shock EL, Helgeson HC (1988) Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures; correlation algorithms for ionic species and equation of state predictions to 5 kb and 1000 C. Geochimica et Cosmochimica Acta, 54, 915–45. Smith DE, Dang LX (1994) Computer simulations of NaCl association in polarizable water. Journal of Chemical Physics, 100, 3757–66. Soper AK (2000) The radial distribution functions of water and ice from 220 to 673 K and at pressures up to 400 MPa. Chemical Physics, 258, 121–37. Soper AK, Bruni F, Ricci MA (1997) Site–site pair correlation functions of water from 25 to 400C: revised analysis of new
and old diffraction data. Journal of Chemical Physics, 106, 247– 54. Sverjensky DA, Shock EL, Helgeson HC (1997) Prediction of thermodynamic properties of aqueous metal complexes to 1000C and 5 kb. Geochimica et Cosmochimica Acta, 61, 1359– 412. Texler NR, Rode BM (1997) Monte Carlo simulations of copper chloride solutions at various concentrations including full 3-body terms. Chemical Physics, 222, 281–8. Texler NR, Holdway S, Neilson GW, Rode BM (1998) Monte Carlo simulations and neutron diffraction studies of the peptide forming system 0.5 mol kg()1) CuCl2-5 mol kg()1) NaCl-H2O at 293 and 353 K. Journal of the Chemical Society Faraday Transactions, 94, 59–65. Todorova T, Seitsonen AP, Hutter J, Kuo IFW, Mundy CJ (2006) Molecular dynamics simulation of liquid water: hybrid density functionals. Journal of Physical Chemistry B, 110, 3685– 91. Tossell JA (1991) Calculations of the structures, stabilities, and Raman and Zn NMR-spectra of ZnCln(OH2)a2)n species in aqueous-solution. Journal of Physical Chemistry, 95, 366–71. Usher A, McPhail DC, Brugger J (2009) A spectrophotometric study of Au(III) halide–hydroxide complexes at 25–80C. Geochimica Cosmochimica Acta, 77, 3359–80. Var’yash LN (1992) Cu(I) complexing in NaCl solutions at 300 and 350C. Geochemistry International, 29, 84–92. te Velde G, Bickelhaupt FM, Baerends EJ, Fonseca Guerra C, van Gisbergen SJA, Snijders JG, Ziegler T (2001) Chemistry with ADF. Journal of Computational Chemistry, 22, 931–67. Xiao Z, Gammons CH, Williams-Jones AE (1998) Experimental study of copper(I) complexing in hydrothermal solutions at 40 to 300C and saturated water vapour pressure. Geochimica et Cosmochimica Acta, 62, 2949–64. Yongyai Y, Kokpol S, Rode BM (1992) Microstructure and species distribution of aqueous zinc-chloride solutions – results from Monte-Carlo simulations. Journal of the Chemical Society, Faraday Transactions, 88, 1537–40. Zhu S, Robinson GW (1992) Molecular-dynamics computer simulation of an aqueous NaCl solution: structure. Journal of Chemical Physics, 97, 4336–48.
Role of saline fluids in deep-crustal and upper-mantle metasomatism: insights from experimental studies R. C. NEWTON AND C. E. MANNING Department of Earth and Space Sciences, University of California, Los Angeles, CA, USA
ABSTRACT Chloride-rich brines are increasingly recognized as playing an important role in high pressure and temperature metamorphic and magmatic systems. The origins of these saline multicomponent fluids are debated, but experimental evidence suggests that regardless of their origin they must be important agents of rock alteration and mass transfer wherever they occur. Studies of the solubility of quartz in H2O, CO2–H2O and salt–H2O solutions provide a framework for understanding the role of brines in the deep crust and upper mantle. While quartz solubility in the system SiO2–H2O–NaCl–CO2 is maximal at a given high pressure and temperature if the solvent is pure H2O, the decline in quartz solubility with NaCl content (salting-out) is less severe than in CO2–H2O fluids at comparable H2O activities. Moreover, at lower pressures, quartz solubility initially salts in at low salt contents, before reaching a maximum and then declining. The behavior of quartz solubility in salt–H2O solutions has not yet been fully explained and is the subject of active debate. Experimental investigations of the solubility of some other rock-forming oxides and silicates show enhancements due to NaCl addition. As illustrated by the well-studied CaO–Al2O3–SiO2–NaCl–H2O system, enhancements initially increase to maxima, and then decline. This behavior can be explained by formation of a range of hydrated aqueous complexes and clusters with specific NaCl:H2O stoichiometries. In contrast, solubilities of calcium salts, including calcite, fluorite, fluorapatite and anhydrite, rise monotonically with increasing NaCl, implying complexing to form anhydrous ionic solutes and ⁄ or ion pairs. The experimental studies offer new insights into fluid-rock interaction in a range of settings, including carbonatite–fenite complexes, granulite-facies metamorphism, porphyry ore deposits and aluminum-silicate vein complexes in high-grade metamorphic terranes. Key words: experimental petrology, metamorphic fluids, mineral solubility, salt effects Received 21 August 2009; accepted 30 November 2009 Corresponding author: C. E. Manning, Department of Earth and Space Sciences, University of California, Los Angeles, CA 90095-1567, USA. Email:
[email protected]. Tel: +1 310 206 3290. Fax: +1 310 825 2779. Geofluids (2010) 10, 58–72
INTRODUCTION The generation and migration of chloride-rich brines is an important geologic process in the lower crust and upper mantle (e.g. Touret 1985; Newton et al. 1998; Yardley & Graham 2002). Fluid inclusions containing chloride brines with >10 wt% NaCl equivalent have been reported from granulites (Touret 1985; Crawford & Hollister 1986; Smit & Van Reenan 1997; Touret & Huizenga 1999; Van den Berg & Huizenga 2001), eclogites (e.g. Philippot & Selverstone 1991; Fu et al. 2001), and mantle diamonds (e.g. Izraeli et al. 2001, 2004; Klein-BenDavid et al. 2007). In addition, crystalline salts have been observed in high-grade
metamorphic rocks (Markl & Bucher 1998). Calcium-rich silicates record equilibration of minerals with saline brines in a range of metamorphic settings (e.g. Munoz 1981; Mora & Valley 1989; Nijland 1993; Kullerud 1996; Markl et al. 1998). Finally, magnetotelluric observations suggest the presence of conductive, saline fluids in deep-crustal fault zones (e.g. Wannamaker et al. 2004). Saline fluids can be produced or concentrated in deep fluid-flow systems by a variety of mechanisms (Newton et al. 1998; Yardley & Graham 2002), including inheritance of an originally saline connate fluid, dissolution of salts of sedimentary origin, H2O loss by preferential partitioning into hydrous minerals during retrograde
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Role of saline fluids in deep-crustal and upper-mantle metasomatism 59 metamorphism or into hydrous silicate liquids during melting and infiltration of externally derived magmatic fluids. Regardless of origin, a fundamentally important factor in the evolution of brines in the deep crust and upper mantle is their immiscibility with CO2-rich aqueous fluid over a wide range of pressure (P) and temperature (T) (Bowers & Helgeson 1983; Johnson 1991; Duan et al. 1995; Heinrich 2007). Fluid inclusion evidence suggests that, in many instances, saline fluids of the lower crust and upper mantle may coexist across a miscibility gap with CO2-rich fluids (e.g. Touret 1985; Gibert et al. 1998). The CO2-rich fluids are more commonly observed as inclusions in minerals because they are trapped more efficiently due to their relatively low ability to wet grain boundaries (Watson & Brenan 1987). Also, postentrapment modification of fluid inclusions due to diffusion (Sterner & Bodnar 1989) or deformation (Barker 2007) is more likely for salty inclusions than for CO2-rich inclusions because of higher mineral solubility and reactivity. However, in spite of their volumetrically minor representation as fluid inclusions, it is likely that the saline components play a more important role than CO2 in metasomatic mass transfer in the deep crust and upper mantle. The potential role of saline fluids in mass transfer has only recently gained experimental attention, but a growing body of data offers important and surprising insights into interactions among salty fluids and minerals at moderate to high P and high T. Here, we review experimental results on mineral solubility in saline fluids at conditions relevant to the lower crust and upper mantle. The data point to the conclusion that alkali halide brine components of deepcrustal and upper-mantle fluids represent powerful solvents capable of significant mass transfer – not just of metals, but of nearly all major rock forming oxides. This behavior helps explain petrologic observations from a wide range of deep geologic environments.
SILICA IN HIGH-GRADE FLUIDS Silica is among the most abundant solutes in H2O-bearing fluids of the deep crust and upper mantle, even when diluted by significant CO2 or salts, or when fluids coexist with bulk compositions possessing low silica activity (Anderson & Burnham 1965; Manning 1994; Newton & Manning 2000, 2002). Thus, investigation of the interactions between fluids and minerals in the lower crust and upper mantle most conveniently begins by consideration of SiO2. Quartz solubility in H2O The dissolution of quartz into H2O at deep-crustal and upper-mantle conditions is strongly dependent on P and T, and on the polymerization of solute silica (Manning
1994; Zotov & Keppler 2000, 2002; Newton & Manning 2003, 2008a). Equilibrium between quartz, H2O and neutral monomeric silica, can be described by the reaction SiO2 þ nH2 O ¼ Si(OH)4 ðn 2ÞH2 O quartz
silica monomer
ð1Þ
Equation 1 includes two forms of interaction between solvent H2O and solute SiO2: for each mole of SiO2, two moles of H2O form hydroxyls, and n ) 2 moles of molecular H2O form a hydration shell (e.g. Weill & Fyfe 1964; Sommerfeld 1967). Polymerization of solute silica is an additional interaction. Taking the H2O of hydration as implicit, homogeneous equilibrium in an H2O–SiO2 fluid can be written 2Si(OH)4 ¼ Si2 O(OH)6 þ H2 O monomer
dimer
ð2Þ
(Newton & Manning 2002a, 2003; Zotov & Keppler 2002). Newton & Manning (2002a, 2003) showed that a simple mixing model which assumes only monomers and dimers adequately describes phase relations in SiO2–H2O fluids to ‡20 kbar and approximately 900C. By equating activities (ai) of the monomer (m) and dimer (d) to their mole fractions (Xi), the equilibrium constant for Equation 2 (Kmd) is Kmd ¼
X d a H2 O Xm2
ð3Þ
When H2O is the only solvent component, aH2 O 1 over a wide range of crustal and mantle conditions, and Kmd values can be described by log Kmd ¼ 1:480 þ 0:0012T þ ð0:000119T 0:1685ÞP ð4Þ where T is in Kelvins and P is in kbar (Newton & Manning 2002a, 2009). This description of quartz solubility fails as temperatures rise above 900C and pressures approach 10 kbar. Along the quartz hydrothermal melting curve, there are increases with rising P in SiO2 solubility in H2O, and in H2O solubility in coexisting melt. These factors lead to critical mixing of SiO2-rich aqueous fluid and hydrous silica melts at a critical end point on the hydrothermal melting curve of quartz near 10 kbar and 1080C (Kennedy et al. 1962; Newton & Manning 2008a). Under these P–T conditions, SiO2-rich fluids and liquids must be highly polymerized, with constitutions similar to pegmatitic rock melts; it is of some interest to inquire what effect saline and carbonic components have on critical mixing of fluids in the system SiO2–H2O at high P and T. Figure 1 shows logKmd isobars at deep-crustal and upper-mantle temperatures. With rising T, logKmd
60 R. C. NEWTON & C. E. MANNING
The solubility of quartz in CO2–H2O fluids at high P and T declines steadily with increasing CO2 content. This is illustrated in Fig. 2A, using data from Newton & Manning (2009) at 800C, 10 kbar. The data indicate that quartz is effectively insoluble in pure CO2, consistent with the absence of interaction or complexing between CO2 and SiO2 in the fluid phase. Thus, CO2-rich solutions are very poor solvents for silica. Data on mixed solvents in which one component does not interact with dissolved silica can be utilized to gain insight into the interactions between the silica and solvent H2O. A priori, it is uncertain whether the number of H2O molecules that hydroxylate and hydrate dissolved SiO2 are fixed or variable, or whether they change with T, P or H2O activity. In the absence of direct in situ observation, the value of n in Equation 1 can be derived from quartz solubility measurements in CO2–H2O fluids because CO2 does not interact detectibly with solute silica. Following
Quartz solubility 800°C 10 kbar
1.2 1
SiO2 molality
Quartz solubility in CO2–H2O fluids
(A) 1.4
0.8 0.6 0.4 0.2 0 0
0.2
0.4
XH2O
0.6
0.8
1
(B) –2.0
log aSiO2 = log amonomer
increases isobarically, indicating increasing fraction of dissolved silica in dimers relative to monomers. In contrast, isothermal increases in pressure yield depolymerization of aqueous silica. Higher oligomers such as the ring trimer become increasingly abundant at 10 kbar within approximately 150C of the critical end point at 1080C in the system SiO2–H2O (Newton & Manning 2008a), but at lower temperatures appear to have negligible concentration in SiO2–H2O fluids.
–2.5
–3.0
n = 1.86
n = 4.02
–3.5
3
CO2-H2O NaCl-H2O –4.0 –0.4
2.5 5 10
–0.3
–0.2
–0.1
0.0
log aH2O 15 20 kbar
Fig. 2. (A) Variation in the molality of SiO2 at quartz saturation with CO2 (filled circles) and NaCl (open circles) at 800C, 10 kbar. Data from Newton & Manning (2000, 2009). (B) Variation in calculated activity of the aqueous silica monomer with H2O activity for same data as in (A). Activity models from Aranovich & Newton (1996, 1999). Slopes from linear fits to data correspond to n, the total hydration number; n in CO2–H2O of 4 indicates 2 molecular H2O of solvation; the lower n and higher solubility of quartz in NaCl–H2O at a given H2O activity implies solubility enhancement; however, the mechanism is debated (see text).
logKmd
2
1.5
1
2Si(OH)4 = Si2O(OH)6 + H2O 0.5 500
600
700
800
900
1000
Temperature (°C) Fig. 1. Variation in the equilibrium constant (Kmd) for homogeneous equilibrium between silica monomers and dimers (Equation 2) as a function of P and T, as calculated from Equation 4 (Newton & Manning 2002a, 2009).
Walther & Orville (1983), the hydration state of solute silica can be derived from the equilibrium constant (K1) for Equation 1 at fixed T and P, K1 ¼
am n aH 2O
ð5Þ
Role of saline fluids in deep-crustal and upper-mantle metasomatism 61
log am ¼ n log aH2 O þ log K1
ð6Þ
from which it can be seen that n is given by the variation in monomer activity with H2O activity. Walther & Orville (1983) proposed that the number of H2O molecules per silica unit is two (i.e. n = 4) based on their quartz solubility measurements in CO2–H2O. However, they had to assume that H2O and CO2 form ideal mixtures at their experimental conditions (2 kbar and 600C) and that SiO2 solubility was low enough that total molality could proxy for monomer activity. Subsequent studies gave n = 3.5 or 2, or suggested that n varies with XCO2 (Newton & Manning 2000; Shmulovich et al. 2006; Akinfiev & Diamond 2009). With the advent of accurate and precise solution models for CO2–H2O (Aranovich & Newton 1999) and aqueous silica (Newton & Manning 2002a, 2003), it is now possible to evaluate n more rigorously. As shown in Fig. 2B, results on quartz solubility in CO2–H2O over a wide range of CO2 contents at 10 kbar, 800C, show that n = 4 (Newton & Manning 2009). The same analysis for the dimer yields n = 7. Other high-quality data on quartz solubility in CO2–H2O at lower P and T yield the same values for n when SiO2 and H2O activities are calculated with the same models. Evidently, the aqueous silica monomer and dimer possess fixed stoichiometry of Si(OH)4Æ2H2O and Si2O(OH)6Æ4H2O – i.e. two solvating H2O molecules per Si – in H2O–CO2 fluids for a large range of crustal and mantle conditions. To the extent that CH4 and CO are also unreactive with SiO2, the simple model for quartz solubility in CO2–H2O should apply generally to quartz solubility in COH fluids over a wide range of oxygen fugacities.
ently with NaCl solutions than with CO2 solutions. Quartz solubility in H2O–NaCl solutions at low but constant P and T displays the interesting behavior of initially increasing with added salt (Fig. 3); i.e. it ‘salts in’. At 1 kbar, quartz solubility increases substantially with increasing NaCl concentration (Novgorodov 1977; Xie & Walther 1993). At 2 kbar and 700C, the solubility first increases, but reaches a maximum at about 10 mol % NaCl and then decreases with further increase in salinity (Fig. 3). This transition from salting-in to salting-out behavior probably has to do with ionization of NaCl, considering that solute NaCl at low pressures, high temperatures is strongly associated (e.g. Quist & Marshall 1968), but at pressures higher than about 4 kbar at 700C becomes progressively ionized at all temperatures and concentrations (Aranovich & Newton 1996; Tropper and Manning 2004). This is supported by the solubility behavior at 4.35 kbar, 700C (Fig. 3). In addition, dissolved silica appears to interact differently with different salts. Shmulovich et al. (2006) investigated quartz solubility in a variety of salt solutions to 9 kbar and 800C. Large-ion alkali halides (with Cs+, I), etc.) have a greater tendency toward salting-in of silica, such that the enhancement persists to high pressures. In contrast, these authors found that CaCl2 produces no enhancement, even at low pressure (Fig. 4). The unique behavior of quartz
0.9
700°C
0.8
Newton & Manning (2000) Manning (1994) Anderson & Burnham (1967)
15 kbar 0.7 0.6
mSiO2
where the activity of quartz is assumed to be unity, am is the activity of the monomeric species, and aH2O is the activity of H2O. Taking logarithms and rearranging leads to
10
Anderson & Burnham (1965) Xie & Walther (1993)
7
Novgorodov (1977)
0.5
Quartz solubility in alkali–halide salt solutions
0.4
Comparison with CO2–H2O At high P and T, quartz solubility in NaCl–H2O solutions behaves as in CO2–H2O solutions: it declines systematically with increasing NaCl (Fig. 2A, Newton & Manning 2000; Shmulovich et al. 2001). Quartz solubility is not greatly different in the two solvent media on an H2O mole fraction basis. More importantly, however, at the same H2O activity, quartz solubility is much higher in the salt solutions (Fig. 2B). Because the difference grows larger with decreasing XH2O, NaCl brines will be more effective solvents than CO2–H2O fluids – not just for metals, but also for SiO2, the major rock forming oxide. The more subdued decline in solubility with decreasing XH2O is not the only way in which quartz interacts differ-
0.3
5 4.35 4 3
Halite sat
2
0.2
1.5 1 kbar
0.1 0 0
0.1
0.2
0.3
0.4
0.5
0.6
XNaCl Fig. 3. Variation in quartz solubility with NaCl mole fraction and pressure at 700C (Newton & Manning 2000). At low pressures, quartz solubility initially rises (‘salts-in’) with increasing NaCl concentration. The behavior disappears above approximately 5 kbar, so that quartz salts out at all NaCl concentrations (with permission from Elsevier).
62 R. C. NEWTON & C. E. MANNING
2.5
Solubility enhancement as solvent modification A conventional line of approach followed by several authors (Xie & Walther 1993; Shmulovich et al. 2001, 2006) makes use of a parameter first defined by Setche´now (1892) to understand the effect of dissolved ionic salts on the solubility of organic compounds. Departures from ideal mixing behavior (i.e. solubility negatively proportional to the mole fraction or molality of the salt) are embodied in the expression:
Newton & Manning (2000) Shmulovich et al (2006)
Relative molality of SiO2
2 CsCl 800 0.5 1.5 NaCl 700 0.2 1
0.5
0
log i ¼ Bi I
KCl 800 0.5 NaCl 800 0.5 CaCl2 800 0.5
0
0.1
0.2
0.3
0.4
0.5
0.6
1 - XH2O Fig. 4. Variation in quartz solubility in salt solutions at various T and P. Lines showing Si molality relative to that in pure H2O are from Shmulovich et al. (2006), with labels identifying salt type, T (in C) and P (in GPa). Filled circles data at 700C, 0.2 GPa, from Newton & Manning (2000).
solubility in different salt solution indicates complex interaction that may have to do with complexing, polymerization, hydration state, mixing statistics (possibly including or excluding ionized units) and local electrical environment of the SiO2 units. Shmulovich et al. (2006) showed that, to a good approximation, the solubility of quartz at a given P and T in a mixed-salt solution is proportional to the sum of the solubilities in the individual salt solutions. This applies also to solvent mixtures of salt and CO2. There are several empirical formulations of the variation in quartz solubility with salt concentration (e.g. Fournier et al. 1982; Fournier 1983; Shibue 1996), but more recent efforts have focused on developing models that explicitly account for the physical chemistry of salting-in and saltingout behavior. The salting-out of silica is readily understood if the solute silica consists of hydrated species, because of the decrease in H2O activity. The physical chemical explanation for salting-in is, however, the subject of ongoing debate. Current hypotheses can be regarded as falling into two categories: (1) the effect is a property of the solvent; i.e. salt components alter electrostatic properties of H2O, especially the dielectric constant, so that uncharged species like silica become more compatible with the hydrous matrix; and (2) the effect is a property of the solute, in that more soluble species are formed by reaction of the silica with salt constituents. Resolution of this question will require additional accurate and specifically directed solubility and spectroscopic studies.
ð7Þ
where ci is the activity coefficient of the organic (or uncharged) solute, I is the ionic strength of the solution (for alkali halides simply the salt molality) and Bi is the Setche´now coefficient, unique to each salt and uncharged solute and generally a function of T and P. This approach cannot explain salting-in followed by salting-out at higher salt concentration unless the Setche´now coefficient is also made a function of salt concentration. For quartz solubility in salt solutions, the expression of Shmulovich et al. (2006) is: log mSiO2 ¼ log mSiO þ 3:5 log XH2 O þ amib 2
ð8Þ
where mi indicates total molality of the subscripted constituent, and the superscript zero refers to initially pure H2O. The third term on the right embodies the Setche´now concept while allowing the term to be a function of salinity and, through the empirical parameters a and b, pressure and to a lesser extent temperature. The second term describes the effect of decreasing H2O activity with salinity, and models the average total hydration number (n) of solute silica as 3.5. The parameter b must be less than unity when salting-in occurs (almost always at low H2O density). The authors give a table from which to obtain a and b by interpolation in the pressure range 1–9 kbar and temperature range 400–800C. Figure 4 compares the quartz solubility predictions of Shmulovich et al. (2006) for various salt solutions, T and P. The advantage of this method is that it assigns all non-ideality in the solubility calculation to a term which is effectively a function of H2O activity. A disadvantage, noted by Evans (2007), is that the model does not directly address the dissociation of NaCl at high pressures and concentrations, so that the mole fraction of H2O is ambiguous (taken by Shmulovich et al. 2006; to be 1 ) XNaCl). Therefore, the model is to some extent non-physical. Also, no account is taken of polymerization of SiO2 in solution, which is an important contributor to SiO2 solution non-ideality at the high concentrations associated with elevated P and T (Newton & Manning 2002a, 2003). Shmulovich et al. (2006) maintain that neglect of these factors is necessary for a unified account of their experimental results for a variety of salts, since
Role of saline fluids in deep-crustal and upper-mantle metasomatism 63
Solubility enhancement as solvent interaction A different approach implies that salting-in is due to a reaction of SiO2 with salt (and possibly H2O) to produce soluble hybrid species. Tanaka & Takahashi (1999) showed by high-resolution mass spectrometry at ambient conditions of quartz-saturated NaCl solutions that there is considerable complexing of Na with polymeric silica forms (monomer, dimer and tetramer). Anderson & Burnham (1967) suggested that enhancement of quartz solubility in KCl solutions at 3 kbar and 600C could be explained by a reaction: SiO2 þKCl þ 2H2 O ¼ KH3 SiO4 þHCl
ð9Þ
The postulated solvent interaction should therefore generate acid solutions under P–T conditions where salting-in occurs; this has not yet been verified for quartz solubility (e.g. by pH measurements, either in situ or on quenched solutions). Anderson & Burnham (1967) did not attempt to show that their hypothesis could account for the phenomenon of salting-in followed by salting-out at higher salinity. Several experimental and theoretical studies (summarized in Frank et al. 2003) have shown that, if Al is present, as in aluminosilicate minerals and melts, acid solutions are generated by reaction with NaCl. Another approach was adopted by Evans (2007). She postulated the dissolution reaction: SiO2 þdNaClþeH2 O ¼ SiO2 ðNaClÞd ðH2 OÞe
ð10Þ
which implies that NaCl is complexed as an undissociated unit. All species in solution are regarded as an ideal mixture, including H2O, NaCl, Na+, Cl), SiO2Æ3H2O (the assumed hydration state of the monomeric complex), and the postulated hybrid complex in Equation 10. Her formulation incorporates non-ideality of the H2O–NaCl solutions as a consequence of pressure-induced ionization, but does not use measured activity values of H2O as primary input; rather, she uses the dissociation constant a as a fitting parameter. Evans (2007) was able to obtain reasonable fits to the solubility data of Newton & Manning (2000) and Shmulovich et al. (2006) with a broad range of d and e values; she chose the values d = 0.5 and e = 0. Figure 5 compares the predictions of Evans (2007) for NaCl–H2O at 700C and various pressures with the experimental data of Newton & Manning (2000). The main advantage of her
1.6
700°C Evans (2007) Newton & Manning (2000)
1.4
Relative mole fraction SiO2
activity coefficients and SiO2 speciation are not as yet known for most salt solutions at high T and P. Akinfiev & Diamond (2009) proposed a similar reconstruction. They showed that experimental measurements on quartz solubility in ionic fluids at high T and P can equally well be represented with n = 2.
1.2 1 0.8 0.2 GPa 0.6 0.435 GPa 1.0 GPa
0.4 0.2 0 0
0.1
0.2
0.3
0.4
0.5
0.6
Mole fraction NaCl Fig. 5. Variation in quartz solubility with NaCl mole fraction and pressure at 700C. Circles and solid lines are from Newton & Manning (2000); dashed lines from Evans (2007). Relative mole fractions are normalized to values in initially pure H2O.
method is that change of the activity coefficients of H2O and dissolved salts is explicitly taken into account; indeed, Evans (2007) showed that this effect can account for the transitions from salting-in at low pressure to salting-out at high pressures. This is evident from Equation 10, where it is seen that decreasing the activity of both NaCl and H2O, which decrease occurs with increasing pressure, destabilizes the hybrid complex. Newton & Manning (2006) attempted to incorporate both pressure-induced ionization of NaCl and polymerization of SiO2 in a single theory. This formalism uses H2O activity measurements of Aranovich & Newton (1996) and makes the assumption that solute SiO2 forms an ideal substituent among all other solution entities (H2O, NaCl, Na+, and Cl)) subject to the formula: XSiO2 ¼
mSiO2
mSiO2 h i ÞXNaCl þ 55:51 1 þ ð1þ XH O
ð11Þ
2
where mole fraction and molality of SiO2 refer to total silica in solution and a, the dissociation parameter, characterizes the non-ideality of NaCl solutions (Aranovich & Newton 1996). Theoretically, a should run from zero for undissociated NaCl to unity for completely dissociated NaCl. In fact, a as fitted by Aranovich & Newton (1996) to their activity data actually exceeds unity by about 10% at the highest pressures (15 kbar). At 10 kbar and 800C it is almost exactly unity. This assumption and the polymerization model of Newton & Manning (2002a, 2006) lead to the formula:
64 R. C. NEWTON & C. E. MANNING
1
aSiO2 ¼ Xm ¼
ð1 þ 8XSiO2 Kmd =aH2 O Þ2 1 4Kmd =aH2 O
ð12Þ
where aH2O is closely equal to XH2O ⁄ (2 ) XH2O), again subject to the same ideal solution model of all ions and neutral molecules and Kmd is the equilibrium constant for dimer formation, Equation 2, assumed to apply to SiO2 solutions in H2O–NaCl solvents. Critique of the quartz solubility models All of the above attempts to quantify quartz solubility and silica activity in salt solutions in terms of physical chemical properties have advantages in data organization, but also significant defects. The Newton & Manning (2006) model yields a Walther–Orville slope of 1.86 (Fig. 2B), very near 2, suggesting that n = 2 in Equation 2a, and consequently that there no hydrogen-bonded H2O molecules in monomeric silica. The parameter n = 2, together with Equations 11 and 12, fits the Newton & Manning (2000) experimental quartz solubility data at 800C and 10 kbar very well. However, the hypothesis leads to a contradiction in that, at the limit of NaCl dilution, the number n must be 4 to be in accord with the results for H2O–CO2 solubility (Newton & Manning 2009). Such a change of Walther– Orville slope at low XNaCl is not evident in the Newton & Manning (2000) data. It seems probable either that the assumption of ideal mixing of neutral silica complexes with charged ions is incorrect or that there is indeed some additional complexing with NaCl, in the manner envisioned by Evans (2007). The quartz solution hypothesis of Anderson & Burnham (1967) embodied in Equation 9 supposes that quite acid solutions result from the dissolution of quartz in alkali halide solutions. Newton & Manning (2006) tested this hypothesis by measuring the pH of quenched fluid from a solubility experiment at 10 kbar and 800C and XNaCl = 0.17. Their measurement of pH 7 was not significantly different from acid–base neutrality at 25C. However, pH measurements have not been performed on fluids quenched from low-pressure quartz solubility experiments; it is possible that Anderson & Burnham’s (1967) model may apply to quartz solubility in the ‘salting-in’ range. Shmulovich et al. (2006) ascribe the effect of dissolved salts on quartz solubility to change in solute properties, according to Setche´now’s formulation. Their discussion is more comprehensive than that of Newton & Manning (2006), who considered only NaCl solutions or than that of Anderson & Burnham (1967), who considered only KCl solutions, but takes no account of such phenomena as hydration state, polymerization or ionization of the salts. Also, as Evans (2007) pointed out, the Setche´now coefficient must change drastically with pressure and salt concentration in order to explain the transition from salting-in to
salting-out. Therefore, much of the utility of the Setche´now formulation is lost. The hypothesis of Evans (2007) that SiO2 reacts with dissolved NaCl to form a neutral hybrid solute complex has the ability to explain salting-in to salting-out in terms of changing NaCl and H2O activities. Her hypothesis also supposes that the SiO2-salt complex is neutral, which supports the observation of pH-neutral quenched fluid observed by Newton & Manning (2006). However, her formalism does not take into account what must be one of the most important properties of solute silica, that of polymerization. In summary, the physical–chemical basis for quartz dissolution in natural fluids, at high P and T, and especially those containing ionized salts, is currently in an imperfect state. Although key data, such as pH measurements, are lacking, empirical equations predict quartz solubility adequately in a variety of simple media which contain both neutral and ionized constituents.
SOLUBILITY ENHANCEMENT OF ROCKFORMING MINERALS BY NaCl In addition to quartz, the solubilities of various simple oxides or silicate minerals in concentrated NaCl solutions have now been measured, though data are still sparse. Nevertheless, the data are sufficient to reveal a general pattern of the enhancement of mineral solubility in H2O–NaCl relative to that in pure H2O at high P and T. Shmulovich et al. (2001) found marked increase in the solubility of diopside, CaMgSi2O6, at 5 kbar and 650C in aqueous fluids of salinity up to XNaCl = 0.26, and Macris & Manning (2006) found a similar enhancement at 800C and 10 kbar. The dissolution is incongruent: the Ca–silicate portion is more soluble and forsterite is left as a residue. Wollastonite (CaSiO3), corundum (Al2O3), and grossular (Ca3Al2Si3O12) show even greater enhancement, reaching concentrations of up to 10 times the pure-H2O solubility at 800C and 10 kbar (Fig. 6; Newton & Manning 2006, 2007). The only silicate mineral investigated thus far which does not show enhancement by NaCl is zircon (ZrSiO4; Newton et al. 2010). The marked solubility enhancements cannot result simply from alteration of the solvent properties, but must reflect strong reactions of the mineral constituents with the dissolved salt components. The example of wollastonite is instructive. The initial salting-in with increasing NaCl concentration is followed by a flat maximum at about XNaCl = 0.35 and then a salting-out at still higher salinity (Fig. 6). Quenched fluids are strongly basic (pH 11–12), which indicates that when Ca enters the solution some OH- is created in the dissolution reaction. We adopt an approach similar to that of Evans (2007) and postulate a wollastonite dissolution reaction of the form:
Role of saline fluids in deep-crustal and upper-mantle metasomatism 65 all molecules and ions in the solution (NaCl counts for two mixing units, Na+ and Cl), in a completely dissociated aqueous fluid). In Newton & Manning’s (2006) corundum solubility measurements, quench pH was 7, implying that neither H+ or OH) are involved in the dissolution reaction. A reaction which satisfies the above observations is:
1.2
800°C 10 kbar
Grossular
1
log X/X°
Wollastonite 0.6
0.4
Halite saturation
0.8
2Al2 O3 þNaClþ6H2 O¼ NaAlðOHÞ4 þAlðOHÞ2 Clþ2AlðOHÞ3
Corundum
0.2
0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
XNaCl Fig. 6. Enhancement of solubilities of corundum, wollastonite and grossular by NaCl in NaCl–H2O solutions. All solubilities are observed mole fraction relative to that in initially pure H2O. Data from Newton & Manning (2006, 2007). Dashed vertical line is halite saturation.
(Newton & Manning 2006), for which a = 1, b = 6, and XNaCl = 0.143. The neutral complex NaAl(OH)4 was proposed by Walther (2001) in his study of corundum solubility in NaCl solutions at low P and T. When SiO2 is added to the system Al2O3–H2O–NaCl, Al2O3 solubility enhancement becomes very strong and, in fact, quartz solubility is simultaneously enhanced over that at the same salinity in the absence of Al2O3 (Newton & Manning 2008b). This mutual enhancement culminates in the precipitation of albite at higher salinities and SiO2 contents of the fluid (Fig. 7). Quenched solutions are very acid (pH 1–2), implying that the dissolution reaction of corundum involves complexing with Na and Si, leaving an excess of Cl). A consistent dissolution reaction must be something like the following: Al2 O3 þ2NaCl þ 6SiðOHÞ4 ¼ 2NaAlSi3 Oð3þdÞ
CaSiO3 þaNaClþbH2 O ¼n hybrid species
ð13Þ
ð16Þ
ðOHÞð102dÞ þð1 þ 2dÞH2 O þ 2HCl
ð17Þ
It may be shown quite generally that, if wollastonite solubility as a function of XNaCl has a maximum (salting-in going to salting-out), then, the maximum in solubility ) determined from the occurs at a NaCl content (XNaCl reaction coefficients by: XNaCl ¼
a aþb
ð14Þ
Equations 13 and 14 state that solubility enhancement is eventually arrested and reversed by decreasing H2O activity with increasing salinity. The observed maximum in the wollastonite solubility curve near XNaCl = 0.35 can be explained most simply by the dissolution reaction: CaSiO3 þNaCl þ 2H2 O ¼ CaClþ þOH þH3 NaSiO4
ð15Þ
= 1 ⁄ 3 and the basic nature of the quenched wherein XNaCl fluids is explained. The neutral complex H3NaSiO4 is analogous to that previously postulated by Anderson & Burnham (1967) for quartz solubility in KCl solutions (see Equation 9). Corundum dissolution in NaCl solutions at 10 kbar and 800C shows a different behavior. There is a nearly ten-fold increase in solubility over that in pure H2O with a maximum near XNaCl = 0.15 (Fig. 6), if, following Evans (2007), the mole fraction of Al2O3 is taken with respect to
Fig. 7. Variation in corundum solubility (as 104 times mole fraction of Al2O3) with NaCl and SiO2 concentration at 800C, 10 kbar. The light solid lines are lines of constant SiO2 mole fraction, and the bold solid line is quartz or albite + silicate liquid saturation; the transition between saturating phases occurs at the maximum in solubility. Dashed vertical line is halite saturation.
66 R. C. NEWTON & C. E. MANNING
SOLUBILITY ENHANCEMENT OF Ca SALTS BY NaCl The greatest solubility-enhancements by brines are found in systems with calcium carbonate, sulfate, fluoride and phosphate. Figure 8 shows enhancement factors for calcite, anhydrite, fluorite and fluorapatite, Ca5(PO4)3F. The solubility versus XNaCl curves are strongly convex to salinity, indicating that the H2O activity is not a major influence, which in turn suggests that the solute species are not hydrates and that multiple hybrid species form upon dissolution. For calcite, the dissolution reaction must be something like: CaCO3 þ2NaCl ¼ CaCl2 þNa2 CO3
ð18Þ
In this reaction calcite, a mineral very insoluble in pure H2O, becomes very soluble in concentrated NaCl solutions at high T and P. Anhydrite undergoes a similar dissolution reaction, but solubility enhancement is extreme. In both systems the solubility becomes so high at high salinity and temperature that critical mixing between saline fluids and mixed-salt melts is anticipated at moderately high temperatures and salinities (probably 900–1000C and XNaCl of 0.3–0.4; Fig. 9).
1000
800°C 1 GPa
F-apatite anhydrite
100
XCa/X°Ca
The average Si ⁄ Al ratio of the solute aluminosilicate species estimated by mass balance by Newton & Manning (2008b) is near three at albite precipitation, which supports the hypothesis of Anderson & Burnham (1983) that Al can exist in Si and Na bearing fluids in the form of a feldspar-like molecule. The hydration state of the complex, characterized by the parameter d, is indeterminate. In view of the marked enhancement of CaSiO3 and Al2O3 solubility in aqueous NaCl and NaCl–SiO2 solutions at high P and T, the solubility of a compound of these components, such as grossular garnet (Ca3Al2Si3O12), is of interest. Newton & Manning (2007) found that grossular solubility is congruent even at very high salinity, with relative enhancement greater than either wollastonite or corundum (Fig. 6). The solubility of Al2O3 can be enhanced over that of corundum in pure H2O by a factor of nearly 100 in solutions containing also NaCl and CaSiO3 (Newton & Manning 2007). Quenched fluids were very basic, showing that Ca–chloride production drives the high solubility. Interestingly, similar enhancement by NaCl solutions at high P and T of the analogous Fe3+ garnet, andradite, does not occur; evidently, Fe2O3 is a much more refractory component than Al2O3 in salty solutions of calc-silicates (Wykes et al. 2008).
Fluorite
Calcite
10
1
0.1 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
XNaCl Fig. 8. Solubility enhancement of Ca salt and oxysalt minerals, including calcite, fluorite, fluorapatite and anhydrite, at 800C, 10 kbar. Enhancement, as mole fraction relative to that in initially pure H2O, was calculated using equations given in Newton & Manning (2002b, 2005), Tropper & Manning (2007) and Antignano & Manning (2008).
APPLICATIONS TO PETROLOGY Solubility studies of various minerals in salty fluids over the last decade have suggested that such fluids may be important agents in petrogenesis, capable of explaining some oftobserved metasomatic phenomena that may not be otherwise readily interpreted. A few examples are discussed below. Deep-crustal fluids in general Various mechanisms have been suggested for the origin of concentrated brines in deep-crustal igneous and metamorphic processes, including involvement of deeply-buried evaporites (e.g. Yardley & Graham 2002), downward circulation of dissolved salts of surficial origin in thermally excited convection cells (McLelland et al. 2002) and magmatic fluids emitted in the late stages of crystallization of igneous rocks at depth (Ryabchikov & Hamilton 1971; Hansteen & Burke 1990). The last hypothesis has gotten considerable momentum from the experiments of Webster et al. (1999), which have shown that concentrated brines can be emitted from Cl-saturated magmas at depth, most abundantly from basaltic intrusions. Whatever the sources of the salty fluids, their high affinity for carbonate and sulfate militates that they must become complex brines in their passage through crustal rocks. The properties of low H2O activity yet relatively high solution capability for
Role of saline fluids in deep-crustal and upper-mantle metasomatism 67 into carbonate-rich and NaCl-rich fractions during cooling (path A), in which case a carbonatite–fenite association may develop, or pass continuously into ultrasaline fluids (path B) capable of voluminous carbonate alteration of country rocks, as in the carbonate-metasomatized megashear zones like the Attur Fault Zone of South India (Wickham et al. 1994). Sulfur-rich magmatism
Fig. 9. Schematic CaCO3–NaCl–H2O pseudo-ternary showing solubilities at 10 kbar, from Newton & Manning (2002b). Critical mixing between a hydrous, NaCl-bearing carbonatite liquid and a CaCO3-rich hypersaline aqueous solution may occur near 1000C and 30 mol % NaCl, 25 mol % CaCO3. A hydrous carbonatite liquid (composition A) with calcite at 1000C will, upon cooling to 950C (point P), emit a concentrated salt solution (point P¢). Both fluids will crystallize calcite upon further cooling over a narrow T interval. At P¢¢, only the CaCO3-rich brine phase remains and deposits most of solute CaCO3 in cooling only 200C further. A calcite-saturated brine of composition B, generated at 1000C, will avoid the two-fluid region and deposit nearly all of its substantial solute carbonate in isobaric cooling past 800C.
quartz and silicate minerals, including actual enhancement for some rock components, indicate that the concentrated brines must be important in rock alteration and mass transport in the deeper parts of the crust. Carbonatite–fenite complexes and carbonated shear zones The association of carbonatite intrusions of obviously igneous origin surrounded by wide tracts of alkali-metasomatized rocks has been described from many localities (e.g. Alno¨ Island, Sweden: Morogan & Wooley 1988). Students of these complexes have largely come to the conclusion that they result from fluid immiscibility in the intrusion process between carbonate-rich and saline aqueous components (e.g. Hamilton et al. 1979). Figure 9 shows a semiquantitative diagram of the pseudo-ternary system CaCO3– NaCl–H2O at 10 kbar, after Newton & Manning (2002b). The very high solubility of calcite in NaCl–H2O solutions at high T and P indicates that such T- and ⁄ or P-dependent fluid immiscibility must take place in this model carbonatite–fenite system. Depending on the initial bulk composition, a supercritical carbonatite magma may either split
The anhydrite solubility study of Newton & Manning (2005) suggested a new mode of anhydrite involvement in Mt. Pinatubo-type S-rich volcanism. Many authorities now envision a ‘basalt trigger’ for the massive 1991 eruption in the Philippines (e.g. Pallister et al. 1992) and, in view of the findings of Webster et al. (1999) of Cl-induced early magmatic outgassing, the possibility arises that the sulfur could have originally been released as sulfate in a magmatic brine. The highly oxidizing fluids which apparently affected the rocks under Mt. Pinatubo (Cameron & Hattori 1994) and which were ultimately vented massively to the surface in the form of sulfate aerosols, could have been in the form of a concentrated brine of magmatic origin. Hattori & Keith (2001) imagined a close genetic connection between Mt. Pinatubo-type eruptions and the porphyry Cu–Mo ores of western North America, which are invariably found in the aureoles of small Late Mesozoic and Tertiary granitic intrusions. These sulfide ore vein deposits are always characterized by massive oxidation and anhydrite deposition. It seems reasonable to suppose that Cl- and S-laden basaltic magmas ‘triggered’ the hypabyssal intrusions in much the same manner as for the Mt. Pinatubo dacite eruptives. Fluid-present granulite-facies metamorphism Deep-crustal granulite-facies metamorphism, which generated the extensive high-grade terranes characteristic of the Precambrian shields of all continents, yields mineral assemblages that indicate formation at low H2O activity, yet possess geochemical and petrological characteristics which in some cases require interaction with a mobile fluid phase (Newton et al. 1998). This evidence, for example from the Archean craton of southern India, includes orthopyroxenebearing metasomatic rocks, massive quartz veins with orthopyroxene-bearing selvages (Fig. 10), whole-rock depletions of Rb and Th across isograds (Hansen et al. 1995), and incipient charnockitic alteration of amphibole-biotite gneisses, as at the amphibolite facies-granulite facies boundary in southern India (Hansen et al. 1987). Early models invoked CO2-rich fluids to explain much of the metasomatism associated with granulite metamorphism (e.g. Newton et al. 1980). Petrologic considerations posed problems for this simple model (e.g. Lamb et al. 1987;
68 R. C. NEWTON & C. E. MANNING
6
10 kbar
CaSO4 molality
5
Fig. 10. Synmetamorphic quartz vein in charnockite with orthopyroxene selvage from near Dharmapuri, Tamil Nadu, South India. Knife for scale.
4
3
900°C 2
halite sat.
700
1
600 500 0
Yardley & Valley 1997); however, mounting evidence suggests that CO2-rich fluids coexisted with an immiscible alkali–halide-rich solution, consistent with the large miscibility gap in the H2O–CO2–NaCl system at high P and T (Johnson 1991; Gibert et al. 1998; Heinrich 2007). Studies of mineral solubility in saline fluids offer insights on the granulite-fluid controversy. Early models were heavily influenced by the plethora of dense CO2-rich fluid inclusions found in many or most granulites, first recognized by Touret (1971). Concentrated brine inclusions in quartz have also been reported from granulites (Touret 1985; Crawford & Hollister 1986); these are rarer and less well preserved than the CO2 inclusions. The fluid inclusion evidence may be questioned in light of the demonstrated likelihood of postentrapment alteration (Sterner & Bodnar 1989; Barker 2007). However, saline fluids trapped in minerals undergoing granulite-facies metamorphism must originally have been very concentrated if they coexisted with an immiscible CO2-rich fluid. The importance of salty fluids has probably been underestimated because of difficulty of capture as fluid inclusions – brines have much greater wetting ability for gain boundaries in silicate mineral aggregates and quartz than does dense CO2 (Watson & Brenan 1987). It is clear from the solubility studies that halide (±carbonate, ±sulfate) solutions at high T and P are much more efficient than dense CO2 in promoting mineral reactions and performing mass transport in the deeper parts of the crust. The amounts of the fluids need not have been excessive to perform important functions such as desiccation and trace element depletion (Aranovich & Newton 1996). For instance, the high solubility of apatite, a principal rare-earth-element carrier, in high P–T brines could explain REE mobility inferred in some granulite terranes (Pan & Fleet 1996). The great wetting ability of salt solutions for mineral grain boundaries, in contrast to CO2, may also be an important factor in deep-crustal metasomatism.
800
0
0.1
halite sat.
0.2
0.3
0.4
XNaCl Fig. 11. Variation of the solubility of anhydrite with NaCl concentration and temperature at 10 kbar, from Newton & Manning (2005). Experiments were unbuffered, but fO2 is inferred to be near Ni–NiO equilibrium.
The high solubility of CaSO4 in alkali halide solutions (Fig. 11) could be a factor in the origin of some highly oxidized terranes, like Wilson Lake, Labrador (Currie & Gittens 1988) and Labwor, Uganda (Sandiford et al. 1987). These extensive Precambrian terranes reveal high oxidation states in their high Mg ⁄ (Mg + Fe) ratios of silicates, even in felsic compositions and in the compositions of the Fe, Ti oxide minerals, including, in some cases, modal hematite. This oxidation could only be the result of pervasive infiltration of highly oxidizing solutions at high grade conditions; highly oxidized protoliths as envisioned by Arima et al. (1986) could not account consistently for the syndrome of related phenomena, which include profound shearing deformation with related vein mineralization, high positive gravity anomalies centered upon the granulites and Archean age of some, in spite of the reducing surficial weathering conditions that prevailed at that time. Cameron & Hattori (1994) ascribed the oxidizing metamorphism to the passage of SO2-bearing fluids released from underplated basaltic magmas; Newton & Manning (2005) modified this concept to include the action of orthomagmatic brines carrying concentrated sulfate. Wannamaker et al. (2004) present evidence from magnetotelluric sounding data that the lower crust under the Great Basin of the western US contains highly conductive, interconnected pore fluids and they suggest that these fluids have their origin in underplated basaltic intrusions. Such a view of regional deep-crustal fluid action is consonant with the ‘Basin-and-Range’ model of granulite-facies metamorphism advanced by Hopgood & Bowes (1990).
Role of saline fluids in deep-crustal and upper-mantle metasomatism 69
Aluminum mobility in high-grade metamorphism The element Al has long been thought to be effectively inert in metasomatic processes. The assumption of fixed Al in metasomatic mass transfer has been used to anchor calculations of mineral reactions in high-grade rocks in open systems (e.g. Hansen et al. 1987). This point of view has been influenced in part by the low hydrothermal solubility of corundum compared to that of quartz at high P and T. However, many authors have pointed to the existence of widespread Al–silicate veining in highgrade terranes (Kerrick 1988; McLelland et al. 2002), where Al transport seems to have been a major process. Anderson & Burnham (1983) argued for the transport of solute Al in the form of feldspar-like complexes. Simpler complexes arising from interaction between Si and Al in alkali-free fluids have also been inferred (Manning 2007). The possible role of Al–Si and alkali–Al–Si complex transport has received only limited attention (e.g. Manning et al. 2010), chiefly because the basic thermodynamic properties of such species are still not tabulated; however, there is evidence that such transport may be further promoted at elevated salinity. The sillimanite-quartz veins described by McLelland et al. (2002) (Fig. 12) are characterized by very salty fluid inclusions in quartz with up to 25 wt% NaCl equivalent. It is possibly significant that this concentration range is the location of a very sharp Al solubility maximum (Fig. 7), according to the solubility experiments of Newton & Manning (2008b). We can propose a scenario in which brines, liberated from the abundant intrusions in the area, increased in salinity as H2O was absorbed into migmatites in the country rocks and the NaCl concentration increased until Al was more or less suddenly mobilized at about 25 wt% NaCl). Further increase in salinity would then begin to have a decreasing effect on Al solubility.
Fig. 12. Sillimanite-quartz vein network in Lyon Mountain granitic gneiss, Adirondack Mountains, New York, USA. Courtesy J. McLelland (McLelland et al. 2002). Lens cap for scale.
The grossular solubility study of Newton & Manning (2007) showed that Ca–Al silicates are also quite soluble in concentrated brines at high T and P. In the case of calc-silicates, the driving force of Al solubility enhancement is the formation of CaCl+ ion, creating very basic fluids in the dissolution process. Chemical studies of calc-silicate lenses in granulites from Australia (Buick et al. 1993) showed that Al is a relatively mobile component, whereas Fe3+ is more inert, tending to remain at fixed levels in calc-silicate lenses interacting with surrounding quartzofeldspathic rocks. This observation concerts with the finding of low solubility of Fe3+ in high P–T NaCl solutions at least at relatively high oxygen fugacity (Wykes et al. 2008). Iron becomes much more soluble at lower oxygen fugacity and lower-grade conditions in rock compositions (Althaus & Johannes 1969; Chou & Eugster 1977).
SUMMARY Solubility studies lead to the conclusion that concentrated brines may be important agents in metasomatic alteration of deep-seated rocks. While providing low H2O activity, which is necessary that aqueous fluids be compatible with anhydrous silicate minerals, especially pyroxenes and garnets, they nevertheless retain relatively high solubility for silicate constituents, high Al-transporting ability in either calcareous or aluminous rocks and high affinity for other salt components, which makes them effective in traceelement mobilization. These specific properties contrast with the limited ability of CO2-rich solutions to mobilize rock constituents. The solubility enhancements by alkali halide solutions at high P and T show contrasting mechanisms in aluminous versus calcareous rocks. Acid solutions are generated in aluminous rocks due to the stability of soluble Na–Al complexes, as shown by a number of earlier workers (Candela 1990; Frank et al. 2003). In calcareous rocks, the stability of Ca ions in chloride solutions drives solubility enhancement. It is certain that concentrated brines will strongly inhibit critical phenomena in rock-H2O systems; this narrows the geologic opportunities for critical mixing between silicaterock melts and intergranular fluids in the crust, because all such fluids are likely to be salty to some degree. However, the high solubility of rock materials in such fluids shows that they are capable of extensive metasomatism under some conditions such that the affected rocks can be reconstituted to almost the same degree as by actual rock melting. Study of trace element behavior and fluid inclusions will doubtless further elucidate the comparative roles of mineralizing fluids and rock melting in the deep crust and upper mantle.
70 R. C. NEWTON & C. E. MANNING
ACKNOWLEDGEMENTS This work was supported by National Science Foundation grant EAR-0711521. The manuscript was completed while CEM was in residence at the Bayerisches GeoInstitute through the auspices of the Humboldt Foundation. Two anonymous reviewers provided invaluable information and advice.
REFERENCES Akinfiev NN, Diamond LW (2009) A simple predictive model of quartz solubility in water-salt-CO2 systems at temperatures up to 1000 C and pressures up to 1000 Mpa. Geochimica et Cosmochimica Acta, 73, 1597–608. Althaus E, Johannes W (1969) Experimental metamorphism of NaCl-bearing aqueous solutions by reaction with silicates. American Journal of Science, 267, 87–98. Anderson GM, Burnham CW (1965) The solubility of quartz in supercritical water. American Journal of Science, 263, 494–511. Anderson GM, Burnham CW (1967) Reaction of quartz and corundum with aqueous chloride and hydroxide solutions at high temperatures and pressures. American Journal of Science, 265, 12–27. Anderson GM, Burnham CW (1983) Feldspar solubility and the transport of aluminum under metamorphic conditions. American Journal of Science, 283-A, 283–97. Antignano A, Manning CE (2008) Fluorapatite solubility in H2O and H2O-NaCl at 700 to 900 C and 0.7 to 2.0 GPa. Chemical Geology, 251, 112–19. Aranovich LY, Newton RC (1996) H2O activity in concentrated NaCl solutions at high pressures and temperatures measured by the brucite-periclase equilibrium. Contributions to Mineralogy and Petrology, 125, 200–12. Aranovich LY, Newton RC (1999) Experimental determination of CO2-H2O activity-composition relations at 600-1000 C and 6-14 kbar by reversed decarbonation and dehydration reactions. American Mineralogist, 84, 1319–32. Arima M, Kerrich R, Thomas A (1986) Sapphirine-bearing paragneiss from the northern Grenville Province in Labrador, Canada. Protolith compositions and metamorphic P-T conditions. Geology, 14, 884–87. Barker AJ (2007) Post-entrapment modification of fluid inclusions due to overpressure: evidence from natural samples. Journal of Metamorphic Geology, 13, 737–50. Bowers TS, Helgeson HC (1983) Calculation of the thermodynamic and geochemical consequences of nonideal mixing in the system H2O-CO2-NaCl on phase relations in geologic systems. Equation of state for H2O-CO2-NaCl fluids at high pressures and temperatures. Geochimica et Cosmochimica Acta, 47, 1247–75. Buick IS, Hardy SL, Cartwright IC (1993) Granulite facies metasomatism: Zoned calc-silicate boudins from the Rauer group, East Antarctica. Contributions to Mineralogy and Petrology, 113, 557–71. Cameron EM, Hattori K (1994) Highly oxidized deep metamorphic zones: occurrence and origin. Mineralogical Magazine, 58A, 142–43. Candela PA (1990) Theoretical constraints on the chemistry of the magmatic aqueous phase. Geological Society of America Special Paper, 246, 11–20. Chou IM, Eugster HP (1977) Solubility of magnetite in supercritical chloride solutions. American Journal of Science, 277, 1296– 314.
Crawford ML, Hollister LS (1986) Metamorphic fluids, the evidence from fluid inclusions. In: Fluid–Rock Interactions During Metamorphism (eds Walther JV, Wood BJ), pp. 1–35. SpringerVerlag, Berlin. Currie KL, Gittens J (1988) Contrasting sapphirine parageneses from Wilson Lake, Labrador and their tectonic applications. Journal of Metamorphic Geology, 6, 609–22. Duan Z, Møller N, Weare JH (1995) Equation of state for NaClH2O-CO2 system. Prediction of phase equilibria and volumetric properties. Geochimica et Cosmochimica Acta, 59, 2869–82. Evans K (2007) Quartz solubility in salt-bearing solutions at pressures to 1 GPa and temperatures to 900 C. Geofluids, 7, 1–17. Fournier RO (1983) A method of calculation of quartz solubilities in aqueous sodium chloride solution. Geochimica et Cosmochimica Acta, 47, 579–86. Fournier RO, Rosenbauer RJ, Bishoff JL (1982) The solubility of quartz in aqueous sodium chloride solution at 350 C and 180 to 500 bars. Geochimica et Cosmochimica Acta, 46, 1975–78. Frank MB, Candela PA, Piccoli PM (2003) Alkali exchange equilibria between a silicate melt and coexisting magmatic volatile phase: an experimental study at 800 C and 100 MPa. Geochimica et Cosmochimica Acta, 67, 1415–27. Fu B, Touret JLR, Zheng YF (2001) Fluid inclusions in coesitebearing eclogites and jadeite quartzite at Shanghe, Dabie Shan (China). Journal of Metamorphic Geology, 19, 529–45. Gibert R, Guillame D, LaPorte D (1998) Importance of fluid immiscibility in the H2O-NaCl-CO2 system and selective CO2 entrapment in granulites: experimental phase diagram at 5–7 kbar, 900 C and wetting textures. European Journal of Mineralogy, 10, 1109–23. Hamilton DL, Freestone IC, Dawson JB, Donaldson CH (1979) Origin of carbonatites by liquid immiscibility. Nature, 279, 52– 54. Hansen EC, Janardhan AS, Newton RC, Prame WKBN, Ravindra Kumar GR (1987) Arrested charnockite formation in southern India and Sri Lanka. Contributions to Mineralogy and Petrology, 96, 225–44. Hansen EC, Newton RC, Janardhan AS, Lindenderg S (1995) Differentiation of Late Archean crust in the Eastern Dharwar Craton, Krishnagiri-Salem area, South India. Journal of Geology, 103, 629–51. Hansteen TH, Burke EAJ (1990) Melt-mineral-fluid interaction in the Oslo Rift, southeast Norway. II. High-temperature fluid inclusions in the Eikorn-Skrim complex. Norges Geologiske Undersokelse Bulletin, 417, 15–32. Hattori K, Keith JF (2001) Contribution of mafic melt for porphyry deposits; evidence from Pinatubo and Bingham. Mineralium Deposita, 36, 799–806. Heinrich W (2007) Fluid immiscibility in metamorphic rocks. Reviews in Mineralogy and Geochemistry, 65, 389–430. Hopgood AM, Bowes DR (1990) Contrasting structural features in the granulite-gneiss-charnockite-granite complex, Lake Baikal, USSR: evidence for diverse geotectonic regimes in early Proterozoic times. Tectonophysics, 174, 279–99. Izraeli ES, Harris JW, Navon O (2001) Brine inclusions in diamonds: a new upper mantle fluid. Earth and Planetary Science Letters, 187, 323–32. Izraeli ES, Harris JW, Navon O (2004) Fluid and mineral inclusions in cloudy diamonds from Koffiefontein, South Africa. Geochimica et Cosmochimica Acta, 68, 2561–75. Johnson EL (1991) Experimentally determined limits for H2OCO2-NaCl immiscibility in granulites. Geology, 19, 925–8.
Role of saline fluids in deep-crustal and upper-mantle metasomatism 71 Kennedy GC, Wasserburg GJ, Heard HC, Newton RC (1962) The upper three-phase region in the system SiO2-H2O. American Journal of Science, 260, 501–21. Kerrick DM (1988) Al2SiO5-bearing segregations in the Lepontine Alps, Switzerland: aluminum mobility in metapelites. Geology, 16, 636–40. Klein-BenDavid O, Izraeli ES, Hauri E, Navon O (2007) Fluid inclusions in diamonds from the Diavik mine, Canada and the evolution of diamond-forming fluids. Geochimica et Cosmochimica Acta, 71, 723–44. Kullerud K (1996) Chlorine-rich amphiboles: interplay between amphibole composition and an evolving fluid. European Journal of Mineralogy, 8, 355–70. Lamb WM, Valley JW, Brown PE (1987) Post-metamorphic CO2rich inclusions in granulites. Contributions to Mineralogy and Petrology, 96, 485–95. Macris CA, Manning CE (2006) The solubility of diopside in H2O-NaCl fluids at 800 C and 10 kbar. Eos, Transactions of the American Geophysical Union, 87. Fall Meeting Supplement, Abstract MR43C-1091. Manning CE (1994) The solubility of quartz in H2O in the lower crust and upper mantle. Geochimica et Cosmochimica Acta, 58, 4831–9. Manning CE (2007) Solubility of corundum + kyanite in H2O at 700C and 10 kbar: evidence for Al-Si complexing at high pressure and temperature. Geofluids, 7, 258–69. Manning CE, Antignano A, Lin HA (2010) Premelting polymerization of crustal and mantle fluids, as indicated by solubility of albite + paragonite + quartz in H2O at 1 GPa and 350–620 C. Earth and Planetary Science Letters, in press. Markl G, Bucher K (1998) Composition of fluids in the lower crust inferred from metamorphic salt in lower crustal rocks. Nature, 391, 781–3. Markl G, Ferry JM, Bucher K (1998) Formation of saline brines and salt in the lower crust by hydration reactions in partially retrogressed granulites from the Lofoten Islands, Norway. American Journal of Science, 298, 705–57. McLelland J, Morrison J, Selleck B, Cunningham B, Olson C, Schmitt K (2002) Hydrothermal alteration of late- to post-tectonic Lyon Mountain Granite Gneiss, Adirondack Mountains, New York: origin of quartz-sillimanite segregations, quartzalbite lithologies and associated Kiruna-type low-Ti oxide deposits. Journal of Metamorphic Geology, 20, 175–90. Mora C, Valley JW (1989) Halogen-rich scapolite and biotite: Implications for metamorphic fluid-rock interaction. American Mineralogist, 74, 721–37. Morogan V, Wooley AR (1988) Fenitization in the Alno¨ Complex, Sweden: distribution, mineralogy and genesis. Contributions to Mineralogy and Petrology, 100, 169–82. Munoz JL (1981) Chloride-hydroxyl exchange in biotite and estimation of relative HCl ⁄ HF activities in hydrothermal fluids. Economic Geology, 76, 2212–21. Newton RC, Manning CE (2000) Quartz solubility in H2O-NaCl and H2O-CO2 solutions at deep crust-upper mantle pressures and temperatures: 2-15 kbar and 500-900 C. Geochimica et Cosmochemica Acta, 64, 2993–3005. Newton RC, Manning CE (2002a) Solubility of enstatite + forsterite in H2O at deep crust ⁄ upper mantle conditions: 4 to 15 kbar and 700 to 900 C. Geochimica et Cosmochimica Acta, 66, 4165–76. Newton RC, Manning CE (2002b) Experimental determination of calcite solubility in H2O-NaCl solutions at deep crust ⁄ upper mantle temperatures and pressures: Implications for metaso-
matic processes in shear zones. American Mineralogist, 87, 1401–9. Newton RC, Manning CE (2003) Activity coefficient and polymerization of aqueous silica at 800 C, 12 kbar, from solubility measurements on SiO2-buffering mineral assemblages. Contributions to Mineralogy and Petrology, 146, 135–43. Newton RC, Manning CE (2005) Solubility of anhydrite, CaSO4, in NaCl-H2O solutions at high pressures and temperatures; applications to fluid-rock interaction. Journal of Petrology, 46, 701–16. Newton RC, Manning CE (2006) Solubilities of corundum, wollastonite and quartz in H2O-NaCl solutions at 800 C and 10 kbar: Interaction of simple minerals with brines at high pressure and temperature. Geochimica et Cosmochimica Acta, 70, 5571– 82. Newton RC, Manning CE (2007) Solubility of grossular, Ca3Al2Si3O12, in H2O-NaCl solutions at 800 C and 10 kbar, and the stability of garnet in the system CaSiO3-Al2O3-H2O-NaCl. Geochimica et Cosmochimica Acta, 71, 5191–202. Newton RC, Manning CE (2008a) Thermodynamics of SiO2H2O fluid near the upper critical end point from quartz solubility measurements at 10 kbar. Earth and Planetary Science Letters, 274, 241–9. Newton RC, Manning CE (2008b) Solubility of corundum in the system Al2O3-SiO2-H2O-NaCl at 800 C and 10 kbar. Chemical Geology, 249, 250–61. Newton RC, Manning CE (2009) Hydration state and activity of silica in H2O-CO2 fluids at high temperatures and pressures. American Mineralogist, 94, 1287–90. Newton RC, Smith JV, Windley BF (1980) Carbonic metamorphism, granulites, and crustal growth. Nature, 288, 45–50. Newton RC, Aranovich LY, Hansen EC, Vandenheuvel BA (1998) Hypersaline fluids in Precambrian deep-crustal metamorphism. Precambrian Research, 91, 41–63. Newton RC, Manning CE, Hanchar JM, Colasanti CV (2009) Free energy of formation of zircon based on solubility measurements at high temperature and pressure. American Mineralogist, doi: 10.2138/am.2010.3213. Nijland TG (1993) Halogen geochemistry of fluid during amphibolite-granulite metamorphism as indicated by apatite and hydrous silicates in basic rocks from the Bamble sector, south Norway. Lithos, 30, 167–89. Novgorodov PG (1977) On the solubility of quartz in H2O + CO2 and H2O + NaCl at 700 C and 1.5 kb pressure. Geochemistry International, 14, 191–3. Pallister JS, Hoblitt RP, Reyes AG (1992) A basalt trigger for the 1991 eruptions of Pinatubo volcano? Nature, 356, 426–8. Pan Y, Fleet ME (1996) Rare earth element mobility during prograde granulite facies metamorphism: significance of fluorine. Contributions to Mineralogy and Petrology, 123, 251–62. Philippot P, Selverstone J (1991) Trace element-rich brines in eclogitic veins: implications for fluid composition and transport during subduction. Contributions to Mineralogy and Petrology, 106, 417–30. Quist AS, Marshall WL (1968) Electrical conductances of aqueous sodium chloride solutions from 0 to 800 C and at pressures up to 4000 bars. Journal of Physical Chemistry, 72, 684–703. Ryabchikov ID, Hamilton DL (1971) Possibility of separation of concentrated chloride solutions in course of acid magma crystallization. Doklady Akademii Nauk SSSR, 197, 933–7. Sandiford M, Neall FB, Powell R (1987) Metamorphic evolution of aluminous granulites from Labwor Hills, Uganda. Contributions to Mineralogy and Petrology, 95, 217–25.
72 R. C. NEWTON & C. E. MANNING Setche´now M (1892) Action de l’acide carbonique sur les solutions des a` sels acids forts. Annales de Chimie et de Physique, 25, 225–70. Shibue Y (1996) Empirical expressions of quartz solubility in H2O, H2O + CO2, and H2O + NaCl fluids. Geochemical Journal, 30, 339–54. Shmulovich KI, Graham CM, Yardley BWD (2001) Quartz, albite and diopside solubilities in H2O-NaCl fluids at 0.50.9 GPa. Contributions to Mineralogy and Petrology, 141, 95– 108. Shmulovich KI, Yardley BWD, Graham CM (2006) The solubility of quartz in crustal fluids: experiments and general equations for salt solutions and H2O-CO2 mixtures at 400-800 C and 0.10.9 GPa. Geofluids, 7, 1–17. Smit CA, Van Reenan DD (1997) Deep crustal shear zones, highgrade tectonites, and associated metasomatic alteration in the Limpopo Belt, South Africa. Journal of Geology, 105, 37–58. Sommerfeld RA (1967) Quartz solution reaction: 400–500 C, 1000 bars. Journal of Geophysical Research, 72, 4253–57. Sterner SM, Bodnar RJ (1989) Synthetic fluid inclusions: VII. Re-equilibration of fluid inclusions in quartz during laboratory simulated metamorphic burial and uplift. Journal of Metamorphic Geology, 7, 243–60. Tanaka M, Takahashi K (1999) The identification of chemical species of silica in sodium hydroxide, potassium hydroxide and sodium chloride solutions by FAB-MS. Analytical Sciences, 15, 1241–50. Touret JLR (1971) Le facie`s granulite en Norwe`ge Me´ridionale. Lithos, 4, 239–249. Touret JLR (1985) Fluid regime in Southern Norway: the record of fluid inclusions. In: The Deep Proterozoic Crust in the North Atlantic Provinces (eds Tobi AC, Touret JLR), pp. 517–49. Reidel, Dordrecht. Touret JLR, Huizenga JM (1999) Precambrian intraplate magmatism: high temperature, low pressure crustal granulites. Journal of African Earth Sciences, 28, 367–82. Tropper P, Manning CE (2004) Paragonite stability at 700 C in the presence of H2O-NaCl fluids: constraints on H2O activity and implications for high-pressure metamorphism. Contributions to Mineralogy and Petrology, 147, 740–9. Tropper P, Manning CE (2007) The solubility of fluorite in H2O and H2O-NaCl at 0.5-2 GPa, 600–1000 C. Chemical Geology, 242, 299–306. Van den Berg R, Huizenga JM (2001) Fluids in granulites of the Southern Marginal Zone of the Limpopo Belt, South Africa. Contributions to Mineralogy and Petrology, 141, 529–45.
Walther JV (2001) Experimental determination and analysis of the solubility of corundum in 0.1 and 0.5m NaCl solutions between 400 and 600 C from 0.5 to 2.0 kbar. Geochimica et Cosmochimica Acta, 65, 2843–57. Walther JV, Orville PM (1983) The extraction-quench technique for determination of the thermodynamic properties of solute complexes: applications to quartz solubility in fluid mixtures. American Mineralogist, 68, 731–41. Wannamaker PE, Caldwell TG, Doerner WM, Jiracek GR (2004) Fault zone fluids and seismicity in compressional and extensional environments inferred from electrical conductivity: the New Zealand southern Alps and US Great Basin. Earth, Planets and Space, 56, 1171–76. Watson EB, Brenan JM (1987) Fluids in the lithosphere: 1. Experimentally-determined wetting characteristics of CO2-H2O fluids and their implications for fluid transport, host-rock physical properties and fluid inclusion formation. Earth and Planetary Science Letters, 85, 497–515. Webster JD, Kinzler RJ, Mathez EA (1999) Chloride and water solubility in basalt and andesite liquids and implications for magmatic degassing. Geochimica et Cosmochimica Acta, 63, 729–38. Weill DF, Fyfe WS (1964) The solubility of quartz in H2O in the range 1000–4000 bars and 400–500 C. Geochimica et Cosmochimica Acta, 28, 1243–55. Wickham SM, Janardhan AS, Stern RJ (1994) Regional carbonate alteration of the crust by mantle-derived magmatic fluids, Tamil-Nadu, south India. Journal of Geology, 102, 379–98. Wykes JL, Newton RC, Manning CE (2008) Solubility of andradite, Ca3Fe2Si3O12, in a 10 mol% NaCl solution at 800 C and 10 kbar: Implications for the metasomatic origin of grandite garnet in calc-silicate granulites. American Mineralogist, 93, 886–92. Xie Z, Walther JV (1993) Quartz solubilities in NaCl solutions with and without wollastonite at elevated temperatures and pressures. Geochimica et Cosmochimica Acta, 57, 1947–55. Yardley BWD, Graham JT (2002) The origins of salinity in metamorphic fluids. Geofluids, 2, 249–56. Yardley BWD, Valley JW (1997) The petrologic case for a dry lower crust. Journal of Geophysical Research, 102, 173–85. Zotov N, Keppler H (2000) In-situ Raman spectra of dissolved silica species in aqueous fluids to 900 C and 14 kbar. American Mineralogist, 85, 600–4. Zotov N, Keppler H (2002) Silica speciation in aqueous fluids at high pressures and high temperatures. Chemical Geology, 184, 71–82.
Potential of palaeofluid analysis for understanding oil charge history J. PARNELL Department of Geology and Petroleum Geology, University of Aberdeen, Aberdeen, UK
ABSTRACT Fluid inclusion data, particularly the distribution of hydrocarbon fluid inclusions and their chemistry, can provide insights into oil charge in a petroleum-prospective region. Examples from the UK Atlantic margin show how we can understand thermal regime, timing and chemistry of oil charge. Data from the UK Atlantic margin based on fluid inclusion temperature profiles shows anomalously high temperatures which are highest at the top of the Triassic–Eocene sequence. This is interpreted as a product of hot fluid flow, probably reflecting hydrothermal activity related to intrusion of sills at depth. The preservation of high temperatures also implies rapid migration from depth through fracture systems. Ar–Ar analysis of oil-bearing K-feldspar cements, and petrographic studies of oil inclusion distribution help delimit timing and migration pathways for the hot fluid charge and later fluid migration events. Coupled with compositional data for oils measured destructively (organic geochemistry) or non-destructively (fluorescence), these approaches allow the development of oil charge histories based on real data rather than theoretical modelling. Key words: Atlantic margin, hydrocarbons, fluid inclusions, sedimentary basins, thermal anomalies Received 10 June 2009; accepted 7 September 2009 Corresponding author: J. Parnell, Department of Geology and Petroleum Geology, University of Aberdeen, Aberdeen AB24 3UE, UK. Email:
[email protected]. Tel: 01224 273464. Fax: 01224 272785. Geofluids (2010) 10, 73–82
INTRODUCTION Hydrocarbon fluid inclusions occur commonly in sedimentary basins, especially where there is a proven petroleum system. They have found value in hydrocarbon exploration, as an indicator of former hydrocarbon conduits or reservoirs (e.g. Munz et al. 1999; Parnell et al. 2001), identification of palaeo-oil-water contacts (e.g. Lisk et al. 2002), timing of oil charge relative to cementation (e.g. Neilson et al. 1998) and as a source of relatively fresh oil for geochemical typing and correlation (e.g. Isaksen et al. 1998; George et al. 2007). They yield information on thermal history through homogenization temperature data, which is commonly integrated with burial (temperature) history modelling to infer the timing of their entrapment and by inference the timing of oil migration through carrier bed or into reservoir (e.g. Nedkvitne et al. 1993). Oil inclusions have also fuelled academic debate about whether diagenesis, especially cementation, continues after emplacement of oil into reservoir porosity (Worden et al. 1998).
These are, typically, the limits to which oil inclusions are used in an exploration programme. Often they are not used at all. However over the period 1995–2008, when hydrocarbon exploration in the UK has included the difficult deep waters of the Atlantic margin (Fig. 1), the wider potential of oil inclusions has been explored in an effort to understand the oil charge history of this region. In this review, several examples are used to show the diversity of aspects through careful study of the distribution of oil fluid inclusions and their chemistry. These include examples which inform us about the thermal regime of oil migration, dating of oil migration, the pathways of oil migration and evidence for the evolution of oil chemistry. The geology of the Atlantic margin region has been described by numerous authors, including Dean et al. (1999), Lamers & Carmichael (1999) and Scotchman et al. (2006). Most of the region is floored by half-graben rift sequences of Permo-Triassic continental clastic sediments deposited upon crystalline Caledonian basement. A thin sequence of Early Jurassic sediments reflects a transition to a marine environment. Renewed rifting occurred
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
74 J. PARNELL
8°W
6°
4°
2°
0°
62°N
Atlantic Ocean
Tertiary lavas
S OE FAR
61°
A
D AN TL E H
SIN BA Quad 206/5 Victory Clair
Devonian and younger sediments
A'
Shetland Islands
Foinaven, Quads 204/19+24
60°
Devonian Orkney - Shetland Platform
Orkney Islands
59°
Dunnet Head
Precambrian + Palaeozoic cover
in the Late Jurassic–Early Cretaceous, resulting in marine sedimentation, including deposition of organic-rich shales and localized sandstone bodies. This sedimentation was followed by a Cretaceous (Turonian) episode of inversion and erosion. Late Cretaceous rifting caused major normal faulting and deposition of thick marine shales. Further localized inversion along some of the faults occurred in the Early Paleocene, but generally deep marine sedimentation persisted through the Paleocene. Paleocene–Eocene volcanic activity was followed by thin paralic to continental sediments of Early Eocene age, in turn followed by thick Eocene–Oligocene marine clastic sedimentation. Several subsequent uplift events are recorded in the Cenozoic, after which a veneer of Pliocene and younger sediments formed the rocks currently on the seafloor. The source rocks for hydrocarbon discoveries are assumed to be of Jurassic age and currently producing fields are in Paleocene sandstone reservoirs.
PALAEOFLUID ANALYSIS Hot fluid pulses There is increasing evidence for the occurrence of hot fluid pulses during the history of basins around the Atlantic
Limit of early Tertiary lavas Limit of FaeroeShetland sill complex
Fig. 1. Map of UK Atlantic margin, showing location of wells West of Shetland used in fluid inclusion study. Note close proximity of eastern limits of Tertiary extrusive and intrusive igneous rocks (adapted from Naylor et al. 1999). Section line A to A¢ indicates location of Fig. 5.
margins and elsewhere. The fluids may have exceeded 200C, well above ambient rock temperatures in most basins. Anomalous fluid temperatures are particularly evident from fluid inclusion analyses and from apatite fission track data. Fluid inclusion temperatures represent minimum trapping temperatures during mineral precipitation rather than present day temperatures. However in sedimentary basins, notably in oilfields, they are commonly taken as a proxy for maximum burial temperatures (see e.g. Osborne & Haszeldine 1993) and display trends of increasing temperature downward which conform to the present geothermal gradient. Data from them is thus used to reach important conclusions about subsurface temperatures, the thermal gradient in the upper crust and the likelihood of oil ⁄ gas generation. The temperatures are used to calculate the timing and duration of cementation events through integration with burial (time-depth ⁄ temperature) histories (e.g. Osborne & Swarbrick 1999). The incorporation of data related to hot fluid pulses into such modelling could lead to significant and costly errors. Evidence for the passage of hot fluids through a section could include: (1) Anomalously high temperatures The simplest evidence for hot fluid pulses is the deduction of temperatures hotter than expected from the
Palaeofluid analysis and oil charge history 75
(2) Irregular thermal profiles Thermal profiles are considered irregular when they do not simply increase with depth and may even show upward increases in temperature. Many examples of inverted thermal profiles have been described. Indeed, Ziagos & Blackwell (1986) state that ‘temperature inversions are ubiquitous in geothermal systems’. The inversions generally occur beneath thermal maxima in permeable aquifers (e.g. Bodvarsson 1973; Ziagos & Blackwell 1986 and references therein) (Fig. 3) and reversals in palaeothermal gradient can become preserved in the geological record, for example in vitrinite reflectance values (e.g. Goodhue & Clayton 1999) and apatite fission track profiles (Duddy et al. 1998).
0
0
5
10
Temperature (C°) 15 20
25
Well 2
Depth (m)
200
300
400
Aquifer
Well 1
500
600
Fig. 3. Inverted temperature-depth profiles related to hot fluid flow along aquifers, Oregon (adapted from Ziagos & Blackwell 1986).
(3) Short-lived thermal events Transient events are deduced from their recognition in fluid inclusions, but absence in kinetically-dependent parameters such as vitrinite reflectance. For example, fluid inclusion studies on cements in the Siluro–Devonian Hedberg Group, central Appalachians, yield a mean homogenization temperature of 203C, with an estimated pressure correction of an additional 50C (Dorobek 1989). These temperatures greatly exceed the maximum values inferred from conodont alteration indices, vitrinite reflectance, illite crystallinity or burial estimates, which were of the order of 150C. Dorobek (1989) estimated that the unaffected vitrinite reflectance data was consistent with heating at 300C for less than 0.1 Myr. O’Brien et al. (1996) concluded a transient heating event from fission track and fluid inclusion data in the Timor Sea, and deduced from modelling of a hot aquifer that flow had lasted for durations of less than 1 Myr. The consensus of these and other attempts to model the duration of hot fluid pulses from palaeotemperature
2
Depth (km)
1
Fig. 2. Conceptual model for fluid flow in oil fields, Railroad Valley, Nevada, showing convection of fluids including oil, partly fault-controlled and perturbation of isotherms. Hot fluids precipitate minerals which enhance sealing capacity of unconformity (after Hulen et al. 1994).
0
–1
Oil
er ne ol d oce -Mi a nd e n c i o oce so z Hol Me l ow df l ui f d re fe r In 50°C
2 km
100°C
–2
30
Aquifer 100
nt ie ad gr al rm No
established burial history. Most sedimentary basins have geothermal gradients in the range 20–30C km)1 and yield fluid inclusion temperatures in the range 80–140C from rocks buried in the range 2–5 km (e.g. Walderhaug 1994). Much of the temperature data measured is in this range, partially reflecting the common precipitation of quartz under these conditions. However in some cases the depth at which this occurred is much less than predicted from the geothermal gradient. For example, a study of oilfields in Palaeozoic rocks in eastern Nevada reported both present day reservoir temperatures (116– 122C) and fluid inclusion temperatures (100–150C) that are anomalously high for depths of 1.1–1.6 km (Hulen et al. 1994). Geothermal gradients of up to 118C km)1 are recorded. As the system is thought to be very young (max. 2.5 Ma), the inclusions also reflect a high gradient. The data are interpreted to reflect a geothermal system in which meteoric waters are heated at depth and returned rapidly up faults (Fig. 2). Hot oil and hot oil inclusions lead Hulen et al. (1994) to describe the system as a ‘geothermal oil reservoir’. The importance of high fault permeability to episodic fluid circulation is also emphasized by Coolbaugh et al. (2005) and Kennedy & van Soest (2007).
76 J. PARNELL parameters (e.g. Lampe et al. 2001; Middleton et al. 2001; Person et al. 2008) is that they last for less than 1 Myr and generally less than 0.1 Myr. Hot fluids at the UK Atlantic margin Hot fluids have been identified in the palaeo-thermal record of numerous wells and outcrops on the UK Atlantic margin from fluid inclusion and fission track data (Parnell et al. 1999; Middleton et al. 2001; Wycherley et al. 2003; Parnell & Middleton 2009), including wells in or adjacent to producing oil fields West of Shetland. Fluid inclusion data was recorded from 12 samples in 204 ⁄ 19–1, Foinaven Field, in quartz cements and ⁄ or healed microfractures through quartz grains which can be shown, petrographically, to postdate deposition of the sandstone (Parnell et al. 2005). Both aqueous and oil inclusions occur. Mean values for primary and early secondary populations of aqueous inclusions are plotted against sample depth in Fig. 4. The same plot also shows vitrinite reflectance data and its equivalence to temperatures, as calculated using the EASYRO programme. The plot shows that vitrinite reflectance increases with depth as expected. The fluid inclusion temperatures are consistent with the reflectance-derived temperatures in the pre-Tertiary section, but in the Paleocene–Eocene they exhibit higher values progressively upwards to the shallowest sample in the Eocene. At some levels a range of values is found, but they include anomalously high temperatures. Consequently, there is an increasing discrepancy between the two estimates of palaeotemperature upward through the Paleocene–Eocene, which at the top exceeds 100C.
T Calculated from Ro Early secondary inclusions Primary inclusions
1000
Tertiary
Depth (m)
2000
3000
4000
pre-Tertiary
5000
60
80
100
120
140
160
180
T (°C) (Inclusion Th, T calc. from Ro) Fig. 4. Mean homogenization temperatures from fluid inclusion populations in sandstones and temperatures derived from vitrinite reflectance data using the EASYRO programme, well 204 ⁄ 19–1. Reflectance-derived temperatures show expected increase with depth, but highest fluid inclusion temperatures are at shallowest levels.
The discrepancy between the temperatures predicted by the vitrinite reflectance model and those observed from the fluid inclusions can be removed by using a tectonicallyderived heat flow history (Scotchman et al. 2006) and a short duration (0.01 Ma) injection of high temperature fluids into discrete sand layers in the Paleocene sandstones. The tectonically-derived heat flow includes a short duration increase in heat flow (56–54 Ma) representing the increase in heat flow associated with the Balder volcanic event, a major episode in the Tertiary magmatic activity which lasted from end-Cretaceous to Eocene and represents a distinct phase in the opening of the North Atlantic (Ritchie et al. 1999). The injection of high temperature fluids is an additional source of heat above that conferred to the sediment column by the Balder event. Anomalous heat flows around basement highs are consistent with up-dip channelling of hot fluids, in the West of Shetland region (Iliffe et al. 1999). The hot fluids are envisaged as being focussed along major extensional faults (Fig. 5, after Scotchman et al. 2006). Some wells or stratigraphic levels in the region do not yield anomalous fluid inclusion data (Parnell et al. 1999), which can be interpreted as reflecting lower proximity and accessibility to fault conduits. The maximum temperatures of about 200C suggest a temperature excess of at least 100C above that expected from conductive heating, implying vertical migration of 3 km or more (Fig. 5). Similar implication of fluid circulation from a deep reservoir is described by Person et al. (2008). A consequence of up-dip migration of hot fluids is that not only may low-permeability units be by-passed, but aquifers low in the succession may also be isolated from the hydrothermal system. On the European margin West of Shetland, maximum fluid inclusion temperatures occur at the top of the Triassic–Eocene sequence, implying that hot fluid transport was limited to the higher aquifers that have the most continuous geometry from deep basin to shallow levels. We may expect to find comparable patterns of fluid flow at other passive margin sites. The high temperatures recorded from Paleocene–Eocene sections in 204 ⁄ 19–1 and other wells (Wycherley et al. 2003) include measurements made from oil inclusions. Oil was being transported in the hot fluids and was itself at anomalously high temperatures, evidenced by oil inclusions entrapped at up to 200C or hotter on the Atlantic margin and elsewhere (Archer et al. 2004). Although these temperatures are at the notional limit of liquid hydrocarbon stability at the surface, stability is not a problem under the high pressures of a confined system. However, the occurrence of these hot oils is an indicator of the migration pathways that the oil is following and thus a pointer towards the oil kitchen. It is also a direct indication that petroleum geologists and petroleum geochemists should consider the potential role of hot fluids.
Palaeofluid analysis and oil charge history 77
A
A′
W
E
NW 204/19-1
secs TWT
1.0
2.0
SE
Rapid flow along faults bypasses lower sandstones
Hot fluids (200+°C) precipitate quartz cements
Top Miocene
Base Tertiary
3.0
Base Cretaceous Top Basement
4.0
Fig. 5. Cross-section across Foinaven region (see Fig. 1 for location), showing modelled up-dip migration of hot fluids. Depths in two-way travel time (TWT). Middle Jurassic and Kimmeridge Clay Formation are source rock units, at depths of about 8 km. Fluid flow along fault conduits allows vertical migration of about 3–5 km. Redrawn from Scotchman et al. (2006).
M. Jurassic Source Kimm. Clay Fm.
5.0
Westray ridge Sjurdur ridge
6.0 Very hot basinal fluids (non-volcanic) heated by regional Balder volcanic heat flows
Foinaven sub-basin
Sandstone Fluid movement
Fractured reservoirs Many oilfields contain fractured reservoirs, in which the reservoir unit is cut by fracture systems including faults (Lonergan et al. 2007). In some cases a rock functions as a reservoir ‘only’ because it has fracture porosity. Depending on the nature of the fractures (connectivity, porosity, stratigraphic extent, timing of formation), fractured reservoirs raise specific issues, including the possibilities of reservoir leakage, heterogeneity and compartmentalization. Insights into these aspects can be gained by the distribution of oil inclusions. A major cause of compartmentalization of sandstone reservoirs is the development of deformation bands, i.e. granulation seams which form by cataclasis in porous sandstones (Antonelli et al. 1994; Shipton & Cowie 2003). Deformation bands have low porosity and permeability, so act as barriers which can isolate compartments and prevent them from being charged by oil or alternatively prevent charged compartments from being drained (Parnell et al. 2004; Sample et al. 2006). Good examples can be seen in Upper Old Red Sandstone aeolian and fluvial sandstones in Dunnet Head, Caithness (Fig. 1) that are likely analogues for the Devono–Carboniferous sandstones of the Clair Field on the Atlantic margin (Parnell et al. 2004). Deformation bands in the Caithness sandstones are mineralized by quartz cement. The bands separate and isolate compartments, some of which contain oil residues and some of which do not (Fig. 6). Quartz cement in the deformation bands contains only aqueous inclusions, whereas quartz overgrowths within the sandstone additionally contain hydrocarbon inclusions (Parnell et al. 2004). These observations indicate that the sandstone was compartmentalized by the bands before oil charge. In the Clair Field itself, fracturing is an important aspect of the reservoirs and hydrocarbon distribution (Coney
0
1
2
Cm
Fig. 6. Sandstone compartmentalized by deformation bands, most containing oil residue (grey), but some free of oil due to pre-charge isolation of bands. Upper Old Red Sandstone, Dunnet Head, Caithness.
et al. 1993; Smith & McGarrity 2001). Deformation bands in the Clair Field sandstones can be placed within a detailed paragenetic sequence of cements. By examination of the distribution of oil inclusions in cements before and after the formation of the deformation bands, the influence of these bands on oil charge can be assessed. The predominant cement is calcite, for which two distinct episodes of precipitation can be distinguished by their cathodoluminescence characteristics, one before and one after the deformation bands. In this case, oil occurs in both preband and postband cements, so the bands have not prevented oil charge. On the contrary, the bands are associated with coeval calcite veins which contain oil inclusions and appear to represent dilatant fractures that allowed oil migration (Baron et al. 2008). Oil-bearing calcite veins also predate deformation bands, showing that fractures had long-term
78 J. PARNELL 130
Oil inclusion in cement and veins
150
Oil inclusion in veins
140
Th 20
130 120
10
110 100
Temperature (°C)
Porosity (%)
30
(ce
me
nt)
Th
(ve
ins
)
Homogenization temperature (°C)
40
Oil charge in hot fluid Cement I n = 10 oil and aqueous
120 110
Uplift, trap breach
100 Cement II aqueous
90
n = 10
80 Equilibration of fluid and rock
70 60 50 40 45.0
Cement III aqueous
50.0
55.0
60.0
n = 11
65.0
70.0
75.0
80.0
85.0
90.0
Ar age (Ma) 0
Calcite I
Calcite II
Fig. 7. Two-stage history of calcite cementation from Clair Field, separated by episode of deformation, including deformation bands. Porosity values show range of porosities measured in four samples at commencement of each cementation stage (two stages distinguished by cathodoluminescence). Temperatures are mean homogenization temperatures measured from fluid inclusions in calcite in cements and veins at each stage, and show decreases from ‘hot’ to ‘normal’ values between stages I and II. Oil inclusions occur in both cements and veins in stage I, but only veins in stage II, indicating change in migration style. All data from Baron et al. (2008).
importance to oil migration. However, there is a significant contrast in distribution of inclusions in pore-filling calcite cements precipitated before and after the bands. Oil inclusions occur in the preband calcite, but not in the postband calcite cement. Microscopy images show that the porosity had decreased from a mean value of 29% during preband cementation, to 10% during postband cementation (Baron et al. 2008) (Fig. 7). The mechanism of oil migration had changed from mixed intergranular porosity ⁄ fractures preband to solely fractures postband. Through this type of study, oil inclusions can be valuable in building a model of how oil charge evolves over time. Dating The dating of oil charge is one of the most critical aspects of predicting a viable petroleum prospect. The timing of oil charge needs to be appropriate relative to other events such as trap formation. In the past, only indirect dating of oil charge was possible, by using the age of minerals precipitated before or after the oil charge, predictions of oil generation and migration from heat flow and subsidence models or using fluid inclusion entrapment temperatures combined with burial (thermal) history plots. However recently it has been shown that K-feldspar cements within sandstones can yield high-resolution Ar–Ar ages, which date any fluids trapped as inclusions within the same cement (Mark et al. 2005). Where oil inclusions occur in the cements, a date for oil occurrence within the sandstone is obtained, which constrains the timing of oil charge (Mark et al. 2005, 2008). An example from Cretaceous
Fig. 8. Cross-plot of temperature (fluid inclusions) and time (Ar–Ar analysis) for three stages of K-feldspar cement, Victory Field. Note temperatures decrease through successive stages and oil inclusions only present in first stage. Values of 50C in third stage are conventional representation of monophase fluid inclusions. Data from Mark et al. (2005).
rocks in the Victory gas field on the Atlantic margin distinguished three successive episodes of K-feldspar cementation and fluid entrapment (Mark et al. 2005). Inclusion studies show that the three episodes represent progressively decreasing fluid temperature and only the first episode contains oil (Fig. 8). The oil-bearing episode yields an Ar–Ar age that is consistent with (slightly postdates) the predicted age of oil generation from the source rock and is a very plausible time for oil in the reservoir. The fluid was relatively hot, reflecting oil migration from depth. The second cementation episode dates a time when oil had been lost from the reservoir, possibly following an uplift event that allowed breaching of the trap. The third cementation episode represents cool fluids that had equilibrated with the shallow sub-surface. The dating of oil charge by this method also has the potential to indicate sources of oil otherwise unrecognized, by providing a date inconsistent with the generation of oil from previously assumed sources. This was demonstrated by Ar–Ar dating of K-feldspar cements in the Clair Field, West of Shetland (Mark et al. 2008). This part of the Atlantic margin is unusual in having oil reserves in Palaeozoic (Devono–Carboniferous) rocks (Allen & Mange-Rajetsky 1992), but the oil has been attributed to the same Jurassic source rocks that feed the petroleum system elsewhere in the region (Rooney et al. 1998; Scotchman et al. 1998). However, K-feldspar cements containing oil inclusions in Palaeozoic sandstones yield dates of late Permian–early Triassic age. This implies source rocks of Permian or earlier age. Source rocks of this age had not been proven previously, although the possibility of mixing of another source with the Jurassic had been suggested (Rooney et al. 1998; Scotchman et al. 1998). In the Clair Field, this older source of oil could be detected because the reservoir rocks predate the
Palaeofluid analysis and oil charge history 79
Oil composition Inclusions oils can be characterized in both destructive and non-destructive procedures. Crushing a rock sample can release oil for analysis by organic geochemistry, while spectroscopic study can allow characterization of physical or chemical properties. Chemistry Crushing of a rock to yield inclusion oils for analysis has seen use in several studies (e.g. Jones & Macleod 2000; George et al. 2007). It does, however, require considerable caution, to avoid mixing of oil residues in the pore spaces (i.e. the current reservoir oil) with that liberated from inclusions. It also requires detailed petrography to check if there is only one population of oil inclusions, otherwise the oil obtained may be a hybrid of multiple generations. Nevertheless, such oils are valuable samples and commonly fresher than what is now in the pore space, hence giving analyses that are more suitable for correlation with other oils or source rocks. In some cases, the inclusion oils will simply be a fresher sample of what is in the pore spaces, but often they represent oil generated earlier in the maturation history, so contribute to an understanding of how oil composition has evolved through time. The chemistry of oils liberated from Atlantic margin sandstones has been reported by Parnell et al. (2005). These have varied in quality, reflecting the fact that absolute quantities of oil are very low in some cases. However, several samples have yielded oils from which quantitative parameters can be calculated, which allows them to be compared with the oils in pore spaces. This is illustrated by pairs of samples from three parts of the West of Shetland region, where in each case the inclusion oil and pore oil have a distinct composition (Fig. 9). The use of two parameters which measure source-based composition and subsequent degradation highlights different histories in different oil fields. In two cases (Clair and Quads 204 ⁄ 19 + 24), the pore oil is more biodegraded than the inclusion oil, and also has a more ‘terrestrial’ signature. This is consistent with a shift from a marine Kimmeridgian (Upper Jurassic) source to addition of a mid-Jurassic source, as proposed by Scotchman et al. (1998). In contrast, data from Quad 206 ⁄ 5 shows (early) entrapment of a degraded oil in inclusions, followed by a charge of a Kimmeridgian-like oil.
204/24 Oil
1.5 Terrestriality
n-alkane C27/C17
time of oil generation. It is likely that most oil from this source rock had been lost by the time the Jurassic-based petroleum system was deposited. Nevertheless, this newly identified source raises the possibility that deep plays may exist in regions of the Atlantic margin that had not previously been considered prospective.
1.0
Clair oil
Biodegradation
206/5 Inclusion
0.5
Clair inclusion 0
0.1
206/5 Oil 204/19 Inclusion
0.2 0.3 Isoprenoid/n-alkane (C14 - C16 )
0.4
0.5
Fig. 9. Cross-plot of n-alkane C27:C17 ratio (increase with waxiness, ‘terrestriality’) against isoprenoids ⁄ n-alkane ratio for C14:C16 (increases with degradation). Comparison of chemistry of inclusion oil and present-day pore oil in Quads 204 ⁄ 19 + 24 and Clair Field show that in both cases pore oil is more terrestrial and more degraded, consistent with a first charge of Kimmeridge-sourced oil followed by addition of a waxier oil. Comparison in Quad 206 ⁄ 5 samples shows a different history, in which Kimmeridgesourced oil arrived later. Partly adapted from Parnell et al. (2005).
Fluorescence Several groups have attempted to characterize the chemistry of inclusion oils by in situ spectroscopic study of individual inclusions. This includes application of confocal scanning laser microscopy (Pironon et al. 1998), Fourier transform-infrared spectroscopy (Pironon et al. 2001) and time of flight-secondary ion mass spectrometry (Li & Parnell 2003). Some of these techniques require relatively large inclusions and at present are unlikely to see widespread use. However, techniques based on fluorescence characteristics can be applied quite readily to multiple inclusions in a single sample and offer potential for discriminating oils of different type (Kihle 1995; Blanchet et al. 2003). Initially, it was believed that fluorescence colour under ultraviolet light could be a semi-quantitative guide to oil gravity (API) (Bodnar 1990; Stasuik & Snowdon 1997). It is, however, evident that the fluorescence colour depends upon several factors, including the type of organic matter and oil fractionation during entrapment and oils of similar composition can yield different fluorescence (George et al. 2001). At best, fluorescence colour serves to compare oils known to have been derived from the same source rock. A more sophisticated characterization is made by measuring fluorescence lifetime, in which the decay time of fluorescence emission is dependent upon the wavelengths of excitation and emission and the composition of the oil (Ryder et al. 2002, 2004; Owens et al. 2008). Reference to a data base of fluorescence lifetimes for oils of known composition allows inferences about the compositions of oils inside inclusions. Two populations of oil inclusions from the Clair Field, initially distinguished by
80 J. PARNELL for their skilled technical support. Vitrinite reflectance data was kindly provided by BP. The manuscript benefited from constructive reviews by M. Person and an anonymous reviewer.
5
Average lifetime (ns)
4
REFERENCES 3
2 Blue fluorescing inclusion Yellow fluorescing inclusion 1 450
500
550 600 Wavelength (nm)
650
700
Fig. 10. Plot of average fluorescence lifetime against emission wavelength, measured with a 405 nm excitation of blue- and yellow-fluorescing oil inclusions in secondary trail through quartz grain, well 208 ⁄ 8–8, depth 1803.1 m, Clair Field. The wavelength dependence patterns follow those typical of mid-maturity crude oils, but with higher API in case of the bluefluorescing inclusion. (after Baron et al. 2008).
yellow (earlier) and blue (later) fluorescence under ultraviolet light, yield distinct fluorescence lifetimes (Fig. 10, after Baron et al. 2008), showing that oil composition was evolving through the paragenesis.
CONCLUSIONS The studies summarized above show that there are numerous approaches that can be taken to obtain useful information from occurrences of hydrocarbon fluid inclusions. They have used the patterns of homogenization temperature data between successive levels, dating of cements that host the inclusions, fluorescence data from inclusion oils, organic geochemical data from inclusion oils and spatial variations in the occurrences of inclusions. This data has helped to understand the thermal regime of fluid movement including oil charge, timing of oil charge, characterization and distribution of oils in a multistage charge history and evidence of reservoir compartmentation. All of this information was obtained sufficiently readily that it makes realistic input to a petroleum exploration programme. More generally, an awareness of the potential of fluid inclusions will allow explorationists to make palaeofluid analysis a valuable part of their strategy.
ACKNOWLEDGEMENTS This review draws upon projects involving numerous collaborators, including M. Baron, D. Mark, D. Middleton, G. Watt, H. Wycherley, M. Feely and P. Green. I acknowledge the cartographic staff at the University of Aberdeen
Allen PA, Mange-Rajetsky MA (1992) Devonian-Carboniferous sedimentary evolution of the Clair area, offshore northwestern UK; impact of changing provenance. Marine and Petroleum Geology, 9, 29–52. Antonelli MA, Aydin A, Pollard DD (1994) Microstructure of deformation bands in porous sandstones at Arches National Park, Utah. Journal of Structural Geology, 16, 941–59. Archer SG, Wycherley HL, Watt GR, Baron ML, Parnell J, Chen H (2004) Evidence for focussed hot fluid flow within the Britannia Field, offshore Scotland, UK. Basin Research, 16, 377–95. Baron M, Parnell J, Mark D, Carr A, Przyjalgowski M, Feely M (2008) Evolution of hydrocarbon migration style in a fractured reservoir deduced from fluid inclusion data, Clair Field, west of Shetland, UK. Marine and Petroleum Geology, 25, 153–72. Blanchet A, Pagel M, Walgenwitz F, Lopez A (2003) Microspectrofluorimetric and microthermometric evidence for variability in hydrocarbon fluid inclusions in quartz overgrowths: implications for inclusion trapping in the Alwyn North field, North Sea. Organic Geochemistry, 34, 1477–90. Bodnar RJ (1990) Petroleum migration in the Miocene Monterey Formation, California, USA: constraints from fluid-inclusion studies. Mineralogical Magazine, 54, 295–304. Bodvarsson G (1973) Temperature inversions in geothermal systems. Geoexploration, 11, 141–9. Coney D, Fyfe TB, Retail P, Smith PJ (1993) Clair appraisal: the benefits of a co-operative approach. In: Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference (ed. Parker JR), pp. 1409–20. Geological Society, London. Coolbaugh MF, Arehart GB, Faulds JE, Garside LJ (2005) Geothermal systems in the Great Basin, western United States: Modern analogues to the roles of magmatism, structure and regional tectonics in the formation of gold deposits. In: Window to the World: Geological Society of Nevada Symposium 2005 (eds Rhoden HN, Steininger RC, Vikre PG), pp. 1063–81. Geological Society of Nevada, Reno. Dean K, McLachlan K, Chambers A (1999) Rifting and the development of the Faeroe-Shetland Basin. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 533–44. Geological Society, London. Dorobek S (1989) Migration of orogenic fluids through the Siluro-Devonian Helderberg Group during late Paleozoic deformation: constraints on fluid sources and implications for thermal histories of sedimentary basins. Tectonophysics, 159, 25–45. Duddy IR, Green PF, Hegarty KA, Bray RJ, O’Brien GW (1998) Dating and duration of hot fluid flow events determined using AFTA and vitrinite reflectance-based thermal history reconstruction. In: Dating and Duration of Fluid Flow and Fluid-Rock Interaction (ed. Parnell J). Geological Society, London, Special Publications, 144, 41–51. George SC, Ruble TE, Dutkiewicz A, Eadington PJ (2001) Assessing the maturity of oil trapped in fluid inclusions using molecular geochemistry data and visually-determined fluorescence colours. Applied Geochemistry, 16, 451–73. George SC, Volk H, Ahmed M (2007) Geochemical analysis techniques and geological applications of oil-bearing fluid inclusions,
Palaeofluid analysis and oil charge history 81 with some Australian case studies. Journal of Petroleum Science and Engineering, 57, 119–38. Goodhue R, Clayton G (1999) Organic maturation levels, thermal history and hydrocarbon source rock potential of the Namurian rocks of the Clare Basin. Marine and Petroleum Geology, 16, 667–75. Hulen JB, Goff F, Ross JR, Bortz LC, Bereskin SR (1994) Geology and geothermal origin of Grant Canyon and Bacon Flat Oil Fields, Railroad Valley, Nevada. American Association of Petroleum Geologists Bulletin, 78, 596–623. Iliffe JE, Robertson AG, Ward GHF, Wynn C, Pead SDM, Cameron N (1999) The importance of fluid pressures and migration to the hydrocarbon prospectivity of the Faeroe-Shetland White Zone. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 601–11. Geological Society, London. Isaksen GH, Pottorf RJ, Jenssen AI (1998) Correlation of fluid inclusions and reservoired oils to infer trap fill history in the South Viking Graben, North Sea. Petroleum Geoscience, 4, 41– 55. Jones DM, Macleod G (2000) Molecular analysis of petroleum in fluid inclusions: a practical methodology. Organic Geochemistry, 31, 1163–73. Kennedy BM, van Soest MC (2007) Flow of mantle fluids through the ductile lower crust: Helium isotope trends. Science, 318, 1433–6. Kihle J (1995) Adaptation of fluorescence excitation-emission micro-spectroscopy for characterization of single hydrocarbon fluid inclusions. Organic Geochemistry, 23, 1029–42. Lamers E, Carmichael SMM (1999) The Paleocene deepwater sandstone play West of Shetland. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 645–59. Geological Society, London. Lampe C, Person M, No¨th S, Ricken W (2001) Episodic fluid flow within continental rift basins: some insights from field data and mathematical models of the Rhinegraben. Geofluids, 1, 42– 52. Li R, Parnell J (2003) In situ microanalysis of petroleum fluid inclusions by time of flight-secondary ion mass spectrometry as an indicator of evolving oil chemistry: a pilot study in the Bohai Basin, China. Journal of Geochemical Exploration, 78–79, 377– 84. Lisk M, O’Brien GW, Eadington PJ (2002) Quantitative evaluation of the oil-leg potential in the Oliver gas field, Timor Sea, Australia. AAPG Bulletin, 86, 1531–42. Lonergan L, Jolly RJH, Rawnsley K, Sanderson DJ (2007) Fractured Reservoirs. Geological Society Special Publication No. 270. Geological Society Publishing House, Bath. Mark DF, Green PF, Parnell J, Kelley SP, Lee MR, Sherlock SC (2008) Late Palaeozoic hydrocarbon migration through the Clair Field, West of Shetland, UK Atlantic margin. Geochimica et Cosmochimica Acta, 72, 2510–33. Mark DF, Parnell J, Kelley SP, Lee M, Sherlock SC, Carr A (2005) Dating of multistage fluid flow in sandstones. Science, 309, 2048–51. Middleton DWJ, Parnell J, Green PF, Xu G, McSherry M (2001) Hot fluid flow events in Atlantic margin basins: an example from the Rathlin Basin. In: Petroleum Exploration of Ireland’s Offshore Basins (eds Shannon PM, Haughton PDW, Corcoran D). Geological Society, London, Special Publications, 188, 91–105. Munz IA (2001) Petroleum inclusions in sedimentary basins: systematics, analytical methods and applications. Lithos, 55, 195– 212.
Munz IA, Johansen H, Holm K, Lacharpagne J-C (1999) The petroleum characteristics and filling history of the Frøy Field and the Rind discovery, Norwegian North Sea. Marine and Petroleum Geology, 16, 633–51. Naylor PH, Bell BR, Jolley DW, Durnall P, Fredsted R (1999) Palaeogene magmatism in the Faeroe-Shetland Basin: influences on uplift history and sedimentation. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 545–58. Geological Society, London. Nedkvitne T, Karlsen DA, Bjørlykke K, Larter SR (1993) Relationship between reservoir diagenetic evolution and petroleum emplacement in the Ula Field, North Sea. Marine and Petroleum Geology, 10, 255–70. Neilson JE, Oxtoby NH, Simmons MD, Simpson IR, Fortunatova NK (1998) The relationship between petroleum emplacement and carbonate reservoir quality: examples from Abu Dhabi and the Amu Darya Basin. Marine and Petroleum Geology, 15, 57– 72. O’Brien GW, Lisk M, Duddy I, Eadington PJ, Cadman S, Fellows M (1996) Late Tertiary fluid migration in the Timor sea: a key control on thermal and diagenetic histories? APPEA Journal, 36, 399–424. Osborne M, Haszeldine S (1993) Evidence for resetting fluid inclusion temperatures from quartz cements in oilfields. Marine and Petroleum Geology, 10, 271–8. Osborne MJ, Swarbrick RE (1999) Diagenesis in North Sea HPHT clastic reservoirs – consequences for porosity and overpressure prediction. Marine and Petroleum Geology, 16, 337–53. Owens P, Ryder AG, Blamey NJF (2008) Frequency domain fluorescence lifetime study of crude petroleum oils. Journal of Fluorescence, 18, 997–1006. Parnell J, Carey PF, Duncan W (1998) History of hydrocarbon charge on the Atlantic margin: evidence from fluid-inclusion studies, West of Shetland. Geology, 26, 807–10. Parnell J, Carey PF, Green P, Duncan W (1999) Hydrocarbon migration history, West of Shetland: integrated fluid inclusion and fission track studies. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 613–25. Geological Society, London. Parnell J, Green PF, Watt G, Middleton D (2005) Thermal history and oil charge on the UK Atlantic margin. Petroleum Geoscience, 11, 99–112. Parnell J, Middleton D, Chen H, Hall D (2001) The use of integrated fluid inclusion studies in constraining oil charge history and reservoir compartmentation: examples from the Jeanne d’Arc Basin, offshore Newfoundland. Marine and Petroleum Geology, 18, 535–49. Parnell J, Middleton DWJ (2009) Pore fluid and hydrocarbon migration histories in North Atlantic basins intruded by Tertiary sills. In: Faroe Islands Exploration Conference: Proceedings of the 2nd Conference (eds Varming T, Ziska H), pp. 124–55. Annales Societatis Scientiarum Faeroensis, Supplementum 50, To´rshavn. Parnell J, Watt GR, Middleton D, Kelly J, Baron M (2004) Deformation band control on hydrocarbon migration. Journal of Sedimentary Research, 74, 552–60. Person M, Banerjee A, Hofstra A, Sweetkind D, Gao Y (2008) Hydrologic models of modern and fossil geothermal systems within the Great Basin: implications for Carlin-type gold mineralization. Geosphere, 4, 888–917. Pironon J, Canals M, Dubessy J, Walgenwitz F, Laplace-Builhe C (1998) Volumetric reconstruction of individual oil inclusions by confocal scanning laser microscopy. European Journal of Mineralogy, 10, 1143–50.
82 J. PARNELL Pironon J, Thiery R, Ayt Ougougdal M, Teinturier S, Beaudoin G, Walgenwitz F (2001) FT-IR measurements of petroleum fluid inclusions: methane, n-alkanes and carbon dioxide quantitative analysis. Geofluids, 1, 2–10. Ritchie JD, Gatliff RW, Richards PC (1999) Early Tertiary magmatism in the offshore NW UK margin and surrounds. In: Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference (eds Fleet AJ, Boldy SAR), pp. 573–84. Geological Society, London. Rooney MA, Vuletich AK, Griffiths CE (1998) Compound-specific isotopes as a tool for characterizing mixed oils: an example from the West of Shetlands area. Organic Geochemistry, 29, 241–54. Ryder AG, Glynn TJ, Feely M, Barwise AJG (2002) Characterization of crude oils using fluorescence lifetime data. Spectrochimica Acta (A), 58, 1025–38. Ryder AG, Przyjalgowski M, Feely M, Szczupak B, Glynn TJ (2004) Time-resolved fluorescence microspectroscopy for characterizing crude oils in bulk and hydrocarbon-bearing fluid inclusions. Applied Spectroscopy, 58, 1106–15. Sample JC, Woods S, Bender E, Loveall M (2006) Relationship between deformation bands and petroleum migration in an exhumed reservoir rock, Los Angeles Basin, California, USA. Geofluids, 6, 105–12. Scotchman IC, Carr AD, Parnell J (2006) Hydrocarbon generation modelling in a multiple rifted and volcanic basin: a case study in the Foinaven Sub-basin, Faroe-Shetland Basin, UK Atlantic Margin. Scottish Journal of Geology, 42, 1–19.
Scotchman IC, Griffiths CE, Holmes AJ, Jones DM (1998) The Jurassic petroleum system north and west of Britain: a geochemical oil-source correlation study. Organic Geochemistry, 29, 671– 700. Shipton ZK, Cowie PA (2003) A conceptual model for the origin of fault damage structure in high-porosity sandstone. Journal of Structural Geology, 25, 333–44. Smith RL, McGarrity JP (2001) Cracking the fractures – seismic anisotropy in an offshore reservoir. The Leading Edge, 20, 18– 26. Stasuik LD, Snowdon LR (1997) Fluorescence micro-spectrscopy of synthetic and natural hydrocarbon fluid inclusions: crude oil chemistry, density and application to petroleum migration. Applied Geochemistry, 12, 229–41. Walderhaug O (1994) Temperatures of quartz cementation in Jurassic sandstones from the Norwegian continental shelf – evidence from fluid inclusions. Journal of Sedimentary Research, 64, 311–23. Worden RH, Smalley PC, Oxtoby NH (1998) Can oil emplacement stop quartz cementation in sandstones? Petroleum Geoscience, 4, 129–38. Wycherley HL, Parnell J, Watt GR, Chen H, Boyce AJ (2003) Indicators of hot fluid migration in sedimentary basins: evidence from the UK Atlantic Margin. Petroleum Geoscience, 9, 357–74. Ziagos JP, Blackwell DD (1986) A model for the transient temperature effects of horizontal fluid flow in geothermal systems. Journal of Volcanology and Geothermal Research, 27, 371–97.
Spatial variations in the salinity of pore waters in northern deep water Gulf of Mexico sediments: implications for pathways and mechanisms of solute transport J. S. HANOR AND J. A. MERCER Department of Geology and Geophysics, Louisiana State University, Baton Rouge, LA, USA
ABSTRACT Spatial variations in the salinity of pore waters in sedimentary basins can provide important insight into basinscale hydrogeologic processes. Although there have been numerous studies of brine seeps in the deep water Gulf of Mexico, much less is known about porewater salinities in the vast areas between seeps. A study has been made of spatial variation in pore water salinities in sediments in an approximately 500 km by 200 km area of the northern deep water (water depth >500 m) Gulf of Mexico sedimentary basin (GOM) to provide insight into pathways and mechanisms of solute transport in this portion of the basin. A second objective was to document salinities in the upper 500 m of the sedimentary section, the approximate depth to which methane hydrates, a potential future energy resource, may be stable. Elevated salinities would reduce the P–T stability range of hydrates. Salinities were calculated from borehole logs using a dual electrical conductivity model. Even though much of the northern GOM is underlain by allochthonous salt most of the undisturbed shallow sedimentary section has not been permeated by hypersaline waters. These waters are limited to areas near brine seeps. Hypersaline waters having salinities in excess of 100 g l)1 become more common at subseafloor depths of 2 km and greater. A field study at Green Canyon 65 and published numerical simulations of fluid flow above tabular salt bodies suggest that brines produced by salt dissolution migrate laterally and pond above salt and ⁄ or within minibasins and that the dominant mechanism of vertical solute transport is a combination of compaction-driven advection and diffusion, not large-scale thermohaline overturn. Superimposed on this diffuse upward flux of dissolved salt is the more focused and localized expulsion of saline fluids up faults. Key words: porewater salinity, Gulf of Mexico, solute transport, methane hydrates Received 30 June 2009; accepted 11 November 2009 Corresponding author: Jeffrey Hanor, Department of Geology and Geophysics, Louisiana State University, Baton Rouge, LA 70803, USA. Email:
[email protected]. Tel: +1 225 578 3418. Fax: +1 225 578 2302. Present address: J. A. Mercer, Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131, USA. Geofluids (2010) 10, 83–93
INTRODUCTION Spatial variations in the salinity of pore waters in sedimentary basins can provide important insight into the pathways and mechanisms of fluid flow and solute transport (e.g., Hanor & McIntosh 2007). Salinities can be determined by two very different techniques, by direct chemical analyses of water samples (see review in Kharaka & Hanor 2007) and by calculation from borehole log response (Bateman 1985; Revil et al. 1998; Ellis & Singer 2008). Borehole
logs often have an advantage in that salinities can often be calculated over large depth intervals, not simply at the depth of a well screen or perforation. Some examples of the use of salinity to delineate basinscale hydrologic processes include studies of topographically driven meteoric fluid flow in foreland basins and deltaic complexes, such as the Western Canada basin (Connolly et al. 1990), the North Slope basin, Alaska (Hanor et al. 2004), and the Beaufort-Mackenzie basin, Canada (Grasby et al. 2009); Pleistocene recharge of
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
84 J. S. HANOR & J. A. MERCER ice-sheet melt waters into saline formation waters in the upper Mid-Continent, USA (McIntosh & Walter 2005; Person et al. 2007), and thermohaline overturn in the Northern Gulf of Mexico basin driven by dissolution of salt (Hanor & McIntosh 2007). Variations in salinity on the smaller field and reservoir scales have been used to document fluid expulsion up faults (Roberts & Nunn 1995; Lin & Nunn 1997) and to provide evidence for reservoir compartmentalization and continuity (Bruno & Hanor 2003). The purpose of the research reported on here was to investigate spatial variations in the salinity formation waters in sediments of the deep-water (>500 m water depth) Gulf of Mexico (GOM) and compare them with what has previously been developed for the northern Gulf onshore and shelf. The motivation in part was simply to extend our knowledge of the regional hydrogeology of the northern Gulf of Mexico basin southward from the Arkansas-Louisiana gulf rim and Gulf coastal plain into the continental shelf and slope. However, there was also an applied component to our salinity research. Gas hydrates are a potential energy resource in the deepwater Gulf of Mexico (Frye 2008), and pore water salinity is one of the key factors controlling the thermodynamic stability of gas hydrates in marine sediments (Tishchenko et al. 2005; Sun & Duan 2007). Specifically, elevated salinities reduce the P–T stability range of gas hydrates. Although there has been extensive work on the seafloor brine seeps in the GOM (e.g., Ruppel et al. 2005 and references therein), much less was known about pore water salinities in the far larger areas between seeps. Because much of the northern GOM is underlain by allochthonous salt bodies which are susceptible to dissolution, the possibility existed that there has been pervasive saturation of the shallow sedimentary section by brines which would significantly reduce the volumetric extent of the hydrate stability zone. There were two components to our salinity study. The first was a km-scale evaluation of variations in salinity in a deep-water oil and natural gas field to assess spatial relations between salinity and proximity to salt. The second was a regional reconnaissance study of subsurface variations in salinity over the northwestern Gulf of Mexico. The results of these borehole log-based studies were then compared with published information on salinities derived from direct porewater sampling and with some recent numerical modeling experiments.
GULF OF MEXICO SEDIMENTARY BASIN: SEDIMENTS, SALT, AND SALINITY Mesozoic Gulf Rim The Gulf of Mexico basin began to form in the late Middle Jurassic as a result of lithospheric stretching and seafloor
Fig. 1. Map of the northern rim of the Gulf of Mexico basin showing features discussed in the text.
spreading associated with the breakup of Pangea (Worrall & Snelson 1989; Salvador 1991). A widespread marine incursion in the Middle Jurassic resulted in the deposition of thick masses of evaporites, known as the Louann Salt, over much of the Gulf region (Fig. 1). The original depositional thickness of the Louann Salt is estimated to have varied from a few hundred meters along the Gulf rim to over a km in the interior salt basins and as much as several km within the area of the present Texas-Louisiana continental slope (Salvador 1991). In the Gulf rim of Texas and Arkansas, the salt is still primarily in its original, near-basal position within the section. Approximately 3 km of Jurassic and Cretaceous marine and non-marine sediments were deposited on top of salt in what is now southern Arkansas. Dickey (1966) was the first to recognize the remarkably linear increase in formation water salinity in southern Arkansas from low salinities at the base of a thin freshwater zone to salinities in excess of 350 g l)1 near the top of salt. Similar relations were documented by Carpenter et al. (1974) for central Mississippi. Hanor (1984) suggested that the systematic linear increase in salinity over a 3 km depth interval represents an on-going, steady-state mass transport of dissolved NaCl from Middle Jurassic evaporites at the base of the section to the base of the shallow meteoric regime. Subsequent numerical modeling of this process by Ranganathan & Hanor (1987) showed that molecular and Soret diffusion are probably not rapid
Spatial variations in the salinity of pore waters 85 enough processes to account for the observed linear profile. Work by Sarkar et al. (1992) showed on the basis of theoretical grounds and previous laboratory experiments that where there is an increase in salinity with depth in a dipping sedimentary sequence, there will be perturbation of isohaline contours at the interface between two sedimentary beds of differing molecular diffusivity. The magnitude of this perturbation is sufficient, in theory, to induce gravitational instability and convection, and dispersion can thus occur if there is a dip to sedimentary bedding even where there is an overall increase in fluid density with depth. See also the extensive discussion of salinity-induced fluid instabilities by Phillips (1991). Cenozoic onshore and shelf The depositional style within the Gulf Basin changed dramatically at the beginning of the Cenozoic with the influx of large masses of sands, silts, and muds from the north and northwest which formed thick fluvial-deltaic and shallow-marine siliciclastic sequences (Galloway et al. 2000). During the lower Miocene, the major depocenters in the Gulf basin shifted from Texas to South Louisiana (Galloway et al. 2000). In south Louisiana and the Louisiana shelf, sediment progradation to the south created a largely sand-dominated section, approximately 3 km thick, which overlies a thick sequence of mudstone-dominated marine sediments. Fluids in the mudstone-dominated section are typically overpressured (Hanor & Sassen 1990). In contrast to the Gulf Rim, where salt is still in its basal position, differential subsidence of salt relative to the surrounding, denser clastic sediments has produced salt diapirs which penetrate shallower and much younger sediments (Worrall & Snelson 1989). Also in contrast to the northern Gulf rim, salinities in the Cenozoic section of south Louisiana Gulf Coast and continental shelf (Fig. 1) do not systematically increase with depth. Instead, the most saline waters occur within the sand-dominated section between subsurface depths of 0.5–3 km and have been produced by the dissolution of salt domes at relatively shallow depths rather than by dissolution of deeply buried bedded salt (Hanor & Sassen 1990). It has been proposed that thermohaline, density-driven free-convection is an important fluid and solute transport mechanism in the shallower fluvial-deltaic sequences (Hanor & Sassen 1990). In the Miocene and younger siliciclastic section pore water salinities between depths of 0.5–3 km typically range from 100 to 150 g l)1, although values as high as 300 g l)1 have been found in the immediate vicinity of some salt structures (Szalkowski & Hanor 2003). In the underlying mudstone-dominated overpressured section, porewater salinities often decrease with depth back to original marine salinities of 35 g l)1 or less presumably as a result of dehydration reactions (Kharaka & Hanor 2007).
Cenozoic offshore Cenozoic sediments of the northern deep water Gulf of Mexico (water depths >500 m) are characterized by bathyal turbidite systems deposited in a complex of minibasins developed on top of and between salt bodies (Weimer et al. 1998). Salt of Middle Jurassic age has detached from its original basal position to form allochthonous bodies having a wide range of geometries, including both diapirs and subhorizontal salt sheets. Some salt sheets have moved large distances laterally and some crop out on the seafloor (Pilcher & Blumstein 2007). Some of the Pliocene-Pleistocene sedimentary sequences within the minibasins between salt are up to 6 km thick. Sands in these turbidite deposits are the principal hydrocarbon reservoirs in the northern deep water Gulf (Weimer et al. 1998).
VARIATIONS IN SALINITY IN DEEP-WATER GULF SEDIMENTS Background The northern deep-water Gulf of Mexico is a major hydrocarbon province, and the exploration for and production of hydrocarbons within this region is regulated by the United States Minerals Management Service (MMS). The United States Federal waters are divided into protraction areas, such as the Green Canyon (GC) area (Fig. 2), which average approximately 110 km by 200 km in areal dimension. The protraction areas are further divided into numbered blocks, which are 4.8 km (three miles) square. Blocks are identified by protraction area and number, for example, GC 65 (Fig. 2). The Green Canyon (GC) protraction area was selected as a focus for the study because of Hanor’s prior work on salinity in Shell’s Bullwinkle field, GC 65 (Hanor 2007),
Fig. 2. Map of the offshore Gulf of Mexico showing the location of boreholes used in the salinity study (black dots), areas discussed in the text (open squares) and protraction areas (large rectangles). The following abbreviations are used: AC = Alaminos Canyon, AV = Atwater Valley, GB = Garden Banks, GC = Green Canyon, KC = Keathley Canyon, MC = Mississippi Canyon, and WR = Walker Ridge.
86 J. S. HANOR & J. A. MERCER and because a significant number of blocks within Green Canyon have been drilled. Additional logs were selected from blocks in five additional protraction areas. However, the discussion here will be largely limited to Green Canyon, but the results for GC are characteristic of the other areas. A total of nearly 1000 salinity calculations were done for 84 boreholes Gulf-wide (Fig. 2). Depending on where logging began and the depth of the hole, salinity determinations extended from a few hundred meters of the seafloor or less to depths of 3 or 4 km or more. Preference was given to logs that appeared to have fairly well-defined sands, as evidenced by gamma ray response, on the assumption that sands are more likely to yield accurate salinity values than mudstones. Techniques Salinities within the GC 65 field area were calculated from spontaneous potential (SP) logs using standard techniques (Bateman 1985). While SP logs do exist for some of the older deep water Gulf of Mexico fields, such as GC 65, the SP response is not included in Logging While Drilling (LWD) logs, which is the current deep water logging technology (Ellis & Singer 2008). As a result, the regional reconnaissance study of formation water salinity, which relied on LWD logs, was made instead using the electrical conductivity technique of Revil et al. (1998). The technique is a refined Archie dual-conductance model (Ellis & Singer 2008), where the measured bulk sediment electrical conductivity is partitioned by calculation between conduction of charge by free ions in aqueous solution, which is a proxy for salinity, and by cations adsorbed on clay minerals. The reader is referred to the original Revil et al. (1998) for the series of equations and constants involved in the salinity calculations or to the recent summary of the dual-conductance technique by Ildefonse et al. (2009). The following information is required as input into the calculations: gamma ray response to calculate the fraction of clay in the bulk sediment, clay mineralogy to calculate the cation exchange capacity of the bulk sediment, bulk sediment electrical resistivity or conductivity, porosity, and temperature. Information on the vertical distance from the seafloor to the top of salt at each borehole location was provided by MMS. The Revil et al. salinity technique was run on a series of digital log data sets from two Gulf of Mexico cruises, IODP Cruise 308 (Flemings et al. 2006) and the JIP 1 Cruise (Kastner et al. 2008) for which downhole values of salinity have been determined by direct chemical analysis of pore fluid samples. However, digital logs for the main part of this study were not available from MMS, and log images had to be used instead. Values for gamma ray and resistivity (or conductivity) were read directly off log images at selected depths for each well. Wherever possible, sands
were selected. However, in many wells there were no welldefined sands in the upper 1000 m of section, and sandy mudstones or mudstones had to be used instead. The cation exchange capacity (CEC) of the sediment is used in partitioning the transport of electrical charge between cations in solution and cations adsorbed on clay minerals. In the absence of direct information on the clay mineralogy and bulk cation exchange capacity of mudstones in the boreholes that were included in this study, a value of 0.0792 meq g)1, the same value used by Revil et al. in their study of salinity-depth relations at Eugene Island 330, was used here also. This is reasonable for sediments derived in large part from the Mississippi River drainage system. Also following Revil et al., an API gamma ray (GR) response of 15 was chosen for end-member sand and 114 for end-member shale. The GR logs were then used to calculate the clay weight fraction and CEC of the sediment. A porosity-depth relation that yields reasonable values for salinity was derived as part of this study from porosity data for fine-grained sediments in IODP Cruise 308, Site U1324 in the Mississippi Canyon area (Flemings et al. 2006). Porosities here were determined by IODP scientists both from log response and by measurement of the moisture content and weight of sediment samples over the entire 600 m cored interval. Downhole temperatures were determined from temperature gradients calculated from proprietary equilibrated bottom hole temperatures (eBHTs) provided by MMS. Results: Green Canyon 65 salinity study The GC 65 field is situated approximately 200 km southwest of New Orleans (Fig. 2). Production has been primarily from sands on the southwest margin of the salt-walled
Fig. 3. Northwest – southeast cross section through the GC 65 (Bullwinkle) minibasin showing spatial variations in formation water salinity (g l)1) as determined from SP response. There are normal seawater salinities of approximately 35 g l)1 down to a depth of approximately 2 km MSL. Brines formed by the dissolution of salt have ponded above salt and a condensed stratigraphic section at the top of the Pliocene, which is shown by a bold wavy line. Log control is shown by the vertical and angled lines below a depth of 500 m.
Spatial variations in the salinity of pore waters 87
KC-151 Salinity (ppt) 0
20
40
60
80
0 50 100 Depth (mbsf)
Bullwinkle minibasin. Brines formed by the dissolution of salt have ponded above a condensed stratigraphic section at the top of the Pliocene, which is represented by the wavy subhorizontal line in Fig. 3. The results of the SP calculations suggest the occurrence of near-normal marine salinities to a depth of ca. 2 km MSL, which represents the regional minimum depth to the top of salt in this part of the Green Canyon area. Below this elevation, salinities within the minibasin increase monotonically with depth (Fig. 3). However, there is a pronounced reversal in salinity below the condensed section, and sediments outside the Bullwinkle minibasin usually show significant reversals with depth, but have normal seawater salinities at shallow depths. Salinity in the hydrocarbon reservoir sands, as determined from chemical analyses of produced waters, decreases laterally away from salt (Hanor 2007).
150 Calculated alinity Lab salinity
200 250 300 350 400 450 Site U1332A Salinity (ppt) 0
10
20
30
40
50
60
70
0
Results: regional salinity study
Depth (mbsf)
50
Salinity results for two digital log sets for which porewater salinity had previously been determined by chemical analysis are shown in Fig. 4. The calculated salinities have a high-amplitude, high-frequency response with depth reflecting similar depth variations in the gamma ray, porosity, and resistivity digital data which go into the calculations (see also figure 10 in Revil et al. 1998). The calculated salinities for IODP Site U1322A, near MC 808 (Fig. 2) are somewhat low in the upper 50 m of section and approximately 10% too high below 100 m. The calculated salinities of the Keathley Canyon 151 site (KC 151 in Fig. 2) have been smoothed using a centered 41-point moving average to try to reduce some of the noise in the signal. The calculated salinities capture in the increase in salinity from seawater values in the upper part of the section, a salinity reversal at approximately 100 m, and salinities of over 50 ppt at the base of the section. Collett (2005) proposed that the section between a depth of 210 and 300 m contains hydrate which has yielded an anomalous resistivity log response. This in turn has yielded anomalously low salinities. Although the Revil technique is not as accurate as directly determining salinities from chemical analysis of porewaters, we conclude that the values of salinity that were obtained from visual log images in this study are sufficiently accurate to allow one to distinguish between formation waters that are of approximately normal seawater salinity (calculated values of ca. 35 g l)1), waters that are obviously hypersaline (calculated values >100– 120 g l)1) and waters of intermediate salinities. The salinity-depth plots in Fig. 5 are representative of the salinity profiles that were obtained using the Revil et al. technique described above. One distinct feature of many, if not most, of the salinity profiles is a systematic increase in salinity with depth below the seafloor (Fig. 5A– C). Formation water salinities typically start out in the
100
150
200
250 Fig. 4. Comparison of porewater salinities calculated using the Revil et al. (1998) dual-conductivity log technique (solid symbols) and salinities determined by laboratory analysis of porewaters squeezed from core samples (open symbols). Data for IODP 308 Site U1322A from Flemings et al. (2006) and data for KC 151 from Collett (2005) and Kastner et al. (2008).
shallow section at or near normal seawater salinities of 35 g l)1 and then increase with depth to values in excess of 100 g l)1 (Fig. 5B,C). However, there are exceptions, such as the GC112 profile (Fig. 5D), where there are hypersaline waters very near the surface. In such cases, the salinity profiles typically reverse with depth, and the shallow hypersaline waters are underlain by waters of normal marine salinities. It is possible that the shallow hypersaline waters reflect the formation of hydrates. An exception to the profiles which show a continuous increase in salinity with depth is also the profile for GC112, which is dominated by waters of normal seawater salinity over a depth range of nearly 3500 m. The salinity-depth trend for GC 9 (Fig. 5A) is unusual in that there are very saline waters at relatively shallow depths (over 250 g l)1 at a subseafloor depth of 1400 m). The information provided to us
88 J. S. HANOR & J. A. MERCER
(A)
Depth (m)
0
0
Salinity (g l–1) 100 150 200
50
250
300
GC 9
400 800 1200 1600
(B) 0
0
50
100
Salinity (g l–1) 150 200 250
300
350
GC 75
1000 Depth (m)
400
2000 3000 4000 5000
(C) 0
0
20
40
Salinity (g l–1) 60 80 100
120
140
160
Depth (m)
GC 110 1000
2000
3000
(D) 0
Depth (m)
1000
0
50
Salinity (g l–1) 100 150
200
250
GC 112
2000
few meters below the seafloor (Fig. 6), which were obtained by visually extrapolating salinity-depth trends for each well upward to the seafloor. Most of the Green Canyon blocks studied here are characterized by near normal seawater salinity and salinities of less than 60 g l)1 to depths of 1000 mbsf. Shallow formation waters appear to be slightly more saline in the northeastern part of the area. A series of salinity cross sections, including a west–east section across the northern portion of Green Canyon (Fig. 7), was constructed using the program Surfer (Rockworks). The light gray areas represent probable areas of normal seawater salinities, and the darker areas represent progressively hypersaline waters. The general pattern is that of near normal salinity in most areas from the sea floor to a depth of 0.5–1 km and then a progressive increase in salinity with depth below that. In some areas there are abrupt increases in salinity at subseafloor depths of 1.5– 2.0 km. We examined possible relations between salinity and proximity to salt using two approaches. In the first the estimated subseafloor depth to the shallowest salt in a given block where we had salinity information was used as a proxy for proximity to salt on the assumption that the dissolution of shallow salt would more readily raise the salinity in shallow sediments within the block than the dissolution of deep salt. In the second approach information provided by MMS on depths to top of salt was used to calculate the subseafloor depth to the top of salt at the specific location of the borehole. We found no clear relation between the depth of calculated near-normal seawater salinity and depth to shallowest salt or the depth to the top of calculated hypersaline (>100 g l)1) pore waters and depth of shallowest salt within given blocks. There is also no clear relation between near-surface salinities and subseafloor depth to the top of salt at individual borehole sites. This may reflect lateral transport of saline waters from shallower salt structures, as noted in the Green Canyon 65 study.
3000 4000
REVIEW OF PREVIOUS STUDIES OF PORE WATER SALINITIES IN THE GULF OF MEXICO
5000 Fig. 5. Representative salinity-depth profiles for boreholes in the Green Canyon area. Depths are reported as below seafloor. Normal seawater salinity is approximately 35 g l)1. The control points represent sands, which are not abundant in the deep water sediments.
indicates that the minimum depth to the top of salt in this block is approximately 3 km. The areal coverage of logs in the Green Canyon area was sufficient to permit making maps showing spatial variations in salinity at specific depths below the seafloor. Our discussion here will be limited to estimated porewater salinities a
Gulf of Mexico drilling cruises There have been several major deep sea coring and drilling cruises in the Gulf of Mexico which have provided important information on variations in pore water salinities, including the pioneering cruise by Manheim & Bischoff (1969), which was the first to establish the relation between the occurrence of hypersaline pore waters in deepwater sediments and the presence of salt. More recent work has focused on documenting porewater salinities on and near seafloor seeps and mounds in relation to the possible occurrence of gas hydrates (Ruppel et al. 2005; Ussler
Spatial variations in the salinity of pore waters 89
28
A'
A
Fig. 6. Map of the Green Canyon study area showing variations in formation water salinity immediately before the seafloor, as determined by projecting salinity-depth trends upward. Dots show borehole control. Line A-A’ is the location of the cross-section in Fig. 5. Seawater salinity is 35 g l)1.
Longitude
27.8
27.6
Salinity (g l–1) 27.4
20
40
–91.8
–91.6
80 –91.2
–91
–90.8
–90.6
–90.4
–90.2
Latitude
Elevation (km from seafloor)
A
Fig. 7. Cross section A-A’ across the northern part of Green Canyon showing spatial variations in formation water salinity. Dots show control. White areas at the top are artifacts of the contouring method. Seawater salinity is 35 g l)1.
60 -91.4
A'
GC 9 I
0 –1 –2 –3 –4 0
2
4
6
8
10
& Paull 2007 & references therein). An intensively sampled and studied site in GOM hydrate exploration is located in Atwater Valley blocks 13 and 14 (Fig. 2), east of Green Canyon (see Wood et al. 2008 and references therein). Among the several studies which have been made in the area are: multi-resolution seismic imagery which has been used to determine mechanisms of methane transport (Wood et al. 2008); electromagnetic (EM) surveys which reflect the influence of elevated temperatures and salinities on EM response (Ellis et al. 2008); heat flow surveys
Fig. 8. Cross section across portions of Atwater Valley blocks 13 and 14 showing estimated top of methane gas and spatial variations in subseafloor temperatures. Labels D and F are seafloor mounds. PC15 and PC14 are the locations of piston core sites shown on Fig. 9. Extracted from data on a figure provided by Warren Wood (Wood personal communication, 2009). Bold arrows show proposed areas of focused upward transport of methane gas and warm waters by Wood et al. (2008).
20
12
14
16
18
Salinity (g l–1)
Distance (10 km units) 40
60
80
100
200
300
350
400
(Coffin et al. 2006); and chemical analyses of pore waters squeezed from shallow cores (Coffin et al. 2006). Fig. 8 is a northwest to southeast cross section through two seafloor mounds, designated D and F, based on information in a figure provided by Warren Wood, U.S. Naval Research Laboratory (Wood, personal communication, 2009). Shown is the top of the free gas zone, as determined by seismic imagery (Wood et al. 2008), and temperature contours extrapolated from shallow measurements by J.M. Gardner (Coffin et al. 2006). We have constructed a cross section based on chloride analyses published by Coffin et al. (2006), which shows spatial variations in chlorinity to a subseafloor depth of 5 m along the part of the transect which traverses Mound F (Fig. 9). Piston cores were not taken across Mound D. All three data sets pictured in Figs 8 and 9 support the hypothesis by the researchers who generated these data is that there is or has been upward focused flow immediately below the two seafloor mounds which has transported heat, saline waters, and methane gas (Wood et al. 2008). Of relevance to the present study is the observation that significantly elevated chlorinities in the transect are restricted to a zone approximately 500 m wide at Mound F. Chloride concentrations are rapidly attenuated with distance to normal marine values away from the mound. Even though piston core PC15 was taken in close proximity to Mound D, its waters are of normal seawater salinity. Similar conclusions that elevated porewater salinities or chlorinities are limited
90 J. S. HANOR & J. A. MERCER
Mound F
Elevation below Sea floor (cm)
PC15
PC14
0 –250 –500
0
500
1000
1500
2000
2500
3000
Chloride (mM)
Distance (m) 500
600
700
800
900
950
1000
Fig. 9. Cross section showing spatial variations in chloride concentrations in pore waters extracted from piston cores between PC15 and PC14 in Fig. 6. Piston coring did not extend across Mound D. Normal seawater Cl = 546 mM. Note zone of elevated chloride approximately 500 m wide below Mound F and the rapid attenuation of Cl to normal marine values away from the mound. PC15 is located very near Mound D (Fig. 8) but shows normal seawater salinities. Contoured from data of Coffin et al. (2005).
to areas within a km or less of a seafloor brine seep can be drawn from the field studies of Ruppel et al. (2005) and Ussler & Paull (2007). Although some of the sites cored and sampled during the western phase of the IODP 308 Cruise, near MC 808 (Fig. 2), were in proximity to salt, the porewaters have normal seawater salinities (Flemings et al. 2006). Sources of saline waters The spatial variations in salinity relative to the location of salt at GC 65 clearly show that the dissolution of salt is responsible for producing elevated salinities in this field (Fig. 3; see also Hanor 2007). Reitz et al. (2007) and Kastner et al. (2008) reported the results of chemical and isotopic analyses of pore waters squeezed from sediment samples from the deepwater Gulf of Mexico. Their results are significant in that each research group proposed an origin for the saline waters they analyzed that invokes some mechanism in addition to or other than the simple dissolution of salt. On the basis of oxygen ⁄ hydrogen isotopic
Fig. 10. Simplified representation of the results of a numerical modeling experiment by Wilson & Ruppel (2007) the dissolution of salt over a million-year model run generated interstitial brines which ponded in structural lows on the top of salt. The upper km of sedimentary section retained its original seawater salinity. Note a general similarity in salinity pattern to that at GC 65 (Fig. 3).
compositions and Na ⁄ Cl and Br ⁄ Cl ratios of these waters from Bush Hill (GC 185) Reitz et al. concluded that rapidly ascending waters underwent a sub-critical phase separation, i.e. boiled (Bischoff 1991), at a subseafloor depth of approximately 1650 m thus increasing the salinity of the residual liquid phase. However, it remains to be demonstrated that this is a physically viable process in this particular field setting. Kastner et al. (2008) concluded that the saline waters they sampled in the Atwater Valley and Keathley Canyon areas were not derived by halite dissolution but rather ‘‘…by advection of a residual brine from within or beyond halite precipitation.’’ This would imply a Middle Jurassic source for the brines, the time the Louann salt was deposited rather than brines formed by the more recent dissolution of more shallow salt. The origins of elevated salinity in GOM sediments need to be studied further. Numerical modeling of thermohaline convection in deep water sediments Wilson & Ruppel (2007) published the results of a numerical modeling study of subseafloor fluid convection driven by spatial variations in temperature and salinity in deep water GOM sediments (Fig. 10). The physical framework of their model is based on an E-W seismic line across Garden Banks block 425 (Fig. 2), immediately west of the Auger field. In their model a 1.5–2.0 km thick sedimentary sequence overlies a tabular salt structure which has a subsurface high which corresponds in position to a seafloor mound. Wilson and Ruppel chose a baseline sediment intrinsic permeability, k, of 10)15 m2, or one millidarcy, in their modeling study. A number of simulations were performed, including increasing and decreasing the baseline k by an order of magnitude and by including fault zones of high permeability. The perturbation of the thermal field by the presence of salt and the generation of salinity by salt dissolution induced density-driven thermohaline circulation of pore water in their model. Increasing and decreasing k by an order of magnitude increased and decreased flow
Spatial variations in the salinity of pore waters 91 rates by approximately an order of magnitude. The temperature contours in their model results parallel the topography of the seafloor to a depth of 1 km. Hence, the model advective fluid flow did not significantly perturb temperature gradients in the upper km of the model sedimentary section. Based on the figures in the Wilson and Ruppel paper, their dissolution model also produced no significant perturbation of the salinity gradient in the uppermost kilometer of sediment. Instead, the interstitial brines produced by salt dissolution ponded in structural lows on either side of the salt crest. The presence of these brines slowed haline convection over time, and molecular diffusion appeared to take over from advection as the dominant solute transport mechanism toward the end of their run (their Fig. 7). The salinity structure shown in their figure is similar to that observed at the Bullwinkle field, GC 65 (Fig. 3). While permeabilities of 10)15 m2 or even higher would be appropriate for the lateral permeability of individual sand beds in GOM turbidite sequences, the vertical permeability of the mudstone-dominated part of the section is likely to be orders of magnitude smaller. Based on the compilation of permeability data for marine sediments of Spinelli et al. (2004), a permeability of 10)15 m2 would be appropriate only for sediments of high porosity at subseafloor depths of less than approximately 20 m. Below these depths, permeabilities rapidly drop by one and two orders of magnitude. Dugan (2008) cites a value of 10)18 m2 at a depth of 300 m in the Keathley Canyon area. It is possible that large-scale fluid convection involving significant vertical transport could not even occur in sediments having such low permeabilities. Two field examples of where fluid density inversions have not been sufficient to drive convective fluid overturn are the condensed stratigraphic section at the GC 65 field, as described previously, and below the Orca basin, a brinefilled seafloor depression in the Garden Banks area at 2656¢N and 9120¢W (Pilcher & Blumstein 2007). Ihizuka et al. (1986) showed that formation water salinities decrease from 237 ppt immediately below the floor of this basin to values of 50 ppt at a depth of approximately 40 m. There are thus significant salinity and fluid density inversions over a very short depth interval. The convex upward nature of the salinity and chlorinity curves is consistent with non-steady-state downward solute transport by molecular diffusion (see earlier diffusion calculations for this site by Addy & Behrens 1980). Downward advection would have produced a convex downward decrease in salinity.
DISCUSSION AND CONCLUSIONS Even though much of the northern Gulf of Mexico is underlain by allochthonous salt and numerous seafloor brine seeps have been documented, our study suggests that
most of the undisturbed portion of the shallow sedimentary section has not been regionally permeated by hypersaline waters. These waters appear to be limited to areas near seafloor brine seeps. Most of the sedimentary section, at least in the boreholes studied, is characterized by pore waters having apparently normal seawater salinities (ca. 35 g l)1) to moderately elevated seawater salinities (<60 g l)1) to a subseafloor depth of a approximately 1 km. Some of the shallow, moderately-elevated salinities may possibly be the result of hydrate formation, which removes free water and can elevate porewater salinities (Liu & Flemings 2006). Hypersaline waters having salinities in excess of 100 g l)1 become more common at subseafloor depths of 2 km, which in the northern Green Canyon area is the approximate regional minimum depth to the top of salt, as determined from MMS data. Many of the salinity-depth profiles generated in this study have a concave-downward trend, similar to salinity profiles in post-Miocene sediments in Messenian evaporite basins of the Mediterranean ocean (McDuff & Gieskes 1976), although over greater depth intervals. The Mediterranean profiles have been shown by McDuff and Gieskes to reflect solute transport dominated by molecular diffusion and compaction-driven advection. Modeling of solute transport by these two processes using the techniques of Ranganathan & Hanor (1988) in deep water Gulf of Mexico sediments would be a logical follow-up to the present study. However, although the ages of the PlioPleistocene reservoir sands are generally well-known from biostratigraphy, the ages of the overlying sediments are less well constrained. The field study at Green Canyon 65 and published numerical simulations of fluid flow above tabular salt (Wilson & Ruppel 2007) suggest that interstitial brines produced by salt dissolution flow laterally and pond above salt and ⁄ or low permeability sedimentary sections within minibasins, thus inhibiting large-scale vertical thermohaline overturn. Detailed kilometer-scale studies of the salinity fields around salt would help to better define transport processes at and away from the sediment-salt interface and the origin of high salinities in GOM sediments. The present study shows that relations between pore water salinity and proximity to salt which have been developed for the onshore areas of the Gulf of Mexico, as discussed earlier in this paper, are not good models for the generally fine-grained, bathyal sediments of the deep water Gulf. Superimposed on a diffuse upward flux of dissolved salt which appears to be dominating much of the deep water sedimentary section is the more focused and localized expulsion of saline fluids up faults (Wood et al. 2008). Additional study of the mechanisms of solute transport in these deep water sediments would not only enhance our understanding of the hydrogeology of this part of the GOM, but would improve our understanding of the potential distribution of methane hydrates. One
92 J. S. HANOR & J. A. MERCER possibility that should be explored further is that on-going sediment deformation driven by salt tectonics has been a factor in inducing fluid flow and solute transport.
ACKNOWLEDGMENTS The authors thank the Editors for the invitation to submit a paper to this special issue of Geofluids. Hanor would like to thank Shell Offshore Inc. for permission to publish the results of the GC 65 study. The regional salinity study was supported by the U.S. Minerals Management Service. We thank Matt Frye and Bill Shedd of MMS and John Grace of Earth Science Associates for valuable discussion and critical data. Warren Wood of the U.S. Naval Research Lab provided important information and a reference figure for the Atwater Valley transect. Candace Guttner played a key role in the initial phase of log interpretation. Our more recent work has been supported in part by NSF Grant ERA-0557555 (Nunn & Hanor). This paper was greatly improved as the result of comments by two anonymous reviewers.
REFERENCES Addy SK, Behrens EW (1980) Time of accumulation of hypersaline anoxic brine in Orca Basin (Gulf of Mexico). Marine Geology, 37, 241–52. Bateman RM (1985) Open-hole Log Analysis and Formation Evaluation. International Human Resources Development Corp., Boston. Bischoff JL (1991) Densities of liquids and vapors in boiling NaCl–H2O solutions: a PVTx summary from 300 to 500C. American Journal of Science, 291, 309–38. Bruno RS, Hanor JS (2003) Large-scale fluid migration driven by salt dissolution, Bay Marchand dome, offshore Louisiana. Gulf Coast Association of Geological Societies Transactions, 53, 97– 107. Carpenter AB, Trout ML, Pickett EF (1974) Preliminary report on the origin and chemical evolution of lead- and zinc-rich oil field brines in central Mississippi. Economic Geology, 69, 1191– 206. Coffin RB, Gardner J, Pohlman J, Downer R, Wood W (2006) Methane Hydrate Exploration, Atwater Valley, Texas-Louisiana shelf: Geophysical and Geochemical Profiles. Naval Research Laboratory NRL ⁄ MR ⁄ 6110-06-9002, Washington DC. Collett TS (2005) Cruise Report: ChevronTexaco Gulf of Mexico JIP drilling program: downhole logging program, DOE-NETL Topical Report, Appendix E. http://www.netl. doe.gov/technologies/oil-gas/publications/Hydrates/reports/ NT41330_Report_CruiseLogging.pdf Connolly CA, Walter LM, Baadsgaard H, Longstaffe FJ (1990) Origin and evolution of formation waters, Alberta Basin, Western Canada sedimentary Basin. I. Chemistry. Applied Geochemistry, 5, 375–95. Dickey PA (1966) Patterns of chemical composition of deep subsurface waters. AAPG Bulletin, 50, 2472–8. Dugan B (2008) Fluid flow in the Keathley Canyon 151 minibasin, northern Gulf of Mexico. Marine and Petroleum Geology, 25, 919–23.
Ellis DV, Singer JM (2008) Well Logging for Earth Scientists, 2nd edn. Springer, New York. Ellis M, Evans RL, Hutchinson D, Hart P, Gardner J, Hagen R (2008) Electromagnetic surveying of seafloor mounds in the northern Gulf of Mexico. Marine and Petroleum Geology, 25, 960–8. Flemings PB, Behrmann J, Davies T, John C, the Expedition 308 Project Team (2006) Proc. IODP, 308. College Station, TX (Integrated Ocean Drilling Program Management International, Inc.), doi:10.2204/iodp.proc.308.101.2006. Frye M (2008) Preliminary Evaluation of In-Place Gas Hydrate Resources: Gulf of Mexico Outer Continental Shelf. Minerals Management Services OCS Report, MMS 2008-004, Washington DC. Galloway WE, Caney-Curry PE, Li X, Buffler RT (2000) Cenozoic depositional history of the Gulf of Mexico basin. AAPG Bulletin, 84, 1743–74. Grasby SE, Chen Z, Issler D, Stasiuk L (2009) Evidence for deep anaerobic biodegradation associated with rapid sedimentation and burial in the Beaufort-Mackenzie basin, Canada. Applied Geochemistry, 24, 536–42. Hanor JS (1984) Variation in the chemical composition of oil-field brines with depth in northern Louisiana and southern Arkansas: Implications for mechanisms and rates of mass transport and diagenetic reaction. Gulf Coast Association of Geological Societies Transactions, 34, 55–61. Hanor JS (2007) Pre-production spatial variation in formation water salinity in a deepwater Gulf of Mexico field. In: Water– Rock Interaction (eds Bullen E, Wang T), pp. 505–8. Taylor and Francis Group, London. Hanor JS, McIntosh JC (2007) Diverse origins and timing of formation of basinal brines in the Gulf of Mexico sedimentary basin. Geofluids, 7, 227–37. Hanor JS, Sassen R (1990) Evidence for large-scale v vertical and lateral migration of formation waters, dissolved salt, and crude oil in the Louisiana Gulf Coast. GCSSEPM, Proceedings 9th Annual Research Conference, Houston, TX, pp. 293–6. Hanor JS, Nunn JA, Lee Y (2004) Salinity structure of the central North Slope foreland basin, Alaska, USA: implications for pathways of past and present topographically-drive fluid flow. Geofluids, 4, 152–68. Ihizuka T, Kawahate H, Aoki S (1986) Interstitial water geochemistry and clay mineralogy of the Mississippi fan and Orca and Pigmy basins, Deep Sea Drilling Project Leg 96. Initial Reports DSDP, 96. Washington (U.S. Government Printing Office), 711–26. Ildefonse B, Drouin M, Violay M, Pezard P (2009) Data report: electrical properties of gabbroic and troctolitic rocks from IODP Hole U1309D. Atlantis Massif: Proc. IODP, 304 ⁄ 305 College Station, TX (Integrated Ocean Drilling Program Management International, Inc.). doi:10.2204/iodp.proc.304305. 204.2009. Kastner M, Claypool G, Robertson G (2008) Geochemical constraints on the origin of pore fluids and gas hydrate distribution at Atwater Valley and Keathley Canyon, northern Gulf of Mexico. Marine and Petroleum Geology, v. 25, 860–72. Kharaka YK, Hanor JS (2007) Deep fluids in the continents: I. Sedimentary basins. In: Treatise on Geochemistry, 5.16 (ed. Drever JI), 1–48. Elsevier, Pergamon, Oxford. Lin G, Nunn JA (1995) Evidence for recent migration of geopressured fluids along faults in Eugene Island, Block 330 from estimates of pore water salinity. Gulf Coast Association of Geological Societies Transactions, 25, 419–24.
Spatial variations in the salinity of pore waters 93 Liu X, Flemings PB (2006) Passing gas through the hydrate stability zone at southern Hydrate Ridge, offshore Oregon. Earth and Planetary Science Letters, 241, 211–26. Manheim FT, Bischoff JL (1969) Geochemistry of pore waters from Shell Oil Company drill holes in the continental slope of the northern Gulf of Mexico. Chemical Geology, 4, 63–82. McDuff RE, Gieskes JM (1976) Calcium and magnesium profiles in DSDP interstitial waters: Diffusion or reaction? Earth Planetary Science Letters, 33, 1–10. McIntosh JC, Walter LM (2005) Volumetrically significant recharge of Pleistocene glacial meltwaters into epicratonic basins: constraints imposed by solute mass balances. Chemical Geology, 222, 292–309. Person M, McIntosh JC, Remenda V, Bense V (2007) Pleistocene hydrology of North America: the role of ice sheets in reorganizing groundwater systems. Reviews of Geophysics, 45, RG3007 ⁄ 2007, 1–28. Phillips OM (1991) Flow and Reaction in Permeable Rocks. Cambridge University Press, Cambridge. Pilcher RS, Blumstein RD (2007) Brine volume and salt dissolution rates in Orca Basin, northeast Gulf of Mexico. AAPG Bulletin, 91, 823–33. Ranganathan V, Hanor JS (1987) A numerical model for the formation of saline waters due to diffusion of dissolved NaCl in subsiding sedimentary basins with evaporites. Journal of Hydrology, 92, 97–120. Reitz A, Haeckel M, Wallman K, Hensen C, Heeschen K (2007) Origin of salt-enriched pore fluids in the northern Gulf of Mexico. Earth and Planetary Science Letters, 259, 266–82. Revil A, Cathles LM, Losh S, Nunn JS (1998) Electrical conductivity in shaly sands with geophysical applications. Journal of Geophysical Research, 103, B10, 23, 925–36. Roberts S, Nunn JA (1995) Episodic fluid expulsion from geopressured sediments. Marine and Petroleum Geology, 12, 195– 204. Ruppel C, Dickens GR, Castellini DG, Gilhooly W, Lizarralde D (2005) Heat and salt inhibition of gas hydrate formation in the northern Gulf of Mexico. Geophysical Research Letters, 32, LO4605, doi: 10.1029/2004GL021909.
Salvador A (1991) Origin and development of the Gulf of Mexico basin. In: The Gulf of Mexico Basin (ed. Salvador A), pp. 389– 444. Geological Society of America, Boulder, CO. Sarkar A, Nunn JA, Hanor JS (1992) Diffusion induced flow in a sloping aquifer with longitudinal salinity gradient. EOS, 73, 134. Spinelli GA, Giambalvo ER, Fisher AT (2004) Sediment permeability, distribution, and influence on fluxes in oceanic basement. In: Hydrogeology of the Oceanic Lithosphere (eds Davis EE, Elderfield H), pp. 151–88. Cambridge University Press, Cambridge. Sun R, Duan Z (2007) An accurate model to predict the thermodynamic stability of methane hydrate and methane solubility in marine environments. Chemical Geology, 244, 248–62. Szalkowski DS, Hanor JS (2003) Compositional systematics of produced waters from southwestern Louisiana. Gulf Coast Association of Geological Societies Transactions, 53, 798–806. Tishchenko P, Hensen C, Wallmann K, Shing C (2005) Calculation of the stability and solubility of methane hydrate in seawater. Chemical Geology, 219, 37–52. Ussler W, Paull CK (2007) Pore-water gradients in giant piston cores from the northern Gulf of Mexico. U.S. Geological Survey Open-File Report, 2004-1358, 8.1–8.24. Weimer P, Rowan MG, McBride BC, Kligfield R (1998) Evaluating petroleum systems of the northern Deep Gulf of Mexico through integrated basin analysis: an overview. AAPG Bulletin, 82, 865–77. Wilson A, Ruppel C (2007) Salt tectonics and shallow subseafloor fluid convection: models of coupled fluid-heat-salt transport. Geofluids, 7, 377–86. Wood WT, Hart PE, Hutchinson DR, Dutta N, Snyder F, Coffin RB, Gettrust JF (2008) Gas and gas hydrate distribution around seafloor seeps in Mississippi Canyon, Northern Gulf of Mexico, using multi-resolution seismic imagery. Marine and Petroleum Geology, 25, 952–9. Worrall DM, Snelson S (1989) Evolution of the northern Gulf of Mexico, with emphasis of Cenozoic growth faulting and the role of salt. In: The Geology of North America; An Overview (eds Bally AW, Palmer AR), pp. 128–56. Geological Society of America, Boulder, CO.
Faults and fault properties in hydrocarbon flow models T. MANZOCCHI, C. CHILDS AND J. J. WALSH Fault Analysis Group, UCD School of Geological Sciences, University College Dublin, Dublin, Ireland
ABSTRACT The petroleum industry uses subsurface flow models for two principal purposes: to model the flow of hydrocarbons into traps over geological time, and to simulate the production of hydrocarbon from reservoirs over periods of decades or less. Faults, which are three-dimensional volumes, are approximated in both modelling applications as planar membranes onto which predictions of the most important fault-related flow properties are mapped. Faults in porous clastic reservoirs are generally baffles or barriers to flow and the relevant flow properties are therefore very different to those which are important in conductive fracture flow systems. A critical review and discussion is offered on the work-flows used to predict and model capillary threshold pressure for exploration fault seal analysis and fault transmissibility multipliers for production simulation, and of the data from which the predictions derive. New flow simulation models confirm that failure of intra-reservoir sealing faults can occur during a reservoir depressurization via a water-drive mechanism, but contrary to anecdotal reports, published examples of production-induced seal failure are elusive. Ignoring the three-dimensional structure of fault zones can sometimes have a significant influence on production-related flow, and a series of models illustrating flow associated with relay zones are discussed. Key words: capillary threshold pressure, fault seal, faults, flow model, hydrocarbon migration, hydrocarbon production, transmissibility multiplier
Received 30 October 2009; accepted 2 February 2010 Corresponding author: T. Manzocchi, Fault Analysis Group, UCD School of Geological Sciences, University College Dublin, Dublin 4, Ireland. Email:
[email protected]. Tel: +353 1 716 2605. Fax: +353 1 716 2607. Geofluids (2010) 10, 94–113
INTRODUCTION This paper concerns the effects of faults on hydrocarbon flow in porous clastic sequences and the manner in which they are modelled in the oil industry. In these circumstances faults are permeability baffles or barriers and can influence the flow of hydrocarbon in three basic ways. First, the geometrical properties of a fault can modify flow paths by juxtaposing stratigraphically distinct permeable or impermeable units against each other. Second, faults can act as membranes by retarding or impeding cross-fault flow between juxtaposed permeable units because of the petrophysical properties of the fault rock. Third, they can provide fault-parallel conduits for flow between vertically separate flow units. Migration or production flow modelling is generally performed in grid-based models that include fault offset as part of the model structure (e.g. Fig. 1A). Hence juxtaposition effects of faults are explicitly represented in the models, and attention is often focused
on the accuracy with which the other two effects are included. Despite this disproportionate focus it is important to remember that considerable uncertainty often surrounds the fault geometry of a modelled volume particularly in regions of poor seismic quality and that the interpreted geometry may have suffered degradation during the model building process. The juxtaposition geometry is often considered to provide the basic plumbing of the model, and if this is wrong, it will almost certainly be impossible to achieve a correct model of the flow in the system by considering only the other two fault-related flow effects (e.g. Nybakken 1991; Clark et al. 2006; Jolley et al. 2007). Similarly, faults are only one part of the flow system, and a correct representation of the stratigraphic architecture and sedimentological flow properties is also an essential component of a flow model. Sedimentological characteristics and their interaction with faultrelated flow properties are, however, beyond the scope of this paper.
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Faults in hydrocarbon flow models 95
(A)
(B)
Fig. 1. Example of a portion of a reservoir production flow model. (A) The overall structure of the model contains explicitly the offsets associated with the larger (seismically visible) faults. The cells in a model are typically 50–100 m wide and 1–5 m thick, and contain appropriately up-scaled values of permeability, porosity and clay content (often, and in this example, using the net:gross ratio value, with non-net portion of each cell representing clay). (B) An Allan diagram of the fault at the centre of the model shown in (A). The coloured polygons represent average Shale Gouge Ratio (SGR) values for each cross-fault cell to cell connection and the black lines mark the edges of these connections. SGR is calculated at connection corners as a function of the model geometry and the clay content (Vshale) of the model cells.
Standard methods for considering flow properties of fault membranes for exploration and production purposes generally follow work-flows which differ only in terms of the input data and output properties (e.g. Bouvier et al. 1989; Yielding et al. 1997; Manzocchi et al. 1999; Jones & Hillis 2003). These methods involve the construction of fault surface projection maps (or Allan diagrams; Downey 1984; Allan 1989) for each fault of interest. Allan diagrams illustrate simultaneously the stratigraphy on both sides of a fault and therefore highlight potential juxtaposition seals and cross-fault flow paths. In exploration fault seal studies, Allan diagrams are generally constructed from gridded surfaces resulting from interpretation of seismic reflection data (e.g. Bouvier et al. 1989), while in production simulation they are most usefully constructed directly from the cells of the 3D flow simulation model, as the resultant diagrams show directly the connections between cells on either side of a fault (non-neighbour connections) to which fault transmissibility multipliers must be assigned (e.g. Fig. 1B). A record of all possible across-fault juxtapositions in a model is provided by the juxtaposition triangle plot which is a generalized Allan diagram (Bentley & Barry 1991; Childs et al. 1997; Knipe 1997). Allan diagrams provide a two-dimensional representation of a fault surface. However, faults are not single surfaces but zones containing a finite thickness of variably deformed rock and may locally comprise several such zones each of which accommodates a proportion of the displacement represented on the Allan diagram. Flow modelling in the oil industry is typically concerned with volumes on the oil-field to basin scales, and it is not possible to explicitly represent all of the internal complexity of fault zones (if
they were known) in models at this scale. Instead, the approach must be to represent those aspects of a fault which are thought to be critical to determining flow and which can be represented in a model at the scale of interest. It is therefore necessary to define fault proxy-properties which are based on simplified conceptual models of a fault zone that can be used to infer the distribution of relevant physical properties and can be calculated as continuously varying fields on the Allan diagrams. The most commonly applied proxy-property is the Shale Gouge Ratio (SGR; Yielding et al. 1997), defined as the fraction of clay within the sequence which has passed each point on a fault, and used to infer both fault rock capillary threshold pressure in an exploration context and fault permeability in a production context (e.g. Fig. 2).
MEMBRANE SEALS IN EXPLORATION Hydrocarbon migration is generally accepted as occurring in localized stringers driven principally by buoyancy forces and opposed by the capillary properties of the rock through which the migration occurs (e.g. England et al. 1987; Carruthers & Ringrose 1998). The most important fault property for membrane fault seal studies is the capillary threshold pressure of the fault-rock. The correlation between capillary threshold pressure and the clay content of laboratory samples (e.g. Fisher & Knipe 1998; Gibson 1998; Sperrevik et al. 2002), coupled with the assumption that SGR provides a proxy for the clay content of fault rock, provides the basis for the most commonly applied fault seal prediction methods (e.g. Yielding et al. 1997). The conceptual basis of the method is summarized in Fig. 3, for an
96 T. MANZOCCHI et al.
Fig. 2. Comparison of assumptions about cross-fault flow generally made in petroleum exploration (column 1) and production (column 2). Column 3 provides a physically more complete consideration of cross-fault flow which is sometimes needed for understanding flow effects in both disciplines. Routine representation of faults in this more complete way in routine flow modelling is an active area of research and development. Light grey, water; dark grey, oil. After Manzocchi et al. (2002).
accumulation formed on the down-thrown side of a membrane seal. Heterogeneity in the reservoir results in a variable SGR profile over the sand-on-sand juxtaposition window (Fig. 3B). Capillary threshold pressure is mapped onto the fault surface (Fig. 3C) using one of the industrystandard relationships (e.g. Sperrevik et al. 2002; Bretan et al. 2003) which link threshold pressure exponentially to SGR. Buoyancy-driven oil migration is stopped by the fault, and the accumulation forms behind it. As the accumulation (A)
(B)
(C)
grows, the capillary pressure at each point in the accumulation increases (dashed lines in Fig. 3C). Eventually, the capillary pressure at some point in the accumulation matches the capillary threshold pressure of the fault rock, allowing migration through the fault, and limiting the height of the fault-bounded accumulation. As Fig. 3C indicates, the critical leak point controlling the column height need not be at the top of the accumulation, at the hydrocarbon–water contact, or at the position with lowest SGR. (D)
Fig. 3. Cartoon illustrating the method for using Shale Gouge Ratio (SGR) as a column height predictor. (A) A hydrocarbon column (grey) on the downthrown side of a fault. The arrow indicates the migration direction and the dashed lines show hydrocarbon–water contacts during the filling of the reservoir. (B) SGR versus depth profile along the length of the sand-sand juxtaposition window. (C) Capillary threshold pressure (curve) is estimated as a direct function of SGR. Also plotted is the capillary pressure in the accumulation in its final state (thick solid line) and with two earlier hydrocarbon–water contacts (dashed lines corresponding to the dashed lines in A). The open circle indicates the depth at which the capillary pressure of the reservoir equals the capillary threshold pressure of the fault rock and therefore this position defines the total seal capacity of this fault. (D) Plot of SGR against reservoir capillary pressure, with the leak point indicated by the circle.
Faults in hydrocarbon flow models 97 Two contrasting approaches underlie the equations linking SGR to capillary threshold pressure. In one approach (e.g. Gibson 1994; Fristad et al. 1997; Yielding et al. 1997, in press; Yielding 2002; Bretan et al. 2003; Bretan & Yielding 2005), data from fault-bounded oil and gas fields with known hydrocarbon–water contacts (and hence known capillary pressure ⁄ depth data) are used to constrain empirical fault seal envelopes (e.g. Fig. 4A). Capillary pressure is plotted against SGR to define a field of data associated with a particular fault surface and, if column height is controlled by a fault membrane, one point in this field represents the critical leak point (e.g. Fig. 3D). When data from multiple reservoirs are combined, the fields of data provide a record of the capillary pressure that can be supported by a particular value of SGR. It seems (Fig. 4A) that faults with SGR = 0.3, for example, can support columns with Pc up to approximately 5 bars in one of the areas studied, and therefore PcT is inferred to be approximately 5 bars if SGR = 0.3. The data set shown in Fig. 4A is a global compilation of normal faults which were active at depths of less than 2 km, and for which across-fault pressure difference was assumed to be equal to the capillary pressure (Yielding 2002). The maximum burial depths of the faults were found to influence the seal envelopes, with deeper faults capable of sealing larger columns, and this dependence is contained in the equations linking capillary threshold pressure to SGR (Fig. 4). Discussion and updated versions of Fig. 4A are provided by Bretan et al. (2003) and Yielding et al. (in press). The second set of approaches is based on applying results from laboratory estimates of capillary threshold pressure which are obtained generally from mercury-air injection tests (e.g. Knipe et al. 1997; Gibson 1998; Sperrevik et al. 2002). The capillary threshold pressure, identified from the complete intrusion capillary pressure curve (Schowalter 1979; Knipe et al. 1997), is cross-plotted against the clay content of the sample (e.g. Fig. 4B) and a best-fit relationship is derived from the range of sample results. In the case of the equations from Sperrevik et al. (2002), capillary threshold pressure is related to clay content as a function of maximum burial depth and depth at the time of faulting. These are proxies for the more fundamental controls of fault rock permeability variability which are the deformation and temperature history (Fisher et al. 2003). As Yielding (2002) pointed out, the most significant threshold pressure from the laboratory data should be the lowest value at a particular clay content as the seal capacity of a fault is controlled by the leakiest position of fault rock. Therefore it should be expected that in the same way as the field data plot on the low capillary pressure side of the critical fault seal envelopes (Fig. 4A), laboratory data should plot on the high threshold pressure side. Indeed, when Yielding (2002) examined the laboratory data of Gibson (1998) in this way, this was found to
(A)
(B)
Fig. 4. Shale Gouge Ratio (SGR) fault seal calibrations. (A) Compilation of SGR versus capillary pressure adjacent to the fault for faults in a variety of extensional basins. Clouds of small points represent entire fault data (e.g. the line in Fig. 3D), while large circles correspond to trap-critical points (e.g. the circle in Fig. 3D). The data are coloured according to present day depth of burial (<3.0 km in blue, 3.0–3.5 km in red and 3.5–5 km in green) and the coloured lines represent fault seal envelopes representative of these three classes of faults. Redrawn from Yielding (2002). (B) Seal-failure envelopes from (A) compared with laboratory measurements of capillary threshold pressure obtained by Gibson (1998) shown as circles, and Sperrevik et al. (2002) shown as crosses, Hydrocarbon–water threshold pressures for these laboratory measurements have been calculated based on a hydrocarbon–water interfacial tension of 40 mN m)1 and a 30 contact angle. The dashed seal-failure envelopes for the deepest and shallowest depth classes are taken from Yielding et al. (in press) and derive from their latest analysis of the global compilation of field data. See text for discussion.
be the case (Fig. 4B). The measurements of Sperrevik et al. (2002), however, tend to contain lower PcT values for the same clay content compared to Gibson’s (1998) data and are therefore contradictory to this conceptual link between laboratory measurements of the capillary properties of fault
98 T. MANZOCCHI et al. there is a critical range in SGR between 0.2 and 0.4 where seal capacity is controlled by SGR, but there is no additional increase in seal capacity with SGR increasing beyond this point. Other studies suggest that the onset of fault seal may be even more abrupt (e.g. Jev et al. 1993; Fristad et al. 1997; Ottesen Ellevset et al. 1998). Childs et al. (2009a) for example, calibrated a critical PcT ⁄ SGR sealing envelope against known column heights in the Oseberg Syd area of the North Sea by randomly selecting possible fault seal envelopes and identifying those that provided the closest match to the observed column heights of several hydrocarbon accumulations when predicted using a raytracing migration modelling tool (Sylta 1991; Krokstad & Sylta 1996; Childs et al. 2002b). A summary of the modelling is given in Fig. 5. The most likely predictions of fault seal envelope deriving from the study all have a sharp increase in seal capacity at SGR = 0.2, and little sensitivity to the specific SGR value beyond this (Fig. 5B).
rocks to measurements of fault seal capacity made from measured column heights (Fig. 4B). The Sperrevik et al. (2002) data imply that the measured columns of Yielding (2002) are too large to be supported by the faults that support them. Revised versions of the seal envelopes derived from the field data (Yielding et al. in press; Fig. 4B) are much more complementary to the laboratory measurements of Sperrevik et al. (2002) than are the original envelopes, and the remaining discrepancy between the laboratory data of Gibson (1998) and Sperrevik et al. (2002) may be a function of different deformation and temperature histories of the fault rocks despite similar burial depth ranges. These comparisons therefore emphasize the importance stressed by many authors of using local rather than global data or calibrations when they are available. Figure 4A implies a gradual increase in fault seal capacity with increasing SGR; however, allied studies (e.g. Bretan et al. 2003; Yielding et al. in press) have concluded that (A)
(B)
(C)
(D)
(E)
Fig. 5. Summary of the migration modelling results of the Oseberg Syd area (Childs et al. 2009a). (A) Structure contour map of the top Tarbert reservoir horizon (scale in kilometres). Black lines outline fault polygons used in the definition of the migration model. Also shown are oil accumulations (dark green) which result from a particular migration model. The open circles show the locations of the seven calibration points used to compare model results with known hydrocarbon distributions, the numbers of these are referred to in the text. (B) Individual fault seal envelopes used in the study coloured according to the total mismatch in column height resulting from the seven accumulations (in metres). The red lines provide the best overall matches. The open circles show the critical points resulting from an individual analysis of each accumulation. (C) Misfit in column height predicted for accumulation 2 as a function of the intersection of seal envelope with the SGR axis (SGR-onset) and the slope of the seal envelope assumed in each model. (D) Ditto, but for accumulation 5. (E) Ditto, but for accumulation 6. (C–E) share the symbols shown to the right of (C). Positive misfits imply over-prediction of the column height. See text for discussion.
Faults in hydrocarbon flow models 99 Significantly, input predictive envelopes that provide the best calibration against the individual columns (i.e. a bestfit line to the seven data points in Fig. 5B) do not provide as good a match from the flow model as predictors with more abrupt increases in seal potential. This is because the SGR-controlled fault membrane seal leakage criteria contained in the flow model govern not only the columns that individual traps can support but also fault-controlled migration paths which can define whether individual traps are accessed or bypassed by hydrocarbon. For traps that are invariably on migration pathways, predictors that seal at low SGR values result in accumulations that are too large and vice versa (e.g. accumulation 2, Fig. 5C). This behaviour is comparable with the prediction that would be made from a reservoir-independent seal analysis, but is observed in only two of the seven accumulations examined (accumulations 2 and 7). Other accumulations exhibit behaviour that would never be predicted from case-by-case, static fault seal analysis as they result from situations where the migration path into, as well as out of, the reservoir is significantly influenced by fault seal. Specifically, the results show situations in which: (1) Intermediate predictors result in approximately the correct column, but predictors that are too sealing or too leaky predict reservoirs that are under-filled (e.g. Fig. 5D). This occurs in four cases (accumulations 1, 3, 4 and 5). (2) Too-leaky faults predict columns that are too high and too-sealing fault columns that are too low. This is observed in one case (accumulation 6, Fig. 3E), and is the opposite of the locally expected behaviour. It results from greater sensitivity of fault properties to reservoir charge rather than seal. These studies imply that the range in SGR over which fault seal becomes significant is similar to the divisions in the widely applied tripartite classification of clastic fault rocks into clay smears for fault rocks with Vshale >0.4, poorly sealing (unless comprehensively cemented) disaggregation zones and deformation bands at Vshale <0.15, and phyllosilicate-framework fault rocks in the intermediate Vshale range of 0.15 to 0.4 (Fisher & Knipe 1998). Phyllosilicate-framework fault rocks therefore seem to be the class of fault rock in which the precise clay content controls seal capacity. These types of fault rocks are the most poorly understood and, unlike the other two classes, are primarily defined as a function of fault rock clay content rather than texture (Knipe et al. 1997; Fisher & Knipe 1998, 2001). Texturally, they exhibit features of both disaggregation zones ⁄ deformation bands and clay smears (Knipe et al. 1997; Fisher & Knipe 1998) and therefore it is consistent with both empirical fault seal calibrations and laboratory data analysis to presume that fault seal is controlled by the preponderance of continuous clay within fault rocks, and that this continuity is achieved more or less abruptly in the
Vshale range occupied by phyllosilicate-framework fault rocks. This is supported by measurements of synthetic binary mixtures of quartz and kaolin which change from being quartz grain framework supported to clay matrix supported at Vshale values in the range of 0.2–0.3 (Crawford et al. 2008). Historically, the SGR algorithm has not been associated with fault seal studies in which the presence of continuous clay smears is considered to be the predominant seal mechanism. Instead, algorithms reliant on the definition of discrete clay layers from which the smears derive have been used as deterministic predictors of clay smear likelihood (e.g. Bouvier et al. 1989, Bentley & Barry 1991; Lindsay et al. 1993; Fulljames et al. 1997; Yielding et al. 1997; Færseth 2006). Recent research attempting to use these algorithms to honour outcrop observations of clay smears introduced a probabilistic flavour to one of these algorithms in order to honour the natural variability present (Childs et al. 2007). Subsequent investigations of the resultant Probabilistic Shale Smear Factor algorithm have revealed close and unexpected parallels with SGR (Childs et al. 2007; Yielding 2009). Unexpected robustness of the SGR approach has also been demonstrated by Dee et al. (2007a) in a study comparing it to predictions made from an approach which includes uncertainties in stratigraphical architecture and sand-shale cut-off values in consideration of juxtaposition seals, but which considers membrane seals to be irrelevant (James et al. 2004). One view of these results is that as soon as realistic uncertainties are introduced through the use of probabilistic methods, it does not really matter which predictive method is used as the uncertainties in the resultant predictions outweigh the determinism implicit in the underlying conceptual model. A corollary to this is therefore that the simplest method, which is based on empirical and deterministic links to SGR, is likely in many cases to be as reliable as more sophisticated predictive methods even though it may not necessarily contain the most plausible conceptual model of fault seal in a specific setting. Importantly, though, SGR should not be considered rigidly as a precise estimate of the clay content of a fault rock, but as a more flexible and abstract proxy-property with a proven track record in empirical fault seal analysis.
FAULT MEMBRANES IN PRODUCTION SIMULATION There is a greater consensus as to how fault membranes should be considered in production simulation rather than migration studies, largely because the controlling properties are inherently more predictable. This is because fault seal capacity relies on outliers of fault property distributions (the sealing capacity of a fault is controlled by the weakest point) while across-fault Darcy flow is controlled
100 T. MANZOCCHI et al. by average properties (specifically the arithmetic average of the ratio of fault rock permeability to thickness). The objective in production simulation is to develop a model of the reservoir which matches observed fluid pressures and production rates at the wells and more recently fluid distributions inferred from 4D seismic data. This model is then used to predict future production and to guide reservoir management decisions. Conventional flow simulation models are built as cornerpoint grids (Ponting 1989) in which the geometrical properties of faults are included explicitly in the simulation model through the definition of the lateral and vertical location of each of the eight corners of each grid-block. Fault membranes are included as transmissibility multipliers. The industry-standard work-flow for calculating geologically meaningful fault transmissibility multipliers as a function of information contained in the flow simulation model was defined by Manzocchi et al. (1999), and consolidated and automated earlier methods (e.g. Bentley & Barry 1991; Acharya et al. 1997; Knai & Knipe 1998; Manzocchi et al. 1998; Walsh et al. 1998a). Fault permeability is usually determined as a function of SGR, using published (e.g. Manzocchi et al. 1999; Sperrevik et al. 2002; Jolley et al. 2007; Crawford et al. 2008) or proprietary (e.g. Myers et al. 2007) predictors. The calibrations are based on databases comparing laboratory measurements of fault rock permeability and clay contents (e.g. Fig. 6A) and, like the fault seal calibrations discussed above, contain the assumption that SGR is equivalent to the clay content of the fault. They also generally include secondary dependencies such as fault throw, depth at the time of faulting, maximum burial depth and perhaps also more complex mechanical considerations of the evolving stress state (Myers et al. 2007). Fault rock thickness is usually predicted as a function of fault throw using a relationship based on compilations of field data (e.g. Robertson 1982; Hull 1988; Marrett & Allmendinger 1990; Childs et al. 1997). There are examples of both successful and unsuccessful reservoir simulation studies in which the fault transmissibility multipliers have been calculated during the model building process using these deterministic calibrations and, as expected by Manzocchi et al. (1999) and demonstrated by Jolley et al. (2007), the more locally relevant the calibration, the more accurate the resultant flow simulation results. Successful applications in the North Sea have been reported for the Heidrun Field (Knai & Knipe 1998), the Scott Field (Yielding 2002), the Snorre Field (Sverdrup et al. 2003), the North Cormorant, Brent and Pelican Fields (Jolley et al. 2007) and the Ringhorne Field (Myers et al. 2007). The input relationships linking fault rock permeability to SGR and fault rock thickness to fault throw used in these studies are shown in Fig. 6B,C if they are known. An influential early study on Cormorant Block IV (Bentley & Barry 1991) used a clay smear rather than the
SGR approach to achieving a geologically meaningful history-matched fault model of the reservoir. Examples where the methods have failed are also instructive. These include Rivenæs & Dart (2002), who found that in order to produce a match on the Brage and Oseberg Fields, the geologically derived transmissibility multipliers values had to be reduced by 2–3 orders of magnitude from those calculated using the standard permeability predictors (i.e. those of Manzocchi et al. 1999 and Sperrevik et al. 2002). Sverdrup et al. (2003) reported a good history match for oil in the Snorre Field using the original equation of Manzocchi et al. (1999), but could not achieve a match for gas. They ascribe this to a need for a two-phase consideration for the gas phase. Zijlstra et al. (2007) were unable to obtain a history match in two Rotliegend reservoirs using single-phase multipliers and therefore devised a means of mimicking some aspects of two-phase flow using a scheme they call the capillary entry height method, which allowed them to achieve much better results. Al-Busafi et al. (2005b) found that geologically derived transmissibility multipliers provided an improved history match on the Pierce Field (Central North Sea), but when the two-phase capillary entry height method was also applied, a yet closer match was achieved (see Fisher & Jolley 2007 for a discussion about this study). Methods for including faults in flow simulation models, as well as some of these successful and unsuccessful studies, have been reviewed recently by Fisher & Jolley (2007). They concluded that the most important aspect is ensuring that the correct juxtapositions are contained in the model and then that geologically reasonable permeability and thickness values are used to calculate transmissibility multipliers. In some situations such as structurally low, high net:gross reservoirs with cataclastic fault rocks, two-phase fault rock properties should also be considered as capillary properties may also be significant. Allied to the need for methods for including these properties is the requirement for two-phase measurements of fault rocks which are only now becoming available (e.g. Al-Hinai et al. 2008). The various methods for attempting to include twophase fault rocks routinely in flow simulation models can be split into three basic classes. The most direct approach is to use local grid refinements so that the relative permeability and capillary pressure curves of the fault rocks can be included explicitly in the simulation models (e.g. Ringrose & Corbett 1994; Manzocchi et al. 1998, 2002; Rivenæs & Dart 2002; Al-Busafi et al. 2005a; Berg & Øian 2007). A similar approach is sometimes advocated for including more realistic fault heterogeneity in flow models (e.g. Fredman et al. 2007). Such approaches may be useful for pilot studies or for validating upscaling methods but are impractical for full-field simulation as the number of cells required in the flow model quickly becomes computationally prohibitive. An additional problem is that widely contrasting flow
Faults in hydrocarbon flow models 101
(A)
(B)
(C)
properties in adjacent cells can sometimes produce erroneous results as discussed in the following section. The second approach aims to capture the two-phase effects of the faults by including them in the up-stream grid-block using standard or innovative dynamic up-scaling methods (e.g. Manzocchi 1999; Manzocchi et al. 2002, 2008a; Vaszi et al. 2005). These methods, however, are often perceived as difficult to implement (e.g. Fisher & Jolley 2007). The third class of approach is to use simplified methods that approximate some of the two-phase effects but unlike the second class of approaches do not attempt
Fig. 6. Fault properties for production simulation. (A) Summary of fault rock permeability data from the North Sea and Norwegian Continental shelf. The shaded area shows Rotliegendes cataclasites, the other four fields of data are Middle Jurassic fault rocks. The arrows indicate the main controls on fault permeability, i.e. increasing burial depth at the time of faulting in clean sandstones (arrow a), and increasing maximum postdeformation burial depth in impure sandstones (arrow b). Redrawn from Fisher & Knipe (2001). (B) Fault permeability relationships used in successful historymatched flow simulation studies. 1: Heidrun Field, Knai & Knipe (1998). 2: Snorre Field, Sverdrup et al. (2003). 3: Overpressure compartmentalization around the Tune Field, Childs et al. (2002a). 4: North Cormorant and Brent Fields, Jolley et al. (2007). (C) Fault rock thickness relationships used in successful history-matched flow simulation models. 1: Heidrun Field, Knai & Knipe (1998) – 1a is used for the Fangst Group, 1b is used for the rest of the model. 2: Two almost identical relationships used in the Brent Group reservoir study by Jolley et al. (2007) and in Ringhorne field by Myers et al. (2007). 3: An alternative relationship used by Jolley et al. (2007). The dots are individual fault rock thickness measurements from outcrop (Childs et al. 2009b).
to capture the full range of possible behaviour. These methods therefore are not applicable to all situations in which two-phase properties may be important and their suitability, which should be assessed on a case by case basis, may vary over the life of a field. The capillary entry height method discussed above (Zijlstra et al. 2007) is one such method. Hence there are now a couple of methods for routinely considering aspects of two-phase flow in fullfield flow simulation (e.g. Manzocchi et al. 2002, 2008a; Fisher & Jolley 2007; Zijlstra et al. 2007) but, as mentioned, neither of them is wholly satisfactory. Devising methods for improving the representation of multiphase fault rock properties in routine flow simulation modelling is therefore still an active area of research. Despite our understanding of the dependencies on the single-phase transmissibility multiplier, constant values for entire faults or fault segments are still often used in history matching studies (e.g. Christoffersen et al. 2008; Stephen & MacBeth 2008). A constant transmissibility multiplier assigned to a fault cutting a sequence with heterogeneous permeability implies that the fault is heterogeneous, but the resultant fault heterogeneity has no geological basis and is simply a reflection of the numerical dependencies contained in the transmissibility multiplier (Manzocchi et al. 1999). Whilst these reservoir models therefore may be history-matched, the implied fault heterogeneity cannot be correct and is almost certainly implausible and as such the models have little or no predictive value. A number of standard models are used by the research community to test history matching and up-scaling methods (e.g. Christie & Blunt 2001; Floris et al. 2001) but unfortunately such models are usually unfaulted. A more recent model (Peters et al. 2009) is faulted, but has all transmissibility multipliers set to 1.0 in the truth model and therefore does not include fault membrane effects. If methods for including fault properties more realistically in automated history
102 T. MANZOCCHI et al. matching are to be improved, an underlying requirement should be that some test models contain realistic faults. A recent threat to the integrity of fault seal methods applied in flow simulation modelling comes from the application of stair-stepping faults (Stern 2005). In certain situations this representation, which aliases a fault surface to the layers of a cellular model as well as to the edges, allows more realistic geometries to be modelled than a conventional representation in which faults are represented only at cell edges (e.g. Hoffman & Neave 2007). Often, however, stairstepped faults are not required but are still used. A stair-stepped fault does not contain the correct layer–layer juxtaposition geometry and throw distribution, and cannot be used in the current methods for automatically assigning fault transmissibility multipliers. The fine-scale details of the layer geometry associated with faults are compromised by this approach, leading to a loss in integrity of the principal control on accuracy of the flow model (i.e. the basic reservoir plumbing; Fisher & Jolley 2007; Jolley et al. 2007). Tectonic faults often occur in power-law size populations with progressively greater frequencies of progressively smaller faults (e.g. Cowie et al. 1996), a characteristic that allows models of seismically invisible faults to be constructed (e.g. Gauthier & Lake 1993; Maerten et al. 2006; Dee et al. 2007b). The largest sub-seismic faults can often be included in flow simulation models explicitly at the edges of grid-blocks, but smaller faults are both too frequent and too small to be represented in this way. Often these faults will have a negligible effect on flow as their offsets are too small to influence the plumbing of the reservoir. There are, however, situations in which small faults are sometimes thought to be important. Small faults can be important in the case of thin-bedded turbidite sequences in which the sandstone beds are stratigraphically poorly connected. In this situation small faults tend to connect up the sandstones even when shale smears are formed, and the faults are a boon to large-scale reservoir flow (Manzocchi et al. 2007). In the case of low permeability deformation bands in high porosity sandstone reservoirs (e.g. Aydin & Johnson 1983; Antonellini & Aydin 1994; Fowles & Burley 1994) the faults are detrimental to flow, and copious research exists aiming to provide generic solutions to the effective permeability of the rock volume as a function of the fault density, the permeability contrast between fault and undeformed sandstone, and aspects of the fault system geometry (e.g. Heath et al. 1994; Manzocchi et al. 1998, 2008b; Walsh et al. 1998b; Odling et al. 2004; Sternlof et al. 2004). In accordance with decades of reservoir engineering practice, the consensus is that despite often having permeabilities 2 or 3 orders of magnitude lower than the undeformed sandstone, pervasive deformation band networks will generally not be significant flow baffles at a reservoir scale because of low volumetric fractions of fault-rock (e.g. Fossen & Bale 2007).
MEMBRANE SEAL FAILURE A topic for which there is abundant anecdotal, but perhaps apocryphal, evidence is the notion that sealing faults sometimes break down during production. Fault seal breakdown may be invoked falsely where oil–water contacts change across a fault, i.e. the fault is a capillary seal, but the fault does not compartmentalize pressures in production. This apparent seal failure can arise because, for example, pressure communication is maintained in the water leg below the oil column. In a production context, changes in the fault transmissibility multipliers required to achieve a history match as production proceeds can be interpreted to indicate a change in fault properties with time. However in many cases this is more likely an indication of poor initial transmissibility multiplier definition. In their classic paper Bouvier et al. (1989) mention how ‘fault break down may occur along weak clay-smear seals as a result of increased pressure differentials from production on one side of a fault’, but do not give examples or explain what they mean. Are these clay smears weak in a mechanical or petrophysical sense? Hydraulic (i.e. mechanical) fault seal failure is generally considered a top-seal rather than fault seal issue in exploration (e.g. Watts 1987; Wiprut & Zoback 2000; Bjørlykke et al. 2005), and is often risked by contouring slip tendency on Allan diagrams to deduce the likelihood of fault reactivation and associated up-fault flow (e.g. Morris et al. 1996; Jones & Hillis 2003). Reservoir depletion can, in principle, induce stress paths capable of reactivating intra-reservoir faults and hence potentially of causing breakdown of their sealing integrity (Chan & Zoback 2002), but examples are elusive. It is not at all clear why pressure depletion should cause capillary seal failure. Many of the earlier explanations confuse capillary pressure with across-fault pressure and many (but not all) of the cartoon scenarios of capillary seal failure of Watts (1987), for example, do not work if capillary pressure is correctly defined as the difference in oil and water pressure adjacent to a fault, and not the difference in pressure across the fault. Are there any well-documented examples of production-induced seal failure? We have been able to find only three publications which attribute observed production behaviour to fault seal breakdown in a production context due to pressure depletion on one side of a fault. The first example is reported as a case of fault seal breakdown, yet the preferred interpretation of the pressure data by the authors of the paper (Jev et al. 1993) is that the behaviour was caused by across-fault flow in water-leg and there was little (if any) across-fault hydrocarbon flow. Flow in the water-leg was possible because the fault is not a seal below the hydrocarbon column as evidenced by decreasing clay smear factors. Hence this example does not show fault seal breakdown, but merely a difference in fault seal
Faults in hydrocarbon flow models 103 integrity on the fault surface and effects of flow of water through the fault. The second example (Davies et al. 2003) is a situation in which the hydrocarbon is in a low-pressure compartment and the water pressure gradient within the fault rock into the high pressure compartment supports the column. As the high pressure compartment is depressurized, the capillary pressure in the fault gradually increases until the capillary threshold pressure is exceeded and the fault becomes permeable to oil. Capillary fault seal failure with this pressure configuration is well understood (e.g. Underschultz 2007) but requires rather specific hydrodynamic conditions. The third reported case of fault seal breakdown is discussed by Gilham et al. (2005), and concerns a fault in the Shearwater High Pressure ⁄ High Temperature reservoir in the Central North Sea. The gas–water contact is the same across the fault, and evidence that the fault was a static seal at the onset of production was interpreted from different fluid compositions measured in samples acquired from wells drilled on each side the fault. Smalley et al. (2004) calculate that lengthy times are required to equilibrate fluid compositions in the absence of pressure gradients, and conclude that such differences are not necessarily evidence of compartmentalization. For example, diffusion-driven geochemical equilibration over distances in the order of only 500 m are possible in the 40 Myr since oil emplacement in the Ross field (Smalley et al. 2004). Initial charge of the Shearwater field occurred at a similar time (Winefield et al. 2005). Dynamic evidence for initial fault sealing and later breakdown in Shearwater derives from interpretation of a P ⁄ Z plot (Gilham et al. 2005). A simple summary of this type of analysis is provided by Zijlstra et al. (2007), and a detailed discussion of the pitfalls associated with it are given by Dake (2001). Gilham et al. (2005) interpreted a sharp change in linear slope in the Shearwater P ⁄ Z plot, but a close examination of the plot indicates that a curved trend may be as likely. This would imply that a low transmissibility (but non-sealing) fault is present throughout the period considered, and does not require any changes in the fault properties (Dake 2001; Zijlstra et al. 2007). In summary, therefore, the case for both static fault seal as well as for subsequent fault seal breakdown in the Shearwater field may not be unequivocal. We have modelled pressure depletion in a simple, onedimensional reservoir model in order to understand if and how we might obtain capillary fault seal failure in a hydrostatic regime. The model is of a gently dipping reservoir with a total column height of 120 m and a fault about 40 m above the oil water contact (Fig. 7). The capillary pressure at this height above the contact is 1.2 bars. The reservoir rock on both sides of the fault is assigned a permeability of 100 mD, and the fault, which is 10 cm thick and represented using discrete grid-blocks, has an intrinsic
Fig. 7. Geometry of the flow simulation model used to examine capillary seal failure. The model is 2 km long and has a total elevation range of 170 m. Water saturation refers to the initial, presimulation state of the model. Numbers 1–5 refer to positions in the model discussed in the text.
permeability of 0.01 mD. Oil and water relative permeability and capillary pressure functions for the fault rock and reservoir were prepared using standard functions (Manzocchi et al. 2002) with an imbibition curve used for the reservoir rock and a drainage curve for the fault rock (Fig. 8). The capillary threshold pressure of the fault rock is 3.77 bars, i.e. significantly larger than the capillary pressure in the reservoir adjacent to the fault. The fault is therefore a seal to oil at the start of the simulation. Pressure depletion in the up-dip compartment was simulated by simply producing oil at a constant (low) bottom hole pressure, and the pressures throughout the model run were monitored (Fig. 9). In the initial stages of the simulation, pressures in the up-dip compartment gradually decrease, but do not change in the down-dip compartment. The fault seal appears robust. After about 200 days, however, pressures in the up-dip compartment remain approximately constant, and the down-dip compartment gradually depressurizes. The seal appears to have become breached during production of the up-dip compartment.
(A)
(B)
Fig. 8. (A) Capillary pressure and (B) relative permeability curves used for reservoir rock and fault rock.
104 T. MANZOCCHI et al.
(A)
(B)
(C)
Fig. 9. Maps of pressure in the model at different times in the simulation run. The onset of depressurization in the down-dip compartment occurs between 180 and 230 days.
(D)
Despite no water being injected into the down-stream compartment, the mechanism for the capillary seal failure in the model is water-drive leakage (e.g. Heum 1996). We explain the process with reference to pressures, saturation and flow rates observed at five positions in the model (Fig. 10). Position 1 is located in the down-dip compartment close to the oil–water contact (Fig. 7). Position 2 is located in the reservoir immediately downdip of the fault. Position 3 is the centre of the fault and Position 4 is in the reservoir immediately up-dip of the fault. Position 5 is at the centre of the up-dip compartment. At the beginning of the simulation, depressurization of the up-dip compartment causes a pressure drop across the fault (Fig. 10A). As the fault is entirely water saturated and is therefore permeable to water, and because there is also some mobility of water in both reservoir compartments adjacent to the fault, water flows through the fault into the up-dip compartment in response to this pressure drop (Fig. 10C). It can do so only at very low rates however, as the relative water permeabilities in the reservoir are very low. Therefore the across-fault pressure differential grows (Fig. 10A). During this early period water is flowing out of the region immediately upstream of the fault (i.e. position 2) and into the fault. This water is predominantly replaced by oil, as oil is both the more abundant fluid in this location and is more mobile than water (Fig. 8B). Hence the water saturation at position 2 gradually decreases, and capillary pressure therefore gradually increases in response following
(E)
(F)
Fig. 10. Model behaviour at positions 1–5 (Fig. 7) through time. (A–D) shows time with a logarithmic scale, while (E–F) shows only the 1-day period in which capillary seal failure occurs.
the imbibition capillary pressure curve of the reservoir rock (Fig. 8A). This increase can clearly be seen after about 100 days (arrowed in Fig. 10B).
Faults in hydrocarbon flow models 105 Eventually (at 190.1 days) the capillary pressure at position 2 equals the capillary threshold pressure of the fault rock, and the fault becomes permeable to oil as well as water. Seal failure by water-drive leakage has occurred. For a brief period (between 190.15 and 190.4 days, Fig. 10E,F) significant water flows out of the fault (Fig. 10E) as oil (which is very mobile at position 2) can flow into the fault much more easily than was previously possible for water, and this oil rapidly replaces the water in the fault. This is accompanied by a corresponding increase in capillary pressure (Fig. 10B) and oil relative permeability within the fault. Like the reservoir rock on either side of the fault, the fault rapidly becomes (at 190.4 days, Fig. 10F) much more permeable to oil than to water, and the depressurization of the down-dip compartment is henceforth facilitated by higher flow rates (of oil) through the fault. On a practical level, performing this modelling in the Eclipse flow simulator (Schlumberger 2005) has required that modifications be made to the two-phase properties of the cell immediately upstream of the fault rock, even though the model contains cells with explicit fault rock relative permeability and capillary pressure curves. Without this modification, which is best thought of as relative permeability pseudoization (e.g. Barker & Thibeau 1997) even though the grid resolution has not been changed, flow across the reservoir ⁄ fault interface and into the fault does not honour the capillary threshold pressure of the fault rock, and water-drive leakage occurs at a capillary pressure that is too low. Details of the required modelling corrections are outside the scope of this paper. The implication however is that published models of water-drive leakage that have not recognized this artefact will contain incorrect seal breakdown which, unlike in the models pre-
(A)
Fig. 11. Cartoons comparing conceptual models of fault zones. (A) A fault in a sand-shale sequence (after Walsh et al. 1998a). (B) A fault in a strong ⁄ weak sequence (after Wibberley et al. 2008). (C) A fault annotated according to the fault core ⁄ damage zone conceptualization of Caine et al. (1996) (after Berg 2004). All cartoons have a normal sense of offset with the downthrown side on the left; (B) also has a dextral strike-slip component of displacement.
sented in this paper, does not honour the capillary threshold pressure of the seal (e.g. Al-Busafi et al. 2005a). The artefact also implies that water-drive leakage would be a model outcome rather more often than it should be. The model confirms that capillary seal failure, via the water-drive leakage mechanism, is possible simply through introducing a pressure gradient across a fault, but gives no indication as to how prevalent the phenomenon might be. Certainly the mechanism will be facilitated by water flow, either from hydrodynamic charge in an exploration setting or through water injection in a production setting (e.g. Heum 1996). Despite the general validity of the mechanism and persistence of unpublished anecdotes, however, there is little evidence in the public domain to support the notion of widespread depletion-induced failure of intra-reservoir sealing faults.
FLOW THROUGH FAULT ZONES We have dealt above with faults using the conventional approach adopted in fault seal studies and in migration and production flow modelling, which is to consider faults as membranes. Faults are actually complex, strongly heterogeneous and anisotropic volumes of more or less deformed rock. That there is no simple way of systematically describing fault zone structure and content, however, is demonstrated by the preponderance of independent, sometimes contradictory, more or less quantitative frameworks that attempt to do so (e.g. Caine et al. 1996; Harris et al. 2003; Kim et al. 2004; Braathen et al. 2009; Childs et al. 2009b; Fig. 11). We have discussed elsewhere (Manzocchi et al. 2008a; Childs et al. 2009b) why our preferred framework for defining fault zone structure for flow modelling considerations is centred around definition of
(B)
(C)
106 T. MANZOCCHI et al.
(A) (C)
(B)
Fig. 12. Model showing flow paths associated with flow within the top layer (Layer 1) of the hangingwall compartment of a multi-layered, faulted model. The model comprises a layer-cake sequence of permeable and impermeable layers, each three grid-blocks thick – for the sake of clarity, only the central gridblocks of each permeable layer are shown in (A), with grid-block coloured by pressure, normalized to the total pressure difference between the injector and producer well. (B) The middle permeable layer in the model highlights the role of connections across the relay-bounding faults at causing pressure perturbations in other layers. (C) Streamlines deriving from the pressure solver, indicating flow paths in every permeable layer in the model. The model does not contain fault-rock, hence the across-fault transmissibilities are a simple function of the connection area and permeabilities of the juxtaposed layers. Simulation performed using 3dsl from Streamsim. Models are 50 · 69 · 39 grid-blocks representing a 500 m · 500 m · 78 m volume. Fault throw is 20 m. Vertical exaggeration ·10 throughout.
the partitioning of displacement within the zone. Essentially it is because in layered clastic rocks, displacement partitioning defines the local plumbing around and within the fault zone which, as discussed in previous sections, is the principal control on flow in both a migration and production context. Outcrop studies describing fault zone structure and emphasizing likely effects of displacement partitioning on flow abound (e.g. Peacock & Sanderson 1994; Antonellini & Aydin 1995; Childs et al. 1996, 1997; Foxford et al. 1998; Shipton & Cowie 2001; Aydin & Eyal 2002; Eichhubl et al. 2005; Færseth et al. 2007; Rotevatn et al. 2007). Field descriptions of exhumed faults in which ancient hydrocarbon, mineralizing or diagenetic fluid flow paths have been identified, are testament to the strong flow localization caused by the three-dimensional heterogeneous structure of fault zones (e.g. Cox 1995; Aydin 2000; Garden et al. 2001; Boles et al. 2004; Shipton et al. 2004; Eichhubl et al. 2009; Dockrill & Shipton in press). Our understanding of spatial and temporal flow heterogeneity arising from fault zone complexity is enhanced further from observations and measurements of present-day natural fault-related flow (e.g. Revil & Cathles 2002; Miller et al. 2004; Haney et al. 2005; Fairley 2009). Faults, obviously, are complicated and are responsible for complicated
flow paths. The requirement for modelling flow in the oil industry, therefore, is to first establish whether the complications are sufficiently common and influential to warrant consideration in particular industrial flow applications and, if they are, to establish how to include the most important of them in a practical manner within the extremely large bounds of uncertainty present in any particular situation. Several recent studies have attempted to understand quantitatively the significance on flow of considering faults as truly three-dimensional zones rather than as twodimensional membranes, particularly in a production context (e.g. Fredman et al. 2007; Harris et al. 2007; Manzocchi et al. 2008a). These studies have drawn no clear conclusions however: sometimes fault zone complexity is significant and sometimes it is not. The reason for this ambiguity is that the dependence on flow of fault zone complexity is hugely case-specific, and depends crucially on the sequence, the fault system, and the scale at which flow is considered. To illustrate this point we present models containing the most significant departure a fault can take from being a 2D membrane, which is when it contains an unbreached relay zone providing continuity of each layer across the fault along a dipping, unfaulted, ramp.
Faults in hydrocarbon flow models 107
(A)
(C)
(B)
(D)
Fig. 13. Models comparing sensitivity of across-fault flow in sequences with different KV ⁄ KH ratio to the presence of a relay ramp. (A) Model containing an unbreached relay. (B) Equivalent model without the relay. No fault-rock is present. Vertical exaggeration ·10. Both models contain a horizontal producer in the bottom footwall layer, and a horizontal injector in the top hangingwall layer. For the models without a relay, flow between the wells must pass through the low permeability horizons; however, when the relay is present juxtapositions across the ramp allow flow exclusively in the high permeability layers. (C, D) Resident fluid (filled circles) and injected fluid (open circles) production profiles for the model with the relay (grey lines) and without it (ornamented lines), for two different KV ⁄ KH ratios. At the lower KV ⁄ KH, the relay is absolutely critical for flow across the fault. No curve is shown for the injected fluid in the ‘no-relay’ model in (D) as the production rate for this is zero throughout the time of the simulation. Simulation performed using 3dsl from Streamsim. Model dimensions as Fig. 12.
The most important flow characteristic of relay ramps is the thoroughly 3D connectivity they provide a reservoir. The prevailing view of an unbreached relay is essentially of a hole in a fault providing a flow path across it (e.g. Rotevatn et al. 2007). Although in many circumstances this may be the most important flow path provided by a relay, the view however represents a gross oversimplification of the complex juxtapositions produced. In addition to these uninterrupted flow paths in each layer, an unbreached ramp provides across-fault juxtaposition between every layer of the sequence displaced by the fault. Figure 12 shows a simple model in which flow in a single layer on the hangingwall of a fault causes pressure perturbations in every layer in both the footwall and hangingwall sides of the fault. Relays therefore represent extremely efficient conduits for vertical pressure communication, as potential flow-paths are formed between every pair of layers in a permeable sequence from the lowermost hangingwall to the uppermost footwall layers, provided only that the rampbounding faults are not impermeable, and that the vertical separation between layers is less than the displacement of the parent fault. In the example discussed above (Fig. 12), the relay provides vertical communication across impermeable layers,
mimicking up-fault flow but actually doing so via multiple stair-casing pathways across the fault made available by the locally high fault displacement gradients produced by the relay zone. If some model layers were not impermeable the relay would be less influential as vertical flow would not rely on the relay alone. The ratio between vertical and horizontal permeability (KV ⁄ KH) is therefore an important variable on the significance of relays on production. The models summarized in Fig. 13 compare across-fault flow rates in layered sequences at two different KV ⁄ KH ratios. At 4 orders of magnitude permeability contrast (Fig. 13C), production rates are similar irrespective of whether or not the fault contains a relay. At 7 orders of magnitude contrast however, the total across-fault flow is approximately 100 times greater if a relay is present (Fig. 13D). During a history matching process models representing the fault as a membrane rather than containing a relay could easily be made to reproduce the production and injection rates observed in the models containing relays by modifying the transmissibility multipliers applied to the membrane (Fig. 13). However, the details of the saturation and pressure distributions (Fig. 14) would not be honoured by such a treatment. This example therefore illustrates the scale-specific nature of the influence of relays on
108 T. MANZOCCHI et al.
(A)
(B)
(C)
(D)
flow: if the correct across-fault flow rate is required, then the relay makes little difference when KV ⁄ KH is 10)4 (Fig. 13). If the pressures and saturations in the vicinity of the fault are required, however, the relay is important (Fig. 14) and should be included in the flow model. Novel geometrical upscaling methods have been developed for including the single-phase transmissibility effects of relays and other forms of fault zone structure in conventional flow models (i.e. those in which faults are represented as membranes) by defining up-fault non-neighbour connections with transmissibilities calculated from high-resolution 3D fault zone models (Manzocchi et al. 2008a). Unlike the conceptually simpler but in practice much more demanding approach of including fault zone structure through explicit geometrical representations in locally refined grids at faults (e.g. Fredman et al. 2007, 2008),
Fig. 14. Details of the models shown in Fig. 13. (A–D) show the combinations of the two KV/KH ratios considered, and the presence or absence of relay ramps. Pressure and saturation at 1000 days show broadly similar distributions for the two models with a higher KV ⁄ KH, but contrast strongly when KV ⁄ KH ratio is low. The local orientation of the streamlines indicates flow direction, and the ‘time of flight’ colour-coding of the streamlines shows the time at which the tracer flood-front would reach each position on the streamline (e.g. Datta-Gupta 2000). At a high KV ⁄ KH ratio there are a significant number of vertical streamline vectors in the model containing relays, and therefore the relay is not essential for permitting vertical flow. At low KV ⁄ KH ratio all streamlines in the model with relays are horizontal, indicating that the flow characteristics of this model are dominated by the ramp. Although vertical streamlines are present in the model without the ramp, these have time of flight >105 days even close to the injector well, indicating the difficulty of draining these compartments. In the higher KV ⁄ KH ratio model, time of flight >105 days occurs only in the upper most hangingwall and lowermost footwall compartments. Vertical exaggeration ·10 throughout.
geometrical upscaling allows fault zone structure to be represented and flow-simulated in field-scale reservoir models (Manzocchi et al. 2008a).
CONCLUDING REMARKS The discussion has focused on flow across fault membranes using the established conceptual framework that hydrocarbon migration and trapping is dominated by capillary and gravity forces, and that hydrocarbon production is dominated by viscous forces (Fig. 2). This standard approach to the study of the trapping of hydrocarbons is based on capillary theory applied to invasion of a non-wetting fluid (hydrocarbon) into a water wet reservoir ⁄ seal system, as described in an earlier section. Application of the theory relies on the assumption that the capillary forces which
Faults in hydrocarbon flow models 109 retard migration are maintained over geological timescales. Some workers have suggested that once a water-wet seal is breached the leak pathway may become oil wet (Aplin & Larter 2005; Larter et al. 2005). If this is the case then the appropriate approach to modelling fault seal in hydrocarbon migration would require consideration not only of the capillary threshold pressure but also of the relative permeability of the seal to hydrocarbon following leakage. In basins where hydrocarbon is being generated at the present day, column heights are likely to be at least partly (and potentially entirely) dependent on a balance between charge and seal permeability rather than exclusively on threshold pressure. It has not been an objective of this paper to review all possible forms of fault-related fluid flow in hydrocarbonbearing basins or to discuss all approaches to modelling flow. We have deliberately not addressed hydrodynamic effects in exploration, where many of the issues associated with two-phase fault rock properties in a seal-breaching context as discussed in this paper are relevant (e.g. Heum 1996; Bjørkum et al. 1998, 1999a,b; Clayton 1999; Rodgers 1999; Grauls et al. 2002; Brown 2003; Teige et al. 2006; Underschultz 2007). Nor have we discussed the fate of reservoirs once seal capacity has been exceeded. In the case of faulted top seal failure for example, quantitative treatments of up-fault flow become important (e.g. Moretti 1998; Faulkner & Rutter 2001; Wilkins & Naruk 2007). The focus has been on issues associated with standard industry challenges, assumptions and modelling approaches. We have tried as far as possible to illustrate fault seal mechanisms, modelling methods and dynamic processes with published examples. This has proved somewhat frustrating and we do not think it is an exaggeration to state that progress in fault seal evaluation is to some extent hampered by the prevalence of anecdotal rules-ofthumb as opposed to well-documented examples. The methods used to evaluate fault seals described above, or variants on them, are standard approaches within the oil industry, but there are very few published accounts that show success or failure of the methods on real fields. The true test and value of the methods is in their capacity to replicate and ultimately predict the effects of faults on fluid flow in the subsurface. Data to conduct these tests are available within oil companies, and although retrospective studies are sometimes conducted, the results of these studies are generally considered to provide a competitive advantage and therefore often remain confidential. Faults have extremely complex 3D structure and petrophysical property distributions. Nevertheless, methods have been developed which are thought to be capable of capturing the most important properties of faults relevant to models of hydrocarbon migration and production. Although it is possible to further refine the modelling methods, we believe that at the moment further progress
requires improvement in both the input to models and the rigour with which currently available methods are applied. In terms of input for example, there is a lack of truly quantitative data related to the internal structure of fault zones. While fault rock thicknesses are known to vary by as much as 3 orders of magnitude on a fault of a given displacement, there are practically no data expressing the variability in thickness over fault surfaces or the correlation lengths of this variability. In this sense, modern structural and fault seal analysis and modelling is guiding the next generation of questions that need to be answered through more focused outcrop studies.
ACKNOWLEDGEMENTS We thank Quentin Fisher and Steve Jolley for their helpful reviews, and Graham Yielding for his comments, data and unpublished article. Schlumberger and Streamsim are thanked for providing use of their Eclipse and 3dsl flow simulators.
REFERENCES Acharya UK, Eisenberg RA, Brenneman RJ, Adeogba AA (1997) Integrated fault seal characterisation for reservoir management. SPE 38090. Presented at the Asia Pacific Oil and Gas conference and Exhibition, Kuala Lumpur, April 14–16. Al-Busafi B, Fisher QJ, Harris SD (2005a) The importance of incorporating the multi-phase flow properties of faults rocks into production simulation models. Marine and Petroleum Geology, 22, 365–74. Al-Busafi B, Fisher QJ, Harris SD, Kendall M (2005b) The impact of fault representation on history match and future generated seismic impedance response in reservoir models – case study for Pierce Field, North Sea. SPE 93429. Proceeding of the SPE Europec ⁄ EAGE Annual Conference, Madrid, Spain, June 13–16. Al-Hinai S, Fisher QJ, Al-Busafi B, Guise P, Grattoni CA (2008) Laboratory measurements of the relative permeability of cataclastic fault rocks: an important consideration for production simulation modelling. Marine and Petroleum Geology, 25, 473–85. Allan US (1989) Model for hydrocarbon migration and entrapment within faulted structures. American Association of Petroleum Geologists Bulletin, 73, 803–11. Antonellini M, Aydin A (1994) Effect of faulting on fluid flow in porous sandstones: petrophysical properties. AAPG Bulletin, 78, 355–77. Antonellini M, Aydin A (1995) Effect of faulting on fluid flow in porous sandstones: geometry and spatial distribution. AAPG Bulletin, 79, 642–71. Aplin AC, Larter SR (2005) Fluid flow, pore pressure, wettability and leakage in mudstone cap rocks. In: Evaluating Fault and Cap Rock Seal (eds Boult P, Kaldi J), American Association of Petroleum Geologists Hedberg Series, 2, 1–12. Aydin A (2000) Fractures, faults, and hydrocarbon entrapment, migration, and flow. Marine and Petroleum Geology, 17, 797– 814. Aydin A, Eyal Y (2002) Anatomy of a normal fault with shale smear: implications for fault seal. AAPG Bulletin, 86, 1367– 81.
110 T. MANZOCCHI et al. Aydin A, Johnson AM (1983) Analysis of faulting in porous sandstones. Journal of Structural Geology, 5, 19–31. Barker J, Thibeau S (1997) A critical review of the use of pseudorelative permeabilities for upscaling. SPE 35491. SPE Reservoir Engineering, 12, 138–43. Bentley MR, Barry JJ (1991) Representation of fault sealing in a reservoir simulation: Cormorant Block IV, UK North Sea. SPE 22667. Proceeding of the 66th Annual Technical Conference and Exhibition, Dallas, Texas, 119–26. Berg SS (2004) The architecture of normal fault zones in sedimentary rocks: analysis of fault core composition, damage zone asymmetry, and multi-phase flow properties. PhD Thesis, University of Bergen. Berg SS, Øian E (2007) Hierarchical approach for simulating fluid flow in normal fault zones. Petroleum Geoscience, 13, 25– 35. Bjørkum PA, Waldehaug O, Nadeau P (1998) Physical constraints on hydrocarbon leakage and trapping revisited. Petroleum Geoscience, 4, 237–9. Bjørkum PA, Walderhaug O, Nadeau P (1999a) Reply to discussion of the paper ‘‘Physical constraints on hydrocarbon leakage and trapping revisited by PA Bjørkum et al.’’ by CJ Clayton. Petroleum Geoscience, 5, 99–101. Bjørkum PA, Walderhaug O, Nadeau P. (1999b). Reply to discussion of the paper ‘‘Physical constraints on hydrocarbon leakage and trapping revisited by PA Bjørkum et al.’’ by S Rodgers. Petroleum Geoscience, 5, 421–3. Bjørlykke K, Hoeg K, Faleide JI, Jahrne J (2005) When do faults in sedimentary basins leak? Stress and deformation in sedimentary basins; examples from the Haltenbanken, offshore Norway. AAPG Bulletin, 89, 1019–31. Boles JR, Eichhubl P, Garven G, Chen J (2004) Evolution of a hydrocarbon migration pathway along basin-bounding faults: evidence from fault cement. AAPG Bulletin, 88, 947–70. Bouvier JD, Kaars-Sijpesteijn CH, Kleusner DF, Onyejekwe CC, van der Pal RC (1989) Three-dimensional seismic interpretation, and fault sealing investigations, Nun River field, Nigeria. American Association of Petroleum Geologists Bulletin, 73, 397– 1414. Braathen A, Tveranger J, Fossen H, Skar T, Cardozo N, Semshaug SE, Bastesen E, Sverdrup E (2009) Fault facies and its application to sandstone reservoirs. AAPG Bulletin, 93, 891– 917. Bretan P, Yielding G (2005) Using buoyancy pressure profiles to assess uncertainty in fault seal calibration. In: Evaluating Fault and Cap Rock Seal (eds Boult P, Kaldi J), American Association of Petroleum Geologists Hedberg Series, 2, 151–62. Bretan P, Yielding G, Jones H (2003) Using calibrated shale gouge ratio to estimate hydrocarbon column heights. AAPG Bulletin, 87, 397–413. Brown A (2003) Capillary effects on fault-fill sealing. AAPG Bulletin, 87, 381–95. Caine JS, Evans JP, Forster CB (1996) Fault zone architecture and permeability structure. Geology, 24, 1025–8. Carruthers D, Ringrose PS (1998) Secondary oil migration: oilrock contact volumes, flow behaviour and rates. In: Dating and Duration of Fluid Flow and Fluid-Rock Interaction (ed. Parnell J), Geological Society of London Special Publication, 144, 205– 20. Chan AW, Zoback MD (2002) Deformation analysis in reservoir space (DARS): a simple formalism for prediction of reservoir deformation with depletion. SPE ⁄ ISRM 78174. Presented at the SPE ⁄ ISRM Rock Mechanics Conference, Irving, TX, October 20– 23.
Childs C, Watterson J, Walsh JJ (1996) A model for the structure and development of fault zones. Journal of the Geological Society, 153, 337–40. Childs C, Walsh JJ, Watterson J (1997) Complexity in fault zone structure and implications for fault seal prediction. In: Hydrocarbon Seals: Importance for Exploration and Production (eds Møller-Pedersen P, Koestler AG), Norwegian Petroleum Society Special Publication, 7, 61–72. Childs C, Manzocchi T, Nell PAR, Walsh JJ, Strand JA, Heath AE, Lygren TH (2002a) Geological implications of a large pressure drop across a small fault in the Viking Graben. In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale R), Norwegian Petroleum Society Special Publication, 11, 127–39. Childs C, Sylta Ø, Moriya S, Walsh JJ, Manzocchi T (2002b). A method for including the capillary properties of faults in hydrocarbon migration models In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale R), Norwegian Petroleum Society Special Publication, 11, 187–201. Childs C, Walsh JJ, Manzocchi T, Strand J, Nicol A, Tomasso M, Schopfer MPJ (2007) Definition of a fault permeability predictor from outcrop studies of a faulted turbidite sequence, Taranaki, new Zealand. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 235–58. Childs C, Sylta O, Moriya S, Morewood N, Manzocchi T, Walsh JJ, Hermanssen D (2009a) Calibrating fault seal using a hydrocarbon migration model of the Oseberg Syd area, Viking Graben. Marine and Petroleum Geology, 26, 764–74. Childs C, Manzocchi T, Walsh JJ, Bonson C, Nicol A, Schopfer MPJ (2009b) A geometric model of fault zone and fault rock thickness variations. Journal of Structural Geology, 31, 117–27. Christie MA, Blunt MJ, (2001) Tenth SPE comparative solution project: a comparison of upscaling techniques SPE 72469. SPE Reservoir Evaluation & Engineering, 4, 308–17. Christoffersen KR, Quin JG, Zweigel P, Hansen OR, Zaostrovski A (2008) Continuous history matching of a HPHT gas condensate field during the first two years of production. SPE 113861. Presented at the 2008 SPE Europec ⁄ EAGE Annual Conference and Exhibition, Rome, June 9–12. Clark SM, Littler M, Burley SD, Williams GD, Hughes D, Coogan S (2006) Modelling the effects of stratigraphic uncertainty in fault seal and trap-fill in faulted structures. Petroleum Geoscience, 12, 143–56. Clayton CJ (1999), Discussion of the paper ‘‘Physical constraints on hydrocarbon leakage and trapping revisited by PA Bjørkum et al.’’. Petroleum Geoscience, 5, 99–101. Cowie PA, Knipe RJ, Main IG (eds) (1996) Scaling laws for fault and fracture populations – analyses and applications. Journal of Structural Geology, 18, 135–383. Cox SF (1995) Faulting processes at high fluid pressures: an example of fault valve behaviour from the Wattle Gully Fault, Victoria, Australia. Journal of Geophysical Research Solid Earth, 100, 12841–59. Crawford BR, Faulkner DR, Rutter EH (2008) Strength, porosity, and permeability development during hydrostatic and shear loading of synthetic quartz-clay fault gouge. Journal of Geophysical Research, 113, B03207. Dake LP (2001). The Practice of Reservoir Engineering (revised edition). Elsevier, Amsterdam. Datta-Gupta A (2000) Streamline simulation: a technology update. Journal of Petroleum Technology, 12, 68. Davies RK, An L, Jones P, Mathis A, Cornette C (2003) Faultseal analysis South Marsh Island Field, Gulf of Mexico. AAPG Bulletin, 87, 479–91.
Faults in hydrocarbon flow models 111 Dee SJ, Yielding G, Freeman B, Bretan P (2007a) A comparison between deterministic and stochastic fault seal techniques. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 259–70. Dee SJ, Yielding G, Freeman B, Healy D, Kuznir NJ, Grant N, Ellis P (2007b) Elastic dislocation modelling for prediction of small-scale fault and fracture network characteristics. In: Fractured Reservoirs (eds Lonergan L, Jolly RJH, Rawnsley K, Sanderson DJ), Geological Society of London Special Publication, 270, 139–55. Dockrill B, Shipton ZK (in press) Structural controls on leakage from a natural CO2 geologic storage site: Central Utah, U.S.A. Journal of Structural Geology. Downey MW (1984) Evaluating seals for hydrocarbon accumulations. American Association of Petroleum Geologists Bulletin, 68, 1752–63. Eichhubl P, D’Onfro PS, Aydin A, Waters J, McCarty DK (2005) Structure, petrophysics and diagenesis of shale entrained along a normal fault at Black Diamond Mines, California – implications for fault seal. AAPG Bulletin, 89, 1113–37. Eichhubl P, Davatzes NC, Becker SP (2009) Structural and diagenetic control of fluid migration and cementation along the Moab fault, Utah. AAPG Bulletin, 93, 653–81. England WA, Mackenzie AS, Mann DM, Quigley TM (1987) The movement and entrapment of petroleum fluids in the subsurface. Journal of the Geological Society, 144, 327–47. Færseth RB, Johnsen E, Sperrevik S (2007) Methodology for risking fault seal capacity: implications of fault zone architecture. AAPG Bulletin, 91, 1231–46. Færseth RB (2006) Shale smear along large faults: continuity of smear and the fault seal capacity. Journal of the Geological Society, 163, 741–51. Fairley JP (2009) Modeling fluid flow in a heterogeneous, faultcontrolled hydrothermal system. Geofluids, 9, 153–66. Faulkner DR, Rutter EH (2001) Can the maintenance of overpressured fluids in large strike-slip fault zones explain their apparent weakness? Geology, 29, 503–6. Fisher QJ, Jolley SJ (2007) Treatment of faults in production simulation models. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 219–33. Fisher QJ, Knipe RJ (1998) Fault sealing processes in siliciclastic sediments. In: Faulting and Fault Sealing in Hydrocarbon Reservoirs (eds Jones G, Fisher QJ, Knipe RJ), Geological Society of London Special Publication, 147, 117–34. Fisher QJ, Knipe RJ (2001) The permeability of faults within siliciclastic petroleum reservoirs of the North Sea and Norwegian Continental Shelf. Marine and Petroleum Geology, 18, 1063–81. Fisher QJ, Casey M, Harris SD, Knipe RJ (2003) Fluid-flow properties of faults in sandstone: the importance of temperature history. Geology, 31, 965–8. Floris FJT, Bush MD, Cuypers M, Roggero F, Syversveen AR (2001) Methods for quantifying the uncertainty of production forecasts: a comparative study. Petroleum Geoscience, 7, S87–96. Fossen H, Bale A (2007) Deformation bands and their influence on fluid flow. AAPG Bulletin, 91, 1685–700. Fowles J, Burley S (1994) Textural and permeability characteristics of faulted, high porosity sandstones. Marine and Petroleum Geology, 11, 608–23. Foxford KA, Walsh JJ, Watterson J, Garden IR, Guscott SC, Burley SD (1998) Structure and content of the Moab Fault Zone, Utah, U.S.A., and its implications for fault seal prediction. In:
Faulting, Fault Sealing and Fluid Flow in Hydrocarbon Reservoirs (eds Jones G, Fisher QJ, Knipe RJ), Geological Society of London Special Publication, 147, 87–103. Fredman N, Tveranger J, Semshaug SI, Braathen A, Svedrup E (2007) Sensitivity of fluid flow to fault core architecture and petrophysical properties of fault rocks in siliciclastic reservoirs: a synthetic fault model study. Petroleum Geoscience, 13, 305–20. Fredman N, Tveranger J, Cardozo N, Braathen A, Soleng H, Roe P, Skorstad A, Syversveen AR (2008) Fault facies modeling: technique and approach for 3-D conditioning and modeling of faulted grids. AAPG Bulletin, 92, 1457–78. Fristad T, Groth A, Yielding G, Freeman B (1997) Quantitative fault seal prediction: a case from Oseberg Syd. In: Hydrocarbon Seals: Importance for Exploration and Production (eds MøllerPedersen P, Koestler AG), Norwegian Petroleum Society Special Publication, 7, 107–24. Fulljames JR, Zijerveld LJJ, Franssen RCMW (1997) Fault seal processes: systematic analysis of fault seals over geological and production time. In: Hydrocarbon Seals: Importance for Exploration and Production (eds Møller-Pedersen P, Koestler AG), Norwegian Petroleum Society Special Publication, 7, 51–60. Garden R, Guscott SC, Burley SD, Foxford A, Walsh JJ, Marshall J (2001) An exhumed palaeo-hydrocarbon migration fairway in a faulted carrier system, Entrada Sandstone of SE Utah, USA. Geofluids, 1, 195–213. Gauthier BDM, Lake SD (1993) Probabilistic modeling of faults below the limit of seismic resolution in Pelican Field, North Sea, offshore United Kingdom. AAPG Bulletin, 77, 761–77. Gibson RG (1994) Fault-zone seals in siliciclastic strata of the Columbus Basin, offshore Trinidad. AAPG Bulletin, 78, 1372– 85. Gibson RG (1998) Physical character and fluid-flow properties of sandstone derived fault gouge. In: Structural Geology in Reservoir Characterisation (eds Coward MP, Johnson H, Daltaban T), Geological Society of London Special Publication, 127, 83–98. Gilham R, Hercus C, Evans A, de Haas W (2005) Shearwater (UK Block 22 ⁄ 30b): managing changing uncertainties through field life. In: Petroleum Geology: North-West Europe and Global Perspectives - Proceedings of the 6th Petroleum Geology Conference (eds Dore AG, Vining BA), pp. 663–73. Geological Society, London. Grauls D, Pascaud F, Rives T (2002) Quantitative fault seal assessment in hydrocarbon-compartmentalised structures using fluid pressure data. In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale R), Norwegian Petroleum Society Special Publication, 11, 127–39. Haney MH, Sneider R, Sheiman J, Losh S (2005) A moving fluid pulse in a fault zone. Nature, 437, 46. Harris SD, McAllister E, Knipe RJ, Odling NE (2003) Predicting the three-dimensional population characteristics of fault zones: a study using stochastic models. Journal of Structural Geology, 25, 1281–99. Harris SD, Vaszi AZ, Knipe RJ (2007) Three-dimensional upscaling of fault damage zones for reservoir simulation. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 353– 74. Heath AE, Walsh JJ, Watterson J (1994). Estimation of the effects of sub-seismic sealing faults on effective permeabilities in sandstone reservoirs. In: North Sea Oil and Gas Reservoirs III (eds Aasen JO, Berg E, Buller AT, Hjelmeland O, Holt RM, Kleppe J, Torsæter O), pp. 173–83. Kluwer Academic Publishers, London. Heum OR (1996) A fluid dynamic classification of hydrocarbon entrapment. Petroleum Geoscience, 2, 145–58.
112 T. MANZOCCHI et al. Hoffman KS, Neave JW (2007) The fused fault block approach to fault network modelling. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 75–87. Hull J (1988) Thickness-displacement relationships for deformation zones. Journal of Structural Geology, 10, 431–5. James WR, Fairchild LH, Nakayama GP, Hippler SJ, Vrolijk PJ (2004) Fault-seal analysis using a stochastic multifault approach. AAPG Bulletin, 88, 889–904. Jev BI, Kaars-Sijpesteijn CH, Peters MPAM, Watts NL, Wilkie JT (1993) Akaso field, Nigeria: use of integrated 3-D seismic, fault slicing, clay smearing, and RFT pressure data on fault trapping and dynamic leakage. AAPG Bulletin, 77, 1389–404. Jolley SJ, Dijk H, Lamens JH, Fisher QJ, Manzocchi T, Eikmans H, Huang Y (2007) Faulting and fault sealing in production simulation models: Brent Province, northern North Sea. Petroleum Geoscience, 13, 321–40. Jones RM, Hillis RR (2003) An integrated, quantitative approach to assessing fault-seal risk. AAPG Bulletin, 87, 507–24. Kim YS, Peacock DCP, Sanderson DJ (2004) Fault damage zones. Journal of Structural Geology, 26, 503–517. Knai TA, Knipe RJ (1998) The impact of faults on fluid flow in the Heidrun Field. In: Faulting and Fault Sealing in Hydrocarbon Reservoirs (eds Jones G, Fisher QJ, Knipe RJ), Geological Society of London Special Publication, 147, 269–82. Knipe RJ (1997) Juxtaposition and seal diagram to help analyze fault seals in hydrocarbon reservoirs. AAPG Bulletin, 81, 187– 95. Knipe RJ, Fisher QJ, Jones G, Clennell MR, Farmer AB, Harison A, Kidd B, McAllister E, Porter JR, White EA (1997) Fault seal analysis: successful methodologies, applications and future directions. In: Hydrocarbon Seals: Importance for Exploration and Production (eds Møller-Pedersen P, Koestler AG), Norwegian Petroleum Society Special Publication, 7, 15–38. Krokstad W, Sylta Ø (1996) Risk assessment using volumetrics from secondary migration modelling: assessing uncertainties in source rock yields and trapped hydrocarbons. In: Quantification and Prediction of Petroleum Resources (eds Dore AG, SindingLarsen R), Norwegian Petroleum Society Special Publication, 6, 219–35. Larter S, Aplin AC, Bennett B (2005) Capillary seal or permeability barrier; advances in experimental and theoretical considerations of caprock wetting state. Abstract: Annual Meeting of the American Association of Petroleum Geologists, 14, A78–9. Lindsay NG, Murphy FC, Walsh JJ, Watterson J (1993) Outcrop studies of shale smears on fault surfaces. In: The Geological Modelling of Hydrocarbon Reservoirs and Outcrop Analogues (eds Flint S, Bryant ID), Special Publication of the International Association of Sedimentologists, 15, 113–23. Maerten L, Gillespie P, Daniel J-M (2006) Three-dimensional geomechanical modeling for constraint of subseismic fault simulation. AAPG Bulletin, 90, 1337–58. Manzocchi T (1999) Towards an Improved Fault Representation in Two-phase Reservoir Simulation. 61st EAGE conference and technical exhibition, Helsinki, June 1999. Extended abstracts, 1, C-05. Manzocchi T, Ringrose PS, Underhill JR (1998) Flow through fault systems in high porosity sandstones. In: Structural Geology in Reservoir Characterisation (eds Coward MP, Johnson H, Daltaban T), Geological Society of London Special Publication, 127, 65–82. Manzocchi T, Walsh JJ, Nell P, Yielding G (1999) Fault transmissibility multipliers for flow simulation models. Petroleum Geoscience, 5, 53–63.
Manzocchi T, Heath AE, Walsh JJ, Childs C (2002) The representation of two phase fault-rock properties in flow simulation models. Petroleum Geoscience, 8, 119–32. Manzocchi T, Walsh JJ, Tomasso M, Strand J, Childs C, Haughton P (2007) Static and dynamic connectivity in bed-scale models of faulted and unfaulted turbidites. in: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 309–36. Manzocchi T, Heath AE, Palananthakumar B, Childs C, Walsh JJ (2008a) Faults in conventional flow simulation models: a consideration of representational assumptions and geological uncertainties. Petroleum Geoscience, 14, 91–110. Manzocchi T, Matthews JD, Strand JA, Carter JN, Skorstad A, Howell J, Stephen KD, Walsh JJ (2008b) A study on the structural controls on oil recovery from shallow marine reservoirs. Petroleum Geoscience, 14, 55–70. Marrett R, Allmendinger RW (1990) Kinematic analysis of faultslip data. Journal of Structural Geology, 12, 973–86. Miller SA, Collettini C, Chiaraluce L, Cocco M, Barchi M, Kaus BJP (2004) Aftershocks driven by a high-pressure CO2 source at depth. Nature, 427, 724–7. Moretti I (1998) The role of faults in hydrocarbon migration. Petroleum Geoscience, 4, 81–94. Morris A, Ferill DA, Henderson DB (1996) Slip tendency analysis and fault reactivation. Geology, 24, 275–8. Myers RD, Allgood A, Hjellbakk A, Vrolijk P, Briedis N (2007) Testing fault transmissibility prediction in a structurally dominated reservoir: Ringhorne field, Norway. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 271–94. Nybakken S (1991) Seal fault traps – an exploration concept in a mature petroleum province: Tampen Spur, northern North Sea. First Break, 9, 209–22. Odling NE, Harris SD, Knipe RJ (2004) Permeability scaling properties of fault damage zones in siliciclastic rocks. Journal of Structural Geology, 26, 1727–47. Ottesen Ellevset S, Knipe RJ, Svava Olsen T, Fisher QJ, Jones G (1998) Fault controlled communication in the Sleipner Vest Field, Norwegian Continental Shelf; detailed, quantitative input for reservoir simulation and well planning. In: Faulting and Fault Sealing in Hydrocarbon Reservoirs (eds Jones G, Fisher QJ, Knipe RJ), Geological Society of London Special Publication, 147, 283–97. Peacock DCP, Sanderson DJ (1994) Geometry and development of relay ramps in normal fault systems. AAPG Bulletin, 78, 147–65. Peters E, Arts RJ, Browwer GK, Geel CR (2009) Results of the Brugge Benchmark study for flooding optimisation and history matching. SPE 119094. Presented at the 2009 SPE Reservoir Simulation Symposium, The Woodlands, TX, February 2–4. Ponting DK (1989) Corner point geometry in reservoir simulation In: The Mathematics of Oil Recovery (ed King PR), pp. 46–65. Clarendon Press, Oxford. Revil A, Cathles LM III (2002) Fluid transport by solitary waves along growing faults a field example from the South Eugene Island Basin, Gulf of Mexico. Earth and Planetary Science Letters, 202, 321–35. Ringrose PS, Corbett PWM (1994) Controls on two-phase fluid flow in heterogeneous sandstones. In: Origin, Migration and Evolution of Fluids in Sedimentary Basins (ed. Parnell J), Geological Society of London Special Publication, 78, 141–50. Rivenæs JC, Dart C (2002) Reservoir compartmentalisation by water saturated faults: is evaluation possible with today’s tools? In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale
Faults in hydrocarbon flow models 113 R), Norwegian Petroleum Society Special Publication, 11, 173– 86. Robertson EC (1982) Continuous formation of gouge and breccia during fault displacement. In: Issues in Rock Mechanics, Proceedings of the 23rd Symposium on Rock Mechanics (eds Goodman RE, Heuze FE), pp. 379–404. American Institute of Mining, Metallurgical and Petroleum Engineers, New York. Rodgers S (1999). Discussion: ‘‘Physical constraints on hydrocarbon leakage and trapping revisited’’ by PA Bjørkum et al. – further aspects. Petroleum Geoscience, 5, 421–3. Rotevatn A, Fossen H, Hesthammer J, Aas TE, Howell JA (2007) Are relay ramps conduits for fluid flow? Structural analysis of a relay ramp in Arches National Park, Utah. In: Fractured Reservoirs (eds Lonergan L, Jolly RHJ, Rawnsley K, Sanderson DJ), Geological Society of London Special Publication, 270, 55–71. Schlumberger (2005) Eclipse Technical Description 2005a. Schlumberger. Schowalter TT (1979) Mechanics of secondary hydrocarbon migration and entrapment. American Association of Petroleum Geologists Bulletin, 63, 723–60. Shipton ZK, Cowie PA (2001) Damage zone and slip-surface evolution over lm to km scales in high porosity Navajo sandstone, Utah. Journal of Structural Geology, 23, 1825–44. Shipton ZK, Evans JP, Kirschner D, Kolesar PT, Williams AP, Heath J (2004). Analysis of CO2 leakage through ‘low-permeability’ faults from natural reservoirs in the Colorado Plateau, east-central Utah. In: Geological Storage of Carbon Dioxide (eds Baines SJ, Worden RH), Geological Society of London Special Publication, 233, 43–58. Smalley C, England WA, Muggeridge A, Abacioglu Y, Cawley S (2004) Rates of reservoir fluid mixing: implications for interpretation of fluid data. In: Understanding Petroleum Reservoirs: Towards an Integrated Reservoir Engineering and Geochemical Approach (eds Cubitt JM, England WA, Larter S), Geological Society of London Special Publication, 237, 99–113. Sperrevik S, Gillespie PA, Fisher QJ, Halvorsen T, Knipe RJ (2002) Empirical estimation of fault rock properties. In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale R), Norwegian Petroleum Society Special Publication, 11, 109–26. Stephen KD, MacBeth C (2008) Reducing reservoir prediction uncertainty by updating a stochastic model using seismic history matching. SPE 100295. SPE Reservoir Evaluation and Engineering, 11, 991–9. Stern D (2005) Practical aspects of scaleup of simulation models. Journal of Petroleum Technology, 57, 74–82. Sternlof KR, Chapin JR, Pollard DD, Durlofsky LJ (2004) Permeability effects of deformation band arrays in sandstone. AAPG Bulletin, 88, 1315–29. Sverdrup E, Helgesen J, Vold J (2003) Sealing properties of faults and their influence on water alternating-gas injection efficiency in the Snorre field, northern North Sea. AAPG Bulletin, 87, 1437–58. Sylta Ø (1991) Modelling of secondary migration and entrapment of a multicomponent hydrocarbon mixture using equation of state and ray-tracing modelling techniques. In: Petroleum Migration (eds England WA, Fleet AJ), Geological Society of London Special Publication, 59, 111–22. Teige GMG, Thomas WLH, Hermanrud C, Øren PE, Rennan L, Wilson OB, Nordga˚rd Bola˚s HM (2006) Relative permeability
to wetting-phase water in oil reservoirs. Journal of Geophysical Research-Solid Earth, 111, B12204. Underschultz J (2007) Hydrodynamics and membrane seal capacity. Geofluids, 7, 148–58. Vaszi AZ, Harris SD, Al-Busafi B, Fisher QJ (2005) An iterative method for the representation of two-phase fault rock properties in reservoir simulators. 13th European Symposium on Improved Oil Recovery, Budapest, Hungary, 25–27 April 2005. Walsh JJ, Watterson J, Heath AE, Childs C (1998a) Representation and scaling of faults in fluid flow models. Petroleum Geoscience, 4, 241–51. Walsh JJ, Watterson J, Heath A, Gillespie PA, Childs C (1998b) Assessment of the effects of sub-seismic faults on bulk permeabilities of reservoir sequences. In: Structural Geology in Reservoir Characterization (eds Coward MP, Johnson H, Daltaban TS), Geological Society of London Special Publication, 127, 99–114. Watts NL (1987) Theoretical aspects of cap-rock and fault seals for single- and two-phase hydrocarbon columns. Marine and Petroleum Geology, 4, 274–307. Wibberley CAJ, Yielding G, Di Toro G (2008) Recent advances in the understanding of fault zone internal structure: a review. In: The Internal Structure of Fault Zones: Implications for Mechanical and Fluid-Flow Properties (eds Wibberley CAJ, Kurz W, Imber J, Holdsworth RE, Collettini C), Geological Society of London Special Publication, 299, 5–33. Wilkins SJ, Naruk SJ (2007) Quantitative analysis of slip-induced dilation with application to fault seal. AAPG Bulletin, 91, 97– 113. Winefield P, Gilham R, Elsinger R. (2005). Plumbing the depths of the Central Graben: towards an integrated pressure, fluid and charge model for the Central North Seal HPHT play. In: Petroleum Geology: North-West Europe and Global Perspectives – Proceedings of the 6th Petroleum Geology Conference (eds Dore AG, Vining BA), pp. 1301–15. Geological Society, London. Wiprut D, Zoback MD (2000) Fault reactivation and fluid flow along a previously dormant normal fault in the northern North Sea. Geology, 28, 595–8. Yielding G (2002) Shale Gouge Ratio – calibration by geohistory. In: Hydrocarbon Seal Quantification (eds Koestler AG, Hunsdale R), Norwegian Petroleum Society Special Publication, 11, 1–15. Yielding G (2009) Using Probabilistic Shale Smear Factor to relate SGR predictions of column height to fault-zone heterogeneity (Abstract). Presented at 2nd International Conference on Fault and Top Seals – From Pore to Basin Scale, Montpellier, France, September 21–24. Yielding G, Freeman B, Needham DT (1997) Quantitative fault seal prediction. AAPG Bulletin, 83, 925–51. Yielding G, Bretan P, Freeman B (in press). Fault seal calibration – a brief review. In: Reservoir Compartmentalization (eds Jolley SJ, Fisher QJ, Ainsworth RB, Vrolijk PJ, Delise S), Geological Society of London Special Publication. Zijlstra EB, Reemst PHM, Fisher QJ (2007) Incorporation of fault properties into production simulation models of Permian reservoirs from the southern North Sea. In: Structurally Complex Reservoirs (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), Geological Society of London Special Publication, 292, 295–308.
Hydrostratigraphy as a control on subduction zone mechanics through its effects on drainage: an example from the Nankai Margin, SW Japan D. M. SAFFER Department of Geosciences, The Pennsylvania State University, University Park, PA, USA
ABSTRACT At many subduction zones, accretionary complexes form as sediments are offscraped from the subducting plate, and excess pore pressures commonly develop as low-permeability marine sediments undergo rapid tectonically driven loading. Mechanical models demonstrate that pore pressure controls the overall geometry of these systems by modifying shear strength both within the accretionary wedge and along its base. At the Nankai margin offshore SW Japan, the taper angle of the accretionary wedge varies markedly along-strike, from 4 along an eastern (Muroto) transect, to 8–10along a western (Ashizuri) transect. Sediment stratigraphy on the subducting plate also varies: along the Ashizuri transect, the lowermost part of the section includes abundant sandy turbidites, whereas along the Muroto transect it is composed of monotonous hemipelagic mudstone. Here, I use a numerical model of fluid flow, together with laboratory measurements that constrain the bulk mudstone permeability, to quantitatively test the hypothesis that the turbidite-rich section along the Ashizuri transect allows drainage at the base of the accretionary complex, resulting in differences in mechanical strength sufficient to cause the differences in taper angle. My results demonstrate that if the turbidite-rich units are 2–100 times more permeable than the mudstone units, the variation in stratigraphy can indeed explain the observed taper angles. In contrast, permeability anisotropy within the turbidite-rich units has only a minor effect; anisotropy ratios of 1000:1 would be required to cause the differences in taper angle. Along the Ashizuri transect, simulated pore pressures result in a basal shear strength ranging from a few MPa at the trench to 20 MPa by 30 km arcward; along the Muroto transect shear strength is substantially lower, reaching only 5 MPa by 30 km. This work shows that lithostratigraphy can strongly influence the mechanical behavior of subduction zone faults, through its control on the distribution and magnitude of excess pore pressure. Key words: critical taper angle, fault mechanics, pore pressure, subduction zones Received 14 July 2009; accepted 18 December 2009 Corresponding author: Demian M. Saffer, Department of Geosciences, The Pennsylvania State University, University Park, PA 16802, USA. Email:
[email protected]. Tel: 814-865-7965 Geofluids (2010) 10, 114–131
INTRODUCTION Pore fluid pressure is one key control on mechanical processes along major faults, including subduction zone plate boundary thrusts and continental and oceanic transforms (e.g., Hubbert & Rubey 1959; Rice 1992; Hickman et al. 1995). By modifying effective normal stress, pore pressure mediates overall fault and wall rock shear strength (e.g., Raleigh et al. 1976; Sibson 1981; Byrne & Fisher 1990), and fault initiation and localization (e.g., Moore & Byrne 1987; Le Pichon et al. 1993). Pore pressure has also been hypothesized to influence
the nature of slip on faults, including both the location of the shallow (up-dip) aseismic–seismic transition (Moore & Saffer 2001) and the occurrence of slow slip and tremor (Peacock 2009). The role of fluids in fault mechanics has been extensively studied at subduction zones, because the combination of rapid tectonically driven loading, abundant fluid supply, and low-permeability typical of marine sediments can result in pore pressures significantly above hydrostatic (e.g., Neuzil 1995), and because the shallowly dipping plate boundary faults are easily studied by drilling and seismic reflection surveys (e.g., Moore et al. 1990, 2001).
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Effects of hydrostratigraphy on drainage 115 One major advance in understanding the link between the mechanical behavior and hydrogeology of subductionaccretion complexes has come from critical taper theory (Davis et al. 1983; Dahlen 1984), in which accretionary wedges and thrust belts are treated as growing, self-similar triangular wedges of rock or sediment that are at Coulomb failure throughout, analogous to the sand in front of a bulldozer blade or snow in front of a snowplow (Davis et al. 1983; see their figure 5). The overall geometry of the deforming and actively accreting wedge is defined by a balance between basal traction and the gravitational forces associated with the wedge’s tapered geometry. The taper angle, given by the sum of surface slope (a) and dip of the basal detachment or de´collement (b) is quantitatively related to the internal strength of the wedge and that of the basal detachment, which depend on the friction coefficients within the wedge (lw) and along its base (lb), and the pore pressure ratios kw and kb (where k = Pf ⁄ Pl, and Pf and Pl are the pore fluid and lithostatic pressures, respectively; all variables and their units are defined in Table 1). Table 1 Notation and units for variables. Symbol
Description
Units
l¢ lb lb¢ lb¢¢ lw k K kd ko ksed kxx kzz L P Pf Ph Pl Q Vp x z C a b / c k k* kb kw qw t x
Effective coefficient of friction Basal coefficient of friction Effective basal coefficient of friction Modified effective basal coefficient of friction Coefficient of friction for accretionary wedge Permeability Hydraulic conductivity De´collement permeability Intercept of log(ksed)-porosity function (Eqn 1) Bulk sediment permeability Permeability in bedding-parallel direction Permeability in bedding-normal direction Drainage path length Pressure Fluid pressure Hydrostatic pressure Lithostatic pressure Fluid source term p-wave velocity Distance from deformation front Depth Fluid source term Surface slope of accretionary wedge De´collement dip angle Porosity Rate of log(ksed) increase with porosity Pore pressure ratio Modified pore pressure ratio Basal pore pressure ratio Wedge pore pressure ratio Fluid density Dynamic viscosity Multiplier for increased permeability of turbidite facies gravitational acceleration
Unitless Unitless Unitless Unitless Unitless m2 m sec)1 m2 m2 m2 m2 m2 m Pa Pa Pa Pa kg m)3 sec)1 m sec)1 m m sec)1 Degrees Degrees Unitless Unitless Unitless Unitless Unitless Unitless kg m)3 Pa s Unitless
g
m sec)2
In most cases, due to lack of detailed information about spatial variations in pore pressure, it is assumed that kw = kb. Despite its simplicity, critical taper theory has been applied widely to thrust belts in active tectonic settings and to compressional regimes at the toes of major delta systems, and has been highly successful in describing their mechanics and pore pressure regimes to first-order (e.g., Davis et al. 1983; Suppe 2007). More recently, observations from drilling and active seismic reflection studies have provided new constraints on the magnitude of excess pore pressures at subduction zones (e.g., Screaton et al. 2002, Tobin & Saffer 2009). Several recent studies have documented underconsolidation and estimated in situ pore pressures that are significantly in excess of hydrostatic using drilling data or p-wave velocity information derived from seismic reflection surveys (Bangs et al. 1990; Cochrane et al. 1996; Screaton et al. 2002). For a few locations at the Northern Barbados and Costa Rican subduction zones, pore pressure has been measured directly in long-term sealed borehole observatories (e.g., Becker et al. 1997; Foucher et al. 1997; Davis et al. 2006; Bekins & Screaton 2007). Although they are limited to specific depth intervals at individual boreholes, these observations document elevated pore pressure near the trench within and surrounding the de´collement zone at both margins. Numerical models have also emerged as powerful tools to study pore pressure development, and in particular to quantitatively explore the relative roles of sediment permeability, sediment thickness, fault zone permeability, loading rate, clay dehydration, and hydrologic communication between the compacting sediments and underlying basaltic ocean crust (e.g., Screaton et al. 1990; Matmon & Bekins 2006; Screaton 2006). For example, several recent studies have investigated the dewatering of sediments underthrust beneath the de´collement using coupled 1D models describing fluid flow and deformation (e.g., Screaton & Saffer 2005; Gamage & Screaton 2006; Screaton 2006; Skarbek & Saffer 2009). These studies have demonstrated that for measured sediment permeabilities and observed deformation rates, pore pressures should be elevated significantly above hydrostatic. Other recent studies have combined 2D hydrological models with critical taper theory, and demonstrate that bulk sediment permeability and the thickness of sediment on the incoming oceanic plate are the dominant factors governing pore pressure generation, and are thus the key underlying controls on taper angle and the overall mechanical strength of subduction–accretion complexes (Saffer & Bekins 2002, 2006; Matmon & Bekins 2006). Here, I use a 2D numerical model describing fluid flow in accretionary complexes (e.g., Screaton et al. 1990; Bekins et al. 1995) to further investigate the relationships
116 D. M. SAFFER between sediment hydrostratigraphy, pore pressure, taper angle, and mechanical strength, using an example from the well-studied Nankai margin offshore SW Japan where drilling, seismic reflection, and laboratory-based studies have provided key constraints on the hydrogeology. This study builds upon previous work by (i) incorporating new detailed permeability measurements for the sediment entering the subduction zone along the Nankai Trough, and (ii) explicitly evaluating the role of sediment hydrostratigraphy in detail and considering the case where kw „ kb, specifically to test the hypothesis that along-strike changes in the lithostratigraphy of incoming sediment drive variations in pore pressure that are sufficient to explain observed along-strike differences in wedge taper angle. I then discuss the role of lithostratigraphy in affecting the mechanical strength of subduction plate boundary decollements, and place the results into a broader framework that considers the competing roles of fluid sources and permeability architecture in governing pore pressure and rock strength.
NANKAI MARGIN Geologic setting The Nankai accretionary complex is forming by the northwestward subduction of the Philippine Sea plate beneath the Eurasian plate at 40 km Myr)1 (Fig. 1) (Seno et al. 1993). The area offshore Shikoku Island has been investigated extensively by drilling, seafloor geophysical and geological studies, and marine seismic reflection surveys (e.g., Moore et al. 1990, 2001; Stoffa et al. 1992). Deep Sea Drilling Project (DSDP) and Ocean Drilling Program (ODP) Legs 31, 87, 131, 190, and 196 drilled and cored several boreholes penetrating the incoming sediment section and accretionary prism along two transects: the Ashizuri transect offshore Cape Ashizuri and the Muroto transect 100 km along-strike to the northeast offshore Cape Muroto (Fig. 1). The prism taper angle varies markedly along-strike, from 4–5 along the Muroto transect to 8–10 along the Ashizuri transect (Fig. 1B,C).
134°
132° land
KP
ku Is
Shiko
136° (A) 34°
CM 1000
1000 00
20
CA 1178 1175
00 20
808 1174 1173 00 00 40
583 582 1177
PP
Depth (km below s.l.)
PSP
Depth (km below s.l.)
0
00 30
20 00
Basin fill
4000
1176
10
EP
33°
3000
Muroto transect
32°
Se Ki am na 4000 ou n nt 4000 s 300 0
Ashizuri transect
31° OOST
(B)
6
808 1174 Turbidites
1173
7 8
Décollement
USB and LSB mudstones
BSR
(C)
1 583
2
Turbidites
USB and LSB 582 (hemipelagic mudstones)
3 4 Décollement
5
LSB (turbidite-rich) 10 km
1177
Fig. 1. (A) Map of the study area offshore SW Japan in region shown by black rectangle in inset, showing locations of the Ashizuri and Muroto transects and DSDP and ODP drill sites. Gray box along Muroto transect shows the extent of a 1999 3D seismic survey. PP, Pacific Plate; PSP, Philippine Sea Plate; EP, Eurasian Plate; CA, Cape Ashizuri; CM, Cape Muroto; KP, Kii Peninsula. (B, C) Interpreted cross-sections based on seismic reflection lines for the Muroto transect (B) and Ashizuri transect (C), after Moore et al. (1990, 2001). Dark gray shading denotes the Upper Shikoku Basin (USB) and Lower Shikoku Basin (LSB) facies; white area denotes the Trench Wedge Turbidite facies.
Effects of hydrostratigraphy on drainage 117
Site 582 Ashizuri transect
Muroto transect
Nankai trench wedge facies
200 Quaternary
300
100
Quaternary
Site 1173
100
Nankai trench wedge facies
Site 1177
?
Upper Shikoku Basin facies
200
400 400
Lower Shikoku Basin mudstone facies
500
Projected decollement
400 500
Lower Shikoku Basin turbidite facies
600
700
500
600
Volcaniclastic facies
800 mbsf
Miocene
Upper Shikoku Basin facies
Miocene
Pliocene
600
300
Projected decollement
Pliocene
Pliocene
300
700 mbsf
Lower Shikoku Basin facies
Basalt
Basalt
Hemipelagic mud
Volcanic ash/tuff
Sandy turbidites
Silty turbidites
Pillow basalt
Fig. 2. Stratigraphic columns for the two transects, after Underwood (2007). The composite stratigraphic column for the Ashizuri transect is defined by drilling at Sites 582 and 1177; dashed line shows an estimated age correlation after Underwood (2007). Locations of samples used in permeability measurements for this study (Site 1177 only) and from Skarbek & Saffer (2009) (Site 1173) are shown by black arrows. Note the abundant sandy turbidites in the LSB facies along the Ashizuri transect.
The incoming sediment on the Philippine Plate consists of four main lithostratigraphic units: (i) an early-middle Miocene volcaniclastic facies overlying basaltic basement, (ii) the Miocene-Pliocene Lower Shikoku Basin (LSB) facies, (iii) the Pliocene-Quaternary Upper Shikoku Basin (USB) facies, and (iv) the Quaternary trench-wedge (TW) facies composed of silty and sandy turbidites interbedded with hemipelagic muds (Fig. 2) (Moore et al. 2001; Underwood 2007). The thickness and lithostratigraphy of the sequence, and of the LSB in particular, vary alongstrike over distances of 100–200 km, with some local variations over distances of 20 km (Moore et al. 2001; Ike et al. 2008). Along both transects, the USB facies is composed of clay-rich hemipelagic mudstone interbedded with thin ashes, and ranges from 150 to 200 m thick. Along the Muroto Transect, the LSB is composed entirely
of hemipelagic mudstone, whereas along the Ashizuri transect it includes abundant sandy turbidites occurring in four distinct sand-rich packages that total 240 m in thickness (Fig. 2) (Moore et al. 2001 Underwood 2007). This difference in stratigraphy is attributed to deflection of turbidite deposition around basement highs throughout the Shikoku Basin from mid-Oligocene to mid-Miocene (Ike et al. 2008). The de´collement localizes within the LSB facies along both transects; the underthrust section is composed entirely of the LSB and volcaniclastic facies, and the accreted wedge incorporates sediment from the uppermost LSB, along with the USB and TW facies (Fig. 1B,C) (Moore et al. 2001). Along both transects, sediment porosity decreases exponentially with depth within the accreting wedge and seaward of the trench (Athy 1930; Hyndman et al. 1993;
118 D. M. SAFFER
Evidence for along-strike differences in hydrogeology Several observations provide indirect and primarily qualitative evidence for along-strike variations in pore pressure. The higher taper angle along the Ashizuri transect suggests that either the wedge itself is weaker than along the Muroto transect, or the basal de´collement is stronger (Davis et al. 1983; Moore et al. 1990). Moore et al. (1990) specifically raised the possibility that the basal de´collement along the Muroto transect is weaker than along the Ashizuri transect, due to either a lower friction coefficient or elevated pore pressure. In seismic reflection data, the de´collement along the Muroto transect is identified as a negative polarity reflection to 30 km arcward of the trench, as would be expected for underconsolidation and elevated pore pressure within the subjacent underthrusting section (Moore et al. 1990; Bangs et al. 2004). In contrast, the de´collement reflector along the Ashizuri transect exhibits normal (positive) polarity, suggesting that the underthrust sediment there is better drained and thus more consolidated than along the Muroto transect. This interpretation is consistent with p-wave velocities in the underthrust section obtained from geophysical surveys and shipboard measurements (Stoffa et al. 1992; Costa-Pisani et al. 2005; Tobin & Saffer 2009). A compilation of these data documents that p-wave velocity increases more rapidly with burial beneath the wedge along the Ashizuri transect than along the Muroto transect (Fig. 3), implying increased porosity loss (consolidation) and thus more efficient drainage. Laboratory shearing experiments have shown that the friction coefficient of sediments in the USB and LSB facies does not vary appreciably between the two transects
3000
Vp (m Sec–1)
Moore et al. 2001). Data from drilling, along with porosities estimated from seismic reflection interval velocities, define an abrupt increase in porosity across the de´collement into the underthrusting sediment along the Muroto transect (Moore et al. 2001; Morgan & Ask 2004; CostaPisani et al. 2005). This porosity inversion is also documented by a negative polarity seismic reflection at the de´collement (Moore et al. 1990; Bangs et al. 2004), and has been interpreted to reflect a combination of increased mean stress due to horizontal compression in the overriding accretionary wedge, and severe underconsolidation in the rapidly buried subducting sediments (e.g., Karig 1990; Saffer 2003; Morgan & Ask 2004). Screaton et al. (2002) and Saffer (2003) estimated excess pore pressures of kb = 0.67–0.77 at Site 1174 (1.5 km landward of the trench) and Site 808 (3.6 km landward) from the anomalously high porosity documented by drilling. Tobin & Saffer (2009) used interval velocities from seismic reflection data to estimate pore pressures of kb = 0.75 along a transect extending to 20 km downdip of the trench.
Ashizuri
2500
Muroto
2000
1500 800
600
400
200
0
–200
–400 –600 –800
Burial depth registered to deformation front (m) Fig. 3. Average p-wave velocity within the underthrust section as a function of burial depth, registered to the deformation front. Data are compiled from seismic reflection interval velocities reported by Stoffa et al. (1992) along the Ashizuri (squares) and Muroto (filled circles) transects, from interval velocities from a pre-stack depth migration of 3D seismic data along the Muroto transect in the region shown in Fig. 1 (Costa-Pisani et al. 2005; Tobin & Saffer 2009) (open circles) [s], and from ODP Sites along the Muroto (gray circles) and Ashizuri (gray square) transects. Error bars are defined as reported by the authors of each study. Note that the rate of velocity increase with burial depth is greater along the Ashizuri transect than the Muroto, suggesting the possibility of more rapid dewatering and porosity loss.
(Brown et al. 2003; Kopf & Brown 2003). These authors suggested that the higher taper angle along the Ashizuri transect could be explained with a pore pressure ratio ranging from k = 0.60 to 0.75 (compared with a pore pressure ratio of k = 0.80–0.90 along the Muroto transect), and although they did not consider differences between kb and kw, they attributed the difference in pore pressure to enhanced overall drainage along the Ashizuri transect allowed by the LSB turbidites, under the assumption that the sediment section underthrust beneath the outer accretionary wedge is similar to the LSB facies in the presentday incoming section. Taken together, these observations have raised the hypothesis that the pore pressure regime – either along the wedge base or within the wedge itself – varies along-strike, and may be linked to along-strike variations in the lithostratigraphy of sediment on the subducting Philippine Sea Plate. Here, I provide a quantitative test of this hypothesis, aimed at investigating whether the observed differences in lithostratigraphy could in fact drive differences in kb and kw sufficient to explain the differences in taper angle.
PERMEABILITY MEASUREMENTS Methods Permeability (k) is a key input to hydrologic models. In particular, bulk permeability of the sediment matrix is the limiting factor in drainage of sediments as they are
Effects of hydrostratigraphy on drainage 119
Permeability measurement results The permeability data define a log-linear relationship between permeability and porosity of the form: log(ksed) = ko + c/, where ksed is sediment permeability, ko is a projected intercept at zero porosity, / is fractional porosity, and c describes the rate of permeability change with porosity. Permeabilities obtained from the CRS method and those from flow-through measurements are highly similar over a porosity range of 12–61% (Fig. 4). Notably, permeabilities of the mudstone lithologies are not significantly different between Sites 1177 and 1173, or between samples from the USB and LSB facies (Fig. 4). Data from this study are consistent with previous flow-through tests conducted on core samples from these boreholes under lower stresses (Gamage 2005; Gamage & Screaton 2006), and with data compilations for argillaceous sediments (Neuzil 1994). Overall, the data are well fit and bounded by the following permeability-porosity function:
1.0
0.8 Compilation for argillaceous shales [Neuzil, 1994 ]
Porosity
mechanically loaded, because even if permeable strata or fractures are present, fluids must first reach the permeable features in order to escape from the system. Meaningful projections of bulk permeability with increased burial depth require obtaining direct measurements over a wide range of porosities and effective stresses. Here, I report on permeability measurements conducted on samples from ODP Site 1177 along the Ashizuri transect, and compile these data with previously published permeability measurements from Sites 1173 and 1177 (locations shown in Fig. 2) (Gamage 2005; Gamage & Screaton 2006; Skarbek & Saffer 2009). All of the permeability data were measured in the vertical or near-vertical. I conducted permeability measurements by two methods: constant rate of strain uniaxial consolidation testing (CRS), and constant head flow-through. These approaches are described only briefly here; a more detailed explanation of the methods and sample geometry is given in Saffer & McKiernan (2005) and Skarbek & Saffer (2009). In CRS tests, the specimen is loaded at a constant strain rate in a fixed-ring consolidation cell via a computer-controlled load frame. During the test, the specimen is drained at one boundary and undrained at the other (e.g., Olson 1986). As the sample is progressively loaded, pore pressure builds and is monitored at the undrained boundary, allowing the permeability to be calculated continuously as a function of consolidation (e.g., Long et al. 2008). The CRS experiments for samples from Site 1177 were conducted over an effective stress range from 1 to 35 MPa. For flowthrough tests, the sample is also loaded in a fixed ring consolidation cell, but is held under a constant effective axial stress, and a pressure gradient is imposed across the sample. Permeability is determined by measuring the resulting steady state flow rate and applying Darcy’s law.
0.6
0.4
0.2 Compilation for sands
0 10–22
10–20
10–18 10–16 Permeability (m2)
10–14
Fig. 4. Permeability data for CRS tests on samples from Site 1173 (black filled circles) (Skarbek & Saffer 2009), Site 1177 (green filled triangles), and for flow-through tests on samples from Sites 1173 (black squares with crosses) and 1177 (filled red squares). Data from Gamage & Screaton (2006) (open squares) for Site 1173 and Gamage (2005) for Site 1177 (partially filled red squares) are shown for comparison. Outlined areas indicate the extent of data from compilations for argillaceous sediment (Neuzil 1994) and sands (Nelson 1994).
logðkÞ ¼ 20:45 0:5 þ 6:93/
ð1Þ
MODELING METHODS I use the finite element code SUTRA to simulate coupled 2D steady-state fluid flow and heat transport within a cross-section perpendicular to the trench along each transect (Voss 1984; see also Screaton et al. 1990; Bekins et al. 1995), in order to investigate the along-strike differences in pore pressure resulting from variations in lithostratigraphy. I then evaluate whether the resulting pore pressures are consistent with the taper angle along each transect. The choice of a 2D model is justified by the fact that the dominant bathymetric slope, and gradients in porosity and compaction-driven fluid expulsion all occur in the dip direction; as a result, hydraulic gradients that drive fluid flow are primarily up-dip rather than trench-parallel (e.g., Screaton & Ge 1997; Saffer & Bekins 2006). This line of reasoning is supported by results from seismic reflection studies, which document that the incoming sediment section and relief on the underlying basaltic basement both vary along-strike, but remain relatively smooth along sections perpendicular to the trench (Ike et al. 2008). It is also consistent with existing 3D modeling studies, which show that large trench-parallel driving forces can drive a component of along-strike flow in high-permeability horizons (k = 10)14–10)12 m2), but the primary fluid flow
120 D. M. SAFFER direction is up-dip toward the trench (e.g., Screaton & Ge 1997; Cutillo et al. 2003; Spinelli & Saffer 2007). Because accretionary wedges both grow and deform internally over geologic timescales (a few to tens of Myr) (e.g., Westbrook 1994; Moore et al. 2001), their overall geometry reflects long-term and spatially averaged internal and basal pore pressures (e.g., Davis et al. 1983). Here, the results of steady-state simulations are taken to represent the time-averaged pore pressure relevant to the maintenance of a critical taper (for further detailed discussion of the steady-state assumption, see Matmon & Bekins 2006; Saffer & Bekins 2006). However, it is important to note that for other processes of interest, such as transient slip, short-timescale pore pressure fluctuations, or chemical transport along faults (Carson & Screaton 1998), transient fluid flow may be significant. Steady-state fluid flow in 2D is described by: kw r ð2Þ ðrPf w g Þ Q ¼ 0; where qw is fluid density (kg m)3), Pf is fluid pressure (Pa), g is gravitational acceleration (m sec)2), k is permeability (m2), t is dynamic viscosity (Pa s), and Q is a fluid source term (kg m)3 sec)1) that includes fluids expelled by both sediment compaction and clay dehydration. Temperatures within the model domain are computed using the heat transport model of Ferguson et al. (1993), which accounts for sediment deformation and heat advection in the sediments as they are subducted or accreted to define a 2D steady-state thermal field (e.g., Bekins et al. 1994, 1995; Saffer et al. 2008). This temperature distribution is then used to define fluid viscosity and density in Eqn (2). Other inputs for the model are (i) hydrologic boundary conditions, (ii) fluid source terms, and (iii) permeability, which is defined separately for the bulk sediments and for the de´collement zone. Each of the model inputs is described briefly in the following sections; for a more detailed discussion of the modeling approach the reader is referred to Bekins et al. (1995) and Saffer & Bekins (1998).
Several recent detailed parametric modeling studies have shown convincingly that bulk sediment permeability (ksed) is the dominant control on pore pressure in subduction– accretion complexes (e.g., Saffer & Bekins 2002), and that de´collement zone permeability (kd) and variations in the porosity distribution and fluid source terms have a smaller effect (Screaton et al. 1990; Bekins & Dreiss 1992; Saffer & Bekins 2006). Thus, the primary focus of this study is to explore the role of bulk sediment permeability in affecting pore pressure; I also explore a range of kd that is consistent with previous studies of the Nankai margin in order to evaluate its effects. Model domain and boundary conditions The model domain extends from 15 km seaward of the trench to 60 km landward, and consists of 91 500 nodes that define 88 940 quadrilateral elements (Fig. 5). The model grid is divided into three hydrologic regimes: the underthrust sediments, the accreted sediments, and the de´collement zone (e.g., Saffer & Bekins 1998; Matmon & Bekins 2006). The surface slope and de´collement dip angle for each transect are approximated by a linear top and base to the model domain. Along the Muroto transect, the surface slope (a) is 1.55 and decollement dip angle (b) is 2.6; along the Ashizuri transect, a is set to 3.8 and b to 4.25 (Moore et al. 1990). The top and seaward edges of the model are specified as hydrostatic constant-head boundaries. The arcward edge of the model is treated as no-flow boundary, under the assumption that at distances >50 km from the trench, permeabilities and compactive dewatering are small and have an insignificant effect on pore pressures (e.g., Bekins et al. 1995). At the base of the sediment section, the 40–55 mthick volcaniclastic facies underlying the LSB contains abundant altered ash (Shipboard Scientific Party 1991, 2001). This stratigraphic interval appears to act as a hydrological seal, based on porosity data that indicate little or no drainage at the base of the sediment column at Sites 1174
Arcward boundary: no flow
10 km Top boundary:
hydrostatic pres
Accreted sediments w
dary: no flo
Basal boun
Underthrust sediments
sure
Fig. 5. Example model grid (top) and cartoon of model domain showing boundary conditions and schematic sediment velocity field (bottom). The de´collement is shown by the thick gray line in both panels. For clarity, in the upper diagram only every 30th column of elements is shown, every third row of elements is shown for the prism, and every fifth row for the underthrust section. The de´collement includes 12 elements at 5 m spacing, which are also not shown.
Effects of hydrostratigraphy on drainage 121 and 808 (Screaton et al. 2002). This is also consistent with down-hole profiles of pore water geochemistry, which document rapid circulation of seawater in the permeable basaltic ocean crust, but a diffusive gradient between the basalt and overlying sediments that suggests hydrological isolation of the two (Spivack et al. 2002; McKiernan 2005; Saffer & McKiernan 2009). Accordingly, the base of the sediment section is prescribed as a no-flow boundary. Although recent studies have illustrated that a permeable ocean crust, if in hydrologic communication with the overlying sediments, may act as an efficient drainage pathway for steeply tapered wedge geometries (Matmon & Bekins 2006), other studies illustrate that for shallow wedge geometries or low-permeability underthrust sediments, relaxing this boundary condition has little effect on simulated pore pressures (Saffer & Bekins 1998; Spinelli et al. 2006).
As sediment packets are transported arcward relative to the trench, their porosity is reduced and pore water is expelled. Defining the porosity distribution empirically for computation of fluid source terms offers several advantages over attempting to simulate fully coupled deformation and dewatering in the accretionary wedge, most notably that (i) although some laboratory studies have been conducted, constitutive relations between mean stress, differential stress, and porosity are generally not adequately defined to allow robust predictions of deformation, porosity loss, or pore pressure generation (e.g., Shi & Wang 1988; Karig 1990; Davis 1996); (ii) the porosity field is reasonably well known for many accretionary complexes from seismic reflection data and drilling (e.g., Fig. 6); and (iii) by using field measurements of porosity as a function of depth compiled for numerous active and exhumed accretionary complexes (Bray & Karig 1985), the additional effects of pressure solution, precipitation and cementation, and timedependent secondary consolidation are implicitly included in the fluid source terms. The fluid sources from sediment compaction are computed from the divergence of sediment velocity, assuming conservation of solid mass and that time-averaged sediment velocities in the wedge can be approximated as uniformly diverging (e.g., Bray & Karig 1985; Bekins & Dreiss 1992; Morgan & Karig 1993). Sediment velocities are computed following Bekins & Dreiss (1992), and are a function of porosity distribution, incoming sediment thickness, plate convergence rate, and wedge geometry. The velocity of accreted sediment decreases with distance into the subduction zone (Fig. 5) due to the combination of consolidation and vertical thickening, as documented by patterns of observed porosity loss and lateral strains
Model porosity and fluid source terms In defining the fluid source terms for regional-scale models of subduction–accretion complexes, one widely adopted approach takes advantage of the fact that accretionary wedges grow self-similarly (Le Pichon et al. 1990; Bekins & Dreiss 1992; Westbrook 1994). This allows fluid sources from compaction and mineral dehydration to be computed in a reference frame fixed at the deformation front, as sediment packets are transported through a prescribed porosity field in the growing accretionary complex (e.g., Bethke 1986; Bekins & Dreiss 1992; Wang 1994). In this reference frame, sediments enter the subduction zone with an initial porosity and moving at the plate convergence rate, and are either offscraped onto the leading edge of the accretionary prism, or underthrust beneath it.
0.5
Muroto
(A)
Ashizuri
(B)
Porosity
0.4 0.3 0.2 t
rthrus
0.1 0 30
Unde
Prism
25
st
rthru
e Und
Prism 20
15
10
5
Distance from deformation front (km)
0
30
25
20
15
10
5
0
Distance from deformation front (km)
Fig. 6. Model porosity in the middle of the underthrust section (solid lines) and in the lowermost accretionary wedge sediments (dashed lines), as a function of distance from the deformation front for the Muroto (A) and Ashizuri (B) transects. The model inputs match inferred porosity within the underthrust section from seismic interval velocities along the Muroto transect to 20 km from the trench (panel A, gray dots) (Tobin & Saffer 2009) and estimated porosities along the Ashizuri transect at locations where Stoffa et al. (1992) report interval velocity, computed following the approach described in Tobin & Saffer (2009) (panel B, gray squares).
122 D. M. SAFFER
/ ¼ 0:75e0:0011z
Muroto transect
ð3aÞ
4000
Porosity
2000
Depth registered to depth at deformation front (m)
measured in mudstone fabrics (Bray & Karig 1988; Morgan & Karig 1993; Henry et al. 2003). The velocity of underthrust sediment is fixed at the plate convergence rate, and the fluid sources are correspondingly larger (Bekins et al. 1995; Saffer & Bekins 1998). The fluid sources from clay dehydration are computed using a kinetic expression for smectite transformation based on temperatures defined by the thermal model, and exposure times computed following Bekins et al. (1994) and Saffer et al. (2008). The dehydration-driven fluid sources are generally small in comparison with the compaction terms (Bekins et al. 1995), but may be significant in some portions of the system (e.g., Spinelli et al. 2006). I define the porosity distribution within the model domain as an exponential function of depth (Athy 1930; Bray & Karig 1985), constrained by data from drilling at Sites 1173 and 1177:
0.60 0.40 0.30 0.20
0
0.10
0.20
–2000
Ashizuritransect
ð3bÞ
where z is depth (m). As noted above, observed porosity increases sharply to higher values below the de´collement, so is specified separately. The rate of porosity loss in the underthrust section is especially important, because the sediment velocities are higher and the source terms are concomitantly larger than in the wedge. Data from drillsites and seismic reflection interval velocities provide information about porosity to 20 km landward of the trench (Stoffa et al. 1992; Moore et al. 2001; Tobin & Saffer 2009) (Figs 6 and 7). In addition, the polarity and amplitude of the de´collement seismic reflection suggests that the increase in porosity across it decreases to near zero by 30–40 km arcward of the trench (Bangs et al. 2004). On the basis of these observations, I specify an offset in the depth used to define porosity in the underthrust section in order to reflect its relative underconsolidation (e.g., Matmon & Bekins 2006). The model porosity honors existing porosity estimates, and converges with that of the overriding accretionary wedge by 40 km from the trench (Fig. 6). Example distributions of porosity and fluid sources are shown in Fig. 7. Permeability architecture The permeability of the hemipelagic mudstones is well defined by the laboratory measurements, and varies remarkably little between drillsites or between the USB and LSB facies (cf. Fig. 4). The mudstone transport properties have also been investigated at larger scales using 1D coupled models of loading and compaction, and the laboratory-derived values appear to adequately describe bulk permeability at a scale of hundreds of meters (Neuzil 1994; Gamage & Screaton 2006; Skarbek & Saffer 2009).
0.30
0.10
–4000 4000
Log source term (VH2O / Vseds–1) (B)
2000
–14 to –14.5 –14.5 to –15
0 –15 to –15.5 –15.5 to –16
/ ¼ 0:79e0:00087z
(A)
–13.5
–2000 –14 to
4
to –1
–14.5
Ashizuri transect V.E. = 4x
–4000 40
30
20
10
0
–10
Distance from deformation front (km) Fig. 7. Example inputs to the flow model along the Ashizuri transect for (A) fractional porosity and (B) log of fluid sources, reported in units of VH2O ⁄ Vsed sec)1 (equivalent to the product qwQ). Note that both porosity and fluid sources within the underthrusting section are considerably larger than in the overlying wedge. V.E., vertical exaggeration.
The fact that the mudstone permeability is well constrained allows other aspects of the overall hydrogeologic system to be explored in detail (e.g., Matmon & Bekins 2006). Here, I consider the three log-linear functions fit to the data as described by Eqn (1) in defining the bulk permeability of the hemipelagic mudstones (Fig. 4). I consider two simple permeability architectures to incorporate the first-order variations in observed lithostratigraphy between the two transects. In a first set of model simulations, I assume that the turbidite-rich facies (both the TW turbidites and the LSB facies turbidites where present along the Ashizuri transect) are more permeable than the hemipelagic mudstones of the USB and LSB. I consider a range of turbidite permeabilities from 2 to 100 times higher than those of the hemipelagic mudstones (delineated by a multiplier x, such that x = 2–100; Table 2). For comparison, drilling data and sedimentological descriptions from Site 1177 document 240 m total thickness of sand packages within the LSB turbidite facies; assuming that the sands are several orders of magnitude more permeable than the mudstones (Nelson 1994), the layer-normal effective permeability of the turbidite-rich
Effects of hydrostratigraphy on drainage 123 Table 2 Range of permeability values used in model simulations. Geological unit
kzz (m2)
kxx:kzz
TW turbidites USB and LSB mudstones De´collement zone
1–100 · mudstones log(k) = )20.45 ± 0.5 + 6.93/ 10)14–1.74 · 10)19x 3 · 10)14–5.22 · 10)19x 3 · 10)15–5.22 · 10)20x 1–100 · mudstones
1 1 1 1 1 1–10 000*
LSB turbidites
*Anisotropy was considered only in simulations for the Ashizuri transect. For these models, permeability of the TW turbidites was specified to be 3, 10 or 30 times higher than that of the hemipelagic mudstones of the USB and LSB, kxx:kzz within the TW turbidites was set equal to 1, and kzz for the LSB facies was set to equal to that of the mudstones.
LSB facies should be 3–10 times higher than for the homogeneous mudstone-dominated LSB along the Muroto transect. If disruption of strata occurs due to tilting or small-scale faulting, the effective bulk permeability could be further increased by interconnection of permeable zones. This simple set of assumptions results in a single zone of enhanced permeability along the Muroto transect corresponding to the TW turbidites (upper 16 elements; Fig. 5); the LSB facies is composed of homogeneous mudstones and its permeability is not increased. Along the Ashizuri transect there are two zones of enhanced permeability corresponding to the TW turbidites (upper 20 elements) and the LSB turbidite facies (lowermost 24 elements) (Fig. 5). In a second set of models, I evaluate the role of permeability anisotropy in the LSB turbidites. For these models, I assume that in the LSB turbidite facies, the beddingnormal permeability (kzz) is the same as for the hemipelagic mudstones, and allow the anisotropy ratio (kxx:kzz) to vary from 1 to 10 000. However, I assign higher permeability in both the x- and z-directions in the TW turbidites, under the assumption that faulting and fracturing within the accreting wedge disrupt bedding at a scale of a few to hundreds of meters and increase overall permeability by mixing of the sandy and mud-rich lithologies. I consider three cases for the TW turbidites, in which they are 3, 10, or 30 times more permeable than the mudstones (Table 2). It is important to note that for both sets of models, by using the present-day incoming section to define the hydrostratigraphy, I implicitly assume that the modern accretionary complex was formed by accretion and underthrusting of a similar sedimentary section in the past. This assumption is consistent with seismic reflection studies which show that (i) the incoming sediment section of the Shikoku Basin varies over distances of 100–200 km along strike but is relatively uniform along lines perpendicular to the trench; and (ii) the province of the Shikoku Basin from Cape Muroto to the southwest is characterized by relatively smooth basement topography and spatially uniform sediment cover (Ike et al. 2008).
Several lines of evidence suggest that the permeability of the de´collement is higher than that of the surrounding sediments, including structural observations that document a 20–30 m-thick brecciated and fractured fault zone at Sites 1174 and 808 (Moore et al. 2001), geochemical anomalies and observed seeps at the seafloor associated with faults (Kastner et al. 1991), and patterns of porosity loss within the underthrusting sediment that suggest upward drainage to a permeable de´collement zone (Saffer 2003). Through detailed sensitivity analyses, previous numerical modeling studies have constrained the timeaveraged de´collement zone permeability to be 10)16– 10)13 m2 along the Muroto transect; studies at other margins have yielded similar values (Bekins et al. 1995; Saffer & Bekins 1998; Spinelli et al. 2006). On this basis, I specify a value of kd = 10)14 m2 at the trench, decreasing by 2 orders of magnitude by 55 km arcward (Table 2). This is consistent with the inferred permeability required to accommodate the volume of fluids released from subjacent sediments by compaction (Skarbek & Saffer 2009), and with increased effective stress accompanying burial that should result in closure of fractures (Rice 1992; David et al. 1994). To assess its effect on modeled pore pressures, I also consider cases in which kd is increased or decreased by one half an order of magnitude. Critical wedge taper and assessment of models To assess whether modeled pore pressures result in stable critically tapered accretionary wedges, I use two criteria. First, the maximum simulated steady-state pore pressure within the model domain cannot exceed the lithostatic pressure (i.e., k < 1.0), because such pore pressures would not be sustained and thus do not represent realistic combinations of permeability and fluid sources (Bekins et al. 1995; Matmon & Bekins 2006). Second, simulated pore pressures must satisfy the equations describing critically tapered accretionary wedges (Davis et al. 1983, eqns 18 and 28; see also Dahlen 1984). In applying the latter criteria, I use values of kw and kb averaged over the front 30 km of the system to define the overall wedge and basal pore pressures, which are the primary control on taper angle (e.g., Breen & Orange 1992). Most previous studies of wedge criticality have assumed that the pore pressure ratio in the wedge and along its base are the same (kb = kw), and either argued that to first order, differences between the two quantities are not well enough known to warrant detailed investigation (Davis et al. 1983; Brown et al. 2003), or they are not significant in controlling wedge geometry (Saffer & Bekins 2002). However, recent work by Matmon & Bekins (2006) has shown that considering the two pore pressure ratios separately may be important, especially for steeply tapered geometries. For the Nankai margin study area, large
124 D. M. SAFFER
MODELING RESULTS AND DISCUSSION Simulated pore pressures Using permeabilities defined by the upper bound and best fit permeability–porosity relationships in Eqn (1), simulated pore pressures are near hydrostatic along both transects, and are too low to be consistent with the observed wedge taper angle along the Muroto transect, regardless of the assumed permeability of the TW turbidite facies or the de´collement zone (Fig. 9A). In contrast, using the lower bound fit to permeability–porosity data, and with turbidite permeabilities in the TW and LSB facies set equal to the mudstone permeabilities (x = 1), maximum simulated pore pressures exceed lithostatic pressure (k > 1.0) within the accretionary wedge along both transects, because the interior of the wedge cannot drain effectively. For higher permeabilities in the TW and LSB turbidites (x‡2), simulated pore pressures are a significant fraction of lithostatic and are consistent with observed taper angles (Figs 8 and 9). This implies that if a lower-bound fit to the permeability data is applicable, the permeability of the turbidite units must be higher than that of the mudstones to allow sufficient drainage. On the basis of the results for the Muroto transect noted above, taken together with results from the previous 1D coupled modeling studies of Gamage & Screaton (2006) and Skarbek & Saffer (2009) which demonstrate that the lower bound fit to permeability data is needed to match observed porosity evolution and inferred pore pressures at Sites 1174 and 808, I consider the lower bound fit to the permeability data as a preferred model. This is justified because between faults or permeable strata, layer-normal permeability should control fluid escape from the sediments; thus a lower bound to the laboratory data is the most appropriate value for the bulk mudstone (Bear & Verruijt 1987). As turbidite permeabilities are increased from 2 to 100 times that of the mudstone (x = 2–100), pore pressures within the wedge along the Muroto transect decrease significantly, from average values of kw = 0.74–0.83 to kw =
(A)
1000
Depth registered to depth at deformation front (m)
contrasts in permeability above and below the de´collement have the potential to cause substantially different basal and wedge pore pressures. Thus, I consider kb and kw separately in assessing critical taper. I define kb using the average value for nodes in the underthrust, and define kw for the wedge interior using the lowermost 12 nodes above the de´collement (e.g., Matmon & Bekins 2006); in both cases I compute the average using values weighted by the nodal area. Based on experimental measurements of friction coefficient for USB and LSB facies mudstones in the study area, I use values of lb ranging from 0.25–0.35, and lw = 0.4 to evaluate the taper angles (Brown et al. 2003; Kopf & Brown 2003).
0 0.60
–1000
0.80
0.70
0.90
–2000
Muroto transect V.E. = 4x 4000
(B)
2000
0.60
0 0.60
0.50
0.60
0.70
–2000
Ashizuri transect V.E. = 4x
–4000 40
30
20
10
0
–10
Distance from deformation front (km) Fig. 8. Example of simulated pore pressures (reported as the normalized pore pressure ratio k) for the Muroto (A) and Ashizuri (B) transects. The Ashizuri transect is better drained along its base, but more poorly drained in the deep interior of the wedge. Although the pore pressure ratio is smallest in the interior of the accretionary wedge, hydraulic head increases monotonically with distance into the accretionary complex in all simulations, such that fluid flow is upward and trenchward, toward the seafloor.
0.53–0.55 (Fig. 9A). Values of kb are less sensitive to the permeability of the turbidite units than are values of kw; as x is increased from 2 to 100, kb decreases only modestly, from 0.78–0.90 to 0.71–0.77. Along the Ashizuri transect, pore pressure in the wedge is similarly affected by permeability of the turbidite facies, decreasing from kw = 0.75– 0.84 to 0.62–0.67. However, the presence of turbidites in the LSB facies results in much greater sensitivity of kb to the permeability assigned to turbidite units than along the Muroto transect, with values decreasing from kb = 0.67– 0.78 to 0.56–0.64 as x is increased from 2 to 100. This is also evident from the differences in slope of relationships between kb and kw defined by the model simulations shown in Fig. 9A. For both transects, the range in reported values of k reflects the range of de´collement permeabilities explored (Fig. 9). For values of x = 10–100, simulated basal pore pressures of kb = 0.77–0.80 (Fig. 9A) are in good agreement with those inferred from seismic reflection interval velocities over the outer 20 km of the Muroto transect (kb = 0.75) (Tobin & Saffer 2009), and those estimated from the 1D coupled model of Skarbek & Saffer (2009). Simulations designed to investigate the role of permeability anisotropy in the LSB turbidite facies demonstrate that it has a small effect on modeled pore pressures. As the
Effects of hydrostratigraphy on drainage 125
1 (A)
Turbidite permeability
(B)
Permeability anisotropy
0.9 to Muro
lb
0.8 10
0.7
100
3
5
w=2
uri
1000
3
A
5 100 30
0.5 0 .5
200
30
z shi
0.6
Kxx:Kzz = 1.0 2–10 30 100
w=2
0.6
10
Kxx:Kzz = 10 000
0.7
0.8
0 .9
lw
0 .5
0 .6
0.7
0.8
0 .9
1 .0
lw
Fig. 9. (A) Simulated values of kw and kb averaged over the front 30 km of the accretionary complex, showing the effect of turbidite facies permeability. Boxes show the range of pore pressure ratios for a range of kd spanning 1 order of magnitude (Table 2); circles show results for the case of kd = 10)14 m2 at the trench. Gray shaded regions denote combinations of kw and kb that define critically tapered accretionary wedges along the two transects as labeled, for lw = 0.40 and lb = 0.25–0.35. For cases in which the turbidite rich units are more permeable than the hemipelagic mudstones, the along-strike differences in simulated pore pressures are sufficient to explain the variations in taper angle between the two transects. (B) Simulated pore pressures for the Ashizuri transect showing the effect of permeability anisotropy within the underthrusting LSB turbidite facies, for cases in which the TW turbidites are three times (black circles), 10 times (squares), and 30 times (black triangles) more permeable than the hemipelagic mudstones. Anisotropy ratios of 1000 are required for simulated pore pressures to explain the observed taper angle along the Ashizuri transect. The dashed boxes denote simulated pore pressures for values of w = 3, 30, and 100 without anisotropy in the LSB (also shown in panel A) for comparison.
permeability anisotropy ratio kxx:kzz is increased from 1:1 to 200:1, simulated values of kb averaged over the front 30 km of the subduction complex decrease from 0.83– 0.86 to 0.76–0.77 (Fig. 9B); increasing the anisotropy ratio to values as high as 10 000:1 reduces kb to 0.54– 0.55. In order to reach a drainage state similar to the case of x = 3 (bulk permeability of the LSB turbidites set equal to three times that of the hemipelagic mudstones), anisotropy ratios of 1000 are required (Fig. 9B). The limited sensitivity of basal pore pressure to permeability anisotropy is attributable to the significant difference in drainage path length for horizontal versus vertical fluid escape (tens of kilometers horizontal distance to a hydrostatic boundary, versus hundreds of meters vertical distance to either the permeable de´collement or the seafloor). This result is also consistent with previous studies, which have shown that even in layered subducting sediment sections, dewatering of underthrust sediment appears to be dominantly vertical (von Huene & Lee 1982; Screaton 2006). Simulated pore pressures exhibit a secondary dependence on de´collement permeability (kd) (Fig. 9A). For example, with x = 3, increasing de´collement permeability along the Muroto transect by 1 order of magnitude results in a relatively modest decrease in kb from 0.89 to 0.76, and in kw from 0.77 to 0.69. This is consistent with several previous studies (e.g., Saffer & Bekins 2006; Spinelli et al. 2006),
and can be explained by the fact that fluid escape from the mudstones above and below the de´collement is primarily limited by the bulk sediment permeability, which governs the access of fluids to the more permeable de´collement. With increased turbidite permeability (x = 30), the effect of increasing kd on simulated pore pressure in the underthrust section is similar, with kb decreasing from 0.77 to 0.72. However, the effect on wedge pore pressure is diminished, with kw decreasing only from 0.57 to 0.55, because drainage is dominated by upward fluid escape through the permeable TW turbidites and as a result, the permeability of the de´collement plays a smaller role. The difference in simulated pore pressures between the two transects is readily explained by permeable turbidites in the underthrust section along the Ashizuri corridor, which increase drainage efficiency along the base of the wedge and result in lower values of kb. In contrast, the absence of a subducted turbidite facies along the Muroto transect leads to higher simulated values of kb, and thus to decreased sensitivity of kb to turbidite permeability. In effect, the subduction accretion complex along the Ashizuri transect is ‘doubly drained’, at its top and base, whereas that along the Muroto transect drains only at the top and to a lesser extent along the de´collement zone. One key implication is that the detailed lithostratigraphy (and thus hydrostratigraphy) of the incoming sediment section plays
126 D. M. SAFFER an important role in controlling spatial variations in pore pressure, which ultimately mediate both the wedge taper angle and the mechanical strength of the wedge and underlying plate boundary fault. Although it is often assumed that accreted sediments are entirely composed of permeable TW turbidites and that the underthrust section contains only fine-grained pelagic sediments (e.g., Le Pichon et al. 1993), drilling at several margins shows that the de´collement actually often localizes within hemipelagic and pelagic sediments, and that a considerable thickness of hemipelagic sediments lies above the detachment and is accreted (e.g., Moore et al. 2001). Furthermore, at many locations, including N. Barbados and parts of the Nankai Margin, the underthrust sediments contain coarse-grained turbidite sequences (Mascle et al. 1988; Moore et al. 2001). Recent results from the Integrated Ocean Drilling Program Nankai Trough Seismogenic Zone Experiment (NanTroSEIZE), focused on a transect 120 km northeast of Cape Muroto offshore the Kii peninsula (Fig. 1), provide another example (Tobin et al. 2009). Along this corridor, the stratigraphy of the incoming section appears similar to that along the Ashizuri transect, in that the LSB is several hundred meters thicker than along the Muroto transect and includes abundant turbidites (e.g., Ike et al. 2008). Preliminary results from drilling are consistent with the modeling results reported here, in that they indicate the system is well-drained beneath the de´collement, perhaps suggesting that the plate boundary system is strong in comparison with the Muroto transect (Screaton et al. 2008). Critical taper angle and mechanical strength of the plate boundary The results described above illustrate that by incorporating a set of simple and plausible assumptions about permeability variations between major hydro-stratigraphic units, along-strike differences in lithostratigraphy generate systematic along-strike variations in pore pressure (Figs 8 and 9), which are sufficient to account for the observed taper angles along the two transects. Specifically, the observed taper angles can be explained if the bulk permeability of the turbidite-rich units is between 2 and 100 times higher than that of the USB and LSB hemipelagic mudstones (x = 2–100) (Fig. 9). In contrast, for permeability anisotropy in the LSB turbidite facies to drive along-strike variations in pore pressure sufficient to explain the observed taper angles, anisotropy ratios of 1000 or more are required (Fig. 9B). The presence of permeable turbidities in the underthrust section along the Ashizuri transect, but not along the Muroto, results in a lower basal pore pressure ratio along the Ashizuri transect. This has two additional consequences. First, simulated pore pressures along the Ashizuri transect
are generally higher in the deepest parts of the accretionary wedge than along the base, such that kb < kw (Fig. 9). This results in stable critically tapered accretionary wedges only if lb is modestly lower than lw (Davis et al. 1983; Deng & Underwood 2001); for the values of lb and lw defined by laboratory shearing experiments simulated pore pressures are indeed consistent with a critical taper (Fig. 9). Second, the wedge base along the Muroto transect should be mechanically weaker than along the Ashizuri, because the vertical effective stress there is lower (Fig. 10A–C). For example, simulations with x = 10–30 result in effective vertical stress along the Ashizuri transect increasing from 6 MPa at the trench to 56–60 MPa within the outer 30 km of the system, whereas that along the Muroto transect increases from 6 MPa to only 14–16 MPa (Fig. 10C). This result is in close agreement with effective stresses inferred from seismic reflection interval velocities by Tobin & Saffer (2009). For a friction coefficient of 0.35, these effective stresses correspond to a shear strength of 20 MPa along the wedge base 30 km from the trench along the Ashizuri transect, and 4.5 MPa along the Muroto transect (Fig. 10D). The simulated shear strength along the Muroto transect is approximately one-fourth of that along the Ashizuri transect. This is a considerably larger difference than expected only on the basis of observed taper angles, owing primarily to the dramatic difference in drainage state along the wedge base that is not explicitly considered in previous analyses of critical taper, which assumed kw = kb (Davis et al. 1983; Brown et al. 2003). One useful way to report the effects of pore pressure on shear strength is by the effective friction coefficient: l¢ = (1 ) kb)l (Davis et al. 1983; Wang & He 1999). However, by this formulation l¢ is lower than the true friction coefficient (l) even in the common case of hydrostatic pore pressure, for which k = 0.40–0.70 depending on the bulk density of the overburden. In order to more clearly illustrate the effect of pore pressure on shear strength, I define a modified effective friction coefficient by lb¢¢ = (1 ) kb*)lb, where k* is the modified pore pressure ratio defined by k* = (Pf ) Ph) ⁄ (Pl ) Ph), and Ph is hydrostatic pressure (Shi & Wang 1988). Values of l¢¢ range from the true friction coefficient for hydrostatic pore pressure, to zero for lithostatic pore pressure (Fig. 10D). Simulated pore pressures at the base of the subduction– accretion complex result in values of lb¢¢ along the Muroto transect that decrease from 0.28 at the trench to 0.10– 0.14 by 13 km landward, and remain low to >30 km landward. This corresponds to a shear strength 29–40% of that expected for the case of hydrostatic pore pressure. In contrast, values of lb¢¢ along the Ashizuri transect increase from 0.22 to 0.30 between the trench and 30 km arcward, corresponding to a shear strength 63–86% of that for the hydrostatically pressured case.
Effects of hydrostratigraphy on drainage 127
60
Lith
Pressure (MPa)
80
ost
atic
40
Hydros
20
tatic
0 30 50
(C)
25
20
15
10
5
0
hiz
ur
i
30 20
Muroto 10 0 30
Lit
ho
25 20 15 10 5 Distance from deformation front (km)
0
sta
tic
60 40
Hyd
rosta
20
(D)
As
40
80
0 30
–5
Ashizuri
(B)
25
20
tic
15
10
5
0
Ashizuri
–5 0.3 0.2
Muroto 20
Ash
izuri
0.1
(1-λ)*μ
Effective stress (MPa)
100
Muroto
(A)
Shear strength (MPa)
Pressure (MPa)
100
0
10
Muroto 0 30
25 20 15 10 5 Distance from deformation front (km)
0
Fig. 10. Hydrostatic pressure, lithostatic pressure, and simulated pore pressure along the base of the wedge (in the uppermost underthrust sediments) along the Muroto (A) and Ashizuri (B) transects. For both transects, simulated pressures are shown for x = 10 (upper curve) and x = 30 (lower curve). (C) Effective stress along the base of the wedge for the two transects, for x = 10 and 30 as in panels (A) and (B). Effective stresses along the Muroto transect inferred from seismic interval velocities by Tobin & Saffer (2009), (grey dots d) are shown for comparison. (D) Shear strength (red) computed from effective stresses in panel (C) and assuming lb = 0.35, and modified effective friction coefficient normalized to a value of lb = 0.35 at hydrostatic pore pressure as described in the text (blue).
Competing roles of geologic forcing and fluid escape in governing mechanical strength Placed in a more general context, the modeling results reported here provide additional support for the idea that development of excess pore pressure is governed by the dynamic interplay between processes that drive excess pore pressure, and those allowing it to dissipate (Neuzil 1995). In the case of subduction zones, continuous tectonic loading and mineral dehydration represent hydrologic forcing that drives pore pressure development. The drainage of excess pressure is mediated by the path length and effective hydraulic conductivity along pathways of fluid escape. In his analysis, Neuzil (1995) illustrated that the balance between these terms can be characterized in a wide variety of settings, and over a wide range of scales, by the simple criterion that excess pore pressures are expected if the dimensionless quantity CL ⁄ K > 1, where C is the fluid source term (VH2O ⁄ Vsed sec)1), L is the characteristic path length for fluid escape, and K is the effective hydraulic conductivity (m sec)1). In subsequent work, Saffer & Bekins (2006) and Spinelli et al. (2006) specifically demonstrated the applica-
bility of this formulation to subduction zones, despite their complex geometry and highly heterogeneous permeability structure. These studies also extended the original analysis of Neuzil (1995) by showing that, in detail, the magnitudes of excess pore pressure and CL ⁄ K co-vary (Fig. 11). I follow the approach of Saffer & Bekins (2006) and Spinelli et al. (2006), and compute CL ⁄ K using representative values of C, L, and K for each model simulation. Because I focus on the strength along the wedge base in the outer 30 km of the subduction zone, I define C as the average fluid source term in the underthrust section between 10 and 20 km arcward of the trench, L as the path length upward from the middle of the underthrusting section to the de´collement and along the de´collement to the trench, and K as the effective hydraulic conductivity computed from the harmonic mean along the assumed flow path (e.g., Saffer & Bekins 2006). For subduction zones, the simple formulation of Neuzil (1995) provides a framework to explain not only the development of elevated pore pressure (Fig. 11A) but also mechanical strength by considering the role of effective stress. As CL ⁄ K increases, pore pressure also increases, and shear strength (as expressed by effective friction coefficient)
128 D. M. SAFFER
0.3 1
(A)
(B) 0.25
Muroto 0.9
0.2
lb
μ'b
0.8 0.7
0.1
ri
izu
Ash
0.6 0.5 0.01
0.15
0.05
0.1
1
10
Γ *L/K
0 0.01
0.1
1
10
Γ *L/K
Fig. 11. Dependence of simulated pore pressure ratio (A) and effective friction coefficient (B) on dimensionless ratio of the product of fluid source terms and mean flow path length normalized by effective hydraulic conductivity (Neuzil 1995). Results are shown for a suite of generic models designed to investigate the effects of bulk permeability (filled circles) and incoming sediment thickness (open circles) reported by Saffer & Bekins (2006), and for simulations reported in this work for the Muroto (red) and Ashizuri (green) transects as shown in Fig. 9A.
decreases (Fig. 11B). The quantity CL ⁄ K is generally smaller for the Ashizuri than the Muroto transect, and as a consequence, kb is lower and lb¢ is correspondingly higher. As noted above, this difference is caused by the higher permeability of underthrust sediment along the Ashizuri transect due to the LSB turbidite facies (both L and C are similar for the two transects). There is some variability in the systematic relationship, which I attribute to a combination of heterogeneity in hydraulic conductivity, the geometry of the flow system, and the flow path assumed when computing the effective values of K and L, none of which are perfectly captured by this simple analysis. For example, along the Muroto transect, pore pressure and effective friction coefficient vary (pore pressure decreases and lb¢ increases) with increased turbidite permeability, yet the quantity CL ⁄ K as computed here does not. This reflects the fact that in defining CL ⁄ K, I assume the path for fluid escape from the underthrusting section is upward and then seaward along the de´collement (e.g., von Huene & Lee 1982; Saffer & Bekins 1999; Screaton & Saffer 2005). However, the fact that permeability in the upper wedge affects kb suggests that in detail, the assumed flow path may not fully represent fluid escape from the system.
SUMMARY AND CONCLUSIONS Using a numerical model of fluid flow, together with laboratory measurements constraining bulk mudstone permeability, I find that along-strike variations in lithostratigraphy at the SW Nankai subduction margin lead to differences in pore pressure sufficient to explain observed differences in the accretionary wedge taper angle. Simulations incorporating the simple assumption that the turbidite-dominated sedimentary units are more permeable than mudstone units produce systematic differences in the pore pressure ratios kb and kw along-strike that can account for
the taper angles. In contrast, permeability anisotropy is less effective in generating variations in pore pressure ratio, and anisotropy ratios of kxx:kzz = 1000 for the turbidite units are required to explain the observations. This set of results provides a robust and quantitative explanation for the along-strike differences in taper angle at the Nankai margin, as has been suggested by others qualitatively (Moore et al. 1990; Stoffa et al. 1992; Brown et al. 2003). The differences in pore pressure, which are strongly mediated by the hydrostratigraphy of the incoming sediment section, lead directly to along-strike differences in shear strength along the wedge base (i.e., the plate boundary), which reaches 20 MPa at 30 km from the trench along the Ashizuri transect and only 5 MPa at the same distance along the Muroto transect. These shear strengths are in close agreement with values reported by previous studies (Skarbek & Saffer 2009; Tobin & Saffer 2009), and correspond to effective friction coefficients of l¢ = 0.10–0.14 for the Ashizuri transect, and values of l¢ = 0.03–0.08 along the more poorly drained Muroto transect. More generally, the results illustrate that the mechanical strength of subduction faults is directly related to the combined effects of fluid sources, fluid flow path length, and permeability, through their effects on pore pressure and effective stress.
ACKNOWLEDGEMENTS I thank Chris Marone, Kelin Wang, Barbara Bekins, and Liz Screaton for helpful discussions and comments during various stages of this work, and two anonymous reviewers for comments that improved the manuscript. This study was supported by NSF-OCE grants 0503905, 0451602, and 0752114 to Saffer; acknowledgment is made to the Donors 006Ff the American Chemical Society Petroleum Research Fund for partial support of this research.
Effects of hydrostratigraphy on drainage 129
REFERENCES Athy LF (1930) Density, porosity, and compaction of sedimentary rocks. American Association of Petroleum Geologists Bulletin, 14, 1–24. Bangs N, Shipley T, Moore J, Moore G (1990) Seismic velocities from the Barbados Ridge Complex; indicators of high pore fluid pressures in an accretionary complex. Journal of Geophysical Research, 95, 8767–82. Bangs NL, Shipley TH, Gulick SPS, Moore GF, Kuromoto S, Nakamura Y (2004) Evolution of the Nankai Trough decollement from the trench into the seismogenic zone: inferences from three-dimensional seismic reflection imaging. Geology, 32, 273–6. Bear J, Verruijt A (1987) Modeling Groundwater Flow and Pollution. Springer, New York. Becker K, Fisher AT, Davis EE (1997) The CORK experiment in hole 949C: long-term observations of pressure and temperature in the Barbados accretionary prism. Proceedings. Ocean Drilling Program. Scientific Results, 156, 247–52. Bekins BA, Dreiss SJ (1992) A simplified analysis of parameters controlling dewatering in accretionary prisms. Earth and Planetary Science Letters, 109, 275–87. Bekins B, Screaton E (2007) Pore pressure and fluid flow in the northern Barbados accretionary complex. In: The Seismogenic Zone of Subduction Thrust Faults (eds Dixon T, Moore JC), pp. 148–70, Columbia University Press, New York, NY. Bekins BA, McCaffrey AM, Dreiss SJ (1994) The influence of kinetics on the smectite to illite transition in the Barbados accretionary prism. Journal of Geophysical Research, 99, 18,147–58. Bekins BA, McCaffrey AM, Dreiss SJ (1995) Episodic and constant flow models for the origin of low-chloride waters in a modern accretionary complex. Water Resources Research, 31, 3205–15. Bethke C (1986) Inverse hydrogeologic analysis of the distribution and origin of gulf coast-type geopressured zones. Journal of Geophysical Research, 91, 6535–45. Bray CJ, Karig DE (1985) Porosity of sediments in accretionary prisms and some implications for dewatering processes. Journal of Geophysical Research, 90, 768–78. Bray C, Karig DE (1988) Dewatering and extensional deformation of the Shikoku Basin hemipelagic sediment in the Nankai Trough. Pure and Applied Geophysics, 128, 725–47. Breen N, Orange DL (1992) The effects of fluid escape on accretionary wedges: 1. Variable porosity and wedge convexity. Journal of Geophysical Research, 97, 9265–75. Brown KM, Kopf A, Underwood MB, Weinberger JL (2003) Compositional and fluid pressure controls on the state of stress on the Nankai subduction thrust: a weak plate boundary. Earth and Planetary Science Letters, 214, 589–603. Byrne T, Fisher D (1990) Evidence for a weak and overpressured decollement beneath sediment-dominated accretionary prisms. Journal of Geophysical Research, 95, 9081–97. Carson B, Screaton EJ (1998) Fluid flow in accretionary prisms: evidence for focused, time-variable discharge. Reviews of Geophysics, 36, 329–51. Cochrane G, Moore JC, Lee HJ (1996) Sediment pore-fluid overpressuring and its effect on deformation at the toe of the Cascadia accretionary prism from seismic velocities. In: Subduction Top to Bottom (ed Bebout G, Platt J, Scholl D, Kirby S), pp. 57–64. Geophysical Monograph 96. American Geophysical Union, Washington, DC. Costa-Pisani P, Reshef M, Moore G (2005) Targeted 3-D prestack depth imaging at Legs 190–196 ODP drill sites (Nankai
Trough, Japan). Geophysical Research Letters, 32, L20309. doi: 10.1029/2005GL024191. Cutillo PA, Screaton EJ, Ge S (2003) Three-dimensional numerical simulation of fluid flow and heat transport within the Barbados Ridge accretionary complex. Journal of Geophysical Research, 108, 2555. doi: 10.1029/2002JB002240. Dahlen FA (1984) Noncohesive critical Coulomb wedges; an exact solution. Journal of Geophysical Research, 89, 10125–33. David C, Wong T-f, Zhu W, Zhang J (1994) Laboratory measurement of compaction-induced permeability change in porous rocks: implications for the generation and maintenance of pore pressure excess in the crust. PAGEOPH, 143, 425–56. Davis DM (1996) Accretionary mechanics with properties that vary in space and time. In: Subduction: Top to Bottom (ed Bebout G, Platt J, Scholl D, Kirby S), American Geophysical Union Geophysical Monograph, 96, 39–48. Davis D, Suppe J, Dahlen FA (1983) Mechanics of fold-and-thrust belts and accretionary wedges. Journal of Geophysical Research, 88, 1153–72. Davis EE, Becker K, Wang K, Obara K, Ito Y, Kinoshita M (2006) A discrete episode of seismic and aseismic deformation of the Nankai trough subduction zone accretionary prism and incoming Philippine Sea plate. Earth and Planetary Science Letters, 242, 73–84. Deng X, Underwood MB (2001) Abundance of smectite and the location of a plate-boundary fault, Barbados accretionary prism. Geological Society of America Bulletin, 113, 495–507. Ferguson IJ et al. (1993) Heat flow and thermal models of the Barbados Ridge accretionary complex. Journal of Geophysical Research, 98, 4121–42. Foucher J-P, Henry P, Harmegnies F (1997) Long-term observations of pressure and temperature in hole 948D, Barbados accretionary prism. Proceedings. Ocean Drilling Program. Scientific Results, 156, 239–45. Gamage KR (2005) Permeabilities of subduction zone sediments and their effect on pore pressure generation, PhD Thesis, The University of Florida, 133 pp. Gamage K, Screaton E (2006) Characterization of excess pore pressures at the toe of the Nankai accretionary complex, Ocean Drilling Program sites 1173, 1174, and 808: results of onedimensional modeling. Journal of Geophysical Research, 111, B04103. doi: 10.1029/2004JB003572. Henry P et al. (2003) Anisotropy of electrical conductivity record of initial strain at the toe of the Nankai accretionary wedge. Journal of Geophysical Research, 108, 2407. doi: 10.1029/2002JB002287. Hickman S, Sibson R, Bruhn R (1995) Introduction to special section: mechanical involvement of fluids in faulting. Journal of Geophysical Research, 100, 12831–40. Hubbert MK, Rubey WW (1959) Role of fluid pressure in mechanics of overthrust faulting. Geological Society of America Bulletin, 70, 115–66. von Huene R, Lee H (1982) The possible significance of pore fluid pressures in subduction zones. In: Studies in Continental Margin Geology (eds Watkins JS, Drake CL), American Association of Petroleum Geologists Memoir, 34, 781–91. Hyndman RD et al. (1993) Velocity, porosity, and pore-fluid loss from the Nankai subduction zone accretionary prism. Proceedings. Ocean Drilling Program. Scientific Results, 131, 211–19. Ike T, Moore GF, Kuramoto S, Park J-O, Kaneda Y, Taira A (2008) Variations in sediment thickness and type along the northern Philippine Plate at the Nankai Trough. The Island Arc, 17, 342–57. doi: 10.1111/j.1440-1738.2008.00624.x. Karig DE (1990) Experimental and observational constraints on the mechanical behaviour in the toes of accretionary prisms. In:
130 D. M. SAFFER Deformation Mechanics, Rheology and Tectonics (eds Knipe RJ, Rutter EH), Geol. Soc. Spec. Publ. London, 54, 383–98. Kastner M, Elderfield H, Martin JB (1991) Fluids in convergent margins; what do we know about their composition, origin, role in diagenesis and importance for oceanic chemical fluxes? Philosophical Transactions of the Royal Society of London, Series A: Mathematical and Physical Sciences, 335, 243–59. Kopf A, Brown KM (2003) Friction experiments on saturated sediments and their implications for the stress state of the Nankai and Barbados subduction thrusts. Marine Geology, 202, 193–210. Le Pichon X, Henry P, Lallemant S (1990) Water flow in the Barbados accretionary complex. Journal of Geophysical Research, 95, 8945–67. Le Pichon X, Henry P, Lallemant S (1993) Accretion and erosion in subduction zones: the role of fluids. Annual Review of Earth and Planetary Sciences, 21, 307–31. Long H, Flemings PB, Germaine JT, Saffer DM, Dugan BE (2008) Data report: consolidation characteristics of sediments from IODP Expedition 308, Ursa Basin, Gulf of Mexico. In Flemings, P.B., Behrmann, J.H., John, C.M., and the Expedition 308 Scientists, Proceedings of the Integrated Ocean Drilling Program, 308. (Integrated Ocean Drilling Program Management International, Inc., College Station, TX). doi:10.2204/ iodp.proc.308.204.2008. Mascle A, Moore JC, Taylor E, Alvarez F, Andreieff P, Barnes R, Beck C, Behrmann J, Blanc G, Brown K, Clark M, Dolan J, Fisher A, Gieskes J, Hounslow M, McLellan P, Moran K, Ogawa Y, Sakai T, Schoonmaker J, Vrolijk P, Wilkens R, Williams C (1988) Proceedings. Ocean Drilling Program. Initial Reports (A), 110, pp. 603. Matmon D, Bekins BA (2006) Hydromechanics of a high taper angle, low-permeability prism: a case study from Peru. Journal of Geophysical Research, 111, B07101. doi: 10.1029/2005JB00 3697. McKiernan AW (2005) Smectite dehydration and updip fluid flow in the Nankai Trough, offshore southwestern Japan. M.S. Thesis, The University of Wyoming, 106 pp. Moore JC, Byrne T (1987) Thickening of fault zones: a mechanism of me´lange formation in accreting sediment. Geology, 15, 1040–3. Moore JC, Saffer DM (2001) Updip limit of the seismogenic zone beneath the accretionary prism of southwest Japan: an effect of diagenetic to low-grade metamorphic processes and increasing effective stress. Geology, 29, 183–6. Moore GF, Shipley TH, Stoffa PL, Karig DE, Taira A, Kuramoto S, Tokuyama H, Suyehiro K, (1990) Structure of the Nankai Trough accretionary zone from multichannel seismic reflection data. Journal of Geophysical Research 95, 8753–65. Moore GF, Taira A, Klaus A, Becker L, Boeckel B, Cragg BA, Dean A, Fergusson CL, Henry P, Hirano S, Hisamitsu T, Hunze S, Kastner M, Maltman AJ, Morgan JK, Murakami Y, Saffer DM, Sa´nchez-Go´mez M, Screaton EJ, Smith DC, Spivack AJ, Steurer J, Tobin HJ, Ujiie K, Underwood MB, Wilson M (2001) New insights into deformation and fluid flow processes in the Nankai accretionary prism: results of Ocean Drilling Project Leg 190. Geochemistry, Geophysics, Geosystems, 2, 10.129/2001GC000166. Morgan JK, Ask MVS (2004) Consolidation state and strength of underthrust sediment and evolution of the de´collement at the Nankai accretionary margin: results of uniaxial reconsolidation experiments. Journal of Geophysical Research, 109, B03102. doi: 10.1029/2002JB002335. Morgan JK, Karig DE (1993) Ductile stains in clay-rich sediment from Hole 808C; preliminary results using X-ray pole figure
goniometry. Proceedings. Ocean Drilling Program. Scientific Results , 131, 141–55. Nelson PH (1994) Permeability-porosity relationships in sedimentary rocks. 35, 38–62. Neuzil CE (1994) How permeable are clays and shales? Water Resources Research, 30, 145–50. Neuzil CE (1995) Abnormal pressures as hydrodynamic phenomena. American Journal of Science, 295, 742–86. Olson RE (1986) State of the art: consolidation testing. In: Consolidation of Soils: Testing and Evaluation: A Symposium (eds Yong RN, Townend FC), pp. 7–70. American Society for Testing and Materials., Philadelphia, PA. Peacock SM (2009) Thermal and metamorphic environment of subduction zone episodic tremor and slip. Journal of Geophysical Research, 114, B00A07. doi: 10.1029/2008JB00597. Raleigh CB, Healy JH, Bredehoeft JD (1976) An experiment in earthquake control at Rangely, Colorado. Science, 191, 1230–7. doi: 10.1126/science.191.4233.1230. Rice JR (1992) Fault stress states, pore pressure distributions, and the weakness of the San Andreas fault. In: Fault Mechanics and Transport Properties of Rocks (eds Evans B, Wong TF), 475– 503, Academic Press, San Diego, CA. Saffer DM (2003) Pore pressure development and progressive dewatering in underthrust sediments at the Costa Rica subduction margin: comparison with northern Barbados and Nankai. Journal of Geophysical Research, 108, 2261. doi: 1029/ 2002JB001787. Saffer DM, Bekins BA (1998)Episodic fluid flow in the Nankai accretionary complex: timescale, geochemistry, and fluid budget. Journal of Geophysical Research, 103, 30,351–70. Saffer DM, Bekins BA (1999) Fluid budgets at convergent plate margins: implications for the extent and duration of fault-zone dilation. Geology, 29, 1095–8. Saffer DM, Bekins BA (2002) Hydrologic controls on the morphology and mechanics of accretionary wedges. Geology, 30, 271–4. Saffer DM, Bekins BA (2006) An evaluation of factors influencing pore pressure in accretionary complexes: implications for taper angle and wedge mechanics. Journal of Geophysical Research, 111, B04101. doi: 10.1029/2005JB003990. Saffer DM, McKiernan AW (2005) Permeability of underthrust sediments at the Costa Rican subduction zone: scale dependence and implications for dewatering. Geophysical Research Letters, 32, L02302. doi: 10.1029/2004GL021388. Saffer DM, McKiernan AW (2009) Evaluation of in situ smectite dehydration as a pore water freshening mechanism in the Nankai Trough, offshore southwest Japan. Geochemistry, Geophysics, Geosystems, 10, Q02010. doi: 10.1029/2008GC00 2226. Saffer DM, Underwood MB, McKiernan AW (2008) Evaluation of factors controlling smectite transformation and fluid production in subduction zones: application to the Nankai Trough. The Island Arc, 17, 208–30. doi: 10.1111/j.14401738.2008.00614.x. Screaton E (2006) Excess pore pressures within subducting sediments: does the proportion of accreted versus subducted sediments matter? Geophysical Research Letters, 33, L10304. doi: 10.1029/2006GL025737. Screaton E, Ge S (1997) An assessment of along-strike fluid and heat transport within the Barbados ridge accretionary complex: results of preliminary modeling. Geophysical Research Letters, 24, 3085–8. Screaton EJ, Saffer DM (2005) Fluid expulsion and overpressure development during initial subduction at the Costa Rica con-
Effects of hydrostratigraphy on drainage 131 vergent margin. Earth and Planetary Science Letters, 233, 361–74. Screaton EJ, Wuthritch DR, Dreiss SJ (1990) Permeabilities, fluid pressures, and flow rates in the Barbados ridge complex. Journal of Geophysical Research, 95, 8997–9007. Screaton EJ, Saffer DM, Henry P, Hunze S, Leg 190 Shipboard Scientific Party (2002) Porosity loss within the underthrust sediments of the Nankai accretionary complex: implications for overpressures. Geology, 30, 19–22. Screaton EJ, Kimura G, Curewitz D, Chester F, Fabbri O, Fergusson C, Girault F, Goldsby D, Harris R, Inagaki F, Jiang T, Kitamura Y, Knuth M, Li C-F, Liljedahl LC, Louis L, Milliken K, Nicholson U, Riedinger N, Sakaguchi A, Solomon E, Strasser M, Su X, Tsutsumi A, Yamaguchi A, Ujiee K, Zhao X (2008) Initial results from the Nankai Trough shallow splay and frontal thrust (IODP Expedition 316): implications for fluid flow. Eos Transactions. AGU, 89, Fall Meet. Suppl., Abstract. Seno T, Stein S, Gripp AE (1993) A model for the motion of the Philippine Sea plate consistent with NUVEL-1 and geological data. Journal of Geophysical Research, 98, 17,941–8. Shi Y, Wang CY (1988) Generation of high pore pressures in accretionary prisms: inferences from the Barbados subduction complex. Journal of Geophysical Research, 93, 8893–910. Shipboard Scientific Party (1991) Site 808. In: Initial Reports of the Ocean Drilling Program (eds Hill IA, Taira A, Firth JV) pp. 71–269. Ocean Drill. Program, College Station, TX. Shipboard Scientific Party (2001), Leg 190 preliminary report: deformation and fluid flow processes in the Nankai Trough accretionary prism. Ocean Drill. Program, College Station, TX. http://www.odp.tamu.edu/publications/prelim/190_prel/190 PREL.PDF. Sibson RH (1981) Fluid flow accompanying faulting: field evidence and models. In: Earthquake Prediction: An International Review (eds Simpon DW, Richards PG), Maurice Ewing Ser., Vol. 4, pp. 593–603. AGU, Washington, DC. Skarbek RM, Saffer DM (2009) Pore pressure development beneath the de´collement at the Nankai subduction zone: implications for plate boundary fault strength and sediment dewatering. Journal of Geophysical Research, 114, B07401. doi: 10.1029/2008JB006205. Spinelli GA, Saffer DM (2007) Trench-parallel fluid flow in subduction zones resulting from temperature differences. Geochem. Geophys. Geosyst., 8, Q09009. doi: 10.1029/2007GC001673. Spinelli GA, Saffer DM, Underwood MB (2006) Hydrogeologic responses to three-dimensional temperature variability, Costa
Rica subduction margin. Journal of Geophysical Research, 111, B04403. doi: 10.1029/2004JB003436. Spivack AJ, Kastner M, Ransom B (2002) Elemental and isotopic chloride geochemistry and fluid flow in the Nankai Trough. Geophysical Research Letters, 13, 1661. doi: 10.1029/ 2001GL014122. Stoffa P, Wood W, Shipley T, Moore G, Nishiyama E, Botelho M, Taira A, Tokuyama H, Suyehiro K (1992) Deepwater high-resolution expanding spread and split spread seismic profiles in the Nankai Trough. Journal of Geophysical Research, 97, 1687–713. B07401. doi: 10.1029/2008JB006205. Suppe J (2007) Absolute fault and crustal strength from wedge tapers. Geology, 35, 1127. doi: 10.1130/G24053A.1. Tobin HJ, Saffer DM (2009) Elevated fluid pressure and extreme mechanical weakness of a plate boundary thrust, Nankai Trough subduction zone. Geology, 37, 679–82. doi: 10.1130/ G25752A. Tobin H, Kinoshita M, Ashi J, Lallemant S, Kimura G, Screaton E, Moe KT, Masago H, Curewitz D, and the Expedition 314 ⁄ 315 ⁄ 316 Scientists (2009) NanTroSEIZE Stage 1 expeditions: introduction and synthesis of key results. In: Proceedings of the Integrated Ocean Drilling Program (eds Kinoshita M, Tobin H, Ashi J, Kimura G, Lallemant S, Screaton EJ, Curewitz D, Masago H, Moe KT, and the Expedition 314 ⁄ 315 ⁄ 316 Scientists), 314 ⁄ 315 ⁄ 316. Integrated Ocean Drilling Program Management International, Inc., Washington, DC. doi: 10.2204/iodp.proc.314315316.101.2009. Underwood MB (2007) Sediment inputs to subduction zones: why lithostratigraphy and clay mineralogy matter. In: The Seismogenic Zone of Subduction Thrust Faults (eds Dixon T, Moore JC), pp. 42–85, Columbia University Press, New York. Voss CI (1984) A finite element simulation model for saturatedunsaturated, fluid density-dependent groundwater flow with energy transport or chemically reactive single-species solute transport. U.S. Geological Survey Water Resources Investigations Report, 84–4369. Wang K (1994) Kinematics of dewatering accretionary prisms. Journal of Geophysical Research, 99, 4429–38. Wang K, He J (1999) Mechanics of low-stress forearcs: Nankai and Cascadia. Journal of Geophysical Research, 104, 15,191–205. Westbrook GK (1994) Growth of accretionary wedges off Vancouver Island and Oregon. In: Proc. ODP, Init. Repts. (eds Westbrook GK, Carson B, Musgrave RJ), Vol. 146, pp. 381–8. Ocean Drilling Program, College Station, TX. doi: 10.2973/odp.proc. ir.146-1.011.1994.
The interplay of permeability and fluid properties as a first order control of heat transport, venting temperatures and venting salinities at mid-ocean ridge hydrothermal systems T. DRIESNER Department of Earth Sciences, ETH Zurich, Switzerland
ABSTRACT While the fundamental influence of fluid properties on venting temperatures in mid-ocean ridge (MOR) hydrothermal systems is now well established, the potential interplay of fluid properties with permeability in controlling heat transfer, venting temperatures, and venting salinities has so far received little attention. A series of numerical simulations of fully transient fluid flow in a generic, across-axis model of a MOR with a heat input equivalent to magmatic supply at a spreading rate of 10 cm year)1 shows a strong dependence of venting temperature and salinity on the system’s permeability. At high permeability, venting temperatures are low because fluid fluxes are so high that the basal conductive heating cannot warm the large fluid masses rapidly enough. The highest venting temperature around 400C as well as sub-seafloor fluid phase separation occur when the permeability is just high enough that the fluid flux can still accommodate all heat input for advection, or for lower permeabilities where advection is no longer capable to transfer all incoming magmatic heat. In the latter case, additional mechanisms such as eruptions of basaltic magma may become relevant in balancing heat flow in MOR settings. The results can quantitatively be explained by the ‘fluxibility’ hypothesis of Jupp & Schultz (Nature, 403, 2000, 880), which is used to derive diagrams for the relations between basal heat input, permeability and venting temperatures. Its predictive capabilities are tested against additional simulations. Potential implications of this work are that permeability in high-temperature MOR hydrothermal systems may be lower than previously thought and that low-temperature systems at high permeability may be an efficient way of removing heat at MOR. Key words: black smoker, convection, fluxibility, heat transport, hydrothermal systems, mid-ocean ridges, permeability Received 19 July 2009; accepted 22 November 2009 Corresponding author: Thomas Driesner, Department of Earth Sciences, ETH Zurich, Clausiusstrasse 25, 8092 Zu¨rich, Switzerland. Email:
[email protected]. Tel: +41 44 632 68 03. Fax: +41 44 632 18 27. Geofluids (2010) 10, 132–141
INTRODUCTION Hydrothermal systems on the ocean floor play an outstanding role in the Earth’s heat flow distribution as they are thought to accommodate about 25% of the total crustal heat output (e.g., Stein & Stein 1994). Much of this is likely happening at mid-ocean ridges (MOR) although the relative contributions of high-temperature, on-axis, black smoker type systems versus diffuse flow or off-axis systems are still debated (e.g., Stein & Stein 1994; Fisher 1998, 2005; Rabinowicz et al. 1998, 1999; Fisher & Becker 2000; Stein & Fisher 2001, 2003; Baker & German 2004; German & Lin 2004; Hutnak et al. 2006; Baker 2007; Fontaine et al. 2008).
Much of this debate simply results from the fact that characterizing hydrothermal systems at great water depth and mapping out the distribution of different hydrothermal system types is a technically very challenging and expensive exercise. Detailed observations so far have only become available for a small fraction of the global MOR ridge length of 60 000–70 000 km (e.g., Baker & German 2004; Baker 2007). In spite of substantial technical improvements (Ramondenc et al. 2006), measurements of hydrothermal heat output at these sites remain subject to large uncertainties (Baker 2007): for fast-spreading ridges, typical heat output is in the range of a few hundred megawatt (MW) per vent field (with an average of 245 ± 170 MW) while in slow-spreading regimes, reported values are typically higher
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
The interplay of permeability and fluid properties 133 (1669 ± 1354 MW). This seems at odds with the fact that magmatic heat input should scale with spreading rate and that calculated steady state magmatic heat supply is always much lower than these values (Sinha & Evans 2004). A possible explanation is that magmatic heat input is episodic and rather local (e.g., Wilcock & Delaney 1996; German & Lin 2004), a view that is further substantiated by the fact that at steady state, heat supply to individual fields would come from unrealistically long lengths along an axis (Baker & German 2004; Baker 2007). A second factor of uncertainty is the still limited fundamental theoretical knowledge about the physics of hydrothermal convection at MOR. Some basic insights have been gained from ‘single pass’ or ‘U-tube’ analytical models (Lowell & Germanovich 2004; and references therein) that prescribed a simple convection geometry and used linearized fluid properties. Insights into the dynamics of these systems and the evolution of convection cell geometries and their influence on heat transport have been obtained from ‘cellular convection’ simulations (e.g., Wilcock 1998; Cherkaoui & Wilcock 1999; Fontaine & Wilcock 2006, 2007). A major step forward was the inclusion of the nonlinear temperature-pressure dependence of water properties (density, specific enthalpy, and viscosity) in transient fluid flow simulations (Jupp & Schultz 2000). It could be demonstrated that these control the onset of convection as well as the maximum venting temperatures of approximately 400C. Once these non-linear fluid properties are accounted for, simple ‘single pass’ models seem to provide quantitative conclusions about several key aspects of MOR hydrothermal system behavior (Jupp & Schultz 2004). More recently, fully transient simulations with accurate non-linear water properties in 2D and 3D have revealed additional insight into the dynamical evolution and fundamental physics of these systems (Coumou et al. 2006, 2008). All studies using real water properties pointed out the importance of non-linear temperature-pressure dependencies for self-organization of convection cells, in particular with respect to their thermal structure and time evolution, a behavior which is absent in simplified ‘single pass’ models and transient simulations with linearized fluid properties. The latest developments in simulation techniques now allow to incorporate the phase relations and fluid properties as a function of temperature, pressure and composition in the system H2O-NaCl (Coumou et al. 2009; Lewis & Lowell 2009). Hence, physical effects such as sub-seafloor phase separation of the saline hydrothermal fluid into vapor and brine can now be studied as a function of parameters such as water depth, heat input and permeability (Coumou et al. 2009). In spite of the advances in simulation technology, some fundamental aspects of MOR hydrothermal systems have so far not been addressed systematically, in particular the potential interplay of permeability and fluid properties in
controlling venting temperatures, heat output, and phase separation. For continental hydrothermal systems, a very strong influence of permeability on how hot the systems can get, on the extent and style of phase separation, and on the efficiency of heat transport has repeatedly been demonstrated (e.g., Hayba & Ingebritsen 1997; Driesner & Geiger 2007; and references therein). These studies consistently showed that continental hydrothermal systems develop upflow temperatures comparable to those of oceanic systems only if they have a system-scale permeability of approximately 1 · 10)15 m)2, a value that is two to four orders of magnitude lower than current estimates for black smoker MOR systems (Lowell & Germanovich 2004; Stroup et al. 2009) or measurements and estimates in offaxis upper oceanic crust (e.g., Becker & Fisher 2000, 2008; Fisher & Becker 2000; Stein & Fisher 2001; Hutnak et al. 2008). Since continental and MOR hydrothermal systems are driven by the same physical process (buoyancy-driven flow above a magmatic heat source), this discrepancy in system permeability appears very large and the question arises whether it is real and may reflect the influence of different geologic conditions (higher pressure at outflow, different temperatures of the heat source as well as different rock and fluid properties) on flow behavior or whether the current picture of permeability and heat transfer in MOR hydrothermal systems requires modification. This paper addresses this question by a series of 2D, state-of-the-art numerical simulations that simulate fully transient fluid flow including the full non-linearities in fluid properties (either pure water or saltwater). After introducing the model setup, results are presented that show a strong dependence of venting temperature and salinity on permeability. In the discussion, it is shown that these findings can quantitatively be explained by including the effect of permeability in the fluxibility hypothesis of Jupp & Schultz (2000, 2004). Predictive capabilities of this approach are demonstrated by a series of simulations with a different basal heat input. Finally, some general conclusions are drawn and an outlook on future developments is given.
MODEL SETUP AND METHODS Simulations were run for a model domain representing a generic across-axis section through a MOR hydrothermal system with a steady axial magma chamber at ca. 1 km depth. The basic setup follows the one used by Coumou et al. (2009): the simulation domain is a rectangular box of 3000 m width, 1000 m height and a nominal thickness of 1 m, discretized with triangular elements. At the bottom, a Gaussian-shaped heat flow profile is applied with the peak centered at 1500 m and a half-width of 500 m, mimicking the thermal effect of the underlying axial
134 T. DRIESNER magma chamber. The peak amplitude is adjusted such that over the 3000 m bottom boundary the desired integrated heat flux is reached. The value of 86 kW for the nominally 1m thick model is equivalent to the total magmatic heat input at a ridge with a steady spreading rate of ca. 10 cm year)1 (Sinha & Evans 2004). The existence of axial magma chambers at slightly more than 1 km depth that are nearly continuous over tens of km along ridges with comparable spreading rates (e.g., the East Pacific Rise, Dunn et al. 2000) is taken as justification for the steady state heat input in the current study. Cold (10C) fluid is allowed to recharge through the top boundary if flow vectors are into the domain, and hot, ascending fluids are allowed to leave the domain through the top by applying a boundary condition that sets dT ⁄ dz = 0 if flow vectors are directed outward (Jupp & Schultz 2000; Coumou et al. 2006). A constant pressure of 25 MPa is applied at the top boundary to simulate an oceanic water column of approximately 2500 m depth. At the left and right sides, no-flow boundary conditions are assigned. Following the philosophy of Coumou et al. (2009), no geometric complexity was added to the model in order to be able to study the first order physics of this simplest generic representation of a fast-spreading MOR system. The following equations for mass balance, heat balance and salt mass balance o½ðSl l þ S þ Sh h Þ ¼ r ð l l Þ r ð Þ ot
o½ð1 Þr hr þ ðSl l þ S þ Sh h Þ ot ¼ r ð l l hl Þ r ð h r ðK rT Þ
o½ðSl l Xl þ S X þ Sh h Xh Þ ot ¼ r ð l l Xl Þ r ð X Þ
ð1Þ
ð2Þ
ð3Þ
þ r ðDrl Xl Þ þ r ðDr X Þ were combined with a two-phase version of Darcy’s law: i ¼ k
kri ½rp i g i
i ¼ l;
ð4Þ
In the equations, / denotes porosity, S are saturations, q density, t time, v Darcy velocity, h specific enthalpy, K thermal conductivity, X mass fraction of salt, D salt diffusivity, k the intrinsic permeability of the rock, kri the relative permeability of phase i (described by a linear relative permeability model with residual saturations of 0 for vapor and 0.3 for liquid), l dynamic viscosity, p fluid pressure, and g is the vector of gravitational acceleration. Subscripts r, h, l, and v stand for rock, halite, liquid, and vapor, respectively.
The equations were solved on the model domain by using the CSMP++ simulation platform (Matthai et al. 2007) with the combined Finite Element-Finite Volume scheme of Geiger et al. (2006a,b) that was modified to describe heat transport on an enthalpy rather than heat capacity basis (Coumou 2008; Coumou et al. 2009). Fluid properties are taken from Haar et al. (1984) in the case of pure water and from Driesner (2007) and Driesner & Heinrich (2007) in the case of saltwater. Specific enthalpies of saltwater were corrected by a factor of 1 ⁄ (1 + XNaCl) missing in Driesner (2007). Fluid properties were stored in lookup tables to improve simulation speed. Resolution of lookup tables varies and is highest around the critical point of water where fluid properties change strongest with temperature, pressure, and composition. In the saltwater simulations, the initial fluid composition as well as that of recharging fluid was set to 3.2 wt% NaCl as a close proxy to seawater. Although permeability measurements in oceanic crust indicate a depth-dependence (Fisher 1998), the model domain is assigned a homogenous isotropic permeability, which is varied from simulation to simulation in order to study the permeability effect on the evolving hydrothermal system. Again, the foremost reason for this choice is to not obscure the first order physical effects in these generic simulations. In addition, previous studies of the effect of a layered permeability structure indicate that black smoker systems will not evolve to the observed hot temperatures when this layering (Lowell et al. 2007) is present. This may imply that these systems develop a very specific permeability structure due to mineral precipitation effects (Fontaine et al. 2007), the accurate modeling of which is, however, beyond current simulation capabilities. Therefore, the homogenous isotropic permeability was chosen because it is the simplest model for which temperatures as observed in nature may develop. As indicated by the governing equations given above, pure porous media flow according to appropriate versions of Darcy’s law is assumed together with thermal equilibrium between fluid and rock. In all simulations, a temperature cutoff was set at 1200C in order to avoid the bottom reaching temperatures higher than the magmatic temperatures of MOR magma chambers. This limit was reached in simulations with low permeability where convective fluid flux was too small to balance the bottom heat input. For those simulations, the de facto heat input was therefore lower than the nominally assigned heat flow and will be discussed in the Discussion section of this paper. In the few cases where temperature in a few cells at the bottom did indeed exceed 1000C (the upper limit of validity of the equations given in Driesner 2007 and Driesner & Heinrich 2007), the respective fluid properties and phase relations were computed by simple linear extrapolation.
The interplay of permeability and fluid properties 135
RESULTS The results are summarized in Fig. 1A,B. The venting temperatures and salinities show a strong dependence on the system’s permeability. A somewhat surprising finding is that the result for venting temperatures is largely independent of whether pure water or saltwater is used in the simulations (Fig. 1A), probably resulting from the similarities of fluid properties of pure water and the low-salinity hot fluids in the hydrothermal convection cells. Venting temperatures in excess of 300C are only observed if the permeability is lower than ca. 3 · 10)14 m)2. In both cases, the highest temperatures are observed as a broad peak for permeabilities between approximately10)15 and approximately10)14 m)2. At lower permeabilities, the temperature decreases again. For high permeabilities that have frequently been reported as the most likely ones related to high-temperature venting (10)11–10)13, e.g., Lowell &
(A)
(B)
Fig. 1. (A) Maximum venting temperature as a function of permeability for pure water and saltwater obtained in simulations with a total heat input of 86 kW into the bottom of the model. (B) Average venting salinities in the saltwater simulations.
Germanovich 2004), temperatures are clearly less than 200C (down to <100C). Salinities (Fig. 1B) are constant at seawater composition for permeabilities above ca. 3 · 10)14 m)2, indicating that no phase separation occurs under these conditions, in agreement with the low temperatures. At lower permeabilities, phase separation occurs in the deeper parts of the hydrothermal convection cell, leading to formation of brine as well as halite precipitation. The salinity of the hydrothermal plume is lowered since the low-salinity vapor phase from the phase-separating regions makes up an increasingly large fraction of the plume. The difference in the anatomy of the system as a function of permeability is pictured in Fig. 2 for three cases: (A) a high permeability (1 · 10)12 m)2) system with no phase separation, (B) a system at a permeability giving the highest vent temperatures (5 · 10)15 m)2), and (C) at a permeability lower that those with peak temperatures (5 · 10)17 m)2). Vectors for fluid fluxes are of identical scale in all three subfigures, demonstrating the massive influence of permeability. In case (A), the fluid flux is very large, meaning that the heat input into the bottom of the model is taken up by a large fluid mass per unit time and, consequently, the fluid experiences only moderate temperature increase with a venting temperature around 100C. In case (B), fluid flux is substantially smaller such that the fluid receives a much higher heat input per unit mass and reaches higher temperatures. These are sufficient to cause phase separation of the original liquid seawater into vapor+liquid (and locally vapor+halite) in the deeper parts of the system. As seawater has rather low salinity and splits upon phase separation into a lower-salinity vapor and a substantially more saline brine, mass balance in combination with density of the two phases requires that the vapor becomes the volumetrically dominant phase. This can be seen from the color coding where the blue end relates to 100% liquid and red to 100% vapor. The vapor+liquid mixture is very buoyant, forming a rapidly rising hydrothermal plume of low salinity. As already pointed out by Coumou et al. (2009), brine saturations in the two-phase plume are most of the time so low that brine remains stagnant in the upflow zone due to relative permeability effects until it reaches a sufficient saturation to become transiently mobile and vent for a short period of time. For very low permeabilities (C), fluid flow is almost stagnant, leading to an overall thermal structure close to a steady state conductive profile, slightly modified by two very weak upflow zones. If such conditions occurred in nature, these outflow regions would likely remain undiscovered as their discharge rates are extremely low. Fluid phase relations in the sub-seafloor reflect the stagnant nature of the system and are determined by the background variation in temperature and pressure that is almost linear
136 T. DRIESNER
(A) 1×10–12 m2, 5000 years
(B)
5×10–15 m2, 5000 years
(C) 5×10–17 m2, 5000 years
Fig. 2. Snapshots of three simulations with different permeability after reaching quasi-steady state. Background color coding indicates liquid saturation (= volume fraction of liquid in pore space; blue: 100% liquid, red: 100% vapor). Contour lines represent temperature in 50C intervals. Flow vectors (white: liquid, blue: vapor) represent fluid flux (Darcy velocity) and have the same scale in all three pictures. See text for further explanations.
with depth. The result is a stable, vapor-dominated bottom region and an upper liquid region. This pattern also explains why the venting temperatures in the pure water and saltwater simulations are somewhat different for the lowest permeabilities.
DISCUSSION The general permeability-dependence of the venting temperatures obtained from the simulations qualitatively matches what has been found in simulations of continental hydrothermal systems above cooling plutons (Hayba & Ingebritsen 1997; Driesner & Geiger 2007). There, the hottest systems occur at a permeability of 1 · 10)15 m)2. Below this value and with decreasing permeability, the decreasing fluid fluxes lead to a smaller and smaller convective contribution and conductive cooling dominates at permeabilities below approximately 1 · 10)16 m)2. Above approximately 1 · 10)14 m)2, fluid fluxes are so high that conductive heat supply from the source is less and less able to heat the large amounts of fluid that increase with increasing permeability. While this picture is qualitatively
intuitive, a more quantitative treatment seems desirable. This is rather complicated in the case of the cooling pluton models because feedbacks between the hydrothermal system and the changing geometries due to cooling of the pluton complicate the analysis. An additional complication is that in continental systems the top boundary is typically at a low pressure of 0.1 MPa, forcing most systems to approach the boiling temperature of approximately 100C at the surface. This is different – and simpler – for MOR systems where the relatively high pressure at the sea-floor allows upwelling fluids to essentially keep their temperature. For the present data set the overall geometry as well as basal heat input are constant, which greatly simplifies the analysis although convection cells are allowed to develop freely. A number of semi-analytical analyses for simplified or ‘single pass’ convection models have been published in the context of MOR systems (see Lowell & Germanovich 2004; and references therein). Most of these did a rather extensive treatment that included feedback on the geometry and extent of the ‘reaction zone’, i.e., the permeable hot zone above the magma chamber where the circulating
The interplay of permeability and fluid properties 137 hydrothermal fluid is heated from essentially ambient temperature to the upwelling temperature, TU. Since effects of heat loss upon upwelling and adiabatic expansion can be shown to be of minor effect, TU can be assumed to be identical to the venting temperature. Under these assumptions, Jupp & Schultz (2004) derived the following expression for the energy balance between heat input from the magmatic source and advective transport by hot hydrothermal fluid upwelling from the ‘reaction zone’: LðTD TU Þ HR ðhU ðTU Þ h0 Þð0 U ðTU ÞÞ 2gkD U LD U ðTU Þ
2
ð5Þ
On the left-hand side, L is the horizontal half-width of the ‘reaction zone’ above the magma chamber, k the crust’s thermal conductivity, TD the ‘driving temperature’ (= essentially the magma near 1200C temperature in real systems), TU the upflow temperature in the hydrothermal plume and HR the height of the conductive boundary layer. On the right-hand side, g is the gravitational constant, kD the permeability in the upflow zone, qU the density of the upflowing fluid, q0 the density of cold (sea)water, hU the specific enthalpy of upflowing fluid, h0 the specific enthalpy of cold water, lU the dynamic viscosity of the upflowing fluid, and LD the half-width of the hydrothermal plume. The term in square brackets is a fluid property termed ‘fluxibility’ that has been shown to control the onset of convection as well as the venting temperature at steady state (Jupp & Schultz 2000, 2004; Coumou et al. 2008). The left-hand side describes the conductive heat input that is heating fluid in the ‘reaction zone’. In the case of the present simulations – when assuming that all heat entering the bottom of the model is picked up by the fluid and advected away – this term can be replaced by the bottom heat flux Q. How much can be advected in total is described by the right-hand side expression, which is a combination of the driving force (essentially the density difference between the hot fluid in the plume versus the cold fluid in the recharge zone), flow resistance (permeability divided by viscosity), and the heat picked up by the fluid, times the plume half width LD. Hence, Eqn (5) can also be written as Q 2gkD
U ðhU ðTU Þ h0 Þð0 U ðTU ÞÞ LD lU ðTU Þ
ð6Þ
Notice that this is for a 2D geometry of unit thickness (1 m) and, therefore, units are W m)1, essentially meaning heat flux per meter ridge length. Assuming that the fluid pressure at mid-height of the modeled hydrothermal system (i.e., 30 MPa) is a good
proxy to describe average fluid properties in the system, Eqn (6) can be used to construct a diagram of the righthand side term as a function of TU for a varying kD. This is shown in Fig. 3. The vertical axis is given on a logarithmic scale in order to be able to display the effect of kD that can vary over orders of magnitude. Each curve represents Eqn (6) for a given kD. The overall curve shape is determined by the fluxibility term and shows a peak just above 400C. A horizontal line shows the heat input of 86 kW m)1 used in the simulations. If the values at which the curves are intersected by this line are plotted onto the same diagram as in Fig. 1, an excellent fit to the simulated data is observed (Fig. 4) if an LD of 50 m is used in Eqn (6). This LD is well in the range of values derived or assumed in previous studies (Jupp & Schultz 2004; Lowell & Germanovich 2004) and close to those observed in the simulations (Fig. 2A,B), demonstrating that the simplified geometry of ‘single pass’ models is in principle sufficient to capture the essence of the freely developing convection cells of the transient simulations. The excellent agreement between Eqn (6) and the simulation results extends from the highest permeabilities down to approximately 10)14 m)2. This is very close to the value of 7 · 10)15 m)2, for which the curve peak in Fig. 3 (dashed line) has the same value as the basal heat input. In other words, for the 86 kW heat input, 7 · 10)15 m)2 is the lowest possible permeability at which the fluid flux is still sufficiently high to allow advective transport to balance this heat input. At lower permeabilities, the fluid flux is
Fig. 3. Solid lines: heat transported advectively by a hydrothermal fluid of temperature TU for a given permeability kD (curve labels). Curves were computed from Eqn (6) using a value of LD = 50 m and for fluid properties at 30 MPa. The unit W m)1 indicates that the data are valid for a twodimensional system of nominal thickness of 1 m, i.e., for conversion into heat input per km ridge length, values on the vertical axis have to be multiplied by 1000. Dotted lines are the equivalent curves for fluid parameters used in published single pass models (Lowell & Germanovich 2004). Dashdotted curve at 86 kW m)1 is the heat input in the simulations shown in Figs 1 and 2. The dashed curve indicates the lowermost permeability for which fluid flux is still high enough to advectively transport this heat. 7.2 kW m)1 is the maximum possible conductive heat transport in the 3000 m wide system assuming a bottom temperature of 1200C and a thermal conductivity of 2 W m)1C)1.
138 T. DRIESNER
Fig. 4. Data from Fig. 1 (symbols) compared to predictions from Eqn (6) for 86 kW heat input into the model domain (thick solid line). Thick dashed curve is computed for 12.5 kW heat input and triangles are simulation results for the same value. Dash-dotted curve is computed for 300 kW and squares are pure water simulations for this value. Thin lines represent the respective curves computed with linearized fluid properties as used by Lowell & Germanovich 2004.
too small and only less heat can be transport by advection. This also became apparent in the simulations as the bottom temperatures reached the cut-off temperature of 1200C. In natural systems where this artificial cut-off does not apply there would be a non-steady state balance between magmatic and hydrothermal heat transfer and the magma chamber will melt rocks at its roof and migrate upwards or the excess magmatic heat is extracted by magma eruptions to the sea floor. In the simulations, due to the temperature cut-off, this rather means that the flat peak in Figs 1A and 4 is reflecting a lower and lower de facto heat input. This is equivalent to the solid curve being shifted to the left, i.e., the system is operating at the highest possible temperature-permeability coordinate for the de facto heat input. The temperature decrease in Figs 1A and 4 at the lowest permeabilities happens at values below approximately 5 · 10)16 m)2 with minor variation depending on whether pure water or saltwater is used in the simulations. When comparing these values with the curves in Fig. 3, it becomes obvious that the temperature decrease starts when conductive heat transport becomes a significant component of heat transfer in the model domain. Using a thermal conductivity of rock, k, of 2 WC)1 m)1as in the simulations, the maximum possible conductive heat transport for the geometry of the model domain is 2 WC)1 m)1 (1200C – 10C) ⁄ 1000 m · 3000 m = 7.2 kW m)1. A curve in Fig. 3 with its peak at this value would be for a permeability just slightly higher than approximately 5 · 10)16 m)2, i.e., exactly the value below which the venting temperatures decrease. Given the fact that this interpretation seems to quantitatively predict the permeability-dependence observed in the
simulations, curves for lower and higher heat fluxes were computed from Eqn (6) to test its predictive capabilities. The lower heat flux value was taken as 12.5 kW m)1 (dashed line in Fig. 4), which is equivalent to a total magmatic heat input for a spreading rate of 2 cm year)1 (Sinha & Evans 2004). Then, a series of saltwater simulations was run for this heat input. The simulated vent temperatures (triangles in Fig. 4) match very well the predicted curve. The peak is occurs at lower permeabilities and is much narrower than for the 86 kW simulation series. The narrow peak top is because the overall heat input is already close to the maximum possible conductive heat transport. As a higher heat flux 300 kW m)1 were chosen, well above the maximum steady magmatic heat input for ridges spreading as fast as 15 cm year)1 (124 kW m)1, Sinha & Evans 2004) but in the range of estimates for some vent fields (Baker 2007; and references, therein) and recent 3D simulations that tried to match those values (Coumou et al. 2008). The simulations were done with pure water only because numerical problems occurred with the saltwater scheme under these extreme conditions. Again, the agreement between simulation and prediction from Eqn (6) is excellent. For comparison, the respective curves were computed with the simplified fluid properties used in published models of hydrothermal convection at MOR (properties taken from Lowell & Germanovich 2004; thin lines). The prediction of the permeability-temperature relation is moderate at best and this model fails to predict the venting temperature limit at around 400C, as was also apparent from the curve shapes in Fig. 3. The difference between models with simplified and full fluid properties demonstrates the crucial importance of including the latter while the simplified geometry – which was assumed the same in both cases – seems to play only an insignificant role. Although the results presented above appear to provide some novel insight into hydrothermal heat transfer at MOR, some limitations of this approach need to be mentioned. One limitation is that the 2D results may not simply be transferable to the real 3D world of MOR systems. Recent findings (Coumou et al. 2008) from 3D simulations suggest that individual hydrothermal plumes may be viewed as roughly radially symmetric systems. Extending the above analysis to such geometry should be relatively easy, however, 3D simulations which would serve as a test for the model are still computationally very expensive. Nevertheless, recent advances in computational techniques will make it possible soon to run enough simulations to allow such tests. More fundamentally, the generic scenario used here is based on an idealized, undisturbed simple geometry with steady heat input from below. Although this may be justified as a good proxy for ridges with well-defined axial magma chambers, geometries of real systems may develop
The interplay of permeability and fluid properties 139 dynamically and discontinuous events such as volcanic eruptions or dike intrusions are likely to occur frequently. On the other hand, the current results seem to indicate that true black smoker temperatures may in fact only arise in a non-steady state situation where magmatic heat input is larger than what can be accommodated by the hydrothermal system. Hence, in such scenarios, transfer of magmatic heat via eruptions to the sea floor may be a necessary consequence. In the case where the magmatic heat input is balanced by the hydrothermal system, dikes may be cooled efficiently and eruptions may be rare. This picture is in first order agreement with previously published concepts (e.g., Wilcock & Delaney 1996) and the simulations presented here may help to test these and develop more rigorous and quantitative models. In real systems, thermal disequilibrium between fluid in fractures and the surrounding rock matrix may exist during the early heating stages of the hydrothermal evolution. This will not be captured in the current simulation approach where thermal equilibrium between fluid and porous rock is assumed and, therefore, time scales until a hot plume reaches the surfaces are likely to be too long in the simulations. Fully addressing these aspects of flow physics in a fractured porous medium, however, is beyond current knowledge of the nature of the oceanic crust and has also not yet been implemented into simulation codes. Once these techniques will become available, we can probably expect new insights into rates and mechanisms operating in these systems. Nevertheless, the fundamental physics described in this contribution is likely to remain a reference frame for the interpretation of many systems.
CONCLUSIONS Probably the most obvious conclusion from this work is that there is a strong permeability dependence of hydrothermal vent temperatures at mid-ocean ridge systems, the systematic behavior of which can quantitatively be described by the fluxibility hypothesis of (Jupp & Schultz 2000, 2004) This demonstrates once more that the nonlinearities in the temperature – pressure – composition dependence of fluid properties are a first order control of the behavior of hydrothermal systems. Some additional and rather fundamental conclusions can be drawn from the curves in Figs 3 and 4: (1) The highest vent temperatures always occur at – or below – the lowest permeability that still allows sufficient fluid flux to advect away all basal heat input from the magma chamber. In other words: the occurrence of high vent temperatures may be indicative for a non-steady state situation, in which the hydrothermal heat transfer is unable to balance the magmatic heat input. For heat fluxes in the order of tens of MW per km ridge length, these permeabilities are
1 · 10)14 m)2 or below, in good agreement with earlier measurements in oceanic crust (Fisher 1998) but lower than some more recent theoretical estimates and – mostly off-axis – field measurements (Becker & Fisher 2000, 2008; Fisher & Becker 2000; Stein & Fisher 2001; Lowell & Germanovich 2004; Hutnak et al. 2006, 2008; Stroup et al. 2009). As is evident from Fig. 3, even very high heat fluxes of up to 1 GW per km ridge length require permeability to be not significantly higher than 1 · 10)13 m)2 if discharge is in the form of high-temperature vents. Hence, permeability in high-temperature MOR systems may be lower than previously estimated from models with linearized fluid properties. (2) The higher the heat input at the base of the system, the higher the permeability at which the maximum vent temperatures occur. This at least partly explains the discrepancies between previous permeability estimates from various modeling approaches (e.g., Coumou et al. 2008 with real water properties versus Lowell & Germanovich 2004 with linearized water properties). For very high heat fluxes, permeabilities may be in the lower end of those predicted from models with linearized water properties. (3) Systems with higher permeability will be cooler but potentially more efficient in removing the heat. Curves in Fig. 3 show that systems at higher permeability always have a higher capacity for advective heat transport (simply, because the fluid flux is higher). While the highest temperature systems may be the most spectacular ones, this could potentially imply that lower temperature ones could in fact be more relevant for the heat budget at MOR. Several studies have provided evidence that at least parts of the oceanic crust may be quite permeable (k > 10)13 m)2) and high-permeability, low-temperature, off-axis systems have been shown to be very efficient in transporting heat (Hutnak et al. 2008). If this also applies to axial crust with active magma input, hydrothermal systems in such environments may be relatively cool (below 200C or even below 100C) and devoid of black smoker type hydrothermal expressions. However, if such systems exist in nature, current search schemes based on light scattering or geochemical tracers in the water column typically applied to find high-temperature systems may not applicable and plumes of colder hydrothermal fluids may not rise in the water column as high as those from high temperature vents.
ACKNOWLEDGEMENTS Sincere thanks go to the CSMP++ community for continuously strengthening and extending the capabilities of this
140 T. DRIESNER simulation tool. In particular, Dim Coumou is thanked for preparing grounds for the simulations presented here. Two anonymous review helped to sharpen the thoughts presented in this paper.
REFERENCES Baker ET (2007) Hydrothermal cooling of midocean ridge axes: do measured and modeled heat fluxes agree? Earth and Planetary Science Letters, 263, 140–50. Baker ET, German CR (2004) On the global distribution of hydrothermal vent fields. In: Mid-Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans (eds German CR, Lin J, Parson LM), pp. 245–66. American Geophysical Union, Washington. Becker K, Fisher AT (2000) Permeability of upper oceanic basement on the eastern flank of the Juan de Fuca Ridge determined with drill-string packer experiments. Journal of Geophysical Research-Solid Earth, 105, 897–912. Becker K, Fisher AT (2008) Borehole packer tests at multiple depths resolve distinct hydrologic intervals in 3.5-Ma upper oceanic crust on the eastern flank of Juan de Fuca Ridge. Journal of Geophysical Research-Solid Earth, 113, 1–12, B07105. Cherkaoui ASM, Wilcock WSD (1999) Characteristics of high Rayleigh number two-dimensional convection in an open-top porous layer heated from below. Journal of Fluid Mechanics, 394, 241–60. Coumou D (2008) Numerical modeling of mid-ocean ridge hydrothermal systems. PhD Thesis, Department of Earth Sciences Zurich, ETH Zurich, 163 pp. Coumou D, Driesner T, Geiger S, Heinrich CA, Matthai S (2006) The dynamics of mid-ocean ridge hydrothermal systems: splitting plumes and fluctuating vent temperatures. Earth and Planetary Science Letters, 245, 218–31. Coumou D, Driesner T, Heinrich CA (2008) The structure and dynamics of mid-ocean ridge hydrothermal systems. Science, 321, 1825–8. Coumou D, Driesner T, Weis P, Heinrich CA (2009) Phase separation, brine formation, and salinity variation at Black Smoker hydrothermal systems. Journal of Geophysical Research-Solid Earth, 114, 1–16, B03212. Driesner T (2007) The system H2O-NaCl. Part II: correlations for molar volume, enthalpy, and isobaric heat capacity from 0 to 1000 degrees C, 1 to 5000 bar, and 0 to 1 X-NaCl. Geochimica et Cosmochimica Acta, 71, 4902–19. Driesner T, Geiger S (2007) Numerical simulation of multiphase fluid flow in hydrothermal systems. In: Fluid-Fluid Interactions. Reviews in Mineralogy and Geochmistry (eds Liebscher A, Heinrich CA), pp. 187–212. Mineralogical Society of America, Chantilly. Driesner T, Heinrich CA (2007) The system H2O-NaCl. Part I: correlation formulae for phase relations in temperature-pressurecomposition space from 0 to 1000 degrees C, 0 to 5000 bar, and 0 to 1 X-NaCl. Geochimica et Cosmochimica Acta, 71, 4880–901. Dunn RA, Toomey DR, Solomon SC (2000) Three-dimensional seismic structure and physical properties of the crust and shallow mantle beneath the East Pacific Rise at 930’N. Journal of Geophysical Research-Solid Earth, 105, 23537–55. Fisher AT (1998) Permeability within basaltic oceanic crust. Reviews of Geophysics, 36, 143–82. Fisher AT (2005) Marine hydrogeology: recent accomplishments and future opportunities. Hydrogeology Journal, 13, 69–97.
Fisher AT, Becker K (2000) Channelized fluid flow in oceanic crust reconciles heat-flow and permeability data. Nature, 403, 71–4. Fontaine FJ, Wilcock WSD (2006) Dynamics and storage of brine in mid-ocean ridge hydrothermal systems. Journal of Geophysical Research-Solid Earth, 111, 1–16, B06102. Fontaine FJ, Wilcock WSD (2007) Two-dimensional numerical models of open-top hydrothermal convection at high Rayleigh and Nusselt numbers: implications for mid-ocean ridge hydrothermal circulation. Geochemistry Geophysics Geosystems, 8, 1–17. Fontaine FJ, Wilcock WSD, Butterfield DA (2007) Physical controls on the salinity of mid-ocean ridge hydrothermal vent fluids. Earth and Planetary Science Letters, 257, 132–45. Fontaine FJ, Cannat M, Escartin J (2008) Hydrothermal circulation at slow-spreading mid-ocean ridges: the role of along-axis variations in axial lithospheric thickness. Geology, 36, 759–62. Geiger S, Driesner T, Heinrich CA, Matthai SK (2006a) Multiphase thermohaline convection in the earth’s crust: I. A new finite element–finite volume solution technique combined with a new equation of state for NaCl-H2O. Transport in Porous Media, 63, 399–434. Geiger S, Driesner T, Heinrich CA, Matthai SL (2006b) Multiphase thermohaline convection in the earth’s crust: II. Benchmarking and application of a finite element–finite volume solution technique with a NaCl-H2O equation of state. Transport in Porous Media, 63, 435–61. German CR, Lin J (2004) The thermal structure of the oceanic crust, ridge-spreading and hydrothermal circulation: how well do we understand their inter-connections? In: Mid-Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans (eds German CR, Lin J, Parson LM), pp. 1–18. American Geophysical Union, Washington. Haar L, Gallagher JS, Kell GS (1984) NBS/NRC Steam Tables: Thermodynamic and Transport Properties and Computer Program for Vapor and Liquid States of Water in SI Units, p. 306. Hemisphere Publishing, Bristol. Hayba DO, Ingebritsen SE (1997) Multiphase groundwater flow near cooling plutons. Journal of Geophysical Research-Solid Earth, 102, 12235–52. Hutnak M, Fisher AT, Zuhlsdorff L, Spiess V, Stauffer PH, Gable CW (2006) Hydrothermal recharge and discharge guided by basement outcrops on 0.7–3.6 Ma seafloor east of the Juan de Fuca Ridge: observations and numerical models. Geochemistry Geophysics Geosystems, 7, 1–36. Hutnak M, Fisher AT, Harris R, Stein C, Wang K, Spinelli G, Schindler M, Villinger H, Silver E (2008) Large heat and fluid fluxes driven through mid-plate outcrops on ocean crust. Nature Geoscience, 1, 611–4. Jupp T, Schultz A (2000) A thermodynamic explanation for black smoker temperatures. Nature, 403, 880–3. Jupp TE, Schultz A (2004) Physical balances in subseafloor hydrothermal convection cells. Journal of Geophysical Research-Solid Earth, 109, 1–12, B05101. Lewis KC, Lowell RP (2009) Numerical modeling of two-phase flow in the NaCl-H2O system: introduction of a numerical method and benchmarking. Journal of Geophysical Research-Solid Earth, 114, 1–18, B05202. Lowell RP, Germanovich LN (2004) Hydrothermal processes at mid-ocean ridges: results from scale analysis and single-pass models. In: Mid-Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans (eds German CR, Lin J, Parson LM), pp. 219–44. American Geophysical Union, Washington.
The interplay of permeability and fluid properties 141 Lowell RP, Gosnell S, Yang Y (2007) Numerical simulations of single-pass hydrothermal convection at mid-ocean ridges: effects of the extrusive layer and temperature-dependent permeability. Geochemistry Geophysics Geosystems, 8, 1–16. Matthai SK, Geiger S, Roberts SG, Paluszny A, Belayneh M, Burri A, Mezentsev A, Lu H, Coumou D, Driesner T, Heinrich CA (2007) Numerical simulation of multi-phase fluid flow in structurally complex reservoirs. In: Structurally Complex Reserviors (eds Jolley SJ, Barr D, Walsh JJ, Knipe RJ), pp. 405–29. The Geological Society, London. Rabinowicz M, Boulegue J, Genthon P (1998) Two- and threedimensional modeling of hydrothermal convection in the sedimented Middle Valley segment, Juan de Fuca Ridge. Journal of Geophysical Research-Solid Earth, 103, 24045–65. Rabinowicz M, Sempere JC, Genthon P (1999) Thermal convection in a vertical permeable slot: Implications for hydrothermal circulation along mid-ocean ridges. Journal of Geophysical Research-Solid Earth, 104, 29275–92. Ramondenc P, Leonid NGA, Von Damm KL, Lowell RP (2006) The first measurements of hydrothermal heat output at 950’N, East Pacific Rise. Earth and Planetary Science Letters, 245, 487– 97. Sinha MC, Evans RL (2004) Geophysical constraints upon the thermal regime of the ocean crust. In: Mid-Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans
(eds German CR, Lin J, Parson LM), pp. 19–62. American Geophysical Union, Washington. Stein JS, Fisher AT (2001) Multiple scales of hydrothermal circulation in Middle Valley, northern Juan de Fuca Ridge: physical constraints and geologic models. Journal of Geophysical Research-Solid Earth, 106, 8563–80. Stein JS, Fisher AT (2003) Observations and models of lateral hydrothermal circulation on a young ridge flank: numerical evaluation of thermal and chemical constraints. Geochemistry Geophysics Geosystems, 4, 1–20. Stein CA, Stein S (1994) Constraints on hydrothermal heat-flux through the oceanic lithosphere from global heat flow. Journal of Geophysical Research-Solid Earth, 99, 3081–95. Stroup DF, Tolstoy M, Crone TJ, Malinverno A, Bohnenstiel DR, Waldhauser F (2009) Systematic along-axis tidal triggering of microearthquakes observed at 950’N East Pacific Rise. Geophysical Research Letters, 36, 1–5, L18302. Wilcock WSD (1998) Cellular convection models of mid-ocean ridge hydrothermal circulation and the temperatures of black smoker fluids. Journal of Geophysical Research-Solid Earth, 103, 2585–96. Wilcock WSD, Delaney JR (1996) Mid-ocean ridge sulfide deposits: evidence for heat extraction from magma chambers or cracking fronts? Earth and Planetary Science Letters, 145, 49–64.
Using seafloor heat flow as a tracer to map subseafloor fluid flow in the ocean crust A. T. FISHER1 AND R. N. HARRIS2 1
Earth and Planetary Sciences Department and Institute for Geophysics and Planetary Physics, University of California, Santa Cruz, Santa Cruz, CA, USA; 2College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR, USA
ABSTRACT We describe how seafloor heat flow is determined, review current understanding of advective heat loss from oceanic lithosphere, and present results from three field areas to illustrate how heat flow measurements are used (along with complementary data) to resolve fluid flow rates and patterns. Conductive heat flow through much of the seafloor is lower than predicted by lithospheric cooling models as a result of hydrothermal circulation; this discrepancy is the basis for global estimates of the magnitude of advective cooling of oceanic lithosphere. Hydrothermal circulation also redistributes heat within the ocean crustal aquifer, leading to local variability. Heat flow studies in Middle Valley, a sedimented spreading center in the northeastern Pacific Ocean, indicate multiple scales of fluid circulation, delineate conditions at the top of a hydrothermal reservoir, and show the influence of primary and secondary convection. Heat flow studies on the eastern flank of the Juan de Fuca Ridge document the thermal influence of isolated basement outcrops surrounded by thick, low-permeability sediments. Warm hydrothermal fluids seep from the crust through a small volcanic edifice, having flowed into the crust through a larger outcrop 50 km to the south. These fluids generate a local geothermal anomaly, but have little influence on regional heat loss from the plate. In contrast, heat flow surveys on part of the Cocos Plate, eastern Equatorial Pacific Ocean, indicate that regional conductive heat loss is just 10–40% of predictions from lithospheric cooling models. Basement outcrops in this area focus massive discharge of cool, hydrothermal fluid and associated heat (4–80 · 103 l s-1 of fluid, 0.8–1.4 GW of heat). Seafloor heat flow studies will be increasingly important in coming years for understanding marine hydrogeologic regimes and the role of fluids in a variety of Earth processes and settings. Key words: seafloor heat flow, subseafloor fluid flow, hydrothermal systems, modeling Received 29 July 2009; accepted 7 November 2009 Corresponding author: A. T. Fisher, Earth and Planetary Sciences Department and Institute for Geophysics and Planetary Physics, University of California, Santa Cruz, Santa Cruz, CA 95064, USA. Email:
[email protected]. Tel.: +831 459 5598. Fax: +831 459 3074. Geofluids (2010) 10, 142–160
INTRODUCTION Soon after collection of the first oceanographic heat flow data (Bullard 1954; Revelle & Maxwell 1952), it was recognized that seafloor heat flow was generally highest and most variable near mid-ocean ridges, areas of elevated basement rocks that lacked thick and continuous sediment cover. At the time when initial heat flow data were collected, theories of seafloor spreading and plate tectonics were not fully developed or widely accepted, and there was considerable debate as to how heat flow measurements should be interpreted (Elder 1965; Langseth et al. 1966;
Sclater 2004). Once convincing evidence for seafloor spreading became apparent, heat flow data presented a conundrum. It was expected that the oceanic lithosphere should release the most heat near spreading centers where the seafloor is youngest (McKenzie 1967; Lister 1972; Parker & Oldenberg 1973), but the variable heat flow seen at young sites commonly included low heat flow values. The difference between theory and observation was resolved by the realization that seafloor heat flow measurements, which document the conductive portion of lithospheric heat loss, generally do not measure the heat that is removed advectively by hydrothermal circulation. In addition,
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Using seafloor heat flow to map subseafloor fluid flow 143 hydrothermal circulation helps to explain the variability commonly observed in regional heat flow surveys, which can result in local redistribution of heat within the crust (whether or not there is net advective heat loss). Theoretical seafloor cooling curves were developed initially in the late 1960s, and have been refined and extended in decades hence, using bathymetric and heat flow data to constrain the physical properties and evolution of oceanic lithosphere as plates age (Davis & Lister 1974; Parsons & Sclater 1977; Stein & Stein 1992). These cooling curves are important for quantifying the magnitude of the conductive heat flow deficit associated with advective heat loss, the extent of local heat redistribution within the lithosphere, the suppression of seafloor heat flow by rapid sedimentation, and other geologic and hydrologic processes (e.g., Anderson & Hobart 1976; Sclater et al. 1980; Davis et al. 1989; Wang & Davis 1992). The last 10–15 years has seen renewed interest in using seafloor heat flow to resolve marine hydrogeologic processes, contemporaneous with increasing application of heat as a tracer in surface water and ground water studies on land (e.g., Anderson 2005; Burow et al. 2005; Constantz et al. 2001). Heat offers many advantages as a tracer of fluid flow. Variations in fluid temperature and heat flow occur naturally within many hydrogeologic systems and at a range of spatial and temporal time scales. Water has a high heat capacity, so relatively small rates of fluid flow can result in significant thermal perturbations. The long-term consistency of bottom water temperatures in the deep ocean imposes a stable boundary condition at the seafloor in many areas, simplifying the application of coupled fluid– thermal flow models. Recent technical advances have made collection of seafloor thermal data easier, cheaper, and more reliable, accurate and precise. Finally, improvements and standardization in ship and instrument navigation allow co-location of complementary mapping, seismic, and chemical data, providing a broader geological context for interpretation of thermal data. In this paper, we review the basis for marine heat flow data collection and processing, and discuss how these data are used to quantify rates and patterns of fluid flow. We discuss thermal data and interpretations based on global data summaries and regional studies of three geologic settings to illustrate key concepts. There are many hydrogeologic settings where researchers have used heat to quantify fluid flow using water column, seafloor, and subseafloor tools, including studies based on transient variations in shallow heat flow (e.g., Becker et al. 1983, 1995; Kinoshita et al. 1996; Veirs et al. 1999; Goto et al. 2005; Hamamoto et al. 2005). Space limitations preclude a comprehensive presentation of these topics; instead, we highlight studies of rapid lateral fluid flow in permeable
volcanic rocks of the upper oceanic crust using shallow heat flow measurement surveys.
METHODS Theoretical basis Seafloor heat flow measurements are based on application of a simplified form of Fourier’s first law for vertical heat transport: q ¼
dT dz
ð1Þ
where q = heat flow (W m-2), k = thermal conductivity (W m-1 C)1), and dT ⁄ dz = thermal gradient (C m-1). Thus determining the conductive seafloor heat flow requires measurements of both the thermal conductivity and the thermal gradient. The negative sign on the righthand side of Eq. (1) indicates that heat flows in a direction opposite to the thermal gradient. In fact, Fourier’s first law is a special case of a more general, one-dimensional transient, conduction equation: o oT oT ¼ oz oz ot
ð2Þ
where j is the thermal diffusivity, the ratio of thermal conductivity to heat capacity. It is often assumed that heat flow through the top of a thermal boundary layer (such as most marine sediments) occurs at steady state, so that the right-hand side of Eq. (2) is set to zero. However, the thermal diffusivity can change with depth, leading to variations in the thermal gradient, even when heat flow is purely conductive (as discussed later). If there is also heat advection by fluid, the one-dimensional transient equation becomes: o oT nvf oT oT ¼
oz oz oz ot
ð3Þ
where n = layer porosity, vf = fluid velocity (nvf is volume flux ⁄ area = specific discharge), and c = the ratio of fluid and saturated formation heat capacities. As a practical matter, it is difficult to resolve the fluid velocity through seafloor materials independent of Eq. (3), so a form of this equation is sometimes used to estimate the fluid flow rate from thermal observations. Equation (3) is one-dimensional, vertical, and based on a constant fluid flow rate. Fluid and heat transport in many seafloor hydrogeologic systems are transient, occur in two or three dimensions, and are fully coupled. Modeling of systems having these complexities may require more sophisticated analytical or numerical approaches, especially if there are spatial or temporal changes in fluid and formation
144 A. T. FISHER & R. N. HARRIS properties, reactive transport, or other non-linear links between fluid and heat flow. Data collection Seafloor heat flow measurements are made in marine sediments with two primary kinds of tools: (i) instruments that penetrate the seafloor and determine temperatures within
Inse
shallow sediments (generally <10 m) and (ii) instruments deployed within sediments while drilling to greater depths (30–400 m). Examples of these tools and their use are summarized briefly in this section. Multipenetration probes comprise a weight stand and data logger, a lance that holds thermal sensors, and a telemetry system that sends data from the seafloor to the surface (Fig. 1A) (Langseth et al. 1965; Davis 1988). Most
rt
rt
Inse
1.1-m lance Data logger
Thermistor sensor
3.5-m lance Weight stand
(C)
Sensor tube housing thermistors and heater wire Acoustic telemetry
(A)
APCT-3 shoe Serial connector Electronics storage case
Thermistor Insert
Board with A/D converter, microprossor, internal clock and memory Plastic isolater
Battery
Pressure case
Thermistor
Batteries Deck box APCT-3 tool frame
Thermistor
ge
og
l ata
(D)
r
D
rt
se
In
(B)
re Co
er lin
Needle probes
(E)
Fig. 1. Examples of seafloor heat flow instrumentation used in marine sediments above a basement aquifer. Arrows labeled ‘‘Insert’’ in panels A–D indicate orientation of instruments when pushed into the seafloor. See text for description and citations for more information on tool design, use, and limitations. (A) Violin-bow, multipenetration heat flow probe used to determine shallow gradients and in-situ thermal conductivity. (B) Autonomous outrigger probes showing construction of logger (top, sketch) and two styles of probes as deployed (bottom, photograph). Probe with fin and long-style thermistor tube (left), and probe with short sensor (right) mounted on a piston core barrel. In both cases the data logger and battery are housed inside a protective pressure case that also functions as a serial connector for programming and data download. (C) Davis–Villinger Temperature Probe used for measurement of sediment temperature at the base of a borehole. Tool (silver) is mounted inside a core barrel (dark gray) for deployment, then pushed into the sediment using the drillstring. (D) Third generation of advanced piston coring temperature tool (APCT-3) used to measure sediment temperature while piston coring with a drillship. Tool frame with thermistor probe (center bottom) is placed inside annular cavity in piston coring shoe (upper right) prior to deployment. (E) Needle probes inserted through core liner to determine the thermal conductivity of recovered sediments.
Using seafloor heat flow to map subseafloor fluid flow 145 modern multipenetration systems have thermistors mounted on outrigger probes attached to a central strength member, or within a single, long ‘violin bow’ outrigger that positions the thermal sensors away from the central strength member (Hyndman et al. 1979). The precision of individual thermistor sensors is typically ±0.001C, and probes commonly have 11 or more thermistors distributed along a 2–5 m lance. Shorter versions of these tools, often with a narrow strength member that houses the thermal sensors, are used with submersibles and remotely operated vehicles. A multipenetration heat flow station begins with the lowering of the instrument towards the seafloor from a ship using a trawl wire and winch. The probe is driven by gravity into sediments and allowed to achieve partial equilibration with ambient formation temperatures. If the instrument has the capability to measure in situ thermal conductivity, a calibrated heat pulse is generated using a wire that extends along the length of the sensor tube. The temperature decay of the heat pulse provides a measure of sediment thermal conductivity at multiple depths. With the completion of a measurement, the instrument is pulled from the seafloor and hoisted 100–500 m (depending on intended measurement spacing), the ship is maneuvered to a new position, and the process is repeated. In this manner a transect of heat flow measurements can be obtained relatively quickly. Multipenetration surveys are typically run along carefully navigated transects, often with collocated swath map and seismic data, during a deployment of 12–48 h. With some systems there is complete data telemetry, and processing can be completed in near real time. But many systems send back just enough information to evaluate tool performance and battery power, and to make a rough estimate of thermal gradients during a survey. More complete and accurate data analysis is completed after the tool is returned to the surface and data are downloaded from logger memory. Some of the earliest oceanographic heat flow measurements were made with temperature probes attached to sediment core barrels, so that a thermal gradient could be determined at the same place and time that sediment was sampled. The technique is similar to that involving multipenetration systems, except that only one gradient is determined during each instrument lowering. Modern outrigger probes used on core barrels are robust, accurate (±0.002– 0.003C), and autonomous, recording and storing data without telemetry (Fig. 1B) (Pfender & Villinger 2002). These systems are ideal for making reconnaissance or supplemental measurements in conjunction with coring operations, which is particularly useful where sediment cover is thin or patchy. Outrigger probe systems on core barrels generally do not provide determinations of sediment thermal conductivity, so the latter measurements must be
made in the laboratory using material recovered during coring (as described below). There is also a system for measuring heat flow through bare rock (Johnson & Hutnak 1996), but this tool is not widely used at present, and the vast majority of seafloor heat flow surveys are conducted on sedimented seafloor. There are two main classes of sediment temperature probes used during scientific ocean drilling (Fisher & Becker 1993): push in tools that require a dedicated tool run, independent of coring operations (Fig. 1C) (Uyeda & Horai 1980; Davis et al. 1997a); and tools that are integrated with piston coring operations (Fig. 1D) (Horai & Von 1985; Fisher et al. 2007). These instruments are generally deployed 1–10 m beyond the drilled depth of a borehole within a few hours of drilling, avoiding the thermal (cooling) influence of pumping cold seawater as a drilling fluid. Additional subseafloor measurements have been made during logging operations in open boreholes, either using sediment instruments on a wireline or using open hole logging instruments made specifically for this purpose (Hyndman et al. 1976; Becker et al. 1983; Fisher et al. 1997; Larson et al. 1993). Strings of thermal sensors have been deployed in subseafloor borehole observatories as part of long-term monitoring installations (e.g., Davis & Becker 1994, 2002; Foucher et al. 1997). Tools used during scientific ocean drilling and in boreholes are not discussed in the rest of this review, which focuses on measurements made in shallow sediments. Thermal gradient measurements are commonly augmented with thermal conductivity measurements made in the laboratory. Even if there is in situ conductivity capability (as with most modern, multipenetration probes), lab measurements can be helpful because they allow more comprehensive assessment of sediment properties at a scale finer than the spacing between adjacent thermistors in a multipenetration probe or outriggers on a core barrel. The needle probe method (Fig. 1E) is most commonly used for determining the thermal conductivity of marine sediments (Von Herzen & Maxwell 1959). Marine sediments are recovered in a plastic core liner. After the core equilibrates to room temperature, a small hole is drilled in the side of the liner, and a needle probe is inserted. The probe contains a wire loop with known heating characteristics, and a thermistor is positioned near the middle of the probe. The temperature response of the probe during continuous or pulse heating through the wire allows thermal conductivity to be determined by fitting observational data to an analytical solution to a radial heat conduction equation. A modified version of this experiment can be conducted on lithified sediments or igneous rock using cut pieces of hard rock pressed against a needle that is backed with a low-conductivity substrate such as acrylic or epoxy.
146 A. T. FISHER & R. N. HARRIS
Data processing Most marine heat flow tools are not kept in the formation long enough for thermistors to fully equilibrate. Full equilibration would require more time that is commonly available, and it can be difficult to recover an instrument pushed into seafloor sediments if the formation around the tool is allowed to ‘settle in’ for an extended time. Instead, equilibration temperatures are computed based on model fits to 6–8 min of temperature data, with the model response extrapolated to infinite time (Bullard & Maxwell 1956; Villinger & Davis 1987; Hartmann & Villinger 2002). Equilibration models are designed based on tool geometry and tool and formation properties. Applying these models requires some understanding of thermal conductivity because statistically equivalent fits of observations and model calculations can be achieved in many cases using a range of plausible thermal conductivity values. If in situ thermal conductivity data are collected, appropriate values can be assigned to individual thermistors. If no thermal conductivity data are available, values are estimated based on earlier surveys in the same (or a similar) area. In either case, Monte Carlo analysis of possible thermal conductivity variations and boundaries between layers allows assessment of uncertainties in equilibrium temperatures associated with these parameters (Stein & Fisher 2001; Hartmann & Villinger 2002). If thermal conductivity is constant with depth, conductive vertical heat flow is calculated with Eq. (1). This equation is modified to account for changes in thermal conductivity with depth (negative sign dropped for convenience) (e.g., Bullard 1939; Louden & Wright 1989): T ¼q
X Dz
ð4Þ
P Dz where is the cumulative thermal resistance, and Dz is the depth interval represented by each value of k. Conductive heat flow is calculated as the slope of a line fit through P Dz points on a plot of T versus .
ERRORS AND UNCERTAINTIES Inexpensive thermistor sensors can be calibrated to provide resolution and absolute accuracy of £0.001C across a useful working range of 0–40C (appropriate for thermal gradient determinations in the upper 5 m of sediment in most locations). Accurate measurements at higher temperatures are sometimes desired, especially in borehole applications, and this can be achieved through selection of thermistors having the appropriate dynamic range. Sensors are commonly calibrated in the lab using a controlled bath, or cross-calibrated in the field by holding the probe stationary in bottom water during a survey. Widely used analog-digital logger components have ‡16 bits of resolution and are
extremely quiet. Thus analytical errors associated with sensors and data loggers tend to be small. Extrapolation of partial cooling records for the sensors, and estimation of in situ thermal conductivity values, can lead to errors in individual temperature values of 0.01– 0.05C, even for high-quality data. The main difficulty is that equilibration records can be made to fit a variety of cooling curves depending on the assumed physical properties, which are rarely known with confidence. However, heat flow determinations are more dependent on the thermal gradient than individual temperatures, so measurement and extrapolation errors for individual sensors are likely to be greater than those for calculated thermal gradients in many cases. When thermal conductivity is measured in situ, the nature and properties of materials between sensor depths may be poorly known. This is a particular problem when making measurements in thin layers of sediments having large differences in thermal properties, such as turbidites. Finally, thermal conductivity is generally not measured with a geometry that is fully consistent with vertical heat flow. In situ measurements determine the horizontal (radial) conductivity, whereas needle probe measurements made on sediment cores determine the geometric mean of horizontal and vertical values (Pribnow et al. 2000). Fortunately, anisotropy in thermal conductivity tends to be small in shallow sediments. Although it is possible to quantify fit statistics when cross-plotting temperature and depth (or cumulative thermal resistance), this approach does not provide a conservative estimate of uncertainty in heat flow measurements. Repeated measurements made during a single instrument lowering suggest that precision is often on the order of 3–5% (Harris et al. 2000), but there may be additional uncertainties associated with undocumented lithologic variations, thermal refraction, or transient processes such as slumping or recent changes in bottom water temperatures. Fortunately for many applications and settings, the hydrogeologic processes inferred from thermal data often involve large differences between measured values and deviations from standard lithospheric reference curves. The best way to avoid or resolve complexities associated with shallow and transient processes is to collect closely spaced heat flow data co-located with seismic reflection and swath map data, so that areas of recent mass movement can be avoided, and to verify the consistency of results through a combination of regional and local measurement campaigns. There may be a nearby oceanographic buoy, or long-term temperature loggers can be left on the seafloor in advance to provide a baseline record from months or years prior to heat flow surveys (Hamamoto et al. 2005). Bottom water temperature variations tend to be small at water depths ‡2–3 km (Davis et al. 2003), but some deep marine environments remain susceptible to
Using seafloor heat flow to map subseafloor fluid flow 147 changes in seafloor temperature (e.g., Barker & Lawver 2000; Fukasawa et al. 2003). Heat flow, q (mW m–2)
50
(A)
Lithospheric cooling reference Global observations (binned, ±1 s.d.)
200
100
10
Ridge-flank hydrothermal circulation
35 25
6 15 2
(B) 5
Portion of lithospheric heat output (%)
Lithospheric cooling models simulate how the lithosphere cools, contracts, and subsides as it ages. Lithospheric cooling rates are estimated from bathymetry (related to the lithospheric heat content) and heat flow measurements from older seafloor, from which heat transfer is thought to be dominantly conductive. These models comprise a reference against which heat flow observations are compared, providing the basis for quantifying anomalies and calculating the extent of advective plate cooling. The two main types of cooling models (constant thickness plate, thickening half space) make similar predictions for young to middle aged seafloor heat flow as a function of lithospheric age (Parker & Oldenberg 1973; Parsons & Sclater 1977; Davis & Lister 1974; McNutt 1995; Harris & Chapman 2004). Both classes of model predict seafloor heat flow to vary as C ⁄ age initially. The dependence continues to great age for the half space model, whereas there is an asymptotic heat flow value for old seafloor predicted by the plate model. When heat flow is in mW m-2 and age is in millions of years, the magnitude of the best-fitting value of C is 475–510, depending on the data and methods used. The plate model is based on constant plate thickness at all ages, even at spreading centers where new seafloor is created. With time the plate cools until equilibrium is achieved between heating from below and heat loss at the top. In contrast, the half space model is based on the lithosphere having zero thickness at the ridge, then cooling and thickening with age. This model predicts heat flow less (and depths greater) than commonly observed on old seafloor. Some differences between the two models can be reconciled if there are additional inputs of heat into the base of older lithosphere. There has been limited recent discussion about the theoretical basis for lithospheric cooling models, but the fundamental issues and the global extent of hydrothermal cooling of the ocean basins are well-established, heavily documented by observational data, and considered reliable by the vast majority of active practitioners (e.g., Von Herzen et al. 2005). Analysis of the global marine heat flow data set shows that the conductive heat flow from young lithosphere is generally lower than predicted by standard cooling models (Fig. 2A). The heat flow deficit is greatest when the seafloor is young, and extends on average to 65 Ma. However, there are numerous older seafloor sites that appear to lose some heat advectively, and additional regions of old seafloor where heat may be redistributed locally by vigorous convection within basement, even if there is no net advective heat loss from the plate (Von Herzen 2004).
Seafloor age (Ma) 30
Heat flow deficit
Cumulative advective heat output (TW)
GLOBAL CONSIDERATIONS
10
Fig. 2. (A) Schematic comparison between standard lithospheric cooling models (solid curve) and global compilations of oceanic heat flow measurements (solid squares = 2 Ma bin averages, error bars = ±1 SD). The difference between modeled and observed values is the conductive heat flow deficit, generally attributed to advective heat loss. The observed variability of measurements is often attributed to local advective redistribution of heat by vigorous convection in basement. (B) Cumulative global advective heat output, based on integrated difference between lithospheric and observed heat flow values, taking into account the area of seafloor occupied by each age bin. Band indicates a range of values from different studies (Sclater et al. 1980; Stein et al. 1995; Mottl 2003). Advective heat loss at seafloor spreading centers comprises a fraction of the total advective heat loss, with the majority occurring on ridge flanks.
When the marine heat flow deficit is integrated across the area of the seafloor as a function of age, the global advective heat loss from the ocean crust is 10 TW, 25% of Earth’s current geothermal heat output, and 35% of the heat lost through the seafloor by conduction and advection combined (Sclater et al. 1980; Stein & Stein 1994; Mottl 2003) (Fig. 2B). Of this advective heat loss, 2–3 TW occurs at or near seafloor spreading centers, and the remaining 7–8 TW occurs on ridge flanks, far from the magmatic influence of lithospheric creation. The total fluid flux (F) implied by the global advective heat output (Q adv) is calculated as: F¼
Q adv ðc Þf DT
ð4Þ
where (qc)f is the heat capacity of hydrothermal fluid, and DT is the difference in temperature between oceanic bottom water and hydrothermal discharge. Thus the calculated fluid flux scales linearly with the characteristic temperature of fluid circulation. Field studies of ridge flank areas from which a significant fraction of lithospheric heat loss is advective show that typical basement temperatures are on the order of 5–40C, which implies that F = 1015– 1016 kg year-1 (Stein & Stein 1994; Mottl 2003), a flow rate approaching that of all of rivers and streams into the ocean.
148 A. T. FISHER & R. N. HARRIS The comparison of thousands of seafloor heat flow observations to well-established models of lithospheric cooling provides a robust estimate of the total amount of water that must flow between the crust and ocean. This information can be combined with highly idealized models to estimate bulk crustal properties on a global or timedependent basis (e.g., Fisher & Becker 2000). However, the comparison of age-binned data to models tells us little about specific processes or properties at individual field sites. Resolving these characteristics requires application of regional and local models. This is especially important for understanding the variability in heat flow values observed in many surveys, once considered to be noise but now understood to result from fundamental physical processes, as described in the following section.
REGIONAL AND LOCAL CONSIDERATIONS Regional and local models Multiple processes can contribute to anomalous seafloor heat flow (Fig. 3). In locations where there is relatively flat seafloor and buried basement relief, conductive refraction tends to increase heat flow immediately above an elevated basement area, because basalt of the upper ocean crust tends to have a thermal conductivity that is greater than that of typical marine sediments (Fig. 3A). Heat flow can be further elevated above a buried basement high if there is vigorous local convection in the shallow crust, making uppermost basement isothermal (Davis et al. 1989; Fisher et al. 1990). In practice, it can be difficult to distinguish between these two processes based on the general pattern of seafloor heat flow, but modeling can quantify the extent of the necessary thermal conductivity contrast or homogeneity in basement temperatures (Davis et al. 1997b). Recharge of bottom seawater though areas of basement exposure, followed by rapid lateral flow and mixing within the crust, can suppress seafloor heat flow on a regional basis (Fig. 3B). The extent of suppression can be calculated through application of the ‘well-mixed aquifer’ (WMA) model, a one-dimensional representation of vertical heat flow through a horizontal flow system (Langseth & Herman 1981). The upper oceanic crust is idealized as having a conductive boundary layer (sediments), underlain by an aquifer (shallow basement), within which there is thermally significant lateral transport. Recharge to the aquifer occurs at the temperature of bottom water, and heat is exchanged with the crust during lateral flow. The aquifer is assumed to be well mixed vertically, having a temperature determined by the balance between heat entering from below and leaving through the conductive upper boundary. Key information required to compare observations to predictions based on the WMA model includes: measured sea-
floor heat flow, the distance from the measurement points to the site of recharge, the thicknesses of the conductive layer and the permeable rock layer through which fluid flows laterally, estimated heat input from below the aquifer (often assumed to be lithospheric), and thermal properties of the sediment and rock ⁄ fluid system. The WMA model can be extended to include a second conductive boundary layer below sediments and above permeable basement (e.g., Rosenberg et al. 2000). The influence of basement relief, local convection, and regional advection may be combined if there is outcrop-tooutcrop circulation (Fig. 3C). Seawater entering the crust, circulating laterally, then existing the crust can carry a significant fraction of lithospheric heat if fluid fluxes are sufficiently large, leaving the conductive seafloor heat flow suppressed between recharge and discharge sites (Davis et al. 1992; Villinger et al. 2002; Fisher et al. 2003b). This process can lead to very low heat flow adjacent to areas where hydrothermal recharge occurs, and locally elevated values adjacent to locations where warmed fluids rise rapidly and exit the crust, even when the rate of fluid flow is insufficient to mine heat from the lithosphere regionally (Fisher et al. 2003a). For each of the situations described above, the sediments overlying basement were assumed to act as a conductive boundary layer, with the most vigorous circulation occurring within underlying basement. Most marine sediments have a permeability orders of magnitude less than that of volcanic rocks from the upper oceanic crust (Spinelli et al. 2004). Because of this, and because driving forces for hydrothermal circulation are limited, there is little fluid flow through sediments that extracts significant quantities of lithospheric heat. The vast majority of seafloor heat flow measurements indicate conductive conditions within shallow sediments. Where there is a component of vertical flow through sediments that perturbs the local geothermal gradient, deviations from conductive conditions can be used to estimate the seepage rate. One model used for interpretation of vertical seepage is based on flow through a layer having constant temperature boundaries at the top and bottom (Fig. 3D) (Bredehoeft & Papadopulos 1965). The assumption that upper and lower boundaries have fixed temperatures is most appropriate when the flow through the layer is modest relative to the size of overlying and underlying reservoirs. In the case of seafloor hydrologic systems, this model has been applied most commonly to marine sediments, bounded at the top by the ocean (essentially an infinite sink for heat), and at the base by a large hydrothermal reservoir. The minimum seepage rate necessary for detection of deviations from conductive conditions depends on the thickness of the boundary layer and the depth extent of measurements. For conductive heat flow of 2 W m-2 across a layer 20-m thick, a flow rate
Using seafloor heat flow to map subseafloor fluid flow 149
q
(A) Local advection
Regional reference
Conduction
Sediments
Basement aquifer
q
(B)
Sediments
Lateral flow and mixing
Basement aquifer
C
(C)
q
Regional reference
Recharge site
Discharge site
Sediments Basement aquifer Thermal conductivity
10 15 20
r
r m/y
30 m
16 mm/y
3 mm/yr
r /yr m/y 30 mm
‡30 mm year-1 may be required. Heat flow measurements made with shorter probes require greater advective transport in order to be detected.
3m
Constant temperature lower boundary
300 mm/yr 3 m/yr
/yr mm r m/y 3m
Depth (m)
5
0
(W/m °C) (D) Temperature (°C) (E) Temperature (°C) (F) 10 20 30 40 10 20 30 40 1.0 1.8 2.6 3 m/yr Fixed Equivalent 30 heat flow thermal 0 m lower conductivty m /yr boundary 90 m m /y r 30
Fig. 3. Illustrations of how regional and local fluid flow can influence seafloor heat. The dashed line labeled ‘‘regional reference’’ in panels A–C could be lithospheric, or could be a lower or higher value that is typical of regional conditions. (A) Buried basement highs can cause conductive refraction, increasing heat flow through overlying sediments. Vigorous local convection can make the sediment–basement interface isothermal, raising heat flow at the seafloor above the basement high even more, depending on the extent of relief and isothermal conditions. B. Recharge of cold bottom water into the seafloor lowers heat flow in nearby areas, with heat flow rising as fluids in the underlying aquifer flow laterally, mix vertically, and are heated (Langseth & Herman 1981). (C) Outcrop-to-outcrop circulation lowers seafloor heat flow immediately adjacent to areas of local recharge, and raises seafloor heat flow immediately adjacent to areas of hydrothermal discharge. Regional heat flow may be suppressed between outcrops if fluid circulation is sufficiently vigorous and extensive (e.g., Davis et al. 1992; Fisher et al. 2003b), or regional heat flow may be essentially unaffected by this form of circulation (Hutnak et al. 2006). (D) Upward fluid seepage can cause curvature in thermal gradients, with temperatures held constant at the top and bottom of the boundary layer (Bredehoeft & Papadopulos 1965). The extent of curvature depends on the thickness of the layer, background gradient, thermal properties, and fluid flow rate. In this example, the layer is 20 m thick, thermal conductivity is 1 W (m C)-1, background heat flow is 2 W m-2, and fluid flow rates are as indicated. (E) As an alternative, the top of the boundary can be held at constant temperature, and total heat flow through the layer (advective and conductive) can be fixed (Wolery & Sleep 1975). Many geological systems are likely to function somewhere these two end members. (F) Thermal conductivity variations that would mimic the flow rates shown by causing curvature in the conductive thermal gradient based on the Bredehoeft & Papadopulos (1965) model (panel D).
Regional reference
An alternative application involves a case when the heat input to the base of the sediment layer is limited, rather than being infinite (Fig. 3E) (Wolery & Sleep 1975). In
150 A. T. FISHER & R. N. HARRIS There are three very young (essentially 0–0.1 Ma) sites where seafloor heat flow surveys have been conducted: Guaymas Basin in the Gulf of California, Escanaba Trough on the Gorda Ridge, and Middle Valley at the northern end of the Juan de Fuca Ridge (Lonsdale & Becker 1985; Abbott et al. 1986; Davis & Villinger 1992). All three are sedimented spreading centers, located adjacent to continental areas where thick sediments trap heat within the subsurface, compressing thermal and chemical gradients near the seafloor, focusing hydrothermal recharge and discharge, and allowing thermal and fluid flow regimes to be characterized through coring, drilling, and heat flow surveys (Davis & Villinger 1992; Gieskes et al. 1982; Kastner 1982; Stein & Fisher 2001; Wheat & Fisher 2007). Studies from these areas provide unique insights on coupled fluid–thermal processes within very young seafloor. Middle Valley (Figs 4A and 5A) is bordered by ridgeparallel, high-angle normal faults and filled with Pleistocene turbidites and hemipelagic clay deposited when sea level was lower and terrigeneous sediments were transported down the continental slope and into Cascadia Basin (Davis & Villinger 1992). As at other sedimented spreading centers, basaltic magma rises from depth and intrudes laterally as sills within thick sediments rather than erupting at the seafloor. Upper lithologic ‘basement’ in this setting comprises alternating sills and indurated sediments having bulk permeability much greater than that of overlying sediments (Becker et al. 1994; Fisher et al. 1994; Langseth & Becker 1994). This sediment–sill sequence hosts vigorous hydrothermal circulation that leads to essentially isothermal (280C) conditions in uppermost basement (Davis & Fisher 1994; Davis & Villinger 1992; Davis & Wang 1994). Although Middle Valley remains tectonically and hydrothermally active, it appears that the primary focus of seafloor spreading is currently shifting to the adjacent West Valley. The total heat flux from a 260 km2 area of Middle Valley suggests heat output at a spatial rate of 16 MW km-1 of ridge (Stein & Fisher 2001), a value consistent with estimated heat output from other (unsedimented) parts of
this case, the conductive gradient is reduced as upward seepage (and heat advection) increases. Conditions in many marine hydrologic systems are likely to be somewhere between these two endmembers. The seafloor boundary condition can reasonably be assumed constant, but the lower boundary condition may be neither constant temperature nor constant heat flow. In practice, interpretations of thermally significant seepage through sediments should be applied cautiously, because the forces available to drive fluids through marine sediments are limited in many settings, and because systematic changes in thermal conductivity can lead to curvature in thermal gradients under conditions of conductive heat flow (Fig. 3F). Vertical analytical models of coupled fluid-heat flow have been modified to account for lateral flow (e.g., Lu & Ge 1996), but this is rare in shallow sediments, because the sedimented seafloor is usually flat or gently sloped and lateral pressure gradients tend to be small. An exception to this latter generalization is presented in the following section, and subsequent sections show how arrays of heat flow measurements can be used to resolve lateral flow rates in basement. Multiple scales of fluid circulation: Middle Valley, northern Juan de Fuca Ridge There are few heat flow studies of active seafloor spreading centers based on measurements of shallow thermal conditions because most such areas are characterized by large regions of basement exposure and limited sediment cover. The most common means for assessing hydrothermal output on young seafloor is through measurement of the heat content of chronic plumes formed by discharge hot, highly altered hydrothermal fluids (Baker & Hammond 1992; Veirs et al. 1999). Plume studies have shown that the advective heat output associated with seafloor spreading scales with spreading rate, and that the advective heat output of these areas is generally consistent with crystallization and cooling of the upper 1–2 km of oceanic crust (Baker 2007).
50°N
North America
Central
46°
Juan de Fuca R idge
50°
48°
16°N America
130°W
Cocos Plate
MV
Cocos - Nazca
BB
Juan de Fuca Plate
126°
8°
Spreading Center (A)
92°W
(B)
84°
Fig. 4. Regional index maps showing locations of field areas discussed in this paper. (A) Northeastern Pacific Ocean. MV = Middle Valley. BB = Baby Bare outcrop. Dashed red box is area of Fig. 6A. (B) Eastern Equatorial Pacific Ocean. Dashed red box is area of Fig. 7.
Using seafloor heat flow to map subseafloor fluid flow 151 the Juan de Fuca Ridge (Baker 2007). But in contrast to more normal seafloor spreading centers, conductive and advective heat outputs from Middle Valley are about equal in magnitude. Bare rock spreading centers are thought to release the vast majority of their heat advectively. The Dead Dog vent field within Middle Valley has been explored with numerous instruments, including a side-scan survey that defined the area of active venting (AAV) based on high acoustic backscatter (Davis & Villinger 1992) (Fig. 5B). Within the AAV there are 20 vents discharging hydrothermal fluid at temperatures up to 280C. A scientific drill hole placed near the center of this AAV cored extrusive basaltic rocks below 260 of turbidites and hemipelagic clay. Repeated heat flow surveys (including data from multipenetration probes and outriggers probes attached to core barrels) show that the highest values are found immediately adjacent to active vents (Fig. 5B). Interestingly, some of the lowest heat flow values were also measured close to active vents. The mean heat flow of 10 W m–2 within the Dead Dog AAV has been interpreted to result from formation of a diagenetic and hydrologic cap above a shallow hydrothermal reservoir (Davis & Villinger 1992; Davis & Fisher 1994; Stein et al. 1998; Stein & Fisher 2001). Fluid con-
T = 280 oC, No flow
6 0.
0.4
0.4
0.8
0.8
(D) No connection to vent
27.6'
48o 20' N
Mean vent field heat flow
0.4
0
>1 W
5
10 km
0
m–2
128o 50' W
40'
128o42.8'W
42.6'
100 200 m
42.4'
B-13 A-12 B-12
10
A-13
Minimum vent field heat flow
10 20 30 Distance from vent (m)
Heat flow (W/m2)
6
20
27.4'
0.6
0.
(C)
B
0.4
0.6
0.4 0.6
T = 2 oC, Free flow
A
No flow
48o 27.2' N
active vent
0.2
30'
(B)
q (W m–2) <2 2-4 4-10 >10
T = 280 oC
(A)
vects vigorously below the cap, helping to maintain isothermal conditions near 280C at 30 m below the seafloor. There may be additional focusing of conductive heat flow because of refraction related to the buried basement edifice and shallow sulfide deposits, both of which have greater thermal conductivity than typical marine sediments (Fig. 3A). The scattering of heat flow values both above and below the mean of 10 W m-2 within the AAV can be explained by proximity to active vents, as demonstrated with coupled numerical modeling (Fig. 5C and D). The vents act as conductive heat sources as hydrothermal fluid rises to the seafloor, warming shallow sediments around the vents, and elevating heat flow to values ‡50 W m-2. In addition, local recharge associated with secondary convection can suppress heat flow to <1 W m-2 close to vents. This secondary convection cannot be a significant mechanism for recharging the main hydrothermal system because fluid samples collected from the vents show no indication of mixing between hydrothermal fluid and ambient seawater, and there is little evidence for water–sediment interaction prior to fluid discharge (Butterfield et al. 1994). Instead, it is likely that recharge of fluids supporting high-temperature discharge in Middle Valley occurs mainly along faults and
0
40
Fig. 5. Diagrams showing data and model results from Middle Valley, northern Juan de Fuca Ridge. Figures modified from Davis & Villinger (1992) and Stein & Fisher (2001). (A) Regional view of Middle Valley, showing valley bounding normal faults (thin lines with ticks) and area of the most complete coverage of seafloor heat flow measurements (within thicker dashed line). Contours show typical heat flow values (units of W m-2), and grey areas show where heat flow is ‡1 W m-2. Dotted box shows area of Fig. 5B. (B) Area of active venting (AAV) including the Dead Dog vent field. Edge of AAV identified by side scan sonar surveys (Davis & Villinger 1992). Open circles indicate location and magnitude of heat flow values, as labeled. Solid squares are individual hydrothermal vents. Crosses indicate at least one thermistor exceeded maximum temperature range of instrument (40C). Stars indicate locations where the heat flow probe fell over without penetrating. (C) Schematic of radial cross-section used in simulations of secondary circulation around a hydrothermal vent. Horizontal (10 and 20 mbsf) and vertical (20 m from vent) 1-m thick channels were used in some simulations, labeled to show flow paths A and B, respectively. (D) Surface heat flow versus distance from vent for simulations with secondary hydrothermal convection. Background heat flow is 10 W m-2. Conductive case is shown for reference with no connection to vent. Cases A and B have a shallower and deeper connection to the hydrothermal vent, respectively, and each is run with connection permeability of 10-12 m2 and 10-13 m2 as shown. The deeper the horizontal layer and the higher its permeability, the more efficient the local heat flow reduction associated with secondary convection.
152 A. T. FISHER & R. N. HARRIS other isolated conduits along the edges of the valley (Davis & Villinger 1992; Stein & Fisher 2001; Wheat & Fisher 2007). In summary, heat flow and other data from Middle Valley resolve multiple scales of fluid circulation, delineate conditions and the depth to the top of a hydrothermal reservoir, and help to define the regional geometry of fluid flow, including the importance of basement outcrops in guiding hydrothermal recharge. The extent to which these results can be applied to more normal (bare rock) seafloor spreading centers remains to be determined.
FLUID FLOW THROUGH OUTCROPS: EASTERN FLANK OF THE JUAN DE FUCA RIDGE Numerous physical and chemical surveys have been completed on 0.7–3.6 Ma seafloor on the eastern flank of the Juan de Fuca Ridge (Davis et al. 1992, 1999; Wheat & Mottl 1994, 2000; Thomson et al. 1995; Elderfield et al. 1999; Fisher et al. 2003a; Hutnak et al. 2006; Wheat et al. 2004). Oceanic basement rocks in this area were formed at the unsedimented Endeavour segment of the Juan de Fuca Ridge, but were subsequently covered at a young age by hemipelagic clay and turbidites. Basement rocks remain
(A)
exposed close to the active spreading center and where seamounts and other basement outcrops are found up to 100 km to the east. Three basement outcrops were identified initially in this area: Papa Bare, Mama Bare, and Baby Bare (Fig. 6A) (Davis et al. 1992; Thomson et al. 1995; Mottl et al. 1998). These outcrops are volcanic cones created off-axis on top of preexisting abyssal hill topography. Baby Bare outcrop is the smallest and most extensively surveyed of the three outcrops, rising 70 m above the surrounding seafloor and covering an area of 0.5 km2. Although the elevated area of this feature is small, the buried Baby Bare edifice rises 600 m above the top of regional volcanic crustal rocks. Analysis of altered rocks, sediment pore fluids, and shallow thermal gradients on Baby Bare outcrop indicate that this feature discharges 5–20 L s-1 of hydrothermal fluid (Thomson et al. 1995; Mottl et al. 1998; Becker et al. 2000; Wheat et al. 2004). Mama and Papa Bare outcrops (Fig. 6A) are also sites of ridge-flank hydrothermal discharge, although these features have not been surveyed as completely as Baby Bare. Mama Bare outcrop is located 14 km northeast of Baby Bare outcrop on the same basement ridge, covers an area of 0.9 km2, and rises 140 m above the surrounding seafloor. Papa Bare outcrop is located on an adjacent buried
(B)
Fig. 6. Maps of field area on the eastern flank of the Juan de Fuca Ridge (index map shown in Fig. 4A). (A) Regional map showing locations of basement outcrops on 3.5–3.6 Ma seafloor (modified from Hutnak et al. 2006). Baby Bare, Mama Bare, and Papa Bare outcrops are sites of hydrothermal discharge, whereas Grizzly Bare is a site of hydrothermal recharge (Fisher et al. 2003a). (B) Fine scale topographic map of Baby Bare outcrop, with locations of heat flow measurements made with the submersible, Alvin (Wheat et al. 2004). Highest heat flow values are associated with a normal fault on the eastern side of the outcrop. Several of these heat flow values are based on curved gradients, as shown in the inset diagram. Numbers in the inset indicate apparent seepage rates, in m year-1, based on application of the model shown in Fig. 3D.
Using seafloor heat flow to map subseafloor fluid flow 153 basement ridge to the east, covers an area of 2.6 km2, and rises 240 m above the surrounding seafloor. As with Baby Bare outcrop, most of the Papa and Mama Bare edifices are buried by thick sediments. Interestingly, no hydrothermal recharge sites have been identified on any of these three features. Consideration of ridge-flank hydrothermal driving forces, basement fluid chemistry, and sediment properties precludes recharge of Baby Bare outcrop through the seafloor surrounding the basement edifice; instead, fluids recharge through Grizzly Bare outcrop 52 km to the south (Fig. 6A) (Fisher et al. 2003a). Grizzly Bare outcrop is conical in shape, 3.5 km in diameter, and rises 450 m above the surrounding seafloor. The consistency of their alignment and strike with regional basement topography (Wilson 1993; Zu¨hlsdorff et al. 2005) suggests that the Grizzly, Baby, and Mama Bare outcrops may have formed along the same buried abyssal hill. Fluid flow in basement between Grizzly Bare and Baby Bare outcrops may be facilitated by enhanced permeability in a direction parallel to the abyssal hill topography (Fisher et al. 2003a, 2008; Wheat et al. 2000). Grizzly Bare outcrop was identified as a site of hydrothermal recharge based initially on patterns of seafloor heat flow immediately adjacent to the edifice. Seafloor heat flow is depressed within a few kilometers of the edge of basement exposure along several transects of measurements (Fisher et al. 2003a; Hutnak et al. 2006). Seismic reflection data allow determination of sediment thicknesses at heat flow measurement locations, and downward continuation of thermal data through the sediment shows that isotherms are swept downwards by cold, recharging fluid in the sediments and shallow basement adjacent to the outcrop edge. In contrast, warm fluid discharge from Baby Bare outcrop causes extremely high seafloor heat flow, and an upward sweeping of isotherms, adjacent to the area of exposed basement (Davis et al. 1992; Fisher et al. 2003a). Baby Bare outcrop was surveyed repeatedly using the submersible Alvin, including high-resolution bathymetric mapping, geological and pore fluid sampling, and heat flow measurements (Wheat et al. 2004) (Fig. 6B). The Alvin heat flow probes used on these surveys were 0.60 m long and contained three or five thermistors. A constant thermal conductivity of 0.89 W (m C)-1 was assumed for all Baby Bare heat flow measurements (Wheat et al. 2004), based on in situ and needle probe measurements of material recovered nearby during subsequent surveys (Fisher et al. 2003a; Hutnak et al. 2006). Most heat flow values measured on Baby Bare outcrop were >1 W m-2, well above values measured on the adjacent flat seafloor, and the highest Baby Bare values were >100 W m-2. There is an area of elevated heat flow aligned along a linear trend on the southwest side of the outcrop, adjacent to a normal fault scarp near the outcrop summit (Fig. 6B). Sediment thick-
ness is not well mapped across Baby Bare outcrop because of irregular topography, but the depth to the 64C isotherm (consistent with regional basement temperatures and fluid chemistry) is generally £50 m (Davis et al. 1992; Elderfield et al. 1999; Hutnak et al. 2006; Wheat & Fisher 2007). Several thermal profiles from Baby Bare outcrop display significant curvature with depth. These thermal profiles were interpreted to indicate fluid flow, through application of the Bredehoeft & Papadopulos (1965) model (Fig. 3D), only if curvature could not be accounted for by reasonable variations in thermal conductivity, and only if the apparent seepage rate exceeded 6.5 m year-1 (Fig. 6B, inset, Wheat et al. 2004). Other profiles, indicating apparently conductive thermal conditions, were evaluated based on nearby geochemical data from push cores. Geochemical data were fit to an advection–diffusion model similar to that used for assessing curvature in thermal data. But because chemical diffusivity for solutes is about three orders of magnitude smaller than thermal diffusivity, chemical data are much more sensitive to fluid flow. Chemical data were used to estimate seepage rates between 0.005 and 2 m year-1, above which chemical concentrations are essentially constant with depth in the upper 0.5 m of sediment. Total heat flow values were calculated as the sum of conductive and advective components (Wheat et al. 2004). Thermal and chemical data were combined to estimate point and total seepage rates, and integrated heat output, from Baby Bare outcrop. These data suggest heat output of 2 MW for Baby Bare and the immediately surrounding seafloor, a value similar to that estimated in earlier seafloor studies (Mottl et al. 1998), and at the lower end of estimates made from thermal anomalies in the overlying water column (Thomson et al. 1995). Estimated Baby Bare heat output is about the same as that from a single high-temperature (350C) hydrothermal vent. The mean heat flow from Baby Bare outcrop is an order of magnitude greater than that estimated from lithospheric cooling models for 3.5 Ma seafloor like that upon which Baby Bare is located. Although there are small areas on Baby Bare outcrop where shimmering water can be seen rising from the seafloor, the thin drape of sediment overlying much of the basement edifice functions effectively as a thermal boundary layer, with vigorous convection in the underlying basement redistributing heat laterally and enhancing the rate of conduction through the top of the edifice. Advective heat loss from Baby Bare outcrop is limited, and the present day circulation system has virtually no influence on regional heat flow (Davis et al. 1999; Hutnak et al. 2006), although it is likely that advective heat loss was considerably greater in the past, before many other outcrops were covered by thick, low-permeability sediments (Hutnak & Fisher 2007).
154 A. T. FISHER & R. N. HARRIS
Massive fluid fluxes resolved with thermal data: Eastern Cocos Plate The 18–24 Ma seafloor of the Cocos Plate offshore the Nicoya Peninsula, Costa Rica (Fig. 4B), comprises a northern region formed at the East Pacific Rise (EPR), and a southern region formed at the Cocos-Nazca Spreading Center (CNS) (Fig. 7) (Meschede et al. 1998; Ranero & von Huene 2000; von Huene et al. 2000; Barckhausen et al. 2001). The boundary between EPR- and CNS-generated seafloor is a combination of a triple junction trace and a fracture zone trace, collectively comprising a ‘plate suture.’ Pre-2000 studies in this area revealed variable thermal conditions (e.g., Langseth & Silver 1996; Vacquier et al. 1967; Von Herzen & Uyeda 1963), but the spatial distribution of warmer and cooler seafloor was incomplete, and the extent and nature of hydrothermal activity was unclear. Later surveys included swath-mapping to identify basement outcrops, and multi-channel seismic reflection data to delineate regional tectonic features, sediment thickness, and basement relief (Fisher et al. 2003b; Hutnak et al. 2007). Multipenetration heat flow data were co-located on seismic reflection profiles to assess heat transport and determine upper basement temperatures; additional heat
(A)
flow measurements were made with autonomous temperature probes attached to core barrels (Pfender & Villinger 2002; Hutnak et al. 2007). Heat flow surveys documented a thermal transition between warm and cool areas of the Cocos Plate (Fisher et al. 2003b; Hutnak et al. 2007) (Fig. 7). The mean heat flow through CNS- and EPR-generated seafloor on the warm part of the plate is consistent with lithospheric thermal reference models, 95–120 mW m-2 (Parsons & Sclater 1977; Stein & Stein 1994). In contrast, the seafloor heat flow through EPR-generated seafloor northwest of the thermal transition is typically 10–40 mW m-2, just 10–40% of lithospheric values. The thermal transition between warm and cool areas of the plate is only a few kilometers wide (Fig. 8), consistent with advective heat extraction from the shallow crust on the cool side of the plate (Fisher et al. 2003b; Hutnak et al. 2007). The thermal transition coincides with the plate suture near where the Cocos Plate begins to subduct in the Middle America Trench. But through most of the survey area, the boundary between warmer and cooler parts of the plate is spatially associated with the occurrence of seamounts and other basement outcrops that penetrate regionally extensive sediments (Fig. 7).
(B)
Fig. 7. A. Bathymetric map of field area on 18–24 Ma seafloor of the Cocos Plate, eastern Pacific Ocean (index map shown in Fig. 4B) (modified from Hutnak et al. 2008). Dorado outcrop is a site of hydrothermal discharge, whereas Tengosed Seamount is a site of hydrothermal recharge. Gray line demarcates thermal boundary between cooler and warmer parts of the plate. (B) Major tectonic features and results of heat flow surveys. Gray area is cool part of Cocos Plate. Black circles and ovals are seamounts and other basement outcrops apparent on satellite and swath data. EPR = East Pacific Rise. CNS = Cocos-Nazca Spreading Center.
Heat flow (mW m–2)
Using seafloor heat flow to map subseafloor fluid flow 155
100 Conductive lithospheric cooling
Heat loss dominantly conductive
20
4000
Two-way travel time (ms)
Heat loss dominantly advective
60
NW
SE
Seafloor Sediments
4400
4800
10°C
Basement
5200
0
20°C 30°C CDP 6000
4
8
12 km
50°C 8000
10 000
12 000
14 000
Fig. 8. Co-located heat flow and seismic data that cross the thermal transition between cool and warm parts of the Cocos Plate (location shown in Fig. 7A). Horizontal axis is common depth point (CDP) of the seismic survey. Diagonal band shows range of lithospheric values predicted by standard models for seafloor of this age. Transition between cool and warm parts of the plate occurs across 2 km, indicating an abrupt change in mechanisms of heat transport in the shallow crust, with vigorous hydrothermal cooling on the northwest side of the transition, and conductive (lithospheric) conditions on the southeast side (Fisher et al. 2003b). Isotherms superimposed on the seismic section were calculated by downward continuing heat flow measurements (Hutnak et al. 2007).
(A)
(B)
Fig. 9. (A) Seismic and heat flow data near and across Dorado outcrop (location shown in Fig. 7A). Basement rises to penetrate 400–500 m of regionally continuous sediment, and is exposed across a small area of seafloor. Isotherms (black lines) created by downward continuation of heat flow data. Heat flow rises abruptly near the edge of the edifice, but basement temperatures are nearly isothermal at 10–20C beyond several kilometers from the outcrop. (B) Seismic and heat flow data adjacent to Tengosed Seamount, 20 km east of Dorado (location shown in Fig. 7A). Note differences in vertical scale relative to panel A. Tengosed Seamount is a large edifice rising >1 km above the seafloor, 2 km above regional basement. The upper crust here is chilled below thick sediments by cold bottom water that recharges through exposed basement. Heat flow (and basement temperatures) rise 10 km south of the outcrop, after passing through the regional thermal transition (dashed red line).
Heat flow and seismic surveys oriented radially away from outcrops indicate that some allow hydrothermal recharge whereas others allow hydrothermal discharge (Hutnak et al. 2007, 2008), a pattern similar to that seen on younger seafloor east of the Juan de Fuca Ridge (Davis
et al. 1992; Fisher et al. 2003a; Hutnak et al. 2006). Seismic and heat flow data across Tengosed Seamount and Dorado ourcrop, areas of basement exposure separated by 20 km, illustrate characteristic features of this circulation pattern (Figs 7 and 9). Cold fluid recharge is indicated by
156 A. T. FISHER & R. N. HARRIS a decrease in seafloor heat flow and a downward sweeping of isotherms where sediment thins in proximity to an outcrop. In contrast, warm fluid discharge results in locally elevated seafloor heat flow, and an upward sweeping of isotherms adjacent to an area of exposed basement. In the latter case, the temperature of the sediment–basement contact often remains nearly isothermal as the contact shallows towards the seafloor. Regional advective power and fluid fluxes from the cold part of the Cocos Plate were quantified using the conductive heat flow deficit (Hutnak et al. 2008). There is a 14,500 km2 area of cool lithosphere on the EPR side of the Cocos Plate. The mean heat flow in this area, away from the local conductive influence basement outcrops and buried ridges, is 29 ± 13 mW m-2 (±1 SD), comprising a regional power deficit of 0.8–1.4 GW. Heat flow profiles adjacent to five of the ten mapped basement outcrops in this area indicate recharge through two, discharge through one (Dorado outcrop), one that both recharges and discharges, and one that shows evidence for neither recharge nor discharge (Hutnak et al. 2008). The mean advective power output of individual discharging outcrops on cooler TicoFlux seafloor is 200–350 MW, a range similar to that determined from plume and point studies of high-temperature vent fields on the southern Cleft segment of the Juan de Fuca Ridge (JdFR) and 21N on the EPR (at the low end) and the Endeavour Main Field on the JdFR and 950’N on the EPR (at the high end) (Baker 2007). But in contrast to these ridge-crest systems, the enormous advection of heat from a few outcrops on the Cocos Plate is conveyed by fluids having temperatures only 5–40C warmer than bottom water (Hutnak et al. 2008). Thermal data from the cool side of the Cocos Plate indicate 4–80 · 103 L s-1 of fluid enters and exits the seafloor in this area. If this fluid flow were distributed across the discharging outcrops, consistent with interpretations from the heat flow surveys, then each discharging outcrop vents 1–20 · 103 L s-1 of cool hydrothermal fluid (Hutnak et al. 2008).
SUMMARY, LESSONS LEARNED, AND PROSPECTS FOR FUTURE WORK Marine heat flow measurements have helped to quantify fluid flow patterns and rates within the seafloor for more than 30 years. Even before the discovery of seafloor hydrothermal vents, researchers used variations in oceanic heat flow to quantify the extent and nature of hydrogeologic processes in the deep sea. Heat flow data are most useful for this purpose when they are carefully navigated and colocated with reflection seismic and mapping data, so that thermal measurements can be placed in a broader context. It is also helpful to combine thermal studies with coring and chemical analyses, and to obtain high-resolution
images or direct observations of seafloor conditions, so that the significance of small-scale variations in heat output can be understood. This has become the standard mode of operation for modern seafloor heat flow studies. Although the studies described in this review were completed in different hydrogeologic settings, they share several common themes. Marine sediments generally function as a conductive boundary layer separating the volcanic crustal aquifer from the overlying ocean. This is a benefit for using seafloor heat flow to map fluid flow, because the (sedimentary) conductive boundary layer provides a relatively continuous medium in which measurements can be made, and across which thermal conditions can be determined. There may be thermally significant fluid flow through marine sediments in some locations, but demonstrating this requires that curvature in thermal gradients cannot be explained by reasonable differences in thermal conductivity. In fact, thermally significant fluid seepage through seafloor sediments is relatively rare, being restricted to locations where sediments are thin or unusually permeable, or where driving forces are unusually high. Patterns of conductive heat flow through marine sediments can be used to map out directions and assess rates of fluid circulation within the underlying crust, and are particularly helpful in identifying locations of fluid recharge and discharge near where basement rocks are exposed. Seafloor heat flow measurements are less helpful in resolving the nature of fluid pathways at depth within the crust, but this issue is being addressed to some extent with analysis of core samples and single-hole and multi-hole measurements and experiments (e.g., Becker et al. 1983; Anderson et al. 1985; Bartetzko et al. 2001; Alt & Teagle 2003; Becker & Davis 2005; Fisher et al. 2008). The continued interest in using heat as a tracer to map fluid flow; more common acquisition of complementary data such as high-resolution digital maps and seismic reflection profiles; and the development of new instruments, analytical techniques, and models, is bringing new attention and new practitioners to the field. Most of the seafloor remains poorly mapped and virtually unexplored with regard to thermal and hydrogeologic conditions. Seafloor heat flow remains an important tool for researchers studying coupled fluid, thermal, chemical, tectonic, magmatic, and biological processes. Heat flow measurements are particularly helpful for quantifying local and regional heat and fluid flow rates, and for mapping the plumbing of subseafloor hydrogeologic systems.
ACKNOWLEDGMENTS This work was supported by National Science Foundation grants OCE-0727952 and OCE-0849354 (ATF) and grants OCE-0637120 and OCE-0849341 (RNH). The authors appreciate thoughtful comments by two
Using seafloor heat flow to map subseafloor fluid flow 157 external reviewers that improved the clarity and focus of this paper.
REFERENCES Abbott DH, Morton JL, Holmes ML (1986) Heat flow measurements on a hydrothermally active, slow spreading ridge: the Escanaba Trough. Geophysical Research Letters, 13, 678–80. Alt JC, Teagle DAH (2003) Hydrothermal alteration of upper oceanic crust formed at a fast spreading ridge: mineral, chemical, and isotopic evidence from ODP Site 801. Chemical Geology, 201, 191–211. Anderson MP (2005) Heat as a ground water tracer. Ground Water, 43, 1–18. Anderson RN, Hobart MA (1976) The relationship between heat flow, sediment thickness, and age in the eastern Pacific. Journal of Geophysical Research, 81, 2968–89. Anderson RN, Zoback MD, Hickman SH, Newmark RL (1985) Permeability versus depth in the upper oceanic crust: in-situ measurements in DSDP Hole 504B, eastern equatorial Pacific. Journal of Geophysical Research, 90, 3659–69. Baker ET (2007) Hydrothermal cooling of midocean ridge axes: Do measured and modeled heat fluxes agree. Earth and Planetary Science Letters, 263, 140–50. Baker ET, Hammond SR (1992) Hydrothermal venting and the apparent magmatic budget of the juan de fuca ridge. Journal of Geophysical Research, 97, 3443–56. Baker ET, Massoth GJ, Feely RA, Embly RW, Thomson RE, Burd BJ (1995) Hydrothermal event plumes from the CoAxial seafloor eruption site, Juan de Fuca Ridge. Geophysical Research Letters, 22, 147–50. Barckhausen U, Renaro C, von Huene R, Cande SC, Roeser HA (2001) Revised tectonic boundaries in the Cocos Plate off Costa Rica: implications for the segmentation of the convergent margin and for plate tectonic models. Journal of Geophysical Research, 106, 19. Barker P, Lawver L (2000) Anomalous temperatures in central Scotia Sea sediments – bottom water variation or pore water circulation in old crust. Geophysical Research Letters, 27, 13–6. Bartetzko A, Pezard P, Goldberg D, Sun Y-F, Becker K (2001) Volcanic stratigraphy of DSDP ⁄ ODP Hole 395A: an interpretation using well-logging data. Marine Geophysical Researches, 22, 111–27. Becker K, Davis EE (2005) A review of CORK designs and operations during the Ocean Drilling Program. In: Proc. IODP, Expedition Reports (eds Fisher AT, Urabe T, Klaus A) College Station, TX, Integrated Ocean Drilling Program, 301, doi:10.2204/iodp.proc.2301.2104. Becker K, Langseth M, Von Herzen RP, Anderson R (1983) Deep crustal geothermal measurements, Hole 504B, Costa Rica Rift. Journal of Geophysical Research, 88, 3447–57. Becker K, Morin RH, Davis EE (1994) Permeabilities in the Middle Valley hydrothermal system measured with packer and flowmeter experiments. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 613–26. Becker NC, Wheat CG, Mottl MJ, Karsten JL, Davis EE (2000) A geological and geophysical investigation of Baby Bare, locus of a ridge flank hydrothermal system in the Cascadia Basin. Journal of Geophysical Research, 105, 23, 557–568. Bredehoeft JD, Papadopulos IS (1965) Rates of vertical groundwater movement estimated from the earth’s thermal profile. Water Resources Research, 1, 325–8.
Bullard EC (1939) Heat flow in South Africa. Proceedings of the Royal Society of London, Series A., 173, 474–502. Bullard EC (1954) The flow of heat through the floor of the Atlantic Ocean. Proceedings of the Royal Society of London, Series A, 222, 408–29. Bullard EC, Maxwell AE (1956) Heat flow through the deep sea floor. Advances in Geophysics, 222, 408–29. Burow KR, Constantz J, Fujii R (2005) Heat as a tracer to estimate dissolved organic carbon flux from a restored wetland. Ground Water, 43, 545–56. Butterfield DA, McDuff RE, Franklin J, Wheat CG (1994) Geochemistry of hydrothermal vent fluids from Middle Valley, Juan de Fuca Ridge. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 395–410. Constantz J, Stonestrom D, Stewart AE, Niswonger R, Smith TR (2001) Analysis of streambed temperatures in ephemeral channels to determine streamflow frequency and duration. Water Resources Research, 37, 317–28. Davis EE (1988) Oceanic heat-flow density. In: Handbook of Terrestrial Heat-Flow Density Determination (eds Haenel R, Rybach L, Stegena L), pp. 223–60. Kluwer, Amsterdam. Davis EE, Becker K (1994) Formation temperatures and pressures in a sedimented rift hydrothermal system: 10 months of CORK observations, Holes 857D and 858G. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 649–66. Davis EE, Becker K (2002) Observations of natural-state fluid pressures and temperatures in young oceanic crust and inferences regarding hydrothermal circulation. Earth and Planetary Science Letters, 204, 231–48. Davis EE, Fisher AT (1994) On the nature and consequences of hydrothermal circulation in the Middle Valley sedimented rift: inferences from geophysical and geochemical observations, Leg 139. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 695–717. Davis EE, Lister CRB (1974) Fundamentals of ridge crest topography. Earth and Planetary Science Letters, 21, 405–13. Davis EE, Villinger H (1992) Tectonic and thermal structure of the Middle Valley sedimented rift, northern Juan de Fuca Ridge. In: Proc. ODP, Init. Repts. (eds Davis EE, Mottl M, Fisher AT) College Station, TX, Ocean Drilling Program, 139, 9–41. Davis EE, Wang K (1994) Present and past temperatures of sediments at Site 857, Middle Valley, northern Juan de Fuca Ridge. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 565–72. Davis EE, Chapman DS, Forster C, Villinger H (1989) Heat-flow variations correlated with buried basement topography on the Juan de Fuca Ridge flank. Nature, 342, 533–7. Davis EE, Chapman DS, Mottl MJ, Bentkowski WJ, Dadey K, Forster C, Harris R, Nagihara S, Rohr K, Wheat G, Whiticar M (1992) FlankFlux: an experiment to study the nature of hydrothermal circulation in young oceanic crust. Canadian Journal of Earth Sciences, 29, 925–52. Davis EE, Villinger H, Macdonald RD, Meldrum RD, Grigel J (1997a) A robust rapid-response probe for measuring bottomhole temperatures in deep-ocean boreholes. Marine Geophysical Researches, 19, 267–81. Davis EE, Wang K, He J, Chapman DS, Villinger H, Rosenberger A (1997b) An unequivocal case for high Nusselt-number hydrothermal convection in sediment-buried igneous oceanic crust. Earth and Planetary Science Letters, 146, 137–50.
158 A. T. FISHER & R. N. HARRIS Davis EE, Chapman DS, Wang K, Villinger H, Fisher AT, Robinson SW, Grigel J, Pribnow D, Stein J, Becker K (1999) Regional heat-flow variations across the sedimented Juan de Fuca Ridge eastern flank: constraints on lithospheric cooling and lateral hydrothermal heat transport. Journal of Geophysical Research, 104, 17, 675–88. Davis EE, Wang K, Becker K, Thomson RE, Yashayaev I (2003) Deep-ocean temperature variations and implications for errors in seafloor heat flow determinations. Journal of Geophysical Research, 108, 2034. doi:2110.1029/2001JB001695. Elder JW (1965) Physical processes in geothermal areas. In: Terrestrial Heat Flow (ed Lee WHK) Washington, DC, American Geophysics Union, 8, 211–39. Elderfield H, Wheat CG, Mottl MJ, Monnin C, Spiro B (1999) Fluid and geochemical transport through oceanic crust: a transect across the eastern flank of the Juan de Fuca Ridge. Earth and Planetary Science Letters, 172, 151–65. Fisher AT, Becker K (1993) A guide for ODP tools for downhole measurements, Technical note 10, doi: 10.2973/odp.tn.10. 1993. Available at: http://www-odp.tamu.edu/publications/ tnotes/tn10/10toc.html (accessed 31 December 2009). Fisher AT, Becker K (2000) Channelized fluid flow in oceanic crust reconciles heat-flow and permeability data. Nature, 403, 71–4. Fisher AT, Becker K, Narasimhan TN, Langseth MG, Mottl MJ (1990) Passive, off-axis convection on the southern flank of the Costa Rica Rift. Journal of Geophysical Research, 95, 9343–70. Fisher AT, Fischer K, Lavoie D, Langseth M, Xu J (1994) Hydrogeological and geotechnical properties of shallow sediments at Middle Valley, northern Juan de Fuca Ridge. In: Proc. ODP, Sci. Res. (eds Mottl MJ, Davis EE, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 627–48. Fisher AT, Becker K, Davis EE (1997) The permeability of young oceanic crust east of Juan de Fuca Ridge determined using borehole thermal measurements. Geophysical Research Letters, 24, 1311–4. Fisher AT, Davis EE, Hutnak M, Spiess V, Zu¨hlsdorff L, Cherkaoui A, Christiansen L, Edwards KM, Macdonald R, Villinger H, Mottl MJ, Wheat CG, Becker K (2003a) Hydrothermal recharge and discharge across 50 km guided by seamounts on a young ridge flank. Nature, 421, 618–21. Fisher AT, Stein CA, Harris RN, Wang K, Silver EA, Pfender M, Hutnak M, Cherkaoui A, Bodzin R, Villinger H (2003b) Abrupt thermal transition reveals hydrothermal boundary and role of seamounts within the Cocos Plate. Geophysical Research Letters, 30, 1550, doi:1510.1029/2002GL016766. Fisher AT, Villinger H, Heesemann M (2007) User Manual for the Third-Generation, Advanced Piston Corer Temperature tool (APCT-3). Integrated Ocean Drilling Program, College Station, TX. Fisher AT, Davis EE, Becker K (2008) Borehole-to-borehole hydrologic response across 2.4 km in the upper oceanic crust: implications for crustal-scale properties. Journal of Geophysical Research, 113, B07106, doi:10.1029/2007JB005447. Foucher J-P, Henry P, Harmegnies F (1997) Long-term observations of pressure and temperature in Hole 948D, Barbados accretionary prism. In: Proc. ODP, Sci. Res. (eds Ogawa Y, Shipley T, Blum P, Bahr J) College Station, TX, Ocean Drilling Program, 156, 239–46. Fukasawa M, Freeland H, Perkin R, Wantanabe T, Uchida H, Nishina A (2003) Bottom water warming in the North Pacific Ocean. Nature, 427, 825–7. Gieskes J, Kastner M, Einsele G, Kelts K, Niemitz J (1982) Hydrothermal activity in the Guayman Basin, Gulf of California,
a synthesis. In: Init. Repts., DSDP (eds Curray J, Moore D), U. S. Govt. Printing Office, Washington, D. C., 64, 1159–68. Goto S, Yamano M, Kinoshita H (2005) Thermal response of sediment with vertical fluid flow to periodic temperature variation at the surface. Journal of Geophysical Research, 110, B01106, doi:10.1029/2004JB003419. Hamamoto H, Yamano M, Goto S (2005) Heat flow measurement in shallow seas through long-term temperature monitoring. Geophysical Research Letters, 32, L21311. Harris RN, Chapman DS (2004) Deep-seated oceanic heat flow, heat deficits, and hydrothermal circulation. In: Hydrogeology of the Oceanic Lithosphere (eds Davis EE, Elderfield H), pp. 311– 36. Cambridge University Press, Cambridge, UK. Harris RN, Von Herzen RP, McNutt MK, Garven G, Jordahl K (2000) Submarine hydrogeology of the Hawaiian archipelagic apron, Part 1, Heat flow patterns north of Oahu and Maro Reef. Journal of Geophysical Research, 105, 21, 353–69. Hartmann A, Villinger H (2002) Inversion of marine heat flow measurements by expansion of the temperature decay function. Geophysical Journal International, 148, 628–36. Horai K, Von Herzen RP (1985) Measurement of heat flow on Leg 86 of the Deep Sea Drilling Project. In: Init. Repts., DSDP (eds Heath GR, Burckle LH), U. S. Govt. Printing Office, Washington, D. C., 86, 759–77. von Huene R, Ranero CR, Wienrebe W, Hinz K (2000) Quarternary convergent margin tectonics of Costa Rica, segmentation of the Cocos Plate, and Central American Volcanism. Tectonics, 19, 314–34. Hutnak M, Fisher AT (2007) The influence of sedimentation, local and regional hydrothermal circulation, and thermal rebound on measurements of heat flux from young seafloor. Journal of Geophysical Research, 112, B12101, doi:10.1029/ 2007JB005022. Hutnak M, Fisher AT, Zu¨hlsdorff L, Spiess V, Stauffer P, Gable CW (2006) Hydrothermal recharge and discharge guided by basement outcrops on 0.7-3.6 Ma seafloor east of the Juan de Fuca Ridge: observations and numerical models. Geochemistry, Geophysics, and Geosystems, 7, Q07O02, doi:10.1029/2006GC 001242. Hutnak M, Fisher AT, Stein CA, Harris R, Wang K, Silver E, Spinelli G, Pfender M, Villinger H, MacKnight R, Costa Pisani P, DeShon H, Diamente C (2007) The thermal state of 18-24 Ma upper lithosphere subducting below the Nicoya Peninsula, northern Costa Rica margin. In: MARGINS Theoretical Institute: SIEZE Volume (eds Dixon T, Moore C, Silver E, Stein S, Furlong K, Brown K), pp. 86–122. Columbia University Press, New York. Hutnak M, Fisher AT, Harris R, Stein C, Wang K, Spinelli G, Schindler M, Villinger H, Silver E (2008) Large heat and fluid fluxes driven through mid-plate outcrops on ocean crust. Nature Geoscience, 1, 611–4. Hyndman RD, Von Herzen RP, Erickson AJ, Jolivet J (1976) Heat flow measurements in deep crustal holes on the MidAtlantic Ridge. Journal of Geophysical Research, 81, 4053– 60. Hyndman RD, Davis EE, Wright JA (1979) The measurement of marine geothermal heat flow by a multipenetration probe with digital acoustic telemetry and in situ conductivity. Marine Geophysical Researches, 4, 181–205. Johnson HP, Hutnak M (1996) Conductive heat flow measured in unsedimented regions of the seafloor. Transactions, American Geophysical Union, 77, 321–4. Kastner M (1982) Evidence for two distinct hydrothermal systems in the Gulf of California. In: Init. Repts., DSDP (eds Curray J,
Using seafloor heat flow to map subseafloor fluid flow 159 Moore D) U. S. Govt. Printing Office, Washington, D. C., 64, 529–42. Kinoshita M, Matsubayashi O, Von Herzen RP (1996) Sub-bottom temperatrue anomalies detected by long-term temperatrue monitoring at the TAG hydrothermal mound. Geophysical Research Letters, 23, 3467–70. Langseth M, Becker K (1994) Structure of igneous basement at Sites 857 and 858 based on Leg 139 downhole logging. In: Proc. ODP, Sci. Res. (eds Davis EE, Mottl MJ, Fisher AT, Slack JF) College Station, TX, Ocean Drilling Program, 139, 573– 83. Langseth MG, Herman B (1981) Heat transfer in the oceanic crust of the Brazil Basin. Journal of Geophysical Research, 86, 10805–19. Langseth MG, Silver EA (1996) The Nicoya convergent margin – a region of exceptionally low heat flow. Geophysical Research Letters, 23, 891–4. Langseth MG, Grim PJ, Ewing M (1965) Heat-flow measurements in the east Pacific Ocean. Journal of Geophysical Research, 70, 367–80. Langseth MG, LePichon X, Ewing M (1966) Crustal structure of the midocean ridges, 5, heat flow through the Atlantic Ocean floor and convection currents. Journal of Geophysical Research, 71, 5321–55. Larson RL, Fisher AT, Jarrard R, Becker K (1993) Highly permeable and layered Jurassic oceanic crust in the western Pacific. Earth and Planetary Science Letters, 119, 71–83. Lister CRB (1972) On the thermal balance of a mid-ocean ridge. Geophysical Journal of the Royal Astronomical Society, 26, 515– 35. Lonsdale P, Becker K (1985) Hydrothermal plumes, hot springs, and conductive heat flow in the southern trough of Guaymas Basin. Earth and Planetary Science Letters, 73, 211–25. Louden KE, Wright JA (1989) Marine heat flow data: a new compilation of observations and brief review of its analysis. In: Handbook of Seafloor Heat Flow (eds Wright JA, Louden KE), pp. 3–67. CRC Press, Boca Raton, FL. Lu N, Ge S (1996) Effect of horizontal heat and fluid flow on the vertical temperature distribution in a semiconfining layer. Water Resources Research, 32, 1449–53. McKenzie D (1967) Some remarks on heat flow and gravity anomalies. Journal of Geophysical Research, 72, 6261–73. McNutt MK (1995) Marine geodynamics: depth-age revisited. Reviews of Geophysics, 33, 413–8. Meschede M, Barckhausen U, Worm H-U (1998) Extinct spreading on the Cocos Ridge. Terra Nova, 10, 211–6. Mottl M (2003) Partitioning of energy and mass fluxes between mid-ocean ridge axes and flanks at high and low temperature. In: Energy and Mass Transfer in Submarine Hydrothermal Systems (eds Halbach P, Tunnicliffe V, Hein J), pp. 271–86. University Press, Berlin, Germany, Dahlem, DWR 89. Mottl MJ, Wheat CG, Baker E, Becker N, Davis E, Feeley R, Grehan A, Kadko D, Lilley M, Massoth G, Moyer C, Sansone F (1998) Warm springs discovered on 3.5 Ma oceanic crust, eastern flank of the Juan de Fuca Ridge. Geology, 26, 51–4. Parker RL, Oldenberg DW (1973) Thermal model of ocean ridges. Nature, 242, 137–9. Parsons B, Sclater JG (1977) An analysis of the variation of ocean floor bathymetry and heat flow with age. Journal of Geophysical Research, 82, 803–29. Pfender M, Villinger H (2002) Miniaturized data loggers for deep sea sediment temperature gradient measurements. Marine Geology, 186, 557–70.
Pribnow DFC, Davis EE, Fisher AT (2000) Borehole heat flow along the eastern flank of the Juan de Fuca Ridge, including effects of anisotropy and temperature dependence of sediment thermal conductivity. Journal of Geophysical Research, 105, 13449–56. Ranero CR, von Huene R (2000) Subduction erosion along the Middle America convergent margin. Nature, 404, 748–52. Revelle R, Maxwell AE (1952) Heat flow through the ocean floor. Nature, 170, 199–200. Rosenberg N, Fisher AT, Stein J (2000) Large-scale lateral heat and fluid transport in the seafloor: revisiting the well-mixed aquifer model. Earth and Planetary Science Letters, 182, 93– 101. Sclater JG (2004) Variability of heat flux through the seafloor: discovery of hydrothermal circulation in oceanic crust. In: Hydrogeology of the Oceanic Lithosphere (eds Davis EE, Elderfield H), pp. 3–27. Cambridge University Press, Cambridge, UK. Sclater JG, Jaupart C, Galson D (1980) The heat flow through oceanic and continental crust and the heat loss of the earth. Reviews of Geophysics and Space Physics, 18, 269–311. Spinelli GA, Giambalvo EG, Fisher AT (2004) Hydrologic properties and distribution of sediments. In: Hydrogeology of the Oceanic Lithosphere (eds Davis EE, Elderfield H), pp. 151–88. Cambridge University Press, Cambridge, UK. Stein JS, Fisher AT (2001) Multiple scales of hydrothermal circulation in Middle Valley, northern Juan de Fuca Ridge: physical constraints and geologic models. Journal of Geophysical Research, 106, 8563–80. Stein CA, Stein S (1992) A model for the global variation in oceanic depth and heat flow with lithospheric age. Nature, 359, 123–37. Stein C, Stein S (1994) Constraints on hydrothermal heat flux through the oceanic lithosphere from global heat flow. Journal of Geophysical Research, 99, 3081–95. Stein CA, Stein S, Pelayo AM (1995) Heat flow and hydrothermal circulation. In: Seafloor Hydrothermal Systems: Physical, Chemical, Biological and Geological Interactions (eds Humphris SE, Zierenberg RA, Mullineaux LS, Thompson RE) American Geophysical Union, Washington, D. C., 91, 425–45. Stein J, Fisher AT, Langseth M, Jin W, Iturrino G, Davis E (1998) Fine-scale heat flow, shallow heat sources, and decoupled circulation systems at two seafloor hydrothermal sites, Middle Valley, northern Juan de Fuca Ridge. Geology, 26, 1115–1118. Thomson RE, Davis EE, Burd BJ (1995) Hydrothermal venting and geothermal heating in Cascadia Basin. Journal of Geophysical Research, 100, 6121–41. Uyeda S, Horai K (1980) Heat flow measurements on Deep Sea Drilling Project Leg 60. In: Init. Repts., DSDP (eds Hussong D, Uyeda S), U. S. Govt. Printing Office, Washington, D. C., 60, 789–800. Vacquier V, Sclater JG, Corry CE (1967) Studies in the thermal state of the earth, the 21st paper: heat-flow, Eastern Pacific. Bulletin of the Earthquake Research Institute, 45, 375–93. Veirs SR, McDuff RE, Lilley MD, Delaney JR (1999) Locating hydrothermal vents by detecting buoyant, advected plumes. Journal of Geophysical Research, 104, 29. Villinger H, Davis EE (1987) A new reduction algorithm for marine heat-flow measurements. Journal of Geophysical Research, 92, 12846–56. Villinger H, Grevemeyer I, Kaul N, Hauschild J, Pfender M (2002) Hydrothermal heat flux through aged oceanic crust: where does the heat escape? Earth and Planetary Science Letters, 202, 159–70.
160 A. T. FISHER & R. N. HARRIS Von Herzen RP (2004) Geothermal evidence for continuing hydrothermal circulation in older (>60 Ma) ocean crust. In: Hydrogeology of the Oceanic Lithosphere (eds Davis EE, Elderfield H), pp. 414–50. Cambridge University Press, Cambridge, UK. Von Herzen RP, Maxwell AE (1959) The measurement of thermal conductivity of deep-sea sediments by a needle probe method. Journal of Geophysical Research, 64, 1557–63. Von Herzen R, Uyeda S (1963) Heat flow through the eastern Pacific floor. Journal of Geophysical Research, 68, 4219–50. Von Herzen R, Davis EE, Fisher AT, Stein CA, Pollack HN (2005) Comments on ‘‘Earth’s heat flux revised and linked to chemistry’’ by A. M. Hoffmeister and R. E. Criss. Tectonophysics, 409, 193–8. Wang K, Davis EE (1992) Thermal effects of marine sedimentation in hydrothermally active areas. Geophysical Journal International, 110, 70–8. Wheat CG, Fisher AT (2007) Seawater recharge along an eastern bounding fault in Middle Valley, northern Juan de Fuca Ridge. Geophysical Research Letters, 34, L20602, doi:10.1029/ 2007GL031347. Wheat CG, Mottl MJ (1994) Hydrothermal circulation, Juan de Fuca Ridge eastern flank: factors controlling basement water composition. Journal of Geophysical Research, 99, 3067–80. Wheat CG, Mottl M (2000) Composition of pore and spring waters from Baby Bare: global implications of geochemical
fluxes from a ridge flank hydrothermal system. Geochimica Cosmochimica Acta, 64, 629–42. Wheat CG, Elderfield H, Mottl MJ, Monnin C (2000) Chemical composition of basement fluids within an oceanic ridge flank: implications for along-strike and across-strike hydrothermal circulation. Journal of Geophysical Research, 105, 13437– 47. Wheat CG, Mottl MJ, Fisher AT, Kadko D, Davis EE, Baker E (2004) Heat and fluid flow through a basaltic outcrop on a ridge flank. Geochemistry, Geophysics and. Geosystems, 5, Q12006, doi:10.1029/2004GC000700. Wilson DS (1993) Confidence intervals for motion and deformation of the Juan de Fuca plate. Journal of Geophysical Research, 98, 16. Wolery TJ, Sleep NH (1975) Hydrothermal reaction at mid-ocean ridges: some implications and constraints. EOS, Transactions, American Geophysical Union, 56, 1073. Zu¨hlsdorff L, Hutnak M, Fisher AT, Spiess V, Davis EE, Nedimovic M, Carbotte S, Villinger H, Becker K (2005) Site Surveys related to IODP Expedition 301: ImageFlux (SO149) and RetroFlux (TN116) expeditions and earlier studies. In: Proc. IODP (eds Fisher AT, Urabe T, Klaus A) College Station, TX, Integrated Ocean Drilling Program, Expedition 301, doi:10.2204/ iodp.proc.2301.2102.2005.
The potential for abiotic organic synthesis and biosynthesis at seafloor hydrothermal systems E. SHOCK1,2 AND P. CANOVAS1 1
GEOPIG, School of Earth & Space Exploration and 2Department of Chemistry & Biochemistry, Arizona State University, Tempe, AZ, USA
ABSTRACT Calculations are presented of the extent to which chemical disequilibria are generated when submarine hydrothermal fluids mix with sea water. These calculations involve quantifying the chemical affinity for individual reactions by comparing equilibrium compositions with the compositions of mixtures in which oxidation–reduction reactions are inhibited. The oxidation–reduction reactions that depart from equilibrium in these systems provide energy for chemotrophic microbial metabolism. Methanogenesis is an example of this phenomenon, in which the combination of carbon dioxide, hydrogen and methane induced by fluid mixing is far from equilibrium, which can be approached if more methane is generated. Similarly, the production of other organic compounds is also favorable under the same conditions that permit methanogenesis. Alkanes, alkenes, alcohols, aldehydes, carboxylic acids and amino acids are among the compounds that, if formed, would lower the energetic state of the chemical composition generated in mixed fluids. The resulting positive values of chemical affinity correspond to the thermodynamic drive required for abiotic organic synthesis. It is also possible that energy release accompanies biosynthesis by chemotrophic organisms. In this way, hydrothermal ecosystems differ radically from familiar ecosystems at Earth’s surface. If captured, the energy released may be sufficient to drive biosynthesis of carbohydrates, purines, pyrimidines and other compounds which require energy inputs. Key words: abiotic organic synthesis, affinity, amino acids, biosynthesis, carbohydrates, hydrothermal, organic acids, seafloor hydrothermal systems Received 12 October 2009; accepted 4 January 2010 Corresponding author: Everett Shock, GEOPIG, School of Earth & Space Exploration, Arizona State University, Tempe, AZ 85287, USA. Email:
[email protected]. Tel: +1-480-965-0631. Fax: +1-480-965-8102. Geofluids (2010) 10, 161–192
INTRODUCTION Twelve years ago, evidence supporting the ideas that abiotic synthesis of organic compounds could occur in hydrothermal processes, and that biosynthesis of organic compounds benefited from hydrothermal conditions, came from theoretical calculations of the geochemical and thermodynamic consequences of mixing a submarine hydrothermal fluid with sea water (McCollom & Shock 1997; Amend & Shock 1998; Shock & Schulte 1998). These studies showed that mixing produced states of disequilibrium in which metabolic and organic synthesis reactions became thermodynamically favored. As an example, it was shown quantitatively that reductive metabolisms, such as methanogenesis and sulfate reduction, would release energy at higher temperatures, and that oxidative metabo-
lisms, such as methanotrophy and sulfide oxidation, would release energy at lower temperatures (McCollom & Shock 1997). It was also shown that present conditions where marine hyperthermophiles thrive are far more conducive to synthesis of all 20 protein-forming amino acids than is synthesis in surface sea water, and that overall synthesis reactions are exergonic (energy releasing) for 11 of them (Amend & Shock 1998). Exergonic biosynthesis reactions are not at all what humans experience at the Earth’s surface where formation of any organic compound in a 20% oxygen atmosphere requires input of energy. Exploration of abiotic organic synthesis during mixing of hydrothermal fluids and sea water on the early Earth and Mars, with sea water compositions far less oxidized than at present, revealed thermodynamic drives for the formation of many classes of organic compounds (Shock & Schulte 1998),
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
162 E. SHOCK & P. CANOVAS and set the stage for further speculations about the emergence of life in hydrothermal systems (Shock 1997; Shock et al. 1998; Russell & Martin 2004; Russell et al. 2005; Martin & Russell 2007). These advances were driven by a single model of the mixing of sea water and a submarine hydrothermal fluid composition based on measurements made on fluids from 21N on the East Pacific Rise (EPR). Over the intervening years, several lines of evidence from experiments and observations of nature show that abiotic organic synthesis is indeed possible at hydrothermal conditions. At the same time, the ability to consider biosynthesis in more detail through theoretical mixing calculations has improved owing to the expansion of the number and variety of biomolecules that can be included. These developments provided the impetus for the present study, which involved a comparison of seven submarine hydrothermal fluids from all of the major recognized biogeographic zones. Recent observations of natural submarine hydrothermal systems show the presence of dissolved organic compounds that may be the products of abiotic organic synthesis. As an example, Proskurowski et al. (2008) provide concentration, stable isotopic and radiocarbon evidence that C1–C4 hydrocarbons have an abiotic source in the fluids venting from Lost City at 30N on the Mid-Atlantic Ridge. The Lost City hydrothermal field occurs in ultramafic rocks, and is characterized by highly basic fluids (pH 9–11) emanating at relatively low temperatures (28–90C). These authors prefer a model of light hydrocarbon generation through a Fischer–Tropsch-type synthesis (see McCollom & Seewald 2007) involving mineral surfaces to catalyze the reduction of CO2(aq) at H2(aq)-enriched conditions. The dissolved hydrogen is a product of the reduction of H2O as ferrous silicates are oxidized to yield magnetite (Fe3O4). Overall, the process involves coupled oxidation of ironbearing minerals and reduction of H2O, accompanied or followed by reduction of CO2(aq), argued to be mantle derived (Proskurowski et al. 2008), to methane and other light hydrocarbons catalyzed by mineral surfaces. Analogous processes are invoked to explain the synthesis of hydrocarbons, carboxylic acids and other organic compounds in higher temperature seafloor systems hosted in ultramafic rocks undergoing serpentinization (Charlou et al. 2002; Konn et al. 2009). Experimental studies have also demonstrated the potential for organic synthesis at conditions meant to mimic those in natural hydrothermal systems. Hydrocarbons were the focus of a study by Foustoukos & Seyfried (2004) who conducted Fischer–Tropsch-type experiments at 390C and 400 bars. Experiments that included the simultaneous formation of the mineral chromite (FeCr2O4) yielded the highest concentrations of methane, ethane and propane, suggesting a catalytic role for chromium. Synthesis was confirmed by using 13C-labeled NaHCO3 as the carbon
source. These experiments were designed to test the idea that abiotic hydrocarbon synthesis happens at high-temperature ultramafic-hosted seafloor systems like those at Rainbow and Logatchev on the Mid-Atlantic Ridge. Nevertheless, there could be similarities in the overall abiotic synthesis processes happening in cooler systems like Lost City, especially if conductive cooling is as extensive as some recent models predict (Foustoukos et al. 2008). While progress was being made in hydrothermal abiotic organic synthesis through experiments and analysis of natural samples, several breakthroughs allowed the expansion of thermodynamic data for aqueous organic compounds at hydrothermal conditions. All of these developments were built on the revised Helgeson–Kirkham–Flowers equation of state for aqueous species (Shock et al. 1992). Data and parameters for carbohydrates were added by Amend & Plyasunov (2001), allowing the first thermodynamic analysis of metabolism of sugars at elevated temperatures and pressures. Group contribution algorithms and other estimation methods made it possible to revise data for amino acids and to make predictions for polypeptides and unfolded proteins (Dick et al. 2006), and for purines, pyrimidines, nucleotides, nucleosides and many other biomolecules involved in energy and electron transfer and the formation of DNA and RNA (LaRowe & Helgeson 2006a,b). These advances make it possible to consider biosynthesis of organic compounds by microbes in hydrothermal systems, together with abiotic organic synthesis that may be occurring in the same systems. As several lines of evidence have advanced the notion that abiotic synthesis of organic compounds can occur in hydrothermal systems, it seems timely to investigate the energetic state of overall organic synthesis reactions that may be occurring, so that greater detail about pathways of abiotic organic synthesis might be suggested. At the same time, identifying novel conditions that are energetically favorable for abiotic organic synthesis can help in predicting new biosynthetic couplings that lower overall energy costs. The appearance of new data and parameters for biomolecules presents an opportunity to begin exploring the overall energy requirements of biosynthesis in thermophilic microbes that populate mixing zones of submarine hydrothermal systems.
DIVERSITY OF SUBMARINE HYDROTHERMAL FLUIDS Hydrothermal systems are known to occur on ridges and in back-arc basins regardless of spreading or convergence rates. Fluids venting from diverse tectonic settings vary in composition in large part because of differences in the composition of the rocks that host the hydrothermal systems. Early discoveries along the EPR and the Mid-Atlantic Ridge revealed hydrothermal systems hosted in basalt,
Seafloor hydrothermal systems 163 and resulting fluids are sometimes considered to be typical of the hydrothermal input to the oceans. Basalt-hosted fluids are relatively reduced, mildly acidic, sulfidic and enriched in metals. Subsequently, it was learned that submarine hydrothermal systems can be hosted in ultramafic rocks, leading to fluids that are more reduced than their basaltic counterparts (Wetzel & Shock 2000; Wetzel et al. 2001). Fluids can also be more acidic than from basalt-hosted systems, notably those from andesitic rocks of back-arc spreading centers. Sediments are present in some fluid-recharge zones and in some cases cover hydrothermally active ridges. Reactions involving sediments and fluids can yield greater concentrations of carbon dioxide and ammonia than unsedimented basalt-hosted fluids.
Compositions of fluids are listed in Table 1 from several seafloor settings including mid-ocean ridges and back-arc basins from the locations shown in Fig. 1. These fluids were chosen to represent a large proportion of the diversity encountered in submarine hydrothermal systems. They also are from each of the major recognized biogeographic zones of seafloor hydrothermal ecosystems. While our understanding of biogeographic zones is far from complete, studies of the evolutionary history of vent and seep invertebrates has allowed the identification of six different biogeographic zones (Van Dover et al. 2002). Many of these tend to correlate well with ocean basins and degrees of isolation along the mid-ocean ridge system (Tunnicliffe & Fowler 1996). Oceanic circulation patterns
Table 1 Compositions of hydrothermal vent fluids and sea water used in model calculations.
System type Host rock Sediments T (C) pH (25C) O2(aq) H2(aq) P H2S(aq) CH4(aq) P CO2 P 2 SO P 43 PO4 NO 3 NHþ 4 SiO2(aq) + Na Cl) Br) K+ Al3+ Ca2+ Mg2+ Fe3+ Fe2+ Ba2+ Mn2+ Cu2+ Zn2+ Pb2+
Rainbow
Guaymas
Endeavour
Lau
9N
TAG
Kairei
BSW
MAR
Basin
Juan de Fuca
Basin
EPR
MAR
CIR
– – 2 7.8 0.1 – – 0.0000003 2.3 27.9 0.00075 0.0003 – 0.16 464 546 0.84 9.8 0.00002 10.2 52.7 0.0000015 – 0.00014 0 0.000007 0.00001 0
MOR Ultramafic No 350 2.8 – 16 1.2 2.5 16 0 – – 0.1 6.9 533 750 1.18 20.4 – 66.6 0 – 24.1 0.008 2.25 0.16 0.185 –
MOR Basalt Yes 315 5.9 – 3.4 59.8 63.4 61.1 0 – – 15.6 13.8 513 637 5.3 48.5 0.0079 41.5 0 – 0.18 0.053 0.236 0.0011 0.04 0.000652
MOR Basalt May be 345 4.5 – 0.62 20 3.4 22 – – – 1.0 17 319 235 0.895 29 0.0019 4.3 0 – 1.36 0.005 0.55 0.021 0.035 0.00051
BAB Basalt ⁄ ande-site No 334 2. – 0.5 13.1 0.06 20 – – – 5.0 14.5 590 489 1.14 79 0.006 41.3 0 – 2.5 0.04 7.1 0.034 3 0.0039
MOR Basalt No 350 2.6 – 1.8 8.71 0.01 11.4 – – – 0.01 20 683 846 1.35 41.5 0.0052 45.6 0 – 12.1 0.008 3.28 0.035 0.106 0.000308
MOR Basalt No 350 3.1 – 0.37 6.7 0.147 3.4 – – – 0.65 22 584 659 1.045 18 0.01 26 0 – 5.17 0.02 0.71 0.13 0.083 0.00011
MOR Basalt ⁄ gab-bro? No 349 3.48 – 8.19 3.97 0.191 5.13 1.56 0.00139 – 0.0388 16.7 511 595 0.95 14.5 0.00528 30.2 0 – 4.6 0.0425 0.829 0.245 0.0895 0.00047
All concentrations in mmolal. Bottom Seawater (BSW), Mid-Atlantic Ridge (MAR), East Pacific Rise (EPR), Central Indian Ridge (CIR), mid-ocean ridge (MOR), back-arc basin (BAB). BSW composition from Bruland (1983) and McCollom (2007, 2008). Rainbow data from Charlou et al. (2002). Guaymas data from Von Damm et al. (1985b, 2005) and Campbell et al. (1988b), and from compilation by Gamo (1995) where original sources are: Piepgras & Wasserburg (1985), Lupton (1983), Campbell et al. (1988b), Welhan & Lupton (1987) and Gieskes et al. (1988). Endeavour vent fluid from Lilley et al. (1993) and Butterfield et al. (1994) in Gamo (1995) and Seewald et al. (2003); Al3+, Ba2+ and Pb2+ from Juan de Fuca Ridge S. Cleft Segment, used in the absence of readily available data for Endeavour, are from Hinkley & Tatsumoto (1987), Trefrey et al. (1994), Philpotts et al. (1987), Von Damm & Bischoff (1987) and P P Evans et al. (1988). Lau Basin vent fluid composition from Fouquet et al. (1991a,b, 1993), and Charlou et al. (1991); methane, CO2, H2(aq), H2S from maximum values presented in Tivey (2007) for back-arc basins; Al3+ from White Lady vent in north Fiji back-arc basin (Ishibashi et al. 1994a,b). EPR 9N compositions from Von Damm et al. (1991, 1997), Shanks et al. (1991), Lilley et al. (1991) and Lupton et al. (1991); Al3+, Ba2+, Zn2+, Pb2+, NH4+, Cu2+ concentrations from EPR 21N OBS (Von Damm et al. 1985a) used in the absence of readily available data. TAG data from Campbell et al. (1988a), Douville et al. (2002) and Charlou et al. (1996, 2002); NH4+ from maximum values presented in Tivey (2007) for mid-ocean ridges. Kairei data from Gallant & Von Damm (2006) and Kumagai et al. (2008).
164 E. SHOCK & P. CANOVAS
Endeavor
Rainbow Guaymas Basin
TAG
9oN EPR
Lau Basin
Kairei
Fig. 1. Geologic and tectonic setting of seafloor hydrothermal systems modeled in this investigation. Color coding is maintined in subsequest figures. Data from Baker et al. (1995), Van Dover et al. (2002), German & Von Damm (2004), Hannington et al. (2005), Koschinsky et al. (2006), Tivey (2007) and German et al. (2008).
and Cenozoic tectonic history were important in defining biogeographic zones and patterns that are recognized today (Van Dover et al. 2002). An example of this is evident in the differentiation exhibited by communities on the EPR that are separated by many hundreds of kilometers from those of the north-east Pacific ridge system. These communities have been diverging since the formerly continuous ridge system was split 28 Ma by the overriding North American Plate (Tunnicliffe 1988). By contrast, on the Mid-Atlantic Ridge there are differences among communities that are relatively close to one another. It is argued that emplacement of the Azores plateau 20 Ma may have served to isolate biological communities at hydrothermal systems to the north and south (Cannat et al. 1999; Van Dover et al. 2002). East Pacific Rise (9N) The EPR exhibits fast spreading with a rate of 11 cm year)1 (Klitgord & Mammerickx 1982; Carbotte & Macdonald 1992). High-temperature hydrothermal vent systems at EPR were first discovered in 1979 (Macdonald et al. 1980; Spiess 1980), and since then the area along 9–11N has become a major focus site for research including towed camera surveys, use of Argo I, ocean bottom seismometers and over 300 Alvin dives, the first of which occurred in March ⁄ April 1991 as part of an extensive fluid sampling program (Fornari et al. 1990, 1998a,b; Haymon
et al. 1991, 1993; Lilley et al. 1991; Lupton et al. 1991; Shanks et al. 1991; Von Damm et al. 1991, 1997; Wright et al. 1995; Gregg et al. 1996; Sohn et al. 1998; Kurras et al. 2000; Engels et al. 2003; Von Damm 2004). The 9N vent field lies at a depth of 2500 m, and evidence from multi-channel seismic data shows the presence of a thin crustal magma chamber 1.5 km beneath the ridge axis (Detrick et al. 1987). The axial summit trough is approximately 60 m across and is the site of active hydrothermal venting with the entire system experiencing extensive volcanism (Fornari et al. 2004). Seafloor eruptive events occurred in 1991 and 1992 (Haymon et al. 1993; Shank et al. 1998), as well as in the months preceding April 2006 which elicited a series of response cruises sponsored by the Ridge 2000 program to verify the recent activity (Lilley et al. 2006; Von Damm et al. 2006; Cowen et al. 2007). Continued volcanism throughout the life span of the ridge at EPR, in conjunction with the high spreading rate and entrainment of fluid and precipitates upward into plumes, is responsible for the relatively diminutive size of the vent deposits (Tivey 2007). However, the continued volcanic activity has also added to the diversity of venting as both black and white smokers dot the vent field and low-temperature diffuse flow is observed exiting from the cracks and crevices in the basaltic floor (Haymon & Kastner 1981). At these locations of diffuse flow there is mixing between cool sea water and hot hydrothermal fluid within the crust. Some chemical species do not behave
Seafloor hydrothermal systems 165 conservatively during mixing, indicating that there may be subsurface microbial communities consuming H2S, H2 and CO2 while at the same time producing other sulfur and carbon species (Von Damm & Lilley 2004). The high-temperature (330–405C) black smokers do not seem to entrain much sea water because the Mg concentrations in samples are very low (Von Damm 2004). All black smoker vents within the hydrothermal system at 9N are acidic and have Cl concentrations indicating that phase separation occurs within the crust (Von Damm 2004). These fluids are enriched in CO2 owing to phase separation and ⁄ or magmatic degassing on this section of the ridge (Lilley et al. 2002; Von Damm & Lilley 2004), which may also account for some of the low pH values and elevated sulfide concentrations. The EPR 9N system has experienced considerable activity over the course of its existence and it is thought that the changing fluid compositions reflect magma migration within the crust on shorter time scales than have been observed in other hydrothermal vent systems (Von Damm 2004). Trans-Atlantic Geotraverse The Trans-Atlantic Geotraverse (TAG) location hosts the single largest known vent deposit on any spreading center (Tivey 2007). TAG has been active for the last 140 Kyr (Lalou et al. 1995) and possesses both black and white smokers as well as zones of diffuse venting. The large sulfide deposit, known as TAG mound, is the focal point for high-temperature activity (Kleinrock & Humphris 1996; Humphris & Tivey 2000) and measures almost 200 m in diameter with a height of almost 50 m (Humphris et al. 1995). Venting at TAG was first sampled during an Alvin expedition in 1986. The typical exit temperature for vent fluids on the mound are in excess of 360C (Campbell et al. 1988a; Chiba et al. 2001; Parker & Von Damm 2005). Newer near-bottom magnetic data seem to imply that there is crustal thinning from long-lived extension on a normal fault (Tivey et al. 2003; deMartin et al. 2007) at the site of TAG. This in turn would indicate that TAG is in fact perched on the hanging wall of an active detachment fault, meaning that it is the tectonic setting of the site rather than the volcanism that is more crucial to the long-lived hydrothermal activity and circulation, which has allowed TAG to grow to such a large size (deMartin et al. 2007). Over the course of its activity, continuous sea water entrainment has instigated precipitation of anhydrite, chalcopyrite and pyrite within the mound and remobilized metals (Edmond et al. 1995; Tivey et al. 1995). A sequence of pyrite, anhydrite, silica and chloritized basalt breccias and stockwork beneath the mound were revealed during the Ocean Drilling Program’s recovery of rock cores (Humphris et al. 1995). These characteristics prove
to be closely matched by those of the Cyprus massive sulfide deposit (Hannington et al. 1998). Endeavour Located near the northern end of the Juan de Fuca Ridge, the 90-km-long Endeavour Segment has an intermediate spreading rate of 6 cm year)1 (Riddihough 1984). In the middle of this segment is a 25-km volcanic high split by a 75- to 200-m-deep, 0.5- to 1-km-wide, steep-sided axial valley (Glickson et al. 2007). As the valley continues south to the end of the segment it attains a width of approximately 3 km. The Main Endeavour Hydrothermal Field (MEF) was discovered in 1982 (Tivey & Delaney 1986; Delaney et al. 1992), and is located at 4757¢N and 12905¢W. Hydrothermal activity at MEF extends along 400 m of the western wall of the 800-m-wide axial valley (Seewald et al. 2003). Bounding normal faults provide conduits for fluids to reach the seafloor and vent locations tightly correlated with these faults (Tivey & Delaney 1985; Delaney et al. 1992). Unlike typical mid-ocean ridge systems, the Endeavour systems feature quite large hydrothermal deposits with steep sides and an abundance of amorphous silica and flanges (Kelley et al. 2002). Some of these structures attain heights of up to 20 m and diameters of 30 m (Seewald et al. 2003). Also out of the ordinary are the high concentrations of CH4, NH4, Br and B, as well as light CH4 carbon isotopic compositions (d13C < )45&), which are consistent with the interaction between sedimentary organic matter and hydrothermal fluids (Tivey & Delaney 1986; Delaney et al. 1992; Lilley et al. 1993; Butterfield et al. 1994; You et al. 1994). Diffuse venting is commonly found along faults and fissures of this basaltic system (Delaney et al. 1992). For some time it was thought that this system was in a tectonically dominated phase of a volcano-tectonic cycle (Kappel & Ryan 1986); however, thinking changed after the discovery of a seismic reflector, imaged at 1.9–4.0 km beneath the central third of the ridge, which was interpreted as the roof of an axial magma chamber extending beneath all vent fields in the system (Detrick et al. 2002; Van Ark et al. 2007). Further exploration led to the discovery of the High Rise, Salty Dawg, Mothra and Sasquatch high-temperature hydrothermal fields. All of these sites are located along a 15-km section of the central ridge segment making the Endeavour segment one of the most active hydrothermal fields ever discovered (Robigou et al. 1993; Kelley et al. 2001a,b). The high density of vent fields and the intensity of their venting has enabled research into the linkages between biological, chemical, geological and geophysical processes associated with this segment of the ridge (e.g. Delaney et al. 1992; Lilley et al. 1993; Sarrazin et al. 1999; Kelley et al. 2002; Tivey & Johnson 2002; Schrenk et al. 2003; Wilcock 2004; Wilcock et al. 2009).
166 E. SHOCK & P. CANOVAS Steep gradients in vent fluid composition and temperature across the MEF are interpreted as stemming from supercritical phase separation (Delaney et al. 1992; Kelley et al. 2002). Studies by Delaney et al. (1992, 1997), Lilley et al. (1993) and Butterfield et al. (1994, 1995) identified highly variable chlorinities (40–540 mmol kg)1) and temperatures along or near the boiling curve for sea water, reflecting both sub- and supercritical processes. In 1999 a magnitude 5.0 earthquake perturbed the system and vent fluids with higher temperatures and salinity indicated the possibility of a minor release of the sequestered brine (Lilley et al. 2000). The long-lived nature of the MEF hydrothermal system and organic-rich composition of its fluids support a diverse and prosperous microbial community with sulfide edifices housing between 105–109 cells g)1 within the chimney structures (Schrenk et al. 1998, 1999). Guaymas Sedimented ridge systems are uncommon because they require the rate of deposition to exceed that of spreading. However, this is just the case at the Guaymas Basin, located at 27N, 111.5W in the central Gulf of California. Overall, the Guaymas Basin is a site of active seafloor spreading associated with the extensional tectonics of the EPR (Simoneit et al. 1992). It is comprised of two northeast-trending grabens known as the Northern and Southern Troughs. The Northern Trough is 40 km long, while the Southern is 20 km long and both are 3–4 km wide (Peter & Scott 1988). Lonsdale & Becker (1985) identified them to be a pair of en echelon, axial rift valleys that overlap at a non-transform offset. Biogenic and terrigenous sediments at this locale are at least 500 m thick (Lonsdale & Lawver 1980). New crust forming at the spreading axis intrudes into overlying sediments as sills and dikes (Einsele et al. 1980; Einsele 1982). Ferromanganese-encrusted sulfide and talc deposits were first discovered in 1977 during submersible dives in the Northern Trough (Lonsdale 1978; Lonsdale et al. 1980). Elevated 3He values, 65–70% higher than atmospheric levels, were interpreted as input from mantle sources (Lupton 1979). Subsequently, in August 1980, extensive hydrothermal deposits were mapped by deep-tow side-scan sonar and photography (Lonsdale 1980), marking the discovery of the Southern Trough hydrothermal area. Samples of hydrothermal precipitates were collected by dredge in 1980, and their mineralogy was described by Koski et al. (1985). In 1982 a series of nine dives by the DSV Alvin was conducted in the Southern Trough within the area of high heat flow (Lonsdale & Becker 1985). The unusually high sedimentation rates of 1–2 m (103 years))1 (Calvert 1966) help shape the biogeochemical processes inherent to this ridge system. Magmatic activity and sill injection produce hydrothermal fluids and
secondary solid phases that differ in chemical and mineralogical composition from those that occur at open-ocean, sediment-starved spreading axes, owing to interaction with the overlying sediments (Kastner 1982; Stout & Campbell 1983; Von Damm et al. 1985a,b; Gieskes et al. 1988). The terrigenous component is derived predominantly from Tertiary volcanics of the Mexican mainland. Sediments accumulating in this shallow basin are hemipelagic, diatomaceous, organic-rich material of low density and high permeability (Curray et al. 1982). Compared with sediment-starved ridges, Guaymas Basin fluids have higher pH, alkalinity, H2S, NH4 and CH4 concentrations, and are depleted in dissolved metals such as Fe, Mn, Cu and Zn (Gieskes et al. 1982a,b, 1991; Von Damm et al. 1985a,b; Welhan & Lupton 1987). Another consequence of the setting that this ridge inhabits is that hydrothermal alteration of sediments produces petroleum products on timescales of tens to thousands of years (Simoneit & Lonsdale 1982; Simoneit 1983a,b, 1985; Peter et al. 1990, 1991). As a consequence, the composition of the vent fluids, sedimentary substrate, and active petroleum generation combine to produce a habitat where a plethora of unusual bacteria, archaea and eukaryotes can thrive (Teske et al. 2002). Lau basin Of the 65 000 km of global ridge crest, less than 10% has undergone systematic exploration for the presence and location of high-temperature venting (Baker & German 2004). Of that entire length, back-arc spreading centers constitute less than 7000 km (Bird 2003) but offer a great variety of hydrothermal fluid compositions, host rock alteration and volatile compositions that is difficult or even impossible to study at mid-ocean ridge systems (German & Von Damm 2004). In this respect, some have argued that the Lau Basin is an ideal study site (German et al. 2006) because of its tectonophysics (Zellmer & Taylor 2001) and the potential for integration of geochemical and geophysical investigations (Turner & Hawkesworth 1998). The Lau Basin hydrothermal systems also attract biological studies (German et al. 2006) owing to the differences between Lau and EPR vent communities (Desbruyeres et al. 1994) and the similarities seen between the Lau and Central Indian Ridge vent communities (Van Dover et al. 2001). Hydrothermal activity in the Lau basin was first discovered and sampled during the 1989 Nautilau cruise aboard the R ⁄ V Nadir between 17 April and 10 May 1989 using the submersible Nautile. During that time 22 dives were performed along the Valu Fa back-arc ridge behind the Tonga–Kermadec trench between latitudes 2125¢S and 2240¢S in water 2000 m deep (Fouquet et al. 1991a,b, 1993). The dive sites were selected on the basis of results from cruises of the R ⁄ V Jean Charcot and R ⁄ V Sonne (Von
Seafloor hydrothermal systems 167 Stackelberg and Shipboard Party 1988) and yielded samples from three major vent fields: White Church, Vai Lili and Hine Hina (Fouquet et al. 1991a,b, 1993; Taylor et al. 1996). In addition, seismic imaging indicated the existence of a magma chamber 3.2 km beneath the Central Valu Fa Ridge (Morton & Sleep 1985; Collier & Sinha 1992). More recently, studies were undertaken to explore the full length of the Valu Fa Ridge and the East Lau Spreading Center (Baker et al. 2005, 2006; Ishibashi et al. 2006; Martinez et al. 2006). These investigations demonstrated that hydrothermal activity progressively increases from south to north (Baker et al. 2005, 2006). The Vai Lili hydrothermal vent field lies at a depth between 1680 and 1740 m and covers an area of approximately 100 · 400 m2 consisting of at least 10 discrete clusters of black and white smokers. This area also contains a massive sulfide mound measuring 50 · 200 m2 with a height of 15 m (Herzig et al. 1998). The host rock associated with the Vai Lili sulfide deposit is predominantly basaltic andesites and rhyodacitic differentiates (Fouquet et al. 1991b) with mineralization being controlled by a normal fault running subparallel to the ridge (Herzig et al. 1993). Portions of the stockwork zone have also been exposed by block rotation (Herzig et al. 1993), which may indicate an advanced stage of tectonic activity (Fouquet et al. 1993). On top of the massive sulfide mound there are active high-temperature black smokers (320–342C) and lower temperature white smokers (250–320C) with 25C pH measurements as low as 2. These vents also have unusually high concentrations of dissolved metals (e.g. up to 7 mM Pb, 3 mM Zn, 35 mM Cu, etc.), which are thought to result from the reaction of andesitic basement rocks with sea water under greenschist conditions (Fouquet et al. 1991b). Kairei The Kairei vent field was the first hydrothermal system discovered and sampled in the Indian Ocean (Gamo et al. 2001; Hashimoto et al. 2001). In 2000 the ROV Kaiko located the black smoker chimneys on the south-western flank of an off-axis knoll 24 km north of the Rodriguez Triple Junction on the Central Indian Ridge (CIR) (Gamo et al. 2001; Hashimoto et al. 2001). The CIR has a spreading rate of 50–60 mm year)1 (DeMets et al. 1990). Active vents at Kairei are located on the south-western flank of a sulfide mound, called Hakuho Knoll, with venting focused in an area approximately 80 m · 40 m (Hashimoto et al. 2001). The black smokers of this basalt-hosted system are similar to others, with some exceptions. Chloride, temperature and silica measurements indicate a complex trajectory of water through the system. The model proposed by Gallant & Von Damm (2006) suggests that there is a
high-temperature source fluid that last equilibrated with quartz at supercritical conditions before undergoing phase separation during the portion of its ascent where it is still above the critical point for sea water. During the remaining part of the ascent conductive cooling is believed to occur. The other major difference is the unusually high H2(aq) concentration in conjunction with a moderately low abundance of CH4(aq). Recently, Nakamura et al. (2009) have proposed that the serpentinization of troctolites in the subsurface is responsible for the low CH4 ⁄ H2 ratio and high Si concentrations observed. This suggests that the deepest high-temperature subsurface reactions occurring at Kairei involve sea water reacting with constituents of deep oceanic crust or portions of the crust ⁄ mantle boundary (Dick et al. 2000; Nakamura et al. 2009). Given the unique geology of the system and its impact on the chemistry of the vent fluids, Karei was proposed to host a hydrogen-based hyperthermophilic subsurface lithoautotrophic microbial ecosystem (HyperSLIME; Takai et al. 2004). Microorganisms in this system would be solely dependent on CO2 and H2 for their primary carbon and energy needs, and this type of ecosystem is proposed to be an analogue for life on Earth before photosynthesis (Takai et al. 2006). Rainbow Hydrothermal activity was discovered along the Mid-Atlantic Ridge in the late 1970’s leading to the sampling of vent fields such as TAG, Lucky Strike, Menez Gwen, Broken Spur and others. All of these are hosted in mafic rock systems and share characteristics similar to those of basaltic systems. The later MICROSMOKE cruise (1995) identified several slow-spreading on-axis ultramafic systems (Konn et al. 2009). The Rainbow hydrothermal field, first discovered in 1997 (Fouquet et al. 1997) shares characteristics with two other ultramafic hydrothermal systems: Logetchev (Sudarikov & Roumiantsev 2000) and Ashadze (Mozgova et al. 2008). The Rainbow hydrothermal field is located south of the Azores on the Mid-Atlantic Ridge at 3614¢N, 3354¢W and a depth of 2300 m (Charlou et al. 2002). The system is at the intersection of a non-transform fault with the ridge faults, and extends 250 m west of the ridge with a width of 60 m (Simoneit et al. 2004). The intensity of the hydrothermal activity, as well as the height of the mounds, increases from west to east and is thought to be related to propagation of fluid channels toward the east caused by cementing and sealing of veins in talus and sediment cover on the west (Simoneit et al. 2004). Within the field are nearly a dozen groups of black smokers distributed across the entire region, all with peridotite host rock compositions and vent fluids issuing large amounts of CH4 and H2 (Charlou et al. 1998, 2002) at 360C (Fouquet et al. 1997).
168 E. SHOCK & P. CANOVAS Rainbow is also characterized by low 25C pH fluids (3–4) with high chlorinity (780 mmol kg)1), high metal concentrations (e.g. total Fe 24 mmol kg)1), high alkaline earth cation levels (Donval et al. 1997; Douville et al. 2002) and low dissolved SiO2 concentrations (7 mmol kg)1). The low silica concentrations are less than half that of other high-temperature basalt or gabbro-hosted vent systems (Von Damm et al. 1985a; Campbell et al. 1988a; Von Damm 1995). The high Fe concentration suggests a relatively low in situ pH (Ding & Seyfried 1992). Low pH and high dissolved Fe and SiO2 concentrations are not typically associated with alteration processes in ultramafic systems (Janecky & Seyfried 1986; Berndt et al. 1996; Wetzel & Shock 2000). Serpentinization of peridotite host rock may be responsible for the concentrations of dissolved methane and hydrogen associated with these systems where levels may be as high as 16 mmol kg)1 for dissolved hydrogen, and 2.5 mmol kg)1 for dissolved methane (Charlou et al. 1998, 2002). The latter is nearly two orders of magnitude greater than basalt-hosted systems with similar dissolved chloride concentrations (Baross et al. 1982; Lilley et al. 1982; Welhan & Craig 1983). Hydrocarbons have also been reported in vent fluids from Rainbow (Holm & Charlou 2001) raising the possibility of organic synthesis. It has been argued that abiotic organic synthesis through Fischer–Tropsch-type processes occurs at Rainbow, driven by the large amounts of hydrogen generated by serpentinization (Konn et al. 2009).
MIXING OF SUBMARINE HYDROTHERMAL FLUIDS WITH SEA WATER Examination of the data in Table 1 reveals that, in addition to being at very different temperatures, sea water and hydrothermal fluids differ radically in chemical composition, especially for elements that occur in more than one oxidation state, including H, O, C, N, S and Fe. The reduced forms of these elements predominate over the oxidized forms in hydrothermal fluids, and the situation in sea water is reversed. As hydrothermal fluids mix with sea water all of these elements in differing oxidation states are brought together, which generates disequilibria that exceed any oxidation–reduction disequilibrium states present in the end-member solutions. If there were no kinetic inhibitions to oxidation–reduction reactions, then these mixtures could re-equilibrate rapidly. However, in every case there are mechanistic complexities to transferring electrons between oxidized and reduced forms of these elements, and these kinetic barriers tend to increase as temperature decreases. As these oxidation–reduction reactions do not easily re-equilibrate, disequilibria persist long enough for microbial processes to operate, and for microbes to use various forms of disequilibria as sources of chemical energy.
In the calculations conducted in this study, oxidation– reduction equilibria were suppressed for all reactions except H2(aq) + 12O2(aq) = H2O, which is required to keep the calculations from departing unrealistically from the range of H2O stability. Although this reaction is also kinetically inhibited during mixing of sea water and hydrothermal fluids, allowing H2(aq) and O2(aq) to equilibrate simulates the transition from reduced conditions at high temperatures to oxidized conditions at lower temperatures as mixing occurs. In a crude way, maintaining this single oxidation–reduction equilibrium simulates the ecosystem transitions that occur as fluids mix and microbes take advantage of the emerging energy supplies. It also provides a lower limit to the activity of H2(aq) in the mixed fluids. One way to think of this is that the calculated fluid mixtures may have less H2(aq) than their natural counterparts, which means that the resulting thermodynamic analyses represent conservative estimates of the energy available from oxidation–reduction disequilibria, especially at higher temperatures. This same approach was taken by McCollom & Shock (1997), Shock & Schulte (1998) and Amend & Shock (1998). One goal of the present study was to consider a suite of fluids representative of the diversity of submarine hydrothermal fluid compositions from basalt-hosted, ultramafichosted, sedimented and back-arc spreading centers, and to compare their potential for supporting abiotic organic synthesis and biosynthesis. Another was to expand upon the variety of organic compounds included in such calculations, especially so that more of the monomers needed for synthesis of biological polymers (proteins, RNA, DNA and lipid membranes) were considered. The EQ3 ⁄ 6 software package (Wolery & Jarek 2003) was used in the present study with a set of equilibrium constants derived from the slop07 database1 for the Supcrt92 (Johnson et al. 1992) code. Mixing is simulated through a titration model in which successive amounts of cold sea water are added to an initial kilogram of hot hydrothermal fluid. The bottom sea water data listed in Table 1 were used to characterize sea water in all of the calculations conducted in this study. Many authors report pH values for submarine hydrothermal fluids measured at 25C, usually onboard ship soon after the samples are collected. These are the pH values listed in Table1. Calculating the consequences of mixing between hydrothermal fluids and sea water requires that the pH and speciation of the hydrothermal fluids at the measured temperature be assessed, which is done by calculating the consequences of heating the fluid to the original measured temperature. Once this is accomplished, 1 The slop07 database includes standard state data and parameters published through 2007, and is available from the GEOPIG website (http://geopig.asu.edu/).
Seafloor hydrothermal systems 169 the resulting fluid composition is used for the mixing calculations. As a consequence, the initial pH values for the high-temperature hydrothermal fluids differ from those reported at 25C. These initial values plot at the hightemperature ends of the curves shown in Fig. 2. It can be seen that the initial temperatures of the hydrothermal fluids vary, with some at or near 350C and others at somewhat lower temperatures. These results show that the diversity of submarine hydrothermal fluids exhibit a large range in high-temperature pH values, spanning from nearly seven down to almost two. It can be seen that the variations in hydrothermal fluid compositions drive considerably different trajectories of pH during mixing with sea water. In some cases, the curves exhibit plateaus and inflections (Kairei, Rainbow, EPR 9N, Lau and Endeavor), and some results show minima in pH during mixing (Guaymas and Endeavor). These changes in pH stem from the changing speciation of multiple elements during mixing, with major contributions from carbonic acid and hydrogen sulfide. Changes in activities of CO2(aq), H2(aq), NHþ 4, CH4(aq), H2O and H2S(aq) for the seven mixing calculations are shown in Fig. 3. The distribution of the curves at high temperatures derives from the differences in the compositions of the hydrothermal fluids. The results for the activity of H2(aq) all show the effects of maintaining
Mixing of vent fluid with seawater
8
Guaymas
7
Kairei Endeavour
pH
6
TAG
5
equilibrium with the reaction H2(aq) + 12O2(aq) = H2O. The individual curves plunge to large negative log activity values at the stage of mixing where the O2(aq) content of sea water overwhelms the H2(aq) content of the vent fluid. The other solutes are unaffected by this behavior as all other oxidation–reductions have been suppressed in the calculations. Trends for the activities of other solutes as mixing with sea water proceeds and temperatures decrease depend increasingly on the composition of sea water, and many reflect perturbations induced by the shifts in pH shown in Fig. 2. As an example, inspection of Table 1 shows that the Guaymas hydrothermal fluid contains the largest concentration of total dissolved carbon dioxide, but the upper left-hand plot in Fig. 3 shows that the activity of CO2(aq) is most positive in the mixture derived from the Lau Basin fluid. As shown in Fig. 2, the pH of the Lau mixture is the lowest of all the seven calculations, and the effects of the low pH on the speciation of carbonic acid in solution drive the activity of the actual CO2(aq) species to greater values, despite the lower total abundance of dissolved carbon dioxide. An analogous explanation applies to the activities of H2S(aq) shown in the lower right-hand plot in Fig. 3. Total sulfide is the greatest in the Guaymas fluid, but the higher pH causes the activity of H2S(aq) in the fluid mixture to be lower than many other calculated values. The fluid that ranks second in total sulfide (Endeavour) has the highest activity of H2S(aq). It can also be seen in Fig. 3 that activities of CH4(aq) and NHþ 4 are the highest in the Guaymas mixture, but the changing pH in this case affects the speciation of NHþ 4 relative to NH3(aq). The other mixtures maintain low enough pH values that the speciation of aqueous ammonia has limited impact on the NHþ 4 activity. Finally, the activities of H2O depicted in the lower lefthand plot do not stray far from unity in these calculations, but the values are slightly different owing to salinity variations that affect ionic strength. The H2O activities shown were used explicitly in the affinity calculations discussed below.
4 EPR 9°N
QUANTIFYING DISEQUILIBRIA
Rainbow
3 Lau
2 0
50
100
150
200
250
300
350
Temperature (°C) Fig. 2. Calculated values of pH resulting from mixing of sea water with each of the hydrothermal fluids listed in Table 1. Initial pH values of the hydrothermal fluids were re-calculated at 250 bars and the temperature reported for each fluid in this study, and do not correspond to the 25C 1-bar values listed in Table 1. Note that the high-temperature ends of the curves correspond to the reported temperatures in Table 1, and the lowtemperature ends converge on the value for sea water listed in Table 1.
The extent to which a chemical system departs from equilibrium can be assessed by evaluating the chemical affinity of various possible reactions within that system. The chemical affinity of a reaction (Ar) is defined as the partial derivative of the change in the overall Gibbs energy of the system (DG) with respect to the progress of that reaction (nr) given by (Helgeson 1979) oDG ð1Þ Ar or P;T ;k where P and T stand for pressure and temperature, and nk represents the progress of all other reactions. As the Gibbs
170 E. SHOCK & P. CANOVAS
0
–0.5
Rainbow
Lau
Guaymas
Kairei
–2
–1 Guaymas
–4
Endeavour
–2 Kairei –2.5 –3
EPR 9°N
TAG
Rainbow
loga H2(aq)
loga CO2(aq)
–1.5
Lau
–6
Endeavour
EPR 9°N
–8
TAG
–10 –12
–3.5 –14 –4
0
50
100
150
200
250
300
0
350
50
Temperature (°C)
150
250
350
Guaymas
Lau
Endeavour
–2
–3
Endeavour
TAG –5
Rainbow
–6
Kairei
loga CH4(aq)
–3 –4
Kairei Rainbow
–4 TAG –5 Lau
–6 EPR 9°N
–7 –8
300
–1 Guaymas
EPR 9°N
0
50
100
150
–7
200
250
300
–8
350
0
50
Temperature (°C)
100
150
200
250
300
350
Temperature (°C)
–0.005
–1 Endeavour
–0.006
EPR 9°N
–2
Kairei
loga H2 S(aq)
–0.007
–0.008 Guaymas
–0.009
Endeavour
Lau
Rainbow TAG
loga H2O
200
Temperature (°C)
–2
loga NH4+(aq)
100
–3 Kairei –4
TAG
Guaymas Rainbow
–5
EPR 9°N Lau –0.01
0
50
100
150
200
250
300
350
–6
0
50
Temperature (°C)
100
150
200
250
300
350
Temperature (°C)
Fig. 3. Selected results of mixing calculations. Activities of CO2(aq) (upper left), H2(aq) (upper right), NHþ 4 (middle left), CH4(aq) (middle right), H2O (lower left) and H2S (lower right) as functions of temperature at 250 bars as each hydrothermal fluid mixes with sea water. Results for H2(aq) show the effects of maintaining equilibrium with the reaction H2(aq) + 12O2(aq) = H2O throughout the mixing calculation, and plunge toward large negative values at the points where the dissolved oxygen in sea water is calculated to overwhelm the dissolved hydrogen in the hydrothermal fluid.
energy decreases for spontaneous reactions, it follows that the chemical affinity is positive for favorable reactions that are not at equilibrium. The overall Gibbs energy of a reaction can be expressed as 0
Dr G ¼ Dr G þ RT ln Qr ; in which Qr stands for the activity product given by
ð2Þ
Qr ¼
Y
ai i; r ;
ð3Þ
i
where mi, r stands for the stoichiometric reaction coefficient of the ith chemical species in the reaction, which is positive for products and negative for reactants, and DrG 0 indicates the standard Gibbs energy of the reaction, given by
Seafloor hydrothermal systems 171 Dr G 0 RT ln Kr ;
Methanogenesis
ð4Þ
which implies that the equilibrium constant, Kr, is a function of temperature, pressure and the choice of standard states.2 These relations can be combined to yield ð5Þ
If Kr > Qr, Ar is positive, and energy is released as the reaction proceeds. By keeping oxidation–reduction reactions from equilibrating in the calculations pursued in this study, values of the chemical affinity for numerous reactions can be evaluated across the entire range of sea water–hydrothermal fluid mixing ratios. The connection between chemical affinity and the potential for a reaction to take place can be illustrated with the example of the overall reaction that occurs during autotrophic methanogenesis. This reaction is given by CO2 ðaqÞ þ 4H2 ðaqÞ ) CH4 ðaqÞ þ 2H2 O;
ð6Þ
for which the logarithmic form of Eqn (3) is log Q6 ¼ log a CH4 ðaqÞ þ 2 log a H2 O log a CO2 ðaqÞ 4 log a H2 ðaqÞ:
ð7Þ
An autotrophic methanogen can only gain energy by making methane when the activity of aqueous methane is below the equilibrium value consistent with the coexisting activities of aqueous carbon dioxide and hydrogen. It is precisely at these same conditions where the chemical affinity of reaction (6) is positive. So, the first step to finding conditions at which autotrophic methanogenesis pays as a metabolic lifestyle is to determine where positive values of the chemical affinity occur. This can be done by combining the activities of CH4(aq), CO2(aq), H2(aq) and H2O shown in Fig. 3 to evaluate Q6 (Eqn 7), and combining the result with independently calculated values of the equilibrium constant (K6) to yield values of A6 using Eqn (5). Equilibrium constants for reaction (6) can be calculated using equations, standard state thermodynamic data and equation-of-state parameters from Shock et al. (1989) and Shock & Helgeson (1990). Resulting chemical affinities for autotrophic methanogenesis from the seven mixing calculations conducted in this study are shown in Fig. 4. It can be seen that chemical affinities for reaction (6) become positive in all cases over at least some of the temperature range of each mixing calculation. Keeping in mind that these calculated values are conservative, owing to the simultaneous equilibration of H2(aq) and O2(aq) with H2O, the results in Fig. 4 suggest 2 Standard states used in this study are unit activity for pure minerals and pure H2O at any temperature and pressure, unit fugacity of pure gases at 1 bar and any temperature, and, for aqueous species, unit activity of a hypothetical 1 molal solution referenced to infinite dilution at any temperature and pressure.
250 bars
Kairei EPR 9°N
Guaymas
Affinity (cal mol–1)
Ar ¼ RT lnðKr =Qr Þ:
Rainbow
40 000
20 000
0 Lau Endeavour
–20 000
TAG –40 000 0
50
100
150
200
250
300
350
Temperature (°C) Fig. 4. Calculated affinities for the overall methanogenesis reaction (6). Positive values indicate temperature ranges where methanogenesis would release energy. Results depend, in part, on the activities of CO2(aq), H2(aq) H2O and CH4(aq) shown in Fig. 3, and reflect the plunge in H2(aq) activities. Suppression of equilibrium between dissolved hydrogen and oxygen would permit affinity values to attain more positive values.
that autotrophic methanogens should be present in the mixing zones around all of these seafloor hydrothermal sites. The common occurrence of thermophilic and hyperthermophilic methanogens at submarine hydrothermal vent locations (Takai et al. 2004; Nercessian et al. 2005; Reed et al. 2009) corroborates these results. As illustrated in Fig. 4, the most positive values of the chemical affinity for reaction (6) are attained by the calculations for mixing between sea water and the fluid compositions from Rainbow and Kairei. It follows that these locations are the most conducive to supporting autotrophic methanogens of the locations considered in this study. Note that the positive values of affinity at the highest temperatures indicate that these fluids start with insufficient CH4(aq) compared with what would be in equilibrium with the measured CO2(aq) and H2(aq) abundances. By contrast, hydrothermal fluid compositions from Lau, TAG and Endeavor have more methane than equilibrium with CO2(aq) and H2(aq) requires, yielding negative values of the affinity at the highest temperatures. It can also be seen in Fig. 4 that all curves plunge to negative values of affinity as temperature decreases, corresponding to the conditions at which sea water, which is far too oxidized to support methanogenesis, begins to overwhelm the ever-diminishing contribution to the mixture from the hydrothermal fluid. Comparison of Fig. 4 with the appropriate four plots of Fig. 3 provides insights into the ranking of the results, and why the various curves take the shapes that they do. At high temperatures the ranking of affinities in Fig. 4 is Rainbow > Kairei > EPR 9N > Guaymas Lau >
172 E. SHOCK & P. CANOVAS Endeavor > TAG. At low temperatures, many of the curves cross, but they plunge to negative values at temperatures that decrease in the order TAG > Lau Endeavour > EPR 9N > Guaymas > Kairei Rainbow. These rankings are similar to the relative activities of H2(aq), with the exception that the activity of H2(aq) is greater at Guaymas than at EPR 9N. The ranking of methanogenesis affinities for these two mixing calculations is inverted because of the high activities of CH4(aq) at Guaymas. It is evident that the activity of H2(aq) exerts greater control over the ranking of affinities than do the other activities in reaction (6). The key is in the stoichiometric reaction coefficients, which are 1 and )1 for CH4(aq) and CO2(aq), respectively, but )4 for H2(aq). All activities of H2O are not far from 1, and do not contribute significantly to the differences in these curves despite its stoichiometric reaction coefficient of 2.
AFFINITIES FOR ABIOTIC ORGANIC SYNTHESIS Positive values of affinity for autotrophic methanogenesis means that the mixtures of hydrothermal fluids and sea water are poised to favor the production of methane. This raises the possibility that the synthesis of other organic compounds may be favored during fluid mixing, which can be tested by evaluating chemical affinities for overall organic synthesis reactions. If concentrations of these compounds were known in sea water and hydrothermal fluids, then activities in mixed fluids could be evaluated, and calculations similar to those described above for methanogenesis could be done. However, compositional data of this type are largely lacking. In the absence of analytical data we can make comparisons between calculated affinities by setting the activities of the organic solutes to some arbitrary value. In this study, an activity of 10)6 was selected to make these comparisons. If activity coefficients are close to unity, then this activity is roughly equivalent to a micromolal concentration, which should be detectable by currently available analytical methods. In addition, such concentrations would be relatively large for intercellular fluids, making the comparisons for biosynthesis reactions compelling. Affinities for arbitrarily selected activities of organic solutes can be assessed from the mixing calculations conducted in this study provided that values of Kr can be calculated. Standard state thermodynamic data for aqueous organic compounds, leading to equilibrium constants for reactions in which they participate, exist for hundreds of organic compounds (Shock & Helgeson 1990; Shock 1992, 1993, 1995; Schulte & Shock 1993; Amend & Helgeson 1997a,b, 2000; Amend & Plyasunov 2001; Dick et al. 2006; LaRowe & Helgeson 2006a,b), and estimation methods permit the inclusion of thousands more (Plyasunov & Shock 2001a,b, 2003; Plyasunov et al. 2004, 2006a,b; Plyasunova et al. 2005).
Examples for several groups of aqueous organic compounds are described here starting with some of the smaller representatives of these groups. C1 compounds Reduction of CO2(aq) can lead to the formation of other compounds in addition to methane. Affinities for forming activities of 10)6 of methanol, formaldehyde, CO(aq) and formic acid are shown in Fig. 5, corresponding to the reactions CO2 ðaqÞ þ 3H2 ðaqÞ ) CH3 OHðaqÞ ðmethanolÞ þ H2 O; CO2 ðaqÞ þ 2H2 ðaqÞ ) CH2 OðaqÞ ðformaldehydeÞ þ H2 O; CO2 ðaqÞ þ H2 ðaqÞ ) COðaqÞ þ H2 O;
ð8Þ
ð9Þ
ð10Þ
and CO2 ðaqÞ þ H2 ðaqÞ ) HOOCHðaqÞ ðformic acidÞ:
ð11Þ
By choosing the same activity for each of these organic compounds, and setting the ranges of the plots to be the same, it should be easier to compare the thermodynamic drives for these synthesis reactions. Note that, as in the case of methanogenesis shown in Fig. 4, affinities become most positive as the temperature decreases. Temperature is controlled by the ratio of sea water to vent fluid; so, as temperature decreases the contribution from the vent fluid is being progressively diluted. As shown in the upper left-hand plot in Fig. 5, affinities are negative for the methanol synthesis reaction (8) in all of the high-temperature (>300C) vent fluids, indicating that these fluids would have equilibrium activities of methanol less than 10)6. However, as fluids from Rainbow, Kairei, Guaymas and EPR 9N mix with sea water, and the temperature of the mixture decreases, thermodynamic drives emerge that are more than sufficient to generate methanol activities of 10)6. Affinities for forming this activity of methanol are barely positive at Lau and Endeavor, and only over a narrow range of temperature between 100 and 200C. The mixture resulting from TAG fluid and sea water never attains positive affinities for a methanol activity of 10)6, which means that only lower activities would be possible if reaction (8) was to proceed toward equilibrium. In contrast to methanol, affinities never turn positive for 10)6 activities of formaldehyde (reaction 9) in any of the mixing calculations conducted in this study (Fig. 5 upper right-hand plot). If formaldehyde forms in these mixtures, its activity is likely to be less. Note that the set of curves
Seafloor hydrothermal systems 173
Methanol synthesis
Formaldehyde synthesis 250 bars
250 bars 40 000
Rainbow Kairei
Guaymas
20 000
Affinity (cal mol–1)
Affinity (cal mol–1)
40 000
EPR 9°N
0
Lau
–20 000
Endeavour
Rainbow
20 000
Kairei
0
–20 000 Lau
TAG
50
100
150
200
250
300
0
350
TAG
Endeavour
–40 000
–40 000 0
Guaymas EPR 9°N
50
Temperature (°C)
100
150
200
250
Carbon monoxide synthesis
350
Formic acid synthesis
250 bars
250 bars
40 000
40 000
Rainbow
20 000
Kairei
Guaymas
Affinity (cal mol–1)
Affinity (cal mol–1)
300
Temperature (°C)
EPR 9°N
0
–20 000
Lau
TAG
Endeavour –40 000
Guaymas
50
100
150
200
250
300
350
Lau
0 Endeavour
EPR 9°N
–20 000
–40 000 0
Kairei
Rainbow
20 000
TAG 0
50
Temperature (°C)
100
150
200
250
300
350
Temperature (°C)
Fig. 5. Chemical affinities calculated for the synthesis of the C1 aqueous compounds methanol, formaldehyde, CO(aq) and formic acid corresponding to reactions (8–11) as functions of temperature during mixing between hydrothermal fluids and sea water at 250 bars.
for formaldehyde is more tightly grouped than the corresponding set for methanol, or that for methane in Fig. 4. In addition, the curves for formaldehyde increase more gently with decreasing temperature than the corresponding curves for methanol or methane. The carbon in formaldehyde is more oxidized than the carbon in methanol or methane, which means that the stoichiometric reaction coefficient for H2(aq) is correspondingly lower. As the contribution of aH2 ðaqÞ to Qr decreases, the resulting curves of affinity versus temperature flatten and become closely spaced. This trend of flattening and bunching of affinity curves reaches its extreme for the cases of CO(aq) and formic acid (reactions 10 and 11), as shown in the lower two plots in Fig. 5. Formic acid can be considered to be a hydrated version of CO(aq) according to the reaction HOOCH ¼ COðaqÞ þ H2 O;
ð12Þ
which helps to explain why the two sets of curves look so similar. At lower temperatures, formic acid is slightly more stable than CO(aq) in aqueous solution, and comparison of the two sets of affinity curves shows that those for formic acid tend to be slightly more positive at temperatures <200C than the corresponding curves for CO(aq). The
observations that the affinity curves for CO(aq) are slightly more positive at higher temperatures can be explained by the fact that dehydration in aqueous solution becomes increasingly favorable as temperature increases (Shock 1993). C2 compounds As the number of atoms in organic molecules increases, so do the possibilities for new structures and arrangements. In this study, affinities for 8 two-carbon compounds were calculated for the seven mixing calculations. These simplest forms of organic compounds with a carbon–carbon bond include hydrocarbons, acids, an alcohol and an aldehyde. As in the case of the C1 compounds in Fig. 5, activities for C2 compounds were all set to 10)6 to facilitate comparisons. Affinities were calculated for the reactions 2CO2 ðaqÞ þ 7H2 ðaqÞ ) C2 H6 ðaqÞ ðethaneÞ þ 4H2 O;
ð13Þ
2CO2 ðaqÞ þ 6H2 ðaqÞ ) C2 H4 ðaqÞ ðetheneÞ þ 4H2 O;
ð14Þ
2CO2 ðaqÞ þ 5H2 ðaqÞ ) C2 H2 ðaqÞ ðethyneÞ þ 4H2 O;
ð15Þ
174 E. SHOCK & P. CANOVAS
2CO2 ðaqÞ þ 6H2 ðaqÞ ) C2 H5 OHðaqÞ ðethanolÞ þ 3H2 O;
ð16Þ
2CO2 ðaqÞ þ 5H2 ðaqÞ ) CH3 CHOðaqÞ ðacetylaldehydeÞ þ 3H2 O;
ð17Þ
2CO2 ðaqÞ þ 4H2 ðaqÞ ) CH3 COOHðaqÞ ðacetic acidÞ þ 2H2 O;
ð18Þ
2CO2 ðaqÞ þ 3H2 ðaqÞ ) CH2 OHCOOHðaqÞ
ð19Þ
ðglycolic acidÞ þ H2 O; and 2CO2 ðaqÞ þ H2 ðaqÞ ) HOOCCOOHðaqÞ ðoxalic acidÞ;
ð20Þ
and the results are plotted in Figs 6 and 7. The ranges of the axes are the same in all of the plots shown in these figures, which should help comparisons among the affinity results. Results for the three hydrocarbons: ethane, ethene (also called ethylene) and ethyne (also called acetylene), together
with those for ethanol are plotted in Fig. 6. Comparisons of these plots shows that the affinities to form ethane (upper left-hand corner) at an activity of 10)6 are the most positive, and those to form the same activity of ethyne (lower left-hand corner) are the most negative. Affinity curves for ethene are intermediate between the other two aqueous hydrocarbons, and are slightly less positive than corresponding curves for ethanol. As in the case of formic acid and CO(aq), ethanol can be thought of as a hydrated form of ethene. Likewise, the affinity values for ethanol tend to be somewhat more positive than the values for ethene, and this difference is more pronounced at low temperatures than at high. Comparison of Figs 4 and 6 shows that, at high temperatures, affinities for ethane, based on an activity of 10)6, are less positive than for methane, based on activities calculated from methane concentration data (see Fig. 3). This makes a direct comparison of the results difficult, but it is evident that affinities for ethane synthesis are large at an ethane activity that is one to five orders of magnitude lower than the initial methane activities of hydrothermal fluids. For reference, ethane concentrations at Lost City are within 20% of the concentrations of methane (Proskurowski et al. 2008) and about 50% of the methane concentration reported for Rainbow (Charlou et al. 2002). Ethene synthesis
Ethane synthesis Rainbow 40 000
250 bars
250 bars 40 000
Kairei
Rainbow Guaymas
Affinity (cal mol–1)
Affinity (cal mol–1)
Guaymas 20 000
EPR 9°N
0 Lau –20 000 Endeavour –40 000 50
100
150
200
0 EPR 9°N –20 000
Lau Endeavour
–40 000
TAG 0
Kairei 20 000
250
300
350
0
TAG 50
100
150
200
250
300
350
Temperature (°C)
Temperature (°C)
Ethanol synthesis
Ethyne synthesis 250 bars 40 000
40 000
250 bars
Rainbow
Kairei
Rainbow 20 000
Kairei Guaymas
0 Lau Endeavour
–20 000
0
50
100
20 000
0
EPR 9°N
–20 000
Lau Endeavour
EPR 9°N
–40 000
Affinity (cal mol–1)
Affinity (cal mol–1)
Guaymas
–40 000 TAG
TAG 150
200
250
Temperature (°C)
300
350
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 6. Calculated affinities for the synthesis of aqueous ethane, ethene, ethyne and ethanol, corresponding to reactions (13–16) as functions of temperature during mixing of seafloor hydrothermal fluids with sea water at 250 bars.
Seafloor hydrothermal systems 175
Acetic acid synthesis
Acetaldehyde synthesis 250 bars Rainbow
40 000
Kairei Guaymas
Affinity (cal mol–1)
Affinity (cal mol–1)
40 000
Rainbow
20 000 Lau 0 EPR 9°N –20 000 Endeavour
250 bars Kairei Guaymas
EPR 9°N 20 000
0 Lau –20 000 Endeavour –40 000
–40 000
TAG
TAG 0
50
100
150
200
250
300
0
350
50
100
150
200
250
Glycolic acid synthesis
350
Oxalic acid synthesis
250 bars
250 bars 40 000
Rainbow 20 000
Guaymas
Kairei
0
Lau
–20 000 EPR 9°N Endeavour
–40 000
Affinity (cal mol–1)
40 000
Affinity (cal mol–1)
300
Temperature (°C)
Temperature (°C)
20 000 Rainbow Kairei
Guaymas
0 Lau –20 000 EPR 9°N
–40 000
Endeavour TAG
TAG 0
50
100
150
200
250
300
350
Temperature (°C)
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 7. Calculated affinities for the synthesis of the aqueous organic compounds acetaldehyde, acetic acid, glycolic acid and oxalic acid according to reactions (17–20) as functions of temperature during mixing of seafloor hydrothermal fluids with sea water at 250 bars.
Calculated affinities for 10)6 activities of acetaldehyde and the three acids: acetic, glycolic and oxalic are plotted against temperature in Fig. 7 for the seven mixing models pursued in this study. Affinities for acetaldehyde and acetic acid reach positive values for all of the calculations except the mixing of TAG hydrothermal fluid with sea water, although the acetaldehyde affinities for the Lau and Endeavor mixing calculations are only barely positive in the 100– 150C range. The negative affinities calculated for 10)6 activities of the hydroxy acid (glycolic) and dicarboxylic acid (oxalic) suggest that these more oxidized organic compounds are considerably less stable than the more reduced carboxylic acid (acetic) and point to a dramatic difference in the behaviors of these three groups of organic acids. The positive values of affinity for acetate synthesis suggest that the metabolic process of acetogenesis may support microbes in mixing zones of seafloor hydrothermal systems. Comparison of Figs 6 and 7 allows us to rank these reactions in terms of affinity. Affinities for ethane are the most positive followed by those for acetic acid, which are slightly more positive than similar values for ethene and ethanol. Acetaldehyde also attains positive affinities, although less so than those for the compounds already mentioned. Glycolic
acid, ethyne and oxalic acid all have negative affinities, with those for oxalic acid being most negative. These results, based on a comparison for constant activities of 10)6, show that, during mixing of hydrothermal fluids and sea water, ethane is the most stable of the C2 products, followed by acetic acid, ethanol, ethene, acetaldehyde, glycolic acid, ethyne and oxalic acid. This suggests that alkanes, carboxylic acids, alkenes, aldehydes and alcohols are likely products of abiotic organic synthesis in hydrothermal systems. Carboxylic acids Results for acetic acid shown in Fig. 7 suggest that other carboxylic acids may be potential products of abiotic synthesis during mixing of hydrothermal fluids and sea water. This is confirmed by the results for four carboxylic acids: propanoic, hexanoic, nonanoic and dodecanoic, shown in Fig. 8. As in Figs 5–7, the activity of each acid was set to 10)6 to calculate values of Qr for the reactions
3CO2 ðaqÞ þ 7H2 ðaqÞ ) CH3 CH2 COOHðaqÞ ðpropanoic acidÞ þ 4H2 O;
ð21Þ
176 E. SHOCK & P. CANOVAS
Propanoic acid synthesis
Hexanoic acid synthesis
200 000
200 000 250 bars
250 bars 150 000
Rainbow
100 000
Affinity (cal mol–1)
Affinity (cal mol–1)
150 000
Kairei Guaymas Lau
50 000 0 EPR 9°N –50 000
Rainbow
Guaymas EPR 9°N
50 000
Lau
0 –50 000
–100 000 0
50
100
150
200
Endeavour
Endeavour
TAG
250
300
–100 000 0
350
TAG 50
Temperature (°C)
100
150
200
250
300
350
Temperature (°C) Dodecanoic acid synthesis
Nonanoic acid synthesis 200 000
200 000 250 bars
Rainbow
Kairei
150 000
Kairei Guaymas
100 000 EPR 9°N 50 000
Lau
0 TAG
250 bars
Rainbow
Affinity (cal mol–1)
150 000
Affinity (cal mol–1)
Kairei
100 000
Guaymas 100 000 EPR 9°N 50 000
Lau
0 Endeavour –50 000
–50 000 Endeavour
–100 000 0
50
100
150
200
TAG
–100 000 250
300
350
0
50
100
150
200
250
300
350
Temperature (°C)
Temperature (°C)
Fig. 8. Calculated affinities for the synthesis of aqueous carboxylic acids: propanoic, hexanoic, nonanoic and dodecanoic according to reactions (21–24) as functions of temperature as seafloor hydrothermal fluids mix with sea water at 250 bars.
6CO2 ðaqÞ þ 16H2 ðaqÞ ) CH3 ðCH2 Þ4 COOHðaqÞ ðhexanoic acidÞ þ 10H2 O; 9CO2 ðaqÞ þ 25H2 ðaqÞ ) CH3 ðCH2 Þ7 COOHðaqÞ ðnonanoic acidÞ þ 16H2 O;
ð22Þ
ð23Þ
and 12CO2 ðaqÞ þ 34H2 ðaqÞ ) CH3 ðCH2 Þ10 COOHðaqÞ ðdodecanoic acidÞ þ 22H2 O;
ð24Þ
together with values of aH2 O , aH2 ðaqÞ and aCO2 ðaqÞ from the seven mixing calculations. Resulting values of Qr were combined with equilibrium constants for these reactions, calculated with equations, data and parameters from Shock et al. (1989) and Shock (1995) to evaluate the affinities shown in Fig. 8. Comparison of the plots in Fig. 8 shows that with increasing molecular size the affinity curves steepen and spread apart. Those for more hydrogen-rich fluids such as Rainbow, Guaymas and Kairei reach increasingly positive affinities for increasingly larger carboxylic acids. Similar,
although less dramatic, results are found for fluids from EPR 9N, Lau and Endeavor, but those for TAG become more negative for increasingly larger compounds. Taking mixing results for the Rainbow fluid with sea water as an example, the maximum in the curve increases from about 40 kcal mol)1 for propanoic to about 190 kcal mol)1 for dodecanoic. This means that energy would be released by the formation of any of these organic acids, and that considerably more energy would be released during synthesis of each mole of larger acids like dodecanoic than for each mole of smaller acids like propanoic. Long-chain carboxylic acids, larger than dodecanoic, are integral to many microbial membranes, and these results suggest that the synthesis of such compounds would also be accompanied by the release of energy in many seafloor hydrothermal systems. The trends shown in Fig. 8 demonstrate that the disequilibria established as hydrothermal fluids mix with sea water not only lead to conditions where energy would be released by organic synthesis, but that the potential release of energy increases for larger compounds. One way of examining these results is to consider the stoichiometric coefficient on H2(aq) in these reactions, which increases
Seafloor hydrothermal systems 177 from 7 for propanoic acid to 34 for dodecanoic acid. These changes amplify the effects of the H2(aq) content of these fluids, as well as differences in H2(aq) contents among the various fluids. Another way to interpret these results is to consider the average oxidation state of carbon in each of these compounds, which can be calculated by assuming each O has a charge of )2 and each H has a charge of 1. The resulting values of the average carbon oxidation state for propoanic, hexanoic, nonanoic and dodecanoic acids are )0.67, )1.33, )1.55 and )1.67 respectively. Reduced conditions that prevail as hydrothermal fluids mix with sea water favor compounds with more reduced average oxidation states for carbon. It follows that more energy would be released per mole of formation of the larger acids.
AFFINITIES FOR BIOSYNTHESIS Conditions that favor abiotic organic synthesis are also energetically favorable for biosynthesis. It seems reasonable to imagine that microbes would take advantage of thermodynamically favorable conditions for making the organic compounds that constitute their cells. If so, then the consequences of fluid mixing relate directly to the thermodynamics of biosynthesis by thermophiles and hyperthermophiles.
The following discussion begins with amino acid syntheses and continues with carbohydrate, purine and pyrimidine syntheses. Amino acids are polymerized into proteins, carbohydrates serve as energy sources and as building blocks of DNA with purines, pyrimidines and other monomers. Amino acids Affinities determined for 10)6 activities of glycine, alanine, valine and leucine are plotted in Fig. 9 for the seven mixing calculations conducted in this study. Values of Qr were obtained with the values of aH2 O, aH2 ðaqÞ, aNHþ 4, aCO2 ðaqÞ and pH for each mixture as shown in Figs 2 and 3, and values of Kr were calculated with equations and parameters from Shock et al. (1989, 1997) and Dick et al. (2006) for the reactions 2CO2 ðaqÞþNHþ 4 þ3H2 ðaqÞ)CH2 NH2 COOHðaqÞ ðglycineÞþHþ þ2H2 O; 3CO2 ðaqÞþNHþ 4 þ6H2 ðaqÞ)CH3 CHNH2 COOHðaqÞðalanineÞþHþ þ4H2 O;
80 000 250 bars
250 bars 60 000
60 000
40 000 Guaymas Rainbow
20 000
Kairei Endeavour
0 –20 000 EPR 9°N
–40 000
Affinity (cal mol–1)
Affinity (cal mol–1)
ð26Þ
Alanine synthesis
Glycine synthesis 80 000
50
Rainbow
40 000
Guaymas Kairei
20 000
Lau
0
Endeavour
–20 000
EPR 9°N
Lau –40 000
TAG 0
100
150
200
250
300
TAG 0
350
50
100
150
200
250
300
350
Temperature (°C)
Temperature (°C)
Leucine synthesis
Valine synthesis 80 000
80 000 250 bars Rainbow Guaymas
EPR 9°N
20 000
Lau 0 Endeavour TAG
–20 000
Guaymas
60 000
Kairei
40 000
250 bars
Rainbow
Affinity (cal mol–1)
60 000
Affinity (cal mol–1)
ð25Þ
Kairei
40 000
EPR 9°N 20 000 Lau 0 Endeavour –20 000
TAG
–40 000
–40 000 0
50
100
150
200
250
Temperature (°C)
300
350
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 9. Calculated affinities at 250 bars for the synthesis of aqueous glycine, alanine, valine and leucine, based on reactions (25–28) as functions of temperature during mixing of seafloor hydrothermal fluids with sea water.
178 E. SHOCK & P. CANOVAS
5CO2 ðaqÞþNHþ 4 þ12H2 ðaqÞ)ðCH3 Þ2 ðCHÞ2 NH2 COOHðaqÞðvalineÞ þHþ þ8H2 O;
ð27Þ
and 6CO2 ðaqÞ þ NHþ 4 þ 15H2 ðaqÞ ) CH3 ðCH2 Þ3 CHNH2 COOHðaqÞ ðleucineÞ þ Hþ þ 10H2 O:
ð28Þ
As in the case of the carboxylic acids, the affinity curves are steeper and more splayed apart as the size of the amino acid increases from C2 glycine to C6 leucine. One consequence of these changes is that affinities for alanine, valine and leucine become positive in the EPR 9N mixture between about 50 and 150C, while those for glycine do not reach positive values at any temperature. Furthermore, the energy that would be released at 50C upon the formation of these acids in the Rainbow mixture increases from 5 kcal mol)1 for glycine to 20 kcal mol)1 for alanine to 50 kcal mol)1 for valine and about 70 kcal mol)1 for leucine. Rather than costing energy, as
amino acid synthesis does in low-temperature, oxidized, surface environments, all that is needed to trigger amino acid synthesis in seafloor hydrothermal ecosystems is catalysis. The positive enhancement of affinities as the size of amino acid molecules increases is also demonstrated by the plots in Fig. 10 for the C4 compounds aspartic acid and asparagine, and the C5 compounds glutamic acid and glutamine. Affinities were evaluated using activities of H2(aq), + H2O, CO2(aq), NHþ 4 and H from the seven mixing calculations, as well as the constraint that the amino acid activities are each 10)6, together with equilibrium constants calculated as for the amino acids in Fig. 9 for the reactions 4CO2 ðaqÞ þ NHþ 4 þ 6H2 ðaqÞ ) HOOCCH2 CHNH2 COOHðaqÞ ðaspartic acidÞ 4CO2 ðaqÞ þ 2NHþ 4 þ 6H2 ðaqÞ ) H2 NCOCH2 CHNH2 COOHðaqÞ ðasparagineÞ þ 2Hþ þ 5H2 O;
Aspartic acid synthesis
250 bars
40 000
40 000 Rainbow
20 000
Guaymas Kairei
0 Lau –20 000 Endeavour
EPR 9°N
Affinity (cal mol–1)
Affinity (cal mol–1)
ð30Þ
Asparagine synthesis
250 bars
–40 000
20 000
Rainbow
Kairei Lau
–20 000
Endeavour EPR 9°N
–40 000 0
50
100
150
200
250
300
Guaymas
0
TAG 350
0
50
Temperature (°C)
TAG
100
150
200
250
250 bars 40 000
Rainbow
Affinity (cal mol–1)
Guaymas 20 000 Kairei 0
Endeavour Lau
–20 000
EPR 9°N TAG
Rainbow Guaymas
20 000 Kairei 0 Lau –20 000
Endeavour EPR 9°N
–40 000
–40 000 0
50
100
150
200
250
Temperature (°C)
350
Glutamine synthesis
250 bars 40 000
300
Temperature (°C)
Glutamic acid synthesis
Affinity (cal mol–1)
ð29Þ
þ Hþ þ 4H2 O;
300
350
0
50
TAG
100
150
200
250
300
350
Temperature (°C)
Fig. 10. Calculated affinities for the synthesis of aqueous aspartic acid, asparagine, glutamic acid and glutamine based on reactions (29–32) as functions of temperature as seafloor hydrothermal fluids mix with sea water at 250 bars.
Seafloor hydrothermal systems 179
5CO2 ðaqÞ þ NHþ 4 þ 9H2 ðaqÞ ) HOOCðCH2 Þ2 CHNH2 COOHðaqÞ ðglutamic acidÞ þ Hþ þ 6H2 O;
ð31Þ
and 5CO2 ðaqÞ þ 2NHþ 4 þ 9H2 ðaqÞ ) H2 NCOðCH2 Þ2 CHNH2 COOHðaqÞ ðglutamineÞ þ 2Hþ þ 7H2 O:
ð32Þ
Note that affinities for the C5 acids (lower plots) are more positive for most of the mixing calculations than for their C4 counterparts (upper plots). These trends are analogous to those for the increasingly positive affinities documented for the successively larger carboxylic acids in Fig. 9 and amino acids in Fig. 10. Glutamic acid shown in Fig. 11 and valine in Fig. 10 are both C5 amino acids. The major difference between them is that glutamic is a dicarboxylic acid, while valine contains just one carboxyl group. Comparison of these figures shows that affinities for a 10)6 activity of glutamic
acid are less positive than those for the same activity of valine. Less positive affinities for a dicarboxylic acid relative to monocarboxylic acids are also evident in Fig. 7 where results for oxalic acid can be compared with those for glycolic acid or acetic acid. Apparently the trend in submarine hydrothermal systems toward lower stability for more oxygen-rich compounds observed for C2 compounds continues when considering amino acids. This theme returns in the discussion of results for carbohydrates below. The examination of amino acid synthesis can be extended to include several other compounds, and examples are shown in Fig. 11 for serine, proline, phenylalanine and tryptophan where the overall synthesis reactions are given by 3CO2 ðaqÞ þ NHþ 4 þ 5H2 ðaqÞ ) HOCH2 CHNH2 COOHðaqÞ ðserineÞ þ Hþ þ 3H2 O;
ð33Þ
5CO2 ðaqÞ þ NHþ 4 þ 11H2 ðaqÞ ) HNðCH2 Þ3 CHCOOHðaqÞ ðprolineÞ þ Hþ þ 8H2 O;
ð34Þ
Serine synthesis
Proline synthesis
100 000
100 000 250 bars
250 bars
Rainbow Guaymas Kairei
0 Endeavour EPR 9°N
–50 000
Affinity (cal mol–1)
Affinity (cal mol–1)
Rainbow 50 000
50 000
Guaymas Kairei
0 Endeavour EPR 9°N Lau
–50 000
Lau TAG
TAG –100 000
–100 000 0
50
100
150
200
250
300
350
0
50
Temperature (°C)
100
150
250
300
350
Tryptophan synthesis
Phenylalanine synthesis 100 000
100 000 250 bars
Rainbow
Guaymas
Kairei EPR 9°N
0
Lau Endeavour
–50 000
Affinity (cal mol–1)
50 000
250 bars
Rainbow
Guaymas
Affinity (cal mol–1)
200
Temperature (°C)
50 000
Kairei EPR 9°N
0 Lau Endeavour –50 000 TAG
TAG
–100 000
–100 000 0
50
100
150
200
250
Temperature (°C)
300
350
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 11. Calculated chemical affinities at 250 bars for the synthesis of aqueous serine, proline, phenylalanine and tryptophan based on reactions (33–36) as functions of temperature during mixing of submarine hydrothermal fluids with sea water.
180 E. SHOCK & P. CANOVAS
9CO2 ðaqÞ þ NHþ 4 þ 20H2 ðaqÞ ) C6 H5 CH2 CHNH2 COOHðaqÞ ðphenylalanineÞ þ Hþ þ 16H2 O;
ð35Þ
and
their effect on carboxylic acids. Comparison of results for serine in Fig. 11 with those for alanine in Fig. 9 shows that affinities are less positive for the hydroxyl-bearing serine, analogous to the comparison of glycolic and acetic acids in Fig. 7. Carbohydrates, purines and pyrimidines
11CO2 ðaqÞ þ 2NHþ 4 þ 23H2 ðaqÞ ) C6 H4 NHCHCCH2 CHNH2 COOHðaqÞ ðtryptophanÞ þ 2Hþ þ 20H2 O:
ð36Þ
Affinities were calculated following the same procedures for quantifying Qr and Kr as described above for other amino acids. Again, the larger amino acids tend to yield more positive affinities for mixing calculations involving the most H2-rich fluids. As an example, for the Rainbow mixing calculation, affinities for phenylalanine reach values that are more positive than those for leucine shown in Fig. 9. At the same time, larger amino acids have less positive affinities in calculations for H2-poor fluids such as TAG, Lau and Endeavor (compare, for example, the results for proline with those for tryptophan). Hydroxyl groups on amino acids seem to have an effect similar to
Unlike carboxylic and amino acids, which have relatively high ratios of carbon to oxygen, calculations for carbohydrates yield less positive values of affinity. This is illustrated by the results shown in Fig. 12 for ribose, ribulose, deoxyribose and glucose. Equilibrium constants were calculated using data and parameters for ribulose and glucose from Amend & Plyasunov (2001) and ribose and deoxyribose from LaRowe & Helgeson (2006a) for the reactions 5CO2 ðaqÞ þ 10H2 ðaqÞ ) C5 H10 O5 ðaqÞ ðriboseÞ þ 5H2 O; 5CO2 ðaqÞ þ 10H2 ðaqÞ ) C5 H10 O5 ðaqÞ ðribuloseÞ þ 5H2 O;
20 000 250 bars
Rainbow
250 bars
Rainbow 0
Kairei –20 000 Guaymas –40 000
Lau
EPR 9°N
–60 000
Endeavour TAG
Affinity (cal mol–1)
0
Affinity (cal mol–1)
ð38Þ
Ribulose synthesis
Ribose synthesis 20 000
Kairei –20 000 Guaymas –40 000 Lau EPR 9°N
–60 000
Endeavour TAG
–80 000
–80 000
–100 000
–100 000 0
50
100
150
200
250
300
0
350
50
100
150
200
250
300
350
Temperature (°C)
Temperature (°C)
Glucose synthesis
Deoxyribose synthesis 20 000
20 000 250 bars
Rainbow
0
Kairei Guaymas
–20 000 EPR 9°N
–40 000
250 bars
Rainbow
Endeavour Lau
–60 000
Affinity (cal mol–1)
0
Affinity (cal mol–1)
ð37Þ
Kairei –20 000 Guaymas
–40 000
Endeavour –80 000
–80 000
Lau
EPR 9°N
–60 000
TAG
TAG –100 000
–100 000 0
50
100
150
200
250
Temperature (°C)
300
350
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 12. Calculated affinities for the synthesis of aqueous ribose, ribulose, deoxyribose and glucose based on reactions (37–40) as functions of temperature during mixing of submarine hydrothermal fluids with sea water at 250 bars.
Seafloor hydrothermal systems 181
5CO2 ðaqÞ þ 11H2 ðaqÞ ) C5 H10 O4 ðaqÞ ðdeoxyriboseÞ þ 6H2 O;
ð39Þ
and 6CO2 ðaqÞ þ 12H2 ðaqÞ ) C6 H12 O6 ðaqÞ ðglucoseÞ þ 6H2 O:
ð40Þ
Activity products were evaluated from the results of mixing calculations shown in Fig. 3, together with the constraint that the activities of the carbohydrates are set to 10)6. Given this constraint, the affinities for ribose, ribulose and glucose never attain positive affinities. By contrast, those for deoxyribose reach positive values for mixing calculations involving the fluids from Rainbow, Kairei and Guaymas. In fact, comparison of the plots in Fig. 12 shows that all of the affinity curves for deoxyribose are more positive than for the other carbohydrates in this study. As the only difference between ribose and deoxyribose is a single oxygen, the comparison between the plots for these two carbohydrates helps to focus on the effect of changing the carbon-to-oxygen ratio in carbohydrates. It is evident that the more reduced carbohydrate, deoxyribose, is more stable than its oxidized counterpart, ribose. It follows that synthesis of deoxyribose is less costly than the same activity of ribose at conditions throughout submarine hydrothermal ecosystems. Deoxyribose is linked with purines (adenine and guanine) and pyrimidines (thymine and cytosine) to form deoxynucleosides that ultimately form parts of the structure of DNA. Data and parameters for purines and pyrimidines from LaRowe & Helgeson (2006a) make it possible to calculate equilibrium constants for synthesis reactions given by 5CO2 ðaqÞ þ 5NHþ 4 þ 5H2 ðaqÞ ) C5 H5 N5 ðaqÞ ðadenineÞ þ 5Hþ þ 10H2 O; 5CO2 ðaqÞ þ 5NHþ 4 þ 4H2 ðaqÞ ) C5 H5 N5 OðaqÞ ðguanineÞ þ 5Hþ þ 9H2 O; 5CO2 ðaqÞ þ 2NHþ 4 þ 8H2 ðaqÞ ) C5 H6 N2 O2 ðaqÞ ðthymineÞ þ 2Hþ þ 8H2 O;
ð41Þ
ð42Þ
ð43Þ
and 4CO2 ðaqÞ þ 3NHþ 4 þ 5H2 ðaqÞ ) C4 H5 N3 OðaqÞ ðcytosineÞ þ 3Hþ þ 7H2 O;
ð44Þ
which were combined with activity products based on the mixing calculations conducted in this study to yield the affinities shown in Fig. 13 after setting activities of the purines and pyrimidines to 10)6. As in the cases of the carbohydrates shown in Fig. 12, most calculated affinities are negative for 10)6 activities of purines and pyrimidines
at the conditions of the mixing calculations. However, those for thymine become positive in the low-temperature ranges for Guaymas, Endeavour and Kairei. The order and arrangement of curves in Fig. 13 differ from those in other figures, owing to the presence of NHþ 4 in the reactions, and its stoichiometic reaction coefficients which are generally larger than most corresponding values for amino acids. As a consequence, calculations for the more ammonia-rich fluids, especially Guaymas, attain enhanced affinities. Likewise, calculated affinities for adenine and guanine in the Endeavour mixing calculations reach values that are more positive than those for the EPR 9N results, reflecting the two order of magnitude greater abundance of ammonia at Endeavour (see Table 1).
DISCUSSION The results summarized above are meant to provide a framework for thinking about the potential for organic synthesis and biosynthesis in submarine hydrothermal ecosystems. Reducing the entire variability of submarine hydrothermal systems to seven representative fluids from around the world is inherently incomplete but meant to capture much of the global diversity. In addition, selecting or assembling a fluid composition for an area of hydrothermal venting cannot reflect the variability within an individual location. It is hoped that local and global variations can be referenced to the set of calculations conducted in this study, at least in a comparative mode. The following discussion includes efforts to examine the global variability in the potential for abiotic organic synthesis and biosynthesis in submarine hydrothermal systems, to consider the energetic consequences for the overall synthesis processes and to summarize the types of variations that would lead to results that would go beyond the scope of the present study. Consequences of global variability As a preface, it should be emphasized that the results presented here are dependent on several assumptions. First, it should be kept in mind that dissolved hydrogen and oxygen are allowed to equilibrate with each other and with H2O throughout these calculations. This is not strictly realistic for fluid mixing, as there are many conditions at which this reaction departs from equilibrium. Evidence for this departure comes from the culturing of microbes capable of gaining their energy from this reaction. Nevertheless, using this constraint means that the resulting affinities for other reactions that are not allowed to equilibrate are minimum values. In a natural system resulting from fluid mixing, there can always be more energy than what was calculated in this study and depicted in the figures shown above. It should be kept in mind that assuming this
182 E. SHOCK & P. CANOVAS
Guanine synthesis
Adenine synthesis 20 000
20 000 250 bars
250 bars 0
Affinity (cal mol–1)
Affinity (cal mol–1)
0 Guaymas
–20 000
Rainbow
–40 000
Kairei
–60 000 Endeavour –80 000 –100 000
Lau
Guaymas Rainbow
–20 000
Kairei
–40 000
Endeavour
–60 000
Lau
–80 000 –100 000 –120 000
–120 000 EPR 9°N
–140 000 0
50
TAG
100
150
–140 000
200
250
300
350
EPR 9°N 0
50
Temperature (°C)
–60 000 –80 000 Endeavour –100 000
Rainbow Kairei
–20 000
Endeavour
–40 000 –60 000 –80 000 –100 000
Lau EPR 9°N
–120 000
–120 000
0
50
100
150
200
350
300
TAG
–140 000
TAG 250
250 bars
Guaymas
Affinity (cal mol–1)
Affinity (cal mol–1)
Lau
EPR 9°N
300
0
–40 000
–140 000
TAG 250
Cytosine synthesis 250 bars
Rainbow Kairei
–20 000
200
20 000
Guaymas
0
150
Temperature (°C)
Thymine synthesis 20 000
100
350
Temperature (°C)
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 13. Calculated affinities for the synthesis of aqueous adenine, guanine, tyrosine and cytosine based on reactions (41–44) as functions of temperature during mixing of submarine hydrothermal fluids with sea water at 250 bars.
equilibrium dramatically affects some of the low-temperature calculations as the abundance of dissolved oxygen in sea water overwhelms the amount of dissolved hydrogen in the hydrothermal fluid. Curves that plunge steeply may diverge from what could be measured in a natural system where hydrogen oxidation is kinetically inhibited. Another assumption about the results shown above is that none of the other oxidation–reduction reactions are allowed to proceed at all in these calculations. This means that the energy represented by affinities for each reaction depend on none of the other reactions having been initiated. A more realistic model would incorporate rates of all of the reactions considered so that the most likely reactions could be predicted. As those reactions would then be allowed to proceed, reactants and products would be consumed and produced, resulting in changes to the activity products and affinities for all of the other reactions. The absence of such kinetic data represents one of the largest gaps in the understanding of geochemistry. With these thoughts in mind, we can nevertheless make some generalizations. The results summarized above demonstrate that the potential for abiotic organic synthesis in submarine hydrothermal systems is the greatest for those fluids that contain
the highest concentrations of H2(aq). The composition selected to represent the Rainbow field contains the most H2(aq), and that for TAG contains the least; mixing calculations for Rainbow typically yield the most positive affinities, while those for TAG are typically the most negative. Plots for all compounds in the C–H–O chemical system (Figs 4–8, and 12) have Rainbow and TAG at the extremes of behavior except for oxalic acid in Fig. 7. It appears that the higher concentrations of dissolved inorganic carbon at Guaymas and Lau allow the corresponding oxalic acid affinities to exceed those for Rainbow. This is the first indication that H2(aq) concentration alone is not sufficient to predict accurately all of the relative locations of curves in the affinity plots shown above. Moving to the C–H–O–N system (Figs 9–11 and 13) reveals that the interplay of calculated affinities becomes considerably more nuanced than simply reflecting the relative magnitudes of H2(aq) contents. The synthesis reac+ tions for amino acids involve NHþ 4 and H , which, together with CO2(aq), H2(aq) and H2O means that there are five variable activities affecting the results. As the number of variables increases, predictability based on any one variable is less robust. Nevertheless, it is possible to gain
Seafloor hydrothermal systems 183 some insights concerning the trends linking fluid composition and affinities for synthesis reactions. Take, for example, the position of the calculated affinity curves for the Guaymas fluid in the plots for glycine in Fig. 9, asparagine and glutamine in Fig. 10 and all of the plots in Fig. 13. As shown in Table 1, the total concentration of ammonia in the Guaymas fluid exceeds most other fluids by more than an order of magnitude (the exception being the Lau fluid, which the Guaymas ammonia concentration exceeds by a factor of three). As the number of N atoms in the organic molecule increases, matched by the stoichiometric reaction coefficient of NHþ 4 , the effects of differing total ammonia concentrations are observable. These effects are the greatest for those compounds with the greatest N:C ratios. Thus, the effect is apparent for glycine, with N:C = 0.5 but not for the other amino acids in Fig. 9 (with N:C ratios of 0.33, 0.2 and 0.16). Likewise, the Guaymas mixing calculations shown in Fig. 10 yield affinities for glutamine and asparagine that exceed at many temperatures those for the Rainbow calculations because of the push supplied by the higher ammonia concentration at Guaymas for these compounds with comparatively higher N ⁄ C ratios than their dicarboxylic counterparts glutamic and aspartic acid. This trend continues and is magnified in the plots in Fig. 13. The details described above emphasize the point that variations in affinities derive from differences in composition. The results of this study allow some generalizations about how those differences affect the potential for abiotic organic synthesis and biosynthesis. Hydrogen content of vent fluid is a major determinant of synthesis potential, and systems hosted in ultramafic rocks will yield greater affinities for organic synthesis reaction than their basaltic, andesitic or rhyolitic counterparts. This is shown by the results for the Rainbow fluid used in this study, and also for the results from Kairei, where fluids seems to be affected by mafic or ultramafic rocks somewhere in the reaction zone. Sedimented systems can also exhibit enhanced hydrogen contents as represented here by the Guaymas fluid. Depending on the content of organic matter in the sediments, such fluids may be enriched in ammonia, which will enhance the potential for synthesis of N-bearing organic compounds like amino acids, purines and pyrimidines, as well as the protein and nucleic acid polymers made from them. Hydrothermal systems hosted at divergent boundaries are not likely to accumulate sediments unless they occur near the continents, but back-arc spreading may accumulate continental-derived sediment far more commonly. The results for the Lau Basin fluid selected in this study may not extrapolate well to other back-arc basin settings nearer to continental or large island arc sediment sources. Nevertheless, these results show what can happen as fluid compositions are influenced by rocks with more silica than what is typically associated with
divergent boundary basalt-hosted systems. However, even in basalt-hosted systems there can be some major differences in synthesis potential based on the fluids chosen for this study. Fluids like the EPR 9N sample that are relatively enriched in H2 provide greater potential for organic synthesis during mixing with sea water than H2-poor samples such as the Endeavour and TAG fluids chosen here. In fact, results for these three fluids can span a considerable range in affinities for given reactions. In many cases the results from EPR 9N and TAG are at the extremes for basalt-hosted systems, and those for Endeavour fall about midway between. This framework should make it possible for estimates to be made for other basalt-hosted systems by comparing compositional data with these three examples. Energetics of organic synthesis and biosynthesis Positive affinities for methanogenesis, as determined in this study (Fig. 4), are a necessity if that metabolic strategy supports autotrophic microbes in hydrothermal ecosystems. The geologic system at ridge environments allows mixing of fluids that are dramatically different in composition, especially for oxidation–reduction processes, and the resulting thermodynamic drive is for methane to be formed from dissolved carbon dioxide and hydrogen. Mechanistic difficulties supply a kinetic barrier to this reaction proceeding on its own; so, catalysts in the form of microbes inhabit the system, trigger the reaction and reap some of the energy that is released. Positive affinities for other organic synthesis reactions might be anticipated owing to how well the geologic system provides for methanogens. Nevertheless, anticipating the magnitude of the thermodynamic drives for abiotic organic synthesis and biosynthesis of various organic compounds may not be intuitive. After all, we live at conditions where organic synthesis in all of its forms is energetically costly. A plant at the surface conducting photosynthesis is working against a 20% O2 atmosphere to reduce CO2 to sugar. Copious energy in the form of solar photons allows the plant to fight this uphill battle and prevail. We take advantage of the energy transferred from the sun by the plant either by directly consuming the plant, or another animal that consumed the plant. The trapped solar energy is dissipated by the surface food web in which we reside, after the initial investment is made by the photosynthetic organisms. The situation in submarine hydrothermal ecosystems is radically different. Positive affinities for the formation of ethane and acetic acid (Figs 6 and 7) suggest that autotrophic ethanogenesis and acetogenesis are metabolic strategies that would release energy and support life in submarine hydrothermal ecosystems. Autotrophic acetogenesis is relatively common (Amend & Shock 2001) and indications of ethanogenesis
184 E. SHOCK & P. CANOVAS (and propanogenesis) are reported based on concentrations and isotopic analyses from deep marine sediments (Hinrichs et al. 2006). Based on the results in Fig. 6, seafloor hydrothermal ecosystems derived from fluids that have reacted with peridotites would be excellent environments in which to continue the search for ethanogenesis and the mysterious ethanogens. The same conditions favor the production of methanol, ethanol, acetaldehyde and ethene, suggesting that additional metabolic discoveries may be anticipated. The potential for amino acid synthesis in seafloor hydrothermal systems was explored earlier using a single composition for a 100C mixture (Amend & Shock 1998). The results obtained in this study for the dozen amino acids depicted in Figs 9–11 suggest that conditions for amino acid synthesis may regularly be far more favorable than that earlier study showed. As examples, the energies released by synthesis of phenylalanine, tryptophan and leucine from the Kairei, Guaymas and Rainbow mixed fluids at 100C all exceed values obtained in the earlier study by factors of two to three. Affinities for these three amino acids in the EPR 9N mixture at temperatures between about 50 and 90C are also more positive than in the 100C mixture used by Amend & Shock (1998). However, in the mixtures generated from the Lau, Endeavour and TAG fluids, syntheses of these amino acids are far from being so favorable. As in the earlier study, there are amino acids, including glycine, asparagine and serine, for which synthesis may not be accompanied by a release of energy. It remains to be seen if these amino acids are less common in the proteins of marine thermophiles and hyperthermophiles, but the increasing availability of genome sequences for these microbes facilitates conducting such tests enormously.
the possibility of fluid compositions that could diverge dramatically from what is presented here. Similarly, a single fluid from the Lau back-arc basin can hardly even do that complex system justice, and is far from representing all of the possibilities that may emerge in back-arc systems given the potential for fractionation to produce a variety of silicarich rocks, and the potential for back-arc spreading to occur near sources of continental sediments. The results for Guaymas show the dramatic changes that continental sediments can impose on a basalt-hosted system, and other compositions await exploration. Results presented here for carboxylic acids, carbohydrates, purines and pyrimidines provide the foundation for exploring the energetics of biosynthesis for many biomolecules. The trend shown by the series of carboxylic acids in Fig. 8 is enticing when extrapolated to lengths representative of molecules found in microbial membranes, but calculations for the appropriate compounds are not yet possible. Likewise, the energetic differences of forming ester and ether bonds, which characterize the differences in archaeal and bacterial membranes, remain to be explored in mixed hydrothermal fluids. Results for ribose, deoxyribose and the purines and pyrimidines suggest that synthesis of genetic material may be costly throughout submarine hydrothermal ecosystems. This can be tested by calculating affinities for nucleotides, nucleosides and their polymerized versions that lead to DNA and RNA. At the same time, it is also possible to evaluate the energy required to make adenosine triphosphate and other molecules involved in the transfer of energy within cells. Taken together, these sorts of results will ultimately make it possible to assess the energy demands of the inhabitants of submarine hydrothermal ecosystems.
CONCLUDING REMARKS Extrapolating beyond the present study The range represented by the seven fluids picked for the present study covers much of the known range of submarine hydrothermal fluid compositions. However, it should be anticipated that what is known may be only a small part of the picture. Basalts are common on the seafloor; so, many systems may produce environments that fall somewhere in the range indicated by the results for TAG, Endeavour and EPR 9N presented here. At the same time, these three fluid compositions do not exhaust the possibilities for basalt-hosted systems, and barely explore the myriad possibilities that keep petrologists sampling seafloor basalts. Likewise, the full range of fluid compositions that may be produced in ultramafic-hosted systems cannot be simulated by the sparse set of examples chosen here. Known ultramafic-hosted systems are in peridotites, but it may be that seafloor hydrothermal systems are hosted in pyroxenites or other varieties of ultramafic rocks, raising
Conditions established as hydrothermal fluids mix with sea water favor the formation of organic compounds at the expense of inorganic starting materials. The mixtures are far-from-equilibrium chemical systems in which the formation of organic compounds would help to dissipate the pent-up energy held by the disequilibrium states. That this is quite an unfamiliar situation compared with the facts of life at the Earth’s surface is evident. Conditions in hydrothermal ecosystems are not well represented by any familiar surface setting. Abiotic organic synthesis has a compelling thermodynamic drive in these systems. Similarly, the energetics of biosynthesis presents a strong contrast to anything familiar. Autotrophic microbes in submarine hydrothermal ecosystems can make many of the monomers that compose their biomolecules by catalyzing already favorable reactions. The overall thermodynamic costs of many biosynthesis reactions can be negative, and may exceed the positive costs of making
Seafloor hydrothermal systems 185 other biomolecules, supporting the conclusion that for many autotrophs in hydrothermal ecosystems biosynthesis acts as ‘a free lunch that they are paid to eat’ (Shock et al. 1998).
ACKNOWLEDGEMENTS This work was funded in part through NSF grants OCE0752541 and OCE-097406. The authors appreciate helpful discussions with Jeff Dick during the course of this study, and the pioneering efforts of Tom McCollom, Jan Amend and Mitch Schulte who helped initiate this type of theoretical analysis.
REFERENCES Amend JP, Helgeson HC (1997a) Calculation of the standard molal thermodynamic properties of aqueous biomolecules at elevated temperatures and pressures. 1. L-a-amino acids. Journal of the Chemical Society, Faraday Transactions, 93, 1927–41. Amend JP, Helgeson HC (1997b) Group additivity equations of state for calculating the standard molal thermodynamic properties of aqueous organic species at elevated temperatures and pressures. Geochimica et Cosmochimica Acta, 61, 11–46. Amend JP, Helgeson HC (2000) Calculation of the standard molal thermodynamic properties of aqueous biomolecules at elevated temperatures and pressures II. Unfolded proteins. Biophysical Chemistry, 84, 105–36. Amend JP, Plyasunov AV (2001) Carbohydrates in thermophile metabolism: calculation of the standard molal thermodynamic properties of aqueous pentoses and hexoses at elevated temperatures and pressures. Geochimica et Cosmochimica Acta, 65, 3901–17. Amend JP, Shock EL (1998) Energetics of amino acid synthesis in hydrothermal ecosystems. Science, 281, 1659–62. Amend JP, Shock EL (2001) Energetics of overall metabolic reactions in thermophilic and hyperthermophilic Archaea and Bacteria. FEMS Microbiology Reviews, 25, 175–243. Baker ET, German CR (2004) On the global distribution of midocean ridge hydrothermal vent-fields. In: The Thermal Structure of the Oceanic Crust and the Dynamics of Seafloor Hydrothermal Circulation, Geophysical Monograph Series, Vol. 148 (eds German CR, Lin J, Parson LM), pp. 245–66. American Geophysical Union, Washington, DC. Baker E, German CR, Elderfield H (1995) Hydrothermal plumes: global distributions and geological inferences. In: Seafloor Hydrothermal Systems: Physical, Chemical, Biological and Geological Interactions, Geophysical Monograph, Vol. 91 (eds Humphris SE, Zierenberg RA, Mullineaux LS, Thomson RE), pp. 47–71. American Geophysical Union, Washington, DC. Baker ET, Massoth GJ, Nakamura K, Embley RW, de Ronde CEJ, Arculus RJ (2005) Hydrothermal activity on near-arc sections of back-arc ridges: results from the Mariana Trough and Lau Basin. Geochemistry Geophysics Geosystems, 6, Q09001, doi: 10.1029/ 2005GC000948. Baker ET, Resing JA, Walker SL, Martinez F, Taylor B, Nakamura K (2006) Abundant hydrothermal venting along melt-rich and melt-free ridge segments in the Lau back-arc basin. Geophysical Research Letters, 33, L07308, doi: 10.1029/ 2005GL025283.
Baross JA, Lilley MD, Gordon LI (1982) Is the CH4, H2 and CO venting from submarine hydrothermal systems produced by thermophilic bacteria? Nature, 298, 366–8. Berndt ME, Allen DE, Seyfried WE Jr (1996) Reduction of CO2 during serpentinization of olivine at 300C and 500 bar. Geology, 24, 351–4. Bird P (2003) An updated digital model of plate boundaries. Geochemistry Geophysics Geosystems, 4, 1027, doi: 10.1029/ 2001GC000252. Bruland KW (1983) Trace elements in seawater. In: Chemical Oceanography, Vol. 8 (eds Riley JP, Chester R), pp. 157–220. Academic Press, London. Butterfield DA, McDuff E, Mottl MJ, Lilley MD, Lupton JE, Massoth GJ (1994) Gradients in the composition of hydrothermal fluids from the Endeavour segment vent field: phase separation and brine loss. Journal of Geophysical Research, 99, 9561–83. Butterfield DA, Roe KK, Massoth GJ, Embley RW, McDuff RE (1995) Spatial and temporal variability in hydrothermal vent fluid composition along the Juan de Fuca Ridge. EOS. Transactions of the American Geophysical Union, 76, 420. Calvert SE (1966) Accumulation of diatomaceous silica in the sediments of the Gulf of California. Geological Society of America Bulletin, 77, 569–96. Campbell AC, Palmer MR, Klinkhammer GP, Bowers TS, Edmond JM, Lawrence JR, Casey JF, Thompson G, Humphris S, Rona PA, Karson JA (1988a) Chemistry of hot springs on the Mid-Atlantic Ridge. Nature, 335, 514–9. Campbell AC, Bowers TS, Edmond JM (1988b) A time series of vent fluid compositions from 21N, EPR (1979, 1981 and 1985) and the Guaymas Basin, Gulf of California (1982, 1985). Journal of Geophysical Research, 93, 4537–49. Cannat M, Briais A, Deplus C, Escartı´n J, Georgen J, Lin J, Mercouriev S, Meyzen C, Muller M, Pouliquen G, Rabain A, da Silva P (1999) Mid-Atlantic Ridge–Azores hotspot interactions: along-axis migration of a hotspot-derived event of enhanced magmatism 10 to 4 Ma ago. Earth and Planetary Science Letters, 173, 257–69. Carbotte SM, Macdonald KC (1992) East Pacific Rise 88108300N: evolution of ridge segments and discontinuities from SeaMARC II and three-dimensional magnetic studies. Journal of Geophysical Research, 97, 6959–82. Charlou JL, Donval JP, Erzinger J, Fouquet Y, Von Stackelberg U (1991) Hydrothermal activity in the Lau Back-Arc Basin: plumes and hot fluid chemistry. Terra Abstract, 3, 406. Charlou JL, Donval JP, Jean-Baptiste P, Dapoigny A, Rona PA (1996) Gases and helium isotopes in high temperature solutions sampled before and after ODP Leg 158 drilling at TAG hydrothermal field (26N MAR). Geophysical Research Letters, 23, 3491–4. Charlou JL, Fouquet Y, Bougault H, Donval JP, Etoubleau J, JeanBabtiste P, Dapoigny A, Appriou P, Rona PA (1998) Intense CH4 plumes generated by serpentinization of ultramafic rocks at the intersection of the 1520¢N fracture zone and the Mid-Atlantic Ridge. Geochimica et Cosmochimica Acta, 62, 2323–33. Charlou JL, Donval JP, Fouquet Y, Jean-Baptiste P, Holm N (2002) Geochemistry of high H2 and CH4 vent fluids issuing from ultramafic rocks at the Rainbow hydrothermal field (3614¢N, MAR). Chemical Geology, 191, 345–59. Chiba H, Masuda H, Lee SY, Fujioka K (2001) Chemistry of hydrothermal fluids at the TAG active mound, MAR 26N, in 1998. Geophysical Research Letters, 28, 2919–22. Collier JS, Sinha MC (1992) Seismic mapping of a magma chamber beneath the Valu Fa Ridge, Lau Basin. Journal of Geophysical Research, 97, 14031–53.
186 E. SHOCK & P. CANOVAS Cowen JP, Fornari D, Shank TM, Love B, Glazer B, Treusch A, Holmes RC, Soule A, Baker ET, Tolstoy M, Pomranig KR (2007) Volcanic eruptions at East Pacific Rise near 95¢N. EOS. Transactions of the American Geophysical Union, 88, 81. Curray JR, Moore DG, Aguayo JE, Aubry MP, Einsele G, Fornari DJ, Gieskes J, Guerrero JC, Kastner M, Kelts K, Lyle M, Matoba Y, Molina-Cruz A, Niemitz J, Rueda J, Saunders AD, Sehrader H, Simoneit BRT, Vacquier V (1982) Guaymas Basin: sites 477, 478 and 481. Initial Reports of the Deep Sea Drilling Project 64, Part I, 211–87. Delaney JR, Robigou V, McDuff RE, Tivey MK (1992) Geology of a vigorous hydrothermal system on the Endeavour Segment, Juan de Fuca Ridge. Journal of Geophysical Research, 97, 19663–82. Delaney JR, Kelley DS, Lilley MD, Butterfield DA, McDuff RE, Baross JA, Deming JW, Johnson HP, Robigou V (1997) The Endeavour hydrothermal system I: cellular circulation above an active cracking front yields large sulfide structures, ‘‘fresh’’ vent water and hyperthermophilic Archaea. RIDGE Events, 8, 11–9. DeMets C, Gordon RG, Argus DF, Stein S (1990) Current plate motions. Geophysical Journal International, 101, 425–78. Desbruyeres D, Alayse-Danet AM, Ohta S, Antoine E, Barbier G, Briand P, Godfroy A, Crassous P, Jollivet D, Kerdoncuff J, Khripounoff A, Laubier L, Marchand M, Perron R, Derelle E, Dinet A, Fialamedioni A, Hashimoto J, Nojiri Y, Prieur D, Ruellan E, Soakai S (1994) Deep-sea hydrothermal communities in southwestern Pacific Back-Arc Basins (the North Fiji and Lau Basins): composition, microdistribution and food-web. Marine Geology, 116, 227–42. Detrick RS, Buhl P, Vera E, Mutter J, Orcutt J, Madsen J, Brocher T (1987) Multi-channel seismic imaging of a crustal magma chamber along the East Pacific Rise. Nature, 326, 35–41. Detrick RS, Carbotte S, van Ark E, Canales JP, Kent G, Harding A, Diebold J, Nedimovic M (2002) New multichannel seismic constraints on the crustal structure of the Endeavour Segment, Juan de Fuca Ridge: evidence for a crustal magma chamber. EOS. Transactions of the American Geophysical Union, 83, Fall Meet. Suppl., Abstract T12B-1316. Dick HJB, Natland JH, Alt JC, Bach W, Bideau D, Gee JS, Haggas S, Hertogen JGH, Hirth G, Holm PM, Ildefonse B, Iturrino GJ, John BE, Kelley DS, Kikawa E, Kingdon A, LeRoux PJ, Maeda J, Meyer PS, Miller DJ, Naslund HR, Niu Y-L, Robinson PT, Snow J, Stephen RA, Trimby PW, Worm H-U, Aaron Y (2000) A long in situ section of the lower oceanic crust: results of ODP Leg 176 drilling at the Southwest Indian Ridge. Earth and Planetary Science Letters, 179, 31–51. Dick JM, LaRowe DE, Helgeson HC (2006) Group additivity calculation of the standard molal thermodynamic properties of aqueous amino acids, polypeptides, and unfolded proteins as a function of temperature, pressure and ionization state. Biogeosciences, 3, 311–36. Ding K, Seyfried WE Jr (1992) Determination of Fe–Cl complexing in the low pressure supercritical region (NaCl fluid)-Iron solubility constraints on pH of subseafloor hydrothermal fluids. Geochimica et Cosmochimica Acta, 56, 3681–92. Donval JP, Charlou JL, Douville E, Knoery J, Fouquet Y, Poncevera E, Jean-Baptiste P, Stievenard M, German C (1997) High H2 and CH4 content in hydrothermal fluids from Rainbow site newly sampled at 3614¢N on the AMAR segment, Mid-Atlantic Ridge (diving FLORES cruise, July 1997). Comparison with
other MAR sites. EOS. Transactions of the American Geophysical Union, 78, 832. Douville E, Charlou JL, Oelkers EH, Bienvenu P, Jove Colon CF, Donval JP, Fouquet Y, Prieur D, Appriou P (2002) The rainbow vent fluids (3614¢N, MAR): the influence of ultramafic rocks and phase separation on trace metal content in Mid-Atlantic Ridge hydrothermal fluids. Chemical Geology, 184, 37–48. Edmond JM, Campbell AC, Palmer MR, Kinkhammer GP, German CR, Edmonds HN, Elderfield H, Thompson G, Rona P (1995) Time series of vent fluids from the TAG and MARK sites (1986, 1990) Mid-Atlantic Ridge: a new solution chemistry model and a mechanism for Cu ⁄ Zn zonation in massive sulphide orebodies. In: Hydrothermal Vents and Processes (eds Parson LM, Walker CL, Dixon DR), Geological Society of London Special Publications, 77–86. Einsele G (1982) Mechanisms of sill intrusion into soft sediment and expulsion of pore water. Initial Reports of the Deep Sea Drilling Project, 64, Part II, 1169–76. Einsele G, Gieskes J, Curray J, Moore D, Aguayo E, Aubry M-P, Fornari D, Guerrero J, Kastner M, Kelts K, Lyle M, Matoba M, Molina-Cruz A, Niemitz J, Rueda J, Saunders A, Sehrader H, Simoneit B, Vacquier V (1980) Intrusion of basaltic sills into highly porous sediments and resulting hydrothermal activity. Nature, 283, 442–5. Engels JL, Edwards MH, Fornari DJ, Perfit MR, Cann JR (2003), A new model for submarine volcanic collapse. Geochemistry Geophysics Geosystems, 4, 1077, doi: 10.1029/ 2002GC000483. Evans SC, White LD, Rapp JB (1988) Geochemistry of some gases in hydrothermal fluids from the southern Juan de Fuca Ridge. Journal of Geophysical Research, 93, 15303–13. Fornari DJ, Haymon RM, Edwards MH, Macdonald KC (1990) Volcanic and tectonic characteristics of the East Pacific Rise crest 909¢N to 954¢N: implications for fine-scale segmentation of the plate boundary. EOS. Transactions of the American Geophysical Union, 71, 625. Fornari DJ, Shank T, Van Damm KL, Gregg TKP, Lilley M, Levai G, Bray A, Haymon RM, Perfit MR, Lutz R (1998a) Time-series temperature measurements at high temperature hydrothermal vents, East Pacific Rise 949¢-51¢N: evidence for monitoring a crustal cracking event. Earth and Planetary Science Letters, 160, 419–31. Fornari DJ, Haymon RM, Perfit MR, Gregg TKP, Edwards MH (1998b) Axial summit trough of the East Pacific Rise 9N to 10N: geological characteristics and evolution of the axial zone on fast-spreading mid-ocean ridges. Journal of Geophysical Research, 103, 9827–55. Fornari D, Tivey M, Schouten H, Perfit MR, Yoerger DR, Bradley A, Edwards M, Haymon R, Scheirer D, Von Damm KL, Shank T, Soule A (2004) Submarine lava flow emplacement at the East Pacific Rise 95¢N: implications for uppermost ocean crust stratigraphy and hydrothermal fluid circulation. In: Mid-ocean ridges: hydrothermal interactions between the lithosphere and oceans (eds German CR, Lin J, Parson LM), pp. 187–218. American Geophysical Union, Washington, DC. Fouquet Y, Von Stackelberg U, Charlou L, Donval JP, Erzinger J, Foucher JP, Herzig P, Mu¨he R, Soakai S, Wiedicke M, Whitechurch H (1991a) Hydrothermal activity and metallogenesis in the Lau Back-Arc Basin. Nature, 349, 778–80. Fouquet Y, Von Stackelberg U, Charlou JL, Donval JP, Foucher JP, Erzinger J, Herzig P, Mu¨he R, Wiedicke M, Soakai S,
Seafloor hydrothermal systems 187 Whitechurch H (1991b) Hydrothermal activity in the Lau back-arc basin: sulfides and water chemistry. Geology, 19, 303– 6. Fouquet Y, Von Stackelberg U, Charlou JL, Erzinger J, Herzig PM, Mu¨he R, Wiedicke M (1993) Metallogenesis in back-arc environments: the Lau Basin example. Economic Geology. Bulletin of the Society of Economic Geologists, 88, 2154–81. Fouquet Y, Charlou JL, Ondreas H, Radford-Knoery J, Donval JP, Douville E, Apprioual R, Cambon P, Landur JY, Normand A, Ponsevera E, German C, Parson L, Barriga F, Costa I, Relvas J, Ribeiro A (1997) Discovery and first submersible investigations on the Rainbow hydrothermal field on the MAR (3614¢N). EOS. Transactions of the American Geophysical Union, 78, 832. Foustoukos DI, Seyfried WE Jr (2004) Hydocarbons in hydrothermal vent fluids: the role of chromium-bearing catalysts. Science, 304, 1002–5. Foustoukos DI, Savov IP, Janecky DR (2008) Chemical and isotopic constraints on water ⁄ rock interactions at the Lost City hydrothermal field 30N Mid-Atlantic Ridge. Geochimica et Cosmochimica Acta, 72, 5457–74. Gallant RM, Von Damm KL (2006) Geochemical controls on hydrothermal fluids from the Kairei and Edmond Vent Fields, 23–25S, Central Indian Ridge. Geochemistry Geophysics Geosystems, 7, Q06018. doi: 10.1029/2005GC001067. Gamo T (1995) Wide variation of chemical characteristics of submarine hydrothermal fluids due to secondary modification processes after high temperature water–rock interaction: a review. In: Biogeochemical Processes and Ocean Flux in the Western Pacific (eds Sakai H, Nozaki Y), pp. 425–45. ERRAPUB, Tokyo. Gamo T, Chiba H, Yamanaka T, Okudaira T, Hashimoto J, Tsuchida S, Ishibashi J, Kataoka S, Tsunogai U, Okamura K, Sano Y, Shinjo R (2001) Chemical characteristics of newly discovered black smoker fluids and associated hydrothermal plume at the Rodriguez Triple Junction, Central Indian Ridge. Earth and Planetary Science Letters, 193, 371–9. German CR, Von Damm KL (2004), Hydrothermal processes. In: Treatise on Geochemistry, Vol. 6 (eds Holland HD, Turekian KK), pp. 181–222. Elsevier, New York. German CR, Baker ET, Connelly DP, Lupton JE, Resing J, Prien R, Walker SD, Edmonds HN, Langmuir C (2006) Hydrothermal exploration of the Fonualei Rift and Spreading Center and the Northeast Lau Spreading Center. Geochemistry Geophysics Geosystems, 7, doi:10.1029/2006GC001324. German CR, Yoerger DR, Jakuba M, Shank TM, Langmuir CH, Nakamura K (2008) Hydrothermal exploration with the Autonomous Benthic Explorer. Deep-Sea Research I, 55, 203–19. Gieskes JM, Elderfield H, Lawrence JR, Johnson J, Meyers B, Campbell A (1982a) Geochemistry of interstitial waters and sediments, Leg 64, Gulf of California. Initial Reports of the Deep Sea Drilling Project, 64, Part II, 675–94. Gieskes JM, Kastner M, Einsele G, Kelts K, Niemitz J (1982b) Hydrothermal activity in the Guaymas Basin, Gulf of California: a synthesis. Initial Reports of the Deep Sea Drilling Project, 64, Part II, 1159–68. Gieskes JM, Simoneit BRT, Brown T, Shaw T, Wang Y-C, Magenheim A (1988) Hydrothermal fluids and petroleum in surface sediments of Guaymas Basin, Gulf of California: a case study. Canadian Mineralogist, 26, 589–602. Gieskes JM, Shaw T, Brown T, Sturz A, Campbell A (1991) Interstitial water and hydrothermal water chemistry, Guaymas Basin, Gulf of California (Edited by J. P. Dauphin and B. R. T. Sim-
oneit). American Association of the Petroleum Geology Memoir, 47, 753–79. Glickson DA, Kelley DS, Delaney JR (2007) Geology and hydrothermal evolution of the Mothra hydrothermal field, Endeavour Segment, Juan de Fuca Ridge. Geochemistry Geophysics Geosystems, 8, 1525–2027. Gregg TKP, Fornari DJ, Perfit MR, Haymon RM, Fink JH (1996) Rapid emplacement of a mid-ocean ridge lava flow on the East Pacific Rise at 946¢–51¢N. Earth and Planetary Science Letters, 144, E1–7. Hannington MD, Galley AD, Herzig PM, Petersen S (1998) Comparison of the TAG mound and stockwork complex with Cyprus-type massive sulfide deposits. In: Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 158 (eds Herzig PM, Humphris SE, Miller DJ, Zierenberg RA), pp. 389–415. Ocean Drilling Program, College Station, TX. Hannington MD, de Ronde CEJ, Petersen S (2005) Sea-floor tectonics and submarine hydrothermal systems. Economic Geology, 100, 111–41. Hashimoto J, Ohta S, Gamo T, Chiba H, Yamaguchi T, Tsuchida S, Okudaira T, Watabe H, Yamanaka T, Kitazawa M (2001) First hydrothermal vent communities from the Indian Ocean discovered. Zoological Science, 18, 717–21. Haymon RM, Kastner M (1981) Hot spring deposits on the East Pacific Rise at 21N: preliminary description of mineralogy and genesis. Earth and Planetary Science Letters B, 53, 363–81. Haymon R, Fornari DJ, Edwards MH, Carbotte S, Wright D, Macdonald KC (1991) Hydrothermal vent distribution along the East Pacific Rise crest 909¢–54¢N and its relationship to magmatic and tectonic processes on fast-spreading mid-ocean ridges. Earth and Planetary Science Letters, 104, 513–34. Haymon RM, Fornari DJ, Von Damm KL, Lilley MD, Perfit MR, Edmond JM, Shanks WCI, Lutz RA, Grebmeier JM, Carbotte S, Wright D, McLaughlin E, Smith M, Beedle N, Olson E (1993) Volcanic eruption of the mid-ocean ridge along the East Pacific Rise crest at 9 degrees 45-520N: direct submersible observations of seafloor phenomena associated with an eruption event in April, 1991. Earth and Planetary Science Letters, 119, 85–101. Helgeson HC (1979) Mass transfer among minerals and hydrothermal solutions. In: Geochemistry of Hydrothermal Ore Deposits (ed. Barnes HL), pp. 568–610. John Wiley, New York. Herzig PM, Hannington MD, Fouquet Y, von Stackelberg U, Petersen S (1993) Gold-rich polymetallic sulfides from the Lau back-arc and implications for the geochemistry of gold in seafloor hydrothermal systems in the Southwest Pacific. Economic Geology. Bulletin of the Society of Economic Geologists, 88, 2182– 209. Herzig PM, Hannington MD, Arribas A Jr (1998) Sulfur isotopic composition of hydrothermal precipitates from the Lau backarc: implications for magmatic contributions to seafloor hydrothermal systems. Mineralium Deposita, 33, 226–37. Hinkley TK, Tatsumoto M (1987) Metals and isotopes in Juan de Fuca Ridge hydrothermal fluids and their associated solid materials. Journal of Geophysical Research, 92, 11400–10. Hinrichs K-U, Hayes JM, Bach W, Spivak AJ, Hmelo LR, Holm NG, Johnson CG, Sylva SP (2006) Biological formation of ethane and propane in the deep marine subsurface. Proceedings of the National Academy of Sciences USA, 103, 14684–9. Holm NG, Charlou JL (2001) Initial indications of abiotic formation of hydrocarbons in the Rainbow ultramafic hydrothermal system, Mid-Atlantic Ridge. Earth and Planetary Science Letters, 191, 1–8.
188 E. SHOCK & P. CANOVAS Humphris SE, Tivey MK (2000) A synthesis of geological and geochemical investigations of the TAG hydrothermal field; insights into fluid-flow and mixing processes in a hydrothermal system. In: Ophiolites and Ocean Crust: New Insights from Field Studies and the Ocean Drilling Program (eds Dilek Y, Moores EM, Elthon D, Nicolas A), Geological Society of America Special Paper, 349. 213–35. Humphris SE, Herzig PM, Miller DJ, Alt JC, Becker K, Brown D, Burgmann G, Chiba H, Fouquet Y, Gemmell JB, Guerin G, Hannington MD, Holm NG, Honnorez JJ, Iturrino GJ, Knott R, Ludwig R, Nakamura K, Petersen S, Reysenbach A-L, Rona PA, Smith S, Sturz AA, Tivey MK, Zhao X (1995) The internal structure of an active sea-floor massive sulphide deposit. Nature, 377, 713–6. Ishibashi J, Grimaud D, Nojiri Y, Auzende J-M, Ruabe T (1994a) Fluctuation of chemical components of the phase-separated hydrothermal fluid from the North Fiji Basin Ridge. Marine Geology, 116, 215–26. Ishibashi J, Wakita JH, Nojiri Y, Grimaud D, Jean-Baptiste P, Gamo T, Auzende J-M, Urabe T (1994b) Helium and carbon geochemistry of hydrothermal fluids from the North Fiji Basin spreading ridge (southwest Pacific). Earth and Planetary Science Letters, 128, 183–97. Ishibashi J, Lupton JE, Yamaguchi T, Querellou J, Nunoura T, Takai K (2006) Expedition reveals changes in Lau Basin hydrothermal system. EOS. Transactions of the American Geophysical Union, 87, 13–6. Janecky DR, Seyfried WE Jr (1986) Hydrothermal serpentinization of peridotite within the oceanic crust: experimental investigations of mineralogy and major element chemistry. Geochimica et Cosmochimica Acta, 50, 1357–78. Johnson JW, Oelkers EH, Helgeson HC (1992) SUPCRT92: a software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000C. Computers & Geosciences, 18, 932–4. Kappel ES, Ryan WBF (1986) Volcanic episodicity and a nonsteady state rift valley along northeast Pacific spreading centers: evidence from Sea MARC I. Journal of Geophysical Research, 91, 13,925–13,940. Kastner M (1982) Evidence for two distinct hydrothermal systems in the Guaymas Basin. Initial Reports of the Deep Sea Drilling Project, 64, Part II, 1143–57. Kelley DS, Delaney JR, Lilley MD, Butterfield DA (2001a) Vent field distribution and evolution along the Endeavour Segment, Juan de Fuca Ridge. EOS. Transactions of the American Geophysical Union, 82, Fall Meet. Suppl., Abstract OS21B-0439. Kelley DS, Delaney JR, Yoerger DR (2001b) Geology and venting characteristics of the Mothra hydrothermal field, Endeavour Segment, Juan de Fuca Ridge. Geology, 29, 959–62. Kelley DS, Baross JA, Delaney JR (2002) Volcanoes, fluids, and life at mid-ocean ridge spreading centers. Annual Review of Earth and Planetary Sciences, 30, 385–491. Kleinrock MC, Humphris SE (1996) Structural control on seafloor hydrothermal activity at the TAG active mound. Nature, 382, 149–53. Klitgord K, Mammerickx J (1982) Northern East Pacific Rise: magnetic anomaly and bathymetric framework. Journal of Geophysical Research, 87, 6725–50. Konn C, Charlou JL, Donval JP, Holm NG, Dehairs F, Bouillon S (2009) Hydrocarbons and oxidized organic compounds in hydrothermal fluids from Rainbow and Lost City ultramafichosted vents. Chemical Geology, 258, 299–314. Koschinsky A, Billings A, Devey C, Dubilier N, Duester A, Edge D, Garbe-Schonberg D, German C, Giere O, Keir R, Lacksche-
witz K, Mai HA, Marbler H, Mawick J, Melchert B, Mertens C, Peters M, Sander S, Schmale O, Schmidt W, Seifert R, Seiter C, Stober U, Suck I, Walter M, Weber S, Yoerger D, Zarrouk M, Zielinski F (2006) Discovery of new hydrothermal vents on the southern Mid-Atlantic Ridge (4S–10S) during cruise M68 ⁄ 1. InterRidge News, 15, 9–15. Koski RA, Lonsdale PF, Shanks W, Brendt ME, Howe SS (1985) Mineralogy and geochemistry of a sediment-hosted hydrothermal sulfide deposit from the Southern Trough of Guaymas Basin, Gulf of California. Journal of Geophysical Research, 90, 6695–707. Kumagai H, Nakamura K, Morishita T, Toki T, Okino K, Shibuya T, Sawaguchi T, Neo N, Joshima M, Sato T, Takai K (2008) Geological background of the Kairei and Edmond hydrothermal vent fields along the Central Indian Ridge: insights into the distinct chemistry between their vent fluids. Geofluids, 8, 239–51. Kurras GJ, Fornari DJ, Edwards MH, Perfit MR, Smith MC (2000) Volcanic morphology of the East Pacific Rise crest 98490-520N. 1. Implications for volcanic emplacement processes at fast-spreading mid-ocean ridges. Marine Geophysical Researches, 21, 23–41. Lalou C, Reyss JL, Brichet E, Rona PA, Thompson G (1995) Hydrothermal activity on a 105-year scale at a slow-spreading ridge, TAG hydrothermal field, Mid-Atlantic Ridge 26N. Journal of Geophysical Research, 100, 17855–62. LaRowe DE, Helgeson HC (2006a) Biomolecules in hydrothermal systems: calculation of the standard molal thermodynamic properties of nucleic-acid bases, nucleosides, and nucleotides at elevated temperatures and pressures. Geochimica et Cosmochimica Acta, 70, 4680–724. LaRowe DE, Helgeson HC (2006b) The energetics of metabolism in hydrothermal systems: calculation of the standard molal thermodynamic properties of magnesium complexed adenosine nucleotides and NAD and NADP at elevated temperatures and pressures. Thermochimica Acta, 448, 82–106. Lilley MD, de Angelis MA, Gordon LI (1982) CH4, H2, CO and N2O in submarine hydrothermal vent waters. Nature, 300, 48–50. Lilley MD, Olson EJ, McLaughlin E, Von Damm KL (1991) Methane, hydrogen and carbon dioxide in vent fluids from the 9N hydrothermal system. EOS. Transactions of the American Geophysical Union, 72, 481. Lilley MD, Butterfield DA, Olson EJ, Lupton JE, Macko SA, McDuff RE (1993) Anomalous CH4 and NHþ 4 concentrations at an unsedimented mid-ocean-ridge hydrothermal system. Nature, 364, 45–7. Lilley MD, Butterfield DA, Kelley DS, Tivey MK, Delaney JR (2000) Long- and short-term variability in hydrothermal activity at the Endeavour RIDGE Observatory site. EOS. Transactions of the American Geophysical Union, 81, 628. Lilley MD, Lupton JE, Olson EA (2002) Using CO2 and He in vent fluids to constrain along axis magma dimension at 9N, EPR. EOS. Transactions of the American Geophysical Union, 83, F1386. Lilley M, Olson E, Lupton J (2006) Changes in volatile concentrations as a result of the 2006 eruption at 9N. EOS. Transactions of the American Geophysical Union, 87 (Abstract #V13C-03). Lonsdale P (1978) Submersible exploration of Guaymas Basin: a preliminary report of the Gulf of California 1977 operations of DSV-4 ‘‘Seacliff’’. Reports, Scripps Institute of Oceanography La Jolla, CA, SIO Ref. 7&1. Lonsdale P (1980) Hydrothermal plumes and baritic sulfide mounds at a Gulf of California spreading center. EOS. Transactions of the American Geophysical Union, 61, 995.
Seafloor hydrothermal systems 189 Lonsdale P, Becker K (1985) Hydrothermal plumes, hot springs, and conductive heat flow in the Southern Trough of Guaymas Basin. Earth and Planetary Science Letters, 73, 211–25. Lonsdale P, Lawver LA (1980) Immature plate boundary zones studied with a submersible in the Gulf of California. Geological Society of America Bulletin, 91, 555–69. Lonsdale P, Bischoff JL, Burns VM, Kastner M, Sweeney RE (1980) A high-temperature hydrothermal deposit on the seabed at a Gulf of California spreading center. Earth and Planetary Science Letters, 49, 8–20. Lupton JE (1979) Helium-3 in the Guaymas Basin: evidence for injection of mantle volatiles in the Gulf of California. Journal of Geophysical Research, 84, 7446–52. Lupton JE (1983) Fluxes of helium-3 and heat from submarine hydrothermal systems: Guaymas Basin versus 21N EPR. EOS. Transactions of the American Geophysical Union, 64, 723. Lupton JE, Lilley M, Olson E, Von Damm KL (1991) Gas chemistry of vent fluids from 9–10N on the East Pacific Rise. EOS. Transactions of the American Geophysical Union, 72, 481. Macdonald KC, Becker K, Spiess FN, Ballard RD (1980) Hydrothermal heat flux of the ‘Black Smoker’ vents on the East Pacific Rise. Earth and Planetary Science Letters, 48, 1–7. Martin W, Russell MJ (2007) On the origin of biochemistry at an alkaline hydrothermal vent. Philosophical Transactions of the Royal Society B, 362, 1887–925. deMartin B, Sohn R, Canales J, Humphris S (2007) Kinematics and geometry of active detachment faulting beneath the TransAtlantic Geotraverse (TAG) hydrothermal field on the MidAtlantic Ridge. Geology, 35, 711–4. Martinez F, Taylor B, Baker ET, Resing JA, Walker SL (2006) Opposing trends in crustal thickness and spreading rate along the back-arc Eastern Lau Spreading Centre: implications for controls on ridge morphology, faulting and hydrothermal activity. Earth and Planetary Science Letters, 245, 655–72. McCollom T (2007) Geochemical constraints on sources of metabolic energy for chemolithoautotrophy in ultramafic-hosted deep-sea hydrothermal systems. Astrobiology, 7, 933–50. McCollom T (2008) Observational, experimental, and theoretical constraints on carbon cycling in mid-ocean ridge hydrothermal systems. Magma to Microbe: Modeling Hydrothermal Processes at Ocean Spreading Centers Geophysical Monograph, 178, 193– 213. McCollom TM, Seewald JS (2007) Abiotic synthesis of organic compounds in deep-sea hydrothermal environments. Chemical Reviews, 107, 382–401. McCollom TM, Shock EL (1997) Geochemical constraints on chemolithoautotrophic metabolism by microorganisms in seafloor hydrothermal systems. Geochimica et Cosmochimica Acta, 61, 4375–91. Morton J, Sleep NH (1985) Seismic reflections from a Lau Basin magma chamber. Earth Sciences Series, Circum-Pacific Council for Energy and Mineral Resources, 2, 441–53. Mozgova NN, Trubkin NV, Borodaev YS, Cherkashev GA, Stepanova TV, Semkova TA, Uspenskaya TYu (2008) Mineralogy of massive sulfides from the Ashadze hydrothermal field, 13N, Mid-Atlantic Ridge. Canadian Mineralogist, 46, 545– 67. Nakamura K, Morishita T, Bach W, Klein F, Hara K, Okino K, Takai K, Kumagai H (2009) Serpentinized troctolites exposed near the Kairei Hydrothermal Field, Central Indian Ridge: insights into the origin of the Kairei hydrothermal fluid supporting a unique microbial ecosystem. Earth and Planetary Science Letters, 280, 128–36.
Nercessian O, Bienvenu N, Moreira D, Prieur D, Jeanthon C (2005) Diversity of functional genes of methanogens, methanotrophs ad sulfate reducers in deep-sea hydrothermal environments. Environmental Microbiology, 7, 118–32. Parker CM, Von Damm KL (2005) Time series fluid compositions from the TAG Hydrothermal Mound, MAR: 1986–2004: EOS. Transactions of the American Geophysical Union, 86, abs. #OS22A–07. Peter JM, Scott SD (1988) Mineralogy, composition and fluidinclusion microthermometry of sea floor hydrothermal deposits in the Southern Trough of Guaymas Basin Gulf of California. Canadian Mineralogist, 26, 567–87. Peter JM, Simoneit BRT, Kawka OE, Scott SD (1990) Liquid hydrocarbon-bearing inclusions in modern hydrothermal chimneys and mounds from the southern trough of Guaymas Basin, Gulf of California. In: Organic Matter in Hydrothermal Systems– Petroleum Generation, Migration and Biogeochemistry (ed Simoneit BRT), Applied Geochemistry, 5, 51–63. Peter JM, Peltonen P, Scott SD, Simoneit BRT, Kawka OE (1991) 14C ages of hydrothermal petroleum and carbonate in Guaymas Basin, Gulf of California – implications for oil generation, expulsion and migration. Geology, 19, 253–6. Philpotts JA, Aruscavage PJ, Von Damm KL (1987) Uniformity and diversity in the composition of mineralizing fluids from hydrothermal vents on the southern Juan de Fuca Ridge. Journal of Geophysical Research, 92, 11327–33. Piepgras DJ, Wasserburg GJ (1985) Strontium and neodymium isotopes in hot springs on the East Pacific Rise and Guaymas Basin. Earth and Planetary Science Letters, 72, 341–56. Plyasunov AV, Shock EL (2001a) Group contribution values of the infinite dilution thermodynamic functions of hydration for aliphatic non-cyclic hydrocarbons, alcohols and ketones at 298.15 K and 0.1 MPa. Journal of Chemical and Engineering Data, 46, 1016–9. Plyasunov AV, Shock EL (2001b) Correlation strategy for determining the parameters of the revised Helgeson–Kirkham– Flowers model for aqueous nonelectrolytes. Geochimica et Cosmochimica Acta, 65, 3879–900. Plyasunov AV, Shock EL (2003) Prediction of the vapor-liquid distribution constants for volatile nonelectrolytes in H2O up to the critical temperature of water. Geochimica et Cosmochimica Acta, 67, 4981–5009. Plyasunov AV, Plyasunova NV, Shock EL (2004) Group contribution values for the thermodynamic functions of hydration of aliphatic esters at 298.15K and 0.1Mpa. Journal of Chemical and Engineering Data, 49, 1152–67, 10.1021/je049850a. Plyasunov AV, Plyasunova NV, Shock EL (2006a) Group contribution values for the thermodynamic functions of hydration at 298.15 K, 0.1 MPa. 3. Aliphatic monoethers, diethers, polyethers. Journal of Chemical and Engineering Data, 51, 276–90, 10.1021/je050390a. Plyasunov AV, Plyasunova NV, Shock EL (2006b) Group contribution values for the thermodynamic functions of hydration at 298.15 K, 0.1 MPa. 4. Aliphatic nitriles and dinitriles. Journal of Chemical and Engineering Data, 51, 1481–90. Plyasunova NV, Plyasunov AV, Shock EL (2005) Group contribution values for the thermodynamic functions of hydration at 298.15 K, 0.1 MPa. 2. Aliphatic thiols, alkyl sulfides, and polysulfides. Journal of Chemical and Engineering Data, 50, 246– 53, 10.1021/je0497045. Proskurowski G, Lilley MD, Seewald JS, Fru¨h-Green GL, Olson EJ, Lupton JE, Sylva SP, Kelley DS (2008) Abiogenic hydrocarbon production at Lost City Hydrothermal Field. Science, 319, 604–7.
190 E. SHOCK & P. CANOVAS Reed AJ, Dorn R, Van Dover CL, Lutz RA, Vetriani C (2009) Phylogenetic diversity of methanogenic, sulfate reducing and methanotrophic prokaryotes for deep-sea hydrothermal vents and cold seeps. Deep-Sea Research II, 56, 1665–74. Riddihough R (1984) Recent movements of the Juan de Fuca Plate System. Journal of Geophysical Research, 89, 6980–94. Robigou V, Delaney JR, Stakes DS (1993) Large massive sulphide deposits in a newly discovered active hydrothermal system, the High-Rise Field, Endeavour Segment, Juan de Fuca Ridge. Geophysical Research Letters, 20, 1887–90. Russell MJ, Martin W (2004) The rocky roots of the acetyl-CoA pathway. Trends in Biochemical Sciences, 29, 358–63. Russell MJ, Hall AJ, Boyce AJ, Fallick AE (2005) On hydrothermal convection systems and the emergence of life. Economic Geology, 100, 419–38. Sarrazin J, Juniper SK, Massoth G, Legendre P (1999) Physical and chemical factors influencing species distributions on hydrothermal sulfide edifices of the Juan de Fuca Ridge, northeast Pacific. Marine Ecology. Progress Series, 190, 89–112. Schrenk MO, Kaye JZ, Baross JA, Kelley DS, Delaney JR (1998) Microbe-mineral associations in intact hydrothermal vent chimneys recovered from the Juan de Fuca Ridge. EOS. Transactions of the American Geophysical Union, 79, 59. Schrenk MO, Kelley DS, Baross JA (1999) Microbial diversity in a 300C black smoker chimney; relationships with thermal and geochemical gradients. American Society of Microbiology Conference on Microbiological Biodiversity, 5–8, August, Chicago, IL. Schrenk MO, Kelley DS, Delaney JR, Baross JA (2003) Incidence and diversity of microorganisms within the walls of an active deep-sea sulfide chimney. Applied Environmental Microbiology, 69, 3580–92. Schulte MD, Shock EL (1993) Aldehydes in hydrothermal solutions: standard partial molal thermodynamic properties and relative stabilities at high temperatures and pressures. Geochimica et Cosmochimica Acta, 57, 3835–46. Seewald J, Cruse A, Saccocia P (2003) Aqueous volatiles in hydrothermal fluids from the Main Endeavour Field, northern Juan de Fuca Ridge: temporal variability following earthquake activity. Earth and Planetary Science Letters, 216, 575–90. Shank TM, Fornari DJ, Von Damm KL, Lilley MD, Haymon RM, Lutz RA (1998) Temporal and spatial patterns of biological community development at nascent deepsea hydrothermal vents (95¢N, East Pacific Rise). Deep-Sea Research II, 45, 465–515. Shanks WC III, Bohlke JK, Seal RR II (1991) Stable isotope studies of vent fields, 9-10 N East Pacific Rise: water–rock interaction and phase separation. EOS. Transactions of the American Geophysical Union, 72, 481. Shock EL (1992) Stability of peptides in high temperature aqueous solutions. Geochimica et Cosmochimica Acta, 56, 3481–91. Shock EL (1993) Hydrothermal dehydration of aqueous organic compounds. Geochimica et Cosmochimica Acta, 57, 3341–9. Shock EL (1995) Organic acids in hydrothermal solutions: standard molal thermodynamic properties of carboxylic acids, and estimates of dissociation constants at high temperatures and pressures. American Journal of Science, 295, 496–580. Shock EL (1997) High temperature life without photosynthesis as a model for Mars. Journal of Geophysical Research, 102. 23,68723,694. Shock EL, Helgeson HC (1990) Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: standard partial molal properties
of organic species. Geochimica et Cosmochimica Acta, 54, 915–45. Shock EL, Schulte MD (1998) Organic synthesis during fluid mixing in hydrothermal systems. Journal of Geophysical Research, 103, 28513–27. Shock EL, Helgeson HC, Sverjensky DA (1989) Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: standard partial molal properties of inorganic neutral species. Geochimica et Cosmochimica Acta, 53, 2157–83. Shock EL, Oelkers EH, Johnson JW, Sverjensky DA, Helgeson HC (1992) Calculation of the thermodynamic properties of aqueous species at high pressures and temperatures: effective electrostatic radii, dissociation constants, and standard partial molal properties to 1000C and 5 kb. Journal of the Chemical Society, Faraday Transactions, 88, 803–26. Shock EL, Sassani DC, Willis M, Sverjensky DA (1997) Inorganic species in geologic fluids: correlations among standard molal thermodynamic properties of aqueous ions and hydroxide complexes. Geochimica et Cosmochimica Acta, 61, 907–50. Shock EL, McCollom T, Schulte MD (1998) The emergence of metabolism from within hydrothermal systems. In: Thermophiles: The Keys to Molecular Evolution and the Origin of Life? (eds Wiegel J, Adams MWW), pp. 59–76. Taylor & Francis, London. Simoneit BRT (1983a) Organic matter maturation and petroleum genesis: geothermal versus hydrothermal. Proc. Syrup. The Role of Heat in the Development of Energy and Mineral Resources in the Northern Basin and Range Province, Geotherm. Res. Council, Special Report No. 13, Davis, CA, pp. 215–41. Simoneit BRT (1983b) Effects of hydrothermal activity on sedimentary organic matter: Guaymas Basin, Gulf of California– petroleum genesis and protokerogen degradation. In: NATOARI: Hydrothermal Processes at Seafloor Spreading Centers (eds Rona PA, Bostrom K, Laubier L, Smith KL Jr), pp. 451–71. Plenum Press, New York. Simoneit BRT (1985) Hydrothermal petroleum: genesis, migration, and deposition in Guaymas Basin, Gulf of California. Canadian Journal of Earth Sciences, 22, 1919–29. Simoneit BRT, Lonsdale PF (1982) Hydrothermal petroleum in mineralized mounds at the seabed of Guaymas Basin. Nature, 295, 198–202. Simoneit BRT, Leif RN, Sturz AA, Sturdivant AE, Gieskes JM (1992) Geochemistry of shallow sediments in Guaymas Basin, Gulf of California: hydrothermal gas and oil migration and effects of mineralogy. Organic Geochemistry, 18, 765–84. Simoneit BRT, Lein AY, Peresypkin VI, Osipov GA (2004) Composition and origin of hydrothermal petroleum and associated lipids in the sulfide deposits of the Rainbow Field (Mid-Atlantic Ridge at 36N). Geochimica et Cosmochimica Acta, 68, 2275–94. Sohn RA, Fornari DJ, Von Damm KL, Hildebrand JA, Webb SC (1998) Seismic and hydrothermal evidence for a cracking event on the East Pacific Rise crest at 9 50¢ N. Nature, 396, 159–61. Spiess FN (with RISE project group) (1980) East Pacific Rise: hot springs and geophysical experiments. Science, 207, 1421–33. Stout PM, Campbell AC (1983) Hydrothermal alteration of near surface sediments, Guaymas Basin, Gulf of California. In: Cenozoic Marine Sedimentation, Pacific Margin, U.S.A. (eds Larue DK, Steel RJ), pp. 223–31. SEPM Pacific Section. Sudarikov SM, Roumiantsev AB (2000) Structure of hydrothermal plumes at the Logatchev vent field, 1445¢N, Mid-Atlantic Ridge: evidence from geochemical and geophysical data. Journal of Volcanology and Geothermal Research, 101, 245–52.
Seafloor hydrothermal systems 191 Takai K, Gamo T, Tsunogai U, Nakayama N, Hirayama H, Nealson KH, Horikoshi K (2004) Geochemical and microbiological evidence for a hydrogen-based, hyperthermophilic subsurface lithoautotrophic microbial ecosystem (HyperSLiME) beneath an active deep-sea hydrothermal field. Extremophiles, 8, 269–82. Takai K, Nakamura K, Suzuki K, Inagaki F, Nealson KH, Kumagai H (2006) Ultramafics–Hydrothermalism–Hydrogenesis–HyperSLiME (UltraH3) linkage: a key insight into early microbial ecosystem in the Archean deep-sea hydrothermal systems. Paleontological Research, 10, 269–82. Taylor B, Zellmer K, Martinez F, Goodliffe A (1996) Sea-floor spreading in the Lau Back-Arc Basin. Earth and Planetary Science Letters, 144, 35–40. Teske A, Hinrichs K, Edgcomb V, Gomez A, Kysela D, Sylva SP, Sogin ML, Jannasch HW (2002) Microbial diversity of hydrothermal sediments in the Guaymas Basin: evidence for anaerobic methanotrophic communities. Applied and Environmental Microbiology, 68, 1994–2007. Tivey MK (2007) Generation of seafloor hydrothermal vent fluids and associated mineral deposits. Oceanogrphy, 20, 50–65. Tivey MK, Delaney JR (1985) Sulfide deposits from the Endeavour Segment of the Juan de Fuca Ridge. Marine Mining, 5, 165–79. Tivey MK, Delaney JR (1986) Growth of large sulfide structures on the Endeavour Segment of the Juan de Fuca Ridge. Earth and Planetary Science Letters, 77, 303–17. Tivey MA, Johnson HP (2002) Crustal magnetization reveals subsurface structure of Juan de Fuca Ridge hydrothermal vent fields. Geology, 30, 979–82. Tivey MK, Humphris SE, Thompson G, Hannington MD, Rona P (1995) Deducing patterns of fluid flow and mixing within the active TAG hydrothermal mound using mineralogical and geochemical data. Journal of Geophysical Research, 100, 12527–55. Tivey MA, Schouten H, Kleinrock MC (2003) A near-bottom magnetic survey of the Mid-Atlantic Ridge axis at 26N: implications for the tectonic evolution of the TAG segment. Journal of Geophysical Research, 108, 2277, doi: 10.1029/ 2002JB001967. Trefrey JH, Butterfield DA, Metz S, Massoth GJ, Trocine RP, Feely RA (1994) Trace metals in hydrothermal solutions from Cleft segment on the southern Juan de Fuca Ridge. Journal of Geophysical Research, 99, 4925–35. Tunnicliffe V (1988) Biogeography and evolution of hydrothermal-vent fauna in the eastern Pacific Ocean. Proceedings of the Royal Society of London Series B, 233, 347–66. Tunnicliffe V, Fowler CMR (1996) Influence of sea-floor spreading on the global hydrothermal vent fauna. Nature, 379, 531–3. Turner S, Hawkesworth C (1998) Using geochemistry to map mantle flow beneath the Lau Basin. Geology, 26, 1019–22. Van Ark EM, Detrick RS, Canales JP, Carbotte SM, Harding AJ, Kent GM, Nedimovic MR, Wilcock WSD, Diebold JB, Babcock JM (2007), Seismic structure of the Endeavour Segment, Juan de Fuca Ridge: correlations with seismicity and hydrothermal activity. Journal of Geophysical Research, 112, B02401, doi: 10.1029/2005JB004210. Van Dover CL, Humphris SE, Fornari D, Cavanaugh CM, Collier R, Goffredi SK, Hashimoto J, Lilly MD, Reysenbach AL, Shank TM, Von Damm KL, Banta A, Gallant RM, Go¨tz D, Green D, Hall J, Harmer TL, Hurtado LA, Johnson P, McKiness ZP, Meredith C, Olson E, Pan IL, Turnipseed M, Won Y, Young CR III, Vrijenhoek RC (2001) Biogeography and ecological setting of Indian Ocean hydrothermal vents. Science, 294, 818–23.
Van Dover CL, German CR, Speer KG, Parson LM, Vrijenhoek RC (2002) Evolution and biogeography of deepsea vent and seep invertebrates. Science, 295, 1253–7. Von Damm KL (1995) Controls on the chemistry and temporal variability of seafloor hydrothermal fluids. In: Seafloor Hydrothermal Systems: Physical, Chemical, Biologic and Geological Interactions, Geophysical Monograph, Vol. 91 (eds Humphris SE, Zierenberg RA, Mullineaux LS, Thomson RE), pp. 222–47. American Geophysical Union, Washington, DC. Von Damm KL (2004) Evolution of the hydrothermal system at East Pacific Rise 95¢N: geochemical evidence for changes in the upper oceanic crust. In: Mid-Ocean Ridges: Hydrothermal Interactions between the Lithosphere and Oceans (eds German CR, Lin J, Parson LM) Geophysical Monograph Series, 148, 285–304. Von Damm KL, Bischoff JL (1987) Chemistry of hydrothermal solutions from the southern Juan de Fuca Ridge. Journal of Geophysical Research, 92, 11334–46. Von Damm KL, Lilley MD (2004) Diffuse flow hydrothermal fluids from 950¢N East Pacific Rise: origin, evolution and biogeochemical controls. In: Subseafloor Biosphere at Mid-Ocean Ridges (eds Wilcock W, Cary C, De Long E, Kelley D, Baross J), Geophysical Monograph Series, 144, 245–68. American Geophysical Union, Washington, DC. Von Damm KL, Edmond JM, Grant B, Measures CI, Walden B, Weiss R (1985a) Chemistry of submarine hydrothermal solutions at 21 N, East Pacific Rise. Geochimica et Cosmochimica Acta, 49, 2197–220. Von Damm KL, Edmond JM, Measures CI, Grant B (1985b) Chemistry of submarine hydrothermal solutions at Guaymas Basin, Gulf of California. Geochimica et Cosmochimica Acta, 49, 2221–37. Von Damm KL, Bischoff JL, Rosenbauer RJ (1991) Quartz solubility in hydrothermal seawater: an experimental study and equation describing quartz solubility for up to 0.5 M NaCl solutions. American Journal of Science, 291, 977–1007. Von Damm KL, Buttermore LG, Oosting SE, Bray AM, Fornari DJ, Lilley MD, Shanks WC III (1997) Direct observation of the evolution of a seafloor ‘‘black smoker’’ from vapor to brine. Earth and Planetary Science Letters, 149, 101–12. Von Damm KL, Parker CM, Zierenberg RA, Lilley MD, Olson EJ, Clague DA, McClain JS (2005) The Escanaba Trough, Gorda Ridge hydrothermal system: temporal stability and subseafloor complexity. Geochimica et Cosmochimica Acta, 69, 4971–84. Von Damm KL, Bates MJ, Carmichael SK, Meana-Prado F, McDermott JM (2006) Response of the 9–10N EPR hydrothermal systems to recent volcanic eruptions. EOS. Transactions of the American Geophysical Union, 87, Abstract #V13C-02. Von Stackelberg U and Shipboard Party (1988) Active hydrothermalism in the Lau Back-arc basin (SW-Pacific): first results from the SONNE 48 cruise (1987). Marine Mining, 7, 431–42. Welhan JA, Craig H (1983) Methane, hydrogen and helium in hydrothermal fluids at 21N on the East Pacific Rise. In: Hydrothermal Processes at Seafloor Spreading Centers, Vol. 12 (eds Rona PA, Bostrom K, Laubier L, Smith K Jr), pp. 391–411. Plenum Press, New York. Welhan JA, Lupton JE (1987) Light hydrocarbon gases in Guaymas Basin hydrothermal fluids: thermogenic versus abiogenic origin. American Association of the Petroleum Geology Bulletin, 71, 215–23. Wetzel LR, Shock EL (2000) Distinguishing ultramafic from basalt-hosted submarine hydrothermal systems by comparing calculated vent fluid compositions. Journal of Geophysical Research, 105, 8319–40.
192 E. SHOCK & P. CANOVAS Wetzel LR, Raffensperger JP, Shock EL (2001) Predictions of hydrothermal alteration within near-ridge oceanic crust from coordinated geochemical and fluid flow models. Journal of Volcanology and Geothermal Research, 110, 319–42. Wilcock WSD (2004) Physical response of mid-ocean ridge hydrothermal systems to local earthquakes. Geochemistry Geophysics Geosystems, 5, Q11009, doi: 10.1029/2004GC000701. Wilcock WSD, Hooft EEE, Toomey DR, McGill PR, Barclay AH, Stakes DS, Ramirez TM (2009) The role of magma injection in localizing black-smoker activity. Nature Geoscience, 2, 509–13. Wolery TW, Jarek RL (2003). Software User’s Manual EQ3 ⁄ 6, Version 8.0. US Department of Energy, Office of Civilian Radioactive Waste Management, Office of Repository Devel-
opment, Software Document Number 10813-UM-8.0-00, 376. Wright DJ, Haymon RM, Fornari DJ (1995) Crustal fissuring and its relationship to magmatic and hydrothermal processes on the East Pacific Rise crest (912¢ to 54N). Journal of Geophysical Research, 100, 6097–120. You C-F, Butterfield DA, Spivack AJ, Gieskes JM, Gamo T, Campbell AJ (1994) Boron and halide systematics in submarine hydrothermal systems: effects of phase separation and sedimentary contributions. Earth and Planetary Science Letters, 123, 227–38. Zellmer KE, Taylor B (2001) A three-plate kinematic model for Lau Basin opening. Geochemistry Geophysics Geosystems, 2, doi: 10.1029/2000GC000106.
Permeability of the continental crust: dynamic variations inferred from seismicity and metamorphism S. E. INGEBRITSEN1 AND C. E. MANNING2 1
US Geological Survey, Menlo Park, CA, USA; 2Department of Earth and Space Sciences, UCLA, Los Angeles, CA, USA
ABSTRACT The variation of permeability with depth can be probed indirectly by various means, including hydrologic models that use geothermal data as constraints and the progress of metamorphic reactions driven by fluid flow. Geothermal and metamorphic data combine to indicate that mean permeability (k) of tectonically active continental crust decreases with depth (z) according to log k )14–3.2 log z, where k is in m2 and z in km. Other independently derived, crustal-scale k–z relations are generally similar to this power-law curve. Yet there is also substantial evidence for local-to-regional-scale, transient, permeability-generation events that entail permeabilities much higher than these mean k–z relations would suggest. Compilation of such data yields a fit to these elevated, transient values of log k )11.5–3.2 log z, suggesting a functional form similar to that of tectonically active crust, but shifted to higher permeability at a given depth. In addition, it seems possible that, in the absence of active prograde metamorphism, permeability in the deeper crust will decay toward values below the mean k–z curves. Several lines of evidence suggest geologically rapid (years to 103 years) decay of high-permeability transients toward background values. Crustal-scale k–z curves may reflect a dynamic competition between permeability creation by processes such as fluid sourcing and rock failure, and permeability destruction by processes such as compaction, hydrothermal alteration, and retrograde metamorphism. Key words: permeability, geothermal, metamorphism, seismicity Received 14 September 2009; accepted 17 January 2010 Corresponding author: S. E. Ingebritsen, US Geological Survey, Menlo Park, CA 94025, USA. Email:
[email protected]. Tel: 1-650-329-4422. Fax: 1-650-329-4463. Geofluids (2010) 10, 193–205
INTRODUCTION Permeability (k) is a measure of the relative ease of fluid flow under unequal pressure. The permeability of the Earth’s crust to aqueous fluids is of great interest because it largely determines the feasibility of important geologic processes such as advective solute transport, advective heat transport, and the generation of elevated fluid pressures by processes such as physical compaction, heating, and mineral dehydration. Yet the measured permeability of the shallow continental crust is so highly variable that it is often considered to defy systematic characterization. The permeability of common geologic media varies by approximately 16 orders of magnitude, from values as low as 10)23 m2 in intact crystalline rock, intact shales, and fault gouge, to values as high as 10)7 m2 in well-sorted gravels. In the upper crust, permeability exhibits extreme heterogeneity, both among geologic units and within particular units. Field-based measurements of layered ash-flow tuff
show up to 104-fold variation between welded and unwelded zones (e.g. Winograd 1971). Similarly large variations have been measured within single soil units (Mitchell 1993). Even larger variations in in situ permeability have been inferred between basalts near the surface of Kilauea volcano (k 10)10 to 10)9 m2) and compositionally identical rocks at 1- to 2-km depth (k 10)16 to 10)15 m2) (Ingebritsen & Scholl 1993). In many geologic environments, there is also large permeability anisotropy, which is conventionally defined as the ratio between the horizontal and vertical permeabilities but may also reflect variously oriented stratigraphic, structural, and ⁄ or tectonic fabrics. Permeability varies in time as well as space. Temporal variability in permeability is particularly pronounced in hydrothermal environments characterized by strong chemical and thermal disequilibrium. Laboratory experiments involving hydrothermal flow in crystalline rocks under pressure, temperature, and chemistry gradients often result in
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
194 INGEBRITSEN & MANNING order-of-magnitude permeability decreases over daily to subannual time scales (e.g. Summers et al. 1978; Morrow et al. 1981, 2001; Moore et al. 1983, 1994; Vaughan et al. 1986; Tenthorey et al. 1998; Cox et al. 2001; Zhang et al. 2001; Polak et al. 2003; Yasuhara et al. 2006). Field observations of continuous, cyclic, and episodic hydrothermal-flow transients at various time scales also suggest transient variations in permeability (e.g. Baker et al. 1987, 1989; Titley 1990; Hill 1993; Urabe 1995; Haymon 1996; Fornari et al. 1998; Sohn et al. 1998; Gillis & Roberts 1999; Johnson et al. 2000; Golden et al. 2003; Husen et al. 2004; Sohn 2007). The occurrence of active, long-lived (103–106 years) hydrothermal systems (Cathles et al. 1997), despite the tendency for permeability to decrease with time, implies that other processes such as hydraulic fracturing and earthquakes regularly create new flow paths (e.g. Rojstaczer et al. 1995). Indeed there have been suggestions that crustal-scale permeability is a dynamically self-adjusting or even emergent property (e.g. Rojstaczer et al. 2008). This study reviews studies of crustal-scale permeability– depth relations in the last decade. Our earlier work emphasized tectonically active regions of the continental crust and focused on permeability averaged over large time and length scales (Ingebritsen & Manning 1999; Manning & Ingebritsen 1999). Here, we show that independent studies generally agree on values of crustal permeability at such scales. However, we extend these results by surveying recent observations of high permeabilities associated with shorter time and length scales, and by considering permeability decay.
CRUSTAL-SCALE PERMEABILITY–DEPTH RELATIONS Permeability is heterogeneous, anisotropic, and transient. Nevertheless, some order has been revealed in globally compiled data. In the early 1980s it was proposed, based on compilations of in situ hydraulic-test data, that the mean in situ permeability of crystalline rocks in the uppermost crust (<1-km depth) is approximately 10)14 m2 (Brace 1980). This result for the very shallow crust is borne out by more recent in situ data (Hsieh 1998), whereas other in situ data suggest an identifiable decrease in permeability with depth (Clauser 1992). Direct in situ measurements of permeability are rare below depths of 2–3 km and nonexistent below 10-km depth. As an alternative, geothermal data and estimates of fluid flux during prograde metamorphism have been used to constrain the permeability of regions of the continental crust undergoing active metamorphism and tectonism. A power-law fit to these data yields
(A)
(B)
Fig. 1. Estimates of permeability based on hydrothermal modeling and the progress of metamorphic reactions showing (A) power-law fit to data and (B) data below 12.5-km depth fitted with a constant value of 10)18.3 m2 (after Manning & Ingebritsen 1999; Ingebritsen & Manning 2002).
where k is in m2 and z is in km (Manning & Ingebritsen 1999). This empirical fit (Fig. 1A) defines a value of log k at 1-km depth ()14) that is equivalent to Brace’s (1980) mean in situ permeability of crystalline rocks. Assuming the depth of the brittle–ductile transition in tectonically active crust to be 10–15 km and fitting the data in each regime separately (Fig. 1B) implies effectively constant permeability of log k )18.3 below 15 km. Townend & Zoback (2000) found eqn 1 to be compatible with data from in situ hydraulic tests and from seismicity induced either by fluid injection or reservoir impoundment. The ‘geothermal–metamorphic’ permeability–depth relation (eqn 1, Fig. 1) has since been used successfully in modeling crustal-scale fluid flow (Lyubetskaya & Ague 2009) and been shown to be reasonably compatible with other independently compiled data (Shmonov et al. 2002, 2003; Saar & Manga 2004; Stober & Bucher 2007). Field permeability measurements (35 soil samples) and lab experiments at high pressure and temperature (11 samples, 237 experimental points to 600C, 200 MPa) by Shmonov et al. (2003) yield a similar relation, log k 12:56 3:225z 0:223 2
log k 14 3:2 log z;
ð1Þ
ð2Þ
with k and z again in m and km, respectively. In this case, )log k at 1-km depth is 15.6. Saar & Manga (2004)
Dynamic permeability of the continental crust 195
which fits their shallow data better and allows for a finite near-surface permeability. In this case, )log k at 1-km depth is approximately 15. Most recently, an empirical fit to in situ data from £4.5-km depth in the Black Forest, Germany, yielded log k 15:4 1:38 log z
ð4Þ
with k and z again in m2 and km, respectively (Stober & Bucher 2007). Figure 2 compares these three proposed crustal-scale permeability–depth relations (eqns 1, 2, and 4), all of which implicitly assume permeability to be isotropic. The permeability trend based on original experimental data (eqn 2) can be viewed as representing the crust under isotropic stress conditions in a state of mechanical and chemical equilibrium (Shmonov et al. 2003). The geothermal–metamorphic curve (Fig. 1A, eqn 1) represents natural systems averaged over large spatial scales and long time scales. Individual metamorphic-permeability values are based on time-integrated fluid flux over the (generally long) time span of a metamorphic event. They are obtained by solving a one-dimensional form of Darcy’s law suitable for variable-density fluids, oðP þ qgzÞ k qx ¼ ; ð5Þ ox l for time-averaged permeability k, Ql k ¼ ; Dtðo½P þ qgz =ox
ð6Þ
where qx is the volumetric fluid flow along the flow path x, q is fluid density, g is gravitational acceleration, z is elevation above a datum, l is the dynamic viscosity of the fluid, (o½P þ qgz =ox) is the energy gradient for flow along x, Q is the time-integrated fluid flux, and Dt is the duration of metamorphism, so that qx = Q ⁄ Dt. The time-integrated fluid flux Q is a key parameter in many metamorphic studies. It is calculated on the basis of fluid-driven changes in rock composition and mineral assemblages. The metamorphic permeabilities (Figs. 1 and 2) represent environments in which fluid flow was or is a consequence of tectonic or magmatic activity. It has been suggested that lower permeabilities might be expected during metamor-
0
10
Depth (km)
developed a model for the permeability structure of the central Oregon Cascades based on several lines of evidence, including springflow characteristics, matching of geothermal data, hydrologically induced seismicity at Mount Hood, and the permeability needed for escape of magmatic volatiles at depth. Their results agreed with the geothermal–metamorphic curve (eqn 1) except at relatively shallow depths (£0.8 km), where they proposed instead z ; ð3Þ k 5 1013 m2 exp 25 km
20
Geothermal-metamorphic (Manning & Ingebritsen, 1999) Experimental (Shmonov et al., 2003) Black Forest (Stober & Bucher, 2007) Nu = 2
30
40 –22
–20
–18 –16 –14 2 Log permeability (m )
–12
–10
Fig. 2. Crustal-scale permeability–depth curves based on geothermal–metamorphic (Manning & Ingebritsen 1999), experimental (Shmonov et al. 2003), and Black Forest (Stober & Bucher 2007) data. The permeability associated with advection-dominated heat transport (Nu > 2) in the lower crust is calculated assuming a driving-force gradient of 10 MPa km)1, a temperature gradient of 25C km)1, and a thermal conductivity of 2 W (m K))1, and ranges from log k )18.1 m2 at 10-km depth to log k )18.6 m2 at 40-km depth. The permeability associated with Nu 2 in the upper crust (0- to 10-km depth) will typically be about 2 orders of magnitude higher (log k )16 m2) because of the lower fluid viscosities and much lower driving-force gradients for fluid flow. The permeability associated with a Sherwood number Sh 2 in the deeper crust would be approximately 104 times lower than that associated with Nu 2 (as per Bickle & McKenzie 1987, their fig. 6); that is, log k )22 m2.
phism associated with cooling and decompression (cf. Yardley & Baumgartner 2007), or in the deep crust in stable cratons (cf. Ingebritsen & Manning 2002). This suggestion is consistent with the fact that mean geothermal–metamorphic permeabilities (eqn 1) are roughly one order of magnitude larger than mean ‘experimental’ permeabilities (eqn 2) (see Fig. 2). The ‘Black Forest’ permeability curve, which represents a tectonically active rifting environment and is empirically constrained only at shallow (<5 km) depths, lies between the experimental and geothermal–metamorphic curves (Fig. 2).
EVIDENCE FOR HIGHER PERMEABILITIES The permeability–depth relations portrayed in Figure 2 are reasonably consistent. However, on short time scales, permeability may reach values significantly in excess of those represented by eqns 1, 2, and 4. There is now a growing body of evidence that allows examination of whether there are systematic variations in this behavior with depth. The evidence includes rapid migration of seismic hypocenters, enhanced rates of metamorphic reaction
196 INGEBRITSEN & MANNING where r is the distance and t is the time (Talwani & Acree 1984; Shapiro et al. 1997). Hydraulic diffusivity is related to hydraulic conductivity K through
in major fault or shear zones, and recent studies suggesting much more rapid metamorphism than has been canonically assumed (Table 1).
K ¼ DSs ; Space–time progression of earthquake fronts
where
Certain well-located earthquake swarms exhibit space–time progression of seismicity fronts that develop roughly as the square root of time, consistent with earthquake triggering by diffusive propagation of an aqueous-fluid pressure front (Fig. 3). Rates of hypocenter migration can be used to calculate a range of hydraulic diffusivities D according to r ¼ ð4pDt Þ1=2 ;
Ss ¼ qg ða þ nbÞ;
Locality
Log k (m2)
Diffusivity (m2 s)1)
Migration of seismic hypocenters Matsuhiro 1965–1967 (V) 0–6 Remiremont 1984 (H) 6–8 Yellowstone 1985 (H) 2.5–9 Mammoth Mtn. 1989 (V) 2–6 Mammoth Mtn. 1989 (H) 5–6 Dobi (Afar) 1989 (H) 5.5–11.6 Antofagasta 1995 (V) 34–38 South Moat, Long Valley 4–9 1997 (H) Umbria-Marche 1997 1–8 West Bohemia 2000 7–10
)12.6 )16 to )13 )12.7 )14.4 to )13.9 )15.2 to )14.9 )8.3 to )7.3 )13.3 )12.7 to )11.8
10 0.2–0.6 0.03–0.06
12–90
)10.4 )14.4
0.27
Table 1 Evidence for relatively high crustal-scale permeabilities. Reference
Cappa et al. (2009) Audin et al. (2002) Waite & Smith (2002) Hill & Prejean (2005) Hill & Prejean (2005) Noir et al. (1997) Nippress & Rietbrock (2007) D.P. Hill, USGS, written communication Miller et al. (2004) Horalek & Fischer (2008)
Locality
Depth
Log k
Reference
Fault-zone metamorphism Hunts Brook, CT Finero, Italy Storo, Greenland Grimsel, Switzerland Broken Hill, Aust. Aar Massif, Switz.
22.3 ± 3.7* 22.3 ± 3.7* 22.3 ± 3.7* 14.9 ± 3.7* 14.9 ± 3.7* 9.7–13.4
)15.8 )16.3 )15.8 )16.45 )16.15 )17.1 to )15.1
Dipple & Ferry (1992) Dipple & Ferry (1992) Dipple & Ferry (1992) Dipple & Ferry (1992) Dipple & Ferry (1992) Challandes et al. (2008)
Locality
Depth
Temporally focused heating Scotland (regional) 12–15.6 Connecticut (reg.) 18.2–29.2
Dt (previous Dt)
0.3 (3 Ma) 2 (13 Ma)
ð9Þ
q is the density of the aqueous phase, g is the gravitational acceleration, a is the bulk compressibility of the medium, n is porosity, and b is the compressibility of the aqueous phase. Permeability k can then be calculated from hydraulic conductivity via
ð7Þ
Depth (km)
ð8Þ
Log k
Reference
)17.4 to )15.6 )17.7 to )16.7
Ague & Baxter (2007) Lancaster et al. (2008)
Locality
Depth
Log k
Reference
Anthropogenic seismicity Rocky Mtn. Arsenal, CO KTB, Germany Soultz, France Basel, Switzerland ‘‘Seismogenic k’’
3.7–7.0 7.5–9 2.85–3.4 4.6–5.0 0–10
)16.2 )16.6 to )16 )14.5 )14.4à )15.3 to )13.3
Hsieh & Bredehoeft (1981) Shapiro et al. (1997) Evans et al. (2005) Ha¨ring et al. (2008) Talwani et al. (2007)
The designations (V) and (H) for seismic hypocenters indicate dominantly vertical and horizontal migration of the seismicity fronts, respectively. The (previous Dt) noted for temporally focused heating refers to the duration of metamorphism assumed by Manning & Ingebritsen (1999 , their table 2). *Dipple & Ferry (1992) do not specify uncertainties; our assumed value is the uncertainly commonly quoted for thermobarometry from metamorphic mineral assemblages Initial (prestimulation) permeability was log k )16.8 (Evans et al. 2005). àInitial (prestimulation) permeability was log k )17 (Ha¨ring et al. 2008).
Dynamic permeability of the continental crust 197 Fault-zone metamorphism
0 D = 0.8 m2/s D = 0.2 m2/s
Depth (km)
2
4
6
8
10
0 (J.D.121)
100 50 150 Time, Days since May 1, 1989
200
Fig. 3. Seismicity propagation rates provide a constraint on (dynamic) hydraulic diffusivity (D = r2 ⁄ 4pt). In the case of the 1989 earthquake swarm beneath Mammoth Mountain, California, D 0.2–0.8 m2 s)1 (from Hill & Prejean 2005).
k ¼ lK =qg;
ð10Þ
where l is dynamic viscocity and g is the gravitational acceleration. The conversion from D to k introduces substantial uncertainty that owes mainly to the uncertainties associated with q, a, n, b, and l. Table 1 lists k values computed by other authors when available (cf. Miller et al. 2004, p. 727). Otherwise, we have converted the reported D values to k by assuming Ss 10)6 m)1, following Saar & Manga (2004, p. 11), and calculating q and l for pure water at the mean seismogenic depth assuming a geothermal gradient of 25C km)1, a mean surface temperature of 10C, and a hydrostatic pressure gradient. The examples of hypocenter migration listed in Table 1 yield values of log k ranging from )16 to )7.3, or 1 to 9 orders of magnitude higher than those indicated by the geothermal–metamorphic curve at comparable depths (Fig. 4). The extreme values of log k ()8.3 to )7.3) are for the Dobi earthquake swarm, central Afar. The Dobi seismic sequence traversed fissured basalts. Although such basalts are the most permeable rocks widely exposed at the Earth’s surface, the permeability of young, unaltered basalt flows is typically somewhat smaller, with mean log k constrained to be in the range of )11 to )9 in diverse geologic settings [the flanks of the mid-ocean ridge (Stein & Fisher 2003), oceanic islands (Ingebritsen & Scholl 1993), and continental volcanic arcs (Manga 1996, 1997)]. The other examples of hypocenter migration yield log k of )16 to )10.4, well within the range observed in various geologic media near the Earth’s surface but unusually high for the given crustal depths.
Our previous compilation of metamorphic-permeability data (Manning & Ingebritsen 1999, their table 2) intentionally omitted major faults and shear zones, as their restricted areal extent and concentration of strain by definition made them anomalous with respect to average properties of the crust. Work on metamorphic data from deep fault zones (Dipple & Ferry 1992, their fig. 4) had already established that fault-zone permeabilities tend to be substantially higher, a finding corroborated by more recent work (Challandes et al. 2008). The six examples of faultzone metamorphism listed in Table 1 yield a mean – and apparently depth-independent (Fig. 4) – permeability of log k )16.1. This is 2 orders of magnitude higher than the depth-independent permeability suggested by the metamorphic data set that excludes fault zones (log k )18.3, Fig. 1B). Temporally focused heating Calculated values of metamorphic permeability are inversely proportional to the duration of metamorphism (Dt in eqn 6). Two recent analyses of metamorphism (Table 1) provide evidence for much more rapid heating than previously assumed, revising the time scale of regional metamorphism from approximately 3 Ma (Ague 1997) to approximately 0.3 Ma in Scotland (Ague & Baxter 2007) and from approximately 13 Ma (Ague 1994) to approximately 2 Ma in Connecticut (Lancaster et al. 2008). These revised time scales increase the calculated permeabilities by roughly an order of magnitude, placing permeability during both events well above the mean geothermal–metamorphic permeability–depth curve (Fig. 4). The recalculated permeabilities are large enough to permit significant heat advection (Fig. 2), consistent with the fact that advectively perturbed geotherms have been inferred in each instance (Ague & Baxter 2007; Lancaster et al. 2008). Anthropogenically enhanced permeability Earthquake triggering by diffusive propagation of an aqueous-fluid pressure front can be initiated by sudden communication between a relatively high-pressure source and lower-pressure surroundings (e.g. Miller et al. 2004; Hill & Prejean 2005). This suggests analogy with anthropogenic earthquake triggering via fluid injection (e.g. Fischer et al. 2008; Shapiro & Dinske 2009) and reservoir filling (Talwani et al. 2007). Studies of waste injection at the Rocky Mountain Arsenal (RMA) (Hsieh & Bredehoeft 1981), the German Continental Deep Drilling Borehole (KTB), and the Soultz and Basel Enhanced Geothermal System (EGS) sites have yielded particularly well-constrained hydraulic parameters. Preinjection permeabilities at
198 INGEBRITSEN & MANNING
(A)
0–2 orders of magnitude higher than those indicated by the geothermal–metamorphic curve (Fig. 4).
Hypocenter migration Fault-zone metamorphism Focused heating Anthropogenic seismicity
–6
PERMEABILITY DECAY AND LOWER BOUNDS
2
Log permeability (m )
–8
log k = –11.7 – 2.9 log z log k = –11.5 – 3.2 log z
The high permeabilities depicted in Fig. 4 (and Table 1) must be localized and transient. If this were not the case, crustal heat transport would be advection-dominated, and crustal temperatures would be generally lower than they are observed or inferred to be. Large-scale crustal permeabilities greater than the approximate threshold for advective heat transport (approximately 10)16 m2 and 10)18 m2 in the upper and lower crust, respectively) must be relatively rare. Further, the high permeabilities depicted in Fig. 4 would preclude the elevated fluid pressures that are believed to be pervasive below the brittle–ductile transition; overpressures typically require large regions of a flow domain (L > 100 m) to be composed of, or bounded by, material with k £ 10)17 m2 (Neuzil 1995; Manning & Ingebritsen 1999).
–10 –12 –14 –16 log k = –14 – 3.2 log z Geothermal-metamorphic data (Manning & Ingebritsen, 1999)
–18 –20
(B)
–6 Brittle
2
Log permeability (m )
–8
Ductile
log k = –9.8 – 5.5 log z Brittle crust – this study
–10
Rates of permeability decay –12 –14 –16 log k = –13.9 – 3.9 log z Brittle crust – Ingebritsen & Manning (2002)
–18 –20 0
0.2
0.4
0.6 0.8 1 1.2 Log depth (km)
1.4
1.6
Fig. 4. Evidence for relatively high crustal-scale permeabilities showing (A) power-law fit to data and (B) data below 12.5-km depth fitted with a constant value. Lower curve in both (A) and (B) is the best fit to geothermal–metamorphic data (Fig. 1). Data points are midpoints in reported ranges in k and z for a given locality (Table 1); error bars depict the full permissible range for a plotted locality and are not Gaussian errors.
the RMA are not well known, so comparison of pre- and poststimulation permeabilities is not possible. Fluid-injection-induced hydraulic fracturing at KTB yielded permeabilities near the upper limits of those determined by previous hydraulic testing (Shapiro et al. 1997). Fluid injection at 2.85- to 3.45-km depth at Soultz and 4.6- to 5.0-km depth at Basel increased permeability 102-fold relative to prestimulation conditions (Evans et al. 2005; Ha¨ring et al. 2008). Finally, a global synthesis of 90 case histories of injection- or reservoir-induced seismicity (Talwani et al. 2007) revealed that each episode seems to be associated with permeabilities in the range of log k = )15.3 to )13.3, above the mean range of crustal permeability at comparable depths. These examples of what we collectively term ‘anthropogenic seismicity’ (Table 1) are
In the absence of active fluid sourcing and tectonism, permeability should tend to decrease due to processes such as mineral precipitation, hydrothermal alteration, and compaction; however, the rate of this decrease is poorly known. Here we examine various constraints on the rates of permeability decay. In the introduction, we cited laboratory experiments involving hydrothermal flow in crystalline rocks that result in order-of-magnitude permeability decreases over subannual time scales. Although many laboratory studies involve strong chemical disequilibrium that may not be representative of most natural systems, there are field observations of hydrothermal-flow transients over comparably short time scales. Further, permeability in hydrothermal upflow zones can be drastically reduced by silica precipitation (Lowell et al. 1993) or thermoelastic stresses (Germanovich & Lowell 1992) over approximately 101 years. Hence, we infer that permeability decay can be very rapid under conditions of strong chemical or thermal disequilibrium. In situ measurements of permeability decay (Table 2) are particularly relevant, albeit scarce and limited to the brittle upper crust. Some such data owe to observations following co-seismic permeability increases caused by ˜ on Flat observatory in strong ground motion. At the Pin the California Coast Ranges, the response of water levels in two shallow (<250 m) wells to solid-Earth tides was used to measure permeability over a 20-year period. Elkhoury et al. (2006) found that permeability increased by as much as a factor of 3–4 coincident with seven regional earthquakes, with the magnitude of increase proportional to the peak ground velocity at the site. Between earthquakes,
Dynamic permeability of the continental crust 199 Table 2 Evidence for changes in in situ permeability in the brittle upper crust.
Locality
Depth (km)
k2 ⁄ k1 (m2)
Reference
Co-seismic permeability increases Pinyon Flat 1988–2006 Loma Prieta 1989 (H) Kobe 1995 (H) Alum Rock 2007 (V)
0–0.2 Shallow* Shallow* Shallow*
£4 10 3–15 3–10
Elkhoury et al. (2006) Rojstaczer et al. (1995) Sato et al. (2000) Manga & Rowland (2009)
Depth (km)
Locality
Postseismic permeability decreases Matsushiro 1965–1970 0–6 Pinyon Flat 1988–2006 Nojimaà 1997–2003 Locality
0–0.2 1.8
Depth (km)
Log k1 (m2)
Log k2 (m2)
Dt (years)
)12.6
)14 to )13
3–5
)14.5 )14.4
)15 )15
2 6
Log k1 (m2)
Log k2 (m2)
Permeability increases from enhanced geothermal system stimulation Soultz, France 2.85–3.4 )16.8 )14.5 Basel, Switzerland 4.6–5.0 )17 )14.4
Reference
Ohtake (1974), Cappa et al. (2009) Elkhoury et al. (2006) Kitagawa et al. (2007) Dt§
Reference
15 d 6d
Evans et al. (2005) Ha¨ring et al. (2008)
For co-seismic permeability increases, V and H denote models inferring dominantly vertical and horizontal fluid flow, respectively. *Models for the Loma Prieta and Kobe responses are based on lateral groundwater flow in systems with total water-table relief of <1 km. The Alum Rock response entailed minor changes in temperature (1–2C), suggesting relatively shallow fluid sourcing. Co-seismic permeability (1965–1967) based on numerical modeling constrained by ground-deformation data (Cappa et al. 2009); postseismic (1970) permeability based on deep-well injection testing (Ohtake 1974). àBased on a series of three injection experiments following the 1995 Kobe earthquake. §Duration of hydraulic stimulation.
permeability decayed steadily toward background values of log k )15 m2 (for one of the two monitored wells) and log k )14.2 m2 (for the other well) over a period of several years (Fig. 5). Numerous investigators have studied the postseismic evolution of permeability in the Nojima, Japan, fault zone following the 1995 Kobe earthquake. Several direct and indirect experiments at Nojima agree
Fig. 5. Permeability response to the 1999 Hector Mine earthquake at the Pin˜on Flat Observatory, California Coast Ranges (from Elkhoury et al. 2010).
that permeability decreased by 40–70% over the 8 years following the earthquake (Tanaka et al. 2007). Both the California Coast Range study and the Nojima studies entailed direct measurement of permeability. Lillemor Claesson and colleagues have inferred postearthquake permeability decay indirectly on the basis of geochemical changes in wells in northern Iceland (Claesson et al. 2007) and northeastern India (Claesson 2007). They inferred substantial permeability decreases over similar time scales of 100 to 101 years. Co-seismic changes in streamflow and groundwater levels in the California Coast Ranges also provide inferential evidence for rates of permeability decay. Prior to the 1989 Loma Prieta earthquake, the water table below ridgelines in the Santa Cruz Mountains was very near the land surface, where it could be tapped by shallow wells. The Loma Prieta earthquake caused a roughly 10-fold increase in shallow permeability, resulting in both temporarily increased streamflow and groundwater-level declines that caused shallow wells to go dry (Rojstaczer & Wolf 1994; Rojstaczer et al. 1995). On the basis of pre-Loma Prieta conditions, one can reasonably infer that water levels (and permeability) on the San Francisco peninsula had reequilibrated between the time of the great 1906 San Francisco earthquake and the 1989 Loma Prieta earthquake. Further, anecdotal reports indicate partial recovery of water levels between 1989 and the time of this writing (S.A.
200 INGEBRITSEN & MANNING
Lower limits of crustal permeability? We have inferred that permeability may decay fairly rapidly from the ‘high’ values listed in Table 1 and depicted in Figs. 4 and 6. How far will it decay? Is there an identifiable lower bound to crustal permeability? Conduction-dominated heat transport seems to be the norm below a few kilometers depth in the crust. The importance of heat advection relative to heat conduction can be represented by the Nusselt number Nu, which is the ratio, in a particular dimension, of the total heat transfer to the heat transfer that would be expected in the absence of advection. For conditions in the deeper crust during prograde metamorphism, we can assume dominantly upward flow of both heat and matter and write Nu ¼
cf qqz T þ ðKm ðTL TU Þ=LÞ ; ðKm ðTL TU Þ=LÞ
ð11Þ
where cf, q, and T are the heat capacity, density, and temperature of fluid flowing upward at a volumetric rate
(A)
0
Depth (km)
10
Manning & Ingebritsen (1999)
20 This study
30
40
(B)
0 Ingebritsen & Manning (2002) Brittle 10
Depth (km)
Rojstaczer, oral communication, 2009). Thus, we can estimate that decay of a roughly 10-fold increase in shallow permeability in the Coast Ranges requires 101 to 102 years. One can also make an inferential argument for rates of fracture healing along the margin of the Skaergaard intrusion, East Greenland, a contact-metamorphic locality where a layered gabbroic intrusion was emplaced at a depth of approximately 3.5 km within a 6- to 7-km-thick section of extrusive basalts. Despite field evidence for multiple episodes of porosity generation and fracturing within about 250 m of the intrusion (Manning & Bird 1991), any associated permeability increase must have been short-lived (approximately <103 years), because the observed metamorphic mineral assemblages require the high temperatures associated with conduction-dominated cooling (Manning et al. 1993). Initial ad hoc attempts to model crustal-scale permeability as a dynamically self-adjusting parameter – conditioned on similar evidence – have assumed that substantial loss of permeability requires times of decades to centuries (Rojstaczer et al. 2008). We infer that in dynamic geologic environments, permeability can decay substantially (by a factor of 2–10 or more) over geologically short time scales of 100 to 103 years. Table 2 assembles quantitative evidence for magnitudes and rates of in situ permeability increase and decrease in the brittle upper crust. Co-seismic shaking associated with regional earthquakes has been observed to cause 3- to 10fold, quasi-instantaneous increases in the permeability of the uppermost crust. Postseismic permeability decreases of similar magnitude have been observed over periods of several years. The relatively strong hydraulic forcing associated with EGS operations has caused 102-fold permeability increases over stimulation periods of days to weeks.
Ductile This study
20
30
40 –22
–20
–18 –16 –14 2 Log permeability (m )
–12
–10
Fig. 6. Evidence for relatively high crustal-scale permeabilities showing (A) power-law fit to data and (B) data below 12.5-km depth fitted with a constant value. Upper curve in both (A) and (B) is the best fit to geothermal– metamorphic data (Fig. 1). ‘High-permeability’ data points are midpoints in reported ranges in k and z for a given locality (Table 1); error bars depict the full permissible range for a plotted locality and are not Gaussian errors. The Dobi (Afar) earthquake swarm (Table 1) is not shown on this plot (it is off-scale).
(m3 m)2) of qz, respectively; Km is the thermal conductivity of the medium; and TL and TU are the temperatures at the upper and lower boundaries of a depth interval L, respectively. Figure 2 shows the approximate lower-crust permeability associated with a Nusselt number of 2 (the level at which advection = conduction) relative to various permeability data. The Nu 2 curve for the lower crust is quite similar to the geothermal–metamorphic and experimental permeability–depth curves. However, all of the high-permeability values of Table 1 would plot well above the Nu 2 curve, and thus seem unlikely to represent long-term stable conditions. Devolatilization-induced metamorphic permeability may be regulated by the heat flow-dependent kinetics of devolatilization. The positive feedback loop depicted in
Dynamic permeability of the continental crust 201 fact, very low permeabilities (cf. log k = )24.3 to )21.3; Audet et al. 2009) have been invoked to explain hydrodynamic phenomena. In a water-saturated medium, the concept of hydrodynamic permeability would seem entirely irrelevant only when and where the transport of oxygen (presumably as molecular water) by grain-boundary diffusion (e.g. Farver & Yund 1995) becomes more effective than Darcian flow.
DISCUSSION
Fig. 7. Feedback between permeability creation and decay during metamorphism. This feedback loop is driven by the heat flow-dependent kinetics of metamorphic devolatilization.
Fig. 7 would tend to constrain long-term average permeability to levels below those at which heat advection becomes significant; that is, Nu £2. Thus, we suggest that permeabilities in excess of Nu 2 would decay, perhaps relatively rapidly. It is difficult to define the lower limit toward which permeability may eventually decay, although the permeability of rocks undergoing retrograde (water-absorbing) metamorphism may be effectively zero (cf. Yardley & Valley 1997; Yardley & Baumgartner 2007). In watersaturated media, the concept of a ‘hydrodynamic’ permeability governed by Darcy’s law (eqn 5) would seem to be less relevant where transport of both heat and solutes is dominantly by diffusion. Because the diffusivity of heat in geologic media is many orders of magnitude greater than diffusivity of any atomic species, there is a large range of permeability within which heat will be largely conducted but solutes largely advected (Bickle & McKenzie 1987). However, below a Sherwood number (mass transfer Nusselt number) of Sh 2, transport of both heat and solutes will be mainly by diffusion. In low-porosity geologic media (0.1% porosity), the permeability associated with Sh 2 will be about 104 times lower (Bickle & McKenzie 1987, their fig. 6) than the permeability associated with Nu 2. Most of the permeability data that we have compiled lie well above the permeability associated with Sh 2. This is necessarily the case because all of the metamorphic data points (Fig. 1), for instance, entail positive geochemical evidence for fluid flow. Even where transport of heat and solutes are largely diffusive, it is possible that hydrodynamic permeability influences fluid pressures, and therefore crustal rheology. In
Some economic geologists, geophysicists, and metamorphic petrologists have long recognized permeability as a dynamic parameter that changes in response to dewatering and fluid production (e.g. Sibson et al. 1975; Walder & Nur 1984; Yardley 1986; Titley 1990; Hanson 1992, 1995, 1997; Dutrow & Norton 1995; Connolly 1997; Cox 2002; Sibson & Rowland 2003; Yardley & Baumgartner 2007). This view is in stark contrast to the hydrogeologic concept of permeability as a static material property that exerts control on fluid flow. The petrologic view of crustal permeability is consistent with indications that fluid pressure is close to the lithostatic load during prograde metamorphism (e.g. Fyfe et al. 1978); sufficiently overpressured fluids cannot be contained in the crust and create the permeability necessary to escape. Recently, it has been suggested that the permeability of the brittle upper crust may also be dynamically self-adjusting, responding to tectonism and external fluid sources as much as the lower crust responds to the magnitude of internal fluid sources (cf. Rojstaczer et al. 2008). The high-permeability data compiled here (Table 1, Figs 4 and 6) seems compatible with the concept of ‘dynamic permeability’ (Cathles & Adams 2005). Like the original compilation of geothermal–metamorphic permeabilities (Fig. 1), the high-permeability data of Table 1 suggest systematic variation with depth (Figs 4 and 6). A quantitative best fit to the data set as a whole yields log k 11:7 2:9 log z;
ð12Þ
with k in m2 and z in km. Fixing the slope at )3.2, the value derived by Manning & Ingebritsen (1999) for ‘geothermal–metamorphic’ data yields log k 11:5 3:2 log z;
ð13Þ
a closely similar result (Fig. 4). These fits are obtained by grouping all of the ‘high-permeability’ data of Table 1. Considered separately, the data from below 12.5 km appear depth-independent, like the geothermal–metamorphic data. The apparently similar organization of the geothermal– metamorphic and grouped ‘high-permeability’ data prompts consideration of the physical implications of the
202 INGEBRITSEN & MANNING empirical constants in the curve fits. The constant )14 of the original power-law geothermal–metamorphic curve (the permeability at 1-km depth from eqn 1) is similar to the mean permeability of the uppermost crust, as defined independently both by in situ well-test data (Brace 1980) and recharge-based calculations (Rojstaczer et al. 2008). The coefficient )3.2 can be inferred to reflect the magnitude of deep metamorphic (or other endogeneous) fluid fluxes. The similar form of the geothermal–metamorphic and high-permeability curves (eqns 1, 12, and 13) may perhaps reflect a confining-pressure dependence of porosity–strain and permeability–strain relations (cf. Cox et al. 2001, his fig. 1). Both the original geothermal–metamorphic data set (Fig. 1) and the ‘high-permeability’ data set (Figs 4 and 6) suggest a high variance and strong depth dependence of permeability at crustal depths of about <10 km, with less variance and essentially no depth dependence below 10-km depth. This supports a general distinction between the hydrodynamics of a brittle upper crust and a ductile lower crust that is dominated by devolatilization reactions and internally derived fluids. Both data sets can reasonably be fitted with a constant value of log k below 10-km depth, again with an offset of about 2 orders of magnitude (log k )18.3 versus log k )16.0). In the deeper crust, the rough coincidence of the geothermal–metamorphic curve and the curve for Nu 2 (Fig. 2) lends credence to the concept of thermally selfregulating metamorphic permeability (Fig. 7), as does the brevity of the episodes of heat advection inferred for metamorphism in Connecticut and Scotland (Table 1; Ague & Baxter 2007; Lancaster et al. 2008). Although the ‘highpermeability’ values summarized in Table 1 may be ephemeral in the context of geologic time, they can be crucially important from the standpoint of heat and mass transport. However, even these ‘high-permeability’ values for metamorphism are probably not the true transient permeabilities. In prograde metamorphism, fluid generation is an intermittent process that switches on an off when reaction boundaries are crossed. Produced fluid migrates through the crust as a high porosity ⁄ permeability wave (Connolly 1997). All of the common petrologic methods yield a time-integrated fluid flux and an average permeability, so that the full cycles of permeability build-up and decay are extremely difficult to resolve. Similarly, the average values of permeability obtained by modeling earthquake-hypocenter migration as a diffusive phenomenon (eqn 7) are smaller than the maximum values obtained when hypocenter migration is modeled as a solitary wave (cf. Miller et al. 2004). In the absence of independent constraints, it is nonetheless reasonable to invoke crustal-scale permeability–depth relations (such as eqns 1, 2, 4, 12, and 13) to make firstorder calculations related to large-scale hydraulic behavior
(e.g. Fulton et al. 2009; Lyubetskaya & Ague 2009) or crustal-scale volatile and solute transport (e.g. Ingebritsen & Manning 2002). However, such permeability–depth relations likely reflect a dynamic competition between permeability creation and permeability destruction. Further, all such relations imply a porous-continuum model for permeability behavior that may be more aptly represented in terms of hydraulic seals (Miller et al. 2003; Audet et al. 2009), two-layer models (Hanano 1998), or multidimensional growth of multiple hydraulic fractures (Hill 1977; Sibson 1996; Miller & Nur 2000). The applicability of continuum modeling to represent (for instance) multiple fractures depends in large part on the size of model elements relative to fracture spacing. More realistic and better-constrained representation of permeability heterogeneity and anisotropy are essential to many practical applications.
ACKNOWLEDGEMENTS We thank Shaul Hurwitz, Kurt Bucher, and an anonymous Geofluids referee for helpful reviews that greatly improved the final version of this paper.
REFERENCES Ague JJ (1994) Mass transfer during Barrovian metamorphism of pelites, south-central Connecticut. II. Channelized fluid flow and the growth of staurolite and kyanite. American Journal of Science, 294, 1,061–134. Ague JJ (1997) Crustal mass transfer and index mineral growth in Barrow’s garnet zone, northeast Scotland. Geology, 25, 73–6. Ague JJ, Baxter EF (2007) Brief thermal pulses during mountain building recorded by Sr diffusion in apatite and multicomponent diffusion in garnet. Earth and Planetary Science Letters, 261, 500–16. Audet P, Bostock MG, Christensen NI, Peacock SM (2009) Seismic evidence for overpressured subducted oceanic crust and megathrust fault sealing. Nature, 457, 76–8. Audin L, Avouac J-P, Flouzat M (2002) Fluid-driven seismicity in a stable tectonic context: the Remiremont fault zone, Vosges, France. Geophysical Research Letters, 29; doi: 10.1029/ 2001GL012988. Baker ET, Massoth GJ, Feely RA (1987) Cataclysmic hydrothermal venting on the Juan de Fuca Ridge. Nature, 329, 149–51. Baker ET, Lavelle JW, Feely RA, Massoth GJ, Walker SL, Lupton JE (1989) Episodic venting of hydrothermal fluids from the Juan de Fuca Ridge. Journal of Geophysical Research, 94, 9,237– 50. Bickle MJ, McKenzie D (1987) The transport of heat and matter by fluids during metamorphism. Contributions to Mineralogy and Petrology, 95, 644–65. Brace WF (1980) Permeability of crystalline and argillaceous rocks. International Journal of Rock Mechanics and Mining Sciences and Geomechanics Abstracts, 17, 241–51. Cappa F, Rutqvist J, Yamamoto K (2009) Modeling crustal deformation and rupture processes related to upwelling of deep CO2rich fluids during the 1965-1967 Matsuhiro earthquake swarm in Japan. Journal of Geophysical Research, 114; doi: 10.1029/ 2009JB006398.
Dynamic permeability of the continental crust 203 Cathles LM, Adams JJ (2005). Fluid flow and petroleum and mineral resources in the upper (20 km) continental crust. Economic Geology 100th Anniversary Volume, 77–110. Cathles LM, Erendi HJ, Barrie T (1997) How long can a hydrothermal system be sustained by a single intrusive event? Economic Geology, 92, 766–71. Challandes N, Marquer D, Villa IM (2008) P-T-t modelling, fluid circulation, and 39Ar-40Ar and Rb-Sr mica ages in the Aar Massif shear zones (Swiss Alps). Swiss Journal of Geosciences, 101, 269–88. Claesson L. (2007). Fluid-rock interaction in two seismically active areas. PhD thesis, Stockholm University, Sweden, 70 pp. + appendices. Claesson L, Skelton A, Graham C, Morth CM (2007) The timescale and mechanisms of fault sealing and water-rock interaction after an earthquake. Geofluids, 7, 427–40. Clauser C (1992) Permeability of crystalline rocks. Eos, Transactions American Geophysical Union, 73, 233. Connolly JAD (1997) Devolatilization-generated fluid pressure and deformation-propagated fluid flow during prograde regional metamorphism. Journal of Geophysical Research, 102, 18,149–73. Cox SF (2002) Fluid flow in mid- to deep crustal shear systems: experimental constraints, observations on exhumed high fluid flux shear systems, and implications for seismogenic processes. Earth Planets Space, 4, 1,121–5. Cox SF, Knackstedt MA, Braun J (2001) Principles of structural control on permeability and fluid flow in hydrothermal systems. Reviews in Economic Geology, 14, 1–24. Dipple GM, Ferry JM (1992) Metasomatism and fluid flow in ductile fault zones. Contributions to Mineralogy and Petrology, 112, 149–64. Dutrow D, Norton D (1995) Evolution of fluid pressure and fracture propagation during contact metamorphism. Journal of Metamorphic Geology, 13, 677–86. Elkhoury JE, Brodsky EE, Agnew DC (2006) Seismic waves increase permeability. Nature, 441, 1,135–8. Elkhoury JE, Niemeijer A, Brodsky EE, Marone C (2010) Dynamic stress stimulates flow in fractures: laboratory observations of permeability enhancement. Journal of Geophysical Research, 115; in press. Evans KF, Genter A, Sausss J (2005) Permeability creation and damage due to massive fluid injections into granite at 3.5 km at Soultz: 1. Borehole observations. Journal of Geophysical Research, 110; doi: 10.1029/2004JB003168. Farver JR, Yund RA (1995) Grain boundary diffusion of oxygen, potassium and calcium in natural and hot-pressed feldspar aggregates. Contributions to Mineralogy and Petrology, 118, 340–55. Fischer T, Hainzl S, Eisner L, Shapiro SA, Le Calvez J (2008) Microseismic signatures of hydraulic fracture growth in sediment formations: observations and modeling. Journal of Geophysical Research, 113; doi: 10.1029/2007JB005070. Fornari DJ, Shank T, Von Damm KL, Gregg TKP, Lilley M, Levai G, Bray A, Haymon RM, Perfit MR, Lutz R (1998) Time-series temperature measurements at high-temperature hydrothermal vents, East Pacific Rise 949¢–51¢N: evidence for monitoring a crustal cracking event. Earth and Planetary Science Letters, 160, 419–31. Fulton PM, Saffer DM, Bekins BA (2009) A critical evaluation of crustal dehydration as the cause of an overpressured and weak San Andreas Fault. Earth and Planetary Science Letters, 284, 447–54. Fyfe WS, Price NJ, Thompson AB (1978) Fluids in the Earth’s Crust. Elsevier Scientific, New York.
Germanovich LN, Lowell RP (1992) Percolation theory, thermoelasticity, and discrete hydrothermal venting in the Earth’s crust. Science, 255, 1564–7. Gillis KM, Roberts MD (1999) Cracking at the magma-hydrothermal transition: evidence from the Troodos Ophiolite, Cyprus. Earth and Planetary Science Letters, 169, 227–44. Golden CE, Webb SC, Sohn RA (2003) Hydrothermal microearthquake swarms beneath active vents at Middle Valley, northern Juan de Fuca Ridge. Journal of Geophysical Research, 108; doi: 10.1029/2001JB000226. Hanano M (1998) A simple model of a two-layered high-temperature liquid-dominated geothermal reservoir as part of a largescale hydrothermal convection system. Transport in Porous Media, 33, 3–27. Hanson RB (1992) Effects of fluid production on fluid flow during regional and contact metamorphism. Journal of Metamorphic Geology, 10, 87–97. Hanson RB (1995) The hydrodynamics of contact metamorphism. Geological Society of America Bulletin, 107, 595–611. Hanson RB (1997) Hydrodynamics of regional metamorphism due to continental collision. Economic Geology, 92, 880–91. Ha¨ring MO, Schanz U, Ladner F, Dyer BC (2008) Characterisation of the Basel 1 enhanced geothermal system. Geothermics, 37, 469–95. Haymon RM (1996) The response of ridge-crest hydrothermal systems to segmented, episodic magma supply. Geological Society Special Publication, 118, 157–68. Hill DP (1977) A model for earthquake swarms. Journal of Geophysical Research, 82, 347–52. Hill DP, Reasenberg PA, Michael A, Arabaz WJ, Beroza G, Brumbaugh D, Brune JN, Castro R, Davis S, dePolo D, Ellsworth WL, Gomberg J, Harmsen S, House L, Jackson SM, Johnston MJS, Jones L, Keller R, Malone S, Munguia L, Nava S, Pechmann JC, Sanford A, Simpson RW, Smith RB, Stark M, Stickney M, Vidal A, Walter S, Wong V, Zollweg J (1993) Seismicity remotely triggered by the magnitude 7.3 Landers, California, earthquake. Science, 260, 1,617–23. Hill DP, Prejean S (2005) Magmatic unrest beneath Mammoth Mountain, California. Journal of Volcanology and Geothermal Research, 146, 257–83. Horalek J, Fischer T (2008) Role of crustal fluids in triggering the West Bohemia ⁄ Vogtland earthquake swarms: just what we know (a review). Studies in Geophysics and Geodesy, 52, 455– 78. Hsieh PA (1998) Scale effects in fluid flow through fractured geologic media. In: Scale Dependence and Scale Invariance in Hydrology (ed. Sposito G), pp. 335–53. Cambridge University Press, New York. Hsieh PA, Bredehoeft JD (1981) A reservoir analysis of the Denver earthquakes: a case of induced seismicity. Journal of Geophysical Research, 86, 903–20. Husen S, Taylor R, Smith RB, Heasler H (2004) Changes in geyser eruption behavior and remotely triggered seismicity in Yellowstone National Park produced by the 2002 M 7.9 Denali fault earthquake, Alaska. Geology, 32, 537–40. Ingebritsen SE, Manning CE (1999) Geological implications of a permeability-depth curve for the continental crust. Geology, 27, 1,107–10. Ingebritsen SE, Manning CE (2002) Diffuse fluid flux through orogenic belts: implications for the world ocean. Proceedings of the National Academy of Sciences, USA, 99, 9,113–6. Ingebritsen SE, Scholl MA (1993) The hydrogeology of Kilauea volcano. Geothermics, 22, 255–70.
204 INGEBRITSEN & MANNING Johnson HP, Hutnak M, Dziak RP, Fox CG, Urcuyo I, Cowen JP, Nabelek J, Fisher C (2000) Earthquake-induced changes in a hydrothermal system on the Juan de Fuca mid-ocean ridge. Nature, 407, 174–7. Kitagawa Y, Fujimori K, Koizumi N (2007) Temporal change in permeability of the Nojima Fault zone by repeated water injection experiments. Tectonophysics, 443, 183–92. Lancaster PJ, Baxter EF, Ague JJ, Breeding CM, Owens TL (2008) Synchronous peak Barrovian metamorphism driven by syn-orogenic magmatism and fluid flow in southern Connecticut, USA. Journal of Metamorphic Geology, 26, 527–38. Lowell RP, Van Cappellen P, Germanovich LN (1993) Silica precipitation in fractures and the evolution of permeability in hydrothermal upflow zones. Science, 260, 192–4. Lyubetskaya T, Ague JJ (2009) Modeling the magnitudes and directions of regional metamorphic fluid flow in collisional orogens. Journal of Petrology, 50, 1,505–31. Manga M (1996) Hydrology of spring-dominated streams in the Oregon Cascades. Water Resources Research, 32, 2,435–9. Manga M (1997) A model for discharge in spring-dominated streams and implications for the transmissivity and recharge of Quaternary volcanics in the Oregon Cascades. Water Resources Research, 33, 1,813–22. Manga M, Rowland JC (2009) Response of Alum Rock springs to the October 30, 2007 Alum Rock earthquake and implications for the origin of increased discharge after earthquakes. Geofluids, 9, 237–50. Manning CE, Bird DK (1991) Porosity evolution and fluid flow in the basalts of the Skaergaard magmatic-hydrothermal system, East Greenland. American Journal of Science, 291, 201– 57. Manning CE, Ingebritsen SE (1999) Permeability of the continental crust: the implications of geothermal data and metamorphic systems. Reviews of Geophysics, 37, 127–50. Manning CE, Ingebritsen SE, Bird DK (1993) Missing mineral zones in contact metamorphosed basalts. American Journal of Science, 293, 894–938. Miller SA, Nur A (2000) Permeability as a toggle switch in fluidcontrolled crustal processes. Earth and Planetary Science Letters, 183, 133–46. Miller SA, van der Zee W, Olgaard DL, Connolly JAD (2003) A fluid pressure-feedback model of dehydration reactions: experiments, modelling, and application to subduction zones. Tectonophysics, 370, 241–51. Miller SA, Collettini C, Chiaraluce L, Cocco M, Barchi M, Kaus BJP (2004) Aftershocks driven by a high-pressure CO2 source at depth. Nature, 427, 724–7. Mitchell JK (1993) Fundamentals of Soil Behavior, 2nd edn. John Wiley and Sons, New York. Moore DE, Morrow CA, Byerlee JD (1983) Chemical reactions accompanying fluid flow through granite held in a temperature gradient. Geochimica et Cosmochimica Acta, 47, 445–53. Moore DE, Lockner DA, Byerlee JD (1994) Reduction of permeability in granite at elevated temperatures. Science, 265, 1,558– 61. Morrow C, Lockner D, Moore D, Byerlee J (1981) Permeability of granite in a temperature gradient. Journal of Geophysical Research, 86, 3,002–8. Morrow CA, Moore DE, Lockner DA (2001) Permeability reduction in granite under hydrothermal conditions. Journal of Geophysical Research, 106, 30,551–60. Neuzil CE (1995) Abnormal pressures as hydrodynamic phenomena. American Journal of Science, 295, 742–86.
Nippress SEJ, Rietbrock A (2007) Seismogenic zone high permeability in the central Andes inferred from relocations of microearthquakes. Earth and Planetary Science Letters, 263, 235–45. Noir J, Jacques E, Bekri S, Adler PM, Tapponnier P, King GCP (1997) Fluid flow triggered migration of events in the 1989 Dobi earthquake sequence of Central Afar. Geophysical Research Letters, 24, 2,335–8. Ohtake M (1974) Seismic activity induced by water injection at Matsuhiro, Japan. Journal of Physics of the Earth, 22, 163–76. Polak A, Elsworth D, Yasuhara H, Grader AS, Halleck PM (2003) Permeability reduction of a natural fracture under net dissolution by hydrothermal fluids. Geophysical Research Letters, 30; doi: 10.1029/2003GL017575. Rojstaczer SA, Wolf S. (1994). Hydrologic changes associated with the earthquake in the San Lorenzo and Pescadero drainage basins. In: The Loma Prieta, California Earthquake of October 17, 1989 – Hydrologic Disturbances (ed. Rojstaczer SA), US Geological Survey Professional Paper 1551-E, E51–64. Rojstaczer SA, Wolf S, Michel R (1995) Permeability enhancement in the shallow crust as a cause of earthquake-induced hydrological changes. Nature, 373, 237–9. Rojstaczer SA, Ingebritsen SE, Hayba DO (2008) Permeability of continental crust influenced by internal and external forcing. Geofluids, 8, 128–39. Saar MO, Manga M (2004) Depth dependence of permeability in the Oregon Cascades inferred from hydrogeologic, thermal, seismic, and magmatic modeling constraints. Journal of Geophysical Research, 109; doi: 10.1029/2003JB002855. Sato T, Sakai R, Furuya K, Kodama T (2000) Coseismic spring flow changes associated with the 1995 Kobe earthquake. Geophysical Research Letters, 27, 1,219–22. Shapiro SA, Dinske C (2009) Fluid-induced seismicity: pressure diffusion and hydraulic fracturing. Geophysical Prospecting, 57, 301–10. Shapiro SA, Huenges E, Borm G (1997) Estimating the crust permeability from fluid-injection induced seismic emission at the KTB site. Geophysical Journal International, 131, F15–8. Shmonov VM, Vitiovtova VM, Zharikov AV, Grafchikov AA (2002) Fluid permeability of the continental crust: estimation from experimental data. Geochemistry International, 40(Suppl. 1), S3–13. Shmonov VM, Vitiovtova VM, Zharikov AV, Grafchikov AA (2003) Permeability of the continental crust: implications of experimental data. Journal of Geochemical Exploration, 78–79, 697–9. Sibson RH (1996) Structural permeability of fluid-driven faultfracture meshes. Journal of Structural Geology, 18, 1031– 42. Sibson RH, Rowland JV (2003) Stress, fluid pressure, and structural permeability in seismogenic crust, North Island, New Zealand. Geophysical Journal International, 154, 584–94. Sibson RH, Moore JMM, Rankin AH (1975) Seismic pumping – a hydrothermal fluid transport mechanism. Journal of the Geological Society of London, 131, 653–9. Sohn RA (2007) Stochastic analysis of exit fluid temperature records from the active TAG hydrothermal mound (Mid-Atlantic Ridge, 26N): 1. Modes of variability and implications for subsurface flow. Journal of Geophysical Research, 112; doi: 10.1029/2006JB004435. Sohn RA, Fornari DJ, Von Damm KL, Hildebrand JA, Webb SC (1998) Seismic and hydrothermal evidence for a cracking event on the East Pacific rise crest at 950¢ N. Nature, 396, 159–61.
Dynamic permeability of the continental crust 205 Stein JS, Fisher AT (2003) Observations and models of lateral hydrothermal circulation on young ridge flank: numerical evaluation of thermal and chemical constraints. Geochemistry Geophysics Geosystems, 4; doi: 10.1029/2002GC000415. Stober I, Bucher K (2007) Hydraulic properties of the crystalline basement. Hydrogeology Journal, 15, 213–24. Summers R, Winkler K, Byerlee J (1978) Permeability changes during the flow of water through Westerly Granite at temperatures of 100–400C. Journal of Geophysical Research, 83, 339–44. Talwani P, Acree S (1984) Pore pressure diffusion and the mechanism of reservoir induced seismicity. Pure and Applied Geophysics, 122, 947–65. Talwani P, Chen L, Gahalaut K (2007) Seismogenic permeability, ks. Journal of Geophysical Research, 112; doi: 10.1029/ 2006JB004665. Tanaka H, Chester FM, Mori JJ, Wang C-Y (2007) Preface – drilling into fault zones. Tectonophysics, 44, 123–5. Tenthorey E, Scholz CH, Aharonov E, Leger A (1998) Precipitation sealing and diagenesis 1. Experimental results. Journal of Geophysical Research, 103, 23,951–67. Titley SR (1990) Evolution and style of fracture permeability in intrusion-centered hydrothermal systems. In: The Role of Fluids in Crustal Processes (eds Bredehoeft JD, Norton DL), pp. 50– 63. National Academy Press, Washington, DC. Townend J, Zoback MD (2000) How faulting keeps the crust strong. Geology, 28, 399–402. Urabe T, et al. (1995) The effect of magmatic activity on hydrothermal venting along the superfast-spreading East Pacific Rise. Science, 269, 1,092–5.
Vaughan PJ, Moore DE, Morrow CA, Byerlee JD (1986) Role of cracks in progressive permeability reduction during flow of heated aqueous fluids through granite. Journal of Geophysical Research, 91, 7,517–30. Waite GP, Smith RB (2002) Seismic evidence for fluid migration accompanying subsidence of the Yellowstone caldera. Journal of Geophysical Research, 107, doi:10.1029/2001JB000586. Walder J, Nur A (1984) Porosity reduction and crustal pore pressure development. Journal of Geophysical Research, 89, 11,539–48. Winograd IJ (1971) Hydrogeology of ash-flow tuff: a preliminary statement. Water Resources Research, 7, 994–1006. Yardley BWD (1986). Fluid migration and veining in the Connemara Schists. In: Fluid-Rock Reactions During Metamorphism, Advances in Physical Geochemistry 5 (eds Walther JV, Wood BJ), pp. 109–31. Springer-Verlag, New York. Yardley BWD, Baumgartner LP (2007). Fluid processes in deep crustal fault zones. In: Tectonic Faults – Agents of Change on a Dynamic Earth (eds Handy MR, Hirth G, Hovius N), pp. 295– 318. The MIT Press, Cambridge. Yardley BWD, Valley JW (1997) The petrologic case for a dry lower crust. Journal of Geophysical Research, 102, 12,173–85. Yasuhara H, Polak A, Mitani Y, Grader AS, Halleck PM, Elsworth D (2006) Evolution of fracture permeability through fluid-rock reaction under hydrothermal conditions. Earth and Planetary Science Letters, 244, 186–200. Zhang S, Paterson MS, Cox SF (2001) Microcrack growth and healing in deformed calcite aggregates. Tectonophysics, 335, 17– 36.
Hydrologic responses to earthquakes and a general metric CHI-YUEN WANG AND MICHAEL MANGA Department of Earth and Planetary Science, University of California, Berkeley, CA, USA
ABSTRACT Hydrologic responses to earthquakes, including liquefaction, changes in stream and spring discharge, changes in the properties of groundwater such as geochemistry, temperature and turbidity, changes in the water level in wells, and the eruption of mud volcanoes, have been documented for thousands of years. Except for some water-level changes in the near field which can be explained by poroelastic responses to static stress changes, most hydrologic responses, both within and beyond the near field, can only be explained by the dynamic responses associated with seismic waves. For these responses, the seismic energy density e may be used as a general metric to relate and compare the various hydrologic responses. We show that liquefaction, eruption of mud volcanoes and increases in streamflow are bounded by e 10)1 J m)3; temperature changes in hot springs are bounded by e 10)2 J m)3; most sustained groundwater changes are bounded by e 10)3 J m)3; geysers and triggered seismicity may respond to seismic energy density as small as 10)3 and 10)4 J m)3, respectively. Comparing the threshold energy densities with published laboratory measurements, we show that undrained consolidation induced by dynamic stresses can explain liquefaction only in the near field, but not beyond the near field. We propose that in the intermediate field and far field, most responses are triggered by changes in permeability that in turn are a response to the cyclic deformation and oscillatory fluid flow. Published laboratory measurements confirm that changes in flow and time-varying stresses can change permeability, inducing both increases and decreases. Field measurements in wells also indicate that permeability can be changed by earthquakes in the intermediate field and far field. Further work, in particular field monitoring and measurements, are needed to assess the generality of permeability changes in explaining far-field hydrologic responses to earthquakes. Key words: liquefaction, permeability change, triggered seismicity, geysers, mud volcanoes Received 22 July 2009; accepted 21 October 2009 Corresponding author: Chi-Yuen Wang, Department of Earth and Planetary Science, University of California, Berkeley, CA, USA. Email:
[email protected]. Tel: (510) 642-2288. Fax: (510) 643-9980 Geofluids (2010) 10, 206–216
INTRODUCTION Over the past two thousand years, a large variety of hydrologic changes following earthquakes has been documented. Examples include liquefaction of soils, changes in the eruption behavior of mud volcanoes and geysers, the formation of new springs, disappearance of previously active springs, increased discharge in streams, and changes in the properties of groundwater such as color, temperature, composition and water pressure (e.g., Montgomery & Manga 2003). It is not unexpected that earthquakes can cause hydrologic changes because the stresses created by earthquakes can be large compared to other time-varying stresses. What is surprising are the large amplitudes of hydrologic responses and the great distances over which
these changes occur: our focus in this paper are hydrologic responses at such large distances. Hydrologic responses to earthquakes are not just curiosities: because earthquakes and water interact with each other through changes in both stress and physical properties of rocks, understanding the origin of hydrological responses can provide new insights into hydrogeologic and tectonic processes at scales in space and time that are otherwise difficult to study. Besides being a matter of academic interest, the study of earthquake-induced hydrologic changes also has important implications for water resources and engineering enterprise. For example, groundwater level changes following earthquakes can affect water supplies (Chen & Wang 2009) and it is sometimes necessary to evaluate the causative role of an earthquake in insurance
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Hydrologic responses to earthquakes and a general metric 207 claims for loss of water supply (Roeloffs 1998). Forensic earthquake hydrology was also applied to evaluate whether an earthquake may have played a causative role in the 2006 eruption of a mud volcano eruption in the Indonesian city of Sidoarjo, in eastern Java, that led to massive destruction of property, infrastructure and evacuation of people (Manga 2007). Debate on the role of an earthquake in this mud volcano continues (Mazzini et al. 2007; Davies et al. 2008). Groundwater level changes following earthquakes may also put some underground nuclear waste repositories at risk (Carrigan et al. 1991; Roeloffs 1998). Earthquake-induced fluid pressure changes can initiate liquefaction of the ground that in turn causes great damage to engineered structures (e.g., Seed & Lee 1966), affect oil well production (Beresnev & Johnson 1994), and sometimes trigger seismicity (Hill & Prejean 2007). Finally, measured changes of the pore pressure in rocks and ⁄ or the chemical composition of groundwater are sometimes taken as signatures of the crustal response to tectonic deformation (e.g., Davis et al. 2001) or even as earthquake precursors (e.g., Silver & Wakita 1996). Earthquakes cause both static and dynamic (shaking) changes of the stress in the crust. Both types of stress change decrease with increasing distance from the earthquake, but at different rates. For example, the peak amplitude of the time-dependent change in the Coulomb failure stress (peak DCFS(t)) diminishes much less rapidly with distance than the static change (DCFS) in the same stress (e.g., Kilb et al. 2002). Thus at close distances the ratio (peak DCFS(t)) ⁄ DCFS is approximately proportional to the source-receiver distance, r, and at larger distances proportional to r2 (Aki & Richards 1980). At distances up to 1 ruptured fault length, the static and the peak dynamic changes are comparable in magnitude, while at distances greater than a few ruptured fault lengths, the peak dynamic change is much greater than the static change. The relative magnitude of static and dynamic stresses is reflected in the hydrologic responses to earthquakes and is critical to understanding the origin of hydrological changes. Hereafter we use the expression ‘near-field’ to denote distances up to 1 ruptured fault length, ‘far-field’ to denote distance many times greater than the ruptured fault length, and ‘intermediate-field’ for distance in between. In the past two decades, there have been rapid development and deployment of geophysical and hydrological instruments around the globe. The number and especially the quality of documented hydrological responses to earthquakes have expanded rapidly, which, in turn, have stimulated more detailed analysis and hypothesis testing. Here we use a common metric, the seismic energy density, to relate the variety of hydrologic responses with each other and, at the same time, to compare the dissipated energy for the hydrologic responses with that measured in laboratory experiments so that some physical constraints may be
provided in the discussion and hypothesis testing of the triggering mechanism of the hydrologic responses. For a more complete discussion of this general topic we refer readers to a volume by Wang & Manga (2010). An earlier and more brief review on the general topic of earthquake hydrology is also available in Manga & Wang (2007).
OBSERVATIONS In Fig. 1 we compile observations of hydrologic responses to earthquakes for several classes of phenomena: liquefaction, changes in streamflow, changes in spring temperature, changes in eruption behavior of geysers, and the eruption of mud volcanoes and magmatic volcanoes. The sources of these observations and data are numerous and we have tabulated them in Table 1, and many are described in more detail in Wang & Manga (2010). These data come from the refereed literature or are based on readily accessed archival data. As noted by many authors (e.g., Mogi et al. 1989; Roeloffs 1998; King et al. 1999; Manga & Wang 2007; Wang & Chia 2008), the distribution of a variety of hydrologic responses may be scaled by the earthquake magnitude M and distance r from the earthquake source. These parameters, i.e., r and M, are used to characterize the occurrences of hydrologic responses because the majority of
Fig. 1. Distribution of earthquake-triggered hydrologic changes as a function of earthquake magnitude and distance. Also plotted are the log r versus M contours of constant seismic energy density e, which is the seismic energy in a unit volume in the seismic wave train; it thus represents the maximum seismic energy available to do work at a given location during the earthquake. No distinction is made among the different magnitude scales because the majority of documentations (many historical) do not note such distinction. For comparison purposes, we also plot earthquake-induced magmatic volcanic eruptions. Data and sources are listed in Table 1.
208 C.-Y. WANG & M. MANGA Table 1 Features of earthquake triggered hydrologic phenomena. Phenomenon
Earthquake
Magnitude
Distance (km)
Reference
Spring discharge Spring discharge Spring discharge Spring discharge Streamflow Spring discharge Streamflow Streamflow Streamflow Streamflow
Alum Rock18 Oct 1989 Alum Rock 30 Oct 2007 Alum Rock 13 Jun 1988 Mount Lewis 31 Mar 1986 Honeydew 17 Aug 1991 Morgan Hill 24 Apr 1984 San Simeon 22 Dec 2003 San Fernando 9 Feb 1971 Northridge 17 Jan 1994 Nisqually 28 Feb 2001
5.0 5.2 5.3 5.7 6.1 6.2 6.5 6.6 6.7 6.8
King et al. (1994) Rowland et al. (2008); Manga & Rowland (2009) King et al. (1994) King et al. (1994) McPherson & Dengler (1992) King et al. (1994) Wang et al. (2004a) Manga et al. (2003) Manga et al. (2003); Quilty et al. (1995) Montgomery et al. (2003)
Streamflow Streamflow
Borah Peak 28 Oct 1983 Loma Prieta 18 Oct 1989
7.0 7.1
5 4 8 15 7 18 38, 72 47 44, 55 13, 16, 18, 20, 23, 24, 26, 29, 30, 34, 40, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 58, 60, 63, 65, 66, 69, 79, 80, 85, 87, 88, 90, 92, 97, 109, 114 12, 13, 40, 51, 64, 68, 78, 100 4, 6, 8, 10, 12, 13, 19, 25, 27, 28, 35, 36, 39, 46, 54, 61, 86
Spring discharge Spring discharge
Loma Prieta 18 Oct 1989 Kobe 17 Jan 1995
7.1 7.2
Streamflow Streamflow Streamflow
Hebgen Lakek 18 Aug 1959 Landers 28 Jun 1992 Kern County 21 Jul 1952
7.3 7.3 7.5
Springs and streamflow
San Francisco 18 Apr 1906
7.9
Streamflow
Alaska 27 Mar 1964
9.2
37, 41, 48, 66, 95, 115, 118, 136, 148, 161, 166, 170, 178, 196, 205, 213, 246, 275, 340, 407 92, 121, 132, 320
Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Mud volcano Water level
16 May 1968 4 Mar 1952 21 Mar 1982 15 Jan 1993 28 Dec 1994 25 Dec 2003 26 Dec 2004 28 Jan 1872 13 Feb 1902 28 Nov 1945 30 May 1935 8 Jul 1895 4 Mar 1977 24 Sept 1848 4 Dec 1957 15 Jun 2006 26 Jan 2001 11 Oct 1915 4 Sept 1895 5 Apr 1781 91 BCE 13 Dec 1990 4 Oct 1978 5 Mar 1828 Alaska 27 Mar 1964
8.2 8.6 6.7 7.6 7.8 8.3 9.1 5.7 6.9 8.1 7.7 8.2 7.2 4.6 8.3 5.8 7.7 5.0 5.0 5.8 5.7 5.7 5.2 5.9 9.2
186 58 23 153 226 145 900 24, 40 45, 51 41, 155, 189 61 141 92 15 75 90 482 21 4 87 15, 16 39 34 56 5000,7215
40 3, 4, 5, 6, 7, 8, 10, 11, 13, 18, 19, 20, 21, 22, 23, 25, 26, 27, 29, 34, 36, 38, 40, 43, 44, 45, 47, 49, 50 37, 41, 46, 57, 60, 62 47, 225 31, 38, 46, 48, 52, 57, 63, 71, 83, 99, 117, 131
Muir-Wood & King (1993) Muir-Wood & King (1993); Briggs (1991); Rojstaczer et al. (1995) as reported in Montgomery et al. (2003) King et al. (1994) Sato et al. (2000)
Muir-Wood & King (1993) Roeloffs et al. (1995); Manga et al. (2003) Muir-Wood & King (1993); Manga et al. (2003); Briggs & Troxell (1955) as reported in Montgomery et al. (2003) Lawson (1908) as compiled in Montgomery et al. (2003) Waller (1966, 1966) as reported in Montgomery et al. (2003) Chigira & Tanaka (1997) Chigira & Tanaka (1997) Chigira & Tanaka (1997) Chigira & Tanaka (1997) Chigira & Tanaka (1997) Manga & Brodsky (2006) Manga & Brodsky (2006) Mellors et al. (2007) Mellors et al. (2007) Delisle (2005) Snead (1964) Mellors et al. (2007) Mellors et al. (2007) Mellors et al. (2007) Rukavickova & Hanzl (2008) Rukavickova & Hanzl (2008) Deville (2008, submitted for publication) Martinelli et al. (1989) in Bonini (2009) Trabucco (1895) in Bonini (2009) De Brignoli di Brunhoff (1836) in Bonini (2009) Guidoboni (1989) in Bonini (2009) D’Alessandro et al. (1996) in Bonini (2009) Silvestri (1978) in Bonini (2009) La Via (1828) in Bonini (2009) Cooper et al. (1965); Vorhis (1968)
Hydrologic responses to earthquakes and a general metric 209 Table 1 (Continued). Phenomenon
Earthquake
Magnitude
Distance (km)
Reference
Water Water Water Water Water Water Water Water Water Water Water Water
Loma Prieta 18 Oct 1989 Landers 28 Jun 1992 Chittenden 18 Apr 1990 Parkfield 20 Oct 1992 Parkfield 14 Nov 1993 Northridge 17 Jan 1994 Parkfield 20 Dec 1994 Hector Mine 16 Oct 1999 Petrolia 1 Sept 1994 Oaxaca 30 Sept 1999 Tokachi-oki 25 Sept 2003 26 Dec 2004
6.9 7.3 5.4 4.7 4.8 6.7 5.0 7.1 7.0 7.4 8.3 9.3
157, 272 433, 451 133 233 269 281 24 421 300, 732 3850 1200 5000, 10800
Water level
Chi-Chi 21 Sept 1999
7.5
15 to 170
Water level Water level Spring temperature Liquefaction Liquefaction Liquefaction
Hengchun 26 Dec 2006 Various 1981–1997 Various 1982–1986 Various 1848–1983 Various 1117 26 May 1983
7.0 3.6 to 8.1 2.3 to 7.9 4.8 to 9.2 5 to 8 7.7
60 to 220 13 to1253 2 to 452 0 to 480 0 to 111 160
Liquefaction Liquefaction
21 Aug 1988 Loma Prieta 18 Oct 1989
6.6 6.9
100 93
Liquefaction
9 Oct 1995
7.3
150
Liquefaction
17 Jan 1995
6.9
40
Liquefaction Liquefaction Liquefaction Liquefaction Liquefaction Liquefaction Liquefaction Magmatic volcanoes Magmatic volcanoes
Izmit1 17 Aug 1999 Ducze 12 Nov 1999 Chi-Chi 21 Sept 1999 Gujarat 26 Jan 2001 Nisqually 28 Feb 2001 Denali 3 Nov 2002 Colima 21 Jan 2003 Various 1730–1960 Various 1973–2005
7.8 7.5 7.5 7.7 6.8 7.9 7.7 8.0 to 8.7 4.8 to 7.8
61 56 80 260 75 300 60 116 to 701 5 to 156
Roeloffs (1998); Roeloffs et al. (2003) Roeloffs (1998); Roeloffs et al. (2003) Roeloffs (1998) Roeloffs (1998) Roeloffs (1998) Roeloffs (1998) Roeloffs (1998) Roeloffs et al. (2003) Brodsky et al. (2003), Roeloffs (1998) Brodsky et al. (2003) Sato et al. (2004) Sil & Freymueller (2006); Kitagawa et al. (2006) 247 observations in Wang & Chia (2008) 105 observations in Wang & Chia (2008) 28 observations by Matsumoto et al. (2003) 8 observations by Mogi et al. (1989) 137 observations in Ambraseys (1988) 315 observations in Galli (2000) http://www.ce.berkeley.edu/~ hausler/sites/NKC001.pdf Bardet & Kapuskar (1993) http://www.lafire.com/ famous_fires/940117_ NorthridgeEarthquake/quake/ 02_EQE_geology.htm sun1.pue.upaep.mx ⁄ servs ⁄ carrs ⁄ GIIS ⁄ manzanillo.html http://www.jrias.or.jp/public/ Hanshin_Earthquake/q1-2e.html Rothaus et al. (2004) geoinfo.usc.edu ⁄ turkey ⁄ Yu et al. (2000) Rajendran et al. (2001) Pierepiekarz et al. (2001) Kayen et al. (2002) geoinfo.usc.edu ⁄ gees ⁄ 11 observations in Manga & Brodsky (2006) 18 observations in Lemarchand & Grasso (2007)
level level level level level level level level level level level level
documentations (many historical) do not note the style of faulting, the directivity of fault rupture, or the distance to the ruptured fault, nor do they make a distinction among the different magnitude scales. Thus they are all simply plotted on a log r versus M diagram, as in Fig. 1. The seismic energy density: a general metric While M and r are useful parameters for comparing and relating the various hydrologic responses, it would be more convenient if we could replace these two parameters with a single parameter to relate the responses. Ideally this parameter would be a physical quantity that can be measured in laboratories, and hence could provide a useful tie between field observations and laboratory measurements. The
parameter ‘seismic energy density’, defined as the maximum seismic energy available in a unit volume to do work on rock or sediment, may serve this purpose. This is because it may be easily estimated from M and r (equation 2) and, at the same time, may be compared with laboratory results on the dissipated energy required to initiate pore pressure change and liquefaction in saturated sediments. The seismic energy density e is defined and given by (e.g., Lay & Wallace 1995) e¼
1X 2 i Ti
Z
vi ðtÞ2 dt
ð1Þ
where q is density, and Ti and vi are the period and particle velocity of the ith mode, and the sum is taken over all
210 C.-Y. WANG & M. MANGA modes for ground motion. For an earthquake of magnitude M, the seismic energy density e at a distance r from the earthquake source may be estimated from the empirical relation (Wang 2007) log r ¼ 0:48M 0:33 log e 1:4;
ð2Þ
where r is in km and e in J m)3. This relation implies that contours of constant e plot as straight lines on a log r versus M diagram – these are added to Fig. 1. The seismic energy density is approximately proportional to the square of the peak ground velocity (Wang et al. 2006) which in turn is proportional to the dynamic strain (e.g., Brodsky et al. 2003), and thus provides a convenient and physically meaningful metric to relate and compare the various hydrologic responses. Figure 1 shows that certain hydrologic responses require much greater seismic energy density (e.g., liquefaction, mud volcanoes) than others (e.g., well level changes, earthquake triggering). On the other hand, each type of hydrologic response spans over four or more orders of magnitude of the seismic energy density values. Scatter may be expected for two reasons. First, if triggering is a threshold process, then for all distances up to the threshold we might expect triggering to be possible. Second, because the hydro-mechanical properties of rocks and sediments are greatly different, the range of sensitivity to seismic shaking may be large. Thus, for any specific hydrologic response to occur, less seismic energy density is required at sites underlain by sediments or rocks more sensitive to seismic disturbances than at sites underlain by less sensitive rocks or sediments. Without a priori knowledge of the seismic sensitivity of the rocks and sediments at a given site, we may compare the different hydrologic responses by focusing on the threshold seismic energy density, i.e., the lower bound of the seismic energy density required to initiate a specific hydrological response in the most sensitive materials. With this simplification, Fig. 1 reveals that different hydrologic responses are bounded by different threshold seismic energy densities. Thus liquefaction, mud volcanoes and streamflow increases are bounded by the contour with e 10)1 J m)3; temperature changes in hot springs are bounded by the contour with e 10)2 J m)3, and most sustained groundwater changes are bounded by the contour with e 10)3 J m)3. An exception for the sustained water-level change is a data point that falls on a contour of e 10)4 J m)3 (Fig. 1). This data point represents the response at a well installed in fractured granite (Brodsky et al. 2003), which appears to be particularly sensitive to seismic disturbances (Woodcock & Roeloffs 1996). Geysers have long been known to be particularly sensitive to earthquakes, as manifested by changes in the time interval between eruptions (Ingebritsen & Rojstaczer 1993). Some geysers in the Yellowstone National Park, for example, have responded to seismic energy density as small
as 10)3 J m)3 (Fig. 1; see also Husen et al. 2004). Given the limited number of data, however, we are unable to confirm whether an M versus log r relationship may also apply to geysers. Triggered seismicity also appears to be particularly sensitive to seismic disturbances and may respond to e as small as 10)4 J m)3 (Fig. 1; Brodsky & Prejean 2005; Hill & Prejean 2007). In their study of triggered earthquakes at Long Valley, California, following the 2004 Denali earthquake in Alaska, Brodsky & Prejean (2005) showed that the early arriving 30 s surface waves triggered local seismicity, while high-frequency seismic waves with comparable cumulative seismic energy density from regional events did not. As a consequence, Brodsky and Prejean dismissed seismic energy density as a threshold for triggering earthquakes because of the timing of the triggered events. Brodsky & Prejean (2005) also noted that these observations imply that low-frequency seismic waves may be more effective in triggering seismicity. In this regard, it is interesting to note that low-frequency seismic waves may also be more effective in causing coseismic water-level changes and liquefaction (Wong & Wang 2007) and triggered eruption of mud volcanoes (Manga et al. 2009). It is important to note that the question whether all triggered seismicity is a hydrological phenomenon is a matter of active debate (e.g., Hill 2008) and it is likely that some triggered earthquakes are not caused by earthquakeinduced re-distribution of pore pressure. Regardless of a clear hydrologic connection, triggered earthquakes by large (M > 9) earthquakes may be global (e.g., West et al. 2005; Velasco et al. 2008). Anti-triggering, that is suppression of small earthquakes, may also occur (Okubo & Wolfe 2008). Part of the differences in the threshold energy between different hydrological responses may be a result of incomplete data. For example, the data for hot spring temperature come from a single hot spring in Japan (Mogi et al. 1989); thus it is not unlikely that, when more hot springs are studied, the lower bound for the earthquake-induced temperature change in hot springs may become lower than 10)2 J m)3. On the other hand, most other data summarized in Fig. 1 are abundant, come from a wide range of geological settings, and thus the differences in the threshold energy may be significant. Undrained consolidation Earthquake engineers have long regarded undrained consolidation as the primary mechanism for pore pressure buildup and liquefaction in saturated sediments (National Science Council 1985). Laboratory experiments have demonstrated that the initiation of liquefaction by this mechanism requires a minimum dissipated energy density of 30 J m)3 (e.g., Green & Mitchell 2004; Wang 2007). This required minimum energy density, however, is more
Hydrologic responses to earthquakes and a general metric 211 than 102 times greater than the seismic energy density at the threshold distance for liquefaction in the field (Fig. 1). Another interesting contradiction between the hypothesis of undrained consolidation and far-field observations arises from the changes of groundwater level. Results from laboratory experiments show that a dissipated energy density of 0.1–10 J m)3 is required to initiate undrained consolidation in saturated sediments (e.g., Ishihara 1996). This is more than two orders of magnitude greater than the seismic energy density at the threshold distance for sustained groundwater-level changes, i.e., 10)3 J m)3 (Fig. 1). Thus it is clear that the mechanism of undrained consolidation may only account for a fraction of the documented liquefaction and groundwater level changes in the field. For liquefaction that occurs at e < 30 J m)3 and sustained pore-pressure changes that occur at distances beyond the bounds for undrained consolidation, a different mechanism is required. Changes in permeability Several mechanisms besides undrained consolidation have been proposed to explain hydrologic responses; these include the static poroelastic strain associated with fault displacement (Wakita 1975; Muir-Wood & King 1993; Quilty & Roeloffs 1997) and an enhanced permeability of the shallow crust by the dynamic strains associated with seismic waves (Mogi et al. 1989; Rojstaczer et al. 1995; Brodsky et al. 2003; Wang et al. 2004a,b). At distances beyond the near field, static poroelastic strain is so small that it cannot easily account for the large amplitude of the observed hydrologic changes (Rojstaczer et al. 1995; Manga & Wang 2007). Furthermore, the model often has difficulty in explaining the sign of the observed groundwater-level changes (Roeloffs 1998; Wang et al. 2001; Koizumi et al. 2004) and the persistent streamflow increases in response to multiple earthquakes of different mechanisms and orientations (Manga et al. 2003). Dynamic strain by itself cannot lead to sustained hydrologic changes, but it can dislodge blockage from fractures and thereby enhance permeability of the shallow crust and to cause re-distribution of pore pressure (Mogi et al. 1989; Roeloffs 1998; Brodsky et al. 2003; Elkhoury et al. 2006, 2009). Mogi et al. (1989) first suggested that seismic waves may dislodge obstacles from passageways feeding hot springs to enhance flow and to cause coseismic increases in hot spring temperature. The same mechanism was used to explain sustained changes in groundwater level (Roeloffs 1998; Brodsky et al. 2003; Wang & Chia 2008) and increased stream discharge (Rojstaczer et al. 1995; Wang et al. 2004a,b) after large earthquakes. Changes in the eruption frequency of geysers can also be caused by
changes in permeability of the conduit and ⁄ or surrounding matrix (Ingebritsen & Rojstaczer 1993). As the permeability of the conduit is very high, changes in the matrix permeability that governs conduit recharge are more likely (Ingebritsen & Rojstaczer 1993; Manga & Brodsky 2006). Elkhoury et al. (2006) documented distinct transient shifts in the phase of the tidal response of water level in wells and interpreted these phase shifts in terms of an enhanced permeability caused by earthquakes. The magnitude of this enhancement increases with increased peak ground velocity, i.e., with increased seismic energy density. Other authors (e.g., Roeloffs 1998) also noticed that the amplitude of the sustained groundwater-level change at a given well increases in proportion to the increased peak ground velocity. These relations to peak ground velocity may be expected since greater seismic energy (more rigorous shaking) may clear more blockage from fluid passageways, resulting in a greater increase in permeability and more efficient redistribution of pore pressure. Taken together, it appears that enhanced permeability in the shallow crust during earthquakes may be a viable mechanism for a broad spectrum of hydrologic responses that occur in the intermediate and far fields. A wide range of experimental studies confirm that timevarying fluid flow can change permeability by dislodging particles or breaking up aggregates of colloidal pore-blocking particles (e.g., Cleasby et al. 1963; Bai & Tien 1997; Bergendahl & Grasso 2000; Gao et al. 2004). This process also occurs in the unsaturated zone (e.g., Saiers & Lenhart 2003). Elkhoury et al. (2009) documented transient increases in permeability in the laboratory following the application of time-varying pore pressure gradients at seismic frequencies. They attributed the changes in permeability to breaking up flocs of small particles. Liu & Manga (2009) applied time-varying stresses under undrained conditions and documented a permeability decrease in the laboratory – an affect they attribute to both particle redistribution within fractures and consolidation of their laboratory sample to reduce the aperture of fractures. In sum, lab experiments confirm that dynamic stresses and time-varying flow can change permeability, and both permeability increases and decreases may be possible. It remains to be shown that seismic waves with energy densities as small as 10)4 J m)3 can still change permeability. While direct measurement is not yet available, indirect estimate may be obtained from field observation. The 2002 M7.9 Denali earthquake, for example, enhanced groundwater flow in Iowa, some 5000 km away, to such an extent that colloidal particles were flushed out of local aquifers and discolored well water (Prior et al. 2003). Referring to Fig. 1, this event occurred at a seismic energy density of 10)4 J m)3. Thus, starting at a threshold energy density of 10)4 J m)3, seismic waves may dislodge minute clogs
212 C.-Y. WANG & M. MANGA from fractures to enhance permeability and redistribute pore pressure, triggering seismicity and causing groundwater-level to change in the most sensitive wells. Increasing seismic energy density may remove larger blockages from fractures to allow more efficient groundwater flow. At 10)3 to 10)1 J m)3, enhanced permeability may be so effective to cause widespread changes in groundwater level in less sensitive wells, changes in geysers eruption frequency and increases in hot spring temperature. Continued increases in pore pressure and removal of grains act to degrade sediment stiffness. At a seismic energy density of 10)1 J m)3, some sensitive sediments may be so weakened to start consolidating. At higher seismic energy density, consolidation and pore-pressure increase may occur in less sensitive sediments. This energy density, though not large enough to induce sediment liquefaction, nonetheless may move the sediments towards a critical state so that they may become liquefied if an additional increment of pore pressure becomes available to push the sediment over the liquefaction limit. We suggest that this additional increment of pore pressure may become available during an earthquake-induced re-distribution of pore pressure from a nearby source, connected by an enhanced permeability, and thus initiates liquefaction in critically pressurized sediments. In such cases, a time delay between the earthquake and the occurrence of liquefaction may occur, as required by the diffusion of pore pressure from the source to the liquefaction site. In the near field where the seismic energy density exceeds 10 J m)3, undrained consolidation may raise pore pressure to the lithostatic limit, directly causing liquefaction. In this case, there should be little delay between the occurrence of liquefaction and earthquake. Earthquakeenhanced permeability must also occur in the near field, but its hydrologic effects in the near field may be obscured by those effects caused by undrained consolidation. An implication of the above discussion is that the evaluation of earthquake hazards at a specific site may be improved if, in addition to the evaluation of soil property and seismic intensity of the site in isolation, one should also consider the hydrologic heterogeneity in the surrounding area and the effect of pore pressure redistribution due to an earthquake-enhanced permeability that connects the site to nearby pore-pressure sources. Why do magmatic and mud eruptions have the same threshold distance? In Fig. 1 we included another example of a geofluid that responded to earthquakes: the triggered eruption of magmatic volcanoes. The examples included in Fig. 1 include only eruptions triggered within days (e.g., Linde & Sacks 1998). We do not include delayed triggered eruptions (e.g., Hill et al. 2002; Marzocchi 2002; Walter & Ame-
lung 2007) as these are less straightforward to establish as triggered events (Eggert & Walter 2009). The mechanism(s) responsible for magmatic eruption are more difficult to identify than the mechanisms for mud volcano eruption as there are a greater number of thermal and mechanical processes that operate in magmatic volcanoes. It is therefore interesting that, on a M versus log r diagram (Fig. 1), the documented triggered eruptions of the magmatic volcanoes show the same threshold distance as the triggered eruptions of mud volcanoes. It is enigmatic why the two types of eruptions, with great differences in many ways, would require the same threshold seismic energy to be triggered. Despite the obvious differences between these two types of eruptions, there are several processes and properties shared by both magmatic and mud volcanoes, including the buoyancy provided by exsolved gases, the fluidization or liquefaction of erupted materials, and an overpressured source (Manga et al. 2009). Could a permeability change, a process proposed for the occurrence of liquefaction and mud volcanism at the threshold distance, also play a role for the occurrence of magmatic eruption at the threshold distance? Magma chambers can become heterogeneous and stratified, as new batches of magma are injected in the magma chamber, and existing magmas differentiate and ⁄ or solidify. The volume change upon solidification, about 10% for typical magmas, will lead to significant changes in pressure in the remaining melt. Owing to the much larger viscosity of silicate melts compared to water, the hydraulic diffusion time needed for pore pressure to re-equilibrate will be correspondingly longer. If the interconnected network of crystals in the solidifying magma – the so-called ‘mush’ – is disrupted or broken by the earthquake, the much larger permeability afforded by fractures will allow for pore pressure to be redistributed much more rapidly. Deeper and pressurized sources may become connected to shallower, critically pressurized regions via such earthquake-enhanced permeability. Re-distribution of pressure through the diffusion of the gas or liquid phase may occur and may push one part of the chamber beyond a critical state leading to eruption. This mechanism may explain why the eruptions of triggered magmatic volcanoes and mud volcanoes require the same seismic energy density to be triggered – a similar energy may be required to disrupt a magmatic mush and a dense sediment. These ideas, of course, are highly speculative and may be difficult to test in the field. At the laboratory scale, however, the energy density required to disrupt a range of materials can be measured. Sumita & Manga (2008) showed that the viscosity of the liquid phase in dense, packed suspensions does not affect the threshold strain amplitudes needed to change pore pressure (Dobry et al. 1982) or rheology.
Hydrologic responses to earthquakes and a general metric 213
DISCUSSION, CONCLUSIONS, AND OUTLOOK The hydrologic responses we compile occurred mostly in the shallow subsurface (<1 km depth), except for mud volcanic eruptions and triggered earthquakes which occurred at depths of many km. The low seismic energy density required at the threshold distances for these responses means that the dynamic permeability enhancement would occur under very low effective stress, that in turn means that they either occur close to the earth’s surface or in highly overpressured portions of the crust. Despite the advances made thus far, much remains to be explored in the hydrologic responses to earthquakes. Two poorly explored aspects include the effect of faulting style and fault rupture directivity. Since the distribution of seismic energy in the near field depends on the style of faulting, the directivity of rupture, and the distance to the ruptured fault, we may expect such dependency to show up in the distribution of the hydrologic responses. Thus the empirical relation among the seismic energy density, earthquake magnitude and hypocentral distance shown in this paper (also Wang 2007) can only be regarded as a first-order approximation. The spatial pattern of changes in streamflow (e.g., Muir-Wood & King 1993) and changes in water level in wells (e.g., Wakita 1975; Jonsson et al. 2003) have been documented (for some cases) to vary with the pattern of static coseismic strain. Part of the reason for our present emphasis on magnitude and distance is that these two parameters are the most easily available, especially for historical earthquakes. Given the recent advances in strong-motion seismology and hydrologic monitoring of groundwater in earthquake prone regions, it should be straightforward to study the dependence of hydrologic responses on the directivity of rupture and on the style of faulting. The result of such efforts can provide important constraints on models on the hydrologic responses in the near- and intermediate-field. The triggering mechanism for the various hydrologic responses remains intriguing. Interpretations need to be quantitative so they can be tested against experimental measurement. The hypothesis that seismic waves can enhance the permeability of the upper crust has long been proposed (e.g., Rojstaczer et al. 1995; Roeloffs 1998; Brodsky et al. 2003; Elkhoury et al. 2006; Wang & Chia 2008). However, the mechanism that causes such an enhancement, i.e., how a transient wave may cause a sustained increase in permeability, remains enigmatic. More laboratory experiments (such as those in Elkhoury et al. 2009; Liu & Manga 2009) are needed to isolate different factors and to identify the responsible mechanism or mechanisms. Theoretical studies are needed to explain the experimental results. Finally, field studies are needed to
verify or to reject the various hypotheses as explanations for the natural phenomena. A surprising finding is that some earthquake-induced water-level changes may be caused by S-waves and Love waves (e.g., Wang et al. 2009). Such changes are inconsistent with the current understanding that in the far field only Rayleigh waves that involve changes in volumetric strain can cause water-level changes (Cooper et al. 1965; Liu et al. 1989; Manga & Wang 2007). It is, however, expected that S-waves and Love waves generate volumetric strains in an anisotropic poroelastic medium (Wang 2000; Brodsky et al. 2003), but quantitative tests are required to demonstrate the validity of the hypothesis. One such test is to deploy a network of broadband stations near the epicenter of an earthquake and to determine the three-dimensional strain tensors at the wells, from which the volumetric strain, if any, can be associated with S- or Love waves. The frequency dependence of liquefaction (e.g., Wong & Wang 2007) and triggered seismicity (Brodsky & Prejean 2005) may provide insight into their mechanisms. However, the results from field and laboratory studies thus far are in conflict. On the one hand, laboratory results show little frequency-dependence of deformation under cyclic shear strain (Yoshimi & Oh-Oka 1975; Sumita & Manga 2008); on the other hand, in situ evidence from seismically instrumented sites show an association of liquefaction and triggered seismicity with low-frequency ground motions (e.g., Brodsky & Prejean 2005; Youd & Carter 2005; Holzer & Youd 2007; Wong &Wang 2007). Future research, including in situ, laboratory, and theoretical work, is required to explain these differences. Another deficiency in the current study of hydrologic response is a complete absence of the ‘site effect’ which is known to have a strong influence on the distribution of seismic energy. Attempts to make predictions regarding liquefaction or other hydrologic responses at a particular site require the incorporation of site-specific geologic data. Such information would be needed in a more detailed study of hydrologic responses than those presented in the simplistic compilation shown in Fig. 1. While it should go without saying that more observations are useful, there is definitely a need for long-term and integrated hydrogeochemical, hydrological, temperature, and deformation measurements. Limited sampling and short time series often limit the ability to test hypotheses and reliably identify hydrologic responses and precursors.
ACKNOWLEDGEMENTS MM is supported by the Miller Institute for Basic Research in Science, and NSF grant EAR 0909701. We thank the reviewers for suggestions and comments.
214 C.-Y. WANG & M. MANGA
REFERENCES Aki K, Richards PG (1980) Quantitative Seismology. Freeman and Company, New York . Ambraseys NN (1988) Engineering seismology. Earthquake Engineering Structural Dynamics, 17, 1–105. Bai R, Tien C (1997) Particle detachment in deep bed filtration. Journal of Colloid and Interface Science, 186, 307–17. Bardet JP, Kapuskar M (1993) Liquefaction sand boil in San Francisco during 1989 Loma Prieta earthquake. Journal of Geotechnical Engineering, 119, 543–62. Beresnev IA, Johnson PA (1994) Elastic-wave stimulation of oil production: a review of methods and results. Geophysics, 59, 1000–17. Bergendahl J, Grasso D (2000) Prediction of colloid detachment in a model porous media: hydrodynamics. Chemical Engineering Science, 55, 1523–32. Bonini M (2009) Mud volcano eruptions and earthquakes in the Northern Apennines and Sicily, Italy. Tectonophysics, 474, 635– 723. Briggs RO (1991) Effects of Loma Prieta earthquake on surface waters in Waddell Valley. Water Resources Bulletin, 27, 991–9. Briggs RC, Troxell HC (1955) Effects of the Arvin-Tehachapi earthquake on spring and stream flow. In: Earthquakes in Kern County (ed. Oakeshott GB). Divisions of Mines Geological Bulletin, California, 171, 81–97. Brodsky EE, Prejean SG (2005) New constraints on mechanisms of remotely triggered seismicity at Long Valley Caldera. Journal of Geophysical Research, 110, B04302. doi.:10.1029/ 2004JB003211. Brodsky EE, Roeloffs E, Woodcock D, Gall I, Manga M (2003) A mechanism for sustained groundwater pressure changes induced by distant earthquakes. Journal of Geophysical Research, 108, 2390. doi.: 10.1029/2002JB002321. Carrigan CR, King GCP, Barr GE, Bixler NE (1991) Potential for water-table excursions induced by seismic events at Yucca Mountain, Nevada. Geology, 19, 1157–60. Chen JS, Wang C-Y (2009) Rising springs along the Silk Road. Geology, 37, 243–46. Chigira M, Tanaka K (1997) Structural features and the history of mud volcanoes in Southern Hokkaido, Northern Japan. Journal of the Geological Society of Japan, 103, 781–91. Cleasby JL, Williamson MM, Baumann ER (1963) Effect of filtration rate changes on quality. Journal of the American Water Works Association, 55, 869–77. Cooper HH, Bredhoeft JD, Papdopulos IS, Bennnett RR (1965) The response of aquifer- well systems to seismic waves. Journal of Geophysical Research, 70, 3915–26. D’Alessandro W, Parello F, Velenza M (1996) Gas manifestations on Mount Etna area: historical notices and new geochemical data (1990–1993). Acta Vulcanologia, 8, 23–9. Davies RJ, Brumm M, Manga M, Rubiandini R, Swarbrick R, Tingay M (2008) The East Java mud volcano (2006–present): an earthquake or drilling trigger? Earth and Planetary Science Letters, 272, 627–38. Davies EE, Wang K, Thomson RE, Becker K, Cassidy JF (2001) An episode of seafloor spreading and associated plate deformation inferred from crustal fluid pressure transients. Journal of Geophysical Research, 106, 21953–63. De Brignoli di Brunhoff G (1836) Relazione accademica dell’ultima eruzione accaduta nel vulcanetto aereo cosi detta Salsa di Sassuolo nel modeenese e considerazioni geognostiche intorno alle salse e alle loro cause. Tipografia Torreggiani et Compagno, Reggio, 66 pp.
Delisle G (2005) Mud volcanoes of Pakistan. In: Mud Volcanoes, Geodynamics and Seismicity (eds Martinelli G, Panahi B), pp. 159–69. NATO Science Series, Earth and Environmental Sciences, Vol. 51, Springer, Dordrecht, Netherlands. Deville E (2008) Mud volcano systems. In: Volcanoes: Formation, Eruptions and Modeling (eds Lewis N, Moretti A), pp. 95–125. Nova publishers, New York. Dobry R, Ladd RS, Yokel FY, Chung RM, Powell D (1982) Prediction of pore water pressure buildup and liquefactionof sands during earthquakes by the cyclic strain method. NBS Building Science Series, U.S., 138, pp. 150. Eggert S, Walter TR (2009) Volcanic activity before and after large tectonic earthquakes: observations and statistical significance. Tectonophysics, 471, 14–26. Elkhoury JE, Brodsky EE, Agnew DC (2006) Seismic waves increase permeability. Nature, 411, 1135–8. Elkhoury JE, Niemeijer A, Brodsky EE, Marone C (2009) Dynamic stress stimulates flow in fractures. Journal of Geophysical Research, (submitted). Galli P (2000) New empirical relationships between magnitude and distance for liquefaction. Tectonophysics, 324, 169–87. Gao B, Saiers JE, Ryan JN (2004) Deposition and mobilization of clay colloids in unsaturated porous media. Water Resources Research, 40, W08602. doi:10.1029/2004WR003189. Green RA, Mitchell JK (2004) Energy-based evaluation and remediation of liquefiable soils. In: Geotechnical Engineering for Transportation Projects (eds Yegian M, Kavazanjian E), ASCE Geotechnical Special Publication, No. 126, 2, 1961–70. Guidoboni E (1989) I terremoti prima del Mille in Italia e nell’area mediterranea: Storia, archeologia, sismologia. Instituto Nazionale di Geofisica-Storia Geofisica Ambiente, Bologna, 766 pp. Hill DP (2008) Dynamic stresses, Coulomb failure, and remote triggering. Bulletin of the Seismological Society of America, 94, 66–92. Hill DP, Prejean SG (2007) Dynamic Triggering in Treatise of Geophysics, Vol. 4, pp. 293–320. Elsevier, Amsterdam. Hill DP, Pollitz F, Newhall C (2002) Earthquake-volcano interactions. Physics Today, 55, 41–7. Holzer TL, Youd TL (2007) Liquefaction, ground oscillation, and soil deformation at the Wildlife Array, California. Bulletin of the Seismological Society of America, 97, 961–76. Husen S, Taylor R, Smith RB, Healser H (2004) Changes in geyser eruption behavior and remotely triggered seismicity in Yellowstone National Park produced by the 2002 M7.9 Denali Fault earthquake, Alaska. Geology, 32, 537–40. Ingebritsen SE, Rojstaczer SA (1993) Controls on geyser periodicity. Science, 262, 889–92. Ishihara K (1996) Soil Behavior in Earthquake Geotechnics. Clarendon Press, Oxford, 350 pp. Jonsson S, Segall P, Pedersen R, Bjornsson G (2003) Post-earthquake ground movements correlated to pore-pressure transients. Nature, 424, 179–83. Kayen R, Thompson E, Minasiah D, Collins B, Moss ERS, Sitar N, Carver G (2002) Geotechnical reconnaissance of the November 3, 2002 M7.9 Denali fault earthquake, Earthquake Spectra, Special Issue on the M7.9 Denali Fault Earthquake of 3 November 2002, 1–27. Kilb D, Gomberg J, Bodin P (2002) Aftershock triggering by complete Coulomb stress changes. Journal of Geophysical Research, 107, 2060. doi: 10.1029/2001JB000202. King C-Y, Basler D, Presser TS, Evans CW, White LD, Minissale AD (1994) In search of earthquake-related hydrologic and chemical changes along the Hayward fault. Applied Geochemistry, 9, 83–91.
Hydrologic responses to earthquakes and a general metric 215 King C-Y, Azuma S, Igarashi G, Ohno M, Saito H, Wakita H (1999) Earthquake-related water-level changes at 16 closely clustered wells in Tono, central Japan. Journal of Geophysical Research, 104, 13,073–82. Kitagawa Y, Koizumi N, Takahashi M, Matsumoyo N, Sato T (2006) Changes in water levels or pressures associated with the 2004 earthquake off the west coast of northern Sumatra (M9.0). Earth Planets Space, 58, 173–79. Koizumi N, Lai W-C, Kitagawa Y, Matsumoto Y (2004) Comment on ‘‘Coseismic hydrological changes associated with dislocation of the September 21, 1999 Chichi earthquake, Taiwan’’ by Min Lee et al.. Geophysical Research Letters, 31, L13603. doi:10.1029/2004GL019897. La Via PD (1828) Sur une eruption fangeuse d’un volcan hydroargileux (Salse) de la Sicile. Bulletin des Sciences Naturelles et de Geologie, 13, 33–35. Lawson AC (1908) The California Earthquake of April 18, 1906, Report of the State Earthquake Investigation Commission, Vol. 1, Carnegie Institution of Washington, Washington DC. Lay T, Wallace TC (1995) Modern Global Seismology. Academic Press, San Diego. Lemarchand N, Grasso J-R (2007) Interactions between earthquakes and volcano activity. Geophysical Research Letters, 34, L24303. Linde A, Sacks IS (1998) Triggering of volcanic eruptions. Nature, 395, 888–90. Liu W, Manga M (2009) Changes in permeability caused by dynamic stresses in fractured sandstone. Geophysical Research Letters, 36, L20307. doi:10.1029/2009GL039852. Liu L-B, Roeloffs E, Zheng X-Y (1989) Seismically induced water level fluctuations in the Wali well, Beijing, China. Journal of Geophysical Research, 94, 9453–62. Manga M (2007) Did an earthquake trigger the May 2006 eruption of the Lusi mud volcano? EOS, 88, 201. Manga M, Brodsky E (2006) Seismic triggering of eruptions in the far field: volcanoes and geysers. Annual Reviews of Earth Planetary Science, 34, 263–91. Manga M, Rowland JC (2009) Response of Alum Rock springs to the October 30, 2007 earthquake and implications for the origin of increased discharge after earthquakes. Geofluids, 9, 237– 50. Manga M, Wang C-Y (2007) Earthquake hydrology. In: Earthquake Seismology (Ed. Kanamori H), Vol. 4 of Treatise on Geophysics. Elsevier, Amsterdam, Ch. 4.10, 293–320. Manga M, Brodsky EE, Boone M (2003) Response of streamflow to multiple earthquakes and implications for the origin of postseismic discharge changes. Geophysical Research Letters, 30, 1214. doi:10.1029/2002GL016618. Manga M, Brumm M, Rudolph ML (2009) Earthquake triggering of mud volcanoes. Marine and Petroleum Geology, 26, 1785–98. doi:10.1016/j.marpetgeo.2009.01.019. Martinelli G, Bassignani A, Ferrari G, Finazzi PB (1989) Predicting earthquakes in Northern Appenines: recent developments in monitoring of Radon 222, Proceedings of the 4th International Symposium on the Analysis of Seismicity and Seismic Risk, Bechyne Castle, Czechoslovakia, 4–9 September 1989, 192–208. Marzocchi W (2002) Remote seismic influence on large explosive eruptions. Journal of Geophysical Research, 107, 2018. doi:10.1029/2001JB000307. Matsumoto N, Kitagawa G, Roeloffs EA (2003) Hydrological response to earthquakes in the Haibara well, central Japan – I. Water level changes revealed using state space decomposition of atmospheric pressure, rainfall and tidal responses. Geophysical Journal International, 155, 885–98.
Mazzini A, Svensen H, Akhmanov GG, Aloisi G, Planke S, Malthe-Sorenssen A, Istadi B (2007) Triggering and dynamic evolution of the LUSI mud volcano, Indonesia. Earth and Planetary Science Letters, 261, 375–88. McPherson RC, Dengler LA (1992) The Honeydew earthquake, August 12, 1991. California Geology, 31–9. Mellors R, Kilb D, Aliyev A, Gasanov A, Yetirmishli G (2007) Correlations between earthquakes and large mud volcano eruptions. Journal of Geophysical Research, 112, B04304. Mogi K, Mochizuki H, Kurokawa Y (1989) Temperature changes in an artesian spring at Usami in the Izu Peninsula (Japan) and their relation to earthquakes. Tectonophysics, 159, 95–108. Montgomery DR, Manga M (2003) Streamflow and water well responses to earthquakes. Science, 300, 2047–49. Montgomery DR, Greenberg HM, Smith DT (2003) Streamflow response to the Nisqually earthquake. Earth Planetary Science Letters, 209, 19–28. Muir-Wood R, King GCP (1993) Hydrological signatures of earthquakes strain. Journal of Geophysical Research, 98, 22. National Research Council (1985) Liquefaction of Soils during Earthquakes. Washington, D.C., 240 pp. Okubo PG, Wolfe CJ (2008) Swarms of similar long-period earthquakes in the mantle beneath Mauna Loa Volcano. Journal of Volcanology and Geothermal Research, 178, 787–94. Pierepiekarz MR, Ballantyne DB, Hamberger RO (2001) Damage report from Seattle. Civil Engineering, American Society of Civil Engineering, 71, 78–83. Prior JC, Boekhoff JL, Howes MR, Libra RD, VanDorpe PE (2003) Iowa’s Groundwater Basics. Iowa Department of Natural Resources, Iowa City, 92 pp. Quilty EG, Roeloffs EA (1997) Water-level changes in response to the 20 December 1994 earthquake near Parkfield, California. Bulletin of the Seismological Society of America, 87, 310–7. Quilty EG, Farrar CD, Galloway DL, Hamlin SN, Laczniak RJ, Roeloffs EA, Sorey ML, Woodcock DE (1995) Hydrologic effects associated with the January 14, 1994 Northridge, California, earthquake, USGS Open File Report 95–813. Rajendran K, Rajendran CP, Thakkar M, Tuttle MP (2001) The 2001 Kutch (Bhuj) earthquake: coseismic surface features and their significance. Current Science, 80, 1397–405. Roeloffs EA (1998) Persistent water level changes in a well near Parkfield, California, due to localand distant earthquakes. Journal of Geophysical Research, 103, 869–89. Roeloffs E, Danskin WR, Farrarr CD, Galloway DL, Hamlin SN, Quilty EG, Quinn HM, Schaefer DH, Sorey ML, Woodcock DE (1995) Hydrologic effects of the June 28, 1992 Landers earthquake. U.S. Geological Survey Open File Report 95-42. Roeloffs EA, Sneed M, Galloway DL, Sorey ML, Farrar CD, Howle JF, Hughes J (2003) Water-level changes induced by local and distant earthquakes at Long Valley caldera, California. Journal of Volcanology and Geothermal Research, 127, 269–303. Rojstaczer S, Wolf S, Michel R (1995) Permeability enhancement in the shallow crust as a cause of earthquake-induced hydrological changes. Nature, 373, 237–9. Rothaus RM, Reinhardt E, Noller J (2004) Regional consideration of coastline change, tsunami damage and recovery along the southern coast of the Bay of Izmit. Natural Hazards, 31, 233– 52. Rowland JC, Manga M, Rose TP (2008) The influence of poorly interconnected fault zone flow paths on spring geochemistry. Geofluids, 8, 93–101. Rukavickova L, Hanzl P (2008) Mud volcanoes in the Khar Angalantyn Nuruu, NW Gobi Altay, Mongolia as manifestation of recent seismic activity. Journal of Geosciences, 53, 181–91.
216 C.-Y. WANG & M. MANGA Saiers JE, Lenhart JJ (2003) Colloid mobilization and transport within unsaturated porous media under transient-flow conditions. Water Resources Research, 39, 1019. doi:10.1029/ 2002WR001370. Sato T, Sakai R, Furuya K, Kodama T (2000) Coseismic spring flow changes associated with the 1995 Kobe earthquake. Geophysical Research Letters, 27, 1219–22. Sato T, Matsumoto N, Kitagawa Y, Koizumi N, Takahashi M, Kuwahara Y, Ito H, Cho A, Satoh T, Ozawa K, Tasaka S (2004) Changes in water level associated with the 2003 Tokachi-oki earthquake. Earth Planets Space, 56, 395–400. Seed HB, Lee KL (1966) Liquefaction of saturated sands during cyclic loading. Journal of Soil Mechanics, Foundation Division, 92, 105–34. Sil S, Freymueller JT (2006) Well water level changes in Fairbanks, Alaska, due to the great Sumatra-Andaman earthquake. Earth Planets Space, 58, 181–4. Silver PG, Wakita H (1996) A search for earthquake precursors. Science, 271, 77–8. Snead RE (1964) Mud volcanoes of Baluchistann, West Pakistan. The Geographical Review, 54, 546–60. Sumita I, Manga M (2008) Suspension rheology under oscillatory shear and its geophysical implications. Earth and Planetary Science Letters, 269, 467–76. Trabucco G (1895) Terremoto della Romagna Toscana del 4 Settembre 1895. Bollettino Societa Geologica Italiana, 14, 284–6. Velasco AA, Hernandez S, Parsons T, Pankow K (2008) Global ubiquity of dynamic earthquake triggering. Nature Geoscience, 1, 375–9. Vorhis RC (1968) Effects outside Alaska, in The Great Alaska of 1964, Hydrology. National Academy of Science Publication 1603, Washington DC, 140–89. Wakita H (1975) Water wells as possible indicators of tectonic strain. Science, 189, 553–5. Waller R (1966) Effects of the March 1964 Alaska earthquake on the hydrology of the Anchorage area, US Geological Survey Professional Paper 544B. Walter TR, Amelung F (2007) Volcanic eruptions following M > 9 megathrust earthquakes: Implications for the SumatraAndaman volcanoes. Geology, 35, 539–42. Wang H (2000) Theory of Linear Poroelasticity with Applications to Geomechanics. Princeton University Press, Princeton, 287 pp. Wang C-Y (2007) Liquefaction beyond the near field. Seismological Research Letters, 78, 512–7.
Wang C-Y, Chia Y (2008) Mechanism of water level changes during earthquakes: Near field versus intermediate field. Geophysical Research Letters, 35, L12402. doi:10.1029/2008GL034227. Wang C-Y, Manga M (2010) Earthquakes and Water, Vol. 114. Lecture Notes in Earth Sciences, Springer, Berlin. 249 pp. ISBN: 978-3-642-00809-2 Wang C-Y, Cheng L-H, Chin C-V, Yu S-B (2001) Coseismic hydrologic response of an alluvial fan to the 1999 Chi-Chi earthquake, Taiwan. Geology, 29, 831–4. Wang C-Y, Manga M, Dreger D, Wong A (2004a) Streamflow increases due to the rupturing of hydrothermal reservoir – evidence from the 2003 San Simeon, California, Earthquake. Geophysical Research Letters, 31, L10502. doi:10.1029/ 2004GL020124. Wang C-Y, Wang C-H, Manga M (2004b) Coseismic release of water from mountains: Evidence from the 1999 (Mw = 7.5) Chi-Chi, Taiwan, earthquake. Geology, 32, 769–72. Wang C-Y, Wong A, Dreger DS, Manga M (2006) Liquefaction limit during earthquakes and underground explosions: Implications on ground-motion attenuation. Bulletin of the Seismological Society of America, 96, 355–63. Wang C-Y, Chia Y, Wang PI, Dreger D (2009) Role of S waves and Love waves in coseismic permeability enhancement. Geophysical Research Letters, 36, L09404. doi:10.1029/2009GL 037929. West M, Sanchez JJ, McNutt SR (2005) Periodically triggered seismicity at Mount Wrangell, Alaska, after the Sumatra earthquake. Science, 308, 1144–6. Wong A, Wang C-Y (2007) Field relations between the spectral composition of ground motion and hydrological effects during the 1999 Chi-Chi (Taiwan) earthquake. Journal of Geophysical Research, 112, B10305. Woodcock D, Roeloffs E (1996) Seismically-induced water level oscillations in a fractured-rock aquifer well near Grants Pass, Oregon. Oregon Geology, 58, 27–33. Yoshimi Y, Oh-Oka H (1975) Influence of degress of shear stress reversal on the liquefaction potential of saturated sand.. Soils and Foundations (Japan), 15, 27–40. Youd TL, Carter BL (2005) Influence of soil softening and liquefaction on spectral accelerations. Journal of Geotechnical and Geoenvironmental Engineering, 131, 811–25. Yu M-S, Shieh B-C, Chung YT (2000) Liquefaction induced by Chi-Chi earthquake on reclaimed land in central Taiwan. SinoGeotech, 77, 39–50 (in Chinese).
The application of failure mode diagrams for exploring the roles of fluid pressure and stress states in controlling styles of fracture-controlled permeability enhancement in faults and shear zones S. F. COX Research School of Earth Sciences, The Australian National University, Canberra, ACT, Australia
ABSTRACT Permeability enhancement associated with deformation processes in faults and shear zones plays a key role in facilitating fluid redistribution between fluid reservoirs in the crust. Especially in high fluid flux hydrothermal systems, fracture-controlled permeability can be relatively short-lived, unless it is repeatedly regenerated by ongoing deformation. Failure mode diagrams in pore fluid factor and differential stress space, here termed k–r failure mode diagrams, provide a powerful tool for analysing how fluid pressure and stress states drive failure, associated permeability enhancement and vein styles during deformation in faults and shear zones. During fault-valve behaviour in the seismogenic regime, relative rates of recovery of pore fluid factor, differential stress and fault cohesive strength between rupture events impact on styles of veining and associated, fracture-controlled permeability enhancement in faults and shear zones. Examples of vein-rich fault zones are used to illustrate how constraints can be placed, not just on fluid pressure and stress states at failure, but also on the fluid pressurization and loading paths associated with failure and transitory permeability enhancement in faults and shear zones. This provides insights about when, during the fault-valve cycle, various types of veins can form. The use of failure mode diagrams also provides insights about the relative roles of optimally oriented faults and misoriented faults as hydraulically conductive structures. The analysis highlights the dynamics of competition between fluid pressures and loading rates in driving failure and repeated permeability regeneration in fracture-controlled, hydrothermal systems. Key words: fracture permeability, faults, veins, stress states, fluid pressures Received 23 October 2009; accepted 26 January 2010 Corresponding author: S. F. Cox, Research School of Earth Sciences, The Australian National University, Canberra, ACT 0200, Australia. E-mail:
[email protected]. Tel: 61-2-61250045. Fax: 61-2-6125 5544. Geofluids (2010) 10, 217–233
INTRODUCTION In the mid to upper crust, flow of hydrothermal fluids is commonly localized along fracture systems that are preserved as vein arrays. In mesothermal gold systems, for example, the high fluid fluxes that are associated with mineralization and attendant hydrothermal alteration are typically localized along arrays of low displacement faults, shear zones and associated fracture networks (e.g. Robert & Poulsen 2001; Cox & Ruming 2004). Similar structures are associated with high fluid flux in fault zones in some non-mineralized, regional deformation environments (Cox
2007). Permeability enhancement in fault zones is associated variously with the production of porous and permeable wear products (gouges, cataclasites and wear breccias) along fault slip surfaces, with the formation of arrays of extension veins adjacent to fault slip surfaces, by dilation and formation of fault-fill veins along parts of fault slip surfaces, and by the formation of dilational breccias adjacent to fault slip surfaces (Figs 1 and 2). Some dilational breccias are interpreted to have been produced by wall-rock implosion driven by co-seismic dilation along fault zones (Sibson 1986). In brittle-viscous transitional deformation regimes, permeability enhancement in shear zones can also be associated
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
218 S. F. COX
(A)
(C)
(B)
(D)
Fig. 1. Examples of vein-development in faults and shear zones. (A) Near-optimally oriented reverse fault with associated extension vein arrays. Wattle Gully gold mine, near Castlemaine, Victoria, Australia (see Cox 1995). Field of view is approximately 1-m wide. (B) Ductile shear zone with associated sigmoidal calcite extension veins. Columbia Icefields Highway, Alberta, Canada. Field of view is 2-m wide. (C) Misoriented, quartz-filled, dilatant reverse fault with associated extension veins. Misorientation angle is approximately 45. Wattle Gully gold mine, near Castlemaine, Victoria, Australia (see Cox 1995). Field of view is 2-m wide. (D) Extension vein arrays associated with normal fault. Wadi Nakhr, Jabal Shams, Oman Mountains (see Hilgers et al. 2006).
with the development of extensional vein arrays (Fig. 1B). But substantial permeability enhancement during viscous creep processes can also be associated with the generation of grain-scale microcrack networks, even where the bulk strain is accommodated largely by intracrystalline plasticity (Zhang et al. 1995). Permeability enhancement in these regimes is especially important in high pore fluid factor environments. Pore fluid factor kv is the ratio of fluid pressure Pf to vertical stress rv (Hubbert & Rubey 1959). Especially in fluid-active regimes at temperatures in excess of approximately 200–300C, compactional processes, cementation of intergranular pores, and healing and sealing of fractures to produce veins can destroy permeability on timescales that are short relative to the lifetimes of hydrothermal systems (e.g. Cox 2005). Accordingly, sustained fluid flow at depth in the crust occurs only if permeability is
repeatedly regenerated. Especially in the seismogenic mid to upper crust, repeated episodes of brittle failure are a key permeability regeneration process. Cyclic generation and destruction of fracture-controlled permeability during the seismic cycle can result in fluid flow being episodic. In the upper crustal brittle regime, rock failure is controlled by the evolution of stress states and fluid pressure states. Tectonic loading usually leads to progressive increase in shear stress on faults with time (Scholz 1990), although sudden, small stress changes (both increases and decreases) may occur on faults in response to slip on nearby faults (Stein 1999). At the same time, fluid pressure changes in fault zones during the seismic cycle may be driven by a number of causes. These include (i) build-up of fluid pressure beneath low permeability segments of otherwise permeable faults or shear zones that are hydraulically linked to
Application of failure mode diagrams 219
(A)
(B)
(C)
(D)
Fig. 2. Damage structures associated with transitory permeability enhancement in fault zones. (A) Banded cataclasite formed by multiple episodes of fault slip and hydrothermal healing of cataclasite between slip events. Porgera gold deposit, Papua New Guinea. Field of view is 2-mm wide. (B) Dilational breccia adjacent to fault zone. Argo gold deposit, St Ives goldfield, Western Australia. Field of view is 2-m wide. (C) Zone of intense development of extension veins adjacent to a thrust fault. Conqueror lode system, St Ives goldfield, Western Australia (see Cox & Ruming 2004). Field of view is 3-m wide. (D) Banded, fault-fill vein and associated breccia in a normal fault zone, Cracow epithermal gold deposit, Queensland, Australia (see Micklethwaite 2009).
overpressured fluid reservoirs; (ii) rapid fluid flow and fluid pressure build-up in the downstream part of a ruptured fault segment that breaches an overpressured fluid reservoir; and (iii) build-up of pore fluid pressure within fault cores due to progressive interseismic porosity destruction and compaction in fault damage products (e.g. Blanpied et al. 1992). Slip events in fault zones relieve stored elastic strain via displacement and co-seismic stress drop. Slip events may also drive co-seismic permeability enhancement and dilatancy, and lead to sudden co-seismic fluid pressure drop. Subsequent interseismic fault sealing and tectonic loading then leads to the recovery of fluid pressure and shear stress. Such cyclic and coupled changes in stress and fluid pressure states, and associated transitory fluid flow events during seismic cycles, are referred to as fault-valve behaviour (Sibson 1981, 1992) and are especially important in influencing the dynamics of fluid flow and fluid–rock interaction in fault-hosted, hydrothermal ore systems (e.g. Cox 1995, 2005). Coupled changes in stress and fluid pressure states also can be significant in the nucleation of earthquakes (Miller et al. 1996; Sibson 2007).
The conditions for brittle failure and proximity of stress states to failure conditions are commonly analysed using Mohr diagrams, which plot failure conditions in shear stress and effective normal stress space (where effective normal stress, rn¢, is normal stress, rn, minus fluid pressure, Pf) (e.g. Secor 1965). Mohr diagrams provide an excellent way of visualizing changes in the maximum and minimum principal stresses (r1 and r3 respectively) and the consequent changes in shear stress (s) and effective normal stress on surfaces inclined at particular angles to r1 and r3. However, changes in fluid pressure are incorporated into changes in effective normal stress, so are not a separate variable that is readily visualized in Mohr space. Sibson (1998, 2000, 2001) introduced the concept of brittle failure mode plots to illustrate conditions required for failure in intact rock and incohesive faults as a function of effective vertical stress rv¢ and differential stress (r1–r3). Such failure mode diagrams are very useful for visualizing how parameters such as tensile strength and friction coefficients influence maximum sustainable fluid pressures at various depths and pore fluid factors. In a related approach,
220 S. F. COX Fournier (1996) analysed the effects of pore fluid factor on conditions for shear failure as a function of differential stress and misorientation angle hr between a fault and the orientation of the maximum principal stress. Similarly, Streit & Cox (2001) analysed failure conditions for cohesive faults in pore fluid factor–depth space and also as a function of hr. Cox et al. (2001) and Cox (2005) employed another type of failure mode diagram, plotted in pore fluid factor–differential stress space, to illustrate how changes in (r1–r3) and kv during fault-valve cycles influence failure modes in intact rock at a given depth. Sibson (2004, 2007) applied similar failure mode diagrams to illustrate fluid pressure conditions associated with gold mineralization and for the generation of an earthquake sequence. In this study, failure mode diagrams in pore fluid factor and differential stress space are used to explore how the evolution of pore fluid factors and differential stresses influences failure modes, and related styles of permeability enhancement and vein development during repeated cycles of fault-valve behaviour. The analysis is especially pertinent to understanding vein distribution and vein development in fracture-controlled hydrothermal systems in intrinsically impermeable rock. It will be shown that failure mode diagrams provide a powerful way to explore when, and under what fluid pressure and stress conditions during fault-valve cycles, particular types of veins may form. Failure mode diagrams in pore fluid factor and differential stress space are here termed k–r failure mode diagrams. The first part of the study outlines macroscopic brittle failure criteria and the construction of k–r failure mode diagrams for both intact rock and during the reactivation of pre-existing, cohesive faults. To illustrate how failure conditions are dependent on tectonic regimes, variations in the shapes of the failure envelope in pore factor–differential stress space are discussed for contractional, extensional and strike-slip faulting regimes. The effects of depth and two key material parameters, cohesive strength and friction coefficient, are also illustrated. As many fault-related hydrothermal ore systems form in faults that are not optimally oriented with respect to the ambient stress field during ore formation, the study next examines how the shape of the failure envelope, and hence the stress and fluid pressure conditions required for permeability regeneration, are sensitively dependent on the misorientation angle hr between the fault and the direction of the maximum principal stress. It will be demonstrated also that time-dependent recovery of cohesive strength due to hydrothermal fault sealing after rupture events is a key factor influencing whether or not extension vein arrays and the associated permeability enhancement can develop in fault zones, or whether permeability regeneration is controlled by shear failure on a pre-existing misoriented fault, or rupture on a new optimally oriented fault zone.
Although many of the results discussed in this study can be obtained from traditional Mohr circle constructions, the major advantage in using k–r failure mode diagrams for exploring stress and fluid pressure states necessary for failure and related permeability enhancement is that k–r failure mode diagrams much more simply illustrate the changes in pore fluid pressures (or pore fluid factor) during deformation than is the case for Mohr diagrams. Many relationships that are hard to recognize on Mohr constructions are clearly displayed on failure mode diagrams. Accordingly, k–r failure mode diagrams make it intuitively easier to understand the interplay of parameters such as fluid factor, stress states, fault misorientation and time-dependent evolution of cohesive strength in driving rock failure and fracturecontrolled permeability enhancement in hydrothermal regimes. Most importantly, k–r failure mode diagrams highlight the critical role, in many hydrothermal systems, of changes in fluid pressures, rather than just stress states, in driving failure and associated permeability enhancement in intrinsically impermeable rock masses at depth in the crust.
FAILURE CRITERIA IN PORE FLUID FACTOR–DIFFERENTIAL STRESS SPACE Macroscopic brittle failure occurs predominantly by either extension mode failure or shear failure (faulting). Extensional mode failure involves the generation of fractures perpendicular to the orientation of r3, with opening of the fracture parallel to r3. Macroscopic shear failure involves displacement parallel to the shear fracture surface. In some cases, failure may occur by hybrid extensional-shear failure, involving components of shear and dilation. The mode of brittle failure is dependent on the relative values of differential stress and tensile strength (Secor 1965). For this analysis, only faults having poles within the r1–r3 plane are considered. In this scenario, the value of the intermediate principal stress r2 does not influence failure conditions (Jaeger & Cook 1979). For simplicity, we also assume an isotropic rock mass with coefficient of internal friction l. The coefficient of static friction for a pre-existing fault is ls. Brittle shear failure conditions For this analysis, the linear Coulomb criterion for shear failure is assumed for compressive effective normal stress conditions in intact rock. Shear failure along a pre-existing fracture is also assumed to follow a Coulomb failure criterion. Other failure criteria could equally well be applied to the construction of k–r failure mode diagrams. The analysis could also be extended for situations where r2 does not lie within the fault plane.
Application of failure mode diagrams 221 Brittle shear failure occurs when shear stress s is given by s ¼ C þ lðrn Pf Þ;
kv ¼
½4C r1 þ 4r3 : 3rv
ð8Þ
ð1Þ
where C is the cohesive strength. For a fault inclined to the orientation of r1 at an angle hr, we have the relationships r þ r r r 1 3 1 3 cos 2hr rn ¼ ð2Þ 2 2
For the special case of reactivation of a fault in which the reactivation angle, hr, is equal to 2hopt, it can also be shown that 1 1 2 kv ¼ 2C þ 2 cos hr l þ r3 : 2lrv l
ð9Þ
and s¼
r r 1 3 sin 2hr : 2
ð3Þ
Substituting eqns 2 and 3 into 1, it can be shown that brittle shear failure occurs where Pf ¼
C r1 þ r3 r1 r3 cos 2hr þ 2 2 l r1 r3 sin 2hr : 2l
ð4Þ
For reverse faulting, if r3 rv and r3 remains constant in magnitude during loading to failure, the shear failure envelope for reverse faults inclined to r1 at 2hopt is a horizontal line in kv (r1–r3) space. However, for both normal and strike-slip faulting, r3 changes during tectonic loading and the shear failure envelope for the special case of hr = 2hopt has a negative slope in kv (r1–r3) space. For the particular case in which l = 0.75 and hr = 2hopt, the shear failure criterion in eqn 9 reduces to kv ¼
Re-casting eqn 4 in terms of pore fluid factor, we obtain 1 ½2C þ lðr1 þ r3 Þ lðr1 r3 Þ cos 2hr 2lrv ðr1 r3 Þ sin 2hr : ð5Þ
1 ½4C þ 3r3 : 3rv
ð10Þ
kv ¼
Accordingly, in kv–differential stress space, the Coulomb shear failure criterion is a straight line whose slope and position are governed by rv, C, hr and l (Fig. 3). The coefficient of internal friction in intact rock is typically in the range 0.5 < l < 1.0 (Jaeger & Cook 1979). Static friction on pre-existing fractures is usually in the range 0.6 < ls < 0.85 (Byerlee 1978). For the particular case of compressional shear failure of intact rock, or reactivation of a pre-existing, optimally oriented fault, the angle hopt between the shear fracture and r1 is given by 1 1 : ð6Þ hopt ¼ tan1 2 l
Extension failure conditions Following Secor (1965), it is assumed that extension failure occurs only for stress conditions in which (r1– r3) < 4T. Additionally, in compressive stress states at depth in the crust, pure extension fracture can only occur when the fluid pressure condition Pf ¼ r3 þ T
ð11Þ
is met. This criterion for hydraulic extension fracturing can be expressed in the form (Fig. 3)
kv ¼
r3 þ T : rv
ð12Þ
Conditions for hybrid extensional-shear failure For optimally oriented brittle shear failure, it can be shown that eqn 5 reduces to 1 cos 2hopt 1 r1 2C tan 2hopt cos 2hopt 2rv 1 þ cos 2hopt þ r3 : cos 2hopt
kv ¼
ð7Þ
For optimally oriented compressional shear failure, with l = 0.75, this shear failure criterion further reduces to
For negative effective normal stresses and rn¢ > )T, the conditions for hybrid extensional-shear failure in intact rock can be approximated by the Griffith failure criterion (Griffith 1924; Jaeger & Cook 1979) ðr1 r3 Þ2 ¼ 8Tðr01 þ r03 Þ;
ð13Þ
where r1¢ = (r1 ) Pf) and r3¢ = (r3 ) Pf). Hybrid extensional-shear fractures can form at angles from 0 to 26.6 for l = 0.75 (Secor 1965). Equation 13 can be re-cast in the form
222 S. F. COX
Extension failure Hybrid extensional shear failure
1
Pore fluid factor, λ v = Pf/σv
Lithostatic Pf
4T
0.4
0
2C/sin 2θopt
Shear failure
Hydrostatic Pf
Differential stress (σ1– σ3), MPa
Comparison of failure envelopes for intact rock in reverse, normal and strike-slip fault zones
Fig. 3. Generic failure mode diagram in pore fluid factor–differential stress space for a given depth in the crust. The shear failure envelope is shown in blue; the extension mode failure envelope is in red; the failure envelope for hydrid extensional-shear mode failure is shown in green. For all stress and pore factor states below the composite failure envelope, the rock mass is elastically strained, but will not fail in brittle mode. Brittle failure occurs when the stress and pore fluid factor states reach the failure envelope. Stress and fluid pressure states above the failure envelope cannot be attained.
Pf ¼
8T ðr1 þ r3 Þ ðr1 r3 Þ2 : 16T
l = 0.75, being approximately an average value for many crustal rocks. For intact rock failure, the simplification that C 2T is used, with tensile strength of 5 MPa. This value is comparable with cohesive strengths of moderately to weakly cohesive rock (Jaeger & Cook 1979). It is noted that most laboratory-determined values of tensile strength involve conditions of dynamic crack growth. Because failure associated with subcritical crack growth could be more important than dynamic crack growth in many fluid-active regimes (Etheridge 1983), cohesive and tensile strengths during extension veining may be substantially less than laboratory determined values.
ð14Þ
Figure 4 compares failure envelopes in k–r space, and hence conditions for generation of fracture permeability, for intact rock in reverse, normal and strike-slip faulting regimes. The extension failure envelope extends between 0 £ r1–r3 £ 4T. The slope of this part of the failure envelope is horizontal in reverse fault regimes, and is negative in strike-slip and normal fault regimes. In each tectonic regime, hybrid extensional-shear failure is modelled as a parabolic failure envelope in the interval 4T £ r1– r3 £ 4T ⁄ sin2hopt. Brittle shear failure occurs on a negative sloping, linear failure envelope at r1–r3 ‡ 4T ⁄ sin2hopt. A feature of k–r failure mode diagrams is that they illustrate clearly how much lower the differential stress is for failure
Accordingly, the extensional shear failure condition in kv (r1–r3) space is given by kv ¼
8T ðr1 þ r3 Þ ðr1 r3 Þ2 : 16T rv
ð15Þ
The hybrid extensional-shear part of the failure envelope in kv (r1–r3) space thus has a parabolic shape (Fig. 3). If the shear failure part of the failure envelope is linear for rn¢ > 0 (eqn 7), then it meets the hybrid extensional shear failure envelope (eqn 15) at ðr1 r3 Þ ¼
2C : sin 2hopt
Lithostatic Pf
1
0.8
Reverse
λ v 0.6
ð16Þ
The maximum differential stress for hybrid extensional shear failure is thus modelled as 2.5C for l = 0.75, and decreases to 2.01C at l = 0.1.
0.4
For simplicity in the following examples, it will be assumed that rv remains constant and is controlled by the depth of overburden. In most of the examples used here, the value of both internal friction and static friction is taken as
Strike-slip
z = 10 km 0.2
μ = 0.75 Normal
C = 10 MPa T = 5 MPa 0
EXAMPLES OF k–r FAILURE MODE DIAGRAMS
Hydrostatic Pf
50
100
150
200
250
300
σ1– σ3, MPa Fig. 4. Failure mode diagrams in kv (r1–r3) space for reverse, normal and strike-slip faulting regimes at a depth of 10 km; C = 10 MPa, T = 5 MPa, l = 0.75. For the strike-slip regime, it is assumed that r2 (r1 + r3) ⁄ 2. The extension failure, hybrid extensional shear failure and shear failure parts of the failure envelope are colour coded as in Fig. 3.
Application of failure mode diagrams 223 and associated permeability enhancement during normal faulting than for reverse faulting at any given fluid pressure. Similarly, they illustrate that higher pore fluid factors are required to drive failure and permeability enhancement in reverse faulting regimes, than in strike-slip and normal faulting regimes, at any given differential stress. Additionally, k–r failure mode diagrams illustrate how the failure envelope limits both the maximum sustainable fluid pressure at any given differential stress (see also Sibson 2004), and the maximum differential stress that can be attained prior to failure and permeability enhancement at any given pore fluid factor.
kv ¼ 1 þ
T T 1þ ; r3 qggz
Lithostatic Pf
1
15 km 10 km 0.8
5 km
λ v 0.6 2 km Hydrostatic Pf
0.4
μ = 0.75
0.2
Depth dependence of conditions for fracture-controlled permeability enhancement in k–r space Examples of the depth dependence of failure envelopes for intact rock failure in reverse and normal faulting regimes are illustrated in Fig. 5. With increasing depth in reverse fault zones, although the absolute pore fluid pressures required to drive extension failure increase in magnitude, the required pore fluid factors decrease with increasing depth (Fig. 5A). The depth dependence of the maximum sustainable pore fluid factor at r1–r3 £ 4T is given by the hydrofracture criterion,
(σ1– σ3) = 4C
(A)
C = 10 MPa T = 5 MPa 0 0
(B)
50
σ1 – σ3, MPa
(σ1 – σ3) = T Lithostatic Pf
1
0.8
15 km
ð17Þ
where q is the rock density, g is the gravitational acceleration and z is the depth. Note also that supralithostatic fluid pressures are required for shear failure at a differential stresses <4C. Additionally, for optimally oriented reverse faults at depths >5 km, fluid pressures need to be substantially above hydrostatic for failure and permeability enhancement to occur at differential stresses <200 MPa. The slope of the brittle shear failure part of the failure envelope becomes increasingly negative in k–r space with decreasing depth. Accordingly, at a given differential stress, the pore fluid factor at the shear failure condition increases with increasing depth. Similarly, at constant pore fluid factor, the differential stress required for permeability enhancement by shear failure increases with increasing depth. Similar relationships hold for normal faulting regimes (Fig. 5B). In this case, the opening of pure extension fractures requires supralithostatic conditions only at r1– r3 < T. At higher differential stresses, extension failure occurs at sublithostatic fluid pressure conditions. But, for depths greater than approximately 2 km in normal fault zones, the formation of extension fractures and associated permeability enhancement still requires overpressured (suprahydrostatic) fluid conditions (i.e. kv > 0.4), for the modelled rock properties. The conditions for failure and permeability enhancement at shallow crustal levels typical for fault-related epithermal
150
100
λ v 0.6
10 km Hydrostatic Pf
0.4
5 km μ = 0.75
0.2
C = 10 MPa T = 5 MPa
2 km 0
0
4T
50
σ1 – σ3, MPa
100
150
Fig. 5. Effects of changes in depth on failure envelopes in k–r failure mode diagrams for intact rock failure in (A) reverse and (B) normal fault regimes. Failure envelopes are given for depths of 2, 5, 10 and 15km, for l = 0.75.
Au vein systems are particularly interesting. These systems tend to form in normal fault zones in near-hydrostatic fluid pressure regimes, and at depths <1–2 km (Simmons et al. 2005). Figure 6A illustrates a failure mode diagram for optimally oriented, normal faulting at a depth of 1 km. For the modelled tensile and cohesive strengths, and for l = 0.75, it is apparent that failure can only occur by extension or hybrid extensional shear modes at this depth if fluid pressures are hydrostatic or larger. Shear failure can only occur on steeply dipping optimally oriented faults at greater depth in the hydrostatic fluid pressure regime. Accordingly, with decreasing depth normal faults are constrained to steepen into the
224 S. F. COX
(A) 1.2
(A) Lithostatic Pf
1 Lithostatic Pf
1
0.85 0.6
0.8
0.8
0.4
Extension failure
λ v 0.6
λ v 0.6
Hydrostatic Pf
0.4
Hydrostatic Pf
0.4
0.2 Hybrid extensional shear failure
0.2
0
μ = 0.75 C = 10 MPa T = 5 MPa 0
10
μ = 0.1
C = 10 MPa z =10 km
Shear failure 20
30
σ1 – σ3, MPa
40
0
50
σ1
(B)
Contractional regime
0.2
0
50
σ1 – σ3, MPa
150
100
(B) 1.1
0 km 1
Lithostatic Pf
C
λv
0.9
=0
MP a
5M
1 km
σ3
0.6 0
Effects of varying friction coefficient and tensile strength on conditions for permeability enhancement during intact rock failure The effects of variations in the coefficient of internal friction on the failure envelope for intact rock at constant depth are illustrated in Fig. 7A. The k–r failure mode diagrams illustrate that the slope of the brittle shear failure
20
a
40 MP a
MP a
Contractional regime z =10 km μ = 0.75
2 km
orientation of subvertical extension mode fractures (Fig. 6B). This relationship is consistent with the observation that many epithermal veins have limited fault-parallel slip, but very substantial dilation (Fig. 2D).
MP
0.8
0.7
Fig. 6. (A) Failure mode diagram for normal faulting regime, typical of epithermal mineralizing environments at a depth of one kilometre; C = 10 MPa, T = 5 MPa, l = 0.75. (B) Schematic illustration of the change in geometry of a normal fault zone with changing depth in the crust.
Pa 10
50
100
150
σ1 – σ3, MPa
200
250
300
Fig. 7. Effects of changes in (A) friction coefficient, and (B) cohesive strength on failure envelopes for reverse faulting of intact rock at 10-km depth.
envelope becomes less negative with the increasing coefficient of friction. Therefore, at constant kv, the stress required for shear failure and associated permeability enhancement increases with increasing friction coefficient. The hydraulic fracture criterion is independent of the friction coefficient in k–r space. The effects of changes in cohesive strength of intact rock (using the relationship C = 2T) on conditions for permeability regeneration are illustrated in Fig. 7B. With increasing cohesive strength, the brittle shear failure envelope moves to higher pore fluid factor and higher differential stress. Similarly, the extension failure envelope moves to higher pore fluid factors and there is an expansion of the
Application of failure mode diagrams 225 field in which permeability enhancement can occur by extension failure and hybrid extensional-shear failure.
(A) 1.15
Reverse faulting regime 1.1
Failure conditions for re-shear and permeability enhancement in pre-existing incohesive fault zones
1.05
As fault zones grow and accumulate displacement over hundreds to thousands of individual slip events (Scholz 1990; Cowie & Roberts 2001), most fault rupture events occur along pre-existing faults. In low temperature regimes, compaction and cementation processes, and associated permeability reduction, can be slow on timescales of slip recurrence, and active faults can remain relatively incohesive between slip events. Despite this, renewed rupture and ⁄ or the development of extension hydrofracture arrays adjacent to faults stimulates the regeneration of permeability. However, at elevated temperatures, and especially in fluid-active regimes, the internal structures of fault rocks and associated vein arrays indicate that fault zones commonly become effectively sealed and partially regain cohesive strength on the timescales of rupture recurrence (Cox 2005). In the following part of the study, conditions for failure and potential permeability enhancement in incohesive faults are discussed. In the subsequent section, the major effects of time-dependent recovery of fault cohesive strength on failure modes and associated permeability regeneration are explored. Failure conditions for incohesive faults in a rock mass containing optimally and non-optimally oriented faults have been discussed previously using traditional Mohr– Coulomb constructions (e.g. Sibson 1985; Thatcher & Hill 1991; Hill & Thatcher 1992). Figure 8 illustrates examples of failure envelopes in k–r space and shows clearly conditions for failure and associated permeability enhancement for incohesive reverse and normal fault regimes as a function of misorientation angle hr. The failure envelope for intact rock (for the case l = ls) is also indicated. For reverse faulting, the slope of the re-shear failure envelope is positive for hr > 2hopt, horizontal for hr = 2hopt, and becomes progressively more negative as hr decreases in the range hopt £ hr < 2hopt. The slope of the re-shear failure envelope can be calculated from eqn 5. For reverse faulting, assuming r3 constant okv 1 ðl l cos 2hr sin 2hr Þ: ¼ oðr1 r3 Þ 2ls r3
ð18Þ
For normal faulting, with r1 constant, okv 1 ð1 þ l cos 2hr þ sin 2hr Þ: ¼ oðr1 r3 Þ 2ls r3
ð19Þ
Figure 9 illustrates the variation in slope of the re-shear failure envelope as a function of hr for 0 £ hr £ 90, for
80° 70°
60° 2θopt
1
Lithostatic
50°
λv
Inta
ct ro
0.95
ck
40°
θopt
0.9 C = 10 MPa μs = 0.75
0.85
z = 10 km 0.8
0
20
40
60
100
80
σ1– σ3, MPa (B)
Normal faulting regime
1.1 80° 70°
1
Lithostatic Pf 60°
0.9
λv
50°
0.8
Int
ac
0.7
t ro
ck
40°
0.6
θopt
C = 10 MPa μs = 0.75
0.5
z = 10 km 0.4
0
20
40
60
σ1– σ3, MPa
80
100
Fig. 8. Effect of misorientation angle hr on failure envelopes for re-shear on cohesionless (A) reverse and (B) normal faults. Misorientation angles from optimal to hr = 80 are indicated for faults with a static friction coefficient of 0.75 and depth of 10 km. The failure envelope for intact rock with l = 0.75 and cohesive strength of 10 MPa (bold line) is also shown.
various friction coefficients during reactivation and any associated permeability enhancement in reverse and normal faults. The minimum slope of the re-shear failure envelope occurs at hopt. Note that for fault reactivation in extensional tectonic regimes, the slope of the re-shear failure envelope is negative for all ls and hr. Changes in the fault misorientation angle significantly change the shape of the failure envelope, and influence whether, for various stress and fluid pressure conditions, failure and permeability enhancement occurs by reactivation of
226 S. F. COX
(A)
1.5/σ3 1.0/σ3
λ v /(σ1– σ3)
0.5/σ3 0 0.85 0.6
–0.5/σ3
0.75
0.4
–1.0/σ3
0.2
–1.5/σ3 –2.0/σ3 –2.5/σ3
0
10
20
30
40
θr
50
60
70
80
90
50
60
70
80
90
(B) 0.5/σ3 0
λ v /(σ1– σ3)
–0.5/σ3 –1.0/σ3 0.85 0.6
–1.5/σ3
0.75
0.4
–2.0/σ3
0.2
–2.5/σ3 –3.0/σ3 –3.5/σ3
0
10
20
30
40
θr
Fig. 9. Variations in slope of failure envelope with misorientation angle hr, in k–r space, as a function of friction coefficient for re-shear of cohesionless (A) reverse and (B) normal faults.
the existing fault, whether new permeable pathways are created by the development of a new, optimally oriented rupture zone, or by the opening of networks of extension fractures or hybrid extensional-shear fractures in the wallrock. Sibson (1985) has shown previously that severely misoriented (i.e. hr > 2hopt), incohesive reverse faults can be reactivated only at supralithostatic fluid pressure conditions. This result is illustrated very clearly in k–r space. For example, for l = 0.75 and at a depth of 10 km, a preexisting, severely misoriented, incohesive reverse fault with hr = 70 can be reactivated at kv >1, but only if differential stress is approximately <10 MPa (Fig. 8A). At higher differential stress, failure and associated permeability enhancement occurs either by extension failure in intact wall-rock
(if 10 MPa < r1–r3 £ 4T), hybrid extensional-shear failure (for 4T £ r1–r3 £ 4T ⁄ sin2hopt) or by development of a new optimally oriented shear failure in intact wall-rock (for r1–r3 ‡ 4T ⁄ sin2hopt). Note also, in this example, that the conditions for generation of extension fracture permeability adjacent to cohesionless reverse faults cannot be accessed if hr is less than approximately 60 and ls = 0.75 (Fig. 8A). Even for normal faulting, near-lithostatic fluid pressures may be required for reactivation and permeability regeneration in severely misoriented fault zones deep in the continental seismogenic regime. For example, in the case modelled in Fig. 8B, faults with hr > 60 can only be reactivated if pore fluid factors are >0.9. Reactivation at these depths also can occur only at very low differential stresses. Where severely misoriented normal faults have hr greater than approximately 60, failure at differential stress >2C ⁄ sin2hopt can only occur by development of a new, optimally oriented rupture in intact rock; this creates new fluid pathways. The conditions for opening of extension fractures in wall-rock adjacent to cohesionless normal faults, at 10-km depth, cannot be accessed if hr is less than approximately 65. Additionally, the maximum shear stress for reactivation of non-optimally oriented normal faults decreases as hr increases (Fig. 8B). This situation also applies to non-optimal reverse and strike-slip faults. One further example illustrates the application of k–r failure mode diagrams to constrain stress and fluid pressure states necessary to drive failure and associated permeability enhancement in strike-slip regimes at seismogenic depths. Figure 10 is a k–r failure mode diagram for a strike-slip regime at a depth of 10 km, constructed assuming r2 = (r1 + r3) ⁄ 2. Failure envelopes for intact rock failure (ls = 0.75, C = 10 MPa, T = 5 MPa), and for re-shear on severely misoriented, cohesionless faults (hr = 70) with static friction coefficients between 0.1 and 0.75, are illustrated. A key point in the diagram is that, if friction coefficients have values close to experimentally determined values for many crustal rocks (i.e. 0.60 < l < 0.85; Byerlee 1978), then near-lithostatic pore fluid factors and very low differential stresses are required for rupture nucleation in both optimally oriented faults and misoriented faults at depth in the seismogenic regime. Indeed, a 70 misoriented strike-slip fault with ls = 0.75 can only be reactivated if kv = 1 and the differential stress is less than approximately 10 MPa. At differential stresses between approximately 10 and 25 MPa, failure and permeability enhancement involves the formation of extension fracture arrays. At higher differential stresses, failure can only occur on optimally oriented faults. At near-hydrostatic fluid pressure conditions, shear failure and related permeability enhancement can only occur at modest differential stresses if friction coefficients are as low as 0.1–0.2. This could be the case for faults containing very low friction materials, such as talc or some clay minerals (Moore & Lockner 2004).
Application of failure mode diagrams 227
θr = 70° 0.75
1
0.4
Lithostatic Pf Intact rock failure μ = 0.75
0.8 0.2 Re-shear
λ v 0.6 Hydrostatic Pf
0.4
μs= 0.1 z = 10 km C = 10 MPa Cf = 0 MPa T = 5 MPa
0.2
0 0
20
40
60
80
100
σ1–σ3, MPa Fig. 10. k–r failure mode diagram for severely misoriented strike-slip faulting at a depth of 10 km, hr = 70, and various values of the static friction coefficient. A failure envelope for intact rock with C = 10 MPa, T = 5 MPa and l = 0.75 is also shown (bold line).
Influence of time-dependent changes in fault cohesive strength for re-shear and permeability enhancement in pre-existing faults Time-dependent changes in fault cohesive strength occur in fluid-active fault zones at depth in the seismogenic regime where temperatures are high enough to promote compaction, sealing and healing of fault damage products (Angevine et al. 1982; Fredrich & Evans 1992; Sibson 1992; Kanagawa et al. 2000). Tenthorey et al. (2003), Tenthorey & Cox (2006) and Giger et al. (2008) demonstrated experimentally that, in hydrous fault zones at midcrustal conditions, substantial recovery of cohesive strength by fluid-assisted compaction and cementation processes occurs on timescales comparable with typical rupture recurrence periods in plate margin regimes. These processes result in very rapid loss of fault zone permeability in the interseismic interval (Giger et al. 2007). Although the effects of interseismic recovery of cohesive strength on failure conditions have been explored by Fournier (1996) and Streit & Cox (2001), failure mode diagrams in k–r space provide further insights about the critical role of timedependent changes in cohesive strength for controlling the modes of fracture-controlled permeability enhancement in fault zones during repeated rupture cycles. Figure 11 illustrates examples of k–r failure mode diagrams for non-optimally oriented reverse and normal faults at a depth of 10 km. For a misoriented reverse fault (i.e. hopt £ hr £ 2hopt) with hr = 50 (Fig. 11A), the effects of progressive changes in fault cohesive strength on the
position of the reactivation failure envelope are indicated. As the cohesive strength increases from 0 to 4 MPa, the reactivation shear failure envelope moves towards higher pore fluid factors. However, as the re-shear envelope moves closer to the intact rock failure envelope, the range of differential stresses for which re-shear and permeability enhancement can occur on the misoriented fault progressively diminishes and the maximum differential stress for reactivation decreases. At (r1–r3) = 0, the re-shear failure envelope starts to move to kv higher than the intact rock failure envelope when the fault cohesive strength Cf exceeds lT. The re-shear envelope moves higher than the entire intact rock failure envelope at Cf greater than approximately 4.5 MPa. Accordingly, re-activation and renewed permeability enhancement within misoriented reverse faults is no longer possible once a fault recovers approximately 50% of the intact rock cohesive strength. Similar results are apparent in Fig. 11B, which illustrates failure conditions for a severely misoriented (i.e. hr > 2hopt) reverse fault zone with hr = 70. In this case, the re-shear envelope moves to higher pore fluid factors and lower differential stresses with increasing recovery of cohesive strength. For severely misoriented reverse faults in general, re-shear is no longer possible once Cf > lT. A key result, therefore, is that even limited recovery of fault cohesive strength makes re-shear and renewed permeability enhancement along severely misoriented faults impossible; optimally oriented shear failure will occur instead, and result in a change to fluid pathways. Additionally, reactivation and renewed permeability enhancement along severely misoriented, cohesive, reverse faults can only occur if differential stresses are very low (<12 MPa for the case illustrated in Fig. 11B). Even then, reactivation can occur only at supralithostatic fluid pressures. The conditions for reactivation of cohesive, non-optimally oriented normal faults at seismogenic depths are not as extreme as for non-optimal reverse faults (Fig. 11C,D). However, as with reactivation of reverse faults, misoriented normal faults can be reactivated only if there has been limited recovery of fault cohesive strength. In the case of severely misoriented normal faults deep within the seismogenic regime (Fig. 11D), re-shear is only possible at very low differential stresses. With increasing recovery of Cf, the field for reactivation shrinks to progressively lower differential stresses and higher pore fluid factors. For cohesive strength recovery of only 20% of the intact rock cohesive strength, reactivation and associated permeability enhancement requires supralithostatic pore fluid factors and differential stresses <5 MPa. At higher differential stresses, intact rock failure occurs. The situation is similar for non-optimally oriented strike-slip faults. In a rock mass containing non-optimally oriented faults, recovery of fault cohesive strength during interseismic period results in the shape of the failure envelope and condi-
228 S. F. COX
(A) 1.05
(B) 1.05
Reverse fault zone
Reverse fault zone 3 2 1 0 MPa
2 3 0 1
1
4 MPa 1
Lithostatic Pf
0.95
Lithostatic Pf
0.95
λv
λv 0.9
0.9 μ = 0.75
μ = 0.75
z = 10 km 0.85
θopt<θr<2θopt
C0 = 10 MPa
θr = 50°
T0 = 5 MPa
0.8
z = 10 km
θr>2θopt
0.85
C0 = 10 MPa
θr = 70°
T0 = 5 MPa
0.8 0
20
40
60
80
0
100
20
σ1 – σ3, MPa
40
60
80
100
σ1 – σ3, MPa
(C) 1.05
(D) 1.1
Normal fault zone
Normal fault zone 1.05
Lithostatic Pf
1
θopt<θr<2θopt 0.95
2
1
θr = 50°
λv
Lithostatic Pf
0
θr>2θopt 0.95
0 2 4 MPa
θr = 70°
λv
0.9
0.9
μ = 0.75
μ = 0.75
z = 10 km
0.85
z = 10 km 0.85
C0 = 10 MPa
C0 = 10 MPa
T0 = 5 MPa
T0 = 5 MPa
0.8 0
10
20
30
40
50
σ1 – σ3, MPa
0.8 0
10
20
30
40
50
σ1 – σ3, MPa
Fig. 11. Effects of changes in fault cohesive strength Cf on shape of failure envelopes for faulting at a depth of 10 km. (A) Misoriented reverse fault zone with hr = 50. (B) Severely misoriented reverse fault zone with hr = 70. (C) Misoriented normal fault zone with hr = 50. (D) Severely misoriented normal fault zone with hr = 70. Thin lines are the shear failure envelopes for reactivation of non-optimally oriented faults for various indicated cohesive strengths. The bold line is the failure envelope for intact rock.
tions for regeneration of fracture permeability changing with time. At depth in the crustal seismogenic regime, and in the absence of recovery of cohesive strength, non-optimally oriented faults can be reactivated only at relatively low shear stresses and elevated pore fluid factors. However, with progressive interseismic fault healing, the failure mode, and hence nature of permeability regeneration (e.g. extension failure, re-shear or optimally oriented shear failure of intact rock), is very sensitive to the fault misorienta-
tion angle and the amount of cohesive strength recovery, as well as the pore pressure and stress conditions. For example, around misoriented fault zones, the extension failure regime cannot be accessed at zero cohesive strength (Fig. 11A,C). Yet, as cohesive strength recovers, and the re-shear failure envelope moves to higher kv, extension failure and associated permeability enhancement adjacent to the fault zone can take place, but only at very low differential stresses. k–r failure mode diagrams provide a very
Application of failure mode diagrams 229 powerful way to analyse conditions for permeability regeneration around non-optimally oriented faults that recover cohesive strength in the interseismic period.
PATHS TO FAILURE: THE RELATIVE ROLES OF CHANGES IN STRESS AND FLUID PRESSURE STATES IN DRIVING FAILURE AND PERMEABILITY REGENERATION In fluid-active faults in the crustal seismogenic regime, the evolution of shear stress, pore fluid factor and fluid flow rates is likely to be coupled in complex ways during the seismic cycle. In scenarios where the pore fluid factor does not change during the seismic cycle, failure is purely stressdriven (e.g. path A in Fig. 12). This is expected to be the typical case in shallow crustal, hydrostatic-pressured regimes. In contrast, if hydrothermal sealing of fracture permeability in fault zones is rapid relative to the duration of interseismic periods, then fault seals can form. If pore fluid pressures build-up beneath seals, for example in response to fluid leakage from deeper-level overpressured reservoirs, pore fluid factors can increase rapidly relative to rates of fault loading. Subsequent failure can then be largely fluid-driven (e.g. paths C to E in Fig. 12). In general in fluid-active fault settings, failure may be induced by a combination of changes in pore fluid factor and shear stress (Sibson 2001). The relative rates of change of pore fluid factor and shear stress during loading to failure are expected to be influenced particularly by factors such as
Extension failure Hybrid extensional shear failure
D E λv
Brittle shear failure
C B A
σ1– σ3 Fig. 12. Schematic k–r failure mode diagram illustrating how changes in fluid pressure and stress states can drive failure. Path A is stress-driven brittle shear failure. On path B, brittle shear failure is driven by changes in both differential stress and pore fluid factor. On path C, brittle shear failure occurs purely in response to fluid pressure change. At low differential stresses, extension failure can be driven by high rates of change of pore fluid factor relative to differential stress, i.e. largely fluid-driven failure (paths D and E).
the rate of development of fault seals, growth of hydrofracture dilatancy, and the evolution of hydraulic connectivity between fluid conductive faults and overpressured fluid reservoirs. Failure mode and styles of permeability enhancement in fault zones are sensitive to the relative rates of recovery of pore fluid factor (¶kv ⁄ ¶t) and differential stress [¶(r1– r3) ⁄ ¶t] between failure events, as well as to the magnitudes of fluid pressure and differential stress at the start of each loading cycle. k–r failure mode diagrams provide an instructive way to analyse the evolution of fluid pressure and stress states during fault-valve cycles and illustrate the key role of fluid-driven failure at depth in the crust. For both optimally oriented and non-optimally oriented faults, the internal structures and vein styles in fault zones (e.g. Figs 1 and 2) can be interpreted, at least qualitatively, in terms of loading and fluid pressurization paths [¶kv ⁄ ¶(r1– r3)] and ¶Cf ⁄ ¶t, as well as kv and stress states, during fault-valve behaviour. This will be illustrated with several examples. Figure 1A illustrates a small reverse fault zone and an associated array of extension veins. This structure formed at depths of approximately 8–12 km as part of a gold-hosting, high fluid flux fault network (Cox 1995). Geometric relationships between the extension veins and the fault zone indicate that the fault was close to optimally oriented. Figure 13A illustrates a k–r failure mode diagram for fault-valve cycles in such a fault zone formed at a depth of 10 km. Assuming that shear stress increases monotonically during the loading cycle, the presence of extension veins indicates that fluid pressure and stress recovery following a fault rupture event must follow a high ¶kv ⁄ ¶(r1–r3) trajectory such as path A. Additionally, each loading cycle must also commence at a low differential stress (i.e. r1– r3 £ 4T). Nucleation, growth and dilation of extension fractures (starting at point 1 in Fig. 13A) limits further rise in pore fluid pressures. Accordingly, regeneration of permeability prior to pervasive shear failure along faults is associated with the opening of arrays of extension fractures adjacent to the fault core. This may localize prerupture fluid flow along the margins of the fault zone. Note also that the geometry of the extension fracture network leads to significant permeability anisotropy with flow preferentially along the r2 direction in the vein array (i.e. parallel to the line of intersection of the extension vein arrays and the fault plane). Continuing interseismic recovery of shear stress will drive the stress state along the failure envelope towards the shear failure condition (Fig. 13A). At the boundary between the extension fracture regime and the hybrid extensional-shear fracture regimes (point 2), hybrid extensional-shear fractures can start to form. Note also that a small fluid pressure drop must occur (presumably accommodated by fracture dilation and associated permeability
230 S. F. COX
(A)
θr = θopt
1.1 1
2 3
1
Lithostatic Pf
B
A λv
C
0.9
D
E
0.8 Reverse z = 10 km μ = 0.75
0.7
C = 10 MPa T = 5MPa 0.6 50
0
σ1 – σ3, MPa
150
100
(B) 1 2
θr = θopt
3
1
4
A 0.8
B C
0.6
Ductile creep regime
λv
• 103ε0 0.4 • 10ε0
• 102ε0
Optimal reverse z = 15 km μ = 0.75 C = 10 MPa T = 5 MPa
•
ε0
0.2
• 3 ε∝σ 0 0
50
σ1 – σ3, MPa
100
150
enhancement) for the stress and fluid pressure states to track further along the failure envelope to the shear failure regime at point 3. At this latter point, shear failure and associated dilatancy and permeability enhancement lead to a sudden drop in both shear stress and pore fluid factor, and the fault-valve cycle can start again. The occurrence of extension veins within fault zones is thus diagnostic of at least some shear failure events and associated permeability regeneration occurring at very low differential stresses. Such low shear stress fault-valve behaviour typically is associated with a paucity of wear products along slip surfaces. Lower ¶kv ⁄ ¶(r1–r3) during interseismic recovery of shear stress and pore fluid pressures can lead to brittle shear failure occurring at higher differential stresses (or
Figure 13. k–r trajectories during fault-valve cycles in optimally oriented fault zones. (A) Intact rock failure in a contractional deformation regime at 10-km depth. From low initial (r1–r3), high ¶kv ⁄ ¶(r1–r3) along path A drives initial extension failure at point 1. Opening of extension fractures inhibits further build-up of fluid pressure. Continued recovery of shear stress drives the fluid pressure and stress state along the failure envelope, through the hybrid extensional shear failure regime (region 2) to the compressional shear failure envelope at point 3. At this point, shear failure occurs, and results in both stress drop and fluid pressure relief. With decreasing ¶kv ⁄ ¶(r1–r3) during interseismic recovery of shear stress and pore fluid factor, brittle shear failure occurs at progressively increasing differential stresses (e.g. paths C to E). (B) Viscous creep may be associated fault-valve cycles at elevated temperatures. The brittle failure envelope is illustrated together with strain rate contours for viscous creep, based on a cubic power law. In path A, during steady state creep at low differential stress, increase in pore fluid factor with time drives pure extension failure when the stress and pore fluid pressure state reaches point 1. At lower ¶kv ⁄ ¶(r1–r3), shear stress increase with time is accompanied by accelerating creep rates. Along path B, extension failure occurs at point 2, and accelerating viscous creep is terminated by brittle shear failure, accompanied by stress drop and fluid pressure relief at point 3. Along path C, accelerating ductile creep is terminated by brittle shear failure at point 4, without growth of spatially associated extension fractures.
shear stresses) and lower pore fluid factor (e.g. paths B to E in Fig. 13A). Failure cycles such as B to E in Fig. 13A are not associated with the formation of fault-related extension veins. Decreased ¶kv ⁄ ¶(r1–r3) in general may to lead to increased intensity of wear damage (cataclasites and wear breccias, e.g. Fig. 2A) as the shear stress and effective normal stress at shear failure increases. If fluid pressure and shear stress recovery during faultvalve behaviour occurs at sufficiently high temperatures, significant viscous creep can occur during the interseismic period. This scenario is illustrated in Fig. 13B, in which viscous creep failure envelopes for various strain rates are indicated at stress and fluid pressure conditions below those of the brittle failure envelope. It is assumed, for illustrative purposes, that creep occurs by a pressure insensitive cubic power law. Where the rate of recovery of fluid pressure relative to the rate of recovery of differential stress is very high (e.g. path A), fluid pressure can increase at approximately constant creep rates. At low enough differential stress, increasing pore fluid factor can ultimately drive pure extension failure during ductile creep when the stress and pore fluid pressure state reaches the extension failure envelope at point 1 in Fig. 13B. Provided creep rate is fast enough to balance the tectonically imposed loading rate, then viscous deformation at constant strain rate can continue during the development of en echelon extension vein arrays in the shear zone, without further build-up of shear stress. Arrays of variably deformed, planar to sigmoidal extension veins associated with ductile shear zones (e.g. Fig. 1B) provide an example of this scenario. The formation of macroscopic extension fractures during viscous creep in shear zones should lead to marked permeability anisotropy favouring
Application of failure mode diagrams 231 flow along the r2 direction. However, it should also be noted that viscous deformation at very high pore fluid factors is also likely to be associated with enhanced microcrack permeability along the core of the shear zone (Zhang et al. 1995). At lower ¶kv ⁄ ¶(r1–r3), increase in differential stress with time is accompanied by accelerating creep rates. Along path B in Fig. 13B, for example, extension failure occurs at point 2, and limits further build-up of fluid pressures. However, ongoing recovery of shear stress can drive the kv and (r1–r3) conditions towards the brittle shear failure envelope. Brittle shear failure, accompanied by permeability enhancement, as well as stress drop and fluid pressure relief, occurs at point 3. Such shear zones are characterized by an association of extension veins, ductile shear fabrics, brittle slip surfaces and possibly associated damage products. Mutually overprinting relationships between these various structures would provide support for repeated fault-valve cycling at low shear stresses. At relatively low ¶kv ⁄ ¶(r1–r3), if ¶kv ⁄ ¶t remains approximately constant, accelerating viscous creep retards the tectonic loading rate, causing a decrease in ¶(r1–r3) ⁄ ¶t with increasing (r1–r3). This produces loading paths such as C in Fig. 13B, in which ¶kv ⁄ ¶(r1–r3) increases with time, and a period of interseismic viscous creep, during shear stress and fluid pressure recovery, may be terminated by brittle shear failure and permeability enhancement along the ruptured zone. Failure occurs without the formation of spatially associated extension veins. In such scenarios, faultvalve cycling will result in mutually overprinting relationships between brittle and viscous shear fabrics in fault zones. These loading paths illustrate how transitions from viscous to brittle behaviour can be related purely to changes in stress and fluid pressure states within fault-valve cycles, and need not necessarily indicate temperature changes during progressive deformation. A range of paths to failure and permeability regeneration is illustrated in Fig. 14A, for scenarios involving a cohesive, misoriented reverse fault at a depth of 10 km and with hr = 50. During loading from very low differential stresses, permeability regeneration by extension failure adjacent to such a misoriented fault requires high ¶kv ⁄ ¶(r1–r3) loading paths such as A. Continuing recovery of shear stress ultimately leads to rupture and permeability regeneration on the misoriented fault, along with relief of shear stress and fluid pressure. For even lower ¶kv ⁄ ¶(r1–r3) paths such as B, permeability enhancement and reactivation of the misoriented fault can take place at low-to-moderate differential stress, without prior opening of associated extension fractures. Even lower ¶kv ⁄ ¶(r1– r3) paths such as C in Fig. 14A, in which (r1–r3) becomes greater than approximately 50 MPa before failure, shear failure can only occur on a new, optimally oriented fault, rather than by reactivation of the existing,
(A) 1.15 Re-shear failure envelope θr = 50° Cf = 4 MPa
1.1
θopt < θr < 2θopt
1.05 Intact rock failure envelope 1
λv 0.95
A B
0.9
C Reverse μ = 0.75 C = 10 MPa z = 10 km
0.85
0.8 0
20
40
60
σ1 –σ3, MPa
(B) 1.15
80
100
θr > 2θopt
1.1
Re-shear failure envelope θr = 70°, Cf = 2 MPa
1.05 1
3 1
B′
B″
λv 0.95
Intact rock failure envelope
2
B C
A
0.9
Reverse μ = 0.75 C = 10 MPa z = 10 km
0.85
0.8 0
20
40
60
σ1 –σ3, MPa
80
100
Figure 14. k–r trajectories during fault-valve cycles in non-optimally oriented fault zones. (A) For fault-valve cycles in misoriented reverse faults (e.g. hr = 50), failure modes are influenced by ¶kv ⁄ ¶(r1–r3) during interseismic recovery of shear stress and pore fluid factor. Path A produces extension failure in intact rock prior to re-shear on the misoriented fault. Path B results in re-shear on the misoriented fault. Path C leads to optimally oriented failure of the intact rock, rather than re-shear of the pre-existing misoriented fault. (B) For fault-valve cycles in severely misoriented reverse fault zones (e.g. hr = 70), very high ¶kv ⁄ ¶(r1–r3) and very low differential stresses (e.g. path A) favour reactivation. Recovery of differential stress and pore fluid factor along path B–B¢ results first in extension failure of intact wall-rock at point 1. Ongoing recovery of shear stress ultimately drives optimally oriented brittle shear failure of the intact rock at point 2. The severely misoriented fault can be reactivated after opening of extension vein arrays in the same fault-valve cycle, only if a stress drop drives the stress and fluid pressure state from points 1 to 3 (path B–B¢¢). Recovery of kv and r1–r3 at low ¶kv ⁄ ¶(r1–r3) (e.g. path C) favours optimally oriented shear failure at differential stresses much higher than required for reactivation of the severely misoriented fault.
232 S. F. COX misoriented fault. In this case, the misoriented fault remains inactive and new high permeability fluid pathways are generated along a new, optimally oriented fault. As noted earlier, brittle shear failure along severely misoriented fault zones, especially reverse faults at seismogenic depths, can only occur in a very restricted range of pore fluid pressure and stress states. Figure 14B illustrates a range of paths for loading cohesive, severely misoriented, reverse fault zones to failure. Severely misoriented faults are especially common in mesothermal gold deposits (e.g. Sibson et al. 1988; Cox 1995), and also in situations where steeply dipping, optimally oriented normal faults are reactivated as high angle reverse faults during tectonic inversion (Sibson 2007). It is clear that reactivation and permeability regeneration along severely misoriented faults can occur only at very low differential stresses and high pore fluid factors, and accordingly requires relatively high ¶kv ⁄ ¶(r1–r3) loading paths during fluid pressure and stress recovery prior to reactivation (e.g. path A). Starting from the same initial stress and fluid pressure state, slightly lower ¶kv ⁄ ¶(r1–r3) loading paths result in initial permeability enhancement by opening of extension veins in intact rock when the stress and fluid pressure states first reach the failure envelope (path B, point 1). Subsequent increase in (r1–r3) due to ongoing stress recovery (path B¢) leads to optimally oriented, intact rock failure, rather than re-shear on the severely misoriented fault. Interestingly, a transition from extensional mode failure to reactivation of a severely misoriented fault during one fault-valve cycle can occur only if the differential stress decreases after extension fracturing commences (path B¢¢ in Fig. 14B). Such reactivation could be triggered, for example, in response to slip on a nearby fault zone causing a sudden drop in shear stress. The common development of extension vein arrays adjacent to severely misoriented faults (e.g. Sibson et al. 1988; Cox 1995) suggests that triggering of permeability enhancement along misoriented faults by stress drop might not be unusual.
CONCLUSIONS Failure mode diagrams in pore fluid factor–differential stress space provide a powerful tool for visualizing and exploring evolution of stress and fluid pressure states during fault-valve cycles. They also provide a useful way for examining how stress and fluid pressure regimes influence failure modes, and hence vein styles, localization of permeability enhancement and timing of permeability regeneration within fault-valve cycles. The shape of the failure envelope in k–r space, and the stress and fluid pressure conditions necessary for permeability regeneration, are sensitive to tectonic regime (contractional, extensional, strike-slip), depth, a range of rock material properties (cohesive strength, tensile strength,
friction coefficient), and the orientation of faults with respect to the maximum principal stress. Relative rates of recovery of pore fluid factor, differential stress and fault cohesive strength during fault-valve cycles are key factors influencing the fluid pressure and stress conditions at failure and whether permeability regeneration occurs via growth of hydraulic extension fractures adjacent to faults, by slip on non-optimally oriented faults, or by rupture on new, optimally oriented faults. For fault-vein systems formed at depth in the crustal seismogenic regime, the presence of extension vein networks around faults, and evidence for repeated reactivation of severely misoriented faults provide key evidence for predominantly fluid-driven failure at very low differential stresses in overpressured hydrothermal systems.
ACKNOWLEDGEMENTS S. Barker, A. Barnhoorn, M. Crawford, S. Giger, R. Henley, S. Micklethwaite and R. Sibson are thanked for providing highly stimulating discussion about failure and fluid flow. J.L. Urai is thanked for introducing the author to the spectacular, fault-related vein systems of the Jabal Akhdar dome in Oman. This research was supported by Australian Research Council grants DP0452448, LP0453836, LP0562164 and CEO561595. Tom Blenkinsop and Steve Miller are thanked for helpful reviews.
REFERENCES Angevine CL, Turcotte DL, Furnish MD (1982) Pressure solution lithification as a mechanism for stick-slip behaviour of faults. Tectonics, 1, 151–60. Blanpied ML, Lockner DA, Byerlee JD (1992) An earthquake mechanism based on rapid sealing of faults. Nature, 358, 574–6. Byerlee JD (1978) Friction of rocks. Pure and Applied Geophysics, 116, 615–26. Cowie PA, Roberts GP (2001) Constraining slip rates and spacings for active normal faults. Journal of Structural Geology, 23, 1901–15. Cox SF (1995) Faulting processes at high fluid pressures: an example of fault-valve behavior from the Wattle Gully Fault Zone, Victoria, Australia. Journal of Geophysical Research, 100, 12841–59. Cox SF (2005) Coupling between deformation, fluid pressures and fluid flow in ore-producing hydrothermal environments. Economic Geology, 100th Anniversary Volume, 39–75. Cox SF (2007) Structural and isotopic constraints on fluid flow regimes and fluid pathways during upper crustal deformation: an example from the Taemas area of the Lachlan Orogen, SE Australia. Journal of Geophysical Research, 112, B08208; doi: 10.1029/2006JB004734. Cox SF, Ruming K (2004) The St Ives Mesothermal Gold System, Western Australia – a case of golden aftershocks? Journal of Structural Geology, 26, 1109–25. Cox SF, Braun J, Knackstedt MA (2001) Principles of structural control on permeability and fluid flow in hydrothermal systems. Reviews in Economic Geology, 14, 1–24. Etheridge MA (1983) Differential stress magnitudes during regional deformation and metamorphism: upper bound imposed by tensile fracturing. Geology, 11, 231–4.
Application of failure mode diagrams 233 Fournier RO (1996) Compressive and tensile failure at high fluid pressure where preexisting fractures have cohesive strength, with application to the San Andreas Fault. Journal of Geophysical Research, 101, 25499–509. Fredrich JT, Evans B (1992) Strength recovery along simulated faults by solution transfer. Proceedings of the US Rock Mechanics Symposium, 33, 121–30. Giger SB, Tenthorey E, Cox SF, Fitz Gerald JG (2007) Permeability evolution of quartz gouges under hydrothermal conditions. Journal of Geophysical Research, 112, B07202; doi: 10.1029/2006JB004828. Giger SB, Cox SF, Tenthorey E (2008) Slip localization and fault weakening as a consequence of fault gouge strengthening – insights from laboratory experiments. Earth and Planetary Science Letters, 276, 73–84. Griffith AA (1924) Theory of rupture. In: Proceedings of the First International Congress on Applied Mechanics (eds Biezeno CB, Burgers JM), pp. 55–63. J Waltman, Delft. Hilgers C, Kirschner DL, Breton JP, Urai JL (2006) Fracture sealing and fluid overpressures in limestones of the Jabal Akhdar dome, Oman mountains. Geofluids, 6, 168–84. Hill DP, Thatcher W (1992) An energy constraint for frictional slip on misoriented faults. Bulletin of the Seismological Society of America, 82, 883–97. Hubbert MK, Rubey WW (1959) Role of fluid pressure in mechanics of overthrust faulting. Part 1. Geological Society of America Bulletin, 62, 355–72. Jaeger JC, Cook NGW (1979) Fundamentals of Rock Mechanics, 3rd edn. Chapman & Hall, London. Kanagawa K, Cox SF, Zhang S (2000) Effects of dissolution-precipitation creep on the strength and mechanical behavior of quartz gouge at high-temperature hydrothermal conditions. Journal of Geophysical Research, 105, 11115–26. Micklethwaite S (2009) Mechanisms of faulting and permeability enhancement during epithermal mineralisation: Cracow gold field, Australia. Journal of Structural Geology, 31, 288–300. Miller SA, Nur A, Olgaard DL (1996) Earthquakes as a coupled shear stress high pore fluid pressure dynamical system. Geophysical Research Letters, 23, 197–200. Moore DE, Lockner DA (2004) Crystallographic controls on the frictional behaviour of dry and water-saturated sheet silicate structure minerals. Journal of Geophysical Research, 109, B03401; doi: 10.1029/2003JB002582. Robert F, Poulsen HK (2001) Vein formation and deformation in greenstone gold deposits. Reviews in Economic Geology, 14, 111–55. Scholz CH (1990) The Mechanics of Earthquakes and Faulting. Cambridge University Press, Cambridge. Secor DT (1965) Role of fluid pressure in jointing. American Journal of Science, 263, 633–46.
Sibson RH (1981)Fluid flow accompanying faulting: field evidence and models. In: Earthquake Prediction: An International Review, Union (eds. Simpson DW, Richards PG), pp 593–603. Maurice Ewing Series 4, American Geophysical Union, Washington, DC. Sibson RH (1985) A note on fault reactivation. Journal of Structural Geology, 7, 751–4. Sibson RH (1986) Brecciation processes in fault zones: inferences from earthquake rupturing. Pure and Applied Geophysics, 124, 159–75. Sibson RH (1992) Implications of fault-valve behaviour for rupture nucleation and recurrence. Tectonophysics, 211, 283–93. Sibson RH (1998) Brittle failure mode plots for compressional and extensional tectonic regimes. Journal of Structural Geology, 20, 655–60. Sibson RH (2000) A brittle failure mode plot defining conditions for high-flux flow. Economic Geology, 95, 41–8. Sibson RH (2001) Seismogenic framework for ore deposition. Reviews in Economic Geology, 14, 25–50. Sibson RH (2004) Controls on maximum fluid overpressure defining conditions for mesozonal mineralization. Journal of Structural Geology, 26, 1127–36. Sibson RH (2007) An episode of fault-valve behaviour during compressional inversion? – The 2004 Mj 6.8 mid-Niigata Prefecture, Japan, earthquake sequence. Earth and Planetary Science Letters, 257, 188–99. Sibson RH, Robert F, Poulsen KH (1988) High-angle reverse faults, fluid-pressure cycling, and mesothermal gold deposits. Geology, 16, 551–5. Simmons SF, White NC, John DA (2005) Geological characteristics of epithermal precious and base metal deposits. Economic Geology, 100th Anniversary Volume, 485–522. Stein RS (1999) The role of stress transfer in earthquake occurrence. Nature, 402, 605–9. Streit JE, Cox SF (2001) Fluid pressures at hypocenters of moderate to large earthquakes. Journal of Geophysical Research, 106, 2235–43. Tenthorey E, Cox SF (2006) Cohesive strengthening of fault zones during the interseismic period: an experimental study. Journal of Geophysical Research, 111, B09202; doi: 10.1029/ 2005JB004122. Tenthorey E, Cox SF, Todd HF (2003) Evolution of strength recovery and permeability during fluid-rock reaction in experimental fault zones. Earth and Planetary Science Letters, 206, 161–72. Thatcher W, Hill DP (1991) Fault orientations in extensional and conjugate strike-slip environments and their implications. Geology, 19, 1116–20. Zhang S, Cox SF, Paterson MS (1995) The influence of room temperature deformation on porosity and permeability in calcite aggregates. Journal of Geophysical Research, 99, 15761–75.
Rates of retrograde metamorphism and their implications for crustal rheology B. W. D. YARDLEY1, D. E. HARLOV2 AND W. HEINRICH2 1
School of Earth and Environment, University of Leeds, Leeds, UK; 2Section 3.3, Chemistry and Physics of Earth Materials, Deutsches GeoForschungsZentrum, Telegrafenberg, Potsdam, Germany
ABSTRACT The rate of reaction of a natural hornblende garnet granulite with water under a range of mid- to lower crustal conditions has been investigated experimentally. In runs of between 7 and 84 days small but measurable amounts of water were consumed, and sheet silicates (300C, 300 MPa and 400C, 400 MPa) and ⁄ or secondary actinolite (400C, 400 MPa and 500C, 500 MPa) were observed to have grown. When normalized to the surface area of the starting materials, hydration rates were in the range of 2–5 · 10)8 g m)2 sec)1. These reaction rates imply that a film of water that infiltrated a planar crack with a half width of 100 lm would be completely consumed within c. 100 years. These results imply that where water infiltrates the crust along faults or underlying shear zones in response to a deformation, it will remain as a free phase for only a finite period of time, which in some cases will be less than the repeat time for major earthquakes in the fault system. Thus, the rheology of fault zones and shear zones is likely to be cyclical, with the zone becoming stronger with time as water is consumed, and then weakened by infiltration of water after each rupture. Key words: crustal rheology, reaction rate, retrogression, shear zone Received 5 October 2009; accepted 10 February 2010 Corresponding author: Bruce W. D. Yardley, School of Earth and Environment, University of Leeds, Leeds LS2 9JT, UK. Email:
[email protected]. Tel: +44-113-343-5227. Fax: +44-113-343-5259. Geofluids (2010) 10, 234–240
INTRODUCTION Retrograde metamorphism – the replacement of relatively anhydrous high-T mineral assemblages by hydrous minerals, often accompanied by carbonates – is a ubiquitous feature of exposed crystalline rocks. Despite this, strongly retrogressed rocks are usually of localized occurrence, being found primarily in shear zones, fault damage zones and in the vicinity of faults and joints, with fresher rocks occurring nearby. There are two possible explanations for the widespread preservation of high-temperature mineral assemblages in rocks that have experienced a long period of residence in the upper crust during exhumation: either the supply of fluid is strictly limited, so that retrograde reactions cannot proceed, or fluids are present for much of the time, but reaction rates are too sluggish away from zones of deformation. The widespread association of retrogression with deformation, which has been noted for many years (see discussion in Turner 1981), can be invoked to support either interpretation. The significance of the issue,
i.e. whether or not crustal rocks remain wet for long periods of time, is that it has important consequences for how the strength of the crust is modelled. In the presence of a free, water-bearing fluid phase, rocks can deform with the aid of intercrystalline deformation mechanisms, such as granular flow, grain boundary sliding, pressure solution and recrystallization, which are not operative in dry rocks (Stunitz & Fitzgerald 1993; Blanpied et al. 1995; Fusseis & Handy 2008; Menegon et al. 2008). Intracrystalline deformation is also enhanced by water, although water may be present in defect cores in rocks that lack a free pore fluid. Geodynamic models have often assumed that the deeper parts of the crust behave as though they were uniformly wet, and are therefore weak (Chen & Molnar 1983). Petrologists have pointed out that this is unlikely because of the persistence of fresh high-grade rocks to the surface (Frost & Bucher 1996; Yardley & Valley 1997), but even if the presence of water in the deep crust was restricted to localized deformation zones it could have a significant
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Rates of retrograde metamorphism 235 impact on the strength of the crust as a whole. If water is injected into deformation zones episodically, in response to seismic activity (Sibson et al. 1988; Muir-Wood & King 1993; Brodsky et al. 2003) it is possible that all this water will be consumed by reaction before the next injection, and hence the rheology of the deformation zone will change from weak to strong between seismic events. By contrast, if hydration reaction rates proceed slowly relative to the interval between injections of water, rheology will not change significantly. This study seeks to evaluate which of these models is most likely to prevail. The rates of retrograde reactions may be controlled by different limiting steps in different circumstances. The rate that we seek to measure is that of the surface hydration reaction in the presence of excess water and at fixed P–T conditions where heat generated by the exothermic hydration reaction is insufficient to inhibit retrogression. A comparable study of the rates of reaction among muscovite, K-feldspar, andalusite, quartz and water was made by Schramke et al. (1986) and will be referred to in further detail below. The experimental approach in this study was to react granulite with a carefully weighed amount of water in an Au capsule. These were run at a series of pressures and temperatures ranging 300–500 MPa and 300–500C, and at the end of the run, capsules were weighed, pierced, dried and reweighed in order to determine the amount of free water remaining at the end of the run. The difference in the weight of water between the beginning and end of the run was taken to be the weight of water combined into hydrous minerals in the course of the experiment, and was normalized to the surface area of the starting material. Starting materials Experiments were carried out on a powdered sample of hornblende granulite from South Harris, kindly supplied by Dr Bob Cliff. Hornblende and plagioclase were the most abundant minerals present, but the sample also contained quartz, clinopyroxene, garnet and minor biotite. The sample was crushed and sieved so that most particles were monomineralic and the grain size was c. 200 lm (Fig. 1). Experimental technique Between 30 and 60 mg of the dried granulite powder was loaded into 3-mm OD gold capsules, c. 12 mm in length, and c. 5 mg distilled water was added while the capsule was on the balance. The rate of evaporative water loss was recorded, as well as the weight, and the time taken for the capsule to be removed and tightly crimped was measured to provide the best possible measure of the water content. Weighings were made to the nearest 0.0001 mg and are believed to be accurate, after allowance for evaporation, to
Fig. 1. SEM image of powdered granulite used as starting material. Most grains are of silicates but the pale grain at the bottom left of the image is ilmenite.
0.0010 mg. Capsules were immediately arc welded shut and reweighed and were then loaded in duplicate into cold-seal autoclaves on a hydrothermal line and run at 300C at 300 MPa, 400C at 400 MPa or 500C at 500 MPa for periods ranging from 7 to 84 days (Table 1). In all cases the experiments took place at substantially lower temperatures than the original metamorphic peak. At the end of the run, the Au capsules were extracted from the autoclave, thoroughly cleaned, dried, weighed and then carefully punctured under a binocular microscope to ensure that no particulate material was lost as residual pressure within the capsule was released. The punctured capsules were dried at 110C for 3 h before being reweighed to determine the weight of water remaining at the end of the run. The change in the weight of water was assumed to be the weight of free water that combined onto new minerals during the run. Results The weighings indicate that a small amount of water chemically reacted with the minerals from the hornblende granulite in most of the experiments. In the majority of the runs, the amount of water fixed in the hydrous phases was in the order of 0.05 mg. Significantly higher values were rejected, and are now believed to have arisen because of incomplete loss of water during drying. In order to confirm that hydration reactions had indeed taken place, a number of capsules were opened and the contents investigated using SEM and XRD. In all cases it was noticeable that, on opening the capsule, the charge was more coherent than the original powder. This was confirmed by observation of significant secondary mineral growth under SE imaging. Primary minerals in runs at 300C showed evidence of partial dissolution, as well as
236 B. W. D. YARDLEY et al. Table 1 Summary of granulite hydration experiments.
Capsule
wt powder (mg)
300C ⁄ 3 kbar 8 63.916 13 32.944 16 52.022 Mean 400C ⁄ 4 kbar 11 41.913 12 51.789 14 42.788 21 48.109 Mean 500C ⁄ 5 kbar 10 48.925 17 59.157 19 65.675 Mean
wt H2O initially (mg)
wt H2O consumed (mg)
Surface area (m2)
Run time (hr)
Rate of water consumed (g sec)1 m)2)
Time to consume 0.1 mm film (years)
5.094 6.199 5.716
0.043 0.089 0.143
0.000685 0.000353 0.000557
164 2016 2016
1.0747E)07 3.4664E)08 3.5226E)08 5.9119E)08
90.7
8.9093E)08 4.0835E)08 1.6995E)08 1.7188E)08 4.1028E)08
77.2
3.7482E)08 4.4130E)08 3.3641E)08 3.8418E)08
82.5
7.084 6.608 5.337 5.020
5.412 8.282 5.334
0.048 0.027 0.005 0.064
0.012 0.101 0.086
(A)
(B)
(C)
(D)
(E)
(F)
0.000449 0.000555 0.000458 0.000515
0.000524 0.000634 0.000704
331 331 164 2016
164 1008 1008
Fig. 2. SEM images of run products from long duration experiments: (A) corroded plagioclase with minor overgrowths of sheet silicate, 300C, 3 kbar; (B) bundles of secondary sheet silicates, 300C, 3 kbar; (C) secondary sheet silicates on an etched substrate, 400C, 4 kbar; (D) etched and fractured primary hornblende with possible secondary amphibole needles, 400C, 4 kbar; (E) garnet surface with prominent etch pits, 500C, 5 kbar; (F) bundle of fine actinolite needles on a hornblende substrate, 500C, 5 kbar.
some mechanical break-up. Growth of clumps of newly formed sheet silicates was also seen (Fig. 2A,B). At 400C there is again evidence for the dissolution of primary mineral grains and the growth of sheet silicates. However,
additional mineral products are also present. Energy dispersive spectrometer (EDS) spectra indicate that some of the plagioclase has been albitized. There are also fine needles of amphibole present, although it is difficult to entirely
Rates of retrograde metamorphism 237 rule out that they may have originated from the fragmentation of the original hornblende (Fig. 2C,D). The most advanced reaction was observed for runs of c. 6 weeks at 500C (Fig. 2E,F). Here, bundles of fibrous actinolite bind together the particles in the charge and even garnet shows well-developed etch pits. Secondary amphiboles in this run are sufficiently well developed to yield EDS spectra and these are readily distinguished from the primary hornblende (Fig. 3). In the secondary amphibole, Na, K and Ti are not detected. The Al and Mg peaks are significantly smaller, while the Fe peak is larger. The XRD traces, while necessarily complex because of the variety of minerals in the starting material, indicate the development of a sheet silicate phase with a new peak at a low 2h in the 300 and 400C run products. In the ˚ , comparable with run at 300C this peak is close to 15 A nontronite, whereas at 400C the peak lies between 12 ˚ and is indicative of illite–montmorillonite and 14 A (Fig. 4). The runs at 500C generally show no peak in this area, although a small peak near 12 A˚ was observed in one run. EDS spectra for these secondary sheet silicates exhibit similar relative peak heights for Al, Fe and Mg, but the 300C clays typically show a small peak for Ca but often no clear peak for K, while at 400C the K peak is clearly present, and the Ca peak is smaller. This supports the XRD evidence for a more illite-like phyllosilicate at 400C. Interpretation The measured fixation of water into hydrous minerals has been normalized to the surface area of the charge, based on the dimensions of the starting material. While the surface area of the charge is clearly greater at the end of the run, this is believed to be due in part to new mineral growth and in part to the fracturing of mineral grains during quench. Etch pits are seen on original grain surfaces, but many surfaces show no evidence of dissolution and are inferred to be newly fractured. On this basis, the average grain is estimated to have a total surface area of 2.4 · 10)7 m2 and a weight of 2.24 · 10)5 g (0.0224 mg). These estimates have been used to calculate the rate of consumption of water through hydration reactions in g sec)1 m)2 from the weight of the mineral powder in the charge and the change in the weight of free water. Results are shown in Fig. 5 and are seen to be closely comparable for all temperatures and run durations investigated. While two experiments at low temperatures yield higher rates, we do not consider that these provide clear evidence for a temperature dependence of hydration rate, within experimental errors, and the rather uniform reaction rate over this temperature range is inferred to arise because DGr increases to a lower temperature, while the reaction mechanism does not require solid diffusion. The
Fig. 3. EDS spectra for amphiboles: (A) primary hornblende showing traces of K and Ti and a prominent Al peak; (B) secondary actinolite with a low level of Al and Ti, K below detection, capsule 17, 500C, 5 kbar.
hydration rates obtained here are about an order of magnitude slower than most of those measured for the hydration of K-feldspar + andalusite to muscovite + quartz by Schramke et al. (1986), which have been normalized to
238 B. W. D. YARDLEY et al.
Fig. 4. XRD traces of run products showing the peaks for secondary sheet silicates from runs at different temperatures. The small peak near 10 A˚ is from biotite in the starting material.
Fig. 5. Calculated rates of water consumption from Table 1, shown as a function of run time.
the surface area of andalusite only, as reaction at this surface controlled the overall reaction rate. The experimental hydration rates have been further used to estimate the rate at which water is consumed by hydration at fracture walls, following rupture of fresh rock. This estimate requires additional assumptions: fractures are treated as simple, parallel-sided features, containing a water film with a half width of 100 lm. The time taken for this film to be consumed by silicate hydration is illustrated in Fig. 6. The estimate of fracture width is based on the size of secondary fluid inclusions decorating microcracks in
Fig. 6. Estimates of the time required for a water film with a half thickness of 10)4 m, coating a fracture in host rock, to be consumed by hydration reactions. Granulite calculations are based on mean rates from Table 1. Pelite data are extracted from Schramke et al. (1986), assuming wall rocks contain 30% andalusite.
quartz, which are commonly 5–10 lm in diameter, and are assumed to form within the open crack, based on CL observations in numerous studies (e.g. Boiron et al. 1992) – thus we make the conservative assumption that a fracture through which water is introduced will be at least an order of magnitude wider than a microcrack in a mineral. The assumption of a simple planar fracture is clearly also conservative; faulted rocks develop damage zones around their cores, and these are the regions in which active retrogression takes place. Typically damage zones contain evidence of multiple fine microcracks, thereby maximizing the
Rates of retrograde metamorphism 239 interface between fluid and rock, with many fine cracks probably containing a thinner film of water than assumed here. Hence, we believe that, despite the uncertainties, the estimate of <100 years for the time taken to consume a water film after it is introduced into freshly fractured rock is unlikely to be an underestimate. Rates calculated for the hydration of pelitic fracture walls, based on the data of Schramke et al. (1986), are even faster than those we have measured for granulites (Fig. 6).
The implications of this study are most significant for relatively young faults and shear zones, which have experienced too little slip to have a completely hydrated fault core. It shows that water can only be present for rather short periods of time in such rocks and that it is likely to be the rate of introduction of water, rather than the rate of hydration reactions, that dominates the rates of retrograde alteration. As a result, rocks remain dry for long periods of their exhumation history.
Implications for fault zone rheology
DISCUSSION AND CONCLUSIONS
The significance of the relatively rapid rates of hydration we have documented is that they imply that water is present for only short periods of time (c. 100 years) in midand lower crustal crystalline basement rocks. Most of the time they must be dry or they would be completely hydrated. Such rocks effectively buffer the H2O activity to relatively low values, hence preserving their high-grade mineral assemblages over long periods of geological time (Yardley & Valley 1997). For the experimental conditions used here and in the quoted experiments of Schramke et al. (1986), the activity of water, were hydration to consume all the water leaving some unreacted granulite, would be at least one and a half orders of magnitude lower than the activity of water if aqueous fluid were present. In the case of water infiltration linked to seismic activity, the residence times we estimate for water in fractured high-grade basement rocks are shorter than typical repeat times of major earthquakes on many faults, which may be c. 500 years or more for an individual fault segment (e.g. Goldfinger et al. 2003). Thus, it seems likely that many crustal faults and shear zones exhibit large variations in strength as they evolve from being freshly fractured and water bearing to being dry, and their cyclical character is not simply dictated by the build-up and release of stress (Cowie 1998). Water remains present in the fracture zone as long as it is open to infiltration, and for a period of the order of 100 years after that. During this time, the fracture is at its weakest because fluidassisted deformation mechanisms are capable of operating. After the water is consumed, the presence of fine-grained hydrous minerals in the fracture zone almost certainly means that it is weaker than the surrounding unfractured rocks, but it is nevertheless stronger than the wet fracture and thus the strength of the fracture zone oscillates between separate episodes of fracturing and water infiltration, increasing as water is consumed by hydration reactions until another rupture occurs. Only when the core of the fault or shear zone is completely hydrated will its strength become uniform over time, as there is no longer an effective sink for water. At this stage, the rock is likely to be dominated by hydrous minerals and this too leads to permanent weakening.
The effect of water in lowering the strength of quartz and other silicate materials has been well documented in a number of studies (Chen et al. 2006; Chernak et al. 2009) and it seems likely that, where fault or shear zone rocks retain sufficient relict high-grade material to react and consume fluids that infiltrate in response to a seismic event, the strength of the deformation zone will vary, depending on whether or not fluid is present. Hence, there are alternative models for the rheology of deformation zones in the crust depending on the relative rates of hydration reactions and fluid infiltration. It is evident that there are four possible hydration states for fault zones and shear zones, provided that they are sufficiently deep (a few kilometres) for mineral growth to take place in response to fluid infiltration. Each will be associated with a different rheology, and with time individual faults and shears may evolve from one state to another. (1) The amount of faulting is limited and the rocks are little altered from their original mineralogy but contain fractures and small amounts of fine-grained secondary minerals. No water is present. (2) Rocks are similarly only partially altered but contain a pervasive fluid phase. (3) Extensive alteration has occurred and the rocks are dominated by fine-grained, hydrous secondary minerals, but free water is absent. (4) Extensively altered rocks also contain a free fluid phase. Of these possible fault rock environments, it is likely that the wet ones, 2 and 4 will be the weakest, but both 3 and 4 can be very weak if the secondary minerals are themselves of low strength. Thus, Moore & Rymer (2007) showed that the rheology of the creeping Parkfield section of the San Andreas fault can be accounted for if talc as well as serpentine is present in the fault gauge. Where weak secondary minerals do not dominate, as in the earlier stages of fault zone development, the presence or absence of water is likely to be the dominant influence on rock strength, with state 2 being much weaker than state 1. State 2 is, however, metastable, and will revert to state 1 rather rapidly as our experiments have shown. After extensive deformation, the core of a fault or shear zone can become completely retrogressed in which case there is no
240 B. W. D. YARDLEY et al. further sink to fix infiltrated fluids locally and state 4 may be the final stable state for the cores of large faults. Certainly most major shear zones lack relict high-grade assemblages in their cores, but the presence of partially hydrated rocks at the margins demonstrates that the adjacent less altered rocks are a permanent sink for water and will always be acting to soak up water from deep deformation zones. Our results demonstrate that free water reacts rapidly with crystalline rocks under mid-crustal conditions, and therefore the widespread persistence of high-grade crystalline rocks through geological time is the result of the dry nature of almost all of the deep crust, almost all of the time. Retrogression is an immediate (c. 100 year timescale) response to water infiltration and results in the complete consumption of the infiltrated water by hydration reactions, except in the fully hydrated cores of large fault or shear zones. The presence or absence of water is the major variable that can affect the strength of crystalline rocks in the midto lower crust (Yardley & Baumgartner 2007). Seismic faulting is a major influence on crustal fluid flow, and can lead to the infiltration of water into dry rock volumes. This results in the localization of strain as the wet rocks are weakened, including the propagation of strain into aseismic shear zones, but we show here that, provided that they retain higher grade minerals with the potential to be hydrated, the water weakening effect is a transient one and the water is rapidly consumed by hydration reactions. Thus, water does not just account for the localization of crustal strain in space, its reactivity also results in strain being localized over time.
ACKNOWLEDGEMENTS We are grateful to Lesley Neve and Eric Condliffe for assistance with characterizing the experimental run products.
REFERENCES Blanpied ML, Lockner DA, Byerlee JD (1995) Frictional slip of granite at hydrothermal conditions. Journal of Geophysical Research, 100, 13045–64. Boiron MC, Essarraj S, Sellier E, Cathelineau M, Lespinasse M, Poty B (1992) Identification of fluid inclusions in relation to their host microstructural domains in quartz by cathodoluminescence. Geochimica et Cosmochimica Acta, 56, 175–85. Brodsky EE, Roeloffs E, Woodcock D, Gall I, Manga M (2003) A mechanism for sustained groundwater pressure changes induced
by distant earthquakes. Journal of Geophysical Research, 108. Article No. 2390. Chen WP, Molnar P (1983) Focal depths of intracontinental and intraplate earthquakes and their implications for the thermal and mechanical properties of the lithosphere. Journal of Geophysical Research, 88, 4183–214. Chen S, Hiraga T, Kohlstedt DL (2006) Water weakening of clinopyroxene in the dislocation creep regime. Journal of Geophysical Research-Solid Earth, 111. Article No. B08203. Chernak LJ, Hirth G, Selverstone J, Tullis J (2009) Effects of aqueous and carbonic fluids on the dislocation creep strength of quartz. Journal of Geophysical Research-Solid Earth, 114. Article No. B04201. Cowie PA (1998) A healing-reloading feedback control on the growth rate of seismogenic faults. Journal of Structural Geology, 20, 1075–87. Frost BR, Bucher K (1996) Is water responsible for geophysical anomalies in the deep continental crust – a petrological perspective. Tectonophysics, 231, 293–309. Fusseis F, Handy MR (2008) Micromechanisms of shear zone propagation at the brittle-viscous transition. Journal of Structural Geology, 30, 1242–53. Goldfinger C, Nelson CH, Johnson JE (2003) Holocene earthquake records from the Cascadia subduction zone and northern San Andreas fault based on precise dating of offshore turbidites. Annual Reviews of Earth and Planetary Sciences, 31, 555–77. Menegon L, Pennacchioni G, Heilbronner R, Pittarello L (2008) Evolution of quartz microstructure and c-axis crystallographic preferred orientation within ductilely deformed granitoids (Arolla unit, Western Alps). Journal of Structural Geology, 30, 1332– 47. Moore DE, Rymer MJ (2007) Talc-bearing serpentinite and the creeping section of the San Andreas fault. Nature, 448, 795–7. Muir-Wood R, King GCP (1993) Hydrologic signatures of earthquake strain. Journal of Geophysical Research, 98, 22035– 68. Schramke JA, Kerrick DM, Lasaga AC (1986) The reaction muscovite + quartz = andalusite + K-feldspar + water; Part 1, Growth kinetics and mechanism. American Journal of Science, 287, 517–59. Sibson RH, Robert F, Poulsen KH (1988) High angle reverse faults, fluid-pressure cycling, and mesothermal gold-quartz deposits. Geology, 16, 551–5. Stunitz H, Fitzgerald JD (1993) Deformation of granitoids at low metamorphic grades. 2. Granular flow in albite-rich mylonites. Tectonophysics, 221, 299–324. Turner FJ (1981) Metamorphic Petrology, 2nd edn. McGraw Hill, New York, 524 pp. Yardley BWD, Baumgartner LP (2007) Fluid processes in deep crustal fault zones. In: Tectonic Faults: Agents of Change on a Dynamic Earth (eds Handy MR, Hirth G, Hovius N), pp. 295– 318. MIT Press, Cambridge, MA. Yardley BWD, Valley JW (1997) The petrologic case for a dry lower crust. Journal of Geophysical Research, 102, 12173–85.
Fluids in the upper continental crust KURT BUCHER AND INGRID STOBER Institute of Geosciences, Geochemistry, University of Freiburg, Freiburg, Germany
ABSTRACT The brittle upper continental crust predominantly consists of granite and gneiss. Fractures form an interconnected network of water-conducting structures with an appreciable permeability also providing substantial fluid-saturated fracture porosity. The chemical composition of fluids in the fracture porosity of granite and gneiss changes with depth. Near the surface Ca–Na–HCO3 waters dominate. With increasing depth, water contains increasing amounts of alkalis and sulfate and grade into chloride-rich waters at greater depth. Total dissolved solids (TDS) of 105 mg l)1 are common at 5-km depth in most basement rocks. All reported deep fluids from the upper crust contain predominantly NaCl and CaCl2. The brines vary from NaCl-rich in granites to CaCl2-rich in mafic reservoir rocks such as amphibolites and gabbros. In regions of the crust with strong topography, fluid flow is important and recharge water may have flushed the basement efficiently, thereby removing old brine components from the granites. Water samples from the new Gotthard Rail Base Tunnel of the Alps represent this type of basement fluid. Analyzed fluids from up to 2.5-km depth differ from basement fluids from areas with less-extreme topography in the following ways. Such waters have relatively low TDS of some 100 mg l)1 and are typically of the Na2CO3–Na2SO4 type. pH tends to be high and varies from 9 to more than 10. Low Ca and ultra-low Mg of such waters result from efficient deposition of secondary Ca–Mg minerals as coatings on fracture walls. Reduction of CO2 to CH4 provides the oxidation capacity for sulfate production from primary rock sulfides. The composition of fluids in fractured continental crust at depths below 1–2 km depends strongly on the topography of the erosion surface. In crust with rugged alpine topography fluids at this depth are low-TDS high-pH waters that derive its composition from fluid–rock interaction alone. In crust with low-to-moderate topography, basement fluids are normally near neutral high-TDS Na–Ca chloride brines that derive the solutes not only from the rock matrix but also from external sources. Key words: brine, crustal fluid, soda, thenardite, water–rock interaction, zeolite Received 2 July 2009, accepted 25 January 2010 Corresponding author: Kurt Bucher, Institute of Geosciences, Geochemistry, University of Freiburg, Albertstr. 23b, D-79104 Freiburg, Germany. Email:
[email protected]. Tel: 49 761 203 6395. Fax: 49 761 203 6407. Geofluids (2010) 10, 241–253
INTRODUCTION The fracture pore space of continental basement rocks is normally filled with an aqueous fluid. Depending on the fracture frequency, characteristic porosity is on the order of about 0.01 (1 vol.%). The fractures normally form an interconnected framework of open permeable space and fluid movement follows the Darcy flow law (Manning & Ingebritsen 1999). The hydraulic conductivity of the fracture network is on the order of 10)9–10)7 m sec)1 (Stober & Bucher 2007) corresponding to permeability in the range of 5 · 10)17 to 3 · 10)15 m2. Significant advective fluid flow is possible if appropriate geological driving forces are
present (Ingebritsen & Manning 1999). In the absence of magmatic heat sources, the prime reason for advective fluid flow in the continental crust are hydraulic heads created by topography. Fracture-related permeability decreases with depth (Ingebritsen & Manning 1999). Interconnected flow porosity vanishes within the brittle–ductile transition (BDT) zone, which is at about 12-km depth depending on temperature, lithology and deformation rate (Wintsch et al. 1995). Deeper in the crust, fluid transport is by other mechanisms than Darcy flow (Frost & Bucher 1994). Above the BDT zone, fractured crystalline basement behaves hydraulically like a homogeneous infinite aquifer (Stober & Bucher 2005). The BDT is normally at
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
242 K. BUCHER & I. STOBER temperatures below the critical point of H2O. Thus, the fluid in the upper continental crust is an aqueous electrolyte solution or simply water (specifically groundwater). Although research drill holes to 9- and 12-km depth exist, uncontaminated water compositions from the basement are not known for depths about >5 km (Pauwels et al. 1993). The crystalline basement of the continents predominantly consists of granites and gneisses. Thus, crustal water is mostly in contact with K-feldspar (Kfs), Na-rich plagioclase (Ab), quartz (Qtz), a small modal amount of biotite (Bt) and various accessory minerals. The fractures are often coated with minerals that formed from alteration reactions of water with the primary minerals of granite and gneiss. Alteration minerals include clays (chlorite, Chl), zeolites, carbonates, sulfates, oxides and occasionally also sulfides (Stalder et al. 1980). The fluid phase is typically not in equilibrium with plagioclase and Bt (and other Fe–Mg silicates) and close to saturation with Kfs and Qtz. With respect to secondary alteration minerals, the wide range of different waters are supersaturated with respect to a distinct assemblage of product phases of the water–rock interaction (WRI) process and undersaturated with respect to other alternative product assemblages. Thus, characteristic alteration assemblages are present in the reaction zone along the fractures reflecting also the distinct nature of water composition that result from the local WRI (Grimaud et al. 1990; Pauwels et al. 1993; Aquilina et al. 1997a,b). Subordinate, although locally important, are mafic rocks such as gabbro and diorite and their metamorphic equivalents, greenschist and amphibolites. Water residing in the fractures of mafic crustal rocks are in contact with Ca-rich plagioclase and a high proportion of pyroxene and ⁄ or amphibole. Alteration produces zeolites and carbonates (Borchardt & Emmermann 1993). The fluids in mafic rocks are distinctly different from ‘normal’ granite and gneiss waters (Stober & Bucher 2005). Other lithologies occurring in continental basement include marbles (calcareous), micaschists and ultramafic rocks. Waters in contact with these less-important lithologies have very distinct compositions and differ markedly from the regular granite and gneiss waters. Continental basement is exposed directly at the surface in the vast areas of the Precambrian shields, locally covered only by Quaternary sediments. In other continental areas, the basement is covered with sedimentary rocks that may reach several kilometers in thickness. The young Hercynian ⁄ Variscan continental crust of central Europe, as an example, crops out at the surface in some erosional windows such as the Black Forest, Vosges, Massive Central, Brittany, Bavarian and Bohemian Forest but is covered with Mesozoic and Tertiary rocks of a thickness ranging from a few meters to 5 km, thus leaving only 7-km basement rock above the BDT zone.
Water occurring in the fractures of brittle continental crust may acquire its observed solute load by a large variety of different processes that are controlled by temperature (depth), lithology of the host rock, advective flow caused by topography, by the presence and the thickness of (former) cover sediments, and by the depth of the water reservoir. Origin and composition of recharge waters may also play an important role in the evolution of deep waters. It is difficult to separate the many different contributions to the composition of basement water. In this study, we give a brief overview on the various types of water found in fractured continental basement rocks and present a compilation of typical water composition data from different depths and geological settings. In addition to the compiled example data, we report new water composition data from water on fractures in granite and gneisses of the new Gotthard rail base tunnel of the central Alps. There the topography is steep and rugged and up to 2.5 km of fractured crystalline basement rocks cover the sampling locations in the tunnel. From this geological situation, it follows that the composition of tunnel waters is predominantly controlled by the chemical interaction of meteoric recharge with primary and secondary minerals of the water-conducting fractures. Thus, the tunnel waters represent near endmember mineral dissolution–precipitation fluids. The derived water–rock reactions permit a better separation of WRI from other contributions to the solute load of deep waters of the continental crust.
WATER IN THE CONTINENTAL UPPER CRUST: VARIABILITY AND COMMON FEATURES Water in the crystalline basement has a wide range of compositions (Table 1) depending on the circumstances summarized above (see also an excellent review by Frape et al. 2004). The total dissolved solids (TDS) vary from a few mg l)1 to more than 3 · 105 mg l)1, a range of 5 orders of magnitude. Groundwater at shallow depth in granite is normally of the Ca–Na–HCO3 type and does not differ from other low-mineralized, near-surface waters. Its chemistry is controlled by the solute content of precipitation (Hardie & Eugster 1970) and processes in the soil zone (White et al. 2005a). With increasing depth, mineralization increases and sulfate becomes an increasingly important anion. Consequently, the waters gradually change to Ca–Na–HCO3– SO4 type (Table 1). This is the characteristic of the depth zone of hundreds of meters below surface. Sulfate is produced from oxidation of sulfide minerals that are accessorily present in basement rocks, mostly pyrite and pyrrhotite. The oxidation consumes oxygen dissolved in the recharge and is commonly biomediated (Vuorinen et al. 1981). Acidity produced by the sulfate reactions is consumed by
Fluids in the upper continental crust 243 Table 1 Selected chemical compositions of water (groundwater, deep water) from fractured crystalline basement. Sample
1
Depth (m) T (C) pH Ca Mg Na K Sr Li Fe Al Mn As HCO3 + CO3 SO4 Cl F Br J NO3 SiO2 HBO2 CO2 TDS Cl ⁄ Br Sample
13
2 27 10.4 7.2 18.84 1.34 7.13 1.06
61 6.28 3.55
3
907
40 15.0 6.25 24.0 11.8 25.0 2.9 0.1 0.03 0.15 0.013 0.61 0.005 58 62 32.2 2.75 0.16
8.9 8.0 1.3 23.0 2.7
76 0.08 14 0.1
7.63 13.85 26 125
230 206 15
5 80 11.5 5.9 389.5 54.4 146.5 7.7
8.17 b.d. 2.00 1529.7 264.6 19.9 1.8
b.d. 17.14 0.14
5.00
130 140 14
4
44.71 3220 2487
16
120 16.0 5.7 198.4 42.6 204.2 17.7 1.03 1.2 3.07 0.53 0.0085 1021 253.5 32.5 2.5 0.4 0.01 0.14 85.11 1.9 3207 1893 81 17
18
6
7
8
650 10.7 6.67 141.0 23.3 7.61 10.4 0.62 b.d. 0.05 b.d.
492
600 28.3 6.6 328.0 27.7 2158 149 18.3 26.7 1.80
470 39.2 8.01 13.2 0.3 298 7.3 0.48 1.4
0.80
0.04
519 353 3705 3.3 22.5 0.22
268 265 132 9.9 0.79
16.93 26.8 418 7365 164
25.86
9.9 89.0 0.4 224.0 0.7
b.d. 66.51 400 3.01 11.8 0.04
18 55 460 5.1
4.71 25.20 0.08
15.00
700 75
867 90 19
9
20
Depth (m) 65 4444 2535 5000 4000 0 1500 1600 T (C) 14 25 115 0.7 18 22 pH 6.50 6.78 5.45 7.84 6.30 5.00 6.55 Ca 593.7 1245 11 700 6600 16 000 199.6 64 000 63 800 Mg 34.4 159 1900 98 2.96 141 5100 78 Na 2891 24 400 63 900 27 400 6382 2870 45 000 18 900 K 252.6 896 503 2860 215 86.8 199 122 Sr 20.5 103 485 400 260 1080 1580 Li 15.4 44.17 41.2 125 6.13 0.07 0.81 Fe 1.83 21.20 36 96 33.6 2.07 Al 1.26 0.02 0.03 1.60 Mn 0.06 6.70 15.4 0.32 4.57 As 3.0 6.5 HCO3 500.4 311.2 31.12 3050 19 58 SO4 384.6 790 1525 159 320 3173 284 223 Cl 5357 38 364 120 500 59 000 38 733 1121 207 000 162 700 F 1.03 31.00 4.70 4.30 0.07 0.05 7.00 Br 18.36 460 726 223 530 1760 1250 J 0.15 2.51 7.00 NO3 2.98 b.d. 12.53 24.08 175.00 79.70 86.40 10.29 20.57 SiO2 HBO2 11.88 90.83 13.40 2508.00 25.13 CO2 264 TDS 10 116 66 971 201 348 97 615 62 629 13 270 324 500 249 100 Cl ⁄ Br 291 83 165 264 73 117 130
10
11
12
200 38.7 7.31 34.3 2.1 142 9.1
660 11.3 7.48 468.0 57.7 1140 43.7 12.2 3.31 2.83 b.d. 0.02 0.026 336 3340 155 1.65 1.1 0.20 2.81 28.30
b.d. 1128 167
b.d. 0.068 221.3 34.4 136 1.5 0.41 0.10 1.70 60.97 0.37 13.2 663 331
603 24.4 8.4 13.1 0.3 205 3.0 0.5 0.69 0.03 0.02 0.03 0.036 163 240 53 10.8 0.43 0.01 0.40 15.70 1.3 6 712 123
21
22
23
0.65 b.d.
1372 289 23 4 13 6.60 8.30 7.60 12 900 10 200 390 268 36 92 10 900 8300 1200 111 19 8.4 671 100 2.4 0.78 9.16 0.20 0.15 4.70 0.00 0.03 3.04 0.51 0.32
26.4 5592 140 24 1090 6.10 1343 2180 33 529 530
28 7.2 95.2 35 746 5.1 320 1689 37 100 30 600 2630 62 050 5.74 1.20 0.41 358 204 9.2 1.01 0.35 17.14 5.80 12.00 6.49 2.84 3.24 0.00 63 133 49 484 4761 101 372 103 150 285
(1) Bleibach, Black Forest BF; (2) Lavia, Finland, Scandinavian Shield (Lahermo & Lampe´n 1987); (3) Pendarve P2, Carnmenellis Granite, Cornwall (Edmunds et al. 1987); (4) Rippoldsau, BF; (5) Bad Peterstal, BF; (6) Clara Mine BF, Level 9.0; (7) Stripa V2, (Lahermo & Lampe´n 1987); (8) Bad Sa¨ckingen, BF; (9) Zurzach, BF; (10) Wildbad BF; (11) Waldshut-Tiengen, BF; (12) Waldkirch BF; (13) Ohlsbach, Black Forest BF (Stober et al. 1999); (14) Urach, BF (Stober & Bucher 2004); (15) Bu¨hl well, basement Rhine Rift valley; (16) Soultz-sous-Forets, geothermal well, basement Rhine Rift valley (Pauwels et al. 1993); (17) KTB, borehole continental deep drilling, Bohemian Massif (BM) (Stober & Bucher 2005); (18) Marianske Lazne, BM (Paces 1987); (19) Thompson 4000–4, Canadian Shield (CS) (Frape & Fritz 1987); (20) Sudbury, CS, N3646A (Frape & Fritz 1987); (21) Yellowknife, CS, 4500-3 (Frape & Fritz 1987); (22) Olkiluoto, Scandinavian Shield (SS), KR12 ⁄ 741 ⁄ 1 (Pitka¨nen et al. 2004); (23) Olkiluoto, SS, KR7 ⁄ 282 ⁄ 1 (Pitka¨nen et al. 2004); (23) Olkiluoto, SS, KR7 ⁄ 282 ⁄ 1 (Pitka¨nen et al. 2004); (24) Rodina Mine, Ukrainian Shield (Vovk 1987). All concentrations in mg l)1. b.d., below detection limit; no entry, not analyzed; TDS, total dissolved solids.
244 K. BUCHER & I. STOBER carbonate dissolution. Calcite and related carbonates may be present as secondary alteration minerals in granites and as coatings on fractures (White et al. 2005b). The consequence is that TDS of the waters increases and that pH is buffered to values ranging from 6 to 8. If waters are CO2 saturated, pH is low (approximately 6) and the waters are typically used as mineral waters in spas and as a high-quality bottled drinking water (Table 1). CO2 originates mostly from carbonate dissolution even in quartzo–feldspathic rocks (Bucher et al. 2009). In some areas, there is ambiguous evidence for deep origin of CO2 (lower crust, mantle, igneous, metamorphic; e.g. Evans et al. 2005). If CO2 is low, water pH increases and approaches 8 if CO2 is close to equilibrium with the atmosphere (Table 1; Stober & Bucher 1999). With further increase in depth, chloride and sodium rise continuously. The amount of NaCl component in basement waters may reach halite saturation whereas pH remains about neutral (Table 1). Waters at several kilometers depth are typically saline brines with NaCl dominating the TDS. The basement brines are the characteristic fluids of continental crust (Frape & Fritz 1987). A number of sources and processes have been proposed and identified for the high salinity of basement brines (e.g. Edmunds et al. 1985; Gascoyne et al. 1987; Beaucaire et al. 1999; Banks et al. 2000; Gascoyne 2004). The chlorinity has probably a common global source because it is characteristically present at depth in all continental basement units worldwide. The increase in TDS with depth is a general trend but the relationship is not simple. In the upper hundreds of meters in the basement, there is little systematic TDS increase visible. Data compilations from the Black Forest basement (Stober & Bucher 1999) and the Canadian Shield (Fig. 1) (Frape & Fritz 1987) show variable but generally low TDS in the uppermost part of the crust and a significant increase of highly mineralized waters at about 1-km depth. Very high TDS waters occur side by side with relatively low TDS waters at depth from 1–1.8 km (Fig. 1). The Canadian Shield waters did not homogenize, although they may have resided for a very long time in the basement rocks. This may indicate that some waters are stagnant and little advection and mixing occurs. The Cl ⁄ Br mass ratio of the Canadian Shield waters does not vary with depth (Fig. 2). All reported waters, including the weakly mineralized shallow groundwaters, have Cl ⁄ Br ratios of 93 ± 31. Thus, seawater can be excluded as the sole source of salinity. The salinity is clearly not derived from evaporitic halite dissolution. In the Black Forest basement, two distinct water types reside in the fracture porosity of granite and gneiss (Fig. 3). The Na–Cl–SO4 type thermal waters and the Ca–Na–HCO3 type mineral waters occur in well-separated small districts but relatively close to each other indicating
Fig. 1. Total dissolved solids (TDS) in g kg)1 versus depth of sampling location of water samples from the Canadian Shield (Frape & Fritz 1987; Table 1).
Fig. 2. Chloride ⁄ bromide mass ratio versus depth for Canadian Shield waters (Frape & Fritz 1987; Table 1).
that hydraulic communication is limited and little mixing occurs, although the waters are younger than those of the Canadian Shield and despite that there is marked topography related advective flow in the Black Forest. Deep drilling into the central European basement has shown that waters at 4- to 5-km depth are brines with TDS of 70–120 g l)1 (Table 1). The brines have low Cl ⁄ Br mass ratios (<100) with the exception of the Soultzsous-Foreˆts geothermal well in granites of the Rhine rift valley, which has Cl ⁄ Br of 270 corresponding to that of seawater (Pauwels et al. 1993). Most brines are NaCl
Fluids in the upper continental crust 245
Fig. 3. Composition of waters from the Variscan crystalline basement of the Black Forest. (A) lowpH CO2-rich mineral waters and (B) NaCl-rich thermal waters.
solutions, the brine from the 4-km deep pilot hole of the Continental Deep Drilling site (KTB) in Germany, however, contains more CaCl2 than NaCl (Mo¨ller et al. 1997; Stober & Bucher 2005). This can be related to the dominance of mafic metamorphic rocks, mostly amphibolites, in the crust forming the fluid reservoir. The amphibolites at the KTB site contain a relatively Ca-rich plagioclase (An28) (Patzak 1991), which reacts with dissolved NaCl to albitize the plagioclase and producing the observed CaCl2 brine (Fig. 4). Canadian Shield brines are also unusually rich in CaCl2 component, which can be explained by reaction of NaCl solution with the mafic, partly gabbroic rock matrix (Fig. 4). The large compositional diversity of basement fluids (Table 1) results from a combination of many different individual contributions that are difficult to separate. For
example, the high total salinity of the waters in Precambrian shields has been related to fossil basinal fluids still residing in the basement after removal of the cover sediments (Gascoyne et al. 1987; Bottomley et al. 1999; Frape et al. 2004). The total mineralization may have been further increased by repeated freezing during cold periods (Bottomley et al. 1999; Frape et al. 2004). However, the high Ca ⁄ Na ratio of many of these fluids may also relate to albitization of plagioclase of mafic host rocks (Fig. 4). Albitization is a common process in low-grade mafic rocks and represents an example of fluid–rock interaction in upper crustal rocks. Thus, anions may have an external origin (basinal brine) whereas at least some of the cations may originate from internal WRI (albitization). The example shows that a given water composition can be the result of many processes and that it is rarely possible to resolve the origin of solutes unequivocally. There is a principal ambiguity about the combination of processes by which crustal water has acquired its chemical composition. In the following, we present new water data from the crystalline basement of the Gotthard area of Alps that may define an endmember case for waters that have acquired the solutes almost exclusively by fluid–rock interaction.
FLUIDS FROM THE GOTTHARD RAIL BASE TUNNEL
Fig. 4. Roozeboom diagram showing the metastable low-temperature distribution of Na and Ca between fluid and plagioclase. Calibration points are Black Forest water data, KTB deep drilling fluids (Stober & Bucher 2005) and unpublished gabbro water data from Seiland, Northern Norway.
Water samples are systematically collected from water-conducting fractures in the Gotthard rail base tunnel drilled through the Central Alps (for details see: http://www.alp transit.ch). The tunnel is more than 57 km long and cuts across crystalline basement rocks representing the upper continental crust. These waters mainly contain solutes derived from reaction of the water with the rock matrix exposed on the fractures along the flow paths. The overburden at the sampling points is up to 2500 m and the maximum temperature is close to 50C. The water samples presented here have been collected in the 11.35-km long Amsteg section of the tunnel (Fig. 5). The area is characterized by a very rugged topography leading to gravity
246 K. BUCHER & I. STOBER
Fig. 5. Geological cross-section of the Gotthard rail base tunnel (Amsteg section) showing the sampling locations.
controlled advective fluid flow. Precipitation in the high mountains with very little vegetation and soil enters the fracture system and migrates to the sampling locations at the opened fractures during tunnel construction. Along the flow path, the water reacts with the minerals of the gneiss and granite exposed on the fracture walls. From this configuration, it follows that the water composition essentially reflects the WRI component in a rather pure form.
ANALYTICAL METHODS: WATER COMPOSITION DATA Water samples for chemical analyses were collected in 500ml acid-cleaned polyethylene bottles. Care was taken to fill the bottle completely to avoid gas exchange. Measured electrical conductivities confirm that no precipitates formed prior to chemical analysis in the laboratory. Temperature, pH and electrical conductivity were measured at the sampling site. After filtration through cellulose acetate membrane filters with a 45-lm pore size, the chemical composition of the water was determined using ion chromatography for the anions, atomic absorption spectroscopy for cations and photometry for silicon and boron. Carbonate and bicarbonate has been titrated with 0.1 N HCl. The data are presented in Tables 2 and 3.
GRANITE, GNEISS AND FISSURES The water samples reported here are from a 1.6-km long section of the southern part of the Amsteg section that includes two lithological units (Fig. 5), the Southern Aar Granite and the Southern gneisses (from 16.195 to 17.844 km). The units dip nearly vertically and the fractures are also steeply oriented and roughly parallel to the alpine gneiss foliation. The Southern Aar Granite predominantly consists of plagioclase, Kfs and Qtz. It also contains Bt and accessory apatite. Plagioclase is pure albite. The rock contains Chl and hematite from Bt alteration but no or very little sulfide (pyrite). The gneisses contain the same minerals as the
granite; however, plagioclase is slightly calcic. The main difference is the presence of a significant amount of sulfide in the gneiss. Fractures and fissures of both granite and gneiss are often coated with fissure minerals. The fissure minerals include: Qtz, adularia, Chl, laumontite, stilbite (Stb), apophyllite, occasionally also calcite and pyrite.
HYDROCHEMISTRY OF THE GOTTHARD TUNNEL WATERS The composition of the Gotthard tunnel waters is markedly different for granite (Table 2) and for gneiss (Table 3). Samples 5–12 (Table 2) are compositionally similar granite waters (Fig. 6A) that were collected from fractures along a 400-m long stretch in the tunnel. Each analysis represents one water-conducting fracture. Typical flow rates are on the order of 1–5 l min)1. Granite water analyses 1 through 4 (Table 2) resemble rather gneiss waters and may have been produced by an unidentified gneiss slab within the granite (Fig. 6A). The pH of granite waters varies from 9.5 to 10.2. Forty percent of the waters have pH above 10. This unusual feature of granite waters (compare with data compilation, Table 1) is coupled with rather low TDS of about 140 mg l)1. The granite waters are dominated by sodium, silica, carbonate and sulfate, and accordingly, can be viewed as high-pH soda waters (Fig. 6A). There is also a significant amount of chloride present. There is a remarkable lack of magnesium in all waters with Mg often below its detection limit of 0.04 mg l)1. Calcium is very low (3 mg l)1). On the contrary, the average SiO2 is around 32 mg l)1 and together with Na the dominant solute in the waters. Fluorine is relatively high (>4 mg l)1). The striking feature of all granite waters is their high pH. The gneiss waters (Table 3) also have generally a high pH, but pH is distinctly lower than in granite waters. TDS varies over a wide range (472–2389 mg l)1) and is much higher than that of granite waters. TDS variation is caused by variable amounts of Ca-sulfate, alkali variation is small. The gneiss waters are dominated by calcium, sodium and sulfate, and thus can be viewed as Ca-sulfate waters
206.00 7.50 1.99 0.08 28.44 0.24 363 93.75
184.00 7.94 1.61 0.09
32.32 0.29 331 88.22
60.90 8.90 2.70 0.09 b.d. 26.44 0.35 164 98.89
116 263 41.50 10.12 10.10 b.d. 39.40 3.72 0.12 b.d. 0.07 5.78 10.80
3
28.00 0.23 274 155.71
142.00 10.90 2.00 0.07
116 621 41.00 10.11 29.60 b.d. 45.10 9.19 0.31 b.d. 0.23 6.46 6.00
4
33.68 0.28 131 87.50
27.10 7.00 2.83 0.08
116 409 32.00 9.79 3.99 0.10 31.10 2.06 0.06 0.20 0.29 1.87 22.21
5
20.80 8.42 3.41 0.09 0.48 35.38 0.24 124 93.56
116 440 41.00 10.09 3.25 b.d. 34.40 2.55 0.06 b.d. 0.08 6.46 15.01
6
27.50 7.70 3.66 0.09 b.d. 35.26 0.26 127 85.56
116 460 37.00 10.23 3.69 b.d. 34.00 0.70 0.04 b.d. 0.10 7.82 14.40
7
42.30 14.30 5.74 0.24 b.d. 23.38 0.46 162 59.58
116 590 42.50 9.82 3.88 0.03 47.90 0.83 0.07 b.d. b.d. 0.68 22.81
8 116 687 42.00 9.72 3.67 b.d. 41.50 4.83 0.32 b.d. b.d. 30.61 3.66 28.50 12.30 3.80 0.19 0.10 35.57 0.33 165 64.74
25.21 1.83 20.30 12.00 4.20 0.15 0.12 37.07 0.51 143 80.00
10
116 612 42.00 10.02 2.90 0.03 33.00 5.09 0.06 0.08 0.15
9
25.21 4.88 20.80 8.50 5.80 0.11 b.d. 24.25 0.47 133 77.27
116 697 40.00 9.60 2.57 0.03 38.20 1.69 0.05 b.d. 0.23
11
9.60 25.02 22.3 15.10 4.30 0.14 b.d. 38.32 0.34 161 107.86
116 811 33.00 9.52 1.03 b.d. 37.50 6.71 0.08 0.02 0.19
12
32.86 0.36 143.18 82.01
3.12 0.03 37.20 3.06 0.09 0.02 0.13 2.10 20.63 4.42 26.20 10.67 4.22 0.14
5–12
Avg
5.75 0.10 17.05 15.45
0.98 0.03 5.44 2.21 0.09 0.02 0.11 3.19 6.98 8.54 7.32 2.15 1.06 0.06
SD
TM, tunnel meter; 116 195 = east tube 16 195 m from entrance in Erstfeld, pH at 25C, concentrations in mg l)1; TDS, total dissolved solids; b.d., below detection limit (Mg 0.04, Fe 0.02, Al 0.02, S2 0.04).
116 205 43.60 9.57 45.00 b.d. 61.70 2.70 1.06 b.d. 0.22 4.37 7.68
116 195 43.30 9.69 36.33 b.d. 58.90 1.88 0.96 0.04 0.13 5.03 6.42
TM T (C) pH Ca Mg Na K Sr Fe Al OH CO3 HCO3 SO4 Cl F Br Sulfide (S) SiO2 HBO2 TDS Cl ⁄ Br
2
1
Sample
Table 2 Analyses of fracture waters from the Southern Aar Granite (NEAT tunnel, Central Alps).
Fluids in the upper continental crust 247
248 K. BUCHER & I. STOBER Table 3 Analyses of fracture waters from the Southern Gneiss Unit (NEAT tunnel, Central Alps). Sample
13
14
15
16
17
18
19
TM T (C) pH Ca Mg Na K Sr Fe Al OH CO3 HCO3 SO4 Cl F Br Sulfide (S) SiO2 HBO2 TDS Cl ⁄ Br
116 900 34.00 7.92 70.20 0.03 67.10 4.68 1.15 0.04 0.08
117 505 36.00 9.03 574.00 0.02 115.00 7.74 1.97 b.d. 0.04 0.51 6.60 0.00 1578.0 73.70 3.94 0.81
117 691 40.00 8.75 95.90 0.05 103.00 20.25 0.70 b.d. 0.10 0.17 9.60 0.00 382.0 57.90 4.27 0.95 26.69 0.15 702 60.95
117 705 41.80 9.32 271.00 0.04 106.00 13.11 1.51 b.d. 0.09 2.21 5.40 0.00 812.0 78.40 4.13 0.88 0.15 25.44 0.59 1319 89.09
117 800 39.00 9.25 66.30 b.d. 93.90 52.90 0.49 b.d. 0.13 0.34 9.00 0.00 292.0 86.10 4.31 0.70 0.11 27.32 0.10 633 123.00
117 844
26.25 0.37 2389 90.99
117 700 37.80 9.41 65.80 b.d. 90.70 1.21 0.49 b.d. 0.10 2.38 5.40 0.00 291.0 33.20 4.07 0.47 0.16 27.07 0.09 520 70.64
0.00 25.02 233.0 36.10 5.62 0.39 b.d. 28.13 0.47 472 92.56
9.72 62.60 b.d. 97.30 1.50 0.32 b.d. 0.12 3.06 7.20 0.00 288.0 31.80 4.19 0.46 0.58 29.50 0.10 523 69.13
Avg
SD
172.26 0.02 96.14 14.48 0.95 0.01 0.11 1.24 6.17 3.57 553.71 56.74 4.36 0.67
192.34 0.02 15.16 18.25 0.62 0.02 0.03 1.26 3.17 9.46 492.67 23.17 0.57 0.23
27.20 0.28 937 85.19
1.32 0.21 703 20.78
TM, tunnel meter; 116 195 = east tube 16 195 m from entrance in Erstfeld, pH at 25C, concentrations in mg l)1; TDS, total dissolved solids; b.d., below detection limit (Mg 0.01, Fe 0.02, Al 0.02, S2 0.04).
(A)
(B)
Fig. 6. Composition of waters from the Gotthard rail base tunnel (Amsteg section): (A) water samples from the Southern Aar Granite; (B) waters from the gneisses.
(Fig. 6B). Because of the lower pH of gneiss waters, silica (SiO2 27 mg l)1) is slightly lower than in granite waters. Fluorine (4.3 mg l)1) is similar to granite waters. Chloride is distinctly higher than carbonate in gneiss waters. Mg is very low.
DISCUSSION The saturation state of the waters with respect to a number of solids has been computed using the code PHREEQC and the LLNL database (Parkhurst & Appelo 1999). As a representative example of the granite waters 5 tp 12 (Table 2), the saturation state of water sample 7 is graphically shown on Fig. 7. The first four analyses of Table 2 resemble
gneiss waters (Table 3) and contain distinctly more Ca and SO4. These waters are close to anhydrite saturation. Type granite water 7 (Table 2) is undersaturated with feldspar components albite and anorthite. It is saturated with calcite and Qtz and strongly undersaturated with clays and other sheet silicates with the exception of Chl. The water is also oversaturated with respect to Ca-zeolites, particularly Stb. The degree of supersaturation is very strong for Chl and Stb. These two phases have been frequently observed as mineral coatings on the walls of water-conducting fractures. It is probable that both minerals are produced by the irreversible mass transfer from granite to the fissure assemblage. The general process of water–rock interaction can be described by a generic reaction of the type:
Fluids in the upper continental crust 249 reduce SI of the two minerals and slows down the product formation. However, this will increase the degree of undersaturation with respect to the reactant minerals and accelerate dissolution reactions. This precipitation–dissolution feedback expressed by reaction (1) probably proceeds at near steady-state conditions and produces the observed granite water(2) (Zhu 2009). The sources of solutes can be identified as follows: Albite is the main source for Na, Bt is source of potassium, magnesium and iron, Kfs can be a source or sink for K, the main source for sulfate in granite is sulfide, and chloride originates from Bt and from fluid inclusions, which is also a minor source of additional Na. Finally, Ca and carbonate formed from dissolution of secondary and fissure calcite and from CO2 in the recharge water. All waters contain sulfur mostly in the form of sulfate, whereas sulfide (H2S) is only a minor component (Tables 2 and 3). Thus iron released by Bt and pyrite alteration is deposited as hematite and not taken up by the water: 2FeS2 þ 19H2 O ¼ 8Hþ þ 4SO4 2 þ 1Fe2 O3 þ 15H2
ð2Þ
The oxidation power for the pyrite to sulfate and hematite reaction is probably provided by carbon dioxide: CO2 þ 4H2 ¼ CH4 þ 2H2 O
ð3Þ
and the overall pyrite oxidation is accomplished in the fracture system by the combination of the reactions (2) and (3): Fig. 7. Saturation state of water from the Southern Aar Granite (PHREEQC; Parkhurst & Appelo 1999).
8FeS2 þ 46H2 O þ 15CO2 ¼32Hþ þ 16SO4 2 þ 4Fe2 O3 þ 15CH4 ð4Þ
albite þ Bt þ waterð1Þ ¼ Chl þ Stb Qtz calcite Kfs þ waterð2Þ
ð1Þ
During WRI, the mineral assemblage of the granite on the left side reacts with low-TDS recharge water(1) and produces the minerals found as fissure coatings and the final water(2) sampled in the tunnel. Chl and Stb are both present as coatings on fissure walls partially covering and overgrowing older fissure minerals that have been originally deposited by earlier and hotter fluids. Chl and Stb are clearly late formations produced by the modern waters. The computed saturation index (SI) for Chl is 4.7 and that of Stb is more than 6 corresponding to a degree of oversaturation by a factor of 5 · 104 for Chl and 106 for Stb. At estimated values for critical saturation (SIc) of 3 for Chl and 4.5 for Stb, where kinetic barriers are overcome and precipitation starts, the waters are oversaturated by a factor of about 30–50 with respect to the two main product minerals. This driving force causes the slow precipitation of Chl and Stb that occurs at large values of SI. Progress of the precipitation reactions causes the dissolution of the reactants, the primary minerals of the granite, to progress at a matching pace. Precipitation of Chl and Stb will
Sulfide is oxidized by CO2 to sulfate and a byproduct of the pyrite dissolution is methane. Abiotic methane (Fritz et al. 1987) has been documented by the presence of methane in fluid inclusions in alpine fissures (Mullis et al. 1993). The protons produced in the process are in turn consumed by the simultaneously progressing albite dissolution. The Al-bearing product phase in reaction (1) producing the Gotthard tunnel waters is Ca-zeolite rather than kaolinite or other clay minerals that are common in surface weathering processes. Thus, Na is released to the water and no mineral sink for Na is present on the fissures. Ca is present in the granite and gneisses as secondary calcite and in very small amounts of apatite. Therefore, it is evident that albite dissolution must be coupled to calcite dissolution: 32NaAlSi3 O8 þ 32Hþ þ 16SO4 2 þ 16CaCO3 þ 16SiO2 þ 96H2 O ¼ 16CaAl2 Si7 O18 7ðH2 OÞ þ 32Naþ þ 16SO4 2 þ 16CO2 ð5Þ Reaction (5) consumes the three common minerals albite, calcite and Qtz of the granite. It produces CO2 required for
250 K. BUCHER & I. STOBER pyrite oxidation according to reaction (4). The reaction also produces the thenardite component (Na2SO4) and, at 45C, the zeolite species Stb (Stober & Bucher 2004). Reaction (5) conserves Al among the silicates. The production of Ca-zeolite keeps Ca-concentration at the observed very low level (Tables 2 and 3). Ca for zeolite precipitation is produced by calcite dissolution, which also liberates CO2 that is used in reaction (4) for producing sulfuric acid. Balancing the proton producing and consuming reactions (4) and (5) results in equation (6) that describes the overall process of reacting recharge water with the minerals of the granite: 32NaAlSi3 O8 þ 8FeS2 þ 16CaCO3 þ 16SiO2 þ 142H2 O ¼ 16CaAl2 Si7 O18 7ðH2 OÞ þ 1CO2 þ 15CH4 þ 4Fe2 O3 þ 32Naþ þ 16SO4 2
ð6Þ
Reaction (6) describes the bulk water–rock reaction that consumes the minerals of the granite and replaces them with Stb and hematite and releases thenardite component to the water. As there is much more albite than pyrite in the granite assemblage, the process stops when all pyrite is used up. The Southern Aar Granite produces about 25 mg l)1 sulfate by reaction (6); the gneisses with their distinctly higher modal sulfide content reach about 400 mg l)1 from the progress of reaction (6). The two sulfate-rich waters on Table 3 require an additional mechanism to explain the very high sulfate content of 812 and 1578 mg l)1. The high sulfate concentrations of samples 14 and 17 (Table 3) are not coupled with corresponding increases in sodium (Fig. 8A) as could be expected from reaction (6). The increase in SO42) is linked to a corresponding increase in calcium, however (Fig. 6B). SO42) is added to the water by the simple anhydrite dissolution reaction: CaSO4 ¼ Ca2þ þ SO4 2
ð7Þ
The main input of sulfate at low sulfate concentration is well described by reaction (6), high sulfate concentrations are caused by reaction (7) (Fig. 8A). Bulk anhydrite disso-
(A)
lution may have two sources: (i) leaching experiments on granite and gneiss powders from the Black Forest have shown that a considerable amount of readily dissolvable CaSO4 is present in basement rocks (Bucher & Stober 2002). It is interesting to note that leachates from various Black Forest gneisses contain much higher SO42) compared with leachates from two different granites of the same area. The dissolvable CaSO4 is probably present as salt films on the silicate grains of the basement rocks although this hypothesis awaits verification. (ii) In all major tunnel constructions across the Central Alps (Gotthard rail base, old Gotthard rail, Lo¨tschberg rail base, Simplon rail), anhydrite has been found as a fissure mineral in basement rocks (e.g. Stalder et al. 1980). In the current Gotthard rail base tunnel, anhydrite occurs as relatively late centimeter-sized euhedral crystals. Fissure anhydrite is commonly coated with Chl telling that it may dissolve in modern waters that precipitate Chl. In sulfate-poor high-pH soda waters of the granite, the Na2CO3 component dominates water composition. Because reaction (5) consumes protons, the waters evolve to high pH (Tables 2 and 3). Thus, with reaction progress, CO32) is buffered to relatively high concentrations and the carbonate anion becomes the dominant solute in addition to Na and sulfate. The mass transfer for the sulfide-poor granite that produces soda water can be summarized by: 2NaAlSi3 O8 þ 1CaCO3 þ 1SiO2 þ 7H2 O ¼ 1CaAl2 Si7 O18 7ðH2 OÞ þ 2Naþ þ 1CO3 2
ð8Þ
The described processes produce the observed Na+, CO32) and SO42) granite waters and the observed fissure assemblage Stb and hematite. They adequately reflect the WRI in granite and gneiss upper crust (Fig. 9). The overall mass transfer for creating waters from the described WRI process can be computed as an inverse model using PHREEQC. As an example: 1 l of the high-pH granite water sample 7 on Table 2 can be created by the mass transfer process (coefficients in mmol):
(B)
Fig. 8. Composition of waters from the Gotthard rail base tunnel (Amsteg section): (A) sodium versus sulfate (mg l)1); (B) sodium versus Na–Cl (mmol l)1). Na–Cl corresponds to sodium contributed by the albite dissolution reaction.
Fluids in the upper continental crust 251
Fig. 9. Photograph of stilbite and hematite on adularia-coating water-conducting fractures in the Gotthard tunnel. Clays are absent.
1:48 albite þ 0:22 halite þ 0:09 fluorite þ 0:45 calcite þ 0:5 Bt þ 0:13 pyrite þ pure H2 O ¼ 0:25 Chl þ 0:06 Qtz þ 0:48 Kfs
Fig. 2). It is also typically two to three times lower than Cl ⁄ Br of modern seawater (288). We conclude from the observed relationships and from results of leaching experiments that the two halogens originate from fluid inclusions or ⁄ and from dissolved grain boundary salts that originally formed at the solidus of the granites (about 650–700C) and contains much higher NaBr component than halite from surface evaporites. Dissolution of grain boundary salt also adds sodium to the waters so that a part of total sodium is related to halite dissolution and only the remaining part is related to albite dissolution (Fig. 8B). It is evident that granite waters receive most of their sodium from Na-silicate dissolution whereas, in gneiss waters, a significant fraction of the sodium is related to halite dissolution (Fig. 8B).
CONCLUSIONS ð9Þ
þ 0:45 Stb þ 0:27 CH4 þ waterð7Þ With Bt = KMg1.5Fe1.5AlSi3O10(OH)2, Chl = Mg3Fe2Al2 Si3O10(OH)8, Stb composition from the LLNL data collection = Ca1.019Na0.136K0.006Al2.18Si6.82O18 • 7.33H2O, all other phases have pure endmember compositions. The inverse model also includes the minor components. Fluorite is present on fissures of the Aar Massiv (Fig. 5). It originates from a hydrothermal system that was active earlier in the alpine cooling and uplift history that formed alpine fissure Qtz and fluorite at about 300C. A minor amount of fluoride is contributed by the Bt to Chl reaction (Seelig & Bucher 2010). However, it should be noted that leaching experiments on granite and gneiss powders from the Black Forest revealed that a significant quantity of promptly dissolvable fluoride is present in basement rocks (Bucher & Stober 2002). In other waters, further north in the Gotthard base tunnel, F was as high as 30 mg l)1 (Seelig et al. 2007). Kfs is formed in process (9) in accordance with the computed saturation state of the water (Fig. 7) and with the presence of late adularia on the fissures (Fig. 9). The WRI also provides a source for chloride in basement waters. Chloride of the tunnel waters ranges from 7.5 to 86.1 mg l)1 (Tables 2 and 3). Granite waters contain about 10 mg l)1. Cl) represents a similar portion of the anion load in granites and gneisses. The source of chloride can be found, similar to anhydrite, within the basement rocks themselves (Savoye et al. 1998; Lira et al. 2007). In contrast to anhydrite, the presence of NaCl salt films on silicate grains has been positively verified (Markl & Bucher 1998). The Cl ⁄ Br mass ratio is about 85 ± 20 in both granites and gneisses, indicating that the source of chloride is the same in both rock types. The Cl ⁄ Br of basement waters is typically far below the corresponding ratio resulting from the dissolution of evaporitic halite (Table 1;
The upper continental crust contains an aqueous fluid in the fracture pore space that shows broad variations as regards composition and origin of solutes. In anorogenic geologically inactive continental crust with little topography, the fluid is typically a very salty aqueous solution. Common features of the fluids are a generally increasing chloride concentration with depth and a commonly low Cl ⁄ Br mass ratio. Fluids at some kilometers depth in the basement are typically saline brines with TDS of 100– 300 g l)1. The waters residing in granites and granitoid gneisses are normally NaCl-dominated brines; waters in amphibolitic rocks tend to contain a significant CaCl2 component. Waters located in gabbros tend to be CaCl2 brines. The variable XNa of deep brines appears to be a consequence of partial equilibration with the plagioclase of the host rock indicating a strong lithology control or buffering of the waters (Fig. 4). In continental collision orogens, such as the Alps, with a strong topography and virtually absent vegetation in the recharge area, ‘old’ crustal waters have been flushed from the basement and the modern waters reflect a pure WRI component and all solutes are exclusively derived from the fractured rocks when the recharge water reacts with the rock minerals along its flow path. The total amount of solutes derived from silicate dissolution reactions is very limited. Thus, the dissolution–precipitation component of WRI may also be small in high-TDS waters from lowland areas. The locally very high salinities of shield waters (Table 1) are not likely derived from silicate dissolution and precipitation processes such as reactions of the type formulated above. However, it is tempting to consider reactions (5) and (6) that massively consume H2O as a means to desiccate the fracture pore space and passively increase the TDS of the residual water. This would be consistent with the generally low Cl ⁄ Br mass ratio of basement waters. However, although Canadian Shield basement does
252 K. BUCHER & I. STOBER contain zeolites locally, (Currie 1965) generally zeolites on fractures appear to be rare, in contrast to the Fennoscandian shield and the Massif Central (Blyth et al. 2009). The deduced development of the chemical composition of Gotthard tunnel waters suggests that the WRI component of high-TDS (>10 g l)1) deep waters in the continental upper crust is probably trivial.
ACKNOWLEDGEMENTS We thank Peter Zbinden from AlpTransit Gotthard AG for giving access to the Gotthard tunnel construction site and permission for using the water composition data presented in this paper. The support and help from Peter Amacher from GeoUri and Beat Frey from AlpTransit in collecting water samples and measuring transient parameters on site is greatly appreciated. We thank the editors Bruce Yardley, Craig Manning and Grant Garven for making this special anniversary edition of Geofluids possible.
REFERENCES Aquilina L, Pauwels H, Center A, Fouillac C (1997a) Water-rock interaction processes in the Triassic sandstone and the granitic basement of the Rhine Graben: geochemical investigation of a geothermal reservoir. Geochimica et Cosmochimica Acta, 61, 4281–95. Aquilina L, Sureau JF, Steinberg M (1997b) Comparison of surface-, aquifer- and pore-waters from a Mesozoic basin and its underlying Palaeozoic basement, southeast France: chemical evolution of waters and relationships between aquifers. Chemical Geology, 138, 185–209. Banks DA, Green R, Cliff RA, Yardley BWD (2000) Chlorine isotopes in fluid inclusions: determination of the origins of salinity in magmatic fluids. Geochimica et Cosmochimica Acta, 64, 1785–9. Beaucaire C, Gassama N, Tresonne N, Louvat D (1999) Saline groundwaters in the Hercynian granites (Chardon Mine, France): geochemical evidence for the salinity origin. Applied Geochemistry, 14, 67–84. Blyth AR, Frape SK, Tullborg E-L (2009) A review and comparison of fracture mineral investigations and their application to radioactive waste disposal. Applied Geochemistry, 24, 821–35. Borchardt R, Emmermann R (1993) Vein minerals in KTB rocks. KTB Report, 2, 481–8. Bottomley DJ, Katz A, Chan LH, Starinsky A, Douglas M (1999) The origin and evolution of Canadian Shield brines: evaporation or freezing of seawater? New lithium isotope and geochemical evidence from the Slave craton. Chemical Geology, 155, 295– 320. Bucher K, Stober I (2002) Water-rock reaction experiments with Black Forest gneiss and granite. In: Water-Rock Interaction, Water Science and Technology Library (eds Stober I, Bucher K), pp. 61–96. Kluwer Academic Publishers, Dordrecht. Bucher K, Zhang L, Stober I (2009) A hot spring in granite of the Western Tianshan, China. Applied Geochemistry, 24, 402– 10. Currie KL (1965) Analogues of lunar craters on the Canadian Shield. Annals of the New York Academy of Sciences, 123, 915– 40.
Edmunds WM, Kay RL, McCartney RA (1985) Origin of saline groundwaters in the Carnmenellis granite (Cornwall, England): natural processes and reaction during hot dry rock reservoir circulation. Chemical Geology, 49, 287–301. Edmunds WM, Kay RLF, Miles DL, Cook JM (1987) The origin of saline groundwaters in the Carnmenellis granites, Cornwall (U.K.): further evidence from minor and trace elements. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), Geological Association of Canada Special Paper, pp. 127– 43. The Runge Press Limited, Ottawa. Evans MJ, Derry LA, France-Lanord C (2005) Hydrothermal flux of metamorphic carbon dioxide from the Central Nepal Himalaya. EOS, Transactions American Geophysical Union, 86, T23C-0581. Frape SK, Fritz P (1987) Geochemical trends for groundwaters from the Canadian Shield. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), Geological Association of Canada Special Paper, pp. 19–38. The Runge Press Limited, Ottawa. Frape SK, Blyth A, Blomqvist R, McNutt RH, Gascoyne M (2004) Deep fluids in the continents: II. Crystalline rocks. In: Surface and Ground Water, Weatherin, and Soils (eds Drever JI, Holland HD, Turekian KK), Treatise on Geochemistry, pp. 541– 80. Elsevier, Amsterdam. Fritz P, Frape SK, Miles M (1987) Methane in the crystalline rocks of the Canadian Shield. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), pp. 211–23. The Runge Press Limited, Ottawa. Frost R, Bucher K (1994) Is water responsible for geophysical anomalies in the deep continental crust? A petrological perspective. Tectonophysics, 231, 293–309. Gascoyne M (2004) Hydrogeochemistry, groundwater ages and sources of salts in a granitic batholith on the Canadian Shield, southeastern Manitoba. Applied Geochemistry, 19, 519–60. Gascoyne M, Davison CC, Ross JD, Pearson R (1987) Saline groundwaters and brines in plutons in the Canadian Shield. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), Geological Association of Canada Special Paper, pp. 53–68. The Runge Press Limited, Ottawa. Grimaud D, Beaucaire C, Michard G (1990) Modelling of the evolution of ground waters in a granite system at low temperature: the Stripaground waters, Sweden. Applied Geochemistry, 5, 515–25. Hardie LA, Eugster HP (1970) The evolution of closed-basin brines. Mineralogical Society of America Special Publication, 3, 273–90. Ingebritsen SE, Manning CE (1999) Geological implications of a permeability-depth curve for the continental crust. Geology, 27, 1107–10. Lahermo PW, Lampe´n PH (1987) Brackish and saline groundwaters in Finland. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), pp. 103–9. The Runge Press Limited, Ottawa. Lira R, Poklepovic MF, Dorais MJ (2007) Solid inclusions of magmatic halite and sylvite in felsic granitoids, Sierra Norte, Co´rdoba, Argentina. Lithos, 99, 363–84. Manning CE, Ingebritsen SE (1999) Permeability of the continental crust: implications of geothermal data and metamorphic systems. Reviews of Geophysics, 37, 127–50. Markl G, Bucher K (1998) Composition of fluids in the lower crust inferred from metamorphic salt in lower crustal rocks. Nature, 391, 781–3. Mo¨ller P, Weise SM, Althaus E, Bach W, Behr HJ, Borchardt R, Bra¨uer K, Drescher F, Erzinger J, Faber E, Hansen BT, Horn EE, Huenges E, Ka¨mpf H, Kessels N, Kirsten T, Landwehr D,
Fluids in the upper continental crust 253 Lodemann M, Machon L, Pekdeger A, Pielow HU, Reutel C, Simon K, Walther J, Weinlich FH, Zimmer M (1997) Plaleofluids and Recent fluids in the upper continental crust: results from the German Continental Deep Drilling Program (KTB). Journal of Geophysical Research, 102, 18245–56. Mullis J, Dubessy J, Poty B, O‘Neil J (1993) Fluid regimes during late stages of a continental collision: physical, chemical, and stable isotope measurements of fluid inclusions in fissure quartz from a geotraverse through the Central Alps, Switzerland. Geochimica et Cosmochimica Acta, 58, 2239–67. Paces T (1987) Hydrochemical evolution of saline waters from crystalline rocks of the Bohemian Massif (Czechoslovakia). In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), pp. 145–56. The Runge Press Limited, Ottawa. Parkhurst DL, Appelo CAJ (1999) PHREEQC (Version 2) – A Computer Program for Speciation, Batch-reaction, One-dimensional Transport, and Inverse Geochemical Calculations. US Geological Survey Water-Resources Investigations Report, Denver, Colorado. 99-4259, 310 pp. (download 2005). Patzak M (1991) Die Metabasite der KTB-Vorbohrung, Oberpfalz (NE-Bayern). PhD Dissertation, University Wu¨rzburg, Germany, unpublished, pp. 103. Pauwels H, Fouillac C, Fouillac AM (1993) Chemistry and isotopes of deep geothermal saline fluids in the Upper Rhine Graben: origin of compounds and water-rock interactions. Geochimica et Cosmochimica Acta, 57, 2737–49. Pitka¨nen P, Partamies S, Luukkonen A (2004) Hydrogeochemical Interpretation of Baseline Groundwater Conditions at the Olkiluoto Site.. Posiva Report OY, Posiva 2003-07, ISBN: 951-652121-5, FIN-27160, 159 pp. Olkiluoto, Finland. Savoye S, Aranyossy J-F, Beaucaire C, Cathelineau M, Louvat D, Michelot J-L (1998) Fluid inclusions in granites and their relationships with present-day groundwater chemistry. European Journal of Mineralogy, 10, 1215–26. Seelig U, Bucher K (2010) Halogens in deep water from the crystalline basement of the new Gotthard rail base tunnel. Geochimica et Cosmochimica Acta, in press, doi:10.1016/j.gca.2010.01. 030. Seelig U, Bucher K, Stober I (2007) High-pH waters from the new Gotthard rail base tunnel, Switzerland. In: Water-Rock Interaction, Vol 12 (eds Bullen T, Wang Y), pp. 251–4. Taylor & Francis, London.
Stalder HA, Sicher V, Lussmann L (1980) Die Mineralien des Gotthardbahntunnels und des Gotthardstrassentunnels N2, 161 pp. Repof Verlag, Gurtnellen, Switzerland. Stober I, Bucher K (1999) Deep groundwater in the crystalline basement of the Black Forest region. Applied Geochemistry, 14, 237–54. Stober I, Bucher K (2004) Fluid sinks within the earth’s crust. Geofluids, 4, 143–51. Stober I, Bucher K (2005) The upper continental crust, an aquifer and its fluid: hydaulic and chemical data from 4 km depth in fractured crystalline basement rocks at the KTB test site. Geofluids, 5, 8–19. Stober I, Bucher K (2007) Hydraulic properties of the crystalline basement. Hydrogeology Journal, 15, 213–24. Stober I, Richter A, Brost E, Bucher K (1999) The Ohlsbach Plume: natural release of deep saline water from the crystalline basement of the Black Forest. Hydrogeology Journal, 7, 273– 83. Vovk IF (1987) Radiolytic salt enrichment and brines in the crystalline basement of the East Europen Platform. In: Saline Water and Gases in Crystalline Rocks (eds Fritz P, Frape SK), pp. 197– 210. The Runge Press Limited, Ottawa. Vuorinen A, Mantere-Alhonen S, Uusinoka R, Alhonen P (1981) Bacterial weathering of Rapakivi granite. Geomicrobiology Journal, 2, 317–25. White AF, Schulz MS, Vivit DV, Blum AE, Stonestrom DA, Harden JW (2005a) Chemical weathering rates of a soil chronosequence on granitic alluvium: III. Hydrochemical evolution and contemporary solute fluxes and rates. Geochimica et Cosmochimica Acta, 69, 1975–96. White AF, Schulz MS, Lowenstern JB, Vivit DV, Bullen TD (2005b) The ubiquitous nature of accessory calcite in granitoid rocks: implications for weathering, solute evolution, and petrogenesis. Geochimica et Cosmochimica Acta, 69, 1455–571. Wintsch RP, Christoffersen R, Kronenberg AK (1995) Fluid-rock reaction weakening of fault zones. Journal of geophysical Research, 100, 13021–32. Zhu C (2009) Geochemical modeling of reaction path and geochemical reaction networks. Reviews in Mineralogy and Geochemistry, 70, 533–69.
Fluid-induced processes: metasomatism and metamorphism A. PUTNIS1 AND H. AUSTRHEIM2 Institut fu¨r Mineralogie, University of Mu¨nster, Mu¨nster, Germany; 2Physics of Geological Processes, University of Oslo, Oslo, Norway 1
ABSTRACT Metamorphism and metasomatism both involve the reequilibration of mineral assemblages due to changes in pressure, temperature and ⁄ or chemical environment. Both processes involve material transport but on different length scales, so every metamorphic reaction is metasomatic on a local scale. Fluids provide a transport mechanism which is orders of magnitude faster than solid state diffusion and induce reequilibration through dissolution of parent phases and reprecipitation of products. Chemical weathering (kaolinitization and serpentinization), and albitization are used as examples to describe the coupling between dissolution and precipitation. Albitization of feldspars in nature and in experiments is a pseudomorphic replacement which generates porosity in the albite. Porosity generation associated with interface-coupled dissolution-precipitation allows rapid material transport and together with fluid induced fracturing, is the mechanism of pervasive fluid flow through reacting crystals. Examples of metamorphic reactions in granulite-eclogite rocks illustrate the role of fluids in inducing chemical changes along fluid pathways. Microstructural criteria for a metamorphic event (i.e. change in P, T) are critically reviewed by describing the corona formed by reaction between kyanite and garnet, as well as partial replacement textures. We conclude that both corona structures and partial replacement textures are equally indicative of a metasomatic reaction (driven by a fluid-induced compositional change) as they may be of a metamorphic reaction driven by a change in P and ⁄ or T. This raises the question of the extent to which fluids play not only a catalytic role but also a thermodynamic role in determining the course of a metamorphic reaction. Key words: dissolution–precipitation, fluid–rock interaction, metamorphism, metasomatism Received 4 February 2010; accepted 5 February 2010 Corresponding author: Andrew Putnis, Institut fu¨r Mineralogie, University of Mu¨nster, Corrensstrasse 24, 48149 Mu¨nster, Germany. Email:
[email protected]. Tel: +49 251 8333451. Fax: +49 251 8338397. Geofluids (2010) 10, 254–269
INTRODUCTION Metasomatism is traditionally defined as metamorphism which involves a change in the chemical composition, excluding the volatile components. This distinction between metasomatism and metamorphism may appear to be well defined, but as no reaction can occur without transport, it appears that this definition requires that we specify an arbitrary length scale of mobility to distinguish the two processes. The fundamental question is therefore whether the two processes rely on completely different reaction mechanisms, that is, does metamorphism occur without the presence of a fluid phase, as opposed to metasomatism where fluid is ubiquitous. Within the general context of mineral–fluid interaction the distinctions become less clear, both in terms of the spatial scale
involved as well as the extent of chemical change between the protolith and the product mineral assemblage. If metamorphism is defined as the process by which a mineral assemblage reequilibrates in response to changes in pressure and temperature (P, T), then metasomatism is the reequilibration of a rock involving a change in the chemical composition. The latter clearly implies an external source and flux of material over a significant distance and this can only be achieved through the interaction of the parent rock with a fluid phase. Metasomatism may be driven by a tectonic event and so be associated with a change in P, T conditions, but as will be shown from the following examples, fluid-driven events need not be associated with tectonism. For a broad understanding of the response of a rock to changes in physicochemical conditions we therefore need to consider not only changes in pressure and temperature
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Fluid-induced processes 255 but also the availability and chemical composition of fluids which may be associated with a metamorphic or metasomatic event. Implicit in any such discussion is the question of the mechanism of reequilibration of a solid mineral assemblage to form a new assemblage, in the presence of a fluid phase. The major difference between the reequilibration of a rock with and without the presence of a fluid phase is the relative kinetics of the mechanisms for mass transport and reaction. In a ‘solid state’ reaction, mass transport can occur by grain boundary diffusion or volume diffusion through crystal grains, both slow mechanisms with large activation energies which are unlikely to have a significant effect on rock textures except perhaps at the highest temperatures. The common observation of microstructures in chemical and isotopic disequilibrium illustrates the fact that equilibration by solid state diffusion can only operate on a length scale much shorter than the scale of the microstructure. In the presence of a fluid phase, mass transport can take place by advection and diffusion through an interconnected fluid with activation energies typically 10 orders of magnitude faster, so that mass transport continues to operate at low temperatures. Reequilibration by dissolution and precipitation becomes a viable mechanism and as we shall see below, the rates of such processes are fast even on a laboratory time-scale. Reaction rates are therefore dependent on the availability of fluid, a point which will be emphasized throughout this paper. Thus on kinetic grounds, it has become clear that fluid flow and mass transport play fundamental roles in both metasomatism and metamorphism (Philpotts & Ague 2009). Metasomatism is clearly recognized on the outcrop scale when a reaction interface can be seen between the protolith and a rock with a different mineralogy, often with a distinct mineral assemblage and chemistry. An example is the albitization of granitic rocks from the Bamble Sector of SE Norway where albitization of a preexisting plagioclase feldspar (An21-23) occurs normal to fractures in essentially
(A)
Fig. 1. Field photographs of albitization. (A) Alteration of grey tonalite to red albitite in the Bamble Sector, SE Norway. (B) Albitization (pale) of grey metasediments in the Curnamona Province, Olary Domain, South Australia.
unaltered rock (Fig. 1A). In the field such a reaction front can be recognized by a reddening of the rock due to hematite precipitation associated with the albitization (Engvik et al. 2008). Such reaction fronts represent the limits of fluid infiltration, and allow a detailed study of the ‘before’ and ‘after’ compositions and textures. In the reaction front region, reaction interfaces in individual crystals can be studied by scanning and transmission electron microscopy and mechanisms of reactions inferred. However, such albitization occurs on a regional scale in the Bamble Sector, but in the absence of a clear reaction front, metasomatized rocks may not be recognized as such, and may merely be mapped as ‘albitites’ with no specific genesis, or even assigned the wrong origin (Elliott 1966; Bodart 1968). In the Curnamona Province in the Olary Domain of South Australia, similar reaction fronts are common (Fig. 1B), often indicating several episodes of fluid infiltration and reaction (Clark et al. 2005). Albitized rocks in this area cover many hundreds of square kilometres, and, as in the Bamble Sector, are spatially related to ore deposits. Perhaps one of the best-studied examples of large-scale sodic metasomatism is in the Mary Kathleen Fold Belt in the Mount Isa Inlier (Queensland, Australia) involving scapolitization and albitization as a result of infiltration of large volumes of highly saline brines at high temperatures (500–700C) (Oliver et al. 1994). Again, these alteration systems are associated with extensive mineralization and evidence of the passage of fluids through rocks, stripping them of their metal content which is reprecipitated from the fluid elsewhere. These and many other examples provide evidence for long-distance mass transfer. Albitization is not however restricted to high temperatures nor to any specific rock type or tectonic setting (Perez & Boles 2005) and can take place at temperatures as low as those in early diagenesis (Saigal et al. 1988; Lee & Lee 1998; Holness 2003; Lee et al. 2003). Regional-scale metasomatism is not confined to sodic alteration, and regional Ca-metasomatism, K-metasomatism,
(B)
256 A. PUTNIS & H. AUSTRHEIM Mg-metasomatism, Fe-metasomatism, Si-metasomatism have all been proposed to account for unusual rock compositions which have no chemical equivalents among sedimentary or igneous parent rocks (Schreyer 1977; Munz et al. 1994; John et al. 2004a; Jo¨ns & Schenk 2004). In the absence of a clear transition zone from unaltered to altered rock, an unequivocal metasomatic origin may be debatable, but there are sufficient well-documented examples to indicate that metasomatism is not merely a local process confined to selvedges and skarns around veins and igneous intrusions (Vernon & Clarke 2008) or blackwall between ultramafic and felsic lithologies (Winter 1998). Metasomatized rocks can also have compositions and mineral assemblages indistinguishable from metamorphic rocks which could have formed isochemically on a relatively large scale, and only a careful textural study of replacement fronts in the field and in individual grains may determine the nature of the protolith. For example, metasomatic fronts have been identified to propose crustal scale Mg-rich fluid infiltration associated with biotite-chlorite formation during metasomatism within the Mont Blanc granite (Rossi et al. 2005, 2007). One of the key issues in any discussion about metasomatism is the mechanism of mass transfer. Mass transfer over the distances observed in metasomatic reactions requires advection of fluid through a parent rock, and diffusion through this fluid. It is generally assumed that the fluid phase can only occupy fractures, pore spaces and intergranular boundaries within the rock, which leaves the problem of how equilibrium can be established within the individual crystals. In the absence of understanding mechanisms of equilibration of solids in the presence of a fluid phase, volume (solid state) diffusion through the crystals has been proposed to be necessary to establish local equilibrium between grains (e.g. Mueller 1967; Vernon & Clarke 2008). However, except at high temperatures solid state diffusion is negligible compared with fluid-induced reaction mechanisms (Putnis 2002; John & Schenk 2003). Understanding the mechanism by which a rock can change its composition and mineralogy at a fluid front is crucial. Although metasomatism can take place on a regional scale, the mechanism takes place on the scale of individual grains, by a coupled dissolution–precipitation process which is explained in more detail in this paper. The link between the mechanism of metasomatism and metamorphic reaction mechanisms was established some time ago by Carmichael (1969) who demonstrated that even in a closed isochemical system, the textural development of a metamorphic rock on a thin section scale can be explained by a sequence of dissolution and precipitation sub-reactions. These individual steps of dissolution–transport–reaction–precipitation are metasomatic reactions, i.e. with local changes in composition and redistribution of material. Thus on a small spatial scale the system is ‘open’ while on
a larger scale it may be closed. However, the fundamental mechanism of metasomatism and metamorphism at the reaction interface is the same. A further important observation in this context is that there are well-documented examples where metamorphism has been inhibited except where fluid has been introduced. A very well-studied example is the eclogitization of Precambrian anorthositic granulites of the Bergen Arcs, western Norway (Austrheim 1987; Jamtveit et al. 1990; Austrheim et al. 1997; Bjørnerud et al. 2002). In this case eclogitization does not involve a major chemical change, but is a function of fluid access and deformation, rather than being controlled by changes in P, T conditions alone. Figure 2 shows a rock outcrop in which a dark band of eclogite associated with a shear plane and fluid infiltration cuts across the paler granulite. The eclogitization is limited by the extent of fluid infiltration laterally from the shear plane. In the absence of fluid the granulite remains essentially unreacted. Assuming that the whole rock outcrop experienced the same P, T conditions, we must conclude that the presence of fluid is responsible for the eclogitization reaction. Fluid-induced reactions are also crucial for converting the lower oceanic crust to eclogite in old and cold subduction zones (Yuan et al. 2000; John & Schenk 2003). The role of chemically reactive aqueous fluids in driving reactions during contact metamorphism is well established (Jamtveit et al. 1992a,b; Ferry et al. 1998, 2002). While the formation of skarns is an unequivocal example of metasomatism associated with fluids mobilized by an igneous intrusion, the formation of wollastonite (CaSiO3) from
Fig. 2. Eclogitization (dark) along a fracture in the paler anorthositic granulite, Bergen Arcs, western Norway.
Fluid-induced processes 257 reaction between calcite and quartz is a very common metamorphic reaction in contact aureoles, and also driven by an infiltrating fluid which transports the CO2 from the reaction site (Ferry & Gottschalk 2009): CaCO3 þ SiO2 ) CaSiO3 þ CO2 ðaqÞ
ð1Þ
The porosity generated by such a reaction, taken simply as the molar volume change in the solid phases, allows fluid transport through the rock. As we shall see below, porosity development also depends on the relative solubilities of the parent and product phases in the specific interfacial fluid, and the thermodynamic stability of wollastonite in the above reaction implies that it is less soluble than a mixture of calcite + quartz in the fluid driving the reaction. This suggests that Ca and Si as well as CO2 may be transported by the fluid, resulting in an even higher porosity generation. The difference between typical skarn formation and wollastonite formation is then a matter of degree rather than principle. Taken a step further, the contact metamorphic facies variations defined in metapelitic assemblages in an aureole around a pluton, usually attributed to decreasing temperature, rarely include the effect of the fluid composition which drives these reactions. In this paper we focus on situations where an essentially dry, low permeability rock is infiltrated by a fluid which induces reequilibration of the mineral assemblage. Such rocks make up most of the continental crust. Fluid infiltration may or may not be associated with a tectonic event with changing pressure and temperature, and the effect of the chemistry of the fluid in changing the chemistry of the rock can vary over a wide spectrum. In the cases we consider, the temperature during infiltration is typically lower than the formation temperature of the parent rock. The situation in a fluid-saturated sedimentary basin undergoing progressive burial, diagenesis and metamorphism may be different, and the scale of mass transfer and fluid transport will be limited. The role of fluids in such a setting has been recently reviewed by Yardley (2009). Here we are not explicitly concerned with the scale of mass transport, but rather with the mechanism of reequilibration at the fluid– mineral interface. With this background linking the mechanism of metasomatism with metamorphism via the interaction of fluids with minerals at the grain scale, we will explore aspects common to both processes. If we take into account the spatial scale of the mass transfer and the extent of chemical change it becomes not merely convenient, but conceptually obvious that in comparing metamorphism in general, with metasomatism, we are dealing with a spectrum of responses of a rock to fluid reaction. The distinction between a metasomatic reaction and a metamorphic reaction at the reaction interface scale will be explored using both natural and experimental examples. The aim of this paper is to give examples of mineral–fluid reactions and their atomic-scale
mechanism and hence to describe how fluids move through rocks.
CHEMICAL WEATHERING AS A METASOMATIC PROCESS Chemical weathering is the in situ response of a rock exposed to the atmosphere. Although the ultimate breakdown product is soil, the early stages of weathering in which a rock retains its integrity, yet changes its mineralogy, are no different, in principle, to the processes which take place during metasomatism. The extent of chemical change can also vary in degree, such as in the breakdown of feldspars to either kaolinite at temperate conditions, or to bauxite under harsher weathering conditions. Long before a rock literally falls apart, chemical weathering alters the mineralogy. Weathering of silicate minerals to clays involves the transport of aqueous solutions to and from the reaction interface. High-resolution transmission electron microscopy of such a reaction interface (Banfield & Barker 1994) in the case of the breakdown of amphibole to smectite shows that the reaction is pseudomorphic, i.e. the volume occupied by the parent and product phases is unchanged (Fig. 3). This is typical of a mineral replacement reaction in which there is a coupled dissolution of the parent and precipitation of the product. The preservation of crystallographic orientation in isovolumetric mineral replacement reactions has been recently discussed by Putnis (2009).
Fig. 3. High-resolution transmission electron microscope image of amphibole isovolumetrically replaced by smectite. Note the crystallographic relationship between the parent and product phases (from Banfield & Barker 1994).
258 A. PUTNIS & H. AUSTRHEIM For the case of amphibole weathering to smectite in the presence of an acidic aqueous solution, Banfield & Barker (1994) have shown that, to preserve volume of the solid parent and product phases, the reaction must involve the removal of Mg, Fe, Si and Al from the reaction interface.
SERPENTINIZATION AND RODINGITIZATION – SEA FLOOR WEATHERING AND METASOMATISM Serpentinization is a common weathering ⁄ hydration reaction in ultramafic rocks, often associated with alteration near mid-ocean ridges, in which olivine and pyroxene are replaced by serpentine minerals and Fe in the parent minerals is released and forms iron oxides. In detail, such reactions are difficult to define precisely, as we need to know the composition of the reacting fluid, how this fluid changes composition as the reaction proceeds and then the composition of the fluid moving out of the ‘system’. For example, the serpentinization of olivine may be written as an isovolumetric replacement reaction: 2Mg2 SiO4 þ2Hþ þH2 O ) Mg3 ½Si2 O5 ðOHÞ4 þMg2þ
ð2Þ
Figure 4 shows the typical ‘mesh structure’ of partially serpentinized olivine, associated with both the precipitation of hematite (from the Fe component in the olivine) as well as magnesite, consistent with the liberation of Mg (as well as the presence of dissolved carbonate in the fluid). The interpretation of the replacement texture in Fig. 4 in terms of volume changes is equivocal. The dark grey areas of serpentine could be considered as an isovolumetric replacement of olivine or could be infilling cracks associated with volume expansion. Without some physical reference frame a case could be made for either interpretation.
Fig. 4. Back-scattered SEM image of a typical mesh texture in serpentinized olivine. The dark grey areas are serpentine, the lighter grey areas are olivine and the black material in the veins is magnesite. The brightest phase in the veins is magnetite.
In general, hydration reactions are thought to increase the solid volume, but this depends totally on the assumptions made about the composition of the fluid into and out of the system. For example, in the serpentinization of orthopyroxene, the reaction could be written as 2MgSiO3 þMg2þ þ3H2 O)Mg3 ½Si2 O5 ðOHÞ4 þ2Hþ
ð3Þ
which preserves the silica composition but implies a 50% increase in volume. Alternatively, the reaction 4MgSiO3 þ 2Hþ þ 5H2 O ) Mg3 ½Si2 O5 ðOHÞ4 þ Mg2þ þ 2H4 SiO4ðaqÞ
ð4Þ
is balanced on volume but involves loss of silica. (The approximate molar volumes of enstatite and lizardite are 25 and 100 cm3 mol)1 respectively). Isovolumetric serpentinization of orthopyroxene can be illustrated by the preservation of undisturbed lamellar structure when in exsolved intergrowths of orthopyroxene and clinopyroxene, only the orthopyroxene is serpentinized (Viti et al. 2005), while in other cases there is evidence of textural disruption and hydraulic fracturing (Shervais et al. 2005). When the rock being serpentinized contains calciumbearing phases such as clinopyroxene, the Ca is released to the fluid phase, which in turn can migrate to react with other mafic or felsic rocks to form rodingites. Rodingitization is a type of Ca-metasomatism leading to assemblages including grossular garnet, diopside, prehnite and chlorite with other minor phases. Assuming that serpentinization and rodingitization take place contemporaneously, Bach & Klein (2009) have used geochemical reaction path modelling to first calculate how seawater composition would change when reacted with peridotite, and then calculate what happens when this new fluid composition reacts with gabbro. The results support the hypothesis that typical rodingite mineral assemblages only form in areas where the composition of the fluids is controlled by serpentinization reactions. Austrheim & Prestvik (2008) observed that the rodingitization of the Leka ophiolite in Norway occurs through mineral replacement reactions and listed six reactions including replacement of olivine by clinopyroxene. In addition to serpentinization and rodingitization which are assumed to involve hydrothermal fluids at temperatures above 150C, pervasive low temperature interaction of peridotites with sea water results in lower MgO and higher SiO2 content relative to unweathered peridotite (Snow & Dick 1995). This type of sea floor weathering is attributed to the incongruent dissolution of olivine and pyroxene, releasing Mg2+ into solution and leaving an enriched silica surface phase behind (Iyer et al. 2008). The mechanism of incongruent dissolution is not well understood, although it is reasonable to assume that it has much in common with other examples of mineral–fluid reaction mechanism described below.
Fluid-induced processes 259
Fig. 5. Back-scattered SEM image of natural oligoclase (An content An22) partially replaced by albite (An2). The parent oligoclase has a smooth textures surface, while the albite (darker grey) is full of small sericite mica inclusions (pale) and pores (black) (from Engvik et al. 2008).
ALBITIZATION – FROM DIAGENESIS TO AMPHIBOLITE FACIES METASOMATISM As mentioned above, albitization of feldspars, and ultimately of all other minerals in a rock can take place over a wide range of temperatures and pressures, and indicates a reaction between a felsic rock and a saline fluid. A detailed study of a reaction front between a more calcic plagioclase (An20) and albite reveals some important features of metasomatism and mineral–fluid reaction mechanisms in general. Figure 5 (Engvik et al. 2008) shows that the interface between the parent oligoclase feldspar and the albite is sharp, that the albite is porous, and that the albitization is also associated with precipitation of hematite and sericitization, indicating introduction of K as well as Na and Si in the fluid phase. Albitization can be readily reproduced experimentally. Ho¨velmann et al. (2010) replaced calcic plagioclase (oligoclase An22) by albite in aqueous solutions containing both Na and Si at 600C within weeks. Figure 6 shows a partially replaced grain with a sharp reaction interface between the parent and product phase. Compositional data indicate extensive mobilization of trace and minor elements, as well as Al, into the fluid phase. Using 18Oenriched aqueous solution as a tracer for the mineral–fluid reaction results in the incorporation of 18O in the product albite, demonstrating that all elements are exchanged, as would be expected from a coupled dissolution–precipitation mechanism (Putnis 2002, 2009). Although the porosity is not obvious in low magnification images such as Fig. 6, higher magnifications and especially transmission electron microscopy show that in experimental feldspar– feldspar replacements the porosity is on a nanoscale (see e.g. Niedermeier et al. 2009). The coarser porosity in the
Fig. 6. Back-scattered SEM image of a cross-section of oligoclase crystals (ol An22) partially replaced by albite (ab) in hydrothermal solution containing Na and Si (image: J. Ho¨velmann).
natural sample (Fig. 5) may also be due to porosity coarsening over time, a phenomenon observed in salt replacement analogues (Putnis et al. 2005). The Engvik et al. (2008) and the Ho¨velmann et al. (2010) studies indicate that albitization is a pseudomorphic replacement reaction in which the dissolution of the parent feldspar is spatially and temporally coupled to the precipitation of albite. Although such coupling is not a prerequisite for a metasomatic reaction, when it does take place it is characterized by a number of distinctive features: (1) The replacement is isovolumetric, i.e. the external dimensions of the parent grains are preserved. (2) A sharp reaction front indicating that volume (lattice) diffusion plays an insignificant or minor role in the reaction. (3) Porosity (assumed to be interconnected) developed in the albite product which allows fluid and mass transfer through the albite product, to the reaction front. (4) Transfer of crystallographic information from the parent to the product. In the case of an isostructural replacement, the crystallographic orientation is preserved across the reaction interface, even though the parent is dissolved and the product reprecipitated. When the replacement is not isostructural some crystallographic features are preserved as in the orientations shown in Fig. 2. In the Bamble Sector, SE Norway albitization affects rocks with a range of lithologies. The example in Fig. 5 is from partially albitized granitic rocks. In the same area, gabbro has also been pervasively affected by Na metasomatism resulting in a complex sequence of replacement processes. From observations across reaction zones between the gabbro protolith and the final assemblage, a complex sequence of progressive replacements can be identified. Feldspars are replaced by Cl-scapolite which is itself replaced by albite.
260 A. PUTNIS & H. AUSTRHEIM
(A)
(B)
Phlogopite replaces ilmenite and hornblende (see also Fig. 10 and the discussion below), while Cl-rich apatite is replaced by OH- and F-rich apatite (Austrheim et al. 2008; Engvik et al. 2009). The replacement of the chlorapatite by hydroxyapatite (Fig. 7) is initiated at grain boundaries and fractures and it is clear from the microtextures that a simple ion-exchange diffusion process cannot be invoked to explain the replacement. However, the microtexture is consistent with a dissolution–precipitation mechanism. It is unclear if the multiple replacement textures observed represent the migration of metasomatic fronts through the rock caused by a single fluid event. All replacement textures document that the chemical inheritance between the parent and product phases may be limited. In the examples above the dissolution of the parent and precipitation of the product are spatially coupled and the textures suggest that the replacement process is at least approximately isovolumetric (apart from the porosity generated in the product). However, as already mentioned in the case of serpentinization, volume changes depend on the composition of the fluid moving into and out of the system. There are numerous documented examples, based on whole rock chemical analyses, where fluid flow through rocks has apparently resulted in compositional and wholerock volume changes (Ague 1994a, 2003a,b; Bucholz & Ague 2010 and references therein). Note that in assessing mass and volume transfer from whole-rock chemical analyses a fundamental issue is the definition of a geochemical reference frame. The usual procedure is to assess changes in bulk chemistry of a rock relative to ‘immobile’ reference elements (see Philpotts & Ague 2009 for a discussion). In experiments or textural studies where it is evident that replacement has been isovolumetric, mass balance calculations are based on preservation of solid volume as in Ho¨velmann et al. (2010) and Banfield & Barker (1994).
METAMORPHIC REACTIONS ALONG FLUID PATHWAYS Eclogite–granulite facies reactions In the Western Gneiss Region of South Norway, eclogite facies rocks have partially reacted to form granulite facies
Fig. 7. Back-scattered SEM images of Cl-rich apatite, (A) partially transformed to OH-rich apatite in patches around the rim of the former phase and also along fractures (B). Note that the replacement is not uniform along fractures but is more extensive near fracture intersections (from Engvik et al. 2009).
Fig. 8. Back-scattered SEM image of a grain boundary between garnet (light) and kyanite (dark grey). A fracture follows part of the grain boundary from lower right to upper left. The reaction products (plagioclase adjacent to the garnet and sapphirine-plagioclase symplectite adjacent to the kyanite) only form where the fracture is present and not on all garnetkyanite boundaries. The kyanite-kyanite grain boundary (large arrow) is filled with the sapphirine + plagioclase symplectite (see also Straume & Austrheim 1999).
assemblages during decompression (Straume & Austrheim 1999) and a detailed study of the textures associated with these reactions illustrates how fluid transport pathways along fractures control the extent of reaction. This is well illustrated by Fig. 8 (Straume & Austrheim 1999) which shows a garnet-kyanite contact boundary in the eclogite, reacting to sapphirine + plagioclase. These granulite facies reaction products do not form on all garnet-kyanite grain boundaries, only on those along which fractures have propagated, and along which fluid has been able to pass. The reaction is not isochemical, as the presence of 10% albite component in the plagioclase clearly shows that the reaction is not only between garnet and kyanite but must involve a sodium source (Straume & Austrheim 1999). The texture in Fig. 8 indicates that not all grain boundaries contained free fluid, and that under fluid-absent conditions the reaction did not proceed. Solid state reaction with volume diffusion was not a significant mechanism. Although grain boundaries are often cited as important transport paths for fluids, there must be a fundamental
Fluid-induced processes 261 difference between fluid transport along grain boundaries and along fractures. The origin of fractures is usually ascribed to deviatoric stresses generated by large-scale tectonic processes, or by local stresses developed during differential expansion of the various minerals during uplift. Grain boundaries are common sites for microfracturing creating a ‘hydraulic permeability’ (Etheridge et al. 1983; Cox & Etheridge 1989). A further origin of fractures may also be due to the highly localized stresses generated by volume changes associated with fluid-induced chemical reactions (Janssen et al. 2008; Jamtveit et al. 2009). Based on the composition of the reactants and products given by Straume & Austrheim (1999), and neglecting for the moment, the Na-content of the plagioclase, we can estimate the volume change for the reaction balanced on Si and O and approximately balanced for the other cations: 2Al2 SiO5 þ ðCa0:74 ; Fe0:78 ; Mg1:46 ÞAl2 Si3 O12 ¼ CaAl2 Si2 O8 þ 0:5Mg2:37 Fe0:45 Al8:6 O12 ½Si2 O8 þ 2SiO2 ð5Þ Using molar volume data from standard databases, the above reaction is very close to isovolumetric if the silica is removed in solution. If the introduction of Na from the fluid into the plagioclase is taken into account, some of the Si will also be consumed by the plagioclase also reducing its Al content. In Fig. 8, the sapphirine + plagioclase product assemblage of the reaction rim has a symplectite texture. The origin of symplectites in metamorphic rocks is also a highly debated topic as similar textures can form under a wide variety of conditions. The example given here shows that symplectite textures can form by reaction involving a fluid phase (i.e. dissolution) whenever two product phases cocrystallize (precipitation) within a restricted environment. Granulite–eclogite facies reactions The granulite to eclogite reaction in the Bergen Arcs of Western Norway has already been mentioned as a wellstudied example of fluid-induced metamorphism (Fig. 2). A detailed study of the transition zone between essentially unreacted granulite and eclogite, i.e. at the limit of fluid infiltration, gives further insights into the reaction mechanisms, specifically on the mechanism of the conversion of diopsidic clinopyroxene in the granulite to omphacitic clinopyroxene in the eclogite. This is an especially important reaction as the jadeite content in omphacite is commonly used as a geobarometer (Holland 1980). Figure 9 shows a polarized light and a back-scattered SEM image of the reaction zones between granulitic clinopyroxene with composition Di0.70CaTs0.20Jd0.10 (labelled zone 1) and omphacitic clinopyroxene with composition Di0.45CaTs0.05Jd0.5 (labelled zone 4). The granulitic clino-
pyroxene can be found free of inclusions and lamellae in the most pristine granulite, but near the eclogites it contains fine amphibole lamellae typically 0.2–0.4 lm in width. Zone 2 is a symplectite which replaces the pyroxene in zone 1. The symplectite is an intergrowth of a pargasitic amphibole with (Na+K) ⁄ Ca 0.75 and Ca ⁄ (Mg+Fe) 0.5 and a clinopyroxene Di0.65CaTs0.10Jd0.25, significantly more sodic than in zone 1. This zone contains numerous small pores. Zone 3 is a coarser amphibole-clinopyroxene symplectite apparently replacing zone 2. The proportion of amphibole in the zone 3 symplectite is reduced but has the same composition as that in zone 2. The pyroxene is more sodic with composition Di0.60CaTs0.10Jd0.30. Zone 3 has large, apparently open pores. The final stage (zone 4) is almost end-member omphacite (Di0.45CaTs0.05Jd0.5.) with little amphibole and a reduced, though still present, porosity. Two possible interpretations of this texture can be suggested. The texture could have evolved during changing P and T conditions so that the higher Jd content corresponds to the maximum P, and the symplectite subsequently formed during uplift. Alternatively the zonal distribution reflects the changing composition of the fluid phase at the extended interfacial region between the granulite and eclogite. The sequence also suggests that the fine amphibole lamellae in zone 1 are the first indication of fluid infiltration associated with the reactions to form the eclogite. Realizing that the reactions take place through a fluid phase may require a modification of the P–T paths suggested for these rocks.
HOW DO FLUIDS MOVE THROUGH ROCKS? It has been recognized for many years that substantial quantities of aqueous fluids are involved in low- and medium-grade metamorphism, in amounts greater than can be supplied by dehydration of surrounding rocks (Fyfe et al. 1978; Etheridge et al. 1983; Ferry 1994; Rumble 1994; Oliver 1996). The driving forces for fluid flow through low permeability rocks have been variously attributed to (i) a response to regional thermal gradients and ⁄ or (ii) deformation-induced hydraulic head gradients (Oliver 1996). In either case, fluid flow depends on the availability of pathways, either preexisting in the intrinsic porosity of the rock, or created by the deformation ⁄ fluid event associated with the metamorphism. In the latter case, dilatancy (the volume change resulting from deformation) is considered to be a key requirement for fluid movement both on a local and a long-range length scale. Hydraulic fracturing, arising from fluid overpressure during devolatilization reactions, has also been recognized as an important mechanism for fluid flow (Thompson & Connolly 1992; Ague 1994b, 2003a; Davies 1999; Oliver & Bons 2001; Zack & John 2007; Lyubetskaya & Ague 2009). Similar considerations of fluid flow paths apply in the case of metasomatism.
262 A. PUTNIS & H. AUSTRHEIM
(A)
(C)
Fig. 9. The reaction zones between diopsidic pyroxene in granulite (zone 1) and omphacite in eclogite (zone 4). Zones 2 and 3 are amphibolesodic pyroxene symplectites in which the omphacite component of the pyroxene increases in a step-wise manner, while zone 4 contains stoichiometric omphacite. (A) Transmitted light micrograph showing the symplectite in zone 2 replacing the diopsidic pyroxene in zone 1. A coarser symplectite (zone 3) with a high porosity [black spots in the BSE SEM image in (B)] replaces zone 2. (C) A higher magnification BSE SEM image showing the successive zones 1–3 (images F. Casarin).
(B)
Apparent permeabilities of rocks have been estimated from time-integrated fluid fluxes inferred from measured progress of metamorphic reactions, and are invariably higher than the permeabilities of similar rocks measured in the laboratory (Oliver 1996; Ingebritsen & Manning 2003). This suggests that during metamorphism and metasomatism, permeability is enhanced. The most popular mechanisms cited are grain-scale dilatancy and hydraulic fracturing. Thus porosity and permeability are transient phenomena associated with the process itself and not necessarily dependent on intrinsic properties of the rock. The concept of ‘reaction-enhanced permeability’ has been discussed for some time and is considered to be dependent on a volume decrease between parent and prod-
(A)
(B)
(C)
uct solid phases in a metamorphic reaction (Rumble & Spear 1983; MacDonald & Fyfe 1985; Cartwright 1997; Ferry 2000). For example, the reaction between quartz and calcite to wollastonite, written in the form of Eqn 1 would result in a molar volume reduction of 33% and hence very significant permeability generation and probable compaction of the wollastonite (Balashov & Yardley 1998). However, a literal mole-for-mole interpretation of such a reaction may not be valid, as reactions may be isovolumetric and therefore reaction equations are more appropriately written to preserve volume, as discussed above, and in more detail in Putnis (2009). Recent advances in understanding the mechanism of pseudomorphic mineral replacement processes (Putnis
(D)
(E)
Fig. 10. When a solid (A) comes into contact with a fluid with which it is out of equilibrium, dissolution of even a few monolayers of this parent may result in an interfacial fluid which is supersaturated with respect to a product phase, which may nucleate on the surface (B). (C)–(E) Continued dissolution and precipitation at the parent-product interface depends on the generation of interconnected porosity in the product phase allowing mass transport through the fluid phase and the migration of the reaction interface from the surface through the parent.
Fluid-induced processes 263 2002, 2009; Putnis & Mezger 2004; Putnis et al. 2005; Putnis & Putnis 2007; Xia et al. 2009a) have shown that reactive fluids can pass through individual crystal grains by a mechanism termed interface-coupled dissolution-precipitation. In the context of fluid flow, porosity generation is a key feature of this mechanism. Figure 5, the replacement of a more calcic plagioclase (oligoclase) by albite, shows that interconnected porosity and hence permeability must have been generated by the replacement process. This allows continuous contact between the fluid and the reaction interface as it moves through the crystal, allowing advective transport of fluid through the pores and providing diffusional pathways for element transport. For example, albitization of plagioclase involves addition of Na and Si and removal of Al and Ca, as well as trace elements all of which are transported through the fluid phase in the pores (Ho¨velmann et al. 2010). The porosity is not only a consequence of a reduction in solid molar volume, which in the case of albite replacing oligoclase is too small to account for the observed porosity, but also depends on the relative solubilities of parent and product phases in the reactive fluid. In other words, more parent material may be dissolved than product reprecipitated, thus generating porosity. Indeed, it has been shown that even in a pseudomorphic reaction in which there is a molar volume increase, porosity may still be generated by the reaction (Putnis et al. 2007; Xia et al. 2009b). The generation of porosity during mineral replacement reactions has been reported in a large number of studies on both natural and experimental samples (Putnis 2009 and references therein). The general principle that some material may be lost to the fluid phase due to solubility differences is not restricted to pseudomorphic replacement, and may equally apply to a metamorphic reaction where dissolution and precipitation are not as spatially coupled as in a pseudomorphic reaction. The recognition of porosity generation in a metasomatic reaction creates a further link between permeability and reaction progress. Combined with hydraulic microfractures it provides a mechanism for the pervasiveness of fluid flow and element transport. The fact that a reactive fluid can move through a single crystal, dissolving the parent at the reaction front and precipitating the product in its wake, eliminates the need to invoke volume diffusion as a mechanism of chemical interaction between fractures and the adjacent ‘wallrock’, as proposed by Cox & Etheridge (1989). Porosity generation associated with contact metamorphism of limestone has recently been quantified using small angle neutron scattering techniques which can be applied to bulk samples, and confirms the micro-scale observations above, that a metasomatic ⁄ metamorphic front is associated with a high permeability zone that moves outwards with the aureole as metamorphism progresses (Anovitz et al. 2009).
THE ROLE OF THE FLUID IN METAMORPHIC AND METASOMATIC REACTIONS If we accept all the petrological, mineralogical, textural, microstructural and stable isotopic data supporting high water fluxes in regional metamorphic belts and metasomatized terrains (Ferry 1994; Ague 2003b; John et al. 2004b; Rossi et al. 2005; Beinlich et al. 2010), then a fundamental question is what is the mechanism by which a mineral or mineral assemblage in a rock equilibrates in the presence of an aqueous fluid. Despite the perhaps obvious proposition that solid state diffusion cannot compete kinetically with dissolution-precipitation as a mechanism of mineral reequilibration (except at temperatures approaching the melting point), the replacement of one mineral by another has often been considered as an essentially solid state reaction, even when the reequilibration involved chemical exchange with a fluid phase. Part of the reason for this is the historical over-reliance that mineralogists placed in materials science (e.g. metallurgy and ceramic science, solid state chemistry and physics) for the development of an understanding of phase transformations and reactions between solids. The geological time-scales invoked for metamorphic reactions also suggested that even slow volume diffusion on enormous time-scales could bring about a complete change in mineral assemblage. At a crystallographic level, replacement processes which preserved both the external shape and internal crystal structure (such as the replacement of one feldspar by another) were obvious candidates for a solid state ion-exchange mechanism. Early experimental work using 18O in the fluid phase in feldspar-fluid equilibration reactions (O’Neil & Taylor 1967; O’Neil 1977) established that oxygen was also exchanged, which suggested the dismantling and reassembly of the whole alumino-silicate framework. Recent work has shown that the preservation of crystal structure in an interface-coupled dissolution–precipitation mechanism is an inevitable consequence of the close coupling when the crystal structures of the parent and product phase are the same. In such a case, the epitaxial relationship between the nucleating phase and the substrate greatly enhances the rate of precipitation (nucleation). When the crystal structures of parent and product are different the parent phase may be polycrystalline, with some degree of lattice matching across the interface, yet may still retain the external shape and volume of the parent. The isochemical nature of many metamorphic reactions implied in the metamorphic facies concept also suggested that a reaction could be achieved by solid state reactions and relatively localized diffusion. Again stable isotope studies demonstrated that metamorphism involved reaction with a fluid which was not in isotopic equilibrium with the parent assemblage. The ability to map oxygen isotope
264 A. PUTNIS & H. AUSTRHEIM distributions with high spatial resolution has confirmed that oxygen isotope reequilibration in individual minerals follows replacement involving the major elements (van Haren et al. 1996; Cole et al. 2004; Bowman et al. 2009). The mechanism of mineral replacement by interface-coupled dissolution-precipitation is illustrated schematically in Fig. 10. In an open experimental system the reequilibration of a mineral in the presence of a fluid with which it is out of equilibrium first involves some dissolution of the parent phase(s). Even the dissolution of a few monolayers of the parent will result in a fluid which is supersaturated with respect to a more stable phase. The precipitation of this phase on the surface of the parent phase is dependent on the kinetics of nucleation: existence of epitaxial (crystallographic) relations between parent and product will favour nucleation, as will a high value of supersaturation at the fluid–mineral interface (i.e. a low solubility of the product phase). Autocatalysis (the feedback between surface nucleation and dissolution rate) also contributes to the overall balance between dissolution and precipitation rates (Anderson et al. 1998a,b). Dissolution and precipitation may be very closely coupled, resulting in an exact pseudomorph preserving nanoscale features of the parent, or they can become spatially separated, depending on whether the overall reaction is dissolution controlled or precipitation controlled respectively (Xia et al. 2009a). In a mineral assemblage in a rock, the reactions will be controlled by the fluid pathways available and the mineralogy through which initial hydraulic or tectonically induced fractures propagate. Although the individual dissolutiontransport-precipitation steps are much more difficult to evaluate (see Carmichael 1969), as are the spatial relationships and coupling between dissolution and precipitation, the basic principles remain the same. In the early stages of a metamorphic or metasomatic reaction when there is the possibility of seeing relationships between reactants and products, the reaction mechanisms are related to fluid pathways (e.g. Fig. 8). When the final mineral assemblages have been formed there may be little textural evidence of the former assemblage. In some special cases where the parent assemblage contained ilmenite with zircon rims, the metasomatism completely removed all the ilmenite, but the zircon rims were unaffected and were thus physical markers of the former existence of the ilmenite, now replaced by phlogopite (Fig. 11) (Austrheim et al. 2008). In this case a Fe-Ti oxide has been replaced by an Mg-silicate, indicating addition of Mg, Si, Al and removal of Fe, Ti from the former site of the ilmenite grain. In a subsequent replacement process prehnite which replaces the phlogopite contains inclusions of titanite, suggesting that the Ti from the phlogopite reacts with Ca and Si from the fluid to form titanite (Fig. 12A). The replacement of the prehnite by albite (Fig. 12B) leaves the titanite inclusions intact within the albite. Such observations of the
(A)
(B)
Fig. 11. (A) BSE image showing the early stages of replacement of ilmenite and hornblende by phlogopite. The bright phase is zircon which marks the original ilmenite grain boundary. (B) After total replacement of all phases by phlogopite, the ‘necklace’ of zircons is the only indication of the grain boundary of the replaced ilmenite grain (from Austrheim et al. 2008).
microstructural evolution record fluid-induced replacement processes and as pointed out by Vernon et al. (2008) can be misinterpreted as indicating parageneses along a specific P, T path. Dissolution–transport–precipitation is a very efficient mechanism of mineral reequilibration and can be driven by very small free energy differences between the parent and product phases. This has been demonstrated by Nakamura & Watson (2001) who showed that reduction in interfacial free energy is sufficient to drive recrystallization. A reduction in strain energy also drives the dissolution of fine-scale coherent intergrowths of cryptoperthite and the reprecipitation of patch perthite (Parsons & Lee 2009). Wood & Walther (1983) and Walther & Wood (1984) have also shown that rates of dissolution of a wide range of minerals under near-equilibrium conditions are very fast on a geological time-scale even at a 1C temperature overstep. The transport and nucleation steps are much more difficult to quantify and depend on the distribution of fluid and its supersaturation. Wood & Walther (1983) argue that the nucleation step is not rate limiting and therefore in the presence of a fluid phase the extent of disequilibrium in
Fluid-induced processes 265
(A)
(B)
Fig. 12. (A) Replacement of prehnite (prh, upper right) by albite (ab, dark). The titanite inclusions in prehnite (white) are also present in the albite, suggesting that titanite was insoluble in the albitizing fluid. The grain in the lower right corner is scapolite. (B) BSE image showing scapolite (grey) and albite (dark). The bright spots in albite are inclusions of titanite. The albite has completely replaced prehnite. Note that albite preserves the form of a former mica, indicating the following sequence of replacements: phlogopite ) prehnite + titanite ) albite + titanite.
metamorphic rocks is likely to be small. More recent textural evidence suggests that nucleation problems may require reactions to be considerably overstepped (Wilbur & Ague 2006) but that subsequent growth is likely to be rapid. However, in the absence of fluid, metamorphic reactions may not occur at all. The fact that fluids increase the rates of reactions is well known, so that there is a tendency to consider the fluid as a catalyst which plays no role in the thermodynamics of the reaction. A catalyst, in the strict sense, makes available a different mechanism with a lower activation energy. Rather than relying on vague concepts such as ‘fluidenhanced solid state diffusion’ to explain why ‘solid-state’ reactions are faster in the presence of the fluid, we suggest that fluids inevitably provide a dissolution-precipitation mechanism for a reaction and this then competes kinetically with solid state diffusion. The fact that the fluid phase is not pure water but contains dissolved CO2 as well dissolved ions suggests that the fluid must play a thermodynamic as well as a kinetic role in the reaction. However as data on fluid thermodynamics over the range of compositions, temperatures and pressures are very limited, this aspect of the role of fluids is usually conveniently ignored. The question of determining whether a reaction is driven by changes in temperature, pressure or chemistry due to infiltration of fluid is a major challenge for the future.
REACTION INTERFACES AND COEXISTING MINERALS In a metamorphic event P–T paths are inferred from sequences of reactions assumed to take place as a rock crosses univariant reaction lines in P–T space. The replacement of one mineral assemblage by another is taken as an indication that such a reaction has or is taking place in a rock. However, the use of mineral microstructures to infer metamorphic events is subject to misinterpretation, as pointed out by Vernon et al. (2008). To define a metamorphic paragenesis requires that the assemblage consists of
minerals which ‘have grown simultaneously or have stably coexisted’ (Vernon et al. 2008). Vernon et al. (2008) consider corona structures to be among the most reliable indicators of a metamorphic reaction. A corona in this sense is a reaction rim between a core mineral and its surroundings, representing an arrested metamorphic reaction which captures a time-frame where the minerals in the corona are in equilibrium with the two reacting minerals on either side. However, the reliability of corona structures as indicators of a metamorphic event is called into question if we consider in more detail the reaction between kyanite and garnet shown in Fig. 8. The reaction rim is no different in principle to a corona. However, the plagioclase in the reaction rim contains 10% albite component and Na is not present in either kyanite or garnet. Furthermore, the reaction rim is not equally developed on all kyanite–garnet interfaces, indicating the reaction has only taken place where the interface coincided with a fracture allowing infiltration of fluid. Equilibration by solid state diffusion of components between the two phases played no significant role in the development of the rim. The dissolution of both garnet and kyanite by the fluid created a compositional gradient in the interfacial fluid, with the fluid at the garnet interface more Ca-rich than the fluid at the kyanite interface. The zonation of the reaction rim, with a layer of plagioclase nucleating on the garnet, indicates a chemical potential gradient in the fluid phase. The importance of the boundary layer of the fluid at the dissolving interface in controlling the precipitation of the product has been emphasized by Putnis & Mezger (2004) and Putnis et al. (2005). The residual fluid containing other components of the dissolution (Ca, Mg, Fe, Al, Si), together with Na from the infiltrating fluid, co-crystallize to form the sapphirine + plagioclase symplectite. The ‘reaction rim’ continues through a kyanite–kyanite contact (arrowed in Fig. 8) where the space is filled with the symplectite, demonstrating the mobility of these elements in the fluid phase through the fracture. Note that the presence of a fluid phase in a reaction does not necessarily imply the
266 A. PUTNIS & H. AUSTRHEIM formation of hydrous minerals even when the fluid components may play a significant role in the thermodynamics. The equilibrium (if any) established in this reaction is between the fluid at the interface (formed from both the infiltrating solution and the dissolving parent) and the product phases in the reaction rim. If the infiltrating solution is out of equilibrium with the parent kyanite + garnet assemblage, then the corona, or rim in this instance, does not represent a reliable indicator of a metamorphic reaction, as the reaction is not isochemical. Although we may infer that the infiltrating fluid contains Na, we can make no assumptions about the composition of the outgoing fluid. The determination of P, T conditions for the reaction becomes problematic, as the fluid is an integral component in the reaction and its thermodynamics must be taken into account. Furthermore, the presence of Na in the plagioclase requires that both the thermodynamic mixing properties of the cations, as well as the change in Al,Si ordering, be taken into account (Carpenter & Ferry 1984) when determining the thermodynamics of the reaction. When a fluid is involved in a metamorphic reaction, the interpretation of the microstructure and phase assemblage in terms of a metamorphic paragenesis becomes highly problematic. Among the microstructural criteria which have been used to infer a sequence of metamorphic reactions in P–T space, partial replacement of one mineral by another has been used to indicate a ‘frozen-in’ metamorphic reaction with the implication that the parent and product phases coexist (Vernon et al. 2008). However, as we have seen from studies of albitization, as just one example (Engvik et al. 2008; Ho¨velmann et al. 2010), metasomatism and interface-coupled dissolution-reprecipitation results in precisely such a partial replacement (Figs 5 and 6). Albitization of labradorite proceeds in the same way as oligoclase (Ho¨velmann et al. 2010) and in both cases the parent plagioclase and product albite are clearly out of equilibrium. The interface between them represents the reaction front between the plagioclase and an Na-Si-bearing fluid. As in the case of coronas, equilibrium, if it exists at all, would be between the interfacial fluid (resulting from the dissolution of the plagioclase) and the precipitating albite.
CONCLUSIONS AND FURTHER IMPLICATIONS From the brief examples above, we can conclude that at the reaction interface level, there is no fundamental mechanistic difference between metasomatism and metamorphism. Except at the highest temperature grades of metamorphism, where solid state diffusion may have a significant role to play in mass transport, or where melting begins, metamorphism involves aqueous solutions. Both metasomatism and metamorphism take place by a sequence of dissolution-transportprecipitation reactions, which in the case of metasomatism
involves a significant change in the bulk chemistry of the rock on a (undefined) macroscopic scale. In cases where the dissolution and precipitation are closely coupled, reaction interfaces can be recognized as partial pseudomorphs. The interpretation of partial pseudomorphs is ambiguous – when do reaction interfaces indicate crossing a reaction line in P, T space and when do they indicate a reaction between an infiltrating fluid out of equilibrium with the parent mineral? In the former case the mineral paragenesis is used to determine a reaction defined in P, T space, treating the fluid as a catalyst which does not alter the thermodynamic relations between parent and product solids. In the latter case the reaction is dominated by the composition of the fluid rather than the P, T conditions, and the extent of reaction by the relative solubilities of the parent and product phases in that fluid. The recognition of when a mineral pair is ‘coexisting’ in the thermodynamic sense requires a detailed study of textural and chemical relations from the field scale to the micron scale. In a system which evolves by dissolution and precipitation, the reaction rates will depend on the kinetics of both processes. Although measured dissolution rates in an open system under laboratory conditions are fast on a geological time-scale (Wood & Walther 1983; Walther & Wood 1984), there is ample evidence of chemical and isotopic disequilibrium on the grain scale (Abart & Sperb 1997; Cole et al. 2004; Villa 2006). Understanding the importance of porosity generation is vital. If a reaction of a parent mineral with a fluid produces an impermeable product, this may effectively isolate the parent against further reaction. Some minerals in the assemblage may react with a fluid and be replaced, while others remain unreacted, although all minerals in the parent assemblage may have been out of equilibrium with the fluid. As pointed out by Vance et al. (2003) and Villa (2006), this opens new possibilities in the dating of geological events, based on the possibility of high spatial resolution isotopic measurements, but most importantly on understanding microtextures, mechanisms of mineral reactions and mass transport on the microscale. A further implication in this context is the continuing debate about the meaning of isotopic ‘closure temperatures’ (Dodson 1973; Glodny et al. 2002). In the examples discussed above, we have proposed that temperature is not the only parameter needed to describe a metamorphic ⁄ metasomatic reaction. In the context of geochronology, Villa (1997) recognized that two entirely different processes affect element (and isotopic) transport: temperature-dependent volume diffusion and fluid-assisted recrystallization, opening new avenues for dating both the thermal history as well as the ‘hygrometric’ history of a rock. A further implication of the role of fluids in metamorphism ⁄ metasomatism is the interaction between deformation, fluid infiltration and induced chemical reactions. The
Fluid-induced processes 267 various possible feedbacks, in which fluid infiltration causes rock weakening and shear deformation related to metamorphism, have been explored for some time (Austrheim & Griffin 1985; Wintsch 1985; Wheeler 1987; Wintsch & Yi 2002). The realization that fluids can pervasively infiltrate impermeable rocks by reaction-induced porosity formation opens the possibility that fluids are far more than merely an initiator of chemical reactions, but may in fact be the principal driver for geodynamics.
ACKNOWLEDGEMENTS This paper has benefitted from discussions with Muriel Erambert, Bjørn Jamtveit, Timm John, Christine Putnis and Bruce Yardley. However the ideas expressed here are those of the authors. We thank Andreas Schmidt-Mumm and Colin Conor for valuable discussions and field guidance in Australia. We are grateful for a very constructive review by Jay Ague. This work was supported by funding from the EU Initial Training Network, Mechanisms of Mineral Replacement Delta-Min (http://www.delta-min. com), the Humboldt Foundation and the Norwegian Research Council grants to the Physics of Geological Processes (PGP) Norwegian Center of Excellence.
REFERENCES Abart R, Sperb R (1997) Grain-scale stable isotope disequilibrium during fluid–rock interaction. 1. Series approximations for advective-dispersive transport and first-order kinetic mineralfluid exchange. American Journal of Science, 297, 679–706. Ague JJ (1994a) Mass transfer during Barrovian metamorphism of pelites, south-central Connecticut. I: evidence for changes in composition and volume. American Journal of Science, 294, 989–1057. Ague JJ (1994b) Mass transfer during Barrovian metamorphism of pelites, south-central Connecticut. II: channelized fluid flow and the growth of staurolite and kyanite. American Journal of Science, 294, 1061–134. Ague JJ (2003a) Fluid flow in the deep crust. In: Treatise on Geochemistry, Vol. 3.06 (eds Holland HD, Turekian KK), pp. 195–228. Elsevier, Amsterdam. Ague JJ (2003b) Fluid infiltration and transport of major, minor, and trace elements during regional metamorphism of carbonate rocks, Wepawaug Schist, Connecticut, USA. American Journal of Science, 303, 753–816. Anderson JG, Doraiswamy LK, Larson MA (1998a) Microphase assisted ‘‘autocatalysis’’ in a solid-liquid reaction with a precipitating product – I. Theory. Chemical Engineering Science, 53, 2451–8. Anderson JG, Larson MA, Doraiswamy LK (1998b) Microphase assisted ‘‘autocatalysis’’ in a solid-liquid reaction with a precipitating product – II. Experimental. Chemical Engineering Science, 53, 2459–68. Anovitz LM, Lynn GW, Cole DR, Rother G, Allard LF, Hamilton WA, Porcar L, Kim MH (2009) A new approach to quantification of metamorphism using ultra-small and small angle neutron scattering. Geochimica et Cosmochimica Acta, 73, 7303–24.
Austrheim H (1987) Eclogitization of lower crustal granulites by fluid migration through shear zones. Earth and Planetary Sciences Letters, 81, 221–32. Austrheim H, Griffin WL (1985) Shear deformation and eclogite formation within granulite-facies anorthosites of the Bergen Arcs, Western Norway. Chemical Geology, 50, 267–81. Austrheim H, Prestvik T (2008) Rodingitization and hydration of the oceanic lithosphere as developed in the Leka ophiolite, north-central Norway. Lithos, 104, 177–98. Austrheim H, Erambert M, Engvik AK (1997) Processing of crust in the root zone of the Caledonian continental collision zone: the role of eclogitzation. In Collision Orogens: Zones of Active Transfer between Crust and Mantle (eds Touret JLR, Austrheim H). Tectonophysics, 273, 129–56. Austrheim H, Putnis CV, Engvik AK, Putnis A (2008) Zircon coronas around Fe-Ti oxides: a physical reference frame for metamorphic and metasomatic reactions. Contributions to Mineralogy and Petrology, 156, 517–27. Bach W, Klein F (2009) The petrology of seafloor rodingites: insights from geochemical reaction path modeling. Lithos, 112, 103–17. Balashov VN, Yardley BWD (1998) Modeling metamorphic fluid flow with reaction – compaction – permeability feedbacks. American Journal of Science, 298, 441–70. Banfield JF, Barker WW (1994) Direct observation of reactantproduct interfaces formed in natural weathering of exsolved, defective amphibole to smectite: evidence for episodic, isovolumetric reactions involving structural inheritance. Geochimica Cosmochimica Acta, 58, 1419–29. Beinlich A, Klemd R, John T, Gao J (2010). Trace-element mobilization during Ca-metasomatism along a major fluid conduit: eclogitization of blueschists as a consequence of fluid–rock interaction. Geochimica et Cosmochimica Acta, 74, 1892–922. Bjørnerud MG, Austrheim H, Lund MG (2002) Processes leading to eclogitization (densification) of subducted and tectonically buried crust. Journal of Geophysical Research 107, B 10 2252. doi: 10.1029/2001JB 000527. Bodart DE (1968) On the paragenesis of albitites. Norsk Geologisk Tidsskrift, 48, 269–80. Bowman JR, Valley JW, Kita NT (2009) Mechanisms of oxygen isotopic exchange and isotopic evolution of 18O ⁄ 16O-depleted periclase zone marbles in the Alta aureole, Utah: insights from ion microprobe analysis of calcite. Contributions to Mineralogy and Petrology, 157, 77–93. Bucholz CE, Ague JJ (2010) Fluid flow and Al transport during quartz-kyanite vein formation, Unst, Shetland Islands, Scotland. Journal of Metamorphic Geology, 28, 19–39. Carmichael DM (1969) On the mechanism of prograde metamorphic reactions in quartz-bearing pelitic rocks. Contributions to Mineralogy and Petrology, 20, 244–67. Carpenter MA, Ferry JM (1984) Constraints on thermodynamic mixing properties of plagioclase feldspars. Contributions to Mineralogy and Petrology, 87, 138–48. Cartwright I (1997) Permeability generation and resetting of tracers during metamorphic fluid flow: implications for advectiondispersion models. Contributions to Mineralogy and Petrology, 129, 198–208. Clark C, Schmidt-Mumm A, Faure K (2005) Timing and nature of fluid flow and alteration during mesoproterozoic shear zone formation, Olary domain, South Australia. Journal of Metamorphic Geology, 23, 147–64. Cole DR, Larson PB, Riciputi LR, Mora CI (2004) Oxygen isotope zoning profiles in hydrothermally altered feldspars: estimating the duration of water-rock interaction. Geology, 32, 29–32.
268 A. PUTNIS & H. AUSTRHEIM Cox SF, Etheridge MA (1989) Coupled grain-scale dilatancy and mass transfer during deformation at high fluid pressures: examples from Mt Lyell, Tasmania. Journal of Structural Geology, 11, 147–62. Davies JH (1999) The role of hydraulic fractures and intermediate-depth earthquakes in generating subduction-zone magmatism. Nature, 398, 142–5. Dodson MH (1973) Closure temperature in cooling geochronological and petrological systems. Contributions to Mineralogy and Petrology, 40, 259–74. Elliott R.B. (1966) The association of amphibolite and albitite, Kragero, south Norway. Geological Magazine, 103, 1–7. Engvik AK, Putnis A, Fitz Gerald JD, Austrheim H (2008) Albitisation of granitic rocks: the mechanism of replacement of oligoclase by albite. Canadian Mineralogist, 46, 1401–15. Engvik AK, Golla-Schindler U, Berndt J, Austrheim H, Putnis A (2009) Intragranular replacement of chlorapatite by hydroxyfluor-apatite during metasomatism. Lithos, 112, 236–46. Etheridge MA, Wall VJ, Vernon RH (1983) The role of the fluid phase during regional metamorphism and deformation. Journal of Metamorphic Geology, 1, 205–26. Ferry JM (1994) A historical review of metamorphic fluid flow. Journal of Geophysical Research, 99, 15487–98. Ferry JM (2000) Patterns of mineral occurrence in metamorphic rocks. American Mineralogist, 85, 1573–88. Ferry JM, Gottschalk M (2009) The effect of fluid salinity on infiltration-driven contact metamorphism of carbonate rocks. Contributions to Mineralogy and Petrology, 158, 619–36. Ferry JM, Sorensen SS, Rumble D (1998) Structurally controlled fluid flow during contact metamorphism in the Ritter Range pendant, California, USA. Contributions to Mineralogy and Petrology, 130, 358–78. Ferry JM, Wing BA, Penniston-Dorland SC, Rumble D (2002) The direction of fluid flow during contact metamorphism of siliceous carbonate rocks: new data for the Monzoni and Predazzo aureoles, northern Italy, and a global review. Contributions to Mineralogy and Petrology, 142, 679–99. Fyfe WS, Price NJ, Thompson AB (1978) Fluids in the Earth’s Crust. Elsevier, New York, 383 pp. Glodny J, Bingen B, Austrheim H, Molina JF, Rusin A. (2002) Precise eclogitization ages deduced from Rb ⁄ Sr mineral systematics: the Maksyutov complex, Southern Urals, Russia. Geochimica Cosmochimica Acta, 66, 1221–35. van Haren JLM, Ague JJ, Rye DM (1996) Oxygen isotope record of fluid infiltration and mass transfer during regional metamorphism of pelitic schist, Connecticut, USA. Geochimica Cosmochimica Acta, 60, 3487–504. Holland TJB (1980) The reaction albite = jadeite + quartz determined experimentally in the range 600–1200C. American Mineralogist, 65, 129–34. Holness MB (2003) Growth and albitisation of K-feldspar in crystalline rocks in the shallow crust: a tracer for fluid circulation during exhumation. Geofluids, 3, 89–102. Ho¨velmann J, Putnis A, Geisler T, Schmidt BC, Golla-Schindler U (2010) The replacement of plagioclase feldspars by albite: observations from hydrothermal experiments. Contributions to Mineralogy and Petrology, 159, 43–59. Ingebritsen SE, Manning CE (2003) Implications of crustal permeability for fluid movement between terrestrial fluid reservoirs. Journal of Geochemical Exploration, 78-79, 1–6. Iyer K, Austrheim H, John T, Jamtveit B (2008) Serpentinization of the oceanic lithosphere and some geochemical consequences: constraints from the Leka Ophiolite Complex, Norway. Chemical Geology, 249, 66–90.
Jamtveit B, Bucher-Nurminen K, Austrheim H (1990) Fluid controlled eclogitisation of granulites in deep crustal shear zones, Bergen Arcs, Western Norway. Contributions to Mineralogy and Petrology, 104, 184–93. Jamtveit B, Bucher-Nurminen K, Stijfhoorn DE (1992a) Contact metamorphism of layered shale-carbonate sequences in the Oslo rift: I. Buffering, infiltration, and the mechanisms of mass transport. Journal of Petrology, 33, 377–422. Jamtveit B, Grorud HF, Bucher-Nurminen K (1992b) Contact metamorphism of layered carbonate-shale sequences in the Oslo Rift. II. Migration of isotopic and reaction fronts around cooling plutons. Earth and Planetary Science Letters, 114, 131–48. Jamtveit B, Putnis CV, Malthe-Sørenssen A (2009) Reaction induced fracturing during replacement processes. Contributions to Mineralogy and Petrology, 157, 127–33. Janssen A, Putnis A, Geisler T, Putnis CV (2008) The mechanism of experimental oxidation and leaching of ilmenite in acid solution. Proceedings of the Ninth International Congress for Applied Mineralogy, 9–12 September, 503–6, Brisbane, Australia. John T, Schenk V (2003) Partial eclogitisation of gabbroic rocks in a late Precambrian subduction zone (Zambia): prograde metamorphism triggered by fluid infiltration. Contributions to Mineralogy and Petrology, 146, 174–91. John T, Scherer E, Haase KM, Schenk V (2004a) Trace element fractionation during fluid-induced eclogitization in a subducting slab: trace element and Lu-Hf ⁄ Sm-Nd isotope systematics. Earth and Planetary Science Letters, 227, 441–56. John T, Schenk V., Mezger K, Tembo F (2004b) Timing and P–T evolution of whiteschist metamorphism in the Lufilian Arc– Zambesi Belt orogen (Zambia): implications for the assembly of Gondwana. Journal of Geology, 112, 71–90. Jo¨ns N, Schenk V (2004) Petrology of whiteschists and associated rocks at Mautia Hill (Tanzania): fluid infiltration during highgrade metamorphism? Journal of Petrology, 45, 1959–81. Lee JI, Lee YI (1998) Feldspar albitisation in Cretaceous non-marine mudrocks, Gyeongsang Basin, Korea. Sedimentology, 45, 745–54. Lee MR, Thompson P, Poeml P, Parsons I (2003) Peristeritic plagioclase in North Sea hydrocarbon reservoir rocks: implications for diagenesis, provenance and stratigraphic correlation. American Mineralogist, 88, 866–75. Lyubetskaya T, Ague JJ (2009) Modeling the magnitudes and directions of regional metamorphic fluid flow in collisional orogens. Journal of Petrology, 50, 1505–31. MacDonald AH, Fyfe WS (1985) Rate of serpentinisation in seafloor environments. Tectonophysics, 116, 123–35. Mueller RF (1967) Mobility of elements in metamorphism. Journal of Geology, 75, 565–82. Munz IA, Wayne D, Austrheim H (1994) Retrograde fluid infiltration in the high-grade Modum Complex, South Norway: evidence for age, source and REE mobility. Contributions to Mineralogy and Petrology, 116, 32–46. Nakamura M, Watson EB (2001) Experimental study of aqueous fluid infiltration into quartzite: implications for the kinetics of fluid redistribution and grain growth driven by interfacial energy reduction. Geofluids, 1, 73–89. Niedermeier DRD, Putnis A, Geisler T, Golla-Schindler U, Putnis CV (2009) The mechanism of cation and oxygen isotope exchange in alkali feldspars under hydrothermal conditions. Contributions to Mineralogy and Petrology, 157, 65–76. Oliver NHS (1996) Review and classification of structural controls on fluid flow during regional metamorphism. Journal of Metamorphic Geology, 14, 477–92. Oliver NHS, Bons PD (2001) Mechanisms of fluid flow and fluid– rock interaction in fossil metamorphic hydrothermal systems
Fluid-induced processes 269 inferred from vein-wallrock patterns, geometry and microstructure. Geofluids, 1, 137–62. Oliver NHS, Rawling TJ, Cartwright I, Pearson PJ (1994) Hightemperature fluid–rock interaction and scapolitisation in an extension-related hydrothermal system, Mary Kathleen, Australia. Journal of Petrology, 35, 1455–91. O’Neil JR (1977) Stable isotopes in mineralogy. Physics and Chemistry of Minerals, 2, 105–23. O’Neil JR, Taylor HP (1967) The oxygen isotope and cation exchange chemistry of feldspars. American Mineralogist, 52, 1414–37. Parsons I, Lee MR (2009) Mutual replacement reactions in alkali feldspars I: microtextures and mechanisms. Contributions to Mineralogy and Petrology, 157, 641–61. Perez R, Boles AR (2005) An empirically derived kinetic model for albitization of detrital plagioclase. American Journal of Science, 305, 312–43. Philpotts A, Ague JJ (2009) Principles of Igneous and Metamorphic Petrology. University Press, Cambridge. Putnis A (2002) Mineral replacement reactions: from macroscopic observations to microscopic mechanisms. Mineralogical Magazine, 66, 689–708. Putnis A (2009) Mineral replacement reactions. MSA Reviews in Mineralogy and Geochemistry, 70, 87–124. Putnis CV, Mezger K (2004) A mechanism of mineral replacement: isotope tracing in the model system KCl-KBr-H2O. Geochimica Cosmochimica Acta, 68, 2839–48. Putnis A, Putnis CV (2007) The mechanism of reequilibration of solids in the presence of a fluid phase. Journal of Solid State Chemistry, 180, 1783–6. Putnis CV, Tsukamoto K, Nishimura Y (2005) Direct observations of pseudomorphism: compositional and textural evolution at a fluid-solid interface. American Mineralogist, 90, 1909–12. Putnis CV, Geisler T, Schmid-Beurmann P, Stephan T, Giampaolo C (2007) An experimental study of the replacement of leucite by analcime. American Mineralogist, 92, 19–26. Rossi M, Rolland Y, Vidal O, Cox SF (2005) Geochemical variations and element transfer during shear-zone development and related episyenites at middle crust depths: insights from the Mont Blanc granite (French-Italian Alps). In: High-Strain Zones: Structure and Physical Properties (eds Bruhn D, Burlini L). Geological Society, London, Special Publications, 245, 373–96. Rossi M, Rolland Y, Vidal O (2007) Evidence for crystal scale fluid infiltration during the Alpine Orogeny. Geophysical Research Abstracts, 9, 06620. Rumble D (1994) Water circulation in metamorphism. Journal of Geophysical Research, 99, 15499–502. Rumble D, Spear FS (1983) Oxygen-isotope equilibration and permeability enhancement during regional metamorphism. Journal of the Geological Society London, 140, 619–28. Saigal GC, Morad S, Bjørlykke K, Egeberg PK, Aagaard P (1988) Diagenetic albitization of detrital K-feldspar in Jurassic, Lower Cretaceous, and Tertiary clastic reservoir rocks from offshore Norway; I, Textures and origin. Journal of Sedimentary Petrology, 58, 1003–13. Schreyer W (1977) Whiteschists: their compositions and pressure– temperature regimes based on experimental, field, and petrographic evidence. Tectonophysics, 43, 127–44. Shervais JW, Kolesar P, Andreasen K (2005) A field and chemical study of serpentinization – Stonyford, California: chemical flux and mass balance. International Geology Review, 47, 1–23. Snow JE, Dick HJB (1995) Pervasive magnesium loss by marine weathering of peridotite. Geochimica Cosmochimica Acta, 59, 4219–35.
Straume AK, Austrheim H (1999) The importance of fracturing during retro-metamorphism of eclogites. Journal of Metamorphic Geology, 17, 637–52. Thompson AB, Connolly JAD (1992) Migration of metamorphic fluid – some aspects of mass and heat transfer. Earth Science Reviews, 32, 107–21. Vance D, Mu¨ller W, Villa IM (2003) Geochronology: linking the isotopic record with petrology and textures – an introduction. Geological Society, London, Special Publications, 220, 1–24. Vernon RH, Clarke GL (2008) Principles of Metamorphic Petrology. University Press, Cambridge, 446 pp. Vernon RH, White RW, Clarke GL (2008) False metamorphic events inferred from misinterpretation of microstructural evidence and P-T data. Journal of Metamorphic Geology, 26, 437– 49. Villa IM (1997) Isotopic closure. Terra Nova, 10, 42–7. Villa IM (2006) From nanometer to megameter: isotopes, atomicscale processes, and continent-scale tectonic models. Lithos, 87, 155–73. Viti C, Mellini M, Rumori C (2005) Exsolution and hydration of pyroxenes from partially serpentinised harzburgites. Mineralogical Magazine, 69, 491–507. Walther JV, Wood BJ (1984) Rate and mechanism in prograde metamorphism. Contributions to Mineralogy and Petrology, 88, 246–59. Wheeler J (1987) The significance of grain-scale stresses in the kinetics of metamorphism. Contributions to Mineralogy and Petrology, 97, 397–404. Wilbur DE, Ague JJ (2006) Chemical disequilibrium during garnet growth: Monte Carlo simulations of natural crystal morphologies. Geology, 34, 689–92. Winter JD (1998) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall, New Jersey. Wintsch RP (1985) The possible effects of deformation on chemical processes in metamorphic fault zones. In: Metamorphic Reactions. Kinetics, Textures and Deformation (ed Thompson AB, Rubie DC), pp. 251–68. Springer, Berlin. Wintsch RP, Yi K (2002) Dissolution and replacement creep: a significant deformation mechanism in crustal rocks. Journal of Structural Geology, 24, 1179–93. Wood BJ, Walther JV (1983) Rates of hydrothermal reactions. Science, 222, 413–15. Xia F, Brugger J, Chen G, Ngothai Y, O’Neill B, Putnis A, Pring A (2009a) Mechanism and kinetics of pseudomorphic mineral replacement reactions: a case study of the replacement of pentlandite by violarite. Geochimica Cosmochimica Acta, 73, 1945– 69. Xia F, Brugger J, Ngothai Y, O’Neill B, Chen G, Pring A (2009b) Three dimensional ordered arrays of nanozeolites with uniform size and orientation by a pseudomorphic coupled dissolutionreprecipitation replacement route. Crystal Growth and Design, 9, 4902–6. Yardley BWD (2009) The role of water in the evolution of the continental crust. Journal of the Geological Society, London, 166, 585–600. Yuan X, Sobolev SV, Kind R, Oncken O, Bock G, Asch G, Scurr B, Graeber F, Rudloff A, Hanka W, Wylegalla K, Tibi R, Haberland Ch, Rietbock A, Giese P, Wigger P, Ro¨wer P, Zandt G, Beck S, Wallace T, Pardo M, Comte D. (2000) Subduction and collision processes in the Central Andes constrained by converted seismic phases. Nature, 408, 958–61. Zack T, John T (2007) An evaluation of reactive fluid flow and trace element mobility in subducting slabs. Chemical Geology, 239, 199–216.
Fluid flows and metal deposition near basement ⁄cover unconformity: lessons and analogies from Pb–Zn–F–Ba systems for the understanding of Proterozoic U deposits M.-C. BOIRON, M. CATHELINEAU AND A. RICHARD G2R, Nancy Universite´, CNRS, CREGU, Vandoeuvre le`s Nancy, France
ABSTRACT Fluid circulation at basement ⁄ cover unconformities is of first importance for metal transfer and especially the formation of Pb–Zn, F, Ba and U-deposits. This is typically the case for world-class Proterozoic U deposits (Canada, Australia, Gabon) in basins, which show many similarities with younger Pb–Zn–F–Ba systems (Irish Paleozoic Pb–Zn deposits, F–Pb–Zn–Ba deposits related to extensional tectonics from Spain, western France and Silesia and fluid movements related to continental rifting in the Rhine graben). As fluid mixing near the basement ⁄ cover unconformity is one of the key factors for ore formation, a series of parameters have been considered for both systems: the time gap between basin formation and metal deposit, the origin and nature of the ore fluids, the temperature of fluid end members and the style of migration. Results show great similarities in all fluid systems: (i) a wide range of fluid salinity indicating the lack of homogeneity of fluid chemistry at the scale of the reservoirs, (ii) the deep penetration of brines through faults and dense networks of microfractures within the basement below the unconformity, (iii) local fluid–rock interaction leading to porosity increase and significant fluid changes in fluid chemistry, (iv) a pulsatory fluid regime during fluid trapping, (v) anisothermal fluid mixing revealed by a systematic temperature gap between brines and recharge fluids, (vi) stages of fluid movements facilitated by discontinuous opening related to later tectonic ⁄ telogenetic stages linked to major geodynamic events, typically without related sedimentation and burial (exception in a few cases characterized by the synchronous production and penetration of surface brines and ore genesis). By analogy with younger systems, the conditions of burial and penetration of brines in the Archean basement suggest that thermal convection drove the brine movements, and was possibly linked to extensional tectonics in a part of the giant mid-Proterozoic U-deposits. Key words: brines, extensional tectonics, fluid movement, metal, unconformity Received 28 August 2009; accepted 2 March 2010 Corresponding author: Marie-Christine Boiron, G2R, Nancy-Universite´, CNRS, CREGU, Boulevard des Aiguillettes, B.P. 70239, 54506 Vandœuvre-le`s-Nancy, France. Email:
[email protected]. Tel: +33 3 83 68 47 30. Fax: +33 3 83 68 47 01. Geofluids (2010) 10, 270–292
INTRODUCTION Fluid circulation near the unconformity between sedimentary cover and underlying basement is of first importance for mass transfer and especially for the genesis of Pb–Zn, F, Ba, Ag and U-ore deposits. In most cases, either the present-day depth of the unconformity is at a few hundred metres to a few kilometres depth, indicating the loss of a significant part of the sedimentary formations, or it is no longer present at all. Such a loss of information about the overlying basin is a major obstacle to the interpretation of the fluid and mass transfer processes, including the most critical stages of ore formation. This is typically the case for
the world – class U deposits hosted by Proterozoic basins. Such deposits occur close to the unconformity (U ⁄ C) between a Proterozoic sandstone cover and an Archean to Lower Proterozoic metamorphic basement. Major host basins are known from Saskatchewan, Canada (Athabasca Basin), Northern Territories, Australia (McArthur Basin, Alligator River Uranium Field) and Oklo, Gabon. In all cases, only a minor part of the basin is preserved, while at least 5–7 km from above the unconformity has been eroded. Although, the nature of the uranium ore-forming process has been investigated since more than 30 years several aspects remain uncertain (references in Kyser & Cuney
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
Fluid flows and metal deposition near unconformity 271 2008). Genetic models invoke large-scale circulation of basinal fluids and U uptake either from monazite-zircon paleoplacers in Proterozoic sandstones, or from the basement, this yielding to two contrasted models: (i) direct penetration of basinal fluids into the basement, with U reduction occurring at the contact with reducing lithologies and ⁄ or reducing fluids (Hoeve & Sibbald 1978), (ii) penetration of brines into the basement, U uptake in the basement itself and U deposition accompanying fluid mixing in the vicinity of major faults (Richard et al. 2010). There are several uncertainties in the genetic models of deposition, in particular in the relative timing and relationships between U deposition and geodynamic processes, the extent of the fluid movements, and the relative role of water rock interaction vs. fluid mixing for the uranium deposition. Several features of the fluid regime during formation of U deposits display similarities with fluid regimes reported from MVT-like deposits (Leach et al. 2005) or more generally carbonate hosted Pb–Zn ores and F–Ba deposits in the vicinity of unconformities. For example, the nature of the fluids, in particular the presence of high salinity, Na–Ca brines with variable Na ⁄ Ca ratios (Derome et al. 2003, 2005, 2007; Richard et al. 2010), the evidence of brine mixing, the penetration of the sedimentary brines in the basement and the localisation of fluid migration near the unconformity. Although such an analogy has not been proposed previously, the similarities between rather old, Proterozoic U deposits and more recent fluid–rock systems in particular Mesozoic Pb–Zn–F–Ba deposits from western Europe have encouraged us to develop a comparative study between the two families of deposits. Understanding of the Pb–Zn–F–Ba deposits has been improved by the direct study of paleofluid chemistry, and the consideration of the time relationships between burial, diagenesis, fluid circulation related to metal deposition and the geodynamic context. These data, generally dispersed in the published reports, or gathered together only for specific types of deposits, have been compiled in this study and are compared for several major deposits from western Europe: Irish Paleozoic Pb–Zn deposits, F–Pb–Zn–Ba deposits related to extensional tectonics, associated with the Atlantic ocean and Gascogne rifting, in Spain, western France and Silesia). In addition, flow in the Rhine graben intracontinental rift, where Triassic formations and their basement have been affected by large-scale fluid flow from Oligocene times to the present day, was also considered (Table 1). In addition, contributions to the understanding of fluid mixing were also considered in this comparative work. The geodynamic context and style of fluid movement for these younger systems are now much better known than for Proterozoic systems and appear better constrained. The main goal of this article is thus to compare Proterozoic ore systems in which basins host
giant U-deposits (the Alligator River Uranium Field in Australia, the Athabasca basin in Canada and the Oklo deposit in Gabon) with the younger Pb–Zn–F–Ba ore systems for which the geodynamic context, and style of fluid movement are better known. The objective was to get key information and models on ore fluid movements in the upper lithosphere and to evaluate their applicability to U deposits.
GIANT U DEPOSITS RELATED TO THE UNCONFORMITY BETWEEN PROTEROZOIC BASINS AND THEIR ARCHEAN ⁄ LOWER PROTEROZOIC BASEMENT The discovery in the 1970–1990s of several giant U deposits near the unconformity between Proterozoic basins and their basement, in particular in the Athabasca Basin, Saskatchewan, (Canada) and in the Alligator River Uranium Field (ARUF), McArthur Basin, Northern Territories, (Australia), has encouraged a systematic evaluation of the nature of the ore fluids and of their origin. Such deposits occur close to the unconformity between a Proterozoic sandstone cover and an Archean to Lower Proterozoic metamorphic basement (see references in the synthesis by Kyser & Cuney 2008). Genetic models invoke large-scale circulation of Na–Ca-rich basinal fluids (Pagel 1975; Pagel & Jaffrezic 1977; Pagel et al. 1980; Derome et al. 2003, 2005, 2007; Richard et al. 2010). There are several important uncertainties in the general model of deposition, and in particular in the timing of U deposition. For example, in Canada, between 1.7 and 1.8 Ga, the ages for the sedimentation and the Grenville orogen, a few geodynamic events have occurred but without clear effects on the Athabasca basin, which is located away from the major deformation zones. Since the 1980s, ages of the U-ores from Proterozoic basins have become successively older, from 1.1 Ga to 1.6 Ga, because of the improvements to analytical techniques which have made possible the analysis of smaller amounts of ore, thereby avoiding possible Pb-loss because of coffinite alteration or uraninite recrystallisation (Alexandre et al. 2009). Thus, in Canada, depending on the age of the U-deposit, the Grenville, Berthoud (Sims & Stein 2003), or Mazatzal (Amato et al. 2008) orogens have been successively considered as the cause of fluid movements or more recently as the cause of recrystallisation of primary ores (Alexandre et al. 2009). Difficulties in relating precisely fluid events to geodynamic events characterized by a specific fluid signature and migration style, complicate the proposition of a metallogenic model. Alligator River Uranium Field (Australia) The Kombolgie sub-basin (northern Territories, Australia) overlies a basement consisting of Archean and Early
Rhine Graben Soultz sous Foreˆts France
Schwarzwald district Germany
North Western French Massif Central
Albigeois South French Massif Central
Malines, Cevennes (South Eastern French Massif Central)
Maestrat basin Eastern Spain
Catalan coastal range North Eastern Spain
Asturias, Spain
Upper Silesia, Poland
Silvermines District Ireland
F–Ba
Pb–Zn–F–Ba
Pb–Zn–F–Ba
F–Ba
Pb–Zn
Pb–Zn
F–Ba
F–Ba
Pb–Zn
Pb–Zn
Location
Trias dolomitized limestones Lower carboniferous limestones
Late Jurassic to early Creataceous sediments Triassic carbonates and evaporites JurassicCreatceous limestones Permo - Triassic series
Triassic calcareous shales silt, evaporite + Jurassic series
Eroded
Marine infra-Lias to Dogger series
Eocene to Oligocene marls and clays Triassic sediments Triassic sandstones Marine carbonate rich sediments
Sedimentary cover
Cambrian to Devonian basement Carboniferous basement Low grade paleozoic metasediments
Granodiorite 287 ± 2.8 Ma (5)
Paleozoic basement
Lower and middle Cambrian dolomitic limestones
Hercynian monzogranites (331 ± 9 Ma) (1) Albtal and Malsburg granites 326 ± 2 Ma and 328 ± 6 Ma (2,3) Hercynian tonalites to granodioirte 350 to 360 Ma (4) Peraluminous intrusions (320 Ma) Paleozoic basement
Basement
Mississipian
Trias
Permo-Trias
Late Cretaceous Early Paleocene Trias
Trias
Trias Hettangian Continental basin
Trias at 200 km Hetttangian
Keuper Trias
Keuper Trias
Age of evaporite
Table 1. Geological setting, geodynamic context and ages of the host formations and mineralisations in the considered areas.
350 (see text for references)
185 ± 28 (Sm–Nd) (8) 135 ± 4 (Rb–Sr) (9)
137 ± 25 (Sm– Nd)(7)
150 (K–Ar) (18) 111 ± 13 (Sm-Nd) (19) at least post Hettangian probably Kimeridgian 62 ± 0.7 (U–Pb) (6)
140–155 (K–Ar, Ar–Ar) (see text for references)
Jurassic (110–150) Tertiary (see text for references)
20 to present day
Ages of mineralisation (Ma)
0
80 Ma
>20 Ma
50–60 Ma
0
55 Ma (Trias) 340 (Cambrian)
40 to 50 Ma
40 to 50 Ma
70 to >130 Ma
180 to 200 Ma
Time gap between sedimentation and fluid movement
Local extensional rift zone or extending tectonics passive margin
Extension Atlantic Ocean opening Pemian grabben
Extension Atlantic Ocean opening
Sligtly post-rift Gascogne gulf
Extensional tectonics
Atlantic Ocean extentional tectonics
Atlantic Ocean extensional tectonics
Oceanic extension - Jurassic Continental extension Tertiary Oceanic extension - Jurassic Continental extension Tertiary
Geodynamic context
272 M.-C. BOIRON et al.
Francevillian
around 1.7 Ga
Francevillian series, Conglomerate to fine sandstones (2100 Ma) (14) U
Athasbaca basin Canada Franceville basin Oklo, Gabon U
1 – Alexandrov et al. (2001), 2 – Schuler & Steiger (1978), 3 – Todt (1976), 4 – Bertrand et al. (2001), 5 – Enrique et al. (1999), 6 – Grandia et al. (2003a), 7 – Pique´ et al. (2008), 8 – Sanchez et al. (2010), 9 – Heijlen et al. (2003), 10 – Ludwig et al. (1987), 11 – Maas (1989), 12 – Cumming & Krstic (1992), 13 – Alexandre et al. (2009), 14 – Stille et al. (1993), 15 – Bonhomme et al. (1982), 16 – Gancarz (1978), 17 – Holliger (1988), 18 – Bonhomme et al. (1987), 19 – Munoz et al. (2005).
continental extension 50–200 Ma
see discussion 200–300 Ma
see discussion 100–200 Ma
1737 ± 20 (U–Pb) (10) 1600–1650 (U–Pb) (11) 1400–1500 (U–Pb) (12, 13) 1960 to 2050 (U– Pb) (16, 17) around 1.7 Ga
Archean to early Proterozoic granito gneiss Paleoproterozic gneiss and pegmatoids. Archean K-rich plutonites 2880 to 2400 Ma (15) Kombolgie sandstones and conglomerates marine origin Proterozoic sandstones Alligator River Australia U
Location
Table 1. Continued.
Sedimentary cover
Basement
Age of evaporite
Ages of mineralisation (Ma)
Time gap between sedimentation and fluid movement
Geodynamic context
Fluid flows and metal deposition near unconformity 273 Proterozoic granito-gneisses (Nanambu Complex), the Kakadu Group gneisses and quartzites, and the metasediments of the epicontinental Cahill Formation. All these rocks underwent the Barramundi and the Top End orogenies between 1870 and 1800 Ma (Page & Williams 1988). A thick (up to 1800 m presently preserved) unmetamorphosed detrital sequence composed of fluvial and aeolian sandstones (the ‘Kombolgie Sub-Group’) unconformably overlies the crystalline and metamorphic basement and is dated at 1822 and 1720 Ma (Sweet et al. 1999). The youngest members of the Kombolgie Sub-group were deposited in marine conditions, as suggested by the occurrence of glauconite and halite crystal casts (Kyser et al. 2000) and by stromatolitic and evaporitic dolostone-sandstone occurring in the McKay formation (Rawlings 1999). All uranium deposits are located in basement rocks with ages between 1600 and 1650 Ma for the Jabiluka, Nabarlek and Koongarra deposits (Maas 1989; Sm–Nd, Rb–Sr data on uraninites, references in Kyser & Cuney 2008). The Kombolgie basin and its basement are characterised by hydraulic breccias, frequently developed along faults. Breccias consist of angular blocks of sandstones cemented by large euhedral drusy quartz crystals, up to 10 cm at their base. Inclusions of two types of brines are present in the drusy quartz crystals from sealed faults (Table 2), together with a low salinity fluid showing clear evidence of mixing with the brines (Derome et al. 2003). The two brines consist of: (i) a Na-rich fluid with ice melting temperatures (Tm ice) ranging from )15 to )35C (mode around )25C) yielding a chlorinity close to 5 mol ⁄ kg H2O, and a homogenisation temperature (Th) between 90 and 160C (mode: 135C). Na ⁄ Ca ratios determined by Laser Induced Breakdown Spectroscopy (LIBS) are around 8; (ii) a Ca-rich fluid characterised by Tm ice ranging from )35 to )52C (mode at )42C and chlorinity between 5.6 and 6 mol ⁄ kg H2O) and Th between 100 and 140C (mode 110C). Na ⁄ Ca ratios are closed to 0.1. The low to moderate salinity fluids display Tm ice in the range of )0.3 to )10C (mode at )2.5C) and Th between 115 and 155C with a mode at 140C. Na ⁄ Ca ratios are variable and in the range of 1–10. The Athabasca uranium deposits The Athabasca basin lies on Archean to Paleoproterozoic basement which consists of gneisses, metapelites and mafic to felsic intrusions affected by the Thelon–Talston (2.0– 1.9 Ga) and Trans-Hudson (1.9–1.7 Ga) orogenies. The sedimentary sequence is now around 1500 m thick at maximum but is thought to have originally reached 5 km, based on pressure–temperature estimation from fluid inclusion studies (Pagel 1975; Derome et al. 2005). The sandstones as well as the basement display local quartz dissolution and precipitation and a major Mg-silicate alter-
274 M.-C. BOIRON et al. Table 2. Summary of paleofluids data available on the different areas. d18O fluid (& )
d D fluid (& )
)3 to 2
)37 to )42
)11. 6 to )3
)10 to )60
5 to 9
)19 to )50
Dubois et al. (1996) Smith et al. (1998) Cathelineau & Boiron (2010) Pauwels et al. (1993) Baatartsogt et al. (2007) Boiron et al. (2002)
Not available
0.1 to 3.2
)15 to )63
Munoz et al.(1999)
150
Not available
5. 5 to 7. 8
)13 to )31
Charef & Sheppard (1988)
)1 to )26
70–130
0 to 3
Not available
)4.9 to )21. 6
85–195 (110)
550–700 1100–2400 300–970 1000–2100
Not available
Not available
Grandia et al. (2003) Pique´ et al. (2008)
)0.1 to )24.3
80–170
0.3 to 7.4
Not available
0 to )23. 3
70–160
100–800 1000–8000 248–560
0 (to )3)
)10 to )30
12–18 wt %
170–240
600–770
Not available
Not available
)0.3 to )52
90–160
150–220
0.9 to 3.3
Not available
)20 to )55
70–195
100–500
)3 to 6
)55 to )147
)1.3 to )63
90–175
70–160
)6 to 0
Not available
CI ⁄ Br (molar)
Location
Tm ice (C)
Th (C)
F–Ba
Rhine Graben Soultz sous Foreˆts France
2000 m : )0.2–)31.6 5000 m : )0.1–)24.6
2000 m : 100–165 5000 m : 150–184
Pb–Zn–F–Ba
Schwarzwald district Germany North Western French Massif Central Albigeois South French Massif Central Malines, Cevennes (South Eastern French Massif Central) Maestrat basin Eastern Spain Catalan coastal range North Eastern Spain Asturias (Spain)
)0.1 to )36
80–200
Not available
)3. 8 to )30.2
80–130
160–1800
)16.7 to )20. 6
125–170
)10 to )16
Pb–Zn–F Ba
F– Ba
Pb–Zn
Pb–Zn F–Ba
F–Ba Pb–Zn Pb–Zn U U U
Upper Silesia, Poland Silvermines District Ireland Alligator River Australia Athasbaca basin Canada Franceville basin Oklo, Gabon
250–800*
References
Sanchez et al. (2009, 2010) Heijlen et al. (2003) Wilkinson et al. (2005) Derome et al. (2007) Richard (2009) Mathieu et al. (2000) Mathieu (1999)
*, unpublished data; Tm ice, melting temperature of ice; Th, homogenisation temperature.
ation assemblage (Mg-rich di-octahedral chlorite (sudoite), Mg tourmaline (dravite) and Mg-illite). Ages proposed for the mineralisation have been estimated in the range of 1510–1540 Ma (Cumming & Krstic 1992; Alexandre et al. 2009). In quartz and carbonate veins, two to three phase fluid inclusions are observed and document two brines: a sodium rich brine (Tm ice from )15 to )30C) and a calcium rich brine (Tm ice from –30 to )55C), which have circulated and mixed at the base of the basin and within the basement at the time of the formation of the uranium mineralisation, at temperature close to 150 ± 30C (Table 2). The calcium-rich brines are inferred to have formed by interaction of the sodium-rich fluid with the basement rocks. Both brines were derived from evaporated seawater, as shown by the low Cl ⁄ Br ratios, the Cl concen-
trations and other recent isotopic data including ome et al. 2005; Richard et al. 2009) (Fig. 1).
37
Cl (Der-
Olko natural reaction zones (Gabon) The Francevillian sedimentary series unconformably overlies the Archean crystalline basement represented by the north Gabon massif in the north and the Chaillu massif in the west (Gauthier-Lafaye 1986; Gauthier-Lafaye & Weber 1989). U ⁄ Pb, Rb ⁄ Sr and K ⁄ Ar age determinations of the granitoids (K-rich plutonites and diorites) from the Chaillu massif yield ages ranging from 2.88 to 2.4 Ga (Bonhomme et al. 1982). The Francevillian series is a 1–4 km thick, unmetamorphosed sequence of clastic and volcanoclastic sediments. The lithostratigraphic succession from the bot-
Fluid flows and metal deposition near unconformity 275
Fig. 1. Model of fluid migration around uranium deposit in Proterozoic series with brine penetration in the basement because of both gravity driven process and thermal convection (example of the Athabasca basin).
tom to the top consists of: (i) conglomerates and fine to coarse sandstones deposited in a fluvial to deltaic environment, (ii) fine-grained marine sediments, mainly pelites and black shales, (iii) massive dolomite and cherts interbedded with black shales, (iv) ignimbrites and epiclastic sandstones with interbedded shales (Fig. 2). Numerous quartz veins, often associated with wall-rock alteration, attest to significant fluid movements along fractures during diagenesis of the Franceville basin. Diagenesis of these formations resulted in a silicification of detrital quartz grains and formation of authigenic illite and chlorite in the matrix (Gauthier-Lafaye 1986; Gauthier-Lafaye & Weber 1989). Fluids were related to the formation of diagenetic minerals, generation of U ore deposits and the migration of hydrocarbons. The extensional fractures in the Archean basement and the sandstones are filled by a quartz–chlorite–calcite–sulphide and a U mineral assemblage (Mathieu et al. 2000). Mineralisations are dated around 1960–2050 Ma (Holliger 1988). Fluids circulating during the quartz–chlorite–calcite–sulphide and U mineral stage can be divided in three major types with the following characteristics (Mathieu et al.
Fig. 2. Fluid migration around the Oklo deposit showing the penetration of recharge fluids from relief zones west of the Franceville rift zone and the mixing of brines with recharge fluids along the fault zones at the edge of the basin, and near the contact between the FA ⁄ FB sedimentary formations (modified from Gauthier-Lafaye 1986).
2000): (i) high salinity Ca–Na-rich brines displaying large Tm ice variations but showing two groups from )63 to )46C (Ca dominated with Na ⁄ Ca ratio around 0.5) and from )36 to )27C (Na dominated with Na ⁄ Ca from 2 to 9) (Table 2). Th ranges from 95 to 155C. The highest salinity fluids display the lowest Th values; (ii) moderate salinity fluids with Tm ice in the range )24 to )14C and Th between 90 and 160C. These fluids are Na dominated with Na ⁄ Ca ratios from 2 to 5.5. (iii) Low salinity fluids having Tm ice in the range of )1.3 to )5.5C and Th between 100 and 175C. Na is the major cation found in these fluids using LIBS, Ca being detected in some inclusions yielding Na ⁄ Ca around 25.
BRINE MIGRATION RELATED TO OCEANIC RIFTING The Poitou High and Pb–F–Ba occurrences of the Northwestern Massif Central Significant fracture and porosity sealing characterises the sedimentary cover-basement interface in the north-western
276 M.-C. BOIRON et al. margin of the Aquitaine basin (France) close to the northwestern part of the French Massif Central, the so-called Poitou High area corresponding to the northern edge of the Aquitaine Basin. Below a ca. 150-m thick sedimentary cover consisting of marine Lias and Dogger formations, the basement lithology is dominated by plutonic rocks from the ‘Tonalitic Lineament’ of the Limousin (Peiffer 1986; Shaw et al. 1993) containing multiple intrusions of medium-K calc-alkaline tonalites and granodiorites dated around 350–360 ± 5 Ma (U–Pb on zircons; Bertrand et al. 2001). Dolomite, calcite, fluorite, barite and quartz constitute most of the fracture fillings attesting fluid circulation either in the sediments or in the basement (Boiron et al. 2002; Fourcade et al. 2002; Cathelineau et al. 2004). These processes (Fig. 3) are large scale as shown by the similarities of mineral sequences, fluid types and general features of most of the F–Ba–Pb–Zn deposits located at the basement-sedimentary cover interface all along the margins of the French Massif Central, (Ziserman 1980; Munoz et al. 1995, 1999) as well as of the U reworking at Le Bernardan (Patrier et al. 1997). Microthermometry and Raman microprobe investigations show the existence of a single dominant fluid type (H2O– NaCl–KCl–CaCl2 ± MgCl2) in all primary two-phase fluid inclusions from all the minerals of the infilling sequence. They show a range of final ice melting temperatures from )3.8 to )30.2C (Table 2). The highest measured chlorinity is around 5.2 mol ⁄ kg. The fluids are Na-dominated with a Na ⁄ Ca ratio around 4 ± 1 measured by LIBS. Homogenisation temperatures are between 65 and 130C for quartz and slightly lower for dolomite (90–115), barite (85– 115C), fluorite (85–100C) and calcite (65–100C) (Boiron et al. 2002). Albigeois fluorite deposits The most important fluorite district in France is considered to have formed during Mesozoic times (150 Ma, Bonhomme et al. 1987) in relation to a third stage of extensional tectonics occurring much after the main rifting of the
Atlantic ocean, and before the Gascogne gulf rifting, considered as Aptian-Albian. Munoz et al. (1999) put forward a model of continental brine circulation (salinity: 20–26 wt% eq. NaCl, Ca–(Mg)–Na brines and Th between 85 and 170C) (Table 2), penetrating deeply into the basement, with the formation of ores along faults where the fluorite deposited as a consequence of temperature drop. Although, the main ores precipitate from saline waters with d18O around 0–3.2& and dD in the range of )15 to )63&, these authors considered that isotopic features are derived from meteoric waters that underwent an isotopic shift by interaction with host rocks. This is the main argument in favour of the continental origin of the brines. It can be also noted that recent paleogeographic reconstruction and fission track data tend to indicate that the French Massif Central was in great part covered by the sea during Lias and Dogger times. The Malines Pb deposit The Malines Pb deposit in Cevennes (south-eastern part of the french Massif Central) is an interesting example with a complex history, subject of several controversies, because it is hosted in an old series of rocks (Cambrian in age) but affected by a series of thermal–diagenetic–hydrothermal events that were difficult to date. Most dates are relative and do not concern strictly the ores. Ambiguous textural relationships have been interpreted either as primary early diagenetic textures (karst infilling below the unconformity) or as epigenetic replacements. Clay ages (K–Ar method, Clauer & Chaudhuri 1995) are interpreted as synchronous with the main fluid stage and considered as representative of general diagenesis, not directly related to the ore deposition (Leach et al. 2001). Remagnetization ages (60–50 Ma) have been presented as related to chemical processes directly linked to the ore stage (Rouvier et al. 2001), but as indicated by Muchez et al. (2005), no definitive argument for a direct genetic link with the bulk of the Pb ores is given. The best relative chronological indicators are based on geology (cross-cutting relationships between
Fig. 3. Schematic model of the fluid circulation in the north-western part of the French Massif Central at the unconformity between the basement and sedimentary cover, showing the upward brine migration below the Toarcian rock and mixing in fault zones (based on data from Boiron et al. 2002).
Fluid flows and metal deposition near unconformity 277 ores and sedimentary formation) and tectonics, and indicate a post-Triassic age at a minimum, most probably early Jurassic (Macquar & Lagny 1981; Macquar et al. 1990; probably post-Hettangian (Lotharingian to Bathonian) following Charef & Sheppard 1988) for the main ore stage and probably Kimmeridgian as sulphide rich layers (sedex) are observed locally in Kimmeridgian formations (Ming-An et al. 1995). This does not preclude an earlier pre-concentration stage. The main ore stage is characterized by the mixing of a cool, dilute fluids, with a metal-rich brine (150C), considered to be a connate water related to the dewatering of the basin by Charef & Sheppard (1988), but which could also result from fluid–rock interaction within the anhydrite-gypsum Triassic series overlying the mineralised Cambrian formations (Table 3). As in many other occurrences from the southern Massif Central, the thickness of the sedimentary overburden cannot explain entirely the rather elevated temperatures, and Charef & Sheppard (1988) inferred the existence of localised thermal anomalies, in part caused by the fluid movements from deeper parts of the Rhoˆne basin in which the thickness of sediment reached up to 8 km. Mesozoic Fluorite veins (Spain) Fluorite veins developed in the Catalan costal range, in faults affecting a basement dominated by a series of calkalcaline granodiorite – monzogranite intrusions overlain by Triassic sediments including Keuper evaporites. Pique´ et al. (2008) inferred the deep circulation of brines, derived from the leaching of Mesozoic evaporites by seawater and ⁄ or meteoric waters, within the basement where brines took up metals and gained K, Ca, Sr and Ba. Resulting Na–Ca–Cl (K–Mg) brines are thought to have circulated during a late Jurassic-Early Cretaceous rifting stage (dated 137 Ma) related to extensional tectonic and to the opening of the Atlantic ocean (Pique´ et al. 2008). The authors note that this period is here characterised by elevated crustal temperatures which can explain the focused convection (Juez-Larre´ & Andriessen 2006). Although rich in metals, ore fluids deposited mainly fluorite, but were not able to deposit metals because of the S content of the fluids (Pique´ et al. 2008). The wide range of fluid inclusion chlorinities points to fluid mixing as an important cause of precipitation. The fluorite deposits of Asturias (Northern Spain) are hosted by Permo-Triassic and Paleozoic rocks. Mineralisation occurs in breccias and fractures and consists of fluorite, barite, calcite, dolomite and quartz with small amounts of sulphides. Mineralisation results from mixing of deep brines with surficial waters. Fluid movements and fluorite deposition occurred between late Triassic and late Jurassic times (Sm–Nd age on fluorite at 185 ± 28 Ma, Sanchez et al. 2010) in relation to rifting events associated
with the Atlantic Ocean opening. Downward penetration of brines into the basement, infiltration of surficial fluids and upwards brine migration into the Mesozoic basin margins are the major processes to explain the fluorite deposition in areas of fluid mixing (Sanchez et al. 2009). Pb–Zn deposits of the Catalan coastal range and Basque Cantabrian area, Spain In the Maestrat basin (Catalan coastal range, Eastern Spain), Zn–Pb deposits occurred in dolomitized Aptian limestones in the margins of the half-graben shape Penyagolosa sub-basin. Metal deposition is thought to be the result of mixing of low temperature–low salinity (although it reaches up to 15 wt% eq NaCl) resident fluids, enriched in sulphur and a high salinity brine at around 80–130C, supposed to be an evaporated seawater derived from sabkha-like environments in Late Cretaceous to Early Paleocene times, as suggested by the age of 62 ± 0.7 Ma for the mineralisation process (Grandia et al. 2000, 2003a). The small thickness of the series at that time would imply large thermal discrepancies between the entering of fluids and the series, and probably large convecting cells, perhaps extending more than 4 km from the surface to reach temperatures of 130C. In that case, the penetration and ascending migration of brines would be synchronous with the rifting stage (Fig. 4). In sediment hosted Zn–Pb deposits of the Basque Cantabrian basin in Northern Spain, fluid mixing has been demonstrated based on microthermometric data and Na–K–Li–Cl–Br systematics of fluid inclusions (Grandia et al. 2003b). Fluids of different origins were involved in the mineralising process: (i) brines (25 wt% eq. NaCl) resulting from seawater evaporation, (ii) brines derived by dissolving halite closed to salt domes, (iii) low salinity fluids (seawater or fresh water). Carbonate hosted Zn–Pb deposits of Upper Silesia, Poland Zn–Pb mineralisation occurs on both sides of a Permian graben in Devonian to Jurassic rocks, especially in dolomitised Muschelkalk (upper Triassic). More than 90% of the ores are hosted by an early diagenetic ⁄ epigenetic dolosparite, but the age of the mineralisation is considered to be younger than upper Triassic, and probably lower Cretaceous as suggested by an age of 135 ± 4 Ma obtained on sphalerite (Rb–Sr on ore stage sulphide (Heijlen et al. 2003). Ores formed from highly saline Na–Ca brines (20– 23 wt.% eq. NaCl, Table 2), thought to be evaporated seawater of Permian–Trias age which penetrated deeply into Carboniferous basement units where they drove albitisation and leached metals and Ca. The expulsion from the basement is inferred to be much later, corresponding to during the extension stages accompanying the opening of the
Schwarzwald district Germany North Western French Massif Central
Albigeois South French Massif Central Malines, Cevennes (South Eastern French Massif Central) Maestrat basin Eastern Spain Catalan coastal range North Eastern Spain Asturias, Spain
Pb–Zn–F–Ba
F–Ba
Silvermines District Ireland Alligator River Australia Athasbaca basin Canada
Franceville basin Oklo, Gabon
Pb–Zn
U X
X
X
X
X
X
X
X
X
X
(X)
(X)
(X)
X
X
X
Connate waters (Rhone basin)- role of Trias evaporites ?
X
Deep aquifer at 300–350C X
Probable
Hot dilute fluid
Ascending hot dilute fluid
Dilute in some places
Hot dilute fluid 170C at 2000 m 200C at 5000 m
Hot fluid
Secondary brine
Primary brine
Cooler S-rich brine
Resident fluid
Resident fluid
Resident fluid
Resident fluid
Cold fluid
X
X
X
X
X (Dolomite stage)
X
X (dilution)
X
X
X
X
X
Mixing
>0.3 km probably more Ca-Na exchange X recharge
Ca–Na exchange
Ca–K gains
Deep penetration Ca–Na exchange
XX
XX
X
A few km
>1 km
7–8 km
>5 km
Fluid in the basement
X indicates the presence of the considered process, XX indicates that the process is relatively important and (X) that the process is suspected. (Correction added on 6 May 2010, after first online publication: The legends of Table 1 and Table 3 were interchanged.)
U
U
Upper Silesia, Poland
Pb–Zn
F–Ba
F–Ba
Pb–Zn
Pb–Zn
Pb–Zn–F–Ba
Rhine Graben Soultz sous Foreˆts France
F–Ba
Location
Fluid recharge
Fluid origin
Table 3. Main characteristics of fluid circulation in the different areas (see Table 2 for references).
Sudoite - illitedravite (quartz dissolution) chlorite
Dolomitisation Albitisation (basement) Dolomitisation in limestones Chlorite
Dolomitisation Silicification
K-micas, chlorite
Dolomitisation
Adularisation (basement) Dolomitisation (infra-Lias) Silicification preceeding Fluorite deposition Dolomitisation
Illitisation
Illitisation
Water- rock interaction
dilution (+ redox processes)
Cooling, mixing of two brines including a S-rich brine Brine mixing, dilution (+ redox processes) Mixing of two brines (+ redox processes)
Mixing with less saline surficial derived fluid present in sediments Mixing Cooling
Dilution at constant temperature
Mixing of 2 brines
Dilution Cooling
Cooling
Dilution Cooling
Dilution
Dilution
Main cause of metal deposition
278 M.-C. BOIRON et al.
Fluid flows and metal deposition near unconformity 279
breccia
Fig. 4. Schematic model of the fluid circulation in the Maestrat Pb–Zn deposit (Spain) (modified from Grandia et al. 2003a).
Northern Atlantic ocean (Fig. 5). After Heijlen et al. (2003), brines (70–160C) are not considered to have migrated following a gravity driven flow model as originally suggested by Leach et al. (1996) but were expelled along deep-seated extensional faults, affecting both basement rocks and their overlying sedimentary basins, through a dilatational pumping mechanism following the seismic model of Muir Wood & King (1993).
IRISH PB–ZN DEPOSITS Irish deposits, which include Navan the largest Zn resource in Europe, are considered to have been formed by the deep penetration of evaporated seawater (Banks et al. 2002), which acquired metals, K and Ca by water–rock interaction in the continental crust at 5–10 km depth
(Russel 1978), and then mixed as they ascended with surficial brines rich in H2S. These brines acquired H2S because of bacterial activity within the lower Carboniferous limestones overlying the lower Paleozoic basement, in fractured zones (Blakeman et al. 2002; Wilkinson et al. 2005). This model emphasises the role of rift zones and local extensional tectonics in driving flow, and considers the topographic flow models proposed by Hitzman & Beaty (1996) as unsuitable for the Irish deposits. It puts forward also the role of thermo-haline density driven convection as the main motor for the fluid movement (Fig. 6). The age (350 Ma, in Wilkinson et al. 2005) of these deposits is in favour of a model of brine circulation nearly coeval with their formation with Mississipian evaporites which are present in the sedimentary sequence, a feature at variance of most other deposits presented in our comparative study. The relationships between the process of seawater evaporation and the heating mechanism of the surface brine is however not fully understood in terms of paleogeography, i.e. the extent of the Mississipian sea controlling the area and volume of evaporated seawater formed at that time and available to sink beneath the surface, and the depth– temperature relationships, including the temperatures reached near the marine sediment–seawater interface.
160 °C
INTRACONTINENTAL RIFTING: THE SOULTZSOUS-FOREˆ TS GEOTHERMAL AREA, RHINE GRABEN (EASTERN FRANCE)
Fig. 5. Fluid migration in Silesian Pb–Zn deposits (Poland) (modified from Heijlen et al. 2003).
The Soultz-sous-Foreˆts geothermal plant is located in the Rhine Graben which constitutes part of the continental rift extending from Frankfurt (Germany) to Basel (Switzerland). The Rhine graben formed during the Oligocene by large-scale extension related to mantle upwelling (Rousset
280 M.-C. BOIRON et al.
Fig. 6. Model of fluid circulation occurring in the Irish carbonate hosted Pb–Zn deposits (modified from Wilkinson et al. 2005).
et al. 1993). The north-western part of the Rhine Graben is filled by an asymmetric sedimentary cover ranging from Triassic to Oligocene in age and overlying Hercynian fractured granitic basement. The sedimentary cover is dominated by a thick Tertiary series, mainly represented by marls and clays. It overlies Triassic sediments which are subdivided into the three classical lithofacies: Keuper, Muschelkalk and about 300 m of Buntsandstein sandstones (Fig. 7). The main granite body is a Hercynian monzogranite dated at 331 ± 9 Ma (U–Pb method, Alexandrov et al. 2001) with no equivalent at the surface. At Soultz-sous-Foreˆts, past to recent fluid movements are recorded by series of healed or sealed micro- to macrostructures, which provide evidence of a rather continuous self-sealing of the open space (Dubois et al. 1996; Smith et al. 1998; Cathelineau & Boiron 2010). This circulation is believed to account for the large changes in thermal gradient at the present day, decreasing from 0.09 to 0.10C ⁄ m in the sedimentary cover to unexpected values of 0.04–0.01C ⁄ m in the upper part of the fractured granites (Clauser & Villinger 1990; Person & Garven 1992). In drill holes, the sedimentary cover and the granite display altered zones characterised by dense network of veins, and alteration of the host rocks (Lede´sert et al. 1999; Sausse 2002; Sausse et al. 2006). Veins are filled by quartz or
Fig. 7. Model of fluid circulation in the basement of the Rhine graben showing deep penetration of sedimentary brines down to 5 km depth and fluid mixing with recharge meteoric water.
barite and healed microfissures reveal a successive self-sealing of the rocks. Previous studies of paleofluid circulation in the EPS-1 (Exploration Puits Soultz) drill hole have shown the circulation of moderate to highly saline fluids both in the sedimentary cover and in the basement down to 2200 m depth (Dubois et al. 1996, 2000), while the GPK2 (Ge´othermie Puits Kutzenhausen) deep drilling documents fluid circulation in the basement down to 5057 m (Cathelineau & Boiron 2010). In two studied zones, either in sandstones and granite close to the unconformity or in deep granite samples, fluids show a large range of Tm ice: )31.6 to )0.2C in the shallowest levels and from )24.6C to )0.1 in the deep granite. These data indicate a probable fluid mixing between two end-members: a brine enriched in divalent cations and a rather dilute fluid. All inclusions homogenised to the liquid phase between 100 and 184C. The measured Th range is between 100 and 165C for the shallow studied zone and from 150 to 185C at 5057 m. The Tm ice – Th diagrams (Fig. 8) show that the fluid inclusions with the highest salinity (i.e. the lowest Tm ice) have the lowest homogenization temperatures, and the low salinities correspond to the highest Th. The mixing trend found at 5057 m is sub-parallel to the mixing trend found for samples from a depth ranging from 1430 to 2207 m. The two extreme salinities are similar in both cases, but a difference in Th of approximately 35–40C is found between the two sets of samples. Although not precisely dated, a series of F–Ba showings are observed on both sides of the Rhine graben (Vosges and Schwartzwald) in the basement and the margins of Mesozoic basins. A part of them have been considered on the basis of structural relationships as Tertiary to recent and linked to the graben activity (Franzke & Lu¨ders 1993; Lu¨ders 1994), or linked to earlier rifting stages during Mesozoic. In the Schwartzwald district, fluorite–barite–quartz vein formation has been related to the circulation of seawater-derived fluids that had strongly interacted with the basement where they gained calcium (21–26 wt% eq. NaCl) and mixed along faults with surface-derived meteoric dilute fluids at tempera-
Fluid flows and metal deposition near unconformity 281 200
200
Kombolgie basin (Australia)
150
Th (°C)
Th (°C)
150 100
100
50 0 –65
Oklo (Gabon)
50
–55
–45
–35
–25
–15
0 –65
–5
–55
–45
Tm ice (°C) 200
200
Northwestern French Massif Central
150
–25
–15
–5
–15
–5
Soultz (Rhine Graben)
150
Th (°C)
Th (°C)
Fig. 8. Tm ice vs. Th diagram applied to aqueous fluid inclusions from two unconformity-related uranium deposits (Kombolgie basin, Australia Derome et al. 2005 and Oklo, Gabon, Mathieu 1999) and two examples of fluid circulation related to extensional tectonics (Northwestern French Massif Central, Boiron et al. 2002 and Rhine Graben, Cathelineau & Boiron 2010).
–35
Tm ice (°C)
100
100
50
50
2000 m 5000 m
0 –65
–55
–45
–35
–15
0 –65
–5
–55
–45
Tm ice (°C)
ture in the range of 100–180C (Lu¨ders & Franzke 1993; Baatartsogt et al. 2007).
–35
–25
Tm ice (°C)
200
150
Th (°C)
PALEOFLUID COMPOSITIONS AND ORIGIN A great part of the data was already published in each case study (see references below) but these data have been brought together in this study to illustrate the main differences and similarities between the main fluid types, in order to compare with the published reports data on reference deposits. Microthermometric data, cation and halogen ratios are compared and presented in Figs 8–12 and Table 2.
–25
100
50
Ca-Mg-Na-K brines
0 –65
–55
–45
Dilute hott Dil t h recharge fluids
Na-Ca-Mg-K brines
–35
–25
–15
–5
Tm ice (°C)
Chlorinity and Tm ice ranges A remarkable feature of a part of the study cases, e.g. the Oklo and Northern territories U deposits, the F–Ba deposits from Massif central, and the Rhine graben examples is the nearly continuous trend of chlorinity between two end-members (Figs 8, 9): a dilute end-member with the highest Tm ice values close to 0C, and a brine end-member. Fluid inclusions with variable Tm ice occur within a single quartz overgrowth, or set of fluid inclusion planes (FIP), and are interpreted as nearly coeval, or corresponding to the same fluid event. This indicates that at each increment of sealing or trapping within veins, the fluids trapped reflect the compositional heterogeneity of the migrating fluids. Such variations are particularly well shown by the Soultz veinlets and deep granite sample FIP, formed within a relatively short period after the Oligocene under nearly constant geothermal gradient but characterised by inclusions ranging from brines to dilute fluids. In the western Massif Central and at Soultz, the lowest Tm ice values are around )32C, as fluids are NaCl-dominated brines. The high chlorinity brines with low Na ⁄ Ca ratios were
Fig. 9. Overview of the Tm ice–Th relationships for several examples of mixing trends between two or three fluid end-members. (i) Full circle: Ca– (Na–Mg–K) brines, (ii) half circle: Na–(Ca–Mg–K) brines, empty circle: dilute hot recharge fluids.
mostly found in Proterozoic basins. In these, two contrasted cases were observed: (i) the Athabasca basin brines of similar chlorinity but exhibiting a large variation in Na ⁄ Ca ratio result in a large Tm ice range ()60 to )20C) but do not show evidence of dilution, and (ii) the Northern Territories U deposits (Australia) and Oklo where mixing of two brines is accompanied by a dilution. Na ⁄ Ca–Na ⁄ K ratios One of the most striking features of the inclusion fluids from the studied U–Pb–Zn–F–Ba deposits is their low Na ⁄ K and Na ⁄ Ca molar ratios. Brines with low Na ⁄ Ca ratio were also reported in more recent systems such as waters from the KTB (Mo¨ller et al. 2005) but with much lower chlorinities (Fig. 10).
282 M.-C. BOIRON et al.
Halogen ratios In most studied cases, a part of the Cl ⁄ Br ratios are lower than that of seawater and may correspond to brines that have evolved beyond halite saturation, and in some instances epsomite saturation (Fig. 11). It can be noted that data points were calculated for mean values as the bulk leachates mix all fluid inclusions, but that the chlorinity range is larger in individual samples than the range of the means. Low Cl ⁄ Br ratios are typical of primary brines, e.g. brines arising from seawater evaporation. In Proterozoic basins such as those from Northern Territories and Athabasca (Mc Arthur, Canada), Cl ⁄ Br values are low whatever the chlorinity, in the range of 100–400, and not scattered. In other Athabasca deposits, a rather wide range of values from 200 up to 800 have been measured (Richard 2009). At Soultz or in the NW Massif Central, Cl ⁄ Br ratio are more scattered, indicating either a mixing of primary brines having a large range of Cl ⁄ Br ratios, which we consider improbable, or a mixing of primary brines with secondary brines as data points are significantly above the mixing lines between seawater (or dilute fluids such as meteoric recharge fluids) and an evaporite brine evolved beyond epsomite saturation. Secondary brines with high Cl ⁄ Br ratios can arise as dilute fluids or seawater entering the fluid system interact with evaporite formations and dissolve halite, as described elsewhere (Hanor & Mcintosch 2007). The ranges and values of Cl ⁄ Br ratios found in ore fluid inclusions in all studied sites are in favour of an evaporitic source for chlorine and of partial to significant mixing of primary and secondary brines. In a few cases, secondary brines predominate (Fig. 12) (Spanish Pb–Zn–F deposits from Catalan and Cantabrian coastal range, Grandia et al. 2003a; b; Pique´ et al. 2008; Sanchez et al. 2009).
Fig. 10. Na ⁄ K vs. Na ⁄ Ca molar ratio diagram applied to fluid inclusion chemical data. 1 – Na-rich brines and 2 – Ca-rich brines, LA-ICPMS data, Mc Arthur River, Athabasca basin (Richard et al. 2010) 3 – Aligator River, crush leach data, (Derome et al. 2005),4 – Oklo, (Mathieu et al. 2000), 5 – Poitou High, crush leach data, (Boiron et al. 2002), 6 – Oil field brines (Carpenter et al. 1974; Kharaka et al. 1987), 7 – Deep basin brines (Fisher & Kreitler 1987), 8 – Silvermines Pb–Zn deposits, crush leach data, (Wilkinson et al. 2005), 9 – Silesia Pb–Zn deposits, crush leach data (Heijlen et al. 2003), 10 – Fluorite deposits in Asturias, LA-ICPMS data, (Sanchez et al. 2009), 11 – Fluorite deposits NE Spain, LA-ICPMS data, (Pique´ et al. 2008). Full circle: KTB fluids, (Mo¨ller et al. 2005), full triangle, Rhine graben fluids (Pauwels et al. 1993) full diamond: seawater, empty diamond: seawater having passed halite saturation, data from Fontes & Matray (1993). H, halite; B, bischofite. Arrows indicate the main trends related to water-rock interactions. Pl, plagioclase, Kf, K-feldspar.
Both ratios reach values near 1 in the most evolved brines, and many paleofluids have Na ⁄ K ratios below 20 with mean values lower than 10 and Na ⁄ Ca ranging from 1 to 10. The low Na ⁄ K ratio are much lower than those predicted by Na–K geothermometers or found in hydrothermal systems for a temperature range of 90–150C, e.g. values ranging from 50 to 150 (Verma & Santoyo 1997).
10 000
Kombol gie basin (Australia) Athabasca
1000
MWSW
G
Cl/Br
Cl/Br
10 000
H
Dilution
100
B
E
Oklo (Gabon)
1000
MWSW
G H E
100
Dilution
B
Primary brines
10
10 0
2000
4000 –1
Cl (mmol kg
6000
2000
Secondary brines
Secondary brines Brine mixing
Primary brines
G H
Dilution
Brine mixing
100
E
B
1000
MWSW
G
Cl/Br
Cl/Br
dilution MW SW
6000
North Western French Massif Central
10 000
Present dayy fluid
1000
4000
Cl (mmol kg–1 solution)
Soultz (Rhine graben)
10 000
0
solution)
H
Dilution E
100
B
10
10 0
2000
4000 –1
Cl (mmol kg
solution)
6000
0
2000
4000
Cl (mmol kg–1 solution)
6000
Fig. 11. Cl ⁄ Br (molar ratio) vs. Cl content of fluid compositions determined by crush leach analysis. Full line: seawater evaporation trend from Fontes & Matray (1993). SW, seawater; MW, meteoric water; G, gypsum; H, halite; E, Epsomite; B, Bischofite. The two dashed lines show the composition of a mixed fluid between either seawater or meteoric water with seawater having reached epsomite saturation. In the diagram for U deposits from Kombolgie basin (upper left diagram), the grey box shows the location of the data obtained for the Mc Arthur and Rabbit Lake uranium deposits (Athabasca basin, Richard 2009). In the lower left diagram (Soultz), the open circles refer to the data obtained on formation waters from Triassic to Oligocene series by Pauwels et al. (1993).
Fluid flows and metal deposition near unconformity 283
Fig. 12. Cl ⁄ Br ratios obtained by crush leach analysis for the different reported occurrences. 1 – Mc Arthur River, Athabasca basin (Richard 2009) 2 – Aligator River, (Derome et al. 2005), 3 – Oklo, (Mathieu et al. 2000), 4 – Silver Mines Pb–Zn deposits (Wilkinson et al. 2005), 5 – Silesia Pb–Zn deposits (Heijlen et al. 2003), 6 – Maestrat basin Pb–Zn deposit, Spain (Grandia et al. 2003a), 7 – Cantabrian Pb–Zn deposits, Spain (Grandia et al. 2003b), the dash line shows high Cl ⁄ Br ratio related to the presence of salt dome, 8 – Poitou High Pb–Zn–F deposits (Boiron et al. 2002), 9 – Fluorite deposits in Asturias (Sanchez et al. 2009), 10 – Fluorite deposits, NE Spain (Pique´ et al. 2008), 11 – Soultz, unpublished data. Seawater evaporation trend from Fontes & Matray (1993), SW, seawater; Hb, halite beginning; Hend, halite end; E, epsomite.
Homogenisation temperatures and the distribution of Tm–Th pairs Figures 8 and 9 are a comparison of Tm–Th pairs for a series of studied cases. The common feature to all FI populations is the Tm–Th trend between two end-members: a brine with a Th around 100C and a more dilute fluid end-member with a higher Th, around 30–40C higher than the Th of the brine. A scattering of Th for a given Tm ice is observed for some FI populations and may arise for several reasons: (i) unavoidable effects of stretching, (ii) the presence in U deposits of radiolytic gases such as H2 which generally cause an increase in Th, a rather common observation in most U deposits, (iii) superimposition of nonsynchronous fluid populations at different temperatures, (iv) fluctuation in fluid pressure above the hydrostatic pressure, a process which is most likely in a deep-seated fluid system when the unconformity reaches a depth of 4–5 km. It can be noted that fluid pressure may be estimated from the consideration of the depth of the unconformity at the time of the deposition, assuming the nature of the pressure (hydrostatic or lithostatic). The depth is generally inferred from temperature estimates, considering geothermal gradients and assuming the nature of the fluid pressure. For depths less than 5 km, the pressure is likely to be hydrostatic, but at lower depths, intermediate or fluctuating pressures could be recorded. This uncertainty of the nature of the fluid pressure makes the interpretation of Th values more difficult. In the case of all trends characterized
by a large and continuous range of Tm ice correlated with Th (Soultz, Oklo), it is highly probable that pressures can be interpreted as hydrostatic. Alternatively, rather large scattering in Th for a given fluid salinity can be interpreted as the result of alternating changes in fluid pressure above the hydrostatic pressure assuming that the temperature was constant. This is the case for the Athabasca U deposits.
DISCUSSION The evolution of the fluid system responsible for the formation of Pb–Zn–(F–Ba)–U deposits in sedimentary basins and in the basement near the unconformity is summarised in the Fig. 13 and Table 3. Main stages are discussed below. Brine origin High chlorinities reaching 4–6 mol ⁄ kg of solution and Cl ⁄ Br ratios much lower than the seawater ratio (Cl ⁄ Br < 600 molar ratio), indicate a probable primary origin of the brines, i.e. by evaporation of seawater. In many circumstances, evaporitic brines are identified on a geochemical basis, and almost no evidence of evaporite remains, this being due either to the dissolution of salt since the basin formed, or to the erosion and complete loss of the formations that might have hosted evaporites. Secondary brines are also suspected in a number of cases explaining the difficulty in finding pristine evaporate-bearing formations.
284 M.-C. BOIRON et al.
Fig. 13. General conceptual model for the genesis of basement and carbonate-hosted deposits near the basin ⁄ basement unconformity.
In the case of old systems, such as the Proterozoic basins, a great part of the basin is now eroded (more than 3 km following P–T reconstructions, see Derome et al. 2003, 2005), and evaporite formations are only found sporadically, frequently at great distances from the mineralised areas. In Australian Proterozoic basins, a few occurrences are present away from the ore zones, such as the evaporitic dolostones in the MacKay formation in the south-eastern part of the McArthur basin (Northern Territories, Rawlings 1999). In most other cases considered in this article, evaporites are among the first sedimentary formation deposited above the unconformity (Figs 1–7): the Hettangian series is in almost direct contact with the unconformity in the NW Massif Central (Poitou High), while Trias evaporites occur near the unconformity in the Rhine graben, in Silesia, and in most Spanish examples. A brine reservoir is thus very close vertically to the basement. As in Proterozoic basins, evaporite formations are not necessarily located in the immediate environment of the deposit, but can be laterally distant: (i) in the Poitou High, only sparse evidence of evaporite is found above the main altered and mineralised zone (relicts of evaporite layers such as boxworks of gypsum in the Infra-Liassic series), while the main Hettangian evaporite layers are found more than 100 km to the south in the Angouleme region, with Trias evaporites at greater distances. (ii) In the Rhine graben, drill holes did not encounter salt formations at Soultz, although salt layers are easily identified within the Trias series at a few tens of kilometres distant. These observations indicate extensive migration of brines away from their source formation. We suggest that long-range migration of brines derived from Triassic evaporites is an important source of mineralisation in many of the examples we have discussed in this study.
Why is deep penetration of brines within the basement is so important? Brines–basement rocks interactions. The penetration of basinal fluids within the basement has been discussed earlier (Yardley et al. 2000; Cathles & Adams 2005), and described in southern Norway (Munz et al. 1995; Oliver et al. 2006; Gleeson et al. 2003), in Spain (Pique´ et al. 2008), and in various places in Europe (Bouch et al. 2006; and see review in Muchez et al. 2005). The depth reached by the brines remains in a great number of cases rather hypothetical, as does the cause of migration. Recent detailed studies of granitoids from Athabasca, and deep samples from the Soultz granite provide some keys for the understanding of the process of brine migration. The deep penetration of brines within the basement through faults is associated with an intimate fluid invading through the matrix permeability. Dense networks of opened and interconnected microfractures are revealed by the healed brine inclusion plane networks within the basement (granites or granitoı¨ds). Microfracturing such as at Soultz is observed in the whole granite down to 5 km depth (Cathelineau & Boiron 2010), and allows brines to interact extensively with the host rocks. FIP from deep samples are dominated by recent reservoir fluids and fluid salinities depict the full range of mixing between the sedimentary brine and the hot dilute fluids. Thus, either sets of micro-fissures have been formed and healed at depth since the Oligocene or the inherited FIP from earlier stages were reopened and healed by the main reservoir fluids. Such microfracturing is also found in basement rocks from Saskatchewan (Athabasca basin, Mercadier 2008; Mercadier et al. 2010). There, the early microfractures, from the late Hudsonian retrograde metamorphism were reopened and healed by evaporitic brines, indicating a fully
Fluid flows and metal deposition near unconformity 285 pervasive brine infiltration through the rocks. The efficiency of the process needs the formation, or re-opening of microfissures at the time of the brine movements, the healing of the cracks which ensured the trapping of the paleobrines indicates that fissures formed probably during repeated micro-seismic events. This penetration of the brine in the microfractured pathways is one of the key factors for metal leaching. Apart of these two examples, microstructural studies of brine migration in basement rocks remain scarce and would merit more attention to generalise such a process. Metal extraction. The presence of an eroded basement with regolith or weathered surface generally found down to a few tens of metres below the unconformity is the second key factor of the metal extraction: after peneplanation, within the weathered part of the outcropping basement, supergene alteration affects most feldspars and phyllosilicates and their deep alteration is a major cause of element release. Pb and Ba are released from feldspars, F and probably Zn from biotite, U from uraninite in biotite and monazite, and metals are subsequently sorbed onto clays and Fe–Ti oxides and hydroxides. In Western Europe, the main metal reservoir is thus the weathered Hercynian terrane, ‘peneplaned’ in the Triassic and affected by supergene alteration. The latter produced a thick weathered and oxidised zone, well exposed in basement quarries showing the unconformity (Mauze´-Toarc¸ais for instance, for the Poitou High). Metal can be easily extracted by brines later on, once most of the metal bearers are deeply altered. It can be noted that in the case of Ag and Ni–Co–As deposits from Morocco, the metals are extracted from weathered ultramafic rocks (middle Precambrian) and transported later on by brines during late Trias or early Lias (Essarraj et al. 1998, 2005) near the unconformity following the same process. Brines penetrating the weathered basement have all the features (temperature, chloride content) required for an optimal metal solubilisation and complexation, as shown by the increased metal concentrations in fluids at increasing chlorinity and temperature (Yardley 2005). High metal concentrations (Cu, Pb, Zn, Fe, Mn) for sedimentary brines have been reported from most carbonate-hosted base metal deposits (Heijlen et al. 2008; Pique´ et al. 2008; Wilkinson et al. 2009). In the case of uranium, Hecht & Cuney (2000) and Mercadier (2008) have documented the dissolution of accessory minerals (zircons and monazite) in the presence of brines. In the basement, uraninite and monazite constitute a major source of U at the time of brine circulation, even in fresh zones located away from the weathered basement. LA-ICPMS analyses have shown that brines from the Athabasca basin could transport significant amounts of U which range from one to several hundreds of ppm (Richard et al. 2010). The role of
basement as the main metal reservoir, the sedimentary cover being insufficient to account for the mass of metal deposited, has been put forward also for other metals such as copper, on the basis of mass and heat transfer modelling (Koziy et al. 2009). When interacting with the basement, chemical features of the sodium dominated brine may drastically evolve towards Ca–K–Mg (Na) rich brines. High Ca ⁄ Na ratios are generally interpreted as the result of a Ca ⁄ Na exchange through mineral–fluid interaction such as albitisation, or Na-metasomatism at greater depth (review in Kharaka & Hanor 2004). The complete conversion of a Na-dominated evaporitic brine into a Ca dominated brine is however still difficult to interpret. Extreme Ca-rich fluids have been found in the present day waters sampled in basements (KTB for instance, Mo¨ller et al. 2005), and in the Saskatchewan deposits where they are considered to be the main brine responsible for U extraction and deposition (Derome et al. 2005; Richard 2009). Fluid mixing Mixing of two or more fluids in the major fluid pathways. Two major fluid types are generally found in cemented pathways: (i) one or more brines frequently identified as primary brines expelled during compaction or by tectonism linked to major geodynamic events. (ii) dilute fluids corresponding to recharge by meteoric (or marine) fluids or local resident fluids. In all the studied cases, brines display a large range of compositions, and in most cases a large range of chlorinity indicative of mixing between brine and dilute fluid end-members. The variety of salinity–temperature pairs indicates that at each stage of fluid circulation and healing of the micro-fissures, a distinct rate of mixing between the two fluids is registered. These different rates of mixing show also that the reservoir fluids were not fully homogenized at the time of healing of the micro-fissures. The dilute end-member of the mixing trends is considered as a recharge fluid of likely meteoric origin. Pristine features of surface fluids are however generally lost, as the d18O of the fluid estimated from the d18O values of quartz are in general typical of fluids buffered by sediment mineral assemblages. d18O values in most cases range from 0 to +4& (Table 2). Exceptions concern fluid flow in large drainage faults, such as those described by Smith et al. (1998) at Soultz. Very similar Th–Tm ice trends with nearly complete series of chlorinities between the two end-members were found in a part of the studied basins as already presented and indicate that fluid mixing occurred in most of the studied locations. In some of the listed examples, e.g. the Asturias F district (Sanchez et al. 2009), and the Maestrat basin, (Grandia et al. 2003a), authors do not report a clear trend of mixing between a dilute end member and a brine but simply the trapping of a spectrum of
286 M.-C. BOIRON et al. fluids covering a large range of chlorinities. It can be noted that in other unconformity deposits such as those from Saskatchewan (Athabasaca basin), the dilute fluid endmember is almost lacking and fluid mixing concerns only the two brine end-members, the Na-dominated and the Ca-dominated (Derome et al. 2005; Richard et al. 2010). The lack of chemical homogenisation of the circulating fluids seems however to characterize most of the basins considered in our comparative study. Cold fluids and hot ascending fluids in thermal disequilibrium with host formations? In a part of the study cases, a systematic temperature gap is recorded between the two fluid end-members. The presence of slightly colder brines indicates that downward penetration of sedimentary brines tends to cool the deep reservoir fluids. The 35–40C gap between the estimated temperatures for highly saline brines and the hotter dilute fluids, may correspond to more than 1 km difference in depth, assuming a mean thermal gradient of about 30C ⁄ km, if the two fluids are in thermal equilibrium with the host rocks in their respective reservoirs. The temperature difference between the brine and the hotter dilute fluid in a given zone, e.g. the area of fracture sealing, is indicative of an anisothermal mixing. At Soultz, anisothermal mixing is found in all parts of the granite reservoir, as shown by the variety of salinity-temperature pairs and the Tm–Th trend which are typical of each depth interval (Cathelineau & Boiron 2010). In a part of the Proterozoic basins (Northern Territories, Australia, Athabasca basin, Canada and Oklo, Gabon), the range of homogenisation temperatures is rather similar to those recorded at Soultz in the upper parts of the reservoir (2 km) or in the north Massif Central. Temperatures can therefore be either converted into depths of 4–5 km assuming similar thermal gradients, or can be considered as temperatures representative of abnormal thermal gradients at much shallower depths of 1–3 km such as those recorded in the Rhine graben. Significant exceptions are: (i) data obtained on Maestrat basin where brines at 130C (23 wt% eq. NaCl) mixed with a cooler and more dilute fluid (Grandia et al. 2003a), (ii) Asturias where no definite mixing trends were obtained (Sanchez et al. 2009), and Malines where the dilute fluid is cooler (70C) than the brines (150C) (Charef & Sheppard 1988). Brine-host rock interactions in the deposit area (mineral dissolution, silicification, dolomitisation and Mg alterations). Two main cases may be distinguished: (i) alteration within sedimentary formations, generally dominated by dolomitisation. In the sedimentary cover, in limestones especially, dolomitisation and calcite dissolution creating porosity were observed in a great number of cases. The main driving force for this process is linked to the high Mg concentration of the evaporated seawater which is the
main cause of the calcite dissolution and the replacement by dolomite. The resulting fluid is depleted in Mg and enriched in Ca. (ii) Mg-silicate alteration within the basement. Mg-rich evaporite fluids are responsible for the intense Mg-metasomatism which affects the basement. Mg alteration from U deposits is characterized either by the crystallization of chlorites (Oklo, Northern territories, Australia) or by a sudoite–Mg illite assemblage in Athabasca (Kister et al. 2006; Mercadier et al. 2010). These alterations display similar mineral assemblage to major talc deposits such as the Trimouns deposit in the Pyrenees, which results from the circulation of evaporitic brine during the extension of the Gascogne gulf (Boulvais et al. 2006; Boiron et al. 2007). As Ca–Mg brines are generally enriched also in potassium, K-alteration (adularisation) may also occur at the top of the basement near the unconformity such as in the case of the Poitou high. Extensive silicification characterizes the main ascending pathways and is generally favoured by the cooling and mixing of hot fluids with lower temperature dilute fluids. The main driving force for massive quartz precipitation is thus the decrease of the quartz solubility at decreasing temperature (Crerar & Anderson 1971). Hence, little quartz was precipitated in the Canadian unconformity deposits, where no hot dilute recharge fluids were found, compared with Australian deposits where large amounts of quartz formed by dilution and cooling of recharge fluids in faults, above the unconformity U deposits. Significant silicification is found at Soultz (quartz filled faults, Smith et al. 1998) and in some deposits from the Poitou High (quartz-galena infillings from Melle, La Charbonie`re and Chatenet quartz-fluorite veins, Boiron et al. 2002) or Albigeois (Munoz et al. 1999). Silicification is thus a major indicator of temperature change, sometimes coupled to dilution in pathways favouring the convection of external fluids. Anisothermal mixing of brines with dilute fluids favours the precipitation of different mineral assemblages depending on the availability of metals, and the presence or absence of sulphur. Dilution provokes the destabilisation of metal complexes, especially U and Pb chlorides. The availability of sulphur is however a key parameter and two main sources may be invoked: the bacterial reduction of sulfate (BSR) which is invoked in the Irish deposits for the for the development of sulphide in the cool, shallow brine and accounts for the very light sulphur isotopic composition of the ore (Blakeman et al. 2002), and the thermochemical reduction of sulphates (Machel 2001) which can produced H2S and therefore be a factor controlling metal deposition when sedimentary formations reached the oil window (case of the Maestrat basin, Grandia et al. 2003a; b). In cases where one fluid involved in the mixing was not sufficiently rich in sulphur, mixing caused mainly the precipitation of fluorite and carbonate, the spent fluid
Fluid flows and metal deposition near unconformity 287 remaining enriched in metals. Such fluids may have kept their potential for a metal deposition in more favourable environments. Metal deposits may be found, therefore, away from the fluorite deposits, as suspected by Pique´ et al. (2008) for the central Catalan coastal range deposits. Geodynamic context The cause of fluid migration frequently remains unknown because of the difficulties in dating precisely the ore events and hence precisely relating fluid migration to specific geodynamic events. These relationships are almost impossible to prove to have occurred over great distances once the basin or the belt is eroded. Extensional vs. compressional tectonic movements were thus largely discussed in the published reports to explain MVT deposits. Gravity driven fluid movements proposed by Leach et al. (2005) for a series of deposits are based on ages whose geological significance is questionable. Thus, in the case of the Cevennes Pb deposits, after a few decades of models favouring a rather old karst infilling, the deposits were alternatively considered as linked to compressional tectonics related to the Pyrenean orogeny (60 Ma) on the basis of remagnetization (Rouvier et al. 2001), or were linked to previous extensional stages during the Malm (see discussion in Muchez et al. 2005). In most other cases where rather young ages have been proposed and interpreted as related to compressional tectonics [46 ± 20 Ma in Silesia (Symons et al. 1995), 15 ± 10 Ma in Asturias (Symons et al. 2009)] it has also been impossible to link precisely the dated remagnetisation of minerals to the ore formation. The arguments in favour of constraining models with reliable ages on ore or gangue (Rb–Sr on sulphides, Sm–Nd on fluorite for instance) have been put forward in a number of the studies discussed in this study (Heijlen et al. 2003; for Silesia, Wilkinson et al. 2005; for Irish deposits, Muchez et al. 2005 for a variety of deposits hosted by Mesozoic formation over Europe including Cevennes). This approach generally supports a link between extensional tectonics and ore formation rather than relating ore-formation to later orogeny. The origin of hot fluids can be found in localised abnormal heat flows, and convection cells favoured by faults cross-cutting both the basement and the sedimentary cover. In the Poitou region, such convection cells have been inferred for the lower Lias formation, where metal deposits have been considered as syn-sedimentary. In that case, hydrothermal activity was considered as synchronous with the Atlantic ocean opening. New geochronological data tend to indicate that these convection cells were active after the main stage of lower Lias sedimentation e.g. around 140–155 Ma (S. Fourcade, M. C. Boiron, M. Cathelineau, G. Ruffet, N. Clauer, O. Belcourt, E. Deloule, Y. Coulibaly and D. A. Banks, unpublished
data). The major stages of fluid movement are linked to permeability increase due to localised mineral dissolution, or fracturing and are thus not related to early diagenesis or the main stages of compaction, but correspond to later major geodynamic events. Most fluid events described in this article occurred 50–200 Ma after the sedimentation (see Table 3), with the notable exception of the Irish deposits (Wilkinson et al. 2005). Style of migration: deep downward penetration and convective cells The main driving forces proposed for the convection of brines in ore systems are: (i) the role of topographic highs area as already proposed for MVT deposits (Leach et al. 2001) and (ii) abnormal heat flow which characterizes rifting tectonics. The patterns of palaeofluid circulation are shown in Figs 1–7: (i) a major system of laterally migrating brines at the basement – sedimentary cover interface where both sandstones and the weathered upper part of the basement had a high permeability, (ii) local convective cells rooted at more than 5 km depth, involving low salinity fluids (presumably surface derived, likely migrating, through basement from structural highs) generally centered on horst fault systems, or faults (reverse in the case of the unconformity U deposits), (iii) deep downward penetration of brines which accounts for the introduction of large amounts of chlorine in the basement microfractures. In the western Massif Central, brines are considered to have migrated along the unconformity below an impermeable cover formation, and then mixed locally with recharge fluids in fault systems. Numerical hydrological modelling of basins aquifers by Evans et al. (1991) shows that: either (i) the density differences related to salinity contrast between the fluid reservoirs are the primary driving forces for fluid flow when sediments surrounding salt contain a dilute fluid, or (ii) thermal effects are necessary to get convection in brinedominated aquifers. The mixing of hot recharged dilute fluids and cooler brines indicates that dilute fluids are able to convect from remote areas, and can be chanelled along faults. Meteoric waters are heated at depth and their density allows them to ascend along drainage faults such as extensional faults (horst and graben systems). Brines are flowing downward to great depth because of their specific density. In all the studied cases, the hydraulic conductivity of the stratigraphic units influences the brine regime on a regional scale, as shown by the specific role of the basal conglomerate and sandstone above the regolith or weathered basement, which in all cases constituted a major aquifer above the fractured basement and below cap rocks such as clay rich formations (Toarcian marls in Poitou, clay-rich formations in the Rhine graben, FB clay
288 M.-C. BOIRON et al. formation at Oklo). Such controls are already shown in similar contexts in the North German basin (Magri et al. 2009). Lessons from relatively recent systems for the understanding of very old brine migration Analogies in fluid chemistry. Fluids found in giant U deposits have many of the same features as brines evolved by basement interactions and found in most carbonate hosted deposits or unconformity related F–Ba–Pb–Zn deposits from western Europe and in basement rocks. Depth of brine penetration within the basement below the unconformity. There are no simple limits to brine penetration within the fractured basement, as shown by young systems such as those related to the continental rifting of the Rhine graben. The analogy between the multiple microfissuring of archean rocks and healing by sedimentary brines and the process observed at Soultz is remarkable. The infiltration of brines below the unconformity in Athabasca may therefore have reached 5 km. Role of thermal convection. The examples of Soultz and the Massif Central provide some keys for the understanding of the old systems where little is known of the dynamics of fluids or the geometry of fluid pathways. In NW-Australia as well as in the NW Massif Central, the dilution of brines in fault systems near the unconformity is observed. In both cases, the origin of hot recharge waters may be found in analog models such as the Soultz geothermal area where recharge fluids are driven from rather remote lateral sources in outcropping areas (meteoric fluid recharge). Similar fluid migration affects the Oklo area (Fig. 2): the main deposit is located at the vicinity of a major fault of the Mounana basement horst, which was considered as related to a double continental rift (Franceville–Lastourville basin and Okondia rift) on both sides of the Ondii–Dambi horst (Gauthier-Lafaye 1986). The main stages of fluid migration are considered to be related to the main extensional phase around 2.05 Ma (Gancarz 1978). The formation of hydraulic breccia attests to contrasts in fluid pressure (Gauthier-Lafaye 1986). The recharge fluids, as at Soultz, likely infiltrated in the basement horst relief (Chaillu granite), and mixed with brines and oil fluids from the sedimentary aquifers (FA conglomerates) along horst faults. It is highly probable that the formation of giant U deposits in Saskatchewan was linked to a specific geodynamic event which caused: (i) the microfissuring and reactivation of pre-existing micropermeability and faults, (ii) the localized heat flows which were necessary for brine convection, as no fluid density contrast characterize the two main brines found in these deposits.
Pulsatory fluid regime. The wide variety of salinity during fluid trapping between the brines which penetrate downward in the basement and the ascending fluids was found at Oklo and in Northern Territories (Australia) but not in Saskatchewan. The analogy with recharge fluids commonly found in relation to convective systems driven by localized heat flows and recharge from emerged zone is only possible in the two latter cases. The lack of homogeneity of the brine composition is in favour of successive inputs of brine within the main fault system, and repeated pulses following models well known in aquifer systems in relation to micro- or macro-seismicity.
ACKNOWLEDGEMENTS This article has benefited from different works carried out at CREGU-G2R by J. Mercadier, D. Derome and R. Mathieu and from fruitful discussions with M. Cuney, D.A. Banks and S. Fourcade. The authors thank CNRS and Areva NC for financial support (BDI PhD grant for A. Richard). Areva NC and Cameco are also acknowledged for providing samples and scientific collaborations. S. Gleeson, P. Muchez and B. Yardley are warmly acknowledged for their constructive comments and editorial handling.
REFERENCES Alexandre P, Kyser K, Thomas D, Polito P, Marlat J (2009) Geochronology of unconformity-related uranium deposits in the Athabasca basin, Saskatchevan, Canada and their integration in the evolution of the basin. Mineralium Deposita, 44, 41–59. Alexandrov P, Royer JJ, Deloule E (2001) 331 ± 9 Ma emplacement age at the Soultz monzogranite (Rhine Graben basement) by U ⁄ Pb ion-probe zircon dating of samples from 5 km depth. Comptes Rendus de l’Acade´mie des Sciences Paris, 332, 747–54. Amato JM, Bouillon AO, Serna AM, Sanders AE, Farmer GL, Gehrels GE, Wooden JL (2008) Evolution of the Mazatzal province and the timing of Mazatzal orogeny: insights from U–Pb geochronology and geochemistry of igneous and metasedimentary rocks in southern New Mexico. GSA Bulletin, 120, 328–46. Baatartsogt B, Schwinn G, Wagner T, Taubald H, Beitter T, Markl G (2007) Contrasting paleofluid systems in the continental basement; a fluid inclusion and stable isotope study of hydrothermal vein mineralization, Schwartzwald district, Germany. Geofluids, 7, 123–47. Banks DA, Boyce AJ, Samson IM (2002) Constraints on the origins of fluids forming Irish Zn–Pb–Ba deposits: evidence from the composition of fluid inclusions. Economic Geology, 97, 471–80. Bertrand JM, Leterrier J, Cuney M, Brouand M, Stussi JM, Delaperrie`re E, Virlogeux D (2001) Ge´ochronologie U–Pb sur zircon de granitoides du Confolentais, du massif de CharrouxCivray (seuil du Poitou) et de Vende´e. Ge´ologie de la France, 1–2, 167–89. Blakeman RJ, Ashton JH, Boyce AJ, Fallick AE, Russell MJ (2002) Timing of interplay between hydrothermal and surface fluids in the Navan Zn + Pb orebody, Ireland: evidence from metals distribution trends, mineral textures and d34S analyses. Economic Geology, 97, 73–91.
Fluid flows and metal deposition near unconformity 289 Boiron MC, Cathelineau M, Banks DA, Buschaert S, Fourcade S, Coulibaly Y, Boyce A, Michelot JL (2002) Fluid transfers at a basement ⁄ cover interface. Part II: large-scale introduction of chlorine into the basement by Mesozoic basinal brines. Chemical Geology, 192, 121–40. Boiron MC, Cathelineau M, Dubessy J, Fabre C, Boulvais P, Banks DA (2007) Na–Ca–Mg rich brines and talc formation in the giant talc deposit of Trimouns (Pyrenees): fluid inclusion chemistry and stable isotope study. Proceeding of the ECROFI XIX Meeting, Bern, July 2007. Bonhomme M, Gauthier-Lafaye F, Weber F (1982) An example of Lower Proterozoic sediments. Precambrian Research, 18, 87– 102. Bonhomme M, Baubron JC, Jebrak M (1987) Mine´ralogie, ge´ochimie, terres rares et aˆge K–Ar des argiles associe´es aux mine´ralisations filoniennes. Chemical Geology, 65, 321–39. Bouch JE, Naden J, Shepherd TJ, McKervey JA, Young B, Benham AJ, Sloane HJ (2006) Direct evidence of fluid mixing in the formation of stratabound Pb–Zn–Ba–F mineralisation in the Alston Block, North Pennine Orefield (England). Mineralium Deposita, 41, 821–35. Boulvais P, De Parseval P, D’Hulst A, Paris P (2006) Carbonate alteration associated with talc-chlorite mineralization in the eastern Pyrenees, with emphasis on the St Barthelemy massif. Mineralogy and Petrology, 88, 499–526. Carpenter AB, Trout ML, Pickett EE (1974) Preliminary report on the origin and chemical evolution of lead- and zinc-rich oil field brines in central Mississippi. Economic Geology, 69, 1191–206. Cathelineau M, Boiron MC (2010) Downward penetration and mixing of sedimentary brines and dilute hot waters at 5 km depth in the granite basement at Soultz-sous-Foreˆts (Rhine graben, France). Comptes Rendus Geosciences, doi: 10.1016/j. crte2009.08.010. Cathelineau M, Fourcade S, Clauer N, Buschaert S, Rousset D, Boiron MC, Meunier A, Lavastre V, Javoy M (2004) Multistage paleofluid percolations in granites: a stable isotope and K–Ar study of fracture illite from Vienne plutonites (N.W. of the french Massif Central). Geochimica Cosmochimica Acta, 68, 2529–42. Cathles LM, Adams JJ (2005) Fluid flow and pretoleum and mineral ressources in the upper (<20 km) continental crust. Economic Geology, 100th Anniversary volume, 77–110. Charef A, Sheppard SMF (1988) The Malines Cambrian carbonate-shale-hosted Pb–Zn deposit, France: thermometric and isotopic (H, O) evidence for pulsating hydrothermal mineralization. Mineralium Deposita, 23, 86–95. Clauer N, Chaudhuri S (1995) Inter-basinal comparison of the diagenetic evolution of illite ⁄ smectite minerals in buried shales on the basis of K–Ar systematics. Clays and Clay Minerals, 44, 818–24. Clauser C, Villinger H (1990) Analysis of conductive and convective heat transfer in a sedimentary basin, demonstration for the Rhine Graben. Geophysical Journal International, 100, 393– 414. Crerar DA, Anderson GM (1971) Solubility and solvation reactions of quartz in dilute hydrothermal solutions. Chemical Geology, 8, 107–22. Cumming GL, Krstic D (1992) The age of unconformity-related uranium mineralization in the Athabasca Basin, northern Saskatchewan. Canadian Journal of Earth Sciences, 29, 1623–39. Derome D, Cuney M, Cathelineau M, Fabre C, Dubessy J, Bruneton P, Hubert A (2003) A detailed fluid inclusion study in silicified breccias from the Kombolgie sandstones (Northern Territory, Australia): inferences for the genesis of middle-Prote-
rozoic unconformity-type uranium deposits. Journal of Geochemical Exploration, 80, 259–75. Derome D, Cathelineau M, Cuney M, Fabre C, Lhomme T, Banks DA (2005) Mixing of sodic and calcic brines and uranium deposition at the McArthur River, Saskatchewan, Canada. A Raman and Laser-Induced Breakdown Spectroscopic study of fluid inclusions. Economic Geology, 100, 1529–45. Derome D, Cathelineau M, Fabre C, Boiron MC, Banks DA, Lhomme T, Cuney M (2007) Paleo-fluid composition determined from individual fluid inclusions by Raman and LIBS: application to mid-Proterozoic evaporitic Na–Ca brines (Alligator Rivers Uranium Field, northern territories Australia). Chemical Geology, 237, 240–54. Dubois M, Ayt Ougougdal M, Meere P, Royer JJ, Boiron MC, Cathelineau M (1996) Temperature of paleo- to modern selfsealing within a continental rift basin : the fluid inclusion data (Soultz-sous-Foreˆts, Rhine graben, France). European Journal of Mineralogy, 8, 1065–80. Dubois M, Lede´sert B, Potdevin JL, Vanc¸on S (2000) Determination des conditions de pre´cipitation des carbonates dans une zone d’alte´ration du granite de Soultz (soubassement du fosse´ Rhe´nan, France): l’enregistrement des inclusions fluides. Comptes Rendus de l’Acade´mie des Sciences Paris, 331, 303–09. Enrique P, Andreichev V, Sole´ J (1999) Rb–Sr dating of a multiintrusive plutonic sequence from the Catalan coastal batholith (NE Spain). Journal of Conference Abstract, 4, 809. Essarraj S, Boiron MC, Cathelineau M, Banks DA, El Boukhari A, Chouaidi MY (1998) Brines related to Ag deposition in the Zgounder silver deposit (Anti-Atlas, Morocco). European Journal of Mineralogy, 10, 1201–14. Essarraj S, Boiron MC, Cathelineau M, Banks DA, Benaharref M (2005) Penetration of surface-evaporated brines into the Proterozoic basement and depositon of Co and Ag at Bou Azzer (Morocco): evidence from fluid inclusions. Journal of African Earth Sciences, 41, 25–39. Evans DG, Nunn JA, Hanor JS (1991) Mechanisms driving groundwater flow near salt domes. Geophysical Research Letters, 18, 927–30. Fisher RS, Kreitler CW (1987) Geochemistry and hydrodynamics of deep-basin brines, Palo Duro Basin, Texas, USA. Applied Geochemistry, 2, 459–76. Fontes JC, Matray JM (1993) Geochemistry and origin of formation brines from Paris Basin, France. 1. Brines associated with triassic salts. Chemical Geology, 109, 149–75. Fourcade S, Michelot JL, Buschaert S, Cathelineau M, Freiberger R, Coulibaly Y, Aranyossy JF (2002) Fluids transfer at the basement ⁄ cover interface. Part I: subsurface recycling of basement trace carbonate (Vienne granitoids, France). Chemical Geology, 192, 99–119. Franzke HJ, Lu¨ders V (1993) Formation of hydrothermal fluorite deposits in the southern Black Forest (SW Germany). Part I: structural control. In: Current Research in Geology Applied Deposits (eds Fenoll Hach A, Torres Ruiz J, Gervilla F), pp. 719–22. University of Granada, Granada. Gancarz J (1978) U–Pb age (2.05 109 years) of the Oklo urnaium deposit. Natural fission reactors. IAEA Symposium Proceedings, Vienna, pp. 513–20. Gauthier-Lafaye F (1986) Les gisements d’uranium du Gabon et les re´acteurs d’Oklo. Mode`le me´talloge´nique de gıˆtes a` fortes teneurs du Prote´rozoı¨que infe´rieur. Me´moire Sciences Ge´ologiques, 78, 206. Gauthier-Lafaye F, Weber F (1989) The Francevillian (Lower Proterozoic) uranium ore deposit of Gabon. Economic Geology, 84, 2267–68.
290 M.-C. BOIRON et al. Gleeson SA, Yardley BWD, Munz IA, Boyce AJ (2003) Infiltration of basinal fluids into high-grade basement, South Norway: sources and behaviour of waters and brines. Geofluids, 3, 33– 48. Grandia F, Asnerom Y, Getty S, Cardellach E, Canals A (2000) U–Pb dating of MVT ore-stage calcite. Implications for fluid flow in a Mesozoic extentional basin from Iberian peninsula. Journal of Geochemical Exploration, 69–70, 377–80. Grandia F, Cardellach E, Canals A, Banks DA (2003a) Geochemistry of fluids related to epigenetic carbonated hosted Zn–Pb deposits in the Maestrat basin, Eastern Spain: fluid inclusion and isotope (Cl, C, O, S, Sr) evidence. Economic Geology, 98, 933– 54. Grandia F, Canals A, Cardellach E, Banks DA, Perona J (2003b) Origin of ore-forming brines in sediment-hosted Zn–Pb deposits of the Basque-Cantabrian basin, Northen Spain. Economic Geology, 98, 1387–411. Hanor JS, Mcintosch JC (2007) Diverse origins and timing of formation of basinal brines in the Gulf of Mexico sedimentary basin. Geofluids, 7, 227–37. Hecht L, Cuney M (2000) Hydrothermal alteration of monazite in the Precambrian crystalline basement of the Athabasca basin (Saskatchewan, Canada): implications for the formation of unconformity-related uranium deposits. Mineralium Deposita, 35, 791–95. Heijlen W, Muchez P, Banks DA, Schneider J, Kucha H, Keppens E (2003) Carbonate-hosted Zn–Pb deposits in upper Silesia, Poland: origin and evolution of mineralizing fluids and constraints on genetic models. Economic Geology, 98, 911–32. Heijlen W, Banks DA, Muchez P, Stensgard BM, Yardley BWD (2008) The nature of mineralizing fluids of the Kipushi Zn–Cu deposit, Katanga, Democratic Republic of Congo: quantitative fluid inclusion analysis using Laser Ablation ICP-MS and bulk crush-leach methods. Economic Geology, 103, 1459–82. Hitzman MW, Beaty DW (1996) Geological setting of the Irish Zn–Pb (Ba–Ag) orefield. In: Society of Economic Geology Special Publication, vol. 4 (ed Sangster DF), pp. 112–43. Hoeve J, Sibbald T (1978) On the genesis of Rabbit Lake and other unconformity-type uranium deposits in northern Saskatchewan, Canada. Economic Geology, 73, 1450–73. Holliger P (1988) Ages U ⁄ Pb de´finis in-situ sur oxydes d’uranium a` l’analyseur ionique: me´thodologie et conse´quences ge´ochimiques. Comptes Rendus de l’Acade´mie des Sciences, Paris, 307, 367–73. Juez-Larre´ J, Andriessen PAM (2006) Tectonothermal evolution of the northeastern margin of Iberia since the break up of Pangea to present, revealed by low-temperature fission-track and (U–Th) ⁄ He thermochronology: a case history of the catalan coastal ranges. Earth and Planetary Science Letters, 243, 159–80. Kharaka YK, Hanor JS (2004) Deep fluids in continents: I sedimentary basins. Treatise on Geochemistry, 5, 499–540. Kharaka YK, Maest AS, Carothers WW, Law LM, Lamothe PJ, Fries TL (1987) Geochemistry of metal-brines from Central Mississippi Salt Dome basin (USA). Applied Geochemistry, 2, 543–61. Kister P, Laverret E, Quirt D, Cuney M, Patrier Mas P, Beaufort D, Bruneton D (2006) Mineralogy and geochemistry of the hostrocks alterations associated with the Shea-Creek unconformitytype uranium deposits (Athabasca basin, Saskatchewan, Canada). Part 2: regional-scale distribution of the Athabasca sandstone matrix minerals. Clays and Clay Minerals, 54, 295–313. Koziy L, Bull S, Large R, Selley D (2009) Salt as a fluid driver, and basement as a metal source, for stratiform sediment-hosted copper deposits. Geology, 37, 1107–10.
Kyser TK, Cuney M (2008) Unconformity-related uranium deposits. In: Recent and Not-So-Recent Developments in Uranium Deposits and Implications for Exploration, vol. 39 (eds Kyser K, Cuney M), pp. 161–219. Mineralogical Association of Canada, Quebec. Kyser TK, Renac C, Hiatt E, Holk G, Durocher K (2000) Comparison of Diagenetic Fluids in Proterozoic Basins, Canada and Australia: Implications for Long Protracted Fluid Histories in Stable Intracratonic Basins. GEOCANADA 2000, Calgary. Leach DL, Viets JG, Kozkowski A, Kibitlewski S (1996) Geology, geochemistry and genesis of the Silesia-Cracow zinc–lead district, southern Poland. Society of Economic Geologists Special Publication, 4, 171–81. Leach DL, Bradley D, Lewchuk MT, Symons DTA, de Marsily G, Brannon J (2001) Mississippi Valley-type lead-zinc deposits through geological time: Implications from recent age-dating research. Mineralium Deposita, 36, 711–40. Leach DL, Sangster DF, Kelley KD, Large RR, Garven G, Allen CR, Gutzmer J, Walters S (2005) Sediment-hosted lead–zinc deposits: a global perspective. Economic Geology, 100th Anniversary volume, 261–607. Lede´sert B, Berger G, Meunier A, Genter A, Bouchet A (1999) Diagenetic-type reactions related to hydrothermal alteration in the Soultz-sous-Foreˆts granite. European Journal of Mineralogy, 11, 731–41. Lu¨ders V (1994) Geochemische Untersuchungen an Gangartmineralen aus dem Bergbaurevier Freiamt-Sexau und dem Badenweiler-Quarzriff, Schwarzwald. In: Die Erz-und Mineralga¨nge im alten Bergbaurevier ‘‘Freiamt-Sexau’’, Mittlerer Schwarzwald, vol. 14 (eds Storch DH, Werner W), pp. 173–91. Abhandlungen des Geologischen Landesamtes, Baden-Wurttemberg. Lu¨ders V, Franzke HJ (1993) Formation of hydrothermal fluorite deposits in the southern Black Forest (SW Germany). Part II: geochemical features. In: Current Research in Geology Applied Deposits (eds Fenoll Hach A, Torres Ruiz J, Gervilla F), pp. 739–42. University of Granada, Granada. Ludwig KR, Grauch RI, Nutt CJ, Nash JT, Frishman D, Simmons KR (1987) Age of uranium mineralization at the Jabiluka and Ranger deposits, Northern Territory, Australia: new U–Pb isotope evidence. Economic Geology, 82, 857–74. Maas R (1989) Nd–Sr isotope constrains on the age and origin of unconformity-type uranium deposits in the Alligator Rivers Uranium Field, Northern Territory, Australia. Economic Geology, 84, 64–90. Machel HG (2001) Bacterial and thermochemical sulfate reduction in diagenetic settings: old and new insights. Sedimentary Geology, 140, 143–75. Macquar JC, Lagny P (1981) Mine´ralisations Pb–Zn ‘‘sous inconformite´’’ des se´ries de plate-formes carbonate´es. Exemple du gisement de Tre`ves (Gard). Relations entre dolomitisations, dissolutions et mine´ralisations. Mineralium Deposita, 16, 293– 307. Macquar JC, Rouvier H, Thieberoz J (1990) Les mine´ralisations Zn, Pb, Fe, Ba, F pe´ri-ce´venoles: cadre structuro-se´dimentaire et distribution spatio-temporelle. Documents BRGM, 183, 143– 58. Magri F, Bayer U, Pekdegger A, Otto R, Thomsen C, Maiwald U (2009) Salty groundwater flow in the shallow and deep aquifer systems of the Schleswig-Holstein area (North German Basin). Tectonophysics, 470, 183–94. Mathieu R (1999) Reconstitution des pale´ocirculations fluides et des migrations e´le´mentaires dans l’environnement des re´acteurs nucle´aires naturels d’Oklo (Gabon) et des argilites de Tournemire (France), p. 519. Unpublished Thesis, INPL, Nancy.
Fluid flows and metal deposition near unconformity 291 Mathieu R, Cuney M, Cathelineau M (2000) Geochemistry of paleofluids circulation in the Franceville basin and around Olko natural reaction zones (Gabon). Journal of Geochemical Exploration, 69–70, 245–49. Mercadier J (2008) Conditions de gene`se des gisements d’uranium associe´s aux discordance prote´rozoı¨ques et localises dans les socles. Exemple du socle du basin d’Athabasca (Saskatchewan, Canada), p. 353. Unpublished Thesis, Nancy Universite´. Mercadier J, Richard A, Boiron MC, Cathelineau M, Cuney M (2010) Brine penetration in the basement rocks of the Athabasca Basin through microfracture networks at the P-Patch U deposit (Canada). Lithos, 115, 121–36. Ming-An H, Disnar JR, Sureau JF (1995) Organic geochemical indicators of biological sulphate reduction in the early diagenetic Zn–Pb mineralization: The Bois-Madame deposit (Gard, France). Applied Geochemistry, 10, 149–435. Mo¨ller P, Woith H, Dulski P, Lu¨ders V, Erzinger J, Ka¨mph H, Pekdeger A, Hansen B, Lodemann M, Banks DA (2005) Main and trace elements in the KTB-VB fluid: composition and hints to its origin. Geofluids, 5, 28–41. Muchez P, Heijlen W, Banks D, Blundell D, Boni M, Grandia F (2005) Extensional tectonics and the timing and formation of basin-hosted deposits in Europe. Ore Geology Reviews, 27, 241–67. Muir Wood R, King GCP (1993) Hydrothermal signatures of earthquake strain. Journal of Geophysical Research, 98, 22035–68. Munoz M, Boyce AJ, Courjault-Rade P, Fallick AE, Tollon F (1995) Multi-stage fluid incursion in the Palaeozoic basementhosted Saint-Salvy ore deposit (NW Montagne Noire, southern France). International Journal of Rock Mechanics and Mining Science and Geomechanics Abstracts, 32, 5, 202A. Munoz M, Boyce AJ, Courjault-Rade P, Fallick AE, Tollon F (1999) Continental basinal origin of ore fluids from southwestern Massif Central fluorite veins (Albigeois, France): evidence from fluid inclusion and stable isotope analyses. Applied Geochemistry, 14, 447–58. Munoz M, Premo WR, Courjault-Rade P (2005) Sm-Nd dating of fluorite from the worldclass Montroc fluorite deposit, southern Massif Central, France. Mineralium deposita, 39, 970–75. Munz IA, Yardley BWD, Banks DA, Wayne D (1995) Deep penetration of sedimentary fluids in basement rocks from southern Norway: evidence from hydrocarbon and brine inclusions in quartz rocks. Geochimica Cosmochimica Acta, 59, 239–54. Oliver NHS, McLellan JG, Hobbs BE, Leverley JS, Ord A, Feltrin L (2006) Numerical models of extensional deformation, heat transfer and fluid flow across basement-cover interfaces during basin related mineralisation. Economic Geology, 101, 1–31. Page RW, Williams IS (1988) Age of the Barramundi Orogeny in northern Australia by means of ion microprobe and conventional U–Pb zircon studies. Precambrian Research, 40–41, 21– 36. Pagel M (1975) De´termination des conditions physico-chimiques de la silicification diage´ne´tique des gre`s Athabasca (Canada) au moyen des inclusions fluides. Comptes Rendus de l’Acade´mie des Sciences, Paris, 280, 2301–04. Pagel M, Jaffrezic H (1977) Analyses chimiques des saumures des inclusions du quartz et de la dolomite du gisement d’uranium de Rabbit Lake (Canada). Aspect me´thodologique et importance ge´ne´tique. Comptes Rendus de l’Acade´mie des Sciences, Paris, 284, 113–16. Pagel M, Poty B, Sheppard SMF (1980) Contribution to some Saskatchewan uranium deposits mainly from fluid inclusion and isotopic data. International Uranium Symposium on the Pine Creek Geosyncline, Vienna, pp. 639–54.
Patrier P, Beaufort D, Bril H, Bonhomme M, Fouillac AM, Aumaitre R (1997) Alteration-mineralization at the Bernardan U deposit (Western Marche, France); the contribution of alteration petrology and crystal chemistry of secondary phases to a new genetic model. Economic Geology, 92, 448–67. Pauwels H, Fouillac C, Fouillac AM (1993) Chemistry and isotopes of deep geothermal saline fluids in the upper Rhine Graben: origin of compounds and water–rock interactions. Geochimica Cosmochimica Acta, 57, 2737–49. Peiffer MT (1986) La signification de la ligne tonalitique du Limousin. Son implication dans la structuration varisque du Massif Central franc¸ais. Comptes Rendus de l’Acade´mie des Sciences, Paris, 303, 305–10. Person M, Garven G (1992) Hydrologic constraints on petroleum generation within the continental rift basins; theory and application to the Rhine Graben. American Association of Petroleum Geology Bulletin, 76, 468–88. Pique´ A, Canals A, Grandia F, Banks DA (2008) Mesozoic fluorite in NE Spain record regional base metal-rich brine circulation through basin and basements during extensional events. Chemical Geology, 257, 139–52. Rawlings DJ (1999) Stratigraphic resolution of a multiphase intracratonic basin system: the Mc Arthur Basin, northern Australia. Australian Journal of Earth Sciences, 46, 703–23. Richard A (2009) Circulation de saumures a` la discordance socle ⁄ couverture se´dimentaire et formation des concentrations uranife`res prote´rozoiques (Bassin de l’Athabasca, Canada), p. 234. Unpublished PhD Thesis, Nancy Universite´. Richard A, Cathelineau M, Boiron MC, Cuney M, Banks DA, France-Lanord C, Pettke T (2009) Origin of U-Mineralizing Brines in the Athabasca Basin, Canada, p. 172, A1098. Goldschmidt Conference, Davos. Richard A, Pettke T, Cathelineau M, Boiron MC, Mercadier J, Cuney M, Derome D (2010) Brine-rock interaction in the Athabasca basement (McArthur River U deposit, Canada): Consequences for fluid chemistry and uranium uptake. Terra Nova, in press. Rousset D, Bayer R, Guillon D, Edel JB (1993) Structure of the southern Rhine graben from gravity and reflection seismic data. (ECORS – DEKORP program). Tectonophysics, 221, 135–53. Rouvier H, Henry B, Macquar JC, Leach D, Le Goff M, Thieberoz J, Lewchuk T (2001) Re´aimantation re´gionale e´oce`ne, migration de fluides et mine´ralisations sur la bordure ce´venole (France). Bulletin de la Socie´te´ ge´ologique de France, 172, 503– 16. Russel MJ (1978) Downward excavating hydrothermal cells and Irish-type ore deposits: importance of an understanding thick Caledonian prism. Transactions of the Institution of Mining and Metallurgy, 87B, 161–71. Sanchez V, Vindel E, Martin-Crespo M, Corbella M, Cardellach E, Banks DA (2009) Sources and composition of fluids associated with fluorite deposits of Asturias (N Spain). Geofluids, 9, 338–55. Sanchez V, Cardellach E, Corbella M, Vindel E, Martin-Crespo M, Boyce AJ (2010) Variability in fluid sources in the fluorite deposits from Asturias (N Spain): further evidences from REE, radiogenic (Sr, Sm, Nd) and stable (S, C, O) isotope data. Ore Geology Reviews, 37, 87–100. Sausse J (2002) Hydromechanical properties and alteration of natural fracture surface in the Soultz granite (Bas Rhin, France). Tectonophysics, 348, 169–85. Sausse J, Fourar M, Genter A (2006) Permeability and alteration within the Soultz granite inferred from geophysical and flow log analysis. Geothermics, 35, 544–60.
292 M.-C. BOIRON et al. Schuler C, Steiger RH (1978) On the genesis of felspar megacryst in granites: Rb ⁄ Sr isotopic study. In: 4th Conference of Geochronology Isotope Geo (ed Zartmann RE), pp. 386–87. Shaw A, Downes H, Thirlwall MF (1993) The quartz diorites of Limousin: elemental and isotopic evidence for Devono-Carboniferous subduction in the Hercynian belt of the French Massif Central. Chemical Geology, 107, 1–18. Sims PK, Stein HJ (2003) Tectonic evolution of the Proterozoic Colorado province, Southern Rocky Mountains: a summary and appraisal. Rocky Mountain Geology, 38, 183–204. Smith MP, Savary V, Yardley BWD, Valley JW, Royer JJ, Dubois M (1998) The evolution of the deep flow regime at Soultzsous-Foreˆts, Rhine graben, eastern France: evidence from a composite quartz vein. Journal of Geophysical Research, 103, B11, 27223–37. Stille P, Gauthier-Lafaye F, Bros R (1993) The Neodymium isotope system as a tool for petroleum exploration. Geochimica Cosmochimica Acta, 57, 4521–25. Sweet IP, Brakel AT, Carson L (1999) The Kombolgie subgroup – a new look at an old ‘‘formation’’. AGSO Research Newsletter, 30, 26–28. Symons DTA, Sangster DF, Leach DL (1995) A Tertiary age from paleomagnetism for Mississipi Valley-type zinc lead mineralization in Upper Silesia, Poland. Economic Geology, 90, 782– 94. Symons DTA, Lewchuk MT, Kawasaki K, Velasco F, Leach DL (2009) The Reocin zinc–lead deposit, Spain: paleomagnetic dat-
ing of a late Tertiary ore body. Mineralium Deposita, 44, 867– 880. Todt W (1976) Zirkon U ⁄ Pb alter des Malsburger granits vom Su¨d Schwarzwald. Neues Jahrbuch fu¨r Mineralogie (Monatshefte), 12, 532–44. Verma S, Santoyo E (1997) New improved equations for Na ⁄ K, Na ⁄ Li and SiO2 geothermometers by outlier detection and rejection. Journal of Volcanic and Geothermal Research, 79, 9–23. Wilkinson JJ, Everett CE, Boyce AJ, Gleeson SA, Rye DM (2005) Intracratonic crustal seawater circulation and the genesis of subseafloor zinc-lead mineralization in the Irish orefield. Geology, 33, 805–08. Wilkinson JJ, Stoffell B, Wilkinson CC, Jeffries TE, Appold MS (2009) Anomalously metal-rich fluids form hydrothermal ore deposits. Science, 323, 764–67. Yardley BWD (2005) Metal concentrations in crustal fluids and their relationship to ore formation. Economic Geology, 100, 613–32. Yardley BWD, Banks DA, Barnicoat AC (2000) The chemistry of crustal brines: tracking their origins. In: Hydrothermal Iron Oxide Copper Gold and Related Deposits, A Global Perspective (ed Porter TM), pp. 61–70. Australian Mineral Foundation, Townsville. Ziserman A (1980) Le gisement de Chaillac (Indre): la barytine des Redoutie`res, la fluorine du Rossignol. Association d’un gıˆte stratiforme de couverture et d’un gıˆte filonien du socle. 26 CGI, Gisements franc¸ais, Fascicule E3.
Magmatic fluids immiscible with silicate melts: examples from inclusions in phenocrysts and glasses, and implications for magma evolution and metal transport VADIM S. KAMENETSKY AND MAYA B. KAMENETSKY ARC Centre of Excellence on Ore Deposits and School of Earth Sciences, University of Tasmania, Hobart, Tas., Australia
ABSTRACT The first occurrence of immiscibility in magmas appears to be most important in the magmatic–hydrothermal transition, and thus studies of magmatic immiscibility should be primarily directed towards recognition of coexisting silicate melt and essentially non-silicate liquids and fluids (aqueous, carbonic and sulphide). However, immiscible phase separation during decompression, cooling and crystallization of magmas is an inherently fugitive phenomenon. The only remaining evidence of this process and the closest approximation of natural immiscible magmatic liquids and vapours can be provided by melt and fluid inclusions trapped in silicate glasses and magmatic phenocrysts. Such inclusions are often used as a natural experimental laboratory to model the process of exsolution and the compositions of volatile-rich phases from a wide range of terrestrial magmas. In this paper several examples from recent research on melt and fluid inclusions are used to demonstrate the significance of naturally occurring immiscibility in understanding some large-scale magma chamber processes, such as degassing and partitioning of metals. Key words: aqueous fluid, carbonate melt, fluid inclusions, immiscibility, melt inclusions, ore deposits, silicate magmas, sulphide melt Received 26 May 2009; accepted 14 November 2009 Corresponding author: Vadim S. Kamenetsky, ARC Centre of Excellence on Ore Deposits and School of Earth Sciences, University of Tasmania, Hobart, Tas. 7001, Australia. Email:
[email protected]. Tel: +61362267649. Fax: +61362262547. Geofluids (2010) 10, 293–311
INTRODUCTION Orthomagmatic models correctly assume that many types of economic mineralization (e.g. Cu–Ni–PGE sulphide deposits, W–Sn skarns and greisens, pegmatites, and Cu–Mo–Au porphyries) relate to magmas and magmaderived fluids. However, among many unanswered questions, there is a mass balance problem when linking magmas and ore deposits. In other words, a common magma-derived ore deposit requires either an unusually metal-rich parental magma or very efficient mechanisms of metal extraction from large magma volumes. It is most likely that both prerequisites are simultaneously necessary to account for formation of ore. Magmas are not routinely available for study, and the rocks that are the products of crystallized magmas are always affected by late-magmatic redistribution of elements, syn-emplacement degassing, and post-magmatic alteration. Likewise, the magma volumes are not easily assessable, and the efficiency of metal extraction from these magmas is poorly known at present.
Existing data on phases transitional between silicate melts and hydrothermal fluids are sparse. This is because igneous petrology deals with processes and composition at temperatures above magma solidification (>800C), whereas temperatures associated with ore deposition (excluding magmatic sulphides) rarely exceed 500–600C, thus leaving a gap in our knowledge of the critical magmatic–hydrothermal transition. It is known, however, that a variety of phases form and disappear in the transition from magmas to ‘mineralizing fluids’. These phases are believed to be genetically related to near-solidus magmatic CO2 and H2O- rich fluids and magma-derived liquids, enriched in sulphide, chloride, carbonate, sulphate, phosphate and other non-silicate components. The origin of ‘mineralizing fluids’ can be linked ultimately to immiscibility (unmixing) in parental silicate magmas (Roedder 1992). The compositional divergence between unmixed phases is extreme, and even though the physical amount of the new phase may be small, its separation and transport can be important (Roedder 1992,
Frontiers in Geofluids, 1st edition. Edited by Bruce Yardley, Craig Manning and Grant Garven. © 2011 by Blackwell Publishing Ltd.
294 V. S. KAMENETSKY & M. B. KAMENETSKY 2003), especially with respect to the formation of many types of ore deposits (Bodnar 1995; Candela & Piccoli 1995). We usually observe only the consequences of unmixing processes, as records of the exact moment of immiscibility do not exist in nature. Additionally, the exact compositions of immiscible phases have proved extremely difficult to document and understand (Kamenetsky et al. 2003; Kamenetsky 2006), largely because of their transient, reactive qualities and small quantities. It has been suggested that melt and fluid inclusions hosted by magmatic phenocrysts and silicate glasses represent ‘snapshots’ of melts or fluids, respectively, at the time of crystallization (e.g. Roedder 1979, 1984; Bodnar 1995; Sobolev 1996; Frezzotti 2001; Danyushevsky et al. 2002; Lowenstern 2003). Although there still remain questions as to the extent to which inclusions are modified after trapping, the application of modern analytical tools in studying inclusions (FTIR and Raman spectroscopy, electron, ion and proton microprobes, scanning electron microscopy, laser ablation ICPMS, etc.) provide great advantages over conventional methods. In this contribution we will review unequivocal occurrences of immiscibility in silicate magmas as recorded in melt and fluid inclusions from a variety of volcanic and intrusive suites in different tectonic settings. This paper is also intended to provide advice for those trying to make sense from the study of similar inclusions. For brevity, most examples given below are from studies in which the authors had first-hand experience.
WHY STUDY MAGMATIC INCLUSIONS? As minerals crystallize from a magma or fluid, they often entrap tiny portions of the media from which they are growing, which then become melt or fluid inclusions (e.g. Roedder 1979, 1984). Trapped as they are, and armoured within the host phase, these inclusions are largely excluded from any further reactions or processes that might take place in an evolving (due to cooling, crystallization, decompression, mixing, degassing, contamination, etc.) magma-fluid environment, and thus record information on a particular evolutionary stage of a magmatic system. Petrological, thermometric, geochemical and isotopic data gathered from melt and fluid inclusions, and appropriate interpretations, are gradually replacing ‘classical’ methods of studying whole rock and natural glass compositions (e.g. Sobolev 1996; Schiano 2003; Kent 2008). For a given magmatic suite, studies of magmatic inclusions provide a far more realistic record of the changing physical conditions and magma composition than a study of whole rocks, which usually represent the net effects of many complex factors, including mixing of liquids formed at various partial melting degrees at various depths in equilibrium with different restite lithologies, heterogeneous mantle sources, chromatographic reactions between rising
partial melts, contamination in the crust, crystal fractionation and ⁄ or accumulation and inevitable degassing. As magmatic volatiles are rarely preserved during eruption or emplacement of mantle-derived magmas into the crust, melt and fluid inclusions provide a reliable record of the evolving magmatic volatile budget (e.g. Metrich & Wallace 2008). Furthermore, magmatic inclusions offer the possibility of ‘seeing through’ alteration, and enabling a much better estimate of pristine magmatic compositions than might be available from whole rock analyses (Kamenetsky et al. 2001b). A methodology for studying immiscible melt and fluid inclusions The successful study of natural samples of immiscible melts and fluids, preserved as magmatic inclusions, requires significant time and effort, and more importantly, a wellbalanced methodological approach. The latter may include sophisticated sample preparation, detailed petrographic and experimental studies and application of in-situ analytical techniques (e.g. Roedder 1984, 1992; Rankin et al. 1992; Bodnar 1995; Heinrich et al. 2003; Kamenetsky et al. 2003; Webster 2004; Rickers et al. 2006; Thomas et al. 2006; Zajacz et al. 2008 and references therein). Sample preparation is the first important step, which should ensure that immiscible phases are preserved, located, described and analysed. A researcher should be aware that melt and fluid inclusions are not always widespread in their host minerals and glasses, and in most cases they are very small (<50 lm), and can be partially or completely lost when exposed by polishing. A few hundred grains should be prepared for examination either in double sided polished sections (with a thickness determined by grain size) or mounted in epoxy resin. Grinding and polishing is better performed in kerosene (or any other light oil lubricant), as non-silicate phases can be water-soluble. In many cases these phases, when exposed on the grain surface, are highly reactive with atmospheric moisture. Thus, they should be preserved inside their hosts for qualitative or semi- quantitative analysis and analysed quantitatively immediately after exposure. As many host grains as possible should be optically examined, using a petrographic microscope, to identify temporal populations of inclusions and their genetic relationships. Primary inclusions are a principal target in the study of magmatic immiscibility, which can be identified in inclusions with variable amounts of two or more supposedly unmixed but coexisting phases (Roedder 1992). Commonly one or more phases in such inclusions have spherical shapes. A researcher should be always on the lookout for non-silicate phases that may represent immiscibility in the silicate melt (e.g. salt crystals, CO2 and H2O vapours and liquids, sulphide globules, etc.) in otherwise normal silicate
Magmatic fluids immiscible with silicate melts 295 melt inclusions. Heterogeneously trapped immiscible phases are difficult to recognize prior to heating of melt inclusions, because of the masking effects of post-entrapment crystallization. To overcome this problem the use of heating and quenching techniques is recommended. Details of thermometric experiments with melt inclusions are available in the modern literature. In the study of immiscibility, heating with visual control is used to melt crystalline silicate daughter phases within devitrified inclusions, liberate immiscible non-silicate phases and promote their coalescence into larger formations (globules or bubbles) in order to successfully analyse them using the micro-beam techniques (Kamenetsky et al. 2003). In some cases experimental heating and cooling may initiate immiscibility within silicate melt inclusions that were trapped homogeneously and naturally quenched above the temperature of immiscibility. Behaviour (e.g. melting, homogenization, crystallization, etc.) inside immiscible phases during thermometric experiments can provide additional constraints on the composition and temperature of phase transformations. Micro-beam methods are now routinely used for analysis of mineral constituents and bulk chemical composition of immiscible phases in melt and fluid inclusions. A silicate melt is usually quenchable into homogeneous glass that can be exposed by polishing and analysed by conventional electron micro-probe methods (EMP) and laser ablation inductively coupled plasma-mass spectrometry (LA–ICPMS). Other immiscible phases, either gaseous or those that can not be quenched to a glass, because of their extremely low viscosity, should be analysed while they are within host grains. Ideally, each inclusion should be first analysed for its individual components by non-destructive, high spatial resolution techniques, such as laser Raman spectroscopy, micro-infrared spectroscopy (FTIR), secondary ion mass spectrometry (SIMS) and proton-induced X-ray emission (PIXE), and then by destructive, but highly sensitive, LA–ICPMS. The latter will provide bulk chemical composition, either as element ratios or absolute concentrations. Immiscibility in silicate magmas recorded by magmatic inclusions Magma-derived fluids are an essential component of virtually all magmatic systems, especially at low pressure, when silicate melts attain saturation in volatiles. Carbon dioxide, H2O, S, Cl and F species are the most voluminous components in magmatic fluids, and their relative proportions are highly variable, depending on the magma composition, temperature and pressure. Unmixing occurs when magmas attain saturation in one or more volatiles, and in many cases this profoundly affects further magma evolution (e.g. the composition of crystallizing phases and the style of
eruption). The composition of the immiscible volatile phase in magmas has been well documented through studies of gaseous emissions associated with volcanism (Symonds et al. 1994). Pioneering work by Edwin Roedder on the silicate melt immiscibility, represented by magmatic inclusions in quartz from granites of Ascension Island (Roedder & Coombs 1967), founded a new approach in the studies of magmatic immiscibility. Since then this method has seen enormous development and expanded rapidly over the last decade, due largely to advances in micro-analytical techniques, and has been at the centre of many recent petrological and geochemical discoveries. We present some examples that collectively demonstrate occurrence and significance of compositionally diverse fluids immiscible with common silicate magmas. Silicate melt–CO2 fluid immiscibility Immiscible CO2 fluids are by far most common in basaltic magmas, because the solubility of CO2 in the silicate melt is strongly reduced at near-surface pressures, in contrast to that of H2O (e.g. Lowenstern 2001). They are represented by near-spherical bubbles of varying density (Figs 1 and 3) in pillow-rim glasses and early magmatic phenocrysts (olivine, plagioclase and clinopyroxene) from mid-ocean and backarc ridges, continental flood basalt suites, oceanic islands and less commonly immature island arcs (Kamenetsky et al. 2002a). Fluid bubbles with essentially CO2 compositions (detected by Raman spectroscopy) and sizes ranging from <5 to 100 lm are typically present in all submarine basaltic glasses (e.g. Moore et al. 1977; Moore 1979; Javoy & Pineau 1991; Kamenetsky et al. 2002a). Apart from their gaseous component, the bubbles often contain solid phases that decorate the bubble-glass interface (Figs 1 and 5). Aggregates of crystals on bubble walls often appear spherical in shape, however they have distinct facets at high magnification (Fig. 3E,F), and in most cases are Cu-bearing Fe-sulphides –Figs 3A,E, 4, 5), but carbonate and sulphate minerals also occur (Figs 1, 2, 3F). Crystal aggregates are partly embedded in the vesicle wall in hemispherical pits, which are observed when the cluster falls out (Fig. 3E). The presence of pits implies that the crystal aggregates were present as solids on the bubble walls, embedded in the meniscus, while the host was still liquid or plastic. Thus, the formation of sulphides inside the CO2 fluid bubbles relates to the magmatic stage. The presence of significant amount of Cu (up to several wt%, Moore & Calc 1971) in the Fe-sulphides, decorating the bubble-glass interface, is best illustrated by the PIXE mapping (Fig. 5A). Laser ablation ICPMS compositional profiles, recorded through the host glass and sulphidebearing bubbles, also show highly elevated Cu and Ag, once the bubbles are opened by the laser beam (Fig. 4).
296 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
(I)
(A)
(D)
(B)
(E)
Fig. 1. Precipitates in heterogeneously trapped fluid bubbles from glassy melt inclusions hosted in primitive olivine (A–H) and clinopyroxene (I). Samples are basalts from the Okinawa Trough (A), eastern Manus backarc basin (B, C), Troodos upper pillow-lavas, Cyprus (D), Hunter Ridge (E), Vanuatu Troughs (F) and Hawaii (G); an andesite from Pukeonake cone, New Zealand (H); and a mafic clast from Ventotene Island, Italy (I). Scale bars are 50 lm.
(C)
(F)
This enrichment can be tentatively linked to gradual depletion of the surrounding glass in Cu (decreasing Cu ⁄ Ca, Fig. 4), which suggests metal diffusion from the melt to the sulphide-forming fluid. Partitioning of Cu from magmas into immiscible magmatic fluids (Lowenstern et al. 1991; Lowenstern 1993; Yang & Scott 1996) should have a profound effect on the composition of residual silicate melts. We demonstrate this point using Macquarie Island basaltic glasses,
Fig. 2. CO2-rich fluid inclusions hosted by primitive olivine in basalts from the Mid-Atlantic Ridge, 43oN (A), Macquarie Island (B), eastern Manus backarc basin (C), Hunter Ridge (D), Heard Island (E) and a picrite from Emeishan, SW China. Note the abundant precipitates decorating the olivine-bubble interface (A–D), and liquid-rich fluids (E, F). Scale bars are 50 lm (solid line) and 25 lm (dashed line).
in which Cu-bearing fluid bubbles (Figs 3A,D, 4, 5A) were found. Macquarie Island ophiolite, an exposed part of the Macquarie Ridge at the Australia–Pacific plate boundary south of New Zealand, contains well-preserved glasses in pillow lavas and hyaloclastites. The glass compositions range from ultra-enriched to ‘normal’ midocean basalt (Kamenetsky et al. 2000; Kamenetsky & Maas 2002). Two compositional groups of melts, based on Mg# - K2O (La ⁄ Sm) relationships, are represented by
Magmatic fluids immiscible with silicate melts 297
(A)
Fig. 3. Precipitates lining the walls of CO2-filled vesicles in oceanic basalt glasses from the Macquarie Island ophiolite, SW Pacific (A–D), Juan de Fuca Ridge (E) and south Atlantic Ridge (F). Precipitates are Cu-bearing Fe monosulphide (A–E) and carbonate (F). Transmitted light photos of unopened vesicles (A–D) and back-scattered electron images of exposed vesicles (E, F). Scale bars are 50 lm.
(B)
(D)
(C)
(E)
(F)
50 µm 105
Glass
Bubble + glass
Ca
Signal intensity
Fig. 4. Time-resolved laser ablation signal (y axis, counts per second) recorded for selected elements (masses) in basaltic glass (Macquarie Island ophiolite) and CO2- and precipitate-bearing bubble in this glass. Initially, carrier gas background was acquired and then the laser was turned on for ablating the glass. The sample was moved under the laser beam, and finally the bubble was breached. Signals for 65Cu and 107Ag increase as the bubble wall is ablated. Note the gradual decrease in Cu ⁄ Ca ratio in the compositional profile towards the bubble. For analysis the bubble was brought by polishing to within 10 lm of the surface.
Background
Ca43 Cu65 Zn66 Ag107
104
Cu 103
Zn 102
Ag A 101
near-primary (highest Mg#, saturated with olivine and Cr spinel only) and fractionated (reduced Mg#, relatively low Al2O3 and CaO, saturated with olivine, plagioclase and clinopyroxene) glasses. The near-primary melts indicate compatible behaviour of both Cu and Zn during mantle melting, whereas in the fractionated melts these metals are strongly decoupled (Fig. 5B). Decompression and cooling of the primary melts are accompanied by crystallization and degassing, which is the most likely explanation of Cu depletion in the fractionated melts (Fig. 5B) and Cu enrichment in the glass-hosted fluid bubbles (Figs 4, 5A). Carbon dioxide bearing fluid inclusions and bubbles in glass inclusions in primitive olivine from a variety of basaltic suites (Kamenetsky et al. 2002a) often contain crystal-
Cu/Ca
Time
line and amorphous precipitates lining the walls of cavities and freestanding crystals (Figs 1, 2A,D). The relative volumes of such precipitates vary considerably between inclusions. Single clear, sometimes, birefringent crystals are represented by variably sized, equant to lath-shaped grains, and often form aggregates (Figs 1, 2A,D). There is a general correlation between bubble to glass ratio for a given inclusion, and the density (amount) of precipitates (Kamenetsky et al. 2001a). Whereas bubbles with the smallest relative volume, carrying little or no precipitates, represent ‘shrinkage bubbles’, those with larger relative volume and more precipitates are interpreted to have been heterogeneously trapped with a melt. Thus the precipitates in such bubbles can be related to crystallization within the immiscible vapour phase.
298 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
(B) 120 110
Cu, ppm
100 90 80 70 60 50 0
2
4
6
8
10
90
Zn, ppm
80 70 60 50 40 0
2
4
6
La/Sm
The precipitates in fluid bubbles are mainly carbonates (magnesite, calcite, dolomite, siderite, ankerite, and rare rhodochrosite and nahcolite), and their presence is consistent with a CO2 –rich fluid (Kamenetsky et al. 2002a). Other phases identified by laser Raman spectroscopy include sulphates (abundant anhydrite and gypsum, and single grains of barite and anglesite) and sulphides (mainly pyrite, and also chalcopyrite and arsenopyrite). Silicate and oxide minerals appear to be present also, although they are not ubiquitous, or at least cannot always be detected using Raman spectroscopy (e.g. salts). The bulk composition of fluid bubbles and the origin of bubble-hosted precipitates is a subject for future research.
(A)
(D)
(B)
(E)
8
10
Fig. 5. Possible relationships between compositions of exsolving fluid (A) and residual melt (B) in magmatic system of the Macquarie Island ophiolite. (A) Optical image and proton-induced X-ray emission (PIXE) Cu map of the fluid bubble with translucent and opaque precipitates showing highly elevated Cu contents (7000 ppm) in the magmatic derived fluid. (B) Cu-rich composition of degassing fluid can result in depletion of fractionating melts (open circles) in Cu compared to their parental melts (closed circles). La ⁄ Sm is used as a proxy for extent of mantle melting. Vertical arrows indicate behaviour of metals (Cu decreasing, Zn increasing) during magma chamber processes, such as degassing and crystallization.
Sulphide spherules decorating vesicles in quenched submarine glasses (e.g. Moore & Calc 1971; Mathez 1976; Yeats & Mathez 1976; Moore et al. 1977; Moore 1979; Alt et al. 1993) and carbonates in CO2-rich fluid bubbles in olivine-hosted melt inclusions (e.g. Sobolev & Nikogosian 1994; Kamenetsky et al. 2001a, 2002a, 2007) have been interpreted to have formed by reaction of sulphurous or CO2-rich volatiles with elements (e.g. Fe, Cu, Mg, Ca, etc.) diffusing from the host melt or crystal (Fig. 4). Alternatively, all the required elements may have been dissolved in the fluid prior to the formation of precipitates (e.g. Lowenstern et al. 1991; Solovova et al. 1991; Lowenstern 1995; Yang & Scott 1996; de Hoog & van Bergen 2000).
(C)
(F)
Fig. 6. Aqueous fluid bubbles in quartz (A, B), quartz-hosted melt inclusions (C–E) and pillowrim glass (F). Samples are rhyolites from Taupo Volcanic Zone, New Zealand (A–C), Rio Blanco, Chile (D, E) and dacite from eastern Manus backarc basin (F). (A, B) Aligned fluid inclusions likely represent fluid expelled during decrepitation of melt inclusions (shown by arrows); (C–F) single and multiple aqueous fluid bubbles with varying density (from vapour-rich to liquid-rich). Scale bars are 25 lm.
Magmatic fluids immiscible with silicate melts 299 Fluid bubbles with liquid CO2 (Fig. 2E,F; e.g. Roedder 1965), although rarely observed, even in basaltic samples, should be studied by microthermometric experiments to determine the CO2 density, which taken together with temperature of crystallization, is helpful in constraining crystallization pressure. The latter can also be determined using the bulk volatile content of silicate melt inclusions (e.g. Anderson & Brown 1993; Lowenstern 2001; Newman & Lowenstern 2002; Metrich & Wallace 2008) after CO2, contained in shrinkage fluid bubbles (up to 80% of the total CO2 in the melt at the time of trapping) is dissolved back to the melt in homogenization experiments (Cervantes et al. 2002). Silicate melt–aqueous saline fluid immiscibility Decompression and crystallization of magmas, especially hydrous subduction-related magmas, results in water saturation. Water can be a significant component in magmatic fluids coexisting with mafic to felsic melts, as observed in arc ⁄ backarc glasses (Fig. 6F; Sobolev & Danyushevsky 1986) and silicate melt inclusions (e.g. Roedder & Coombs 1967; Dunbar et al. 1989; Lowenstern & Mahood 1991; Solovova et al. 1991; Danyushevsky et al. 1992; Dunbar & Hervig 1992; Dunbar & Kyle 1993; Naumov et al. 1993, 1994, 1996; Lowenstern 1995; Thomas & Webster 2000; Lowenstern 2003; Davidson et al. 2005; Kamenetsky 2006; Spilliaert et al. 2006; Thomas et al. 2006; Davidson & Kamenetsky 2007; Zajacz et al. 2008). The degassed amounts of H2O can be quantified by comparison of the H2O contents of quenched glasses with melt inclusions from the same suites (e.g. Sobolev & Chaussidon 1996). Although immiscibility between silicate melts and aqueous fluids can be approached in many different ways, the study of natural coexisting inclusions of the immiscible phases has an advantage of direct observation and allowing mass-balance calculations. Once a H2O-bearing silicate melt is entrapped and isolated within host phenocrysts, crystallization of daughter silicate minerals increases volatile abundances in the residual melt and reduces internal pressure. As a result, the melt reaches saturation in one or more volatiles, and a separate fluid phase appears. In the case of quartz-hosted melt inclusions the build-up of fluid pressure can cause hydraulic fracturing. Decrepitation, possibly occurring at b–a quartz inversion at 573C, can explain the spatial association of crystallized melt inclusions with swarms of elongated, often tubular, aqueous vapourrich and two-phase inclusions (Fig. 6A,B), some with salt crystals. These either heal fractures radiating from silicate melt inclusions, or decorate what appear to be oriented structural defects (dislocations, channels) in the quartz (Fig. 6A,B). In many cases the exsolution of the aqueous fluid may occur during, or prior to, crystallization (and trigger crys-
tallization of plagioclase, whose liquidus temperature is significantly depressed by H2O). Individual fluid inclusions and fluid bubbles, which comprise a large proportion of the melt inclusion volume (Fig. 6C–F), are a good indication of syn-crystallization saturation in volatiles and immiscibility. However, if crystallization occurred earlier than saturation in H2O, an immiscible aqueous fluid would not be trapped, and the search for inclusions representing such fluid can be a tiring and futile task. On the other hand, a homogeneous H2O-undersaturated melt trapped at higher pressure should crystallize and evolve in a closed system (such as within a melt inclusion) towards saturation and immiscibility (e.g. Naumov 1979; Roedder 1984; Thomas & Webster 2000). Glass inclusions described in many studies (e.g. Roedder & Coombs 1967; Lowenstern & Mahood 1991; Solovova et al. 1991; Naumov et al. 1993, 1994; Lowenstern 1995; Naumov et al. 1996; Thomas & Webster 2000; Lowenstern 2003; Davidson et al. 2005; Thomas et al. 2006; Davidson & Kamenetsky 2007; Zajacz et al. 2008) may contain in situ exsolved (i.e. ‘shrinkage’) aqueous vapour, aqueous liquid, or both, sometimes with cubic chloride crystals (Fig. 7A,B). Silicate melt – aqueous fluid immiscibility that occurs in homogeneously trapped silicate melt inclusions during natural cooling can be distinguished from heterogeneous trapping by examining the relative proportions of silicate melt and aqueous fluid in individual inclusions. Bubbles of an aqueous fluid typically have a spherical shape (Figs 6C,D, 7A,B), although sometimes they appear to be plastically deformed by the contracting glass (Fig. 6F; Davidson & Kamenetsky 2007). Aqueous onephase liquid bubbles in silicate glass (Fig. 7A,B) are likely to be a metastability phenomenon, reflecting the surface tension effects in very small containers (Roedder 1984). The nucleation of vapour phase inside such bubbles may occur during heating and freezing experiments, and PIXE or Raman measurements (Fig. 7A). Crystal-free, one- and two-phase aqueous bubbles may have elevated salinity (up to 17 wt% NaCl eq.), estimated from the melting temperature of ice in freezing experiments (Davidson & Kamenetsky 2007). Heating stage experiments are used to understand phase changes inside aqueous bubbles, and where possible to induce melt-fluid immiscibility. Details of such experiments are rarely published, so we refer here to a study of Taupo rhyolites (Davidson & Kamenetsky 2007). In addition to fluid bubbles at room temperature, new bubbles form at 450C, and an aqueous liquid phase may appear in originally vapour-rich bubbles (Fig. 7C,D). Clear cubic minerals (salts) crystallize in such bubbles upon cooling (Fig. 7E,J). More bubbles appear in the inclusions at 573C, following expansion of the host quartz by 1 vol.%. At >630C the bubbles shrink and dissolve, resulting in a homogeneous melt at a minimum temperature of
300 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
(B)
(C)
(D)
(E)
560°C
(F)
(G)
(H)
(I)
730C. Using repeated cycling of temperature between 650 and 800C it is possible to coalesce several bubbles into one and observe a vapour-liquid phase boundary. Such ‘artificially’ produced bubbles homogenize into liquid, and crystallize salts upon cooling below 400C (Figs 7E,H, 8B,C). Microbeam analyses of hydrous saline bubbles demonstrate the presence of chlorine and bromine, and a number of metals, most importantly Cu and Ag, and to a lesser extent Zn, Mn, Ba, and Pb (Fig. 8). The exceptionally high Cu abundances (up to 3–3.5 wt%) in post-heating bubbles (Fig. 8B,C) can be a result of Cu diffusion through quartz from the experimental environment (Kamenetsky & Danyushevsky 2005; Zajacz et al. 2009), however, Cu, and sometimes Ag, are also enriched in naturally occurring aqueous bubbles from the same samples (Fig. 8A). The strong partitioning of Cu into magmatic-derived aqueous fluids, recorded by studies of natural samples (e.g. Bodnar 1995; Kamenetsky et al. 2001a; Audetat & Pettke 2003; Harris et al. 2003; Sun et al. 2004; Davidson et al. 2005; Cauzid et al. 2007; Davidson & Kamenetsky 2007; Zajacz et al. 2008), raises the question of the fate of Cu
(J)
Fig. 7. Aqueous saline fluid bubbles in quartzhosted melt inclusions from Taupo volcanic zone rhyolites. (A, B) Unheated partly recrystallized melt inclusions with enlargements of shrinkage bubbles filled with aqueous liquid and crystal (note that the vapour phase in (A), right hand close-up only appeared after laser Raman analysis, prior to this it was single phase aqueous liquid only). (C, D) A partly crystallized melt inclusion showing the exsolution of bubbles and the appearance of a phase boundary in the existing bubble upon heating to 560C. (E) Close up of the bubble in this inclusion (boxed in D) after cooling to room temperature, note clear crystals in the liquid. (F–J) Liquid-rich (F–H) and vaporrich (I, J) crystal-bearing bubbles formed during heating of melt inclusions. (F, G) Enlargements of inclusions shown on Figs 8B,C, respectively; (H) and (I) are within 10 lm in the same inclusion. Solid scale bars are 5 lm, dotted scale bars are 20 lm. All photos are at room temperature, except (D).
and other fluid-compatible metals (e.g. Au, Ag, Re, etc.) in the context of syn-eruptive degassing. For example, strongly outgassed rhyolitic glasses of the Okataina (Taupo Volcanic Zone) almost lack any Cu, in contrast to their precursor melts (Kamenetsky & Danyushevsky 2005). This, taken together with almost complete exhaustion of H2O in the Okataina glasses and their relative depletion in Cl, suggests that Cu was lost from the magmas in an aqueous saline fluid. Tentative calculations, applied to the 47 km3 of the Okataina rhyolites erupted over 20 ka, would suggest a total of 107 tons of released Cu, or 1300 kg ⁄ day. This seems realistic if compared to 300 kg Cu daily flux from White Island (Le Cloarec et al. 1992; Hedenquist et al. 1993), a relatively small active volcano in the same volcanic province. This amount of Cu would be sufficient to form a giant deposit, providing suitable trapping and precipitation conditions existed. Silicate melt–hydrosaline melt immiscibility Exsolution of a moderately saline aqueous fluid from a silicate melt, and its later separation into vapour and brine fluids, has been a cornerstone of the orthomagmatic ore deposit model (e.g. Burnham 1979). It was also
Magmatic fluids immiscible with silicate melts 301
(A) Cu
105 104
Ag
103 102 20 µm
101
(B)
30 µm
Cl
Cu
(C)
Fig. 8. Optical images, time-resolved laser ablation signal (A) and proton-induced X-ray emission element maps (B, C) of aqueous fluid bubbles inside unheated (A) and heated (B, C) melt inclusions in rhyolitic quartz from Taupo Volcanic Zone, New Zealand. The colour scale in each element image is normalized to its own maximum. The outlines on the element maps mark the boundaries of the melt inclusions and spherical globules. Note strong partitioning of Cl and Cu (and other metals) from the silicate melt into immiscible bubbles.
30 µm
demonstrated that the melt composition controls ‘the type of volatile phase that exsolves’ (Webster 2004), and at certain conditions some magmas are capable of direct exsolution of highly concentrated chloride liquids (brine, salt melt, or hydrosaline fluid; see definitions in Halter & Webster 2004). Given the existing evidence from magmatic inclusions, a continuum from low density aqueous vapours (e.g. Cline & Bodnar 1994; Shinohara 1994; Bodnar 1995) to hydrosaline melts (e.g. Roedder & Coombs 1967; Reyf & Bazheyev 1977; Reyf 1984; Solovova et al. 1991; Frezzotti 1992; Rankin et al. 1992; Roedder 1992; Solovova et al. 1992; Lowenstern 1994; De Vivo et al. 1995; Reyf 1997; Kamenetsky et al. 1999; Fulignati et al. 2001; Gilg et al. 2001; Audetat & Pettke 2003; Harris et al. 2003; Kamenetsky et al. 2004b; Renno et al. 2004; Davidson & Kamenetsky 2007; Zajacz et al. 2008), coexisting with each other and the silicate melt, occurs in nature.
Mn
Cl
Cu
Zn
Pb
A rare case of in-situ immiscibility in inclusions of homogeneously trapped melt is presented on Fig. 9A. Despite variable size these inclusions in the core of leucite crystal show approximately similar ratios of brownish alkaline silicate glass, translucent crystalline mass of chlorides, carbonates, and sulphates (according to Raman spectroscopy) and Cu-bearing Fe-sulphide globule, usually on the meniscus between silicate and non-silicate phases. In all other samples, studied by the authors to date, heterogeneous trapping of immiscible silicate and hydrosaline melts in so-called composite inclusions (with variable proportions of co-trapped phases, Figs 9B,I, 10, 11) is dominant. After unmixing the immiscible melts can be incorporated in a growing crystal as coexisting inclusions of pure end-member compositions (Figs 10D, 11; e.g. Audetat & Pettke 2003; Kamenetsky et al. 2004b). Co-trapping of silicate melt with hydrosaline liquid is most evident in the inclusions that escaped post-trapping
302 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
(I)
crystallization of their silicate melt component (i.e., glass; Figs 9B,I, 10D). However, in most cases fine-grained crystalline silicate masses inside composite inclusions obscure evidence for non-silicate phases that can be seen after experimental heating melts silicate and liberates non-silicate phases. Heating of hydrosaline melt immiscible globules trapped together with the silicate melt (Fig. 10A,C) shows no mixing even at temperatures exceeding the liquidus (900–1100C). First melting in the globules occurs at 150–190C, followed by the vapour bubble acquiring a spherical shape and moving in the continuously forming melt. Further heating results in complete melting at 670– 735C and bubble dissolution at 740–935C (Fig. 10A,C). Hydrosaline melt, entrapped homogeneously and in composite inclusions in the same phenocryst (Fig. 10D; e.g. Harris et al. 2003; Kamenetsky et al. 2004b), show similar behaviour during heating, although temperatures of phase transformations can vary for different samples. Experimental heating has also proved informative when the silicate component crystallized on the wall of the cavity and mixed back with the hydrosaline melt component during natural cooling (Fig. 10B, for details see De Vivo et al. 2006). Post-heating, melt inclusions consist of silicate glass and spherical globules, containing micro-crystals and a significant quantity (up to 50 vol.%) of vapour phase (Fig. 11). The number (from 1–2 to 100s) and sizes (<1–15 lm) of globules vary significantly even in neighbouring composite inclusions. In larger inclusions the globules commonly form an emulsion (globules suspended in the glass
Fig. 9. Examples of immiscibility between silicate melt (glass) and non-silicate melts (mostly appear as multiphase globules). (A) Melt inclusions in the core of leucite crystal from Ventotene Island, Italy with relatively constant proportions of silicate glass, carbonate-sulphate-chloride aggregate and sulphide (opaque); (B–C) melt inclusions in quartz from granitic clast in rhyolite, Gawler Range Volcanics, South Australia show variable amounts of co-trapped silicate and non-silicate melts; (D–I) heterogeneously trapped immiscible melts, vapour and crystal phases in individual inclusions in clinopyroxene (D: Ventotene; E – Vesuvius endoskarn), quartz (F: Gawler Range) and nepheline (G–I: Vesuvius endoskarn). Scale bars are 25 lm.
‘matrix’; e.g. Harris et al. 2003, 2004; Kamenetsky et al. 2004b; Davidson et al. 2005; Davidson & Kamenetsky 2007). This presents the most convincing case of magmatic immiscibility, in which at least two liquids coexist during quartz growth. The size of the trapped silicate melt phase is random, depending on the size of the growth irregularities on the surface of the host crystal (Roedder 1984), whereas the size of the trapped globules is thought to represent the actual size of dispersed hydrosaline melt phase at the time of immiscibility, and before coalescence (Reyf 1984). The crystalline content of the hydrosaline globules is dominated by chlorides (from PIXE and LA-ICPMS analyses; Figs 11 and 12), however, other minerals (sulphates, carbonates, fluorides, oxides and sulphides) have also been identified. All globules are metal-rich (100s–10 000s ppm; Figs 11 and 12), but element abundances and ratios can vary significantly, even for co-trapped hydrosaline melts (Rankin et al. 1992; Kamenetsky et al. 2002b, 2004b). Although, Cu (and some other elements, e.g. Na, Li, Ag) can be introduced during experimental heating (Kamenetsky & Danyushevsky 2005; Zajacz et al. 2009), the natural occurrence of Cu-chloride hydrous melt (globules of clinoatacamite in silicate glass) unmixed with the silicate melt and Cu-Fe-sulphide melt have been reported (Renno et al. 2004). Compositional variability among co-existing hydrosaline globules suggests a disequilibrium character for exsolution and strong compositional fractionation between immiscible phases. If immiscibility is a continuous process in highly
Magmatic fluids immiscible with silicate melts 303
(C)
(A)
20°C
20°C
423°C
646°C
629°C
587°C
sl
sl
hsl 734°C
hsl
936°C
876°C
(D) 640°C
hsl
1050°C
868°C
(B)
18°C
300°C
540°C
745°C
sl
hsl 496°C
20°C
610°C
517°C
hsl hsl 861°C
917°C
18°C
300°C
540°C
745°C
sl
hsl
sl
938°C
Fig. 10. Phase transformations (melting, dissolution of vapour,etc) during heating stage experiments in essentially chloride melt co-trapped with silicate liquid in K-feldspar from Ventotene Island, Italy (A, B), quartz from Omsukchan granite, NE Russia (C) and clinopyroxene from the Vesuvius endoskarn (D). Note on Fig. 10D occurrence of two hydrosaline melt inclusions in the same growth plane with the inclusion of silicate liquid (sl), which contains a globule of hydrosaline liquid (hsl). Insets on Fig. 10D show snapshots at different temperatures of the hydrosaline melt inclusions that homogenize completely (including dissolution of opaque phases) at 745C. Opaque phase on Fig. 10A is a Cu-rich sulphide (see Fig. 11B for PIXE maps). Scale bars are 15 lm.
evolved magmatic systems, the components of a dispersed hydrosaline phase must therefore have varying composition, because of the variability of diffusion rates for different elements. In the residual granitic system, where crystallization and immiscibility drive chemical fractionation to the extreme, chemical disequilibrium may occur on very small spatial and temporal scales, and is maintained by slow element diffusion and delayed mixing in a relatively cold, viscous and strongly crystalline environment. Composite magmatic inclusions appear to represent a ‘broth’ of the silicate melt and hydrosaline metal-enriched fluids, characteristic of the magmatic to hydrothermal transition that is a prerequisite to the formation of economic deposits. Silicate melt–carbonate melt immiscibility Silicate carbonate melt immiscibility appears to be a widespread phenomenon and has been documented in many intrusive (e.g, Dawson & Hawthorne 1973; Egorov et al. 1993; Kogarko 1997) and volcanic (e.g. Kjarsgaard & Peterson 1991; Macdonald et al. 1993; Bailey et al. 2005, 2006) alkaline rocks and upper mantle peridotite and
eclogite xenoliths (e.g. Pyle & Haggerty 1994; Church & Jones 1995; Kogarko et al. 1995; Frezzotti et al. 2002; Rosatelli et al. 2007). This form of liquid immiscibility has been experimentally modelled at different physical and chemical parameters in numerous works. Such good coverage from studies of rocks and experiments seemingly implies no need for laborious melt inclusion investigations. However, most naturally occurring carbonatite immiscible liquids are reported as having alkali-poor or alkali-free calcitic and dolomitic compositions. On the other hand, the immiscible carbonate melts observed in modern Oldoinyo Lengai volcano in Tanzania (e.g. Church & Jones 1995) are characteristically rich in alkalies and volatile elements (S, F and Cl) and classified as natrocarbonatites. Moreover, similar compositions have been recorded in melt inclusions in phenocrysts from several magmatic suites worldwide (e.g. Rankin & Le Bas 1974; Le Bas & Aspden 1981; Kogarko et al. 1991; Nielsen et al. 1997; Veksler et al. 1998a; Andreeva et al. 2001; Fulignati et al. 2001; Gilg et al. 2001; Panina et al. 2003; Kamenetsky et al. 2004a; Panina 2005; Andreeva et al. 2006).
304 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
Cl
Cu
Zn
Ca
Mn
As
Br
(B)
Zn
Cl
Cu
Br
Pb
Spectacular example of immiscibility between alkali-carbonatite and peralkaline Fe-rich nephelinite melt at Oldoinyo Lengai is described by Mitchell (2009), where nepheline-hosted melt inclusions represent heterogeneously trapped silicate melt and varying number of globules of the
23Na 35Cl 39K 55Mn 57Fe 66Zn 95Mo 118Sn 139La 182W 208Pb
Signal intensity
106
105
102
crystallized natrocarbonate liquid (Fig. 13). During heating experiments the globules show complete melting at low temperature (570C, Fig. 13C), which coincides with the temperature of eruption of natrocarbonatite lavas (Krafft & Keller 1989; Dawson et al. 1990).
Na Pb
K
10 µm
Mn
104
103
Fig. 11. Optical images and proton-induced X-ray emission (PIXE) element maps of heated and quenched melt inclusions (silicate glass with a multiphase globule of hydrosaline liquid) in quartz from Omsukchan granite, NE Russia (A) and K-feldspar from Ventotene Island, Italy (B). The colour scale in each element image is normalized to its own maximum. The outlines on the element maps mark the contact between silicate glass and hydrosaline globule. Note strong partitioning of Cl, Br and metals into the non-silicate phases. Scale bars are 15 lm.
Cl Sn
Zn Mo
La
Fe W
101
Time
Fig. 12. Time-resolved laser ablation signal (y axis, counts per second) recorded for selected elements (masses) in the hydrosaline fluid inclusion in quartz, from a granitic clast in rhyolite, Gawler Range Volcanics (South Australia). For analysis this multiphase inclusion (see optical image) was brought by polishing to within 10 lm of the sample’s surface. Element concentrations (in wt%) are estimated (assuming 10 wt.% Na) as follows: K – 11.3, Mn – 1.6, Fe – 8.9, Zn –0.51, Rb – 0.24, Sn – 0.06, La – 0.03, W – 0.07, Pb – 0.36.
Magmatic fluids immiscible with silicate melts 305
(A)
(B)
(C)
440°C
570°C
Fig. 13. Immiscible melts of nephelinitic and natrocarbonitite composition co-trapped in melt inclusions in nepheline phenocrysts from the Oldoinyo Lengai volcano, Tanzania. (C) shows complete melting of solids and reduction of the vapour bubble in a carbonatite globule during a heating stage experiment. Scale bars are 15 lm.
The contrasting compositions of quenched carbonate melts occurring in rocks (alkali-poor) and melt inclusions (alkali-rich) bring a question about true chemical identity of carbonatitic melts. The loss of alkalies and other elements (e.g. REE, Cl) from silicate-hosted calcite and dolomite globules in rocks during syn- and post-magmatic alteration has been proposed in several studies (e.g. Macdonald et al. 1993; Kogarko 1997), but still requires insights from melt inclusions. Importantly, in many cases carbonate-rich melts and fluids are intimately associated with chloride, phosphate, fluoride, sulphate and sulphide components (e.g. Kogarko et al. 1995; Panina & Usol’tseva 2000; Fulignati et al. 2001; Gilg et al. 2001; Andreeva et al. 2004; Kamenetsky et al. 2004a; Panina 2005) that are also immiscible with silicate liquids. The major fractionation of rare earth and some other lithophile elements between carbonatitic (and other non-silicate) liquids and conjugate silicate melts (e.g. Veksler et al. 1998b; Suk 2001; Veksler 2004) is instrumental in building up a potential source of mineralization, for which the melt inclusion studies may provide critical information. Silicate melt–sulphide melt immiscibility Silicate melt–sulphide melt immiscibility occurs in a variety of mantle and crustal magmas, and the record of this process is widely available from studies of sulphide globules in submarine glasses and phenocrysts, sulphide segregations in the groundmass of ultramafic (e.g. kimberlite & komatiite) to felsic (e.g. dacite and rhyolite) rocks, and products of partial melting in mantle peridotites and eclogites. The occurrence and compositions of sulphide globules in
glasses and olivine phenocrysts in basalts from mid-ocean rifts, backarc basins and oceanic seamounts and islands, documented by numerous studies (e.g. Desborough et al. 1968; Mathez 1976; Mathez & Yeats 1976; Czamanske & Moore 1977; MaClean 1977; Gurenko et al. 1987; Chaussidon et al. 1989; Francis 1990; Peach et al. 1990; Stone & Fleet 1991; Nikogosian & Sobolev 1994; Ackermand et al. 1998; Roy-Barman et al. 1998) have been used to put constraints on the process of sulphide immiscibility and partitioning of sulphur and metals, however, application of studies of tiny sulphide melt samples (usually <100 lm) is still problematic. In part, this relates to the fact that droplets of formerly homogeneous sulphide melt are composed of several phases at room temperature (Fig. 14B,E). Although the mineral assemblage is dominated by Fe-sulphides (intermediate and monosulfide solid solution), other phases, such as Ni-rich (pentlandite) and Cu-rich (chalcopyrite) Fe-bearing sulphides and various Fe oxides and hydroxides, are usually present and complexly intergrown with each other (Fig. 14B,E). We believe that sulphide and oxide phases (Fig. 14A,B) represent unmixing from a homogeneous precursor melt. The relative proportions of different minerals measured in a given section of a sulphide globule cannot be considered representative for entire globule. Thus, for the estimate of a bulk composition each sulphide globule should be heated, homogenized at magmatic temperature and quenched. Although the sulphide melt will crystallize (not quench to a glass), it will form a fine-grained, more homogeneous mineral aggregate than a naturally cooled inclusion (Fig. 14C,E) prior to heating. This method has proved efficient for sulphide melt globules enclosed in phenocrysts, but it cannot be applied to sulphide globules in glasses and groundmass. Another natural occurrence of immiscible sulphide melts is as small sulphide globules in homogenized silicate melt inclusions (Fig. 14G,I; e.g. Gurenko et al. 1987; Metrich et al. 1999). The sizes of globules and host melt inclusions are correlated, indicating that the sulphide globule is a daughter phase and that immiscibility occurred in a closed system. In this case immiscibility is caused by Fe loss from the silicate melt into host olivine and the consequent reduction in sulphur solubility during cooling (Danyushevsky et al. 2002). Bulk chemical compositions (including trace elements, e.g. Au, Ag, Pb, Zn, Te, Se, As, PGE) of immiscible sulphide melts can be obtained by LA-ICPMS, if whole sulphide globules are ablated. The compositions of coexisting sulphides and silicate melts (glasses and melt inclusions) may potentially yield useful information on the metal and sulphur budgets in evolving magmatic system (Francis 1990). For example, immiscible sulphide globules in an unusually Ni-rich (292 ppm) and Cu-poor (29 ppm) basaltic andesite glass from near the Bouvet triple junction in south Atlantic (Kamenetsky et al. 2001c) are characteristically enriched in
306 V. S. KAMENETSKY & M. B. KAMENETSKY
(A)
(D)
(B )
(C )
(E)
(F)
(I)
(H)
(G) S
b Sp S
Ni and depleted in Cu (Fig. 14B). However, a researcher should be aware of that sulphide immiscibility can be also a consequence of factors operating on a local scale (e.g. at the melt – crystal interface, Nikogosian & Sobolev 1994) and thus studies of sulphide globules in phenocrysts and melt inclusions may not be directly applicable to the origin of magmatic sulphide deposits.
CONCLUDING REMARKS Research on melt ⁄ fluid inclusions has undoubtedly demonstrated that liquid ⁄ fluid immiscibility at the brink of magma solidification has been starting point for many mineralizing processes. It is also possible that a process that results in the ore formation is simply an endpoint on a continuum of immiscibility. However, the composition of immiscible phases, their evolution at cooling, and when and how these phases precipitate economic minerals, still remain unsubstantiated. Since magmatic immiscibility and consequent separation of volatile-rich phases from a cooling silicate magma is a keystone of the orthomagmatic models of ore formation, more insights into late magmatic – early hydrothermal processes and phases are required. If we are to apply melt ⁄ fluid inclusion studies to unravel the details of immiscibility event(s), then the choice of samples becomes critical. Apart from technical problems (e.g. presence of phenocrysts, availability and preservation of inclusions) we
S
Fig. 14. Sulphide melt globules in clinopyroxene (A, circled; other inclusions are Fe oxides), pillow rim glasses (B, F), olivine (C–E) and homogenized melt inclusions in olivine (G–I; s – sulphide, b – vapour bubble, sp – Cr-spinel). Samples are basalts from the Azores (A), South Atlantic Ridge (B, D-F) and Kurile Islands (C), and picrites from Gorgona Island (G, H) and Emeishan, SW China (I). Coarse-grained sulphide minerals with variable Fe, Ni and Cu abundances (B, D) may coexist with Fe oxides and hydroxides in a single globule (B) and single phenocryst (A). After heating and quenching sulphide globules demonstrate more homogeneous distribution of various sulphide minerals (C, E). Scale bars are 50 lm (solid line) and 25 lm (dashed line).
face a dilemma: which rocks are most suitable for such a study? At first glance, the choice of rocks spatially and temporally related to mineralization (e.g. host intrusions) seems most attractive, because such rocks are interpreted as originating from magmas that supplied the volatiles and ore-forming elements. However, if this viewpoint is correct, such rocks (and likely their melt inclusions) must be depleted in volatiles and metals relative to the precursor (parental) magma. Moreover, in such rocks the magmatic signature is commonly overprinted by later hydrothermal events. On the other hand, rocks cogenetic with those having a demonstrated direct relationship to mineralization, but without a proven link of their own, may be a case in which the evolution from silicate magmas to mineralizing fluids has not gone to completion. By studying examples in which immiscibility does not go all the way to ore formation, it should be possible to use melt ⁄ fluid inclusion approach to record consecutive ‘snapshots’ of immiscibility process and immiscible phases.
ACKNOWLEDGEMENTS This work was inspired by late Edwin Roedder, who first recognized the significance of approaching immiscibility processes and immiscible phases by studying naturally trapped inclusions. We are grateful to Alexander Sobolev and Vladimir Naumov who shared their research expertise and passion for a beautiful world of melt and fluid
Magmatic fluids immiscible with silicate melts 307 inclusions. Discussions with I. Andreeva, R. Bodnar, C. Heinrich, J. Lowenstern, L. Panina, S. Smirnov, I. Solovova, C. Szabo, I. Veksler, P. Wallace, J. Webster and J. Wilkinson significantly refined our understanding of magmatic–hydrothermal transition. Our immiscibility studies benefited from the help of many colleagues and friends, who generously provided samples, gave considerable guidance, and helped to shape ideas and present results. Special thanks to S. Allen, A. Crawford, L. Danyushevsky, P. Davidson, B. De Vivo, S. Eggins, M. Elburg, P. Fulignati, K. Go¨mann, A. Gurenko, R. Maas, P. Marianelli, A. McNeill, T. Mernagh, N. Me´trich, R. Mitchell, I. Nikogosian, C. Ryan, R. Thomas, V. Sharygin, WD Sun, O. Vasyukova, G. Yaxley and L. Zhitova. We are indebted to S. Gilbert, S. Meffre, L. Danyushevsky, K. Go¨mann, C. Ryan, T. Mernagh, and D. Steele for assistance with analytical work and A. McNeill for improving the text of the manuscript. The comments from two anonymous reviewers and the Editor Bruce Yardley helped to improve the manuscript. V.S.K. was supported by an Australian Research Council Professorial Fellowship and Discovery Grant, and M.B.K. acknowledges receipt of an Australian Postgraduate Scholarship.
REFERENCES Ackermand D, Hekinian R, Stoffers P (1998) Magmatic sulfides and oxides in volcanic rocks from the Pitcairn hotspot (South Pacific). Mineralogy and Petrology, 61, 149–62. Alt JC, Shanks WC III, Jackson MC (1993) Cycling of sulfur in subduction zones: The geochemistry of sulfur in the Mariana Island Arc and back-arc trough. Earth and Planetary Science Letters, 119, 477–94. Anderson AT Jr, Brown GG (1993) CO2 contents and formation pressures of some Kilauean melt inclusions. American Mineralogist, 78, 794–803. Andreeva IA, Kovalenko VI, Naumov VB (2001) Crystallization conditions, magma compositions, and genesis of silicate rocks of the Mushugai-Khuduk carbonatite-bearing alkalic complex, southern Mongolia: evidence from melt inclusions. Petrology, 9, 489–515. Andreeva IA, Kovalenko VI, Naumov VB, Kononkova NN (2004) Compositions and formation conditions of silicate and salt magmas forming the garnet syenite porphyries (Sviatonossites) of the carbonatite-bearing Mushugai-Khuduk complex, southern Mongolia. Geochemistry International, 42, 497–512. Andreeva IA, Kovalenko VI, Kononkova NN (2006) Natrocarbonatitic melts of the Bol’shaya Tagna Massif, the eastern Sayan region. Doklady Earth Sciences, 408, 542–6. Audetat A, Pettke T (2003) The magmatic–hydrothermal evolution of two barren granites: A melt and fluid inclusion study of the Rito del Medio and Canada Pinabete plutons in northern New Mexico (USA). Geochimica et Cosmochimica Acta, 67, 97– 121. Bailey K, Garson M, Kearns S, Velasco AP (2005) Carbonate volcanism in Calatrava, central Spain: a report on the initial findings. Mineralogical Magazine, 69, 907–15. Bailey K, Kearnsi S, Mergoil J, Daniel JM, Paterson B (2006) Extensive dolomitic volcanism through the Limagne Basin, cen-
tral France: a new form of carbonatite activity. Mineralogical Magazine, 70, 231–6. Bodnar RJ (1995) Fluid-inclusion evidence for a magmatic source of metals in porphyry copper deposits. In: Magmas, Fluids, and Ore Deposits (ed. Thompson JFH), Mineralogical Association of Canada Short Course Series, 23, 139–52. Burnham CW (1979) Magmas and hydrothermal fluids. In: Geochemistry of Hydrothermal Ore Deposits (ed. Barnes HL), pp. 71– 136. Wiley and Sons, New York. Candela PA, Piccoli PM (1995) Model ore-metal partitioning from melts into vapor and vapor ⁄ brine mixtures. In: Magmas, Fluids, and Ore Deposits (ed. Thompson JFH), Mineralogical Association of Canada Short Course Series, 23, 101–27. Cauzid J, Philippot P, Martinez-Criado G, Menez B, Laboure S (2007) Contrasting Cu-complexing behaviour in vapour and liquid fluid inclusions from the Yankee Lode tin deposit, Mole Granite, Australia. Chemical Geology, 246, 39–54. Cervantes P, Kamenetsky V, Wallace P (2002) Melt inclusion volatile contents, pressures of crystallization for Hawaiian picrites, and the problem of shrinkage bubbles. In: Eos Trans. AGU Fall Meet. Suppl., 83(47), V22A–1217. Chaussidon M, Albare`de F, Sheppard SMF (1989) Sulphur isotope variations in the mantle from ion microprobe analyses of micro-sulphide inclusions. Earth and Planetary Science Letters, 92, 144–56. Church AA, Jones AP (1995) Silicate-carbonate immiscibility at Oldoinyo Lengai. Journal of Petrology, 36, 869–89. Cline JS, Bodnar RJ (1994) Direct evolution of brine from a crystallizing silicic melt at the Questa, New Mexico, molybdenum deposit. Economic Geology, 89, 1780–802. Czamanske GK, Moore JG (1977) Composition and phase chemistry of sulfide globules in basalt from the Mid-Atlantic ridge rift valley near 37o N lat. Geological Society of America Bulletin, 88, 587–99. Danyushevsky LV, Sobolev AV, Kononkova NN (1992) Methods of studing magma inclusions in minerals during investigations on water-bearing primitive mantle melts (Tonga trench boninites). Geochemistry International, 29, 48–62. Danyushevsky LV, McNeill AW, Sobolev AV (2002) Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: an overview of techniques, advantages and complications. Chemical Geology, 183, 5–24. Davidson P, Kamenetsky VS (2007) Primary aqueous fluids in rhyolitic magmas: Melt inclusion evidence for pre- and post-trapping exsolution. Chemical Geology, 237, 372–83. Davidson P, Kamenetsky V, Cooke DR, Frikken P, Hollings P, Ryan C, van Achterbergh E, Mernagh T, Skarmeta J, Serrano L, Vargas R (2005) Magmatic precursors of hydrothermal fluids at the RHO Blanco Cu–Mo deposit, Chile: links to silicate magmas and metal transport. Economic Geology, 100, 963–78. Dawson JB, Hawthorne JB (1973) Magmatic sedimentation and carbonatitic differentiation in kimberlite sills at Benfontein, South Africa. Journal of the Geological Society of London, 129, 61–85. Dawson JB, Pinkerton H, Norton GE, Pyle DM (1990) Physicochemical properties of alkali carbonatite lavas: Data from the 1988 eruption of Oldoinyo Lengai, Tanzania. Geology, 18, 260–3. De Vivo B, Torok K, Ayuso RA, Lima A, Lirer L (1995) Fluid inclusion evidence for magmatic silicate ⁄ saline ⁄ CO2 immiscibility and geochemistry of alkaline xenoliths from Ventotene Island, Italy. Geochimica et Cosmochimica Acta, 59, 2941–53. De Vivo B, Lima A, Kamenetsky V, Danyushevsky LV (2006) Fluid and melt inclusions in sub-volcanic environments from
308 V. S. KAMENETSKY & M. B. KAMENETSKY volcanic systems: Examples from the Neapolitan area and Pontine Islands, Italy. In: Melt Inclusions in Plutonic Rocks (ed. Webster JD) Mineralogical Association of Canada Short Course Series, 36, 211–37. Desborough GA, Anderson AT, Wright TL (1968) Mineralogy of sulfides from certain Hawaiian basalts. Economic Geology, 63, 636–44. Dunbar NW, Hervig RL (1992) Volatile and trace element composition of melt inclusions from the Lower Bandelier Tuff: Implications for magma chamber processes and eruptive style. Journal of Geophysical Research, 97, 15151–70. Dunbar NW, Kyle P (1993) Lack of volatile gradient in the Taupo plinian-ignimbrite transition: evidence from melt inclusion analysis. American Mineralogist, 78, 612–8. Dunbar NW, Hervig RL, Kyle PR (1989) Determination of preeruptive H2O, F and Cl contents of silicic magmas using melt inclusions: examples from Taupo volcanic center, New Zealand. Bulletin of Volcanology, 51, 177–84. Egorov LS, Melnik AY, Ukhanov AV (1993) First finding of a kimberlite dyke containing syngenetic calcite carbonatite schlieren in Antarctica. Transactions (Doklady) of the Russian Academy of Sciences, 328, 230–3. Francis RD (1990) Sulfide globules in mid-ocean ridge basalts (MORB), and the effect of oxygen abundance in Fe–S–O liquids on the ability of those liquids to partition metals from MORB and komatiite magmas. Chemical Geology, 85, 199–213. Frezzotti ML (1992) Magmatic immiscibility and fluid phase evolution in the Mount Genis granite (southeastern Sardinia, Italy). Geochimica et Cosmochimica Acta, 56, 21–33. Frezzotti ML (2001) Silicate-melt inclusions in magmatic rocks: applications to petrology. Lithos, 55, 273–99. Frezzotti ML, Touret JLR, Neumann ER (2002) Ephemeral carbonate melts in the upper mantle: carbonate-silicate immiscibility in microveins and inclusions within spinel peridotite xenoliths, La Gomera, Canary Islands. European Journal of Mineralogy, 14, 891–904. Fulignati P, Kamenetsky VS, Marianelli P, Sbrana A, Mernagh TP (2001) Melt inclusion record of immiscibility between silicate, hydrosaline and carbonate melts: Applications to skarn genesis at Mount Vesuvius. Geology, 29, 1043–6. Gilg HA, Lima A, Somma R, Belkin HE, De Vivo B, Ayuso RA (2001) Isotope geochemistry and fluid inclusion study of skarns from Vesuvius. Mineralogy & Petrology, 73, 145–76. Gurenko AA, Poliakov AI, Kononkova NN (1987) Immiscible sulfide segregations in minerals of early crystallization stages of basaltic rock series. Doklady Akademii Nauk SSSR, 293, 439– 43. Halter WE, Webster JD (2004) The magmatic to hydrothermal transition and its bearing on ore-forming systems. Chemical Geology, 210, 1–6. Harris AC, Kamenetsky VS, White NC, van Achterbergh E, Ryan CG (2003) Melt inclusions in veins: Linking magmas and porphyry Cu deposits. Science, 302, 2109–11. Harris AC, Kamenetsky VS, White NC, Steele DA (2004) Volatile phase separation in silicic magmas at Bajo de la Alumbrera porphyry Cu–Au deposit, NW Argentina. Resource Geology, 54, 341–56. Hedenquist JW, Simmons SF, Giggenbach WF, Eldridge CS (1993) White Island, New Zealand, volcanic–hydrothermal system represents the geochemical environment of high-sulfidation Cu and Au ore deposition. Geology, 21, 731–4. Heinrich CA, Pettke T, Halter WE, Aigner-Torres M, Aud(tat A, Gmnther D, Hattendorf B, Bleiner D, Guillong M, Horn I (2003) Quantitative multi-element analysis of minerals, fluid
and melt inclusions by laser-ablation inductively-coupled-plasma mass-spectrometry. Geochimica et Cosmochimica Acta, 67, 3473–97. de Hoog JCM, van Bergen MJ (2000) Volatile-induced transport of HFSE, REE, Th and U in arc magmas: evidence from zirconolite-bearing vesicles in potassic lavas of Lewotolo volcano (Indonesia). Contributions to Mineralogy and Petrology, 139, 485–502. Javoy M, Pineau F (1991) The volatiles record of a ‘popping’ rock from the Mid-Atlantic Ridge at 14o N: chemical and isotopic composition of gas trapped in the vesicles. Earth and Planetary Science Letters, 107, 598–611. Kamenetsky VS (2006) Melt inclusion record of magmatic immiscibility in crustal and mantle magmas. In: Melt Inclusions in Plutonic Rocks (ed. Webster JD) Mineralogical Association of Canada Short Course Series, 36, 81–98. Kamenetsky VS, Danyushevsky LV (2005) Metals in quartz-hosted melt inclusions: Natural facts and experimental artefacts. American Mineralogist, 90, 1674–8. Kamenetsky VS, Maas R (2002) Mantle-melt evolution (dynamic source) in the origin of a single MORB suite: a perspective from magnesian glasses of Macquarie Island. Journal of Petrology, 43, 1909–22. Kamenetsky VS, Wolfe RC, Eggins SM, Mernagh TP, Bastrakov E (1999) Volatile exsolution at the Dinkidi Cu–Au porphyry deposit, Philippines: a melt-inclusion record of the initial oreforming process. Geology, 27, 691–4. Kamenetsky VS, Everard JL, Crawford AJ, Varne R, Eggins SM, Lanyon R (2000) Enriched end-member of primitive MORB melts: petrology and geochemistry of glasses from Macquarie Island (SW Pacific). Journal of Petrology, 41, 411–30. Kamenetsky VS, Binns RA, Gemmell JB, Crawford AJ, Mernagh TP, Maas R, Steele D (2001a) Parental basaltic melts and fluids in eastern Manus backarc basin: implications for hydrothermal mineralisation. Earth and Planetary Science Letters, 184, 685– 702. Kamenetsky VS, Crawford AJ, Meffre S (2001b) Factors controlling chemistry of magmatic spinel: an empirical study of associated olivine, Cr-spinel and melt inclusions from primitive rocks. Journal of Petrology, 42, 655–71. Kamenetsky VS, Maas R, Sushchevskaya NM, Norman M, Cartwright I, Peyve AA (2001c) Remnants of Gondwanan continental lithosphere in oceanic upper mantle: evidence from the South Atlantic Ridge. Geology, 29, 243–6. Kamenetsky VS, Davidson P, Mernagh TP, Crawford AJ, Gemmell JB, Portnyagin MV, Shinjo R (2002a) Fluid bubbles in melt inclusions and pillow-rim glasses: high-temperature precursors to hydrothermal fluids? Chemical Geology, 183, 349–64. Kamenetsky VS, van Achterbergh E, Ryan CG, Naumov VB, Mernagh TP, Davidson P (2002b) Extreme chemical heterogeneity of granite-derived hydrothermal fluids: An example from inclusions in a single crystal of miarolitic quartz. Geology, 30, 459– 62. Kamenetsky VS, De Vivo B, Naumov VB, Kamenetsky MB, Mernagh TP, van Achterbergh E, Ryan CG, Davidson P (2003) Magmatic inclusions in the search for natural silicate-salt melt immiscibility: methodology and examples. In: Developments in Volcanology, 5. Melt inclusions in volcanic systems: methods, applications and problems (eds De Vivo B, Bodnar RJ), pp. 65–82. Elsevier, Amsterdam. Kamenetsky MB, Sobolev AV, Kamenetsky VS, Maas R, Danyushevsky LV, Thomas R, Pokhilenko NP, Sobolev NV (2004a) Kimberlite melts rich in alkali chlorides and carbonates: a potent metasomatic agent in the mantle. Geology, 32, 845–8.
Magmatic fluids immiscible with silicate melts 309 Kamenetsky VS, Naumov VB, Davidson P, van Achterbergh E, Ryan CG (2004b) Immiscibility between silicate magmas and aqueous fluids: a melt inclusion pursuit into the magmatic– hydrothermal transition in the Omsukchan Granite (NE Russia). Chemical Geology, 210, 73–90. Kamenetsky VS, Pompilio M, Metrich N, Sobolev AV, Kuzmin DV, Thomas R (2007) Arrival of extremely volatile-rich highMg magmas changes explosivity of Mount Etna. Geology, 35, 255–8. Kent AJR (2008) Melt inclusions in basaltic and related volcanic rocks. Reviews in Mineralogy, Geochemistry: Minerals, Inclusions and Volcanic Processes, 69, 273–331. Kjarsgaard B, Peterson T (1991) Nephelinite-carbonatite liquid immiscibility at Shombole volcano, East Africa: petrographic and experimental evidence. Mineralogy and Petrology, 43, 293– 314. Kogarko LN (1997) Role of CO2 on differentiation of ultramafic alkaline series: liquid immiscibility in carbonate-bearing phonolitic dykes (Polar Siberia). Mineralogical Magazine, 61, 549–56. Kogarko LN, Plant DA, Henderson CMB, Kjarsgaard BA (1991) Na-rich carbonate inclusions in perovskite and calzirtite from the Guli Intrusive Ca-carbonatite, Polar Siberia. Contributions to Mineralogy and Petrology, 109, 124–9. Kogarko LN, Henderson CMB, Pacheco H (1995) Primary Ca-rich carbonatite magma and carbonate-silicate-sulphide liquid immiscibility in the upper mantle. Contributions to Mineralogy and Petrology, 121, 267–74. Krafft M, Keller J (1989) Temperature measurements in carbonatite lava lakes and flows from Oldoinyo Lengai, Tanzania. Science, 245, 168–70. Le Bas MJ, Aspden JA (1981) The comparability of carbonatitic fluid inclusions in ijolites with natrocarbonatite lava. Bulletin of Volcanology, 44, 429–38. Le Cloarec M-F, Allard P, Ardouin B, Giggenbach WF, Sheppard DS (1992) Radioactive isotopes and trace elements in gaseous emissions from White Island, New Zealand. Earth and Planetary Science Letters, 108, 19–28. Lowenstern JB (1993) Evidence for a copper-bearing fluid in magma erupted at the Valley of Ten Thousand Smokes, Alaska. Contributions to Mineralogy and Petrology, 114, 409–21. Lowenstern JB (1994) Chlorine, fluid immiscibility, and degassing in peralkaline magmas from Pantelleria, Italy. American Mineralogist, 79, 353–69. Lowenstern JB (1995) Applications of silicate-melt inclusions to the study of magmatic volatiles. In: Magmas, Fluids, and Ore Deposits (ed. Thompson JFH). Mineralogical Association of Canada Short Course Series, 23, 71–99. Lowenstern JB (2001) Carbon dioxide in magmas and implication for hydrothermal systems. Mineralium Deposita, 36, 490–502. Lowenstern JB (2003) Melt inclusions come of age: Volatiles, Volcanoes, and Sorby’s Legacy. In: Developments in Volcanology, 5. Melt Inclusions in Volcanic Systems: Methods, Applications and Problems (eds De Vivo B, Bodnar RJ), pp. 1–21. Elsevier, Amsterdam. Lowenstern JB, Mahood GA (1991) New data on magmatic H2O contents of pantellerites, with implications for petrogenesis and eruptive dynamics at Pantelleria. Bulletin of Volcanology, 54, 78– 83. Lowenstern JB, Mahood GA, Rivers ML, Sutton SR (1991) Evidence for extreme partitioning of copper into a magmatic vapor phase. Science, 252, 1405–9. Macdonald R, Kjarsgaard BA, Skilling IP, Davies GR, Hamilton DI, Black S (1993) Liquid immiscibility between trachyte and
carbonate in ash-flow tuffs from Kenya. Contributions to Mineralogy and Petrology, 114, 276–87. MaClean WH (1977) Sulfides in Leg 37 drill core from MidAtlantic Ridge. Canadian Journal of Earth Sciences, 14, 674– 83. Mathez EA (1976) Sulfur solubility and magmatic sulfides in submarine basalt glass. Journal of Geophysical Research, 81, 4269– 76. Mathez EA, Yeats RS (1976) Magmatic sulfides in basalt glass from DSDP Hole 319A and Site 320, Nazca Plate. In: Initial Reports of the Deep-Sea Drilling Project (eds Yeats RS, Hart SR) U.S. Government Printing Office, Washington, D.C., 34, 363– 73. Metrich N, Wallace PJ (2008) Volatile abundances in basaltic magmas and their degassing paths tracked by melt inclusions. Reviews in Mineralogy & Geochemistry: Minerals, Inclusions and Volcanic Processes, 69, 363–402. Metrich N, Schiano P, Clocchiatti R, Maury RC (1999) Transfer of sulfur in subduction settings: an example from Batan Island (Luzon volcanic arc, Philippines). Earth and Planetary Science Letters, 167, 1–14. Mitchell RH (2009) Peralkaline nephelinite–natrocarbonatite immiscibility and carbonatite assimilation at Oldoinyo Lengai, Tanzania. Contributions to Mineralogy and Petrology, 158, 589– 98. Moore JG (1979) Vesicularity and CO2 in mid-ocean ridge basalt. Nature, 282, 250–3. Moore JG, Calc L (1971) Sulfide spherules in vesicles of dredged pillow basalt. American Mineralogist, 56, 476–88. Moore JG, Batchelder JN, Cunningham CG (1977) CO2-filled vesicles in mid-ocean basalt. Journal of Volcanology and Geothermal Research, 2, 309–27. Naumov VB (1979) Determination of concentration and pressure of volatiles in magmas from inclusions in minerals. Geochemistry International, 16, 33–40. Naumov VB, Solovova IP, Kovalenker VA, Rusinov VL (1993) Immiscibility in acidic magmas: Evidence from melt inclusions in quartz phenocrysts of ignimbrites. European Journal of Mineralogy, 5, 937–41. Naumov VB, Kovalenker VA, Rusinov VL, Kononkova NN (1994) Inclusions of high-density magmatic water in phenocrysts from acid volcanics of the Western Carpathians and Central Tien Shan. Petrology, 2, 480–94. Naumov VB, Tolstykh ML, Kovalenker VA, Kononkova NN (1996) Fluid overpressure in andesite melts from Central Slovakia: Evidence from inclusions in minerals. Petrology, 4, 265–76. Newman S, Lowenstern JB (2002) VolatileCalc: a silicate melt– H2O–CO2 solution model written in Visual Basic for excel. Computers & Geosciences, 28, 597–604. Nielsen TFD, Solovova IP, Veksler IV (1997) Parental melts of melilitolite and origin of alkaline carbonatite: Evidence from crystallised melt inclusions, Gardiner complex. Contributions to Mineralogy and Petrology, 126, 331–44. Nikogosian IK, Sobolev AV (1994) Immiscibility processes in magmas of Hawaii (Pacific Ocean) and Reunion (Indian Ocean). Doklady Akademii Nauk, 338, 214–8. Panina LI (2005) Multiphase carbonate-salt immiscibility in carbonatite melts: data on melt inclusions from the Krestovskiy massif minerals (Polar Siberia). Contributions to Mineralogy and Petrology, 150, 19–36. Panina LI, Usol’tseva LM (2000) The role of liquid immiscibility of calcitic carbonatites from the Malyi Murun Massif (Aldan). Geologiya i Geofizika, 41, 655–70.
310 V. S. KAMENETSKY & M. B. KAMENETSKY Panina LI, Stoppa F, Usol’tseva LA (2003) Genesis of melilitite rocks of Pian di Celle volcano, umbrian kamafugite province, Italy: evidence from melt inclusions in minerals. Petrology, 11, 365–82. Peach CL, Mathez EA, Keays RR (1990) Sulfide melt-silicate melt distribution coefficients for noble metals and other chalcophile elements as deduced from MORB: implication for partial melting. Geochimica et Cosmochimica Acta, 54, 3379–89. Pyle JM, Haggerty SE (1994) Silicate-carbonate liquid immiscibility in upper-mantle eclogites - implications for natrosilicic and carbonatitic conjugate melts. Geochimica et Cosmochimica Acta, 58, 2997–3011. Rankin AH, Le Bas MJ (1974) Liquid immiscibility between silicate and carbonate melts in naturally-occurring ijolite magma. Nature, 250, 206–9. Rankin AH, Ramsey MH, Coles B, van Langevelde F, Thomas CR (1992) The composition of hypersaline, iron-rich granitic fluids based on laser-ICP and Synchrotron-XRF microprobe analysis of individual fluid inclusions in topaz, Mole granite, eastern Australia. Geochimica et Cosmochimica Acta, 56, 67–79. Renno AD, Franz L, Witzke T, Herzig PM (2004) The coexistence of melts of hydrous copper chloride, sulfide and silicate compositions in a magnesiohastingsite cumulate, Tubaf Seamount, Papua New Guinea. Canadian Mineralogist, 42, 1–16. Reyf FG (1984) Microemulsion state of fluid saturated granitic magmas: characteristics and petrological implications. Transactions (Doklady) of the USSR Academy of Sciences, 276, 1197– 201. Reyf FG (1997) Direct evolution of W-rich brines from crystallizing melt within the Mariktkan granite pluton, west Transbaikalia. Mineralium Deposita, 32, 475–90. Reyf FG, Bazheyev YD (1977) Magmatic chloride solutions and tungsten mineralizations. Geochemistry International, 14, 45– 51. Rickers K, Thomas R, Heinrich W (2006) The behavior of trace elements during the chemical evolution of the H2O-, B-, and F-rich granite–pegmatite–hydrothermal system at Ehrenfriedersdorf, Germany: a SXRF study of melt and fluid inclusions. Mineralium Deposita, 41, 229–45. Roedder E (1965) Liquid CO2 inclusions in olivine-bearing nodules and phenocrysts from basalts. American Mineralogist, 50, 1746–82. Roedder E (1979) Origin and significance of magmatic inclusions. Bulletin de Mineralogie, 102, 487–510. Roedder E (1984) Fluid inclusions. In: Reviews in Mineralogy, 12, Mineralogical Society of America. Roedder E (1992) Fluid inclusion evidence for immiscibility in magmatic differentiation. Geochimica et Cosmochimica Acta, 56, 5–20. Roedder E (2003) Significance of melt inclusions. In: Developments in Volcanology, 5. Melt Inclusions in Volcanic Systems: Methods, Applications and Problems (eds De Vivo B, Bodnar RJ) pp. xv–xvi. Elsevier, Amsterdam. Roedder E, Coombs DS (1967) Immiscibility in granitic melts, indicated by fluid inclusions in ejected granitic blocks of Ascension Island. Journal of Petrology, 8, 417–51. Rosatelli G, Wall F, Stoppa F (2007) Calcio-carbonatite melts and metasomatism in the mantle beneath Mt. Vulture (Southern Italy). Lithos, 99, 229–48. Roy-Barman M, Wasserburg GJ, Papanastassiou DA, Chaussidon M (1998) Osmium isotopic compositions and Re-Os concentrations in sulfide globules from basaltic glasses. Earth and Planetary Science Letters, 154, 331–47.
Schiano P (2003) Primitive mantle magmas recorded as silicate melt inclusions in igneous minerals. Earth-Science Reviews, 63, 121–44. Shinohara H (1994) Exsolution of immiscible vapor and liquid phases from a crystallizing silicate melt: Implications for chlorine and metal transport. Geochimica et Cosmochimica Acta, 58, 5215–21. Sobolev AV (1996) Melt inclusions in minerals as a source of principle petrological information. Petrology, 4, 228–39. Sobolev AV, Chaussidon M (1996) H2O concentrations in primary melts from supra-subduction zones and mid-ocean ridges: Implications for H2O storage and recycling in the mantle. Earth and Planetary Science Letters, 137, 45–55. Sobolev AV, Danyushevsky LV (1986) Proof of the magmatic origin of water and its estimated concentration in residual boninite melt. Transactions (Doklady) of the USSR Academy of Sciences, 288, 119–22. Sobolev AV, Nikogosian IK (1994) Petrology of long-lived mantle plume magmatism: Hawaii, Pacific and Reunion Island, Indian Ocean. Petrology, 2, 111–44. Solovova IP, Girnis AV, Naumov VB, Kovalenko VI, Guzhova AV (1991) Mechanism of degassing of silicic magmas: formation of two fluid phases during pantellerite crystallisation, Pantelleria. Transactions (Doklady) of the USSR Academy of Sciences, 320, 982–5. Solovova IP, Girnis AV, Guzhova AV, Naumov VB (1992) Magmatic salt inclusions in minerals from eastern Pamirs alkaline basalts. Geokhimiya, 1, 68–77. Spilliaert N, Allard P, Metrich N, Sobolev AV (2006) Melt inclusion record of the conditions of ascent, degassing, and extrusion of volatile-rich alkali basalt during the powerful 2002 flank eruption of Mount Etna (Italy). Journal of Geophysical Research-Solid Earth, 111, B04203. Stone WE, Fleet ME (1991) Nickel-copper sulfides from the 1959 eruption of Kilauea Volcano, Hawaii: Contrasting compositions and phase relations in eruption pumice and Kilauea Iki lava lake. American Mineralogist, 76, 1363–72. Suk NI (2001) Experimental study of liquid immiscibility in silicate-carbonate systems. Petrology, 9, 477–87. Sun W, Arculus RJ, Kamenetsky VS, Binns RA (2004) Release of gold-bearing fluids in convergent margin magmas prompted by magnetite crystallization. Nature, 431, 975–8. Symonds RB, Rose WI, Bluth GJS, Gerlach TM (1994) Volcanicgas studies: methods, results, and applications. In: Volatiles in Magmas (eds Carroll MR, Holloway JR) Mineralogical Society of America, 30, 1–66. Thomas R, Webster JD (2000) Strong tin enrichment in a pegmatite-forming melt. Mineralium Deposita, 35, 570–82. Thomas R, Webster JD, Rhede D, Seifert W, Rickers K, Forster HJ, Heinrich W, Davidson P (2006) The transition from peraluminous to peralkaline granitic melts: Evidence from melt inclusions and accessory minerals. Lithos, 91, 137–49. Veksler IV (2004) Liquid immiscibility and its role at the magmatic–hydrothermal transition: a summary of experimental studies. Chemical Geology, 210, 7–31. Veksler IV, Nielsen TFD, Sokolov SV (1998a) Mineralogy of crystallized melt inclusions from Gardiner and Kovdor ultramafic alkaline complexes: implications for carbonatite genesis. Journal of Petrology, 39, 2015–31. Veksler IV, Petibon C, Jenner GA, Dorfman AM, Dingwell DB (1998b) Trace element partitioning in immiscible silicate-carbonate liquid systems: An initial experimental study using a centrifuge autoclave. Journal of Petrology, 39, 2095–104.
Magmatic fluids immiscible with silicate melts 311 Webster JD (2004) The exsolution of magmatic hydrosaline chloride liquids. Chemical Geology, 210, 33–48. Yang KH, Scott SD (1996) Possible contribution of a metal-rich magmatic fluid to a sea-floor hydrothermal system. Nature, 383, 420–3. Yeats RS, Mathez EA (1976) Decorated vesicles in deep-sea basalt glass, eastern Pacific. Journal of Geophysical Research, 81, 4277– 84.
Zajacz Z, Halter WE, Pettke T, Guillong M (2008) Determination of fluid ⁄ melt partition coefficients by LA-ICPMS analysis of coexisting fluid and silicate melt inclusions: Controls on element partitioning. Geochimica et Cosmochimica Acta, 72, 2169–97. Zajacz Z, Hanley JJ, Heinrich CA, Halter WE, Guillong M (2009) Diffusive reequilibration of quartz-hosted silicate melt and fluid inclusions: Are all metal concentrations unmodified? Geochimica et Cosmochimica Acta, 73, 3013–27.
INDEX
Page numbers in italics refer to figures. Page numbers in bold refer to tables. ab initio molecular dynamics 44, 45 copper(I) complexation 52 copper(II) complexation 51 gold(III) complexation 54 water 46–7 zinc chloride 50 abiotic organic synthesis 161–85 accretionary wedges 115 critical taper 123–4 critical taper angle and mechanical strength 126, 127 model porosity and fluid source terms 121–2 modeling methods 119 acetaldehyde 174, 175 acetic acid 174, 175 acetogenesis 183 active venting 151 adenine 181, 182 advective heat flow/transport 139, 143, 147, 150, 200 alanine 177, 178 Albigeois South French Massif Central 272, 274, 276, 278 albite 65, 66 albitization 254, 255, 259–60, 263 alkali–halide salt solutions solubility enhancement 62–4 vs CO2–H2O fluids 61–2 Alligator River Uranium Field, Australia 271, 273, 274, 278, 281, 282 aluminum mobility 69 Alvin heat probes 153 amino acids 177–80, 182, 183, 184 ammonium ion 170, 183 amphibolite facies metasomatism 259–60 anhydrite solubility 67, 68 anthropogenic seismicity 196, 197–8 apatite solubility 25, 26, 27, 28, 29, 68 volumetric properties 31 aqueous fluids 3–15 binary and ternary systems 7–10 phase relations in one- and two-component systems 5, 6, 7 silicate melt immiscibility 299–300 thermodynamic model for mineral solubility 20–35, 39–40 water properties at high P–T 10–13 water–silicate systems at high P–T 13–15 see also brines; fluids Ashizuri transect 116, 117, 118, 121, 122 critical taper angle and mechanical strength 126, 127 geologic forcing and fluid escape 128 simulated pore pressures 124–6 asparagine 178
aspartic acid 178 Asturias, Spain 272, 274, 277, 278 Athabasca uranium deposits, Canada 273–4, 274, 275, 278 Atlantic margin, UK 73, 74, 77, 78 hot fluids at 76 oil composition 79–80 autotrophic methanogenesis 171–2 back-arc basin 163, 166–7 Balder volcanic event 76 barium, lead–zinc–fluorine–barium deposits 270–88 basement/cover unconformity 270–88 binary aqueous fluid systems 7–10 biosynthesis 161–85 black smoker systems 133, 134, 139, 165, 167 borehole logs 83, 86, 87–8 Born theory 21–2, 42 Bottom Seawater 163, 168–9 brines basement rocks interactions 284–5 in deep-crustal and upper-mantle metasomatism 58–69 migration styles 287–8 oceanic rifting 275–9 origin, Proterozoic deposits 283–4 see also salinity; sodium chloride brines brittle failure 219–20 conditions 220–1 depth dependence 223–4 envelopes 222–3, 225, 226, 228 friction coefficient and tensile strength 224–5, 226, 227 in incohesive fault zones 225–7 stress and fluid pressure changes 229–32 brittle–ductile transition zone 241–2 bromine/chlorine ratios 244, 251, 282, 283 bulk permeability 118–19, 120 calcite cementation 77–8 solubility 25, 26, 27, 28, 29, 31 volumetric properties 31 calcium salts, solubility 66, 67, 68 calcium/sodium ratios 281–2 caloric equation of state 22 capillary seal failure 102, 103, 104 capillary threshold pressure 95–6, 97, 103, 105 carbohydrates 180–1 carbon dioxide CO2–H2O fluids 60–2 H2O–NaCl–CO2 system 9 mixing calculations 169, 170
Index silicate melt immiscibility 295–9 carbon monoxide 172, 173 carbonate melts 303–5 carbonatite–fenite complexes 67 carboxylic acids 174, 175–7, 184 Car–Parrinello method 44–5, 52 cassiterite 52–3 Catalan coastal range, Spain 272, 274, 277, 278 cementation, calcite 77–8 Cenozoic offshore Gulf 85 Cenozoic onshore Gulf and shelf 85 chemical affinity abiotic organic synthesis 172–7 biosynthesis 177–81 definition 169–70 and disequilibrium 169–72 chemical weathering 254, 257–8 chlorine/bromine ratios 244, 251, 282, 283 chlorinity, and Tm ice ranges 282 clay and capillary threshold pressure 97 and fault permeability 101 smear seals 99, 102 clinopyroxene 296, 302, 303, 306 CO2 see carbon dioxide Cocos Plate, eastern Pacific 154, 155, 156 cohesive strength, faults 227–9 conductive heat flow/transport 138, 143, 147, 148, 151, 200 conductive refraction 148, 149 constant head flow-through 119 constant rate of strain tests 119 continental crust fluids in 241–52 permeability 193–202 convective cells 287–8 copper, in magma inclusions 295–6, 297, 298 copper(I) complexation 51–2, 53 copper(II) complexation 50–1 corona 254, 265–6 corundum solubility 25, 26, 27, 28, 29, 31 solubility in NaCl 65 volumetric properties 31 co-seismic permeability 198, 199 Coulomb failure stress 207 critical point, water 5, 46 critical taper angle 126 critical taper theory 115 critical wedge taper 123–4 angle 115, 118 cross-fault flow 96 crustal permeability depth relations 194–5 lower limits 200–1 crustal rheology 234–40 Cu see copper cytosine 181, 182 dating, oil charge 78–9 deep-crustal metamorphic fluids, salinity 58–69 deep downward penetration 287–8 deep-water Gulf of Mexico 85–8, 90–1 deformation bands 77, 78 density seismic energy density 207, 209–10, 212, 213
sodium chloride–water mixtures 49 solvent density 23, 24, 26, 29 water at high P–T 10–11 density functional theory 42–3 deoxyribose 180, 181 depth brine penetration 288 and fracture-controlled permeability 223–4 permeability–depth relations 194–5 and water composition 242, 243, 244 diagenesis 259–60 dielectric constant reciprocal 22 water 11, 12–13, 42, 46 dilatancy 261, 262 diopside 64 disequilibria, quantifying 169–72 dissociation equilibrium 11–12 dissolution–precipitation 254, 256, 262, 263–5 dodecanoic acid 176 drilling cruises, Gulf of Mexico 88–90 dual-conductance technique 86, 87 dynamic stress/strain 207, 211 dynamic viscosity 11, 13 earthquakes hydrologic responses 206–13 magnitude and distance from source 207, 208, 209, 212, 213 space–time progression of fronts 196–7 see also seismicity East Pacific Rise 163, 164–5 eclogitization 256, 260–1, 262 electric permittivity 12 Endeavour 163, 165 equations of state 10, 22 Ergodic hypothesis 44 ethane 173, 174 ethanogenesis 183, 184 ethanol 174 ethene 173, 174 ethyne 173, 174 exchange-correlation functional 43, 46–7 extension failure 220, 221, 223, 224, 229, 230, 231 failure envelopes 222–3, 225, 226, 228 failure mode diagrams 217–32 examples of g–s diagrams 222–9 far-field 207, 211 fault mechanics modeling methods 119–24 modeling results and discussion 124–8 permeability measurements 118–19 subduction zones 114–28 fault membranes production simulation 99–102 seal capacity 97, 98 seal failure 102–5 seals 95–9 fault zones flow through 105–8 fracture permeability in 217–32 metamorphism 196, 197, 198 reverse 218, 225, 228
313
314 Index faults cohesive strength 227–9 in hydrocarbon flow models 94–109 incohesive 225–7 permeability 100, 101, 102, 103, 107 feldspars 259, 303 flow models, hydrocarbon 94–109 fluid escape 127–8 fluid flow, at basement/cover unconformity 270–88 fluid flux 135, 136, 139, 147 precipitated minerals from 32, 33, 34, 39–40 thermal data resolution 154–6 time-integrated 32–4, 39–40 fluid inclusions 293–306 hydrocarbon 73–80 fluid pressure, and fracture permeability 217–32 fluid properties, mid-ocean ridge hydrothermal 132–9 fluids hot fluid pulses 74–6 metasomatism and metamorphism 254–67 mixing 285–7 movement through rocks 261–3 upper continental crust 241–52 see also aqueous fluids; brines; hydrothermal fluids; water fluorescence 79, 80 fluorine, lead–zinc–fluorine–barium deposits 270–88 fluorite solubility 25, 26, 27, 28, 29 volumetric properties 31 fluxibility hypothesis 139 formaldehyde 172, 173 formic acid 172, 173 Fourier’s first law for vertical heat transport 143 fracture permeability, in faults and shear zones 217–32 fractured reservoirs 77–8 friction coefficient 224–5, 226, 227 gamma ray logs 86 gas hydrates 84 geodynamics 287 geologic forcing 127–8 geothermal gradients, mineral solubilities along 30, 32 geysers 207, 210, 211 Gibbs energy of a reaction 170 Gibbs energy of dissolution 23 glasses, inclusions in 293–306 glucose 180, 181 glutamic acid 178, 179 glutamine 178, 179 glycine 177, 178 glycolic acid 174, 175 gneisses 242, 246, 248, 251 gold mines 218, 219 gold(III) complexation 54 Gotthard rail base tunnel 245–51, 247, 248 grain boundaries 260–1 granites 242, 246, 247, 248–51 granulite facies metamorphism 67–8, 260–1, 262 hydration 235, 236, 237–9 Green Canyon salinity study 85–8, 89 grossular 65, 66, 69 guanine 181, 182 Guaymas Basin 163, 166
Gulf of Mexico Cenozoic offshore 85 Cenozoic onshore and shelf 85 drilling cruises 88–90 Mesozoic Gulf rim 84–5 numerical modeling of thermohaline convection 90–1 pore water salinity 83–92 salinity variations in deep-water Gulf 85–8 sources of saline waters 90 H2O see water heat transport/flow advective 139, 143, 147, 150, 200 conductive 138, 143, 147, 148, 151, 200 mid-ocean ridges 132–9 ocean crust 142–56 heating, temporally focused 196, 197, 198 Helgeson–Kirkham–Flowers model 21, 42, 162 hexanoic acid 176 high temperature and pressure aqueous fluids at 3–15 properties of pure water 10–13 water–silicate systems at 13–15 hot fluid pulses 74–6 hydration granulite 235, 236, 237–9 ion 47–8 lattice breakdown and 22 hydraulic diffusivity 196, 197 hydraulic fracturing 261, 262 hydrocarbons flow models 94–109 fluid/oil inclusions 73–80 migration 95, 96, 98, 99 synthesis 162, 173–4 hydrogen concentrations 182 mixing calculations 169, 170 hydrogen sulfide 169, 170 hydrologic responses, earthquakes 206–13 hydrosaline melts 300–3 hydrothermal fluids abiotic organic synthesis 172–7 biosynthesis 177–81 composition 162–9, 163 diversity of 162–8 and fracture permeability 217–32 global variability 181–3 metal complexation and ion association in 41–55 mixing with seawater 168–9 quantifying disequilibria 169–72 hydrothermal systems at mid-ocean ridges 132–9 back arc 166–7 in the ocean crust 142–56 seafloor 161–85 hypocenter migration 196, 197, 198 immiscibility, magmas 293–306 in situ permeability measurements 194, 198, 199 incohesive faults 225–7 integrated fluid flux 32, 33, 34, 39–40 interatomic potentials 45, 46 copper(II) complexation 50–1
Index zinc chloride 49–50 interface-coupled dissolution–precipitation 263–4 intracontinental rifting 279–81 ion hydration and clustering 47–8 ion product 11 Juan de Fuca Ridge 150–3, 165 K see potassium chloride; potassium/sodium ratios Kairei vent field 163, 167 kinetic energy functional 42 Kohn–Sham equations 42, 43, 44–5 Krichevskii parameter 23 Lagrangian mechanics 44, 45 lattice breakdown and hydration 22 Lau basin 163, 166–7 lead–zinc–fluorine–barium deposits 270–88 leucine 177, 178 liquefaction 207, 209, 210–11, 212 lithospheric cooling models 147–8 Logging While Drilling logs 86 Louann Salt 84 Love waves 213 Maestrat basin, Spain 272, 274, 278, 279 magma immiscibility 293–306 magmatic volcanoes 207, 209, 212 Malines, Cevennes, France 272, 274, 276–7, 278 mass transfer 32–3, 34, 39–40, 255, 256 melt inclusions 293–306 membrane seals 95–9 capacity 97, 98 failure 102–5 production simulation 99–102 Mesozoic Gulf rim 84–5 metal complexation copper(I) 51–2, 53 copper(II) 50–1 gold(III) 54 in concentrated NaCl brines 47–9 molecular dynamics 44–5, 46 quantum chemistry of 42–3 tin(II) 52–4 and water at ambient/supercritical conditions 45–7 zinc chloride solutions 49–50 metal extraction 285 metamorphic fluids, salinity 4, 58–69 metamorphism 254–67 continental crust 193–202 fault-zone 196, 197, 198 granulite facies 67–8, 260–1, 262 permeability feedback loop 201 retrograde 234–40 metasomatism 254–67 methane 172 hydrates 84 mixing calculations 169, 170 methanogenesis 171–2, 183 methanol 172, 173 micro-beam methods 295 mid-ocean ridge hydrothermal systems composition of fluids 163–8 hydrocarbon synthesis 162
315
permeability and fluid properties 132–9 mineral solubility along geothermal gradients 30, 32 at constant pressure 30–1, 32 lattice breakdown and hydration 22 retrograde 30, 31 rock-forming minerals 25, 26, 27, 28–30, 64–6 thermodynamic model 20–35, 39–40 volumetric solute–solvent interactions 22–3 minerals pseudomorphic replacement 262–3 volumetric properties 31 misorientation angle 225, 226 Mohr diagrams see failure mode diagrams molecular dynamics copper(II) complexation 50–1 metal complexation 44–5, 46 NaCl brines 48–50 water 46–7 see also ab initio molecular dynamics mud volcanoes 207, 208, 210, 212, 213 Muroto transect 116, 117, 118, 121 critical taper angle and mechanical strength 126, 127 geologic forcing and fluid escape 128 simulated pore pressures 124–6 Na see sodium NaCl see sodium chloride brines; water–sodium chloride system Nankai Margin, SW Japan 114–28 near-field 207, 212 needle probes 144, 145 neutron diffraction 45, 46 9°N vent field 163, 164–5 nonanoic acid 176 North Western French Massif Central 272, 274, 275–6, 278, 281, 282 Nose–Hoover thermostat 45 Nusselt Number 200–1 ocean crust, subseafloor heat flow 142–56 oceanic rifts, brine migration 275–9 oil charge history 73–80 oil composition chemistry 79 fluorescence 79, 80 Oklo Franceville Basin, Gabon 273, 274–5, 278, 281, 282 olivine 258, 296, 306 one-component fluid systems 5 organic synthesis 161–85 chemical affinity for 172–7 energetics 183–4 Oseberg Syd area 98 outcrop-to-outcrop circulation 148, 149 outcrops, seafloor heat flow 152–6 outrigger probes 144, 145 oxalic acid 174, 175 oxidation–reduction equilibria 168, 171 oxygen–oxygen pair distributions 46, 48 pair distribution function Cu–Cl 52, 53 Cu–O pair distribution O–O pair distributions Sn–Cl pair distribution
45 51 46, 48 54
316 Index palaeofluid analysis dating 78–9 fractured reservoirs 77–8 hot fluid pulses 74–6 hot fluids at the UK Atlantic margin 76 oil composition 79–80 palaeofluids, compositions and origin 281–3 Pb see lead–zinc–fluorine–barium deposits Perdew, Burke and Ernzerhof functional 43 permeability architecture 122–3 bulk permeability 118–19, 120 changes after earthquakes 211–12 continental crust 193–202 evidence for higher 195–8 fault permeability 100, 101, 102, 103, 107 feedback loop 201 fracture permeability 217–32 lower limits of crustal 200–1 measurements 118–19, 194, 198, 199 mid-ocean ridge 132–9 rates of decay 198–200 reaction-enhanced 262 turbidite 122, 124–6 permeability–depth relations 194–5 petroleum production 96 pH 168–9 phase relations binary systems 7, 8 in one- and two-component systems 5, 6, 7 ternary systems 9, 10 phenocrysts, inclusions in 293–306 phenylalanine 179, 180 plate boundary, mechanical strength 126, 127 pore fluid factor–differential stress space 220–2 pore pressure 114, 115 along-strike differences 118 critical taper angle and mechanical strength 126, 127 critical wedge taper 123–4 and earthquakes 210, 212 geologic forcing and fluid escape 127–8 modeling methods 119–20 simulated 124–6 pore water salinity 83–92 porosity distribution 121–2 porosity generation 254, 257, 263 portlandite solubility 25, 27, 28, 29, 31 volumetric properties 31 postseismic permeability 199 potassium chloride see also alkali–halide salt solutions 8, 63 potassium/sodium ratios 281–2 precipitation–dissolution 254, 256, 262, 263–5 pressure aqueous fluids at high T–P 3–15 capillary threshold 95–6, 97, 103, 105 depletion and seal failure 102, 103, 104 fluid, and fracture permeability 217–32 and mineral solubility 30–1, 32 see also pore pressure primary inclusions 294 production simulation, fault membranes in 99–102 proline 179 propanoic acid 175, 176
Proterozoic U deposits 270–88 proton-induced X-ray emission (PIXE) maps 298, 301, 304 protraction areas 85–6 pseudomorphic mineral replacement 262–3 purines 181, 182 pyrimidines 181, 182 pyroxene 258, 261, 262 clinopyroxene 296, 302, 303, 306 quantum chemistry, metal complexation 42–3 quartz retrograde solubility 31 solubility 25, 26, 27, 28, 29, 31 solubility in alkali–halide salt solutions 61–4 solubility in CO2–H2O fluids 60–1 solubility in water 59–60 solubility models 64 volumetric properties 31 Rainbow hydrothermal field 163, 167–8 reaction-enhanced permeability 262 reaction zone, mid-ocean ridges 136–7 relay ramps 107, 108 reservoir production flow model 95 reservoirs, fractured 77–8 re-shear 225–7, 227–9, 231 retrograde metamorphism 234–40 retrograde solubility 30, 31 reverse fault zones 218, 225, 228 rheology, crustal 234–40 Rhine Graben Soultz sous Fôrets, France 272, 274, 278, 279–81, 282 rhyolite 298, 300, 301, 302 ribose 180, 181 ribulose 180, 181 rodingitization 258 Roozeboom diagram 245 rutile solubility 25, 26, 27, 28, 29, 31 volumetric properties 31 S-waves 213 saline fluids see brines salinity aqueous fluids 4 metamorphic fluids 4, 58–69 pore water 83–92 upper crustal waters 244 and venting temperature 133, 135–6, 139 see also brines; sodium chloride brines salting-in effect 62, 63 salting-out effect 9, 62 saturation index 249 Schwarzwald district, Germany 272, 274, 278 seafloor heat flow 142–56 global considerations 147–8 instrumentation 144, 145 regional/local considerations 148, 149, 150–2 through outcrops 152–6 seafloor hydrothermal systems 132–9, 142–56, 161–85 seawater and methanogenesis 171 bottom 163, 168–9 chemical composition 163
Index mixing with submarine hydrothermal fluids 168–9 sediment temperature probes 144, 145 sedimentary basins oil inclusions 73–4 pore water salinity 83–92 thermal anomalies 75 seismic energy density 207, 209–10, 212, 213 seismicity anthropogenic 196, 197–8 continental crust 193–202 hypocenter migration 196, 197, 198 triggered 207, 210, 212, 213 self-dissociation, water 11–12, 24 serine 179 serpentinization 258 Setchénow coefficient 62 Shale Gouge Ratio 95, 96, 97, 98–9, 100, 101 shear failure 220–32 shear zones fracture permeability in 217–32 retrograde metamorphism 234–40 Sherwood number 201 silica in high-grade fluids 59–64 see also quartz silicate melts 293–306 aqueous fluid immiscibility 299–300 carbon dioxide immiscibility 295–9 carbonate melt immiscibility 303–5 hydrosaline immiscibility 300–3 sulfide melt immiscibility 305–6 water–silicate systems 13–15 silicate–water systems 13–15 Silvermines District, Ireland 272, 274, 278, 279, 280 simple point charge (SPC/E) model 46, 47, 49 slip events 219 Smith & Dang potentials 47, 49 sodium Na/Ca ratios 281–2 Na/K ratios 281–2 sodium chloride brines concentrated 47–9 quartz solubility in 61, 63 solubility enhancement of Ca salts 66 solubility enhancement of rock-forming minerals by 64–6 zinc complexation in 50 zinc speciation in 49 see also alkali halide salt solutions; water–sodium chloride system sodium chloride dissociation/solubility 24 solubility see mineral solubility solute transport 83–90 solute–solvent interactions 22–3 solvent density 23, 24, 26, 29 space–time progression, earthquake fronts 196–7 spatial variations, pore water salinity 83–92 spontaneous potential logs 86, 87 spring discharge 207, 208, 211 spring temperature 207, 209, 210, 211 stair-stepping faults 102 static permittivity see dielectric constant static stress/strain 207, 211 stratigraphic columns 117 streamflow 207, 208, 210 stress states
dynamic/static 207, 211 and fracture permeability 217–32 subduction zones fault mechanics 114–28 modeling methods 119–24 modeling results and discussion 124–8 permeability measurements 118–19 subseafloor heat flow 142–56 sulfate 242, 250 sulfide melts 305–6 sulfides 165, 167 sulfur-rich magmatism 67 SUPCRT database 42, 49, 51 supercritical fluids 13 SUTRA 119 TAG see Trans-Atlantic Geotraverse taper angle 115, 118, 126 TDS see total dissolved solids temperature aqueous fluids at high T–P 3–15 spring 207, 209, 210, 211 Tm ice ranges 282 Tm–Th pairs 281, 283 venting temperature/salinity 133, 135–6, 139 see also heat transport/flow temporally focused heating 196, 197, 198 tensile strength 224–5 ternary aqueous fluid systems 7–10 thermal anomalies 74–5 thermal conductivity measurements 145, 146, 149 thermal convection 288 thermal events, transient 75–6 thermistor sensors 146 thermodynamic model 21–3 calibration 25, 26, 27, 28 conclusions 34 discussion 28–30 equivalences 39 geological applications 30–4 hydrothermal conditions 42 for mineral solubility 23–5 parameters 25 thermohaline convection 90–1 thermometric experiments 295 threshold seismic energy density 210, 212, 213 thrust belts 115 thymine 181, 182 time-dependency, fault cohesive strength 227–9 time-integrated fluid flux see fluid flux time-resolved laser ablation 297, 301, 304 tin(II) complexation 52–4 Tm ice ranges 282 Tm–Th pairs 281, 283 total dissolved solids 242, 243, 244, 246, 247, 248, 251 Trans-Atlantic Geotraverse 163, 165, 169, 170, 171 transmissibility multipliers 100, 101, 102 transport theory 32, 39–40 triggered seismicity 207, 210, 212, 213 tryptophan 179, 180 turbidite permeabilities 122, 124–6 two-component fluid systems 5, 6, 7 Type 1 systems 5, 6, 7 Type 2 systems 6, 7, 13
317
318 Index U deposits, Proterozoic 270–88 unconformity, basement/cover 270–88 undrained consolidation 210–11 upper continental crust, fluids in 241–52 upper-mantle metamorphic fluids, salinity 58–69 Upper Silesia, Poland 272, 274, 277, 278, 279 valine 177, 178, 179 vein arrays 218, 219, 220 vent fields 164–8 venting temperature/salinity 133, 135–6, 139 Verlet algorithm 44 viscosity, water 11, 13 viscous creep 230, 231 vitrinate reflectance 75, 76 volumetric properties, minerals 31 volumetric solute–solvent interactions 22–3 water ab initio molecular dynamics 46–7 at ambient and supercritical conditions 45–7 critical point 5, 46 dielectric constant 11, 12–13, 42, 46 Gotthard tunnel 245–51, 247, 248 mixing calculations 169, 170 quartz solubility in 59–60 simulations using classical potentials 46
upper continental crust 242, 243, 244–5 see also aqueous fluids; fluids; hydration; seawater water–albite system 13–14, 15 water at high P–T density 10–11 dielectric constant 12–13 self-dissociation 11–12 viscosity 11, 13 water-drive leakage 104, 105 water level changes, from earthquakes 206–7, 208, 209, 210, 211–12, 213 water–potassium chloride system 8 water–rock interaction process 242, 249, 250, 251 water–salt systems, phase relations 5, 6 water–silicate systems 13–15 water–sodium chloride system 7, 8 density and phase diagram 49 metal complexation 47 mineral solubility in 64, 65 water–sodium chloride–carbon dioxide system 9 well-mixed aquifer model 148 white smokers 167 wollastonite 64–5 zinc, lead–zinc–fluorine–barium deposits 270–88 zinc chloride solutions 49–50
