Geochemical Rate Models

Geochemical Rate Models
An Introduction to Geochemical Kinetics
Rate models (kinetic models) are used to predict rates of geochemical

processes in the near-surface environment. Based on the author’s 30 years
of teaching and research experience, this combination of reference and
textbook provides a systematic, comprehensive description of rate models,
developed from fundamental kinetic theory and presented using consistent
terminology and notation.
Major topics include rate equations, reactor theory, transition state
theory, surface reactivity, advective and diffusive transport, aggregation
kinetics, nucleation kinetics, and solid–solid transformation rates. The the-
oretical basis and mathematical derivation of each model is presented in
detail and illustrated with worked examples from real-world applications
to geochemical problems. The book is also supported by online resources:
self-study problems put students’ new learning into practice; and spread-
sheets provide the full data used in figures and examples, enabling students
to manipulate the data for themselves.
This is an ideal overview for advanced undergraduate and graduate
students, providing a solid understanding of geochemical kinetics. It will
also provide researchers and professional geochemists with a valuable
reference for solving technical and scientific problems.
J. Donald Rimstidt is a Professor Emeritus of Geochemistry at Virginia

Polytechnic Institute and State University (Virginia Tech). During his 30-year
career there he has taught several graduate courses, including Aqueous
Geochemistry, Environmental Geochemistry, Geochemical Kinetics, and
Geochemical Thermodynamics. The effective pedagogic methods from those
courses, for building quantitative skills and instructing modeling tech-
niques, underlie this book. Professor Rimstidt is the author or co-author
of more than 100 papers, including his seminal article, “The kinetics of
silica-water reactions” (with Hu Barnes, 1980), which forms the basis of the
quasi-thermodynamic presentation of transition-state theory presented in
this book.
“Don Rimstidt brings 30 years of teaching experience at a world class

geosciences department to the writing of this book. With its beautifully
clear exposition of the theoretical and practical aspects of geochemical
kinetics, and use of real-world examples, this book is essential reading for
students and professionals with interests in geochemical rate models.”
David J. Vaughan, Research Professor, University of Manchester, and

President of the Mineralogical Society of America
Geochemical Rate Models
An Introduction to Geochemical Kinetics
J. Donald Rimstidt
Virginia Polytechnic Institute and State University

University Printing House, Cambridge CB2 8BS, United Kingdom
Published in the United States of America by Cambridge University Press, New York
Cambridge University Press is part of the University of Cambridge.
It furthers the University’s mission by disseminating knowledge in the pursuit of

education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107029972
© J. Donald Rimstidt 2014
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2014
Printed and bound in the United Kingdom by TJ International Ltd., Padstow, Cornwall
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data

Rimstidt, J. Donald, 1947– author.
Geochemical rate models : an introduction to geochemical kinetics /
J. Donald Rimstidt, Virginia Polytechnic Institute and State University.
pages cm
Includes bibliographical references and index.
ISBN 978-1-107-02997-2 (hardback)
1. Geochemical modeling. 2. Chemical kinetics. 3. Geochemistry–Mathematical
models. I. Title.
QE515.5.G43R56 2014
551.9–dc23 2013022101
ISBN 978-1-107-02997-2 Hardback
Additional resources for this publication at www.cambridge.org/rimstidt
Cambridge University Press has no responsibility for the persistence or accuracy of

URLs for external or third-party internet websites referred to in this publication, and
does not guarantee that any content on such websites is, or will remain, accurate or
appropriate.
For my students
Contents
Preface page ix
1 Geochemical models 1
2 Modeling tools 9
3 Rate equations 36
4 Chemical reactors 56
5 Molecular kinetics 79
6 Surface kinetics 102
7 Diffusion and advection 128
8 Quasi-kinetics 156
9 Accretion and transformation kinetics 182
10 Pattern formation 205
References 210
Index 228
vii
Preface
This book is intended for use as a textbook for an upper-level undergradu-

ate or a first-year graduate course and as a handy reference for profession-
als who need to refresh their memories about a particular rate model. It
focuses on rate models that are well understood, widely used, and pertinent
to geochemical processes. Its scope is limited to an amount of material that
can be covered in a one-semester course and the depth of presentation is
restricted to the principal aspects of each model. The reference list directs
the reader to sources that contain the additional details needed to produce
more sophisticated models.
As the scientific and technological enterprise grows larger, scientists cope
with the onslaught of new knowledge by retreating to more narrowly defined
sub-disciplines. Increasing specialization tends to make each sub-discipline
less relevant to the overall scientific and technological enterprise. This tempts
scientists to pursue questions of trivial importance relative to the larger pic-
ture. Furthermore, the terminology that grows up in these enclaves becomes
abstruse and incomprehensible. Geochemistry is challenged to avoid this
trap by aggressively integrating knowledge from cognate disciplines in order
to remain vibrant and meaningful. This is my justification for choosing
many of the topics in this book from scientific and technological areas that
might not be considered as traditional geochemistry. Hopefully this intel-
lectual cross-fertilization will improve geochemistry’s vigor. My choice of
topics was predicated on the question: What should everyone know about
rates of geochemical processes? I hope that this book answers that ques-
tion. If I have done my job properly, this book will provide an intellectual
foundation for anyone, including scientists and technologists from cognate
disciplines, who deal with geochemical processes. I have tried to meld these
topics by using a standard notation and terminology that is consistent with
that already used in geochemistry.
I believe that understandable and reliable models are the basis of scien-
tific knowledge, so this book emphasizes the development of conceptual
and quantitative models. In addition, effective models make predictions that
have practical value so these models are a valuable contribution to society.
The essence of the scientific method is to compare a model’s predictions
ix
x Preface
with actual outcomes to test the validity of the underlying knowledge. Every
scientist should understand and appreciate the importance of this model-
building and -testing scenario.
This book does not delve into computer models of rate processes because
there are already many good sources of information about this approach.
Instead it focuses on fundamental and relatively simple models that are the
foundations of the models incorporated into the various computer codes.
I have enjoyed the process of writing this book and learned a great deal
from teaching this material to students. I hope that the readers will find it
equally interesting and enlightening.
Chapter 1
Geochemical models
Modern geochemistry studies the distribution and amounts of the chemical elem-
ents in minerals, ores, rocks, soils, waters, and the atmosphere, and the circulation
of the elements in nature, on the basis of the properties of their atoms and ions.
(Goldschmidt, 1958)
The distribution and circulation of the chemical elements in and on the Earth
is influenced by a myriad of chemical and physical factors, many of which
have changed over geological time. Understanding the role of these factors
in geological processes requires us to condense information about elemental
abundances and distributions into models. This book is about geochemical
models for situations where time plays a key role. Geoscientists have always
appreciated the importance of time in fashioning the Earth. Many geological
processes require time spans that are far too long for human observation,
but we can use models to extrapolate rates based on short-term observations
to predict geochemistry in deep time. Equally important are models that
forecast the future behavior of geochemical systems because those models
are needed for environmental management and resources recovery projects.
Some of the models described in this book were developed by geochemists
but many others come from applied sciences and engineering. Because of
this diverse provenance, the models in their original form used a confusing
mix of units, terminology, and notation. This book attempts to remedy that
problem by recasting the models using internally consistent notation, units,
and terminology familiar to geochemists. Furthermore, whenever possible
the models are developed from fundamental theory showing a sufficient
number of intermediate steps to allow the reader to follow the derivations.
Thermodynamic models have been a mainstay of geochemistry since the
early twentieth century. They are especially effective for deep earth condi-
tions where local equilibrium conditions prevail. However, at and near the
Earth’s surface extensive amounts of mass transport and low temperatures
keep many reactions from reaching equilibrium. Kinetics models are needed
to properly describe these situations. This makes the models of kinetic and
dynamic processes described in this book complementary to thermodynamic
1
2 Geochemical models
models. The thermodynamic models describe the equilibrium endpoint for

geochemical processes, and the kinetic and dynamic models predict how
long it might take to reach that endpoint.
Kinetics and dynamics

Element redistribution in nature is the result of chemical reactions that are
described by geochemical kinetics models and physical transport that is
described by geochemical dynamics models. These geochemical kinetics and
geochemical dynamics models are adapted from a diverse range of scientific
and technological sources. Geochemical kinetics is grounded in well-estab-
lished chemical kinetics theory (Laidler, 1987a, 1987b; Lasaga, 1998b) but
that theory must be combined with practical methods to deal with the com-
plexities of geological problems and technological applications. Valuable
contributions to the field of geochemical kinetics and dynamics come from
chemical engineering (Hill Jr., 1977; Levenspiel, 1972a), mineral processing
(Burkin, 2001), civil and environmental engineering (Brezonik, 1994; Weber
Jr. and DiGiano, 1996), soil science (Sparks, 1989), and even food process-
ing (van Boekel, 2009). There are also several very useful books about mass
transfer processes (Denny, 1993; Probstein, 1989; Vogel, 1994) that contain
models that are readily adapted to deal with geochemical dynamics. Finally,
there are many books that treat various aspects of geochemical kinetics
(e.g. Berner, 1980; Lasaga, 1998b; Lerman, 1979; Stumm, 1990; Zhu and
Anderson, 2002). Each of these sources has a particular scientific style that
is reflected in different ways of making observations and constructing mod-
els. The challenge is to select relevant knowledge from these sources and to
recast it in a way that makes it useful to geoscientists.
Model construction
Models organize our knowledge about the world. Doing science is the pro-
cess of making observations, using those observations to develop a model,
and then verifying the model’s effectiveness by comparing its predictions
with additional observations. The methods and scope of model-making
vary from discipline to discipline, but the goal of creating reliable predictive
tools is always the same. Developing a useful model requires a combination
of creative thinking and disciplined use of modeling tools. There are many
good discussions of modeling strategies (Aris, 1994; Bender, 1978; Bunge,
1997; Hall and Day, 1977; Hall et al., 1977; Harte, 1988; Overton, 1977;
Rescigno and Thakur, 1988; Shoemaker, 1977) but developing an effective
model is in many ways a creative process. Good models are elegant and
powerful. Elegance means that the model is expressed in the simplest, easi-
est to understand terms. Power means that the model explains the widest
Model construction 3
possible range of behaviors of the system of interest. Finally, good models

are beautiful. Perhaps some of this beauty comes from power and elegance,
but some of the beauty arises when the model is congruent with our thought
processes. The esthetics of scientific creation is a difficult concept to learn
(Wechsler, 1988), but it is a necessary part of making a lasting contribution
to science.
Models can be categorized in various ways. Predictive models forecast
the future behavior of a system, whereas conceptual models are used to
understand relationships between system parts and processes. Deterministic
models are constructed from mathematical functions that unambiguously
relate cause and effect so that a particular set of input parameters produces
a clearly related set of predicted results. Probabilistic models use statistical
data to estimate the chance that an event or condition will occur. Forward
models predict the future behavior of a system, whereas inverse (or reverse)
models are used to extract fundamental data or mathematical relationships
from past observations.
Models can relate information in three formats. Visualization illustrates
spatial, mathematical, or sequential relationships using diagrams, graphs,
and images. Narration uses words to explain how processes occur and
are related to each other. Mathematics uses computational operations to
quantitatively relate processes and features to each other. Disciplined use
of notation, reliable data sources, and homogeneous units (Bender, 1978)
along with appropriate mathematical methods and effective error analysis
are features of effective mathematical models. However, it would be impos-
sible to understand most mathematical models without accompanying
visualization and narration. Visualization plays an equally important role
in the development of scientific ideas (Wainer, 2005). For example, flow
charts showing subsystems as boxes connected by arrows are often used
to show how mass and/or energy moves through a system (McClamroch,
1980). This method of model construction, which is reprised in the ideal
chemical reactor models described in Chapter 4, is especially useful because
it leads to quantitative relationships in a simple and natural way. Graphs
that allow us to visualize mathematical relationships are critical to under-
standing how dependent variables are influenced by independent variables.
Harris (1999) provides extensive and detailed descriptions of many kinds of
graphical formats. Effective graphical illustrations are easy to understand
without needing to consult descriptive text. Effective graphs and illustra-
tions maximize the information to ink ratio (Tufte, 2001). All models are
built upon a conceptual foundation. Although visualizations often nicely
summarize the conceptual basis of a model, it is usually impossible to fully
explain the model without the use of some text. Well-organized and clearly
stated narrative descriptions of the conceptual basis of a model are a key
part of model development.
Model reliability
No model is perfect (Oreskes et al., 1994). Even the best models fail under
some conditions. In addition, making mistakes is a natural part of the mod-
eling process. Systematic methods should be used to find, analyze, and cor-
rect mistakes and to define the valid range of a model. Determining the cause
of a model’s failure and repairing the model is a key task in model building.
Models are used to understand how complex behaviors arise from the
interaction of simple processes. Ideally models are built upon reliable prin-
ciples such as the conservation of matter, energy, and charge or the prin-
ciple of detailed balancing. Often less-reliable relationships such as empir-
ical rate equations must also be used. All quantitative models require input
data and relationships that come from measurements, which always contain
some error. This error can propagate through a model in unexpected ways,
especially if the model simulates nonlinear interactions. This challenges the
modeler who must decide whether an unexpected result is a legitimate pre-
diction or simply an artifact arising from an unfortunate combination of
errors. Three classes of errors occur in models. Formal errors are incorrect
assumptions and/or formulations. They include errors in the conceptual
foundation of the model as well as errors in the input data. Structural errors
are errors in mathematical manipulations such as programming errors or
algebraic errors. They include software bugs. Computational errors are
errors in numbers caused by incorrect rounding or by addition or multi-
plication errors. Because there are so many ways that a model might fail,
models and their predictions must be verified and validated to delineate the
bounds of their reliability.
Verification tests whether the model is internally consistent, incorpo-
rates the correct relationships in the correct ways, and uses correct data.
It is a good idea to develop a set of standards and practices that can be
used in geochemistry model-making to insure the validity of the resulting
model. Table 1.1 is an example of a checklist that might be used. Models
should use a consistent system of notation and units to avoid structural
errors. Equations should be tested using dimensional analysis. The input
data should be reviewed to insure its correctness and internal consistency.
Verification should determine the expected precision of the model’s predic-
tions based on the propagation of errors through the model. Error ana-
lysis is probably the most undervalued part of model development. For
simple models, error propagation can be done using simple algebraic meth-
ods described in Chapter 2. For complex models, simple error propagation
is often not practical. In such cases sensitivity analysis is a good strategy.
Sensitivity analysis uses various schemes to systematically vary, within
the expected range of error, the values of important modeling parameters
to see how those variations affect the model’s predictions. The utility of
Model reliability 5
Table 1.1. Checklist of recommended validation and verification procedures
Verification Questions
Are the equations used dimensionally homogeneous?

Are the standard and reference states of the variables internally consistent?
Is the notation internally consistent and clearly stated in the documentation for
the model?
Are the computational methods correct? Have numerical simulations been
tested against analytical solutions?
Are the chemical reactions correctly balanced? Are all the important chemical
reactions and chemical species accounted for?
Does error analysis demonstrate that the predicted values are accurate enough
to be useful?
Has the range and domain of the model been documented and is the situation
being modeled within the range and domain?
Are the model’s outputs clearly related to the inputs? Does sensitivity analysis
show that each input has a significant effect on the model output?
Are the spatial and conceptual relationships consistent and properly
documented?
Validation Questions
Are the predictions geologically reasonable? Are they consistent with reasonable
estimates?
Are the predictions consistent with other scientific observations and knowledge?
Are the quantitative predictions sufficiently close to the behavior of one or more
natural analogs?
a particular model depends upon whether it predicts the behavior of the

system of interest within a practical range of uncertainty. For example, a
model of a process used to control the lead content of drinking water would
be useless if uncertainty in the predicted lead content was larger than the
regulatory level. Typically models are constructed using very conservative
assumptions that are frequently chosen on an arbitrary basis. Error analysis,
on the other hand, clearly demonstrates quantitatively which of the input
data most strongly affects the uncertainty of the predicted behavior.
Validation determines whether the model correctly predicts observable
outcomes with reasonable accuracy and precision. Showing that a model can
predict an analytical solution might be a useful validation test. Validation
should compare a model’s predictions with geological analogs. Models are
hypotheses; they can never be proven true (Nordstrom, 2012). However, a
model with a record of consistently successful predictions is more useful
than an untested one. Finally, collecting new observations to test a model’s

predictions allows the model to be further refined. It is important to restate
that the validating process does not prove that a model makes correct predic-
tions; instead validation only decreases the probability of a model’s failure.
Some technology models, usually related to resource recovery or waste
disposal, involve predicting the behavior of the Earth after human interven-
tion. Because there are frequently no precedents for such cases, models of
the effects of human activities are sometimes validated using natural ana-
logs, geological situations that are similar to the technological situation. In
tests of technological models a clear definition of accuracy is needed. This
definition is complicated because geochemical parameters almost always
show a range of values, so comparison of a single model run with a single
field observation has very little meaning. Instead, the input parameters for
the model must be varied within the expected range to create a range of
predictions that can be compared with the observed range of values. Then a
statistical test can be used to estimate how often the model will fail.
Example 1.1. Estimation method for validation of a chemical

weathering denudation rate model
Denudation is the lowering of the elevation of the Earth’s surface by chemical

weathering, erosion, and mass wasting. A model of chemical weathering would
account for the rate of removal of material from soil and underlying rock by
various dissolution processes. A first step toward validating such a model is to
develop some independent estimates of the chemical denudation rate.
One estimate might be based on geological knowledge. We know that denuda-
tion rates must be smaller than uplift rates. Otherwise the continents would
be reduced to the elevation of sea level. The maximum uplift rate is probably
about 10% of the maximum ocean ridge spreading rate, which is ~10 cm/yr. That
makes uplift rates smaller than 1 cm/yr. The uplift rate of the average crust is
likely to be 100 to 1000 times smaller than the maximum rate. We might choose
the geometric mean of these values and estimate that the overall crust uplift rate
is 0.03 mm/yr or 3 mm/century. A terrain with such a low uplift rate is likely
to be relatively flat so mass wasting and erosion might account for two-thirds
of the denudation, which means that chemical weathering would account for
about one-third. This gives an estimate of denudation by chemical weathering
of ~1 mm/century.
Another way to estimate the chemical denudation rate is based on the realiza-
tion that most of the soluble products of chemical weathering are carried away
by infiltrating groundwater. If we consider 1 m2 of land surface in a setting with
about 100 cm of rainfall per year with 10% infiltration of the rainwater, ~100 L
of water passes through each square meter annually. As that water infiltrates
Interpretation of results 7
into the water table, the concentration of dissolved species increases to perhaps
500 mg/L. The total mass of rock and soil dissolved away each year is 5 × 104 mg
or ~50 g. If the density of the rock and soil minerals is ~3 g/cm3, then 17 cm3
(= 1.7 × 10–5 m3) of material is carried away annually. If that volume is removed
from under the 1 m2 of land surface, the surface elevation will be lowered by
1.7 × 10–5 m (~0.02 mm). So the land surface elevation is reduced by
0.02 mm/yr or ~2 mm/century.
These estimates show that chemical weathering denudation rates should be
on the order of a few millimeters per century. They might be as high as a few
centimeters per century in humid tropical settings or as low as a few hundred
microns per century in arid cold environments. Because these estimates con-
strain model predictions to fall within this range of values, they can be used as a
validation test of a chemical weathering denudation model.
Harte (1988) and Weinstein and Adam (2008) provide detailed explanations
about how to make meaningful estimations and they give a large number of
practice examples.
Interpretation of results
The results of simple deterministic models are generally easy to interpret
and apply. For example, a model of radioactive decay will predict that the
radioactivity of a substance will be 1/8 of the original value after three
half-lives have passed. As models become more complex they become more
difficult to interpret. This suggests that model building should begin with
highly idealized and simplified cases that are easily understood and tested.
The next stage of model building involves testing the effect of potentially
important variables on the model output to determine which variables are
important enough to include and which can be neglected. This process of
building models in a stepwise fashion not only eliminates a large number of
unneeded independent variables but it helps the modeler develop a concep-
tual understanding of the situation being modeled.
The most important task of model building is to communicate expert
knowledge developed by the model builder to others who can use that
knowledge to solve problems. The modeling endpoint should be a report
that contains thorough documentation of the model and an explanation
of how to use its predictions. This report should contain a clear descrip-
tion of the model’s conceptual basis in a well-written narrative section with
accompanying illustrations and explanations of the mathematical methods.
The report should illustrate relationships between the model’s predictions
and input parameters using response maps. These graphs of predictions ver-
sus parameters are much easier to understand than equations and tables.
Many of the figures in this book are response maps. The report should also
describe the implementation of the modeling algorithm, including descrip-
tions of special computational methods. The rationale for selecting the
input data, including a discussion of their uncertainty, should be described.
The range and domain of the model should be specified. Finally, the report
should explain how to interpret the model’s output.
Chapter 2
Modeling tools
Before any model is ready for use it must be verified. The verification step
is greatly simplified if the model is constructed using conventional com-
putational methods, notation, and units. This chapter reviews some pro-
cedures and conventions that are recommended for geochemical model
construction.
Balancing chemical reactions

The first step in building a geochemical model is to write balanced equa-
tions that describe the governing chemical reactions. It is often possible to
recognize these reactions based on past experiences, but when experience is
lacking a general strategy is needed to identify these key reactions. The strat-
egy should recognize that, with few exceptions, the key reactions involve the
most abundant phases and chemical species. Creating a mineral inventory
listing the possible hosts for the elements of interest is a first step toward
selecting the solid phases to include in the model. Similarly, a chemical ana-
lysis of the aqueous phase can be used along with an aqueous speciation
model to identify important aqueous species. The reactions among these
mineral and aqueous species are expressed as balanced chemical reactions
and these reactions become the basis for the subsequent model.
Once the key solid phases and aqueous species have been recognized, the
next step is to write the balanced chemical reactions that link them. Most
geochemical reactions can be balanced by inspection, which involves the
stepwise selection and adjustment of tentative reaction coefficients. All reac-
tion balancing is based on the concept of conserved atomic identity, which
means that reactions only change the associations among the atoms with-
out changing their atomic identity. This means that there must be an equal
number of each kind of atom on both sides of the reaction. Furthermore,
the conservation of charge requires that the total charge must be the same
on both sides of a balanced chemical equation. These requirements are
enforced in five steps.
9
10 Modeling tools
1. Write the chemical formulae for the reactant and product species on opposite
sides of an equal sign.
2. Find the element that is least represented in these species and insert integer (or
sometimes small fraction) coefficients into the equation to make equivalent
amounts of this element on each side of the equation.
3. Repeat this step with the second least represented element and so on until all the
elements are balanced. For reactions involving aqueous solutions, it is usually
necessary to add H2O, O2, H2, or H+ to balance the hydrogen and oxygen.
4. Adjust the coefficients of the charged species to produce an equivalent amount
of charges on each side of the equation.
5. Finally, divide the entire equation through by integers to reduce these coefficients
to small integer values (although sometimes small fractions are acceptable).
Although balancing equations by inspection is the most convenient method,

the process can be daunting for reactions involving complicated molecules.
For the more challenging reactions, various algebraic methods are recom-
mended. The reaction coefficients are related by a series of linear equations,
so complex reactions can be balanced by solving these equations either by
the substitution method or by using matrix algebra (Campanario, 1995;
Krishnamurthy, 1978; Presnall, 1986). The various matrix methods are well
suited for incorporation into computer codes.
Example 2.1. Pyrite oxidation reaction
Acid mine drainage is a common environmental problem associated with disposed

mine wastes that contain iron sulfide minerals, especially pyrite. Acid mine drainage
is a low pH solution containing relatively high concentrations of dissolved iron and
sulfate. Acid mine drainage streams are stained yellow, brown, and red by precipi-
tated ferric oxyhydroxides and hydroxysulfates. We know that oxidation of pyrite
(FeS2) to produce sulfuric acid is the primary cause of acid mine drainage and that,
although there are often several different ferric oxyhydroxides and hydroxysulfate
minerals present, goethite (FeOOH) is one of the most important. Therefore, con-
structing a model of acid mine drainage geochemistry might begin by developing a
chemical equation that describes the overall acid mine drainage process.
1. We know that during the formation of acid mine drainage, pyrite is destroyed
and goethite is formed so the first step is to put pyrite on the left side of the
equation and goethite and sulfate on the right.
FeS2 = FeOOH + SO42−
2. Inspection of this unbalanced “equation” shows that an equal number of

iron atoms occur on each side so no action is needed to balance the iron.
FeS2 = FeOOH
Balancing chemical reactions 11
3. However, there must be two sulfates on the right-hand side to balance the two
sulfur atoms in the pyrite:
FeS2 = FeOOH + 2 SO42−
4. Next, there must be 0.5 H2O on the left-hand side to balance the H in the
goethite on the right-hand side.
FeS2 + 0.5 H2O = FeOOH + 2 SO42−
5. Because the sulfur in the pyrite is reduced, with a formal charge of −1, and the
sulfur in the sulfate is oxidized, with a formal charge of +6, we know that an
oxidant, in this case O2, must occur on the left-hand side of the reaction. There
are already 10 oxygen atoms on the right-hand side of the reaction and only
0.5 on the left so 9.5 oxygen atoms (4.75 O2) are needed on the left-hand side.
FeS2 + 4.75 O2 + 0.5 H2O = FeOOH + 2 SO42−
6. Finally, four hydrogen ions are needed on the right-hand side to charge bal-
ance the sulfate. These can come from the addition of two more water mol-
ecules to the left-hand side.
FeS2 + 4.75 O2 + 2.5 H2O = FeOOH + 4 H+ + 2 SO42−
However, this adds too much oxygen to the left-hand side, so one O2 must be
removed.
FeS2 + 3.75 O2 + 2.5 H2O = FeOOH + 4 H+ + 2 SO42−
7. This reaction is often written with decimal fractions as shown above but the
coefficients can be written as simple fractions.
FeS2 + 15/4 O2 + 5/2 H2O = FeOOH + 4 H+ + 2 SO42−
or the equation can be multiplied by 4 to get

4 FeS2 + 15 O2 + 10 H2O = 4 FeOOH + 16 H+ + 8 SO42−
A more systematic alternative to this approach begins by expressing the reaction

coefficients as algebraic variables and then solving the resulting equations by the
substitution method.
a FeS2 + b O2 + c H2O = d FeOOH + e H+ + f SO42−
Then the relationships between the coefficients that are required by the conser-
vation of mass and charge are expressed by simple equations.
Fe: a = d
S: 2a = f
O: 2b + c = 2d +4f
H: 2c = d + e
chg: e = 2 f
12 Modeling tools
These equations are solved in a stepwise fashion to find the numerical value for
each coefficient.
Let a = 1.
From Fe: d = a = 1.
From S: f = 2a = 2.
From chg: e = 2f = 4.
From H: 2c = 1 + 4 so c = 5/2
From O: 2b = 2d +4f – c = 2 + 8 – 5/2 = 15/2 so b = 15/4
These coefficients produce a balanced equation.

1 FeS2 + 15/4 O2 + 5/2 H2O = 1 FeOOH + 4 H+ + 2 SO42−
In all cases, the value for one of the coefficients must be arbitrarily chosen. The
other coefficients are then scaled to that value.
Notation
The notation used in this book follows geochemistry conventions as closely
as possible but has been modified where necessary to avoid redundancy or
confusion. Geochemical modeling requires an easy to recognize and intern-
ally consistent set of notation. Inconsistent notation complicates model
verification and leads to errors. The notation used in geochemistry is often
different from that used in engineering, chemistry, and biology, and some-
times models require the introduction of new notation. In all cases, notation
should be clearly defined in the model’s documentation and it is good prac-
tice to include a table of notation, with units, in every report.
Thermodynamic functions, especially free energy, enthalpy, and entropy,
are often used in kinetics models. Because thermodynamics is so widely used
in science, many different kinds of notation have developed. The convention
chosen for this book is illustrated using Gibbs free energy as an example.
For the free energy change related to a reaction, the delta in front of ∆Gr°
means that this variable reflects the change in free energy in going from the
reactants to the products. The subscript, r , designates this variable as the
free energy of reaction. The superscript, °, indicates that both the reactants
and the products are in the standard state, 298.15 K (25°C) and either 1 atm
or 100 kPa (1 bar) total pressure. For the free energy of formation of a sub-
stance from the elements, the subscript, f, associated with ∆Gf° means that
this is the free energy related to the formation from the elements (this is
essentially the same as ∆Gr° except that the free energy of formation of the
elements in their most stable state at 298.15 K (25°C) and either 1 atm or
1 bar total pressure is defined as zero. It is often useful to use an appended
Concentration 13
notation for ∆Gf° and ∆Gr° to indicate what reaction or substance the vari-
able refers to. For example, the free energy of reaction for a reaction listed
in equation (1) would be ∆Gr°(1) and the free energy of formation of quartz
would be ∆Gf°(qz).
Concentration and activity notation can be very confusing. Chemists have
long used square brackets, [i], to indicate the activity of species i and paren-
theses, (i), to indicate the concentration of species i, but geochemists have
occasionally reversed this convention. To make things worse, curly brackets
are used in some engineering disciplines to indicate activity. This confusion
is best avoided by abandoning the bracket-parenthesis notation in favor of
using ai to denote activity and mi to indicate molal concentration. Although
molar concentration units are best avoided in geochemistry, when used they
can be represented as Mi. Concentrations given other units such as mol/m3,
ppm, or mg/L units are best represented as ci.
pX notation indicates the negative logarithm (base 10) of quantity X.
Most people are familiar with pH, which is the negative logarithm of the
hydrogen ion activity, i.e. pH = –log aH . By analogy pK = –log K and
+
pCl = –log a Cl .
−
Kinetics variable notation is quite different from discipline to discipline

and often inconsistent within a single discipline. In this book, ri refers to the
change in the number of moles of species i in a system per unit time (mol/
sec), Ri refers to the change in concentration of species i per unit time (molal/
sec), and Ji refers to the flux of species across an interface (mol/m2sec).
Units
The International System of Units was devised to facilitate communication
between scientific and engineering disciplines. Using SI units is highly rec-
ommended because it minimizes errors that arise from complicated units
conversions. The International System of Units is based on seven base units
(Table 2.1). Decimal multiples of these units and derived units are named
using the prefixes listed in Table 2.2.
Concentration
Concentration units listed in Table 2.3 are frequently used by geochemists.
Thermodynamic calculations are typically based on molal concentration
and concentration should be expressed in molal units whenever possible.
Some disciplines use molar or millimolar concentration units supposedly
because of the convenience of mixing of solutions using volumetric flasks.
Unfortunately molar concentrations change whenever the solution density
changes because of changing temperature, pressure, or composition. This
14 Modeling tools
Table 2.1. Base units and important related units in the SI
Quantity Name Symbol
SI base units
length meter m
mass gram g
time second s
electric current ampere A
temperature (thermodynamic) kelvin K
amount of substance mole mol
luminous intensity candela cd
Derived units
volume liter 1 L = 10–3 m3
force newton 1 N = 1 kg m–1
sec–2
energy joule 1J=1Nm
calorie1 1 cal = 4.184 J
pressure pascal 1 Pa = 1 N m–2
bar 1 bar = 105 Pa
atmosphere 1 atm = 101325 Pa
power watt W
electrical charge coulomb C
electrical potential volt V
1
Widely used but not recommended.
Table 2.2. Decimal multiples of SI units
Factor Prefix Symbol Factor Prefix Symbol
1024 yotta Y 10–1 deci d

1021 zetta Z 10–2 centi c
1018 exa E 10–3 milli m
1015 peta P 10–6 micro µ
1012 tera T 10–9 nano n
109 giga G 10–12 pico p
106 mega M 10–15 femto f
103 kilo k 10–18 atto a
102 hecto h 10–21 zepto z
101 deka da 10–24 yocto y
Concentration 15
Table 2.3. The most commonly used concentration units for aqueous solutions
Unit Definition
milligrams per liter, mg/L milligrams solute/1 L solution

millimolar, mM moles solute/103 L solution
molal, m moles solute/1 kg water
molar, M moles solute/1 L solution
mole fraction, X moles solute/∑(moles of all solutes and solvents)
parts per billion, ppb grams solute/109 grams solution
parts per million, ppm grams solute/106 grams solution
per mil, ‰ grams solute/103 grams solution
means that a complex transformation is needed to make molar concentra-

tion consistent with thermodynamic standard states (mole fraction or molal
scales). Molar and molal concentrations are nearly equivalent when the
solutions are dilute (<~0.1 M or m) and near 25°C because the solution
density is near 1 g/cm3.
Some conventions for reporting concentrations of aqueous species are
often confusing to the uninitiated. Sometimes conventions are chosen
because they make calculations convenient but some are based on historical
practice. To avoid confusion, it is useful to follow the most widely used con-
ventions. However, there are cases when these conventions should be vio-
lated to simplify a calculation. Whenever conventions are violated the new
usage should be carefully explained in the documentation that accompanies
the calculations and models. A good example of a convention that makes
calculations easier is the case of dissolved CO2. Carbon dioxide dissolves in
water via two different reactions.
CO2(v) = CO2(aq) (2.1)
CO2(v) + H2O = H2CO3(aq) (2.2)
This means that both CO2(aq) and H2CO3(aq) are present in the solution.
Because there is no simple way to determine the concentration of these
individual species, the concentration of unionized, dissolved CO2 is usu-
ally reported as the sum of the concentrations of CO2(aq) and H2CO3(aq).
This convention is quite useful for thermodynamic calculations and the
reported equilibrium constants for reactions involving H2CO3(aq) as well
as the ∆Gf°(H2CO3(aq)) are based on the convention of lumping these two
species together. However, this convention may cause a problem in kinetics
models because the rate of reaction of CO2(aq) is likely to be different from
the rate of reaction of H2CO3(aq).
16 Modeling tools
A convention that is based on historical practice is reporting dissolved

silica concentrations as mg SiO2/L or as ppm SiO2 rather than in terms of
Si or H4SiO4. This convention appears to be a relic from the time when the
concentration of elements in minerals was reported in terms of oxide com-
ponents, i.e. weight percent oxides.
There are several properties of aqueous systems that are described using
well-defined operations. The variables derived from these operations can be
used in rate models, but their usefulness is often restricted because of their
incomplete link to fundamental (SI) units. Most of the important oper-
ational methods are carefully laid out in Greenberg et al. (1992) and their
significance is explained in Hem (1985). Sometimes comparable values can
be derived by equivalent operations; for example, hardness is now deter-
mined by measuring the calcium and magnesium concentrations of solu-
tions rather than by using a soap test. Although operationally defined values
often have important technological uses, they are seldom useful for detailed
modeling of the thermodynamic or kinetic reactivity of the solution but
they often can be converted into variables that are useful for modeling.
Acidity is measured by titrating the sample with a NaOH solution to
raise its pH to a defined endpoint (usually 8.3). This measures not only the
concentration of free hydrogen ions in the sample but also the amount of
hydrogen ions that are liberated by various hydrolysis reactions. Acidity is
reported as mg CaCO3/L equivalent. This can be interpreted as the amount
of calcite that would have to be added to the solution to raise the pH to 8.3.
Solutions like acid mine drainage or oil field brines contain reduced redox
sensitive species (e.g. Fe2+, H2S) that upon oxidation and hydrolysis produce
hydrogen ions. These solutions are treated with hydrogen peroxide before
the titration to release this latent acidity. Kirby and Cravotta III (2005a)
and Kirby and Cravotta III (2005b) provide a detailed explanation of latent
acidity in acid mine drainage.
Alkalinity is a measure of the acid-neutralizing capacity of the solution.
It is measured by titrating the solution to pH 8.3 to determine phenol-
phthalein alkalinity and to a pH near 4.5 (see Greenberg et al. (1992) for
exact values) to determine total alkalinity. Note that most values reported
as alkalinity are total alkalinity. Alkalinity is also reported as mg CaCO3/L
equivalent, which can be interpreted as the equivalent amount of calcite
needed to consume the amount of acid titrated. Because bicarbonate is usu-
ally the predominate anion in most non-marine surface waters and shallow
groundwater, total alkalinity values are sometimes interpreted as a reflection
of the bicarbonate concentration of the solution. This interpretation should
be applied with caution because several other kinds of anions, including
hydroxide, organic acids, phosphates, silicates, carbonate, and borate, con-
tribute to the measured alkalinity.
Dimensional analysis 17
Conductivity is a measure of the ability of an aqueous solution to carry

an electric current between two chemically inert electrodes. To avoid polar-
ization of the electrodes, an alternating current signal is used. Because the
current is carried by ions, conductivity is related to the amount of dissolved
electrolytes so that the salinity of the solution can be estimated from the
measured conductivity (see Greenberg et al. (1992)).
Eh measures the electrical potential difference, in volts, between an inert
electrode (usually Pt) and a reference electrode (usually Ag-AgCl) that has
been calibrated relative to the standard hydrogen electrode using ZoBell’s
solution (Nordstrom, 1977). This value is frequently not equivalent to
thermodynamic Eh values because the slow rates of many redox reactions
keep them from attaining equilibrium at the Pt electrode.
Hardness was originally designed to measure the capacity of the water to
react with and precipitate a standard soap. Although other cations, includ-
ing H+ and Fe2+, can react with soaps, hardness is currently defined as the
sum of the Ca2+ and Mg2+ concentrations because these cations are usually
the most abundant ones in natural waters. Water hardness is reported as mg
CaCO3/L equivalent and this quantity is calculated as
mg CaCO3/L = 2.5 (mg/L) Ca + 4.1 (mg/L) Mg (2.3)
Salinity is a measure of the mass of dissolved salts in the solution. This

quantity is commonly estimated using an empirical relation between some
measured physical quantity, like conductivity, refractive index, sound speed,
or density, and the dissolved salt content of a standard solution. The cur-
rently recommended methods to determine salinity use electrical conductiv-
ity or density. Details of the methods are given in Greenberg et al. (1992).
Salinity is reported as a dimensionless number such as % or ‰.
TDS is a measure of the total dissolved solids, including unionized species
like SiO2, in the solution. This measurement is slightly different from salin-
ity, which measures only the amount of ionizable salts. TDS is determined
from the weight of the residue of a filtered solution evaporated to dryness at
180°C. It is reported in units of mg total dissolved solids/L.
Dimensional analysis
A dimension is a quantity such as time, mass, temperature, or distance,
that can be expressed numerically in terms of one or more standard units.
Dimensions have homogeneous linear scales so that a 2-meter interval mea-
sured at the beginning of a 1-kilometer traverse is identical in length to a
2-meter interval measured near the end of the traverse. The magnitude of
a dimension is expressed in terms of a pure number and a unit of measure.
Units are standards to which dimensional quantities are compared to obtain
a numerical ratio, which is the pure number. For example, comparison of
18 Modeling tools
a football field to a yardstick gives the ratio of 100:1 allowing us to say

that the football field is 100 units long and the unit of measure is a yard
which has a universally agreed-upon standard length. This means that we
can define any unit of measure for a dimension as long as it is related to any
other unit of measure for that dimension by a constant ratio. This allows us
to convert dimensional measurements from one set of units to another using
the method of unit conversions discussed below. The concept of dimen-
sion requires that mathematical models of dimensional systems have homo-
geneous dimensions so that the individual terms of the defining equations
must have the same units. The methods of dimensional analysis and simili-
tude are widely used in engineering and physics to identify relationships
between variables used in quantitative models and to argue that these rela-
tionships should apply over a wide range of scales. When nonlinear relation-
ships between variables occur, the variables must be expressed in terms of
dimensionless numbers to avoid violating the principle of similitude. There
are many books about dimensional analysis but most of the following was
drawn from Douglas (1969), Schepartz (1980), and Zlokarnik (1991).
One of the simplest applications of dimensional analysis is to convert
a measured or calculated dimension from one system of units to another.
This simply involves multiplying the quantity by conversion factors until
all the units cancel out except the desired ones. For example, if a solution
contains 80 ppm Ca2+, the molal concentration is found by the following
unit conversions:
80 mg 1g 1 mol mol
80 ppm ≈ × × = 2 × 10 −3 = 2 × 10 −3 m (2.4)
1 kg 1000 mg 40 g kg
The universal gas constant can be converted from volume and pressure units
to energy units.
8.314 x10-2 dm 3 bar 1000 cm 3 1 J/bar 8.314 J

× × = (2.5)
mol K 1 dm 3 10 cm 3 mol K
A fundamental principle of dimensional analysis is that after all unit conver-

sions are completed the same units must appear on the right- and left-hand
side of an equation. This principle of homogeneous units is an extremely
useful way to verify a calculation. If the terms of an equation do not have
homogeneous units then the equation is incorrect. Even though homoge-
neous units are not a complete guarantee of the correctness of an equation,
passing this test does give substantial confidence in the equation.
Because geochemical processes are often quite complex, situations for
which no quantitative model exists are often encountered. These situ-
ations can be handled by developing a model using dimensional analysis.
Dimensional analysis 19
This approach assumes that the behavior of the system is unrelated to

its scale. For example, we assume that if the reaction between Ca2+ and
HCO3− forms calcite in 100 g of solution in a laboratory beaker contain-
ing synthetic sea water, it will form calcite in exactly the same way in the
vast reaches of the ocean as long as the solution concentrations are the
same. This idea of scale invariant behavior of systems is called similitude.
Principles of similitude have long been used to build physics and engineer-
ing models (Bridgman, 1931; Huntley, 1967; Kline, 1965; Murphy, 1950;
Staicu, 1982). Although this principle is very powerful and the resulting
models are typically valid over orders of magnitude of scale, quantum
effects can cause similitude-based models to fail at scales approaching the
size of atoms.
Example 2.2. Surface area to volume ratio of a sphere
The surface area to volume ratio of a sphere is used in several models in this
book. Dimensional analysis can be used to find the relationship between the sur-
face area (A) with dimensions of [L2] and the internal volume (V) with dimen-
sions of [L3] for any convex regular solid. The surface area is related to the vol-
ume by a dimensionless constant, b.
[ L2 ] = b[ L3 ]2 3 (2.6)
A = bV 2 / 3 (2.7)
The first step toward finding the value of b, substitutes the definitions of A and
V for a sphere into Eq. (2.7).
2 /3
4 
4π r 2 = b  π r 3  (2.8)
3 
Equation (2.8) can be solved to find b.
4π 4π 1/ 3
b= 2 /3
= 2 /3
= 4.84 (2.9)
 4  4
  π 2 /3  
3 3
For cubes, b = 6. Other regular convex solids have intermediate values of b.

Note that when this relationship between area and volume is scaled downward,
we eventually find a situation where solids do not occupy space in a way that is
compatible with this model. That is because the actual dimensions of an atom
are best expressed in terms of probability density functions rather than as abso-
lutely fixed spatial dimensions.
20 Modeling tools
The concept of similitude is further formalized by the Buckingham π the-

orem that states that an equation containing n variables can be rewritten as
an equation containing n − m dimensionless numbers, where m is the overall
number of independent dimensions in the model.
Dimensionless numbers are needed whenever nonlinear equations are
encountered because transcendental and polynomial functions cannot
have units. Taking the logarithm of 10 moles or raising e to the power of
20 minutes is a meaningless calculation. This does not mean that there
cannot be units within these functions but it does mean that those units
must cancel out. For example, it is perfectly acceptable to raise e to kt
where t is time and the rate constant, k, has the dimension of reciprocal
time so that the product kt has no units. Recasting a variable into a dimen-
sionless number for use in a nonlinear equation can be done in various
ways, but all those ways are subject to the restriction of the Buckingham
π theorem.
Logarithms
Logarithmic transformations are quite common in geochemical models. The
variables used in geochemical calculations often range over many orders
of magnitude so that it is more convenient to work with the logarithms of
these numbers rather than the numbers themselves. In addition, many rela-
tionships in thermodynamics and kinetics are linearized using logarithmic
transformations.
A logarithm is the power (p) to which a base (a) must be raised to produce
a number (N).
ap = N (2.10)
Adding logarithms is equivalent to multiplying numbers so it is often con-

venient to express numbers as powers of ten. Adding the exponents is
equivalent to multiplying the numbers.
245 × 63 = (10 2.39 )(101.80 ) = 10 2.39 +1.80 = 10 4.19 = 1.6 × 10 4
Likewise division is accomplished by subtracting powers of ten.
245 63 = 10 2.39 101.80 = 10 2.39 −1.80 = 10 0.59 = 3.9
Base 10 logarithms are so often used in geochemical calculations that it is

helpful to memorize approximate values for the logarithms of the numbers
1 through 9. The following table offers some memory aids.
Errors 21
log 1 = 0.0 100

log 2 = 0.301 ≈ 0.3 Memorize
log 3 = 0.477 ≈ 0.5 Memorize
log 4 = 0.602 ≈ 0.6 = 22 = 2 log(2)
log 5 = 0.699 ≈ 0.7 Halfway between 4 and 6
log 6 = 0.778 ≈ 0.8 = (2)(3) = log(2) + log(3)
log 8 = 0.903 ≈ 0.9 = 23 = 3 log(2)
log 10 = 1.0 101
Note that the largest errors associated with this method are
for the logarithms of 3 and 6, both of which are incorrect by
~0.02 log units.
Errors
There is some uncertainty in all data, and model building must take this
error into account. The first step in error management is error detection,
error reduction, and error quantification. There are three types of error:
systematic error, random error, and blunders. Improved experimental pro-
tocol can reduce all these, but designing progressively better experiments
eventually leads to diminishing returns so that at some point it is necessary
to use some kind of error analysis to manage the uncertainty in the variable
being quantified.
Systematic errors affect the accuracy but not the precision of the result.
They are usually errors in calibration or observation where the same incor-
rect protocol is applied to all measurements. They displace all measurements
from the true value by the same amount so they cannot be detected by a
statistical analysis of only one data set. However, systematic error can be
detected and reduced by comparing data sets from several different sources
using meta-analysis and systematic review (Rimstidt et al., 2012).
Random error displaces individual measurements from the true value, but
as the number of determinations increases their average becomes closer to
the true value. In a series of independent measurements that contain only
random errors, the most reliable estimate of the true value is the mean of all
the measurements. So that for a series of N measurements, x1, x2, x3, … xN,
the best estimate of the true value of x is x .
∑x i
x= i =1
(2.11)
N
22 Modeling tools
The standard deviation of the measurements, σx, is a measure of the amount

of random error associated with the measurements but it is not a measure of
the error associated with x .
∑ (x − x)
2
i
σx = (2.12)
N −1
The best estimate of the uncertainty in x is the standard error of the

mean, σ x .
σx
σx = (2.13)
N
This uncertainty is reported as x ±σ x or as x (σ x ). The latter convention is

used in this book.
Blunders are one-time mistakes that are random but not normally distrib-
uted. They might be the result of misreading an instrument or an incorrect
notebook entry. The most egregious blunders can be identified and rejected
from the data set using Chauvenet’s criterion. Chauvenet’s criterion is based
on the idea that a typical set of experimental data will have a normal dis-
tribution of errors (Taylor, 1982). That is, 68.3% of the data will fall within
one standard deviation of the mean; 95.5% will fall within two standard
deviations; and 99.7% will fall within three standard deviations. If the data
set contains values that do not conform to this pattern, then these values
are likely to be the result of some error(s) other than the expected random
errors that produce a normal distribution. Because these unusual values are
not part of the expected pattern they should be discarded from the data
set. Keep in mind that this approach assumes a normal distribution of
error. If the measurement produces a different kind of error distribution,
Chauvenet’s criterion cannot be used.
The procedure for applying Chauvenet’s criterion to a series of measure-
ments, x1, x2, x3, … xN, is as follows:
1. Calculate the mean and standard deviation of all the data.
2. Select a suspicious datum (xs) and calculate the number of standard deviations
(ts) between the point and the mean.
xs − x
ts = (2.14)
σs
3. Use a probability table to determine the probability (Ps) that a normally distrib-
uted measurement will differ from the mean by ts standard deviations (Ps = 1
− Ptable).
Significant digits 23
4. Determine the number of measurements in the data set that are expected to be as
bad as xs by multiplying the total number of data (N) by the probability (Ps).
N s = NPs (2.15)
5. If Ns < 0.5, then xs fails Chauvenet’s criterion and is rejected along with all data
that differ from the mean by more than xs.
6. After the data are rejected, calculate a new mean and uncertainty using the
remaining data.
7. Chauvenet’s criterion can be applied to a data set only one time.
Significant digits
The significant digits in a number give an indication of the uncertainty asso-
ciated with that value. This means that the number has been rounded to the
same order of magnitude as its uncertainty. The number of significant digits
is determined by the following rules:
1. The leftmost nonzero digit is the most significant one.
2. In numbers with no decimal point, the rightmost nonzero digit is the least
significant one.
3. In numbers with a decimal point, the rightmost digit is the least significant one,
even if it is a 0.
4. All digits between the least and most significant are counted as significant digits.
For example, 65000 has two significant digits; 0.1010 has four significant
digits; and 65000.1010 has nine significant digits. When numbers are trun-
cated to the proper number of significant digits, the least significant digit
must be rounded up or down according to the following rules:
1. If the digit following the one to be retained is 6, 7, 8, or 9, round the retained digit
up by one.
2. If the digit following the one to be retained is 0, 1, 2, 3, or 4, simply drop it with-
out changing the retained digit.
3. If the digit is 5 and the digit to be retained is even, do not change it. If it is odd,
round it up by one.
The simplest method of error propagation involves keeping track of signifi-

cant digits during a calculation and rounding the answer to the same num-
ber of significant digits as occurs in the least significant number used for the
calculation. This procedure gives a very conservative account of the possible
error in a calculated value. As a rule of thumb, the number of significant
digits in a reported value should not exceed the smallest number of signifi-
cant digits in any number used to calculate that value. For a number with a
24 Modeling tools
reported uncertainty, the least significant digit in the stated value should be
of the same order of magnitude (in the same decimal position) as the most
significant digit of the uncertainty. It is customary to perform the calcula-
tion using numbers with additional significant digits, to avoid introducing
rounding errors and then rounding the final answer to the correct number of
significant digits. For a number without a reported uncertainty, one should
assume that all reported digits are significant. Geochemical measurements
typically have more than 1% error so that, if no uncertainty is stated, three
significant digits are likely to be sufficient to express the reliable part of the
number.
Propagation of errors
Precision is lost as variables are manipulated through formulae. This loss of
precision is taken into account by propagating the uncertainty associated
with the model’s variables through the equations. There are several standard
references that provide detailed discussions of error propagation and man-
agement (Bevington, 1969; Deming, 1943; Mandel, 1964; Taylor, 1982).
For the propagation of uncertainties in equations with addition or sub-
traction, the uncertainty in the calculated value is the square root of the sum
of the squares of the absolute uncertainties.
q = x + y + za − b − c (2.16)
δq = (δ x )2 + (δ y)2 + (δ z )2 + (δ a )2 + (δ b)2 + (δ c )2 (2.17)
For the propagation of uncertainties in multiplication or division, the frac-

tional uncertainty in the calculated value is the square root of the sum of the
squares of the fractional uncertainties.
xyz
q= (2.18)
abc
2 2 2 2 2 2
δq  δx  δ y  δz  δa  δb  δc 
=   +  +  +  +  +  (2.19)
q  x  y  z  a  b  c
For the propagation of uncertainty where a constant is known exactly, the

uncertainty in the calculated value is the product of the constant and the
uncertainty in the variable.
q = Bx (2.20)
Propagation of errors 25
δq = B δx (2.21)
For the propagation of uncertainties through a function of one variable, the

uncertainty in the calculated result is the product of the derivative of the
function and the derivative of the variable.
q = q (x) (2.22)
dq
δq = δx (2.23)
dx
Conversion between logarithms and antilogarithms is so common in geo-

chemical models that it is worth showing the error propagation for these
special cases of the application of Eq. (2.23). The antilog transformation
converts log k to k so q(x) = 10x.
dq = (10 x ) ( ln10 ) δ x (2.24)
δ x = 2.303(10(log x ) )δ log x (2.25)
The logarithmic transformation converts k to log k so q(x) = log(x).
log e
δq = δx (2.26)
x
0.434 δx
δ (log x ) = δ x = 0.434 (2.27)
x x
The last term in this equation is the fractional error in x.

For power law functions, the fractional uncertainty is the product of
the absolute value of the exponent and the fractional uncertainty in the
variable.
q = xn (2.28)
δq δx
= n (2.29)
q x
In the most general case, the uncertainty in the dependent variable of a

function with several independent variables is given by Eq. (2.31).
q = f ( x, y,... z ) (2.30)
26 Modeling tools
2 2 2
 ∂q   ∂q   ∂q 
δq =  δx +  δ y + ... +  δ z  (2.31)
 ∂x   ∂y   ∂z 
In all cases, the maximum fractional uncertainty in a calculated result is

never larger than the sum of the fractional uncertainties in the variables.
δq δx δ y ... δz
≤ + + + (2.32)
q x y z
Example 2.3. Some examples of error propagation
Addition and/or subtraction: It is sometimes necessary to composite samples in

order to have enough material for a chemical analysis. If three samples with
masses of 251 mg, 172 mg, and 302 mg that were weighed on a balance with an
error of 2 mg are composited, the total mass of sample is 725 mg and the uncer-
tainty associated with this value is found using Eq. (2.17).
δ q = 2 2 + 2 2 + 2 2 = 3.46 (2.33)
This means that the mass of the composited sample is 725(3.46) mg.

Multiplication and/or division: The rate of reaction in a mixed flow reaction is
calculated by multiplying the concentration (mol/g) of a chemical species in the
effluent solution by the flow rate (g/sec). If the concentration is 2.31(0.0122) ×
10−4 mol/g and the flow rate is 0.124(0.00131) g/sec, the rate is 1.53 × 10−5 mol/sec
and the uncertainty associated with this value is found using (2.19).
2 2
δq  1.22 × 10 −6   0.00131
=  −4 
+  = 0.0118 (2.34)
q  2 .31 × 10   0.124 
The fractional error in the rate is 1.18% so the absolute error is (0.0118)(1.53 ×
10−5) = 1.81 × 10−7. This means that the rate is 1.53(0.0181) × 10−5 mol/sec.
Multiplication by a constant with no error: A common laboratory activity is to
prepare a solution by weighing a mass of solid and then dissolving it in water.
The concentration is usually reported in terms of the number of moles of solid
rather than the mass. If the mass of NaCl (WM = 58.44) is 40.00(0.01) g, the
number of moles is 40.00/58.44 = (40.00)(0.0171) = 0.6845 moles. We can assume
that there is no random error associated with the molecular weight of NaCl, so
the error for this value is found using Eq. (2.21).
δ q = 0.0171 (0.01) = 0.000171 (2.35)
This conversion shows that the amount of NaCl added to the solution is
0.6845(0.00017) mol.
Regression models 27
Antilog transformation: The intercept of a double logarithm plot was used to

find that log k = −4.18 ± 0.27. Applying the antilog transform shows that k =
6.61 × 10−5. The error for this value is found using Eq. (2.25).
δ k = (2.303) (6.61 × 10 −5 ) (0.27 ) = 4.11 × 10 −5 (2.36)
The antilog transformation shows that k = 6.61(4.11) × 10−5.

Logarithmic transformation: If a rate constant is determined to be k = 6.61(4.11)
× 10−5 the fractional error in k is 0.622. Equation (2.27) is used to find the error
in log k.
δ (log k ) = (0.434 )(0.622 ) = 0.270 (2.37)
The logarithmic transform of this value shows that log k = −4.18(0.27).
Regression models
Fitting data to equations is an important modeling activity that is supported
by an abundance of computer software. The mathematical basis for this
software is described in many statistics books and will not be elaborated
here. Methodologies that are especially suited for kinetics data are reviewed
in Chapter 7 of van Boekel (2009). This section explains some strategies that
can guide fitting kinetic data to equations.
One reason to fit data to a function is to summarize the data and report it
in a way that allows easy interpolation or other mathematical manipulation.
If this is the main objective, the data can be fit to an arbitrary function so
long as that function makes a smooth and parsimonious summary of the
data and is compatible with desired mathematical manipulations. Typically
a low-order polynomial is suitable for this purpose. It is important to avoid
over-fitting the data. The goal is to filter out random error without losing
the information contained in the data. Increasing the number of regression
variables tends to fit the data better as reflected by smaller residuals and a
higher correlation coefficient. This is because the additional variables make
the fitting function more flexible, with more minima and maxima, so it can
pass closer to the data points. However, this additional flexibility can cause
the function to flex wildly and to predict values that are beyond the range
of the data. This unreasonable result can be avoided by beginning the fitting
process using the simplest feasible equation and then adding variables one
at a time until the residuals of the fit are about the same size as the errors in
the data. Sometimes data can be converted to a linear or other simple form
using a logarithmic or power transformation. The fitted equation should not
be extrapolated beyond the domain of the original data.
28 Modeling tools
Example 2.4. Finding the rate of Fe(III) reaction with pyrite
Table 2.4 lists Fe3+ concentration versus contact time with pyrite from a plug
flow reactor experiment (PFR10) (Rimstidt and Newcomb, 1993). The rate of
ferric iron consumption is the time derivative of these data. Finding this deriva-
tive for t = 0 is especially useful because the initial conditions of the experiment
are accurately known so their effect on the rate can be clearly established. One
strategy for finding the initial rate is to fit the first few concentration versus time
Table 2.4. Concentration of Fe3+ versus contact

time between solution and pyrite in a plug flow
reactor experiment (PFR10) (Rimstidt and
Newcomb, 1993)
time, sec Fe3+, molal
0 1.80 × 10−3
4.7 1.16 × 10−3
6.3 1.12 × 10−3
9.1 1.04 × 10−3
20.5 8.24 × 10−4
0.002
0.0015
• •
mFe3+, molal
0.001 •
0.0005
0
0 5 10 15 20 25
t, sec
Figure 2.1. Fit of the PFR10 data to a second-order polynomial (solid line, R 2 =
0.96) and a fourth-order polynomial (dashed line, R 2 = 1.00).
data to a polynomial. Computing the derivative of the polynomial at t = 0 gives

the initial rate.
mFe3+ = a + bt + ct 2 + ... (2.38)
dmFe3+
= b + 2 ct + ... (2.39)
dt
When t = 0, the rate equals b. Figure 2.1 shows fits of a second and a fourth-
order polynomial to the data.
mFe3+ = 1.73 × 10 −3 − 1.23 × 10 −4 t + 3.80 × 10 −6 t 2 (2.40)
mFe3+ = 1.80 × 10 −3 − 3.07 × 10 −4 t + 5.34 × 10 −5 t 2 − 4.11 × 10 −6 t3 + 1.04 × 10 −7 t 4

(2.41)
The second-order fit predicts an initial rate of −1.23 × 10−4 m/sec and the fourth
fit predicts a much faster rate of −4.85 × 10−4 m/sec. Both functions fit the data
well and the fourth-order polynomial fits it so well that the function passes
through all the points so there are no residuals. However, the rate based on the
fourth-order polynomial appears to be too fast because that function does not
smooth out the random errors in the data. Both functions incorrectly predict
that the concentration of Fe3+ increases after ~18 seconds of reaction because
they are simply approximations for a short interval of the true concentration
versus time function.
A second reason to fit data to a function is to test whether the data are con-
sistent with a model or to use a theoretical function to extrapolate experimen-
tal results to conditions otherwise not attainable. The goodness of fit of the
data to a theoretical function can be gauged by the coefficient of determin-
ation (R2), which is the correlation coefficient squared. R2 is often interpreted
as the fraction of the variability of the response variable that is explained
by its functional relationship to the independent variable. For example, if
R2 = 0.8 then 80% of the variation in the response variable is explained by
the model and 20% of the variation is the result of factors, including ran-
dom error, that are not part of the model. Some geochemical processes occur
under conditions or over time intervals that are difficult to simulate by exper-
iments, making it necessary to use a theoretical model to predict their behav-
ior. This approach uses data from experiments performed under easily attain-
able conditions to quantify parameters that are used to calibrate a theoretical
model. These parameters are then used to predict the value of the variable at
30 Modeling tools
0.8
0.6
relative error 0.4
Figure 2.2. The 0.2

relative error in a
statistically estimated
value decreases as
the reciprocal of the 0
square root of the 0 20 40 60 80 100
number of data. number of data
the unattainable conditions. This kind of extrapolation exposes the predic-

tion to two kinds of error. First, the theoretical function may be incorrect or
improperly chosen. This possibility is tested by determining (R2) how well
the experimental data fit to the function for the attainable conditions and by
comparing predicted values to independently estimated values. The second
kind of error is the result of random error in the data that is transformed into
uncertainty in the fitted parameters and this uncertainty is further transmit-
ted to the predicted variable. This random error can be significantly reduced
by increasing the size of the data set (Gauch Jr, 1993). Figure 2.2 shows that
the relative error drops rapidly as the number of data increases. However, this
graph also shows that there are diminishing returns for adding data beyond
those needed to reduce the error to an acceptable level. Once the data set is
established, the regression parameters should be determined using a com-
puter code that calculates their uncertainty. This uncertainty is then forward
propagated to provide an estimate of the uncertainty in values predicted by
the theoretical function. This approach allows for a quantitative determin-
ation of the uncertainty associated with a model’s predictions.
Example 2.5. Kaolinite to illite transformation under diagenetic

conditions
Rates of many mineral transformations under diagenetic conditions are too

slow to measure in the laboratory. In order to model the rate at which kaolinite
would transform to illite during burial diagenesis, Chermak and Rimstidt (1990)
measured the reaction rates from 250 to 307°C and used the Arrhenius equation
Table 2.5. Rate constants for the kaolinite to illite

conversion reaction
T, °C k, sec–1
250 1.18 × 10−10

275 7.78 × 10−10
289 1.00 × 10−9
307 4.48 × 10−9
–6
–8
•
••
–10 •
–12
log k
–14
–16 log k = 16.1
–18
100°C
–20
0.0015 0.002 0.0025 0.003 0.0035
1/T, K–1
Figure 2.3. Arrhenius plot used to extrapolate rate constants for the kaolinite to
illite transformation reaction from high temperature measurements to find the
rate constant at 100°C.
to extrapolate those rates to lower temperatures. In this example the goal is to

predict the rate constant at 100°C. The rate constants from the high temperature
experiments are listed in Table 2.5. These rate constants can be extrapolated to
lower temperatures using the Arrhenius equation.
k = Ae − Ea RT
(2.42)
This equation can be log-transformed to a linear form.
 − Ea   1 
log k = log A +  (2.43)
 R   T 
32 Modeling tools
Fitting the data in Table 2.5 to this equation gives the line shown in Figure 2.3.
This fit explains 97% of the variation in log k (R2 = 0.97).
−8039(992 )
log k = + 5.45(1.8) (2.44)
T
The numbers in parentheses are 1 standard error of the parameter.

Solving this equation for 100°C (373 K) predicts that log k is −16.1 (k = 7.94 ×
10−17 sec−1). The uncertainty in the log k value is found by forward propagation
of the standard errors associated with the fitting parameters in Eq. (2.44).
δ log k = (0.18)2 + (992 T )2 = 2.67 (2.45)
So log k = −16.1(2.7). Even though the data fit the equation quite well, the
uncertainty in log k is nearly 3 log units. Most of this error is due to the uncer-
tainty associated with the slope of the line. This example stands as a warning to
model builders who extrapolate data without considering error propagation.
Numerical differentiation
Kinetic processes cause a change in a quantity over time. If this quantity is
measured at evenly spaced time intervals, it is relatively easy to find the rate
by computing a numerical derivative (Pollard, 1977). The simplest method
of numerical differentiation is based on the observation that the chord of a
graph has a slope that closely approximates the tangent at an intermediate
point.
This leads to a simple scheme for calculating the numerical derivative
from a data set (Pollard, 1977). The derivative at the first point in the data
set ( f−′1) gives more weight to the first two points.
1
f −′1 = ( −3 f−1 + 4 f0 − f1 ) (2.46)
2h
The derivative at each intermediate point ( f0′ ) is the slope of the chord con-
necting the surrounding points (Figure 2.4).
1
f0′ = ( − f−1 + f1 ) (2.47)
2h
The derivative at the last point in the data set ( f+′1) gives more weight to the
last two points.
1
f+′1 = ( f−1 − 4 f0 + 3 f1 ) (2.48)
2h
Numerical differentiation 33
f–1 f0 f+1
h
Figure 2.4. Schematic
f(x) f 0 = 1 ( f 1+ f1)
2h illustration of how
the slope of a chord
approximates the
tangent of a graph at an
x intermediate point.
Note that the equations for the endpoints of the data set are less dependable
than the equation for the intermediate points and they tend to give unrea-
sonable results if the slope is steep. Because numerical differentiation takes
the difference between values of the dependent variable it magnifies the scat-
ter in those data. Some of this scatter can be eliminated by various smooth-
ing schemes, but they tend to bias the data and are best avoided.
Example 2.6. Rate of reaction of Fe3+ with pyrite in a batch reactor
Ferric iron reacts with pyrite to produce ferrous iron and sulfuric acid.
14 Fe3+ + FeS2(py) + 4 H2O = 15 Fe2+ + 2 SO42− + 8 H+ (2.49)
Rimstidt and Newcomb (1993) measured the concentration of Fe3+ in a solution

in contact with pyrite every 300 sec (= h) for 3000 sec. Their data for experiment
BR5 are shown in Table 2.6 and Figure 2.5 along with the numerical derivative,
which is the rate of reaction. The computed rates are negative because Fe3+ is
consumed by the reaction. The rate for the first point (t = 0) is found using Eq.
(2.46).
 1 
f −′1 =   ( −3 (9.79 × 10 ) + 4 (9.61 × 10 ) − 9.45 × 10 ) = −6.33 × 10
−4 −4 −4 −8
 ( 2 )(300 ) 
(2.50)
The rate for the next point (t = 300) is found using Eq. (2.47).
 1 
f0′ =   ( −9.79 × 10 + 9.45 × 10 ) = −5.67 × 10
−4 −4 −8
(2.51)
 (2 )(300 ) 
The subsequent points are treated the same way. The rate for the last point (t =
3000) is found using Eq. (2.48).
 1 
f+′1 =   (9.04 × 10 − 4 (8.98 × 10 ) + 3 (8.91 × 10 )) = −2.50 × 10
−4 −4 −4 −8
 ( 2 )(300 ) 
(2.52)
34 Modeling tools
(a) 1.0× 10–3
9.5× 10–4
concentration, molal
9.0× 10–4
8.5× 10–4
0 1000 2000 3000
time, sec
(b) 0.0
–2.0× 10–8
rate, molal/sec
–4.0× 10–8
–6.0× 10–8
0 1000 2000 3000

time, sec
Figure 2.5. (a) Concentration of Fe3+ versus time from experiment BR5 (Rimstidt
and Newcomb, 1993). (b) Numerical derivatives of the concentration of Fe3+
versus time data. Early in the experiment the rates are fast but they slow to a
nearly constant rate after ~300 sec.
Numerical differentiation 35
Table 2.6. Numerical differentiation of data from experiment BR5

reported in Rimstidt and Newcomb (1993)
t, sec m, molal dm/dt, molal/sec
0 9.79 × 10−4 –6.33 × 10−8

300 9.61 × 10−4 –5.67 × 10−8
600 9.45 × 10−4 –3.83 × 10−8
900 9.38 × 10−4 –2.00 × 10−8
1200 9.33 × 10−4 –1.83 × 10−8
1500 9.27 × 10−4 –2.33 × 10−8
1800 9.19 × 10−4 –2.67 × 10−8
2100 9.11 × 10−4 –2.50 × 10−8
2400 9.04 × 10−4 –2.17 × 10−8
2700 8.98 × 10−4 –2.17 × 10−8
3000 8.91 × 10−4 –2.50 × 10−8
Chapter 3
Rate equations
Chemical thermodynamics explains why reactions occur and chemical kin-

etics investigates how they occur. Reacting systems do not reach equilibrium
instantly and metastable phases and species can persist for geologic time peri-
ods. For example, even though graphite is the thermodynamically stable form
of elemental carbon at Earth surface conditions, diamonds have persisted at
the Earth’s surface for hundreds of millions of years. Metastable persistence
can have a profound effect on element distributions and geological processes,
so a goal of geochemical kinetics is to model metastable persistence. The
time needed for metastable species to convert to stable species is quantified
using rate equations. This chapter deals with examples of rate equations that
are frequently encountered in geochemical models and shows how the differ-
ential rate equations are integrated to quantify metastable persistence.
A chemical reaction converts one or more reactants into one or more
products and the rate equation describes how fast this conversion occurs.
The reaction rate (r, mol/sec) is the time derivative of the number of moles
of chemical species consumed or formed.
dn
r= = k ∏ aini (3.1)
dt
The rate constant, k, incorporates the effects of variables such as tempera-

ture, pressure, or ionic strength on the rate. The rate constant can be thought
of as a measure of the resistance to the conversion and small values of k pro-
duce small values of r. The effect of activity, concentration, or partial pres-
sure of the reacting species on the rate appears in the rate equation as explicit
terms each raised to an appropriate power. The product of these terms is the
driving force for the reaction, so that large values of activity, partial pressure,
or concentration produce large values of r. The exponents for the concentra-
tion, activity, or partial pressure terms are called partial orders (ni) and the
sum of those exponents for a rate equation is the overall reaction order, no.
no = ∑ ni (3.2)
36
Rate equations 37
An elementary reaction step is a reaction that converts reactants directly

to products through a single transition state (see Chapter 5). The reaction
order for an elementary reaction step usually reflects the molecularity of the
reaction. The molecularity of an elementary reaction step is the number of
species that come together to form the activated complex.
Composite reactions consist of multiple elementary reaction steps that
occur in series, in parallel, or both. Many geochemical reactions are com-
posites of several elementary reaction steps. This makes elucidating their
reaction mechanisms very challenging because their reaction order and
molecularity are not related in a simple way. Marin and Yablonsky (2011)
offer extensive guidance about dealing with composite reactions.
There are many ways to express reaction rates, and keeping track of
notation for different kinds of rates along with the units of their accompa-
nying rate equations is challenging. For a simple rate equation such as (3.1),
the rate and the rate constant have units of mol/sec, which are the units
expected from transition-state theory (Chapter 5). A reaction rate can also
be expressed in terms of the time rate of change of concentration of a spe-
cies (R, mol/kg sec = molal/sec), by dividing both sides of Eq. (3.1) by the
mass of water (M) in the system.
dm r 1
R= = = k ∏ aini (3.3)
dt M M
If the reaction takes place on an interface that separates two phases and the
rate equation expresses how fast a component is transferred to or from that
interface, this quantity is actually a flux (J, mol/m2sec).
r 1
J= = k ∏ aini (3.4)
A A
From Eq. (3.3) we see that r = RM and from Eq. (3.4) we see that r = JA. If
we combine these expressions we find how the rate of change of concentra-
tion in a solution is related to the dissolution flux.
dm A A
R= = J= k ∏ aini (3.5)
dt M M
Equations (3.3) and (3.5) contain both activity and concentration variables.
In order to integrate the equation and to make the equations compatible
with transition-state theory (Chapter 5), the concentration in the derivative
term of these equations can be converted to activity.
γ m
a= (3.6)
γ ° m°
38 Rate equations
The conventional hypothetical ideal 1 molal standard state standard makes

γ ° = 1 and m° = 1 mol/kg, so the units on both sides of the equation are
sec−1. The resulting equation is simplified by combining all the extraneous
terms into an apparent rate constant (k′, sec−1).
da  1 γ  γ k
= r =   ∏ aini = k ′∏ aini (3.7)
dt  m°M γ °  M
This approach can also be applied to Eq. (3.5).
da  A γ   A
=  J =   γ k ∏ aini = k ′∏ aini (3.8)
dt  m°M γ °  M
More often, the activity terms on the right-hand side are converted to con-
centration terms by rearranging Eq. (3.6). This allows the equation to be
integrated to give the concentration as a function of time, but can result
in complicated and often meaningless units for the apparent rate constant.
For example, if one or more of the reaction orders is a fraction, as often
is the case for mineral dissolution rates, the apparent rate constant will
have units of molal raised to a fractional power, which has no physical
meaning.
dm  k   k ∏ γ ini 
R= =   ∏ γ ini mini =   ∏ mi = k ′∏ mi
ni ni
(3.9)
dt  M   M 
dm  A  A  
R= ==   k ∏ γ ini mini =    k ∏ γ ini  ∏ mini = k ′∏ mini (3.10)
dt  M    M  
These rate equations describe the rates of production or consumption of

chemical species. It is often desirable to report the overall advancement
for a chemical reaction in terms of the extent of reaction, ξ (mol). The
extent of reaction is defined in terms of the number of moles of species i
(ni, mol) consumed or produced and the stoichiometric coefficient (νi) for
that species.
( ni )t = ( ni )t = 0 + νiξ (3.11)
The reaction rate is then expressed as the time derivative of the extent of
reaction.
d ξ 1 dni
= (3.12)
dt ν i dt
Unopposed reactions (n ≠ 1) 39
Example 3.1. Oxidation of pyrite by ferric iron
The oxidation of pyrite by ferric iron produces ferrous sulfate and sulfuric acid.
14 Fe3+ + FeS2(py) + 8 H2O = 15 Fe2+ + 2 SO42− + 16 H+ (3.13)
The reaction rate can be written in terms of any of the species in Eq. (3.13)
where the negative sign indicates that the species is consumed.
dξ 1 dnFe3+ 1 dnpy 1 dnH2 O 1 dnFe2+ 1 dnSO24− 1 dnH+

=− =− =− = = = (3.14)
dt 14 dt 1 dt 8 dt 15 dt 2 dt 16 dt
The choice of a reaction progress variable for experimental determination of

the reaction rate is limited by practical considerations. Monitoring the rate of
water consumption would be nearly impossible because the amount of water
consumed is minuscule compared to the 55.5 moles of water per kilogram of
solution. Likewise keeping track of the rate of pyrite destruction using a weight-
loss technique is very insensitive compared to determining the concentration
of an aqueous species. Using the rate of hydrogen ion production as a reaction
progress variable is confounded by the fact that the hydrolysis of Fe3+ and Fe2+
involves hydrogen ions. This leaves measuring the concentration of Fe3+, Fe2+,
or SO42− concentrations as the best way for monitoring the reaction progress.
Ideally the rate of consumption of Fe3+ along with the rates of production of
Fe2+ and SO42− would be determined and compared to see if they are consistent
with Eq. (3.14). For example, if the rate of production of sulfate is not ~1/7 the
rate of Fe3+ consumption, either the analytical procedure has a systematic error
or reactions in addition to Eq. (3.13) are consuming or producing these species.
Integrated rate equations

Many geochemical models are based on the integrated form of rate equa-
tions. Some typical examples are shown here and the integrated versions
of other forms can be found in various kinetics textbooks. Capellos and
Beielski (1972) give a comprehensive treatment of integrated rate equations
for simple as well as some composite reactions.
Unopposed reactions (n ≠ 1)

When the rate for the reverse reaction is sufficiently small it can be ignored,
so the overall reaction rate equals the rate of the forward reaction.
A R
+
→ products (3.15)
40 Rate equations
In the simplest case, only one aqueous species affects the reaction rate and
the rate of consumption of species A is proportional to the rate constant (k)
multiplied by its concentration (m, mol/kg) raised to an arbitrary power (n).
Values of n and k are determined empirically.
dm
= − km n (3.16)
dt
This equation is rearranged and integrated between the initial (t = 0, m =

mo) and final conditions (t = t, m = m) to produce a relationship between the
concentration of A and time.
m t
dm
∫m n
= − k ∫ dt (3.17)
mo 0
m(− n +1) − mo(− n +1) = − ( − n + 1) kt (3.18)
(− n +1)

 m
= 1−
( − n + 1) kt (3.19)
 m  mo(− n +1)
o
Equation (3.19) can be simplified by recasting it in terms of the fraction of

material remaining (p).
m
p= (3.20)
mo
 1 
 ( − n + 1) kt   (− n +1)
 
p = 1 − (3.21)
 m(− n +1)  o
In some cases it is better to recast the equation in terms of the fraction

reacted (α = 1 – p).
 ( − n + 1)kt 
( 1− n +1)
α = 1 − 1 − (3.22)
 m( − n +1)  o
Example 3.2. Dissolved oxygen consumption by reaction with pyrite
When air-saturated surface waters infiltrate pyritic mine wastes, the dissolved
oxygen (DO) is consumed by reaction with pyrite.
FeS2 + 7/2 O2 + H2O = Fe2+ + 2 H+ + 2 SO42− (3.23)
Equation (3.21) can be used to find how long it will take for all the DO to be
consumed if 1 kg of air-saturated solution (mDO = 2.7 × 10−4 = 10−3.57 mol/kg)
contacts 1 m2 of pyrite surface.
Unopposed reactions (n ≠ 1) 41
0.8
0.6
p
0.4
0.2
0
0 10 20 30 40 50 60 70
time, hr
Figure 3.1. Fraction of DO remaining as a function of time for 1 kg of air-

saturated solution reacting with 1 m2 of pyrite surface.
Williamson and Rimstidt (1994) report the following rate equation for reaction
(3.23).
J py = A(10 −8.19 mDO

0.5
mH−0+.11 ) (3.24)
The rate of DO consumption is related to the rate of pyrite destruction by the

stoichiometry of Eq. (3.23).
dξ 2
= − J py = − J DO (3.25)
dt 7
Equations (3.5), (3.24), and (3.25) can be combined to produce a rate equa-
tion for DO consumption for a system with 1 kg of water and 1 m2 of pyrite
surface area.
RDO = − 7( 2 )(10 10 0.77 ) mDO

−8.19 0.5
= − (10 0.54410 −8.1910 0.77 ) mDO
0.5
= −10 −6.88 mDO
0.5
(3.26)
In this equation, the rate constant, k = 10−6.88; the initial DO concentration, mo =

10−3.56; and the reaction order, n = 0.5. If α is set equal to zero in Eq. (3.21), the
resulting relationship can be rearranged to give the time needed to completely
consume the DO.
42 Rate equations
t=
mo(1− n ) (10 −3.57 )0.5
= −0.30 −6.88 = 105.40 = 2.48 × 105 sec = 69.0 hr (3.27)

(1 − n )k 10 10
This can be verified by using Eq. (3.21) to create a graph (Figure 3.1) of p as a

function of time.
Unopposed reactions (n = 1)

The above model applies for all values of n other than one. However first-
order reactions (n = 1) are relatively common and the integrated form of a
first-order unopposed rate equation is much simpler.
 dm 
  = − km (3.28)
dt 
Rearranging and integrating this equation between the initial (t = 0, m = mo)

and final conditions (t = t, m = m) gives an equation for the concentration
as a function of time.
m t
dm
∫ m
= −k ∫ t (3.29)
m0 0
 m
ln   = − kt (3.30)
 mo 
It is often useful to express this result in terms of the fraction reacted (α) or
the fraction remaining (p).
m
α = 1− p = = e − kt (3.31)
mo
Characteristic time
Reaction rates are often compared using a characteristic time. For example,
the characteristic time might be defined as the time needed for the destruc-
tion of 50% of the reactants (α = p = 0.5). This time is called the reaction’s
half-life, t½. When t = t½, α = p = 0.5. The half-life for an unopposed reac-
tion where n ≠ 1 is found by setting m = 0.5 in Eq. (3.18).
(− n +1)
mo(− n +1) − (0.5 mo )
t1 ⁄ 2 = (3.32)
( − n + 1) k
Characteristic time 43
The half-life for an unopposed reaction where n = 1 is found by setting m =

0.5 in Eq. (3.30).
ln 0.5 0.693
t1 ⁄ 2 = = (3.33)
−k k
The time constant is another characteristic time that is sometimes used to

describe processes that follow exponential growth or decay as described by
Eq. (3.34). The time constant is defined as tc = 1/k (sec). If t = tc, the reaction
has proceeded to 63.2% of completion.
p = 1 − e −1 = 1.0 − 0.368 = 0.632 (3.34)
After 5 tc, p = 99.3 and for all practical purposes the reaction can be said to
be complete.
Example 3.3. Oxidation of hydrogen sulfide
Hydrogen sulfide is produced as a byproduct of the decay of organic matter in

sediments. When this H2S escapes to the overlying water column, it is oxidized
by reaction with dissolved oxygen. The persistence of this hydrogen sulfide is
strongly dependent upon pH, so it is instructive to express the half-life of H2S in
air-saturated fresh water (I ≈ 0) as a function of pH.
The reaction is H2S + O2 → products. This reaction involves several electron
transfer steps with one, or at most two, electrons transferred per elementary
step, so it is not surprising that there are several observed reaction intermediates
including SO32−, SO42−, and S2O32− (Avrahami and Golding, 1968; Zhang and
Millero, 1993).
In spite of the complexity of the reaction, the rate equation that describes
H2S disappearance is first order in H2S and first order in O2, making it an overall
second-order reaction (Millero, 2001).
RH2 S = − kmH2 S mO2 (3.35)
This is an important and well-studied reaction so the effects of temperature

and ionic strength on k are known. When they are taken into consideration,
the rate constant (k, hr−1) can be expressed in terms of the hydrogen ion
concentration.
14.67 + 2.84 × 10 −7 mH−1+.16

k= (3.36)
1 + 1.05 × 10 −7 mH−1+.0
The concentration of dissolved oxygen in equilibrium with air is 2.7 × 10−4 molal.
This can be combined with the expression for the rate constant above to find an
overall rate constant, k′.
44 Rate equations
200
180
160
140
120
t , hr
100
80
60
40
20 H2S HS–
0
4 5 6 7 8 9 10
pH
Figure 3.2 Half-life of H2S in air-saturated solution as a function of pH at 25°C.

The rapid decrease in half-life near pK1 for H2S suggests that HS− is much more
reactive than H2S.
 14.67 + 2.85 × 10 −7 mH−1+.16 

k ′ = (2.70 × 10 −4 )  (3.37)
 1 + 1.05 × 10 −7 mH−1+.0 
Combining Eqs (3.35) and (3.37) gives an equation that expresses the rate as a
function of hydrogen sulfide concentration.
dmH2 S  3.96 × 10 −3 + 6.70 × 10 −11 mH−1+.16 

RH2 S = =  mH2 S = k ′mH2 S (3.38)
dt  1 + 1.05 × 10 −7 mH−1+.00
This equation is integrated to produce an expression for the fraction remaining

as a function of time.
 m
ln   = ln p = − k ′t (3.39)
 mo 
p = e − (k ′t) (3.40)
Substituting α = 0.5 into this equation gives a relationship for the half-life of
H2S (t½, hr) as a function of hydrogen ion concentration.
Opposed reactions: principle of detailed balance 45
ln 0.5 0.693 0.693

t1 ⁄ 2 = = = (3.41)
−k ′ k′  3.96 × 10 −3 + 6.70 × 10 −11 mH−1+.16 
 1 + 1.05 × 10 −7 mH−1+.0 
0.693 (1 + 1.05 × 10 −7 mH−1+.0 ) 0.693 + 6.71 × 10 −8 mH−1+.0

t1 ⁄ 2 = = hr (3.42)
3.96 × 10 + 6.70 × 10
−3 −11 −1.16
m
H+ 3.96 × 10 −3 + 6.70 × 10 −11 mH−1+.16
A graph of t½ versus pH (Figure 3.2) shows that the half-life of H2S is relatively

long and constant between pH 4 to 6, but declines rapidly as the pH approaches
pK1 of H2S (6.98), suggesting that HS− oxidizes more quickly than H2S.
Opposed reactions: principle of detailed balance

If a reaction proceeds in both the forward and backward direction simulta-
neously, the concentrations of the reacting species change until the reaction
reaches equilibrium.

A← 
→B
R+
(3.43)
R−
For the very simple reaction described by Eq. (3.43), the rate of the for-
ward reaction is the product of the forward rate constant (k+) and the
activity of A.
dmB
R− = = k + aA (3.44)
dt
The rate of the backward reaction is the product of the backward rate con-
stant (k−) and the activity of B.
dmA
R+ = = k − aB (3.45)
dt
The overall reaction rate is the difference between the forward and back-
ward rates.
R = k+ aA − k− aB (3.46)
Regardless of the original activities of A and B, their activities will change

until the system reaches equilibrium. The principle of detailed balance (also
called microscopic reversibility) states that at equilibrium the rate of the for-
ward reaction is exactly equal to the rate of the backward reaction so that
the net rate is zero. When RA = RB, k+aA, eq = k−aB, eq, and that means that the
equilibrium constant is the quotient of the rate constants, K = k+/k−.
46 Rate equations
1.2
0.8
R/R+
0.6
0.4
Figure 3.3. Reaction rate
normalized to the far-from-
equilibrium rate versus 0.2
chemical potential driving
the reaction. The overall
rate approaches zero as 0
the driving force for the –10 –8 –6 –4 –2 0
reaction approaches zero. ∆µr /RT
k+ aB , eq
= =K (3.47)
k− aA, eq
This relationship appears to be valid even when the reaction is not at equi-
librium so long as the reaction mechanism (i.e. transition state) remains the
same. This means that the rate equation can also be written to make the rate
proportional to the activity ratio, Q = aA/aB.
 ka   Q
R = − k+ aA 1 − − B  = − k+ aA 1 −  (3.48)
 k+ aA   K
The chemical potential driving the reaction is a function of Q/K.
Q
∆µ r = − RT ln   (3.49)
K
− ∆µ r
Q
= e RT (3.50)
K
Combining Eqs (3.48) and (3.50) gives an equation that relates the reac-
tion rate to the chemical potential driving the reaction. This relationship
can also be expressed in terms of affinity where A = −∆µr. Affinity is a key
concept in non-equilibrium thermodynamics, which is discussed briefly in
Chapter 10.
First-order forward and first-order backward 47
 − ∆µ r

R = k+ aA 1 − e RT  (3.51)
 
When the reaction is far from equilibrium (∆µr < ~−3 RT) the reverse reac-
tion rate is effectively zero, so the far-from-equilibrium reaction rate (R+)
≈ k+aA. This means that Eq. (3.51) can be written to show how the overall
rate declines as the reaction approaches equilibrium (Figure 3.3).
− ∆µ r
R R
= = 1 − e RT (3.52)
R+ k+ aA
This relationship is valid only if the reaction mechanism remains the same
over the entire range of conditions. If the reaction mechanism changes, the
relationship between R and µ becomes complicated.
Opposed reaction: first-order forward and first-order

backward
The simplest form of an opposed reaction involves only one reactant and
one product species.
RA = k+ mA

A←  →B
 (3.53)
RB = k− mB
The net rate of appearance of B is the difference between the forward and
reverse rates.
dmB
= RB − RA = k+ mA − k− mB (3.54)
dt
Aris (1989) presents three different schemes to integrate this differential rate
equation for an initial condition where the concentration of B is zero, i.e.
mBo = 0 when t = 0. The following derivation is a more general derivation
where mBo ≥ 0. The trick to this derivation is to identify relationships, which
allows Eq. (3.54) to be recast into a form that is easily integrated. The most
useful of these relationships are the conservation of mass and the principle
of detailed balance. The conservation of mass requires that the concentra-
tion of A and B must sum to a constant value so that the sum of concentra-
tions of A and B at any time must equal the sum of their concentrations at
equilibrium (mAe + mBe).
mA + mB = mAe + mBe (3.55)

48 Rate equations
The principle of detailed balance requires that the quotient of the equilib-
rium concentrations be equal to the equilibrium constant. This applies for
dilute solutions where the activity coefficients are very near one.
k+ mBe
K= = (3.56)
k− mAe
The conservation of mass equation (3.55) can be used to write the mA term
in Eq. (3.54) in terms of mB.
dmB
= k+ (( mAe + mBe ) − mB ) − k− mB (3.57)
dt
Rearranging and integrating this equation between the initial (t = 0, mB =

mBo) and final conditions (t = t, mB = mB) gives an equation for the concen-
tration of B as a function of time.
dmB
= k+ ( mAe + mBe ) − ( k+ + k− ) mB (3.58)
dt
mB t
dmB
∫ k+ ( mAe + mBe ) − ( k+ + k− ) mB ∫0
= dt (3.59)
mB 0
The integral is evaluated between the initial and final conditions.
 k ( m + mBe ) − ( k+ + k− ) mB 
ln  + Ae  = − ( k+ + k − ) t (3.60)
 k+ ( mAe + mBe ) − ( k+ + k− ) mBo 
The mA term is eliminated by substitution of a rearranged version of Eq.

(3.56); mA = mBe/K.
 + ( K Be )
 k mBe + m − k + k m 
( + −) B 
 = − ( k+ + k − ) t
( )
ln  (3.61)
 k+ mBe + mBe − ( k+ + k− ) mBo 
 K 
Next the equilibrium constant is eliminated using the principle of detailed

balance, k+/K = k−.
 ( k + k− ) mBe − ( k+ + k− ) mB   mBe − mB 
ln  +  = ln   = − ( k+ + k − ) t (3.62)
 ( k + + k − ) mBe − ( k + + k − ) mBo   mBe − mBo 
 mB 
 mBe − mB  1 − mBe 
ln   = ln   = ( k+ + k− ) t (3.63)
 mBe − mBo  1 − mBo 
 mBe 
This equation can be transformed and rearranged to give the ratio of the
concentration of B to its equilibrium concentration.
 mB   mBo  − (k+ + k− )t
 m  = 1 − 1 − m  e (3.64)
Be Be
If mBo = 0, this equation is relatively simple.
 mB  − (k+ + k− )t
 m  = 1 − e (3.65)
Be
mB = mBe (1 − e − (k+ + k− )t ) (3.66)
Equation (3.65) can be rearranged to find a characteristic time, defined as

the time needed for mB to change from 0 to 0.5 mBe.
0.5 =
mB
mBe
( − ( k + k )t
= 1 − e + − 12 ) (3.67)
ln 0.5 −0.693 0.693

t1 = = = (3.68)
2 − ( k+ + k − ) − ( k + + k − ) k + + k −
Example 3.4. Hydration of CO2
CO2 dissolves in water as CO2(aq), which then reacts with water to form car-
bonic acid, H2CO3.

CO2 (aq ) + H2 O ← 
R+
→ H2CO3 (3.69)
R−
This hydration–dehydration reaction is slow enough to make its rate significant

in some biochemical processes and it may be slow enough to influence the course
of certain geochemical processes. We can get a sense of how fast this reaction
proceeds by calculating its half-life along with the time needed to achieve 99%
of equilibrium.
The rate of the forward reaction is the forward rate constant times the concen-
tration of CO2(aq). Water is in great excess so its activity is set equal to 1.
dmH2 CO3
R+ = = k+ aCO2 aH2 O = k+ aCO2 (3.70)
dt
The rate of the reverse reaction is the reverse rate constant times the concentra-
tion of H2CO3.
50 Rate equations
Table 3.1. Values of k+ and k− at 25°C compiled by Garg

and Maren (1972) along with the average of the values. The
standard error of the mean is given in parentheses.
k+, molal/sec k−, molal/sec
0.0275 31
0.0257 25.9
0.026 20.6
0.043 25.5
0.0358 27
0.0375 15.1
0.044 17.5
0.036 13.7
0.037 18
16.7
Average 0.0347(0.002) 21.1(1.9)
dmCO2
R− = = k− aH2 CO3 (3.71)
dt
The reacting species have zero charge, so we can set their concentrations equal to
their activity. The overall rate of appearance of H2CO3 is the difference between its
rate of formation and the rate of destruction as defined by Eqs (3.70) and (3.71).
dmH2 CO3
R+ − R− = = k+ mCO2 − k− mH2 CO3 (3.72)
dt
It is not uncommon to find multiple determinations of rate constants for impor-

tant reactions like the CO2 hydration/dehydration reaction. Table 3.1 lists several
values of k+ and k− that were tabulated in Garg and Maren (1972). Each of these
values came from an independent experimental study, so a reasonable way to
find the most probable values for these rate constants is to average the reported
values.
The principle of detailed balance can be used along with the average values
for the rate constants to calculate the equilibrium constant for reaction (3.69).
aH2 CO3 k+ 0.0347

K= = = = 1.64 × 10 −3 (3.73)
aCO2 k− 21.1
This relatively small value means that at equilibrium only 0.16% of the dissolved
CO2 is in the form of H2CO3. This is due to the small size of the forward rate
1.0
0.8
0.6
mB/mBe
0.4
0.2 50 % 99 %
0
0 0.1 0.2 0.3
time, sec
Figure 3.4. Graph of mB/mBe versus time showing the time needed for the H2CO3
concentration reaction to reach 50% and 99% of the equilibrium value.
constant relative to the backward rate constant, i.e. H2CO3 forms slowly and
dissociates quickly.
Equation (3.66) can be rearranged to find the time needed for H2CO3 to reach
a fraction of the equilibrium concentration.
1  m 
t=− ln 1 − B  (3.74)
( k+ + k − )  mBe 
The graph of mB/mBe versus time is shown in Figure 3.4.

If mB/mBe = 0.5,
1
t=− ln (1 − 0.5) = − (0.0473) ( −0.301) = 0.0327 sec (3.75)
(21.13)
If mB/mBe = 0.99,
1
t=− ln (1 − 0.99) = − (0.0473) ( −4.61) = 0.218 sec (3.76)
( .13)
21
Although these times may seem short compared to the characteristic times for
many geochemical processes, they are actually very slow compared to the rates
of interaction of most aqueous ions with water molecules.
52 Rate equations
Opposed reaction: zero-order forward and

first-order backward
In some cases, the activity (or concentration) of the reactants remains con-
stant over the duration of a reversible reaction. Mineral dissolution, where
mineral A dissolves to form aqueous species B, is a good example of this
situation.
RA = k+′ aA = k+′

A(s ) ←  → B(aq )
 (3.77)
RB = k−′ aB
Phase A is a solid, so aA = 1 at all times. Because the reaction occurs at

the interface between the solid and the solution, the rate is directly propor-
tional to the area of that interface. This proportionality is incorporated into
apparent forward (k+′ ) and reverse (k−′ ) rate constants.
k+′ = Ak+ and k−′ = Ak− (3.78)
The overall rate of appearance (or disappearance) of B is the difference

between the dissolution and precipitation rate.
daB
RB − RA = = k+′ − k−′ aB (3.79)
dt
This equation can be rearranged, integrated, and evaluated from an initial

condition where aB = aBo when t = 0 to aB = aB when t = t.
aB t
daB
∫ k+′ − k−′ aB ∫0
= dt (3.80)
aBo
 k ′ − k−′ aB 
ln  +  = − k−′ t (3.81)
 k+′ − k−′ aBo 
 k−′ aB   aB 
 1 − k+′   1− K 
ln   = ln   = − k−′ t (3.82)
 1 − k−′ aBo  a
 1 − Bo 
 k+′   K 
The equilibrium constant for Eq. (3.77) is equal to the equilibrium activity
of species B (K = aBe) so K in Eq. (3.82) can be replaced by aBe.
 aB   aBo  − k ′ t
 a  = 1 − 1 − a  e (3.83)
−
 Be   Be 
If no B is present in solution at the beginning of the reaction, aB /aBe = 0.

This activity ratio increases as the reaction proceeds, so that when aB /aBe =
1 the reaction has reached equilibrium.
Zero-order forward and first-order backward 53
 aB 
 a  =1− e
− k−′ t
(3.84)
 Be 
This equation can be rearranged to find a characteristic time (t½, sec), which
is defined as the time needed for aB /aBe to change from 0 to 0.5.
aB
0.5 = = (1 − e − k−′ t1 ⁄ 2 ) (3.85)
aBe
ln 0.5 −0.693 0.693

t1 ⁄ 2 = = = (3.86)
− k−′ − k−′ k−′
Example 3.5. Quartz equilibrium times
Quartz makes up about 15% of the Earth’s crust, so it is not surprising that it
often controls the dissolved silica concentration in adjacent pore waters, espe-
cially at high temperatures. However, at low temperatures the ability of quartz
to buffer the dissolved silica concentration is limited by its slow dissolution and
precipitation rates. We can visualize the time needed for a solution to equilibrate
with quartz as a function of temperature using the rate constants from Rimstidt
and Barnes (1980) to construct a graph showing the concentration (activity) of
dissolved silica versus time for solutions that initially contain no dissolved silica
at temperatures of 25°, 50°, and 75°C. This model considers a system consisting
of 1 m2 of quartz in contact with 1 kg of water.
Quartz dissolves to release silicic acid into solution.
SiO2(qz) + 2 H2O(l) = H4SiO4(aq) (3.87)
K = aH4 SiO4 e (3.88)
The differential rate equation describing quartz dissolution is analogous to

Eq. (3.79).
daH4 SiO4
= k+′ − k−′ aH4 SiO4 (3.89)
dt
The integrated form of this rate equation is analogous to Eq. (3.81).
 k+′ − k−′ aH4 SiO4 

ln   = − k−′t (3.90)
 k+′ − k−′ aH4 SiO4 , o 
If there is no dissolved silica when t = 0, the integrated rate equation becomes

analogous to Eq. (3.84).
aH4 SiO4 = (aH4 SiO4 e ) (1 − e − k−′ t ) (3.91)

54 Rate equations
Table 3.2. Values of the precipitation rate constant and the equilibrium

constant needed for this calculation.
T, °C K k−′
25 1.11 × 10−4 3.80 × 10−10
50 2.50 × 10−4 1.79 × 10−9
75 4.94 × 10−4 6.77 × 10−9
0.0005
0.0004 75°C
silica concentration, molal
0.0003
50°C
0.0002
0.0001
25°C
0
0 20 40 60 80 100
time, yr
Figure 3.5. Graph showing the concentration versus time values for 1 m2 of

quartz surface dissolving into 1 kg of water. The vertical lines show the time
needed to reach 50% of the equilibrium concentration for each temperature.
Because H4SiO4 is an uncharged species, we can assume that its activity coef-
ficient is near 1, and because the activity of quartz and water are 1 we can set
aH4 SiO4 e = K .
This allows Eq. (3.91) to be rewritten in terms of the equilibrium constant and
the precipitation rate constant.
mH4 SiO4 = K (1 − e − k−′ t ) (3.92)
The values of the precipitation rate constant and the equilibrium constant at
25°, 50°, and 75°C (Rimstidt and Barnes, 1980) given in Table 3.2 were used to
calculate the concentration versus time graphs shown in Figure 3.5.
Zero-order forward and first-order backward 55
100
80
60
t1⁄2, yr
40
20
0
0 20 40 60 80 100
temperature, °C
Figure 3.6. t½ for the quartz precipitation–dissolution reaction as a function of

temperature.
Figure 3.5 shows that even though the equilibrium concentration of dis-
solved silica increases considerably with increasing temperature, the time needed
to reach 50% of that equilibrium concentration decreases significantly from
58 years at 25°C to 3.2 years at 75°C.
Another way to visualize this is to create a graph that shows t½ as a function
of temperature using the relationship t1 ⁄ 2 = 0.693 k−′ where log k− = −0.707–
2589/T (K) (Rimstidt and Barnes, 1980). This graph is shown in Figure 3.6. At
temperatures above 100°C the equilibration rates become relatively fast.
Chapter 4
Chemical reactors
Reactor models define the environment in which chemical reactions occur.

Chemical engineers developed the concept of ideal chemical reactors to
model processing equipment in chemical plants (Aris, 1956; Hill Jr., 1977;
Levenspiel, 1972a, 1972b; Schmidt, 1998). Figure 4.1 summarizes the essen-
tial features of the three types of ideal chemical reactors discussed in this
chapter. Batch reactors (BR) are well mixed and they are closed to mass
transfer but open to energy transfer making them equivalent to closed
thermodynamic systems. Mixed flow reactors (MFR) are also well mixed
and they are open to both mass and energy transfer so they are equiva-
lent to open thermodynamic systems. Engineers usually refer to mixed flow
reactors as continuously stirred tank reactors (CSTR) and geochemists
sometimes call them “flow through” reactors. Plug flow reactors (PFR) are
also open to both mass and energy transfer but the reacting solution passes
through the reactor as slugs that do not mix with each other. The slugs of
fluid are closed to fluid loss or gain so they act as if they are batch reactors
moving through space.
Quantitative models of reactor performance are based on the principles
of conservation of mass and energy. This chapter focuses on the conser-
vation of mass, but modeling the energy budget of reactors uses the same
approach. The rate of accumulation of a substance in the reactor equals the
rate of addition minus the rate of removal plus the rate of generation.
racc = rin − rout + rgen (4.1)
Ideal chemical reactors can be linked in many different ways so that the
output of one reactor becomes the input of the next so that the reactions
occur in stages. Staged reactor models can be adapted to simulate most, if
not all, real-world scenarios. This means that the behavior of complex inter-
acting systems can be simplified to combinations of simple reactors, each
of which is relatively easy to model. Ideal chemical reactor models can be
applied as a first approximation to most natural situations and the concept
easily leads to mathematical descriptions of those situations. For example,
56
Chemical reactors 57
Figure 4.1. The three types

of ideal chemical reactors.
The ideal batch reactor
(BR) is well mixed but
closed to mass transfer.
The ideal mixed flow
batch reactor mixed flow
reactor (MFR) is well mixed
reactor
and subject to continuous
mass transfer. The fluid in
an ideal plug flow reactor
(PFR) moves as slugs,
which are closed to mass
transfer with each other
and therefore act as batch
plug flow
reactor reactors moving through
space.
a shallow pond or lake that is well stirred by wind action can be modeled
as a batch reactor. A pool in a low gradient stream, which is stirred by the
stream flow, can be modeled as a mixed flow reactor; and a parcel of fluid
flowing in a rock fracture can be modeled as a plug flow reactor. In addition
to using chemical reactors models to understand natural situations, they are
the basis for designing experiments to measure rates.
Chemical reactors can be embellished to provide effective mixing and
to establish well-defined physical relationships among reacting phases.
Figure 4.2 illustrates some examples of methods devised to bring solids into
contact with solutions under controlled hydrodynamic conditions. Mixing by
an overhead stirrer can suspend small particles to produce a homogeneous
slurry. Magnetic stirrers can also be used for mixing, but the motion of the
magnet on the reactor bottom can grind the particles, which changes the
surface area of the solid to the mass of the solution in an unpredictable way.
Shaking, rocking, or rotating a reactor that has a free space (gas phase) can
also produce effective mixing. This scheme is especially useful for batch reac-
tors that must be sealed to avoid the gain or loss of a vapor phase. Reactors
containing a slurry, packed bed, or monolith can be mixed using an external
recycle loop containing a pump that provides a mixing flow. If the loop is
closed, the system is a batch reactor. If solution is added and removed from
the loop, the system is a mixed flow reactor. Larger grains can be held in a
packed bed through which the solution is forced by pumping. If the bed is
long compared to its lateral dimensions, the reactor will perform as a plug
58 Chemical reactors
overhead externally
stirrer recycled
packed bed
shaken
Figure 4.2. Examples of
schemes bring solids
into contact with fluids in magnetic rotating disk
chemical reactors. stirrer
flow reactor. If the bed is thin relative to its lateral dimensions, it will approxi-
mate the behavior of a mixed flow reactor. Monolith samples, which can be
solid blocks of material or particles imbedded in an inert solid matrix, are
desirable because they can be cleaved or ground and polished to present well-
defined surfaces to the solution and these surfaces can be observed during and
after their reaction with the solution. Monoliths can be presented to the solu-
tion in several ways ranging from simply suspending them in a stirred solu-
tion to flowing the solution past them using various schemes. A widely used
scheme is to embed the monolith into the face of a disk, which is attached to
a rotating shaft. This rotating disk exposes the solid to a well-defined hydro-
dynamic situation, which makes this apparatus useful for studying the rate
of transport-limited dissolution (Alkattan et al., 1998; Levich, 1962; Sjöberg
and Rickard, 1983). Many other schemes are possible, but regardless of the
scheme, the reactor should be designed to operate as closely as possible to an
ideal chemical reactor to simplify modeling its behavior.
Tracer kinetics 59
Tracer kinetics
Tracer kinetics interpreted using the continuity equation is an effective
way to characterize the mixing and flow behavior of chemical reactors.
Tracers are inert substances that are added to reactors, and the tracer con-
centration in the reactor or in the effluent stream is used to infer some-
thing about the reactor performance. Ideal tracers are inert so they do
not react with anything in the reactor. This means that the rgen term in the
continuity equation (4.1) equals zero. Tracers are frequently used to deter-
mine reactor dimensions or flow rates in field settings where direct phys-
ical measurements are impractical or impossible. In addition to displaying
conservative behavior, field tracers must be environmentally benign, inex-
pensive, and easy to quantify at low concentrations. Bromide and lithium
ions meet these criteria and are frequently used. The environmental and
cost requirements for tracers used in laboratory, pilot scale, and indus-
trial settings are less stringent because smaller amounts are used and the
resulting solutions can be captured and treated before disposal. Dyes or
even radioactive tracers might be used under these more controlled con-
ditions. Natural situations are often complicated so the models in this
section should be viewed as approximations of real systems. The models
described by Rubinow (1975) illustrate how tracer models are applied to
more challenging cases.
Tracer dosing follows one of two end member scenarios; the tracer is either
added as an instantaneous spike or it is added continuously beginning at a
known time. Adding spikes is the simplest method but if the reactor is large
a single spike may not produce a detectable signal. On the other hand, con-
tinuous addition requires a metering pump and a larger amount of tracer.
Serendipitous natural tracers can be used, but they are often introduced in
unpredictable and irregular amounts. This makes them useful for tracing
flow paths even if they cannot be used to develop quantitative models of
reactor volumes or flow rates.
Tracer behavior is different in each of the ideal chemical reactors, so the
first step in developing a tracer kinetics model is to decide which reactor
type best simulates the real situation and which kind of tracer dosing has
occurred. If the amount and timing of tracer introduction is known, a for-
ward model can be developed. If the amount and timing of tracer detected
is known, an inverse model can be produced. The equations derived in the
following section use concentration units of mg/L because these units are
typical of field studies. For laboratory experiments the models would use
molal concentration units. The mass (M) variables would be replaced by
mole quantities (n). The volume variables (V) would be replaced by mass of
water (M) and the flow rate (Q) would have units of kg/sec.
Tracer spike addition to a batch reactor

Adding a tracer spike to a batch reactor produces a step increase in the
tracer concentration in the reactor and the concentration remains constant
thereafter. No tracer is discharged from the reactor and no tracer is gener-
ated in the reactor.
M acc = M in − M out + M gen = M in − 0 + 0 = M in (4.2)
The concentration in the reactor after the spike is added (C, mg/L) is the
sum of the initial mass of tracer (Mo, mg) and the mass of tracer added in
the spike (Ms, mg) divided by the total volume of solution (Vs, L).
M o + M s CoVo + C sVs
C = = (4.3)
Vo + Vs Vo + Vs
The initial amount of the tracer in the reactor is the original concentration
of tracer (Co, mg/L) times the original volume of solution (Vo, L).
M o = CoVo (4.4)
The mass of the tracer in the spike is its concentration (Cs, mg/L) times the
volume of the spike solution.
M s = C sVs (4.5)
Example 4.1. Volume of a hot spring pool
Hot spring pools are usually well stirred by thermal convection and rising gas
bubbles, so they can be modeled as ideal batch reactors and the volume of water
in the pool can be determined using a tracer spike.
A typical spike might consist of 1 L of a 4.6 × 105 mg/L bromide solution. Once
the spike is added to the hot spring, it takes a short time for it to mix throughout
the spring pool. If the completely mixed pool water contains 4.6 mg/L, Eq. (4.3)
can be rearranged to calculate the volume of the pool (Vo,, L). This relationship
can be further simplified by realizing that Cs >> C so C can be neglected in the
numerator and that C >> Co so that Co can be neglected in the denominator.
C −C  C 
V = Vs  s ≈ Vs  s  (4.6)
 C − Co  C
 mg 
4.6 × 105
 Cs   L  = 1 × 105 L
V ≈ Vs   = 1 L  (4.7)
C mg 
 4.6 
L
The volume of the spring pool is 1 × 105 L.

Tracer spike addition to a mixed flow reactor 61
Tracer spike addition to a mixed flow reactor

Adding a tracer spike to an ideal mixed flow reactor produces a step increase
in the tracer concentration just as in the batch reactor case, but in this case
the tracer’s concentration declines over time as the reactor’s effluent carries
it away. If the amount of solution in the reactor remains constant and no
more tracer is added by the feed solution or by generation in the reactor, the
continuity equation simplifies to make the rate of accumulation equal to the
rate of loss by the effluent flow.
racc = rin − rout + rgen = 0 − rout + 0 = − rout (4.8)
The rate of removal of the tracer by the effluent solution is directly pro-
portional to its flow rate (Q, L/sec) divided by the reactor volume (V, L).
This quotient (kQ, sec−1) behaves like the rate constant in a first-order rate
equation.
dC Q
= −   C = − kQC (4.9)
dt V 
Equation (4.9) can be rearranged and integrated from the initial tracer
concentration (Co, mg/L) to the tracer concentration at some later time
(C, mg/L).
C t
dC
∫ C = − kQ ∫ dt (4.10)
Co 0
C
ln   = − kQt (4.11)
 Co 
This equation can be transformed to give the concentration of tracer in the

reactor as a function of time.
Q
−  t
C = Coe − kQ t = Coe V 
(4.12)
Example 4.2. Hot spring flow rate
Once the tracer concentration has homogenized in the hot spring, as described
in Example 4.1, discharge from the spring carries away some of the tracer so the
concentration in the spring pool declines as shown in Figure 4.3.
Equation (4.11) can be used along with the data from this graph to find the
discharge of the hot spring. After 5 hours (18,000 sec) have passed, the concen-
tration is 0.76 mg/L so C/Co = 0.165 and ln(C/Co) = −1.80. This means that kQ =
−1.8/−1.8 × 104 = 1.0 × 10−4 sec−1. The volume of water in the spring pool is 1.0
× 105 L so the discharge Q = kQV = (1.0 × 10−4 sec−1)(1.0 × 105 L) = 10 L/sec.
3
C, mg/L
0
0 2 4 6 8 10
time, hr
Figure 4.3. Concentration of tracer in a hot spring as a function of time. After 5

hours have passed, the tracer concentration in the spring pool is 0.76 mg/L.
Tracer spike addition to a plug flow reactor

Adding a tracer spike to an ideal plug flow reactor creates a slug from which
no tracer is lost or gained as it traverses the length of the reactor. For an
ideal plug flow reactor, the concentration of the tracer in the slug is the same
regardless of its position in the reactor. In real plug flow reactors, diffusion
and dispersion cause some of the tracer to advance ahead of the slug and
some to lag behind the slug. The peak of tracer concentration is usually
chosen as the location of the slug. The time needed for the slug to travel a
given distance is used to characterize the flow path. For example, the time
needed for a slug to pass through a packed bed can be used to find the vol-
ume of fluid in the packed bed if the flow rate (Q, L/sec) and bed porosity
(φ) are known.
Qt
V= (4.13)
φ
If the reactor does not contain a packed bed, φ = 1.

Plug flow reactor models must be constructed with a frame of reference
in mind. An observer riding along on the tracer slug would see no change
in concentration over the modeled time and could infer nothing about the
reactor.
Continuous tracer addition to a mixed flow reactor 63
Continuous tracer addition to a batch reactor

If a solution containing a dissolved tracer is added to an ideal batch reactor
at a constant rate, the rate of accumulation of both the tracer and the solu-
tion is described by the continuity equation.
racc = rin − rout + rgen = rin − 0 + 0 = rin (4.14)
The mass of tracer in the reactor (M, mg) is the initial mass (Mo, mg) plus the
mass added over the interval of tracer injection (ΔM, mg) and that amount
equals the product of concentration in the tracer feed (Cin, mg/L) and the
feed flow rate (Qin, L/sec) times the time interval (t, sec).
M = M o + ΔM = M o + CinQint (4.15)
The volume of solution in the reactor at the end of the tracer injection (V,
L) is the initial volume (Vo, L) plus the volume added by the injection flow.
V = Vo + Qint (4.16)
The concentration of tracer (C, mg/L) in the reactor at time = t (sec) is the
mass of tracer from Eq. (4.15) divided by the volume of solution from Eq.
(4.16).
M M o + CinQt
C= = (4.17)
V Vo + Qt
Continuous tracer addition to a mixed flow reactor

If a tracer is added to an ideal mixed flow reactor at a constant rate, the
rate of accumulation of the tracer in the reactor amounts to the difference
between the rate of input by the feed stream and the rate of removal in the
effluent stream.
racc = rin − rout + rgen = rin − rout + 0 = rin − rout (4.18)
dC  Q  Q
=   Cin −   C = kQ (Cin − C ) (4.19)
dt  V  V 
The concentration in the feed stream is constant so this equation can be

rearranged and integrated.
C t
dC
∫ Cin − C = kQ ∫ dt (4.20)
Co 0
 C − Co 
ln  in = kqt (4.21)
 Cin − C 
This equation can be rearranged to find the concentration of the tracer in

the reactor as a function of time.
C = Cin − (Cin − Co ) e − kq t (4.22)
Continuous tracer addition to a plug flow reactor

This case is very similar to the previously discussed case for the injection of
a tracer spike. The time needed for the first spiked slug to reach the reactor’s
discharge is found by rearranging Eq. (4.13).
Vφ
t= (4.23)
Q
All subsequent slugs contain the same tracer concentration as the first.
Measuring reaction rates using chemical reactors

Rate measurements are best performed with experiments that closely
approximate the behavior of ideal chemical reactors. The concept of ideal
chemical reactors is well developed (Aris, 1989; Levenspiel, 1972a, 1972b;
Rimstidt and Newcomb, 1993), so that the methods of extracting rates from
the experimental data are relatively simple and reliable. Chemical reactors
can have many different design elements, some of which are illustrated in
Figure 4.2, but as long as their behavior approximates an ideal chemical
reactor the following analysis methods are applicable.
Batch reactor experiments

Batch reactor experiments are easy to design and execute but are difficult to
interpret. Because batch reactors are closed systems, the rate of accumula-
tion of a species in a batch reactor equals that species’ rate of generation.
The concentration versus time data from batch reactor experiments must be
manipulated to extract the rate of generation. If an elementary reaction is
responsible for the rate of generation, the rate equation will contain a con-
centration term raised to an integer power and it is possible to identify an
appropriate integrated rate equation by fitting the concentration versus time
data to selected integrated rate equations until a good match is found. This
guess method is often impractical because there are innumerable choices for
the integrated rate equation. Furthermore, fits of concentration versus time
Batch reactor experiments 65
data to integrated rate equations, either using linearized forms or using non-
linear regression, will produce dubious results unless the data span a very
large extent of reaction (Rimstidt and Newcomb, 1993).
A more common approach to analyzing batch reactor data is to deter-
mine rates by taking the derivative of the concentration versus time data.
If the concentration is determined at equally spaced times, the numerical
derivative of the concentration versus time data can be used to find the rate
of generation at each sampling time. A rate equation can then be developed
by correlating the rate at each sample time with the conditions that existed
at that time. Single experiments seldom span a large enough extent of reac-
tion to produce a reliable correlation, so rates from multiple experiments
that span a large range of concentrations must be pooled.
An alternative strategy is to perform numerous short-term experiments
that span a wide range of initial conditions and then use the derivative of
a polynomial fit of the concentration versus time data for each experiment
to find the rate at the beginning of each experiment. An advantage of find-
ing this initial rate is that all the rate-controlling parameters are fully con-
strained by the experimental design at the beginning of the experiment. This
method finds the derivative, at t = 0, of an arbitrary function that fits the
concentration versus time data. The simplest fitting function is a straight line
and the slope of the line equals the rate. This works for cases where the rate
is nearly constant over the duration of the experiment. If the rate changes
significantly during the experiment, the concentration versus time graph will
be curved and a straight line underestimates the rate. Fitting the data to a
second (or higher) order polynomial will accommodate some curvature.
m = a + bt + ct 2 (4.24)
The rate is approximated by the derivative of Eq. (4.24).
dm
= b + 2ct (4.25)
dt
The initial rate (t = 0) is b. When t = 0, the rate-controlling parameters are

known so this rate can be correlated with those conditions. This method is
quite effective if the extent of the reaction is small (<~5% of the reactant
consumed) otherwise Eq. (4.25) will systematically underestimate the ini-
tial rate. A higher-order polynomial will better fit data for greater extents
of reaction, but that approach is not recommended because it sometimes
causes unusual flexing of the fit near t = 0 leading to an incorrect estimate
of the initial rate.
An alternative to the polynomial method is the chord method. The chord
method is recommended for experiments that reach larger extents of reac-
tion (Casado et al., 1986; Piscitelle, 1990; Waley, 1981). This method extracts
the initial rate from the slopes, (mt − mo)/∆t, of chords drawn between each
concentration (mt) point and the starting concentration (mo). The slopes of
these chords are graphed versus time and this function is extrapolated to t =
0. This gives the slope of the tangent to the concentration versus time graph
at t = 0, which is the initial rate. Graphs of the chords’ slopes versus time
are nearly linear for small extents of reaction but do show some curvature,
which can be fit by a second-order polynomial. The initial rate is found by
solving this polynomial for the slope at t = 0.
Example 4.3. Initial rate of scorodite dissolution using the polynomial

and chord method
Harvey et al. (2006) measured the rate of scorodite (FeAsO4•2H2O) dissolu-

tion using a batch reactor. One of the experiments produced the concentra-
tion of arsenic versus time data given in Table 4.1. These data can be analyzed
to find the initial rate using either the polynomial fit method or the chord
method.
Table 4.1. Concentration versus time data for a scorodite dissolution rate

experiment
t m ∆m/∆t
sec molal molal/sec
0 1.89 × 10 −6
900 2.95 × 10−6 1.19 × 10−9

1800 3.81 × 10−6 1.07 × 10−9
3600 4.95 × 10−6 8.50 × 10−9
7200 5.77 × 10−6 5.39 × 10−9
10800 6.34 × 10−6 4.12 × 10−9
14400 7.12 × 10−6 3.63 × 10−9
The heavy line in Figure 4.4 is a second-order polynomial fit of all the concen-
tration versus time data in Table 4.1.
mAs = −2.59 × 10 −14 t 2 + 6.89 × 10 −10 t + 2.34 × 10 −6 (4.26)
The “b” term for this fit is 6.89 × 10−10. A comparison of this fit to the first few
data points shows that the slope of the line at t = 0 is too small. This problem
can be remedied by fitting only the first three data points to produce the light
line shown in Figure 4.4.
Batch reactor experiments 67
8×10–6
•
•
6×10–6 •
•
4×10–6 •
2×10–6 •
0
0 4000 8000 12000 16000
time, sec
Figure 4.4. Concentration of arsenic released by scorodite dissolution versus

time fit to a second-order polynomial. The heavy line is a fit to all the data and
the light line is a fit to the first three data.
mAs = −1.32 × 10 −13 t 2 + 1.31 × 10 −9 t + 1.89 × 10 −6 (4.27)
The “b” term for this fit is 1.31 × 10−9, which is nearly twice the value obtained
by fitting all the data points. This example illustrates that a relatively large
amount of error can result from fitting data for too large an extent of
reaction.
The chord method is an improvement over the polynomial fit method but it
requires slightly more manipulation of the data. First the slopes of the chords
shown in Figure 4.5a are calculated and tabulated. Then the chord slopes are
graphed versus time as shown in in Figure 4.5b and these points are fit to a
second-order polynomial.
s = 5.56 × 10 −18 t 2 + 1.45 × 10 −13 t + 1.31 × 10 −9 (4.28)
When t = 0, the slope of the chord, which approximates the tangent, is 1.31 ×
10−9 molal/sec, which is the initial rate.
(a) 8× 10–6
6× 10–6
4× 10–6
2× 10–6
0
0 4000 8000 12000 16000
time, sec
(b) 1.5× 10–9

chord slope, molal/sec
1.0× 10–9
5.0× 10–10
0
0 4000 8000 12000 16000
time, sec
Figure 4.5. (a) Concentration of arsenic versus time graph with chords

connecting the concentration at zero time and the concentration at the time of
each sampling. (b) A graph of the slope of the chords versus time. Extrapolation
of this graph to zero time gives slope of the chord at t = 0, which is the
initial rate.
Mixed flow reactor experiments 69
Mixed flow reactor experiments

Mixed flow reactor experiments are relatively difficult to design and exe-
cute but they produce data that are easy to analyze. When the experiment
reaches steady state, the concentration of all species in the reactor is con-
stant. This means that the rate of species accumulation is zero, so the rate
of generation is the difference between the input rate, which is the flow
rate (Q, kg/sec) multiplied by the feed concentration (min, mol/kg), and the
output rate, which is the flow rate multiplied by the effluent concentration
(mout, mol/kg).
rgen = rout − rin = ( min − mout )Q (4.29)
Usually min is set by the experimental design and remains constant so

only the concentration in the effluent needs to be determined. It is advisable
to determine the flow rate (Q, kg/sec) at each sample time by weighing the
sample container before and after a sample is taken for chemical analysis
because even the best pumps do not maintain constant flow rates over long
time intervals. Weighing samples is more precise and more accurate than
volumetric measurements. If the reactor is well stirred, the effluent stream
is a representative sample of the solution in the reactor so its analysis estab-
lishes the chemical conditions that produced the observed rate.
Example 4.4. Rate consumption of Fe3+ by reaction with pyrite in a MFR
Rimstidt and Newcomb (1993) performed an experiment to measure the rate of

pyrite reaction with ferric iron in a mixed flow reactor. The reactor, containing
2 g of pyrite with a specific area of 0.047 m2/g, was fed by a 1.0 × 10−3 m Fe3+
solution at a flow rate of 2.70 × 10−5 kg/sec. After a time, the experiment reached
steady state where the concentration of Fe3+ in the effluent solution maintained
a constant value of 1.86 × 10−4 molal.
Based on this information, the rate of Fe3+ consumption in the reactor was
found using Eq. (4.29).
mol
rgen = ( min − mout )Q = (10.0 × 10 −4 − 1.86 × 10 −4 ) (2.70 × 10 −5 ) = 2.20 × 10 −8
sec
The flux of Fe3+ to the surface of the pyrite was determined by dividing this
rate by the total surface area of the pyrite.
mol
2.20 × 10 −8
J Fe3+
r 3+
= Fe = sec = 2.34 × 10 −7 mol (4.30)
A  m2  m 2sec
(2 g )  0.047 
 g 
Plug flow reactor experiments

Plug flow reactor experiments are more complicated to design and execute
than batch reactor experiments, and they produce concentration versus
time data that must be analyzed using the same methods that are used for
batch reactors. As a result, plug flow reactor experiments are seldom used to
determine rates. However, the very high surface area to solution volume of
packed beds achievable in plug flow reactors makes them useful for meas-
uring the dissolution or precipitation rates of very unreactive solids. A key
assumption for such experiments is that the surface reactivity of the solid
does not change during the experiment.
Either the polynomial method or the chord method can be used to find an
initial rate from plug flow reactor data. The chord method is especially con-
venient in this case because the difference between the concentration in the
effluent and feed streams (∆m, molal) divided by the transit time for a slug
of solution is a chord. The transit time is the reactor volume (V, L) times the
porosity (φ, no units) divided by the flow rate (Q, L/sec).
Vφ
Δt = (4.31)
Q
Setting Q to different values produces different transit times with corre-

sponding different values of ∆m. Dividing the resulting ∆m by the transit
time gives the slope of a chord. Extrapolating a graph of these chord slopes
to ∆t = 0 gives the initial rate.
The chord method determines the initial rate, R (molal/sec). To convert
this rate into the flux of species to or from the surface of the solid, R must be
divided by the A/M ratio for the packed bed as described in Chapter 3.
Example 4.5. Rate consumption of Fe3+ by reaction with pyrite in a PFR
Rimstidt and Newcomb (1993) performed an experiment in a plug flow reactor

where a feed solution containing 1.0 × 10−3 m Fe3+ passed through a packed
bed of pyrite with an A/M of 334 m2/kg at different flow rates. The data for this
experiment are shown in Table 4.2.
Figure 4.6 shows that regression of ∆m/∆t versus ∆t to a second-order polyno-
mial gives an excellent fit with R2 = 0.997.
The chord slopes given in the table were correlated with transit time using a
second-order polynomial (Figure 4.6).
Δm
= 6.25 × 10 −8 t 2 − 2.23 × 10 −6 t + 2.42 × 10 −5 (4.32)
Δt
Plug flow reactor experiments 71
Table 4.2. Data for Fe3+ reaction with pyrite plug flow experiment.
Q mout ∆t ∆m ∆m/∆t
L/sec molal sec molal molal/sec
3.92 × 10−5 3.94 × 10−6 20.50 9.61 × 10−5 4.69 × 10−6

8.83 × 10−5 1.56 × 10−5 9.10 8.44 × 10−5 9.27 × 10−6
1.28 × 10−4 2.19 × 10−5 6.30 7.81 × 10−5 1.24 × 10−5
1.71 × 10−4 3.06 × 10−5 4.70 6.94 × 10−5 1.48 × 10−5
2.17 × 10−4 3.66 × 10−5 3.70 6.34 × 10−5 1.71 × 10−5
2.5×10–5
2.0×10–5
1.5×10–5
∆m/∆t
1.0×10–5
5.0×10–6
0
0 5 10 15 20 25
time, sec
Figure 4.6. Slope of the chords versus time for the Fe3+ reaction with pyrite in a
plug flow reactor experiment. The intercept of this graph is the initial rate.
Extrapolation of this function to ∆t = 0 gives an initial reaction rate of 2.42 ×

10−5 molal/sec. To obtain the flux of Fe3+ from the solution to the pyrite surface
this rate is divided by the A/M for the experiment to get J = 6.72 × 10−8 mol/
m2sec.
Double log method

To produce geochemical rate models, rates determined by reactor experi-
ments must be converted into rate equations that summarize how the rate
varies with solution composition, temperature, and other rate-determining
variables. If the rates are determined at near-equilibrium conditions, the rate
data must be fit to an equation that takes into account both the forward and
reverse rate. Most geochemical rate experiments are designed to measure
rates for far-from-equilibrium conditions where the reverse reaction rate is
effectively zero. These experimental rates can be fit to a simple equation that
relates the rate to the product of the concentration (m, molal) of each react-
ing species raised to a power (n).
r = k ∏ mini (4.33)
This equation is nonlinear; direct fitting requires the use of nonlinear regres-
sion, which can be a daunting task. This problem can be avoided by log-
transforming Eq. (4.33) to a linear form.
log r = log k + ∑ ni log mi (4.34)
Fitting the data to this equation using linear regression is a simple way to
find the rate constant and the apparent reaction orders. This equation can
be further generalized by transforming the log k term using the Arrhenius
equation (see Chapter 5). The activation energy (Ea, kJ/mol) term in this
equation expresses how the rate changes with changing temperature (T, K).
E  1
− a 
k = Ae  R T
(4.35)
A is a pre-exponential factor with the same units as the rate constant and R
is the gas constant (8.314 J/mol K). Equation (4.35) can be log-transformed
to express log k in terms of 1/T.
 Ea  1
log k = log A −  (4.36)
 2.303R  T
Combining Eqs (4.34) and (4.36) produces a function that relates the
observed rate to temperature and solution composition.
 Ea  1
 2.303R  T ∑
log r = log A −  + n log mi (4.37)
Multiple linear regressions must be used to fit data to Eq. (4.37). See for
example Rimstidt et al. (2012).
Non-ideal chemical reactors 73
Example 4.6. Fluorapatite dissolution flux as a function of pH
Chaïrat et al. (2007) measured the rate of calcium release from dissolving fluora-
patite over the pH range of 3 to 7. The predominant reaction produces calcium
and fluoride ions and undissociated phosphoric acid.
Ca5(PO4)3F + 9 H+ = 5 Ca2+ + 3 H3PO4 + F− (4.38)
Figure 4.7 shows that the logarithm of the Ca2+ release flux versus pH is linear
over this pH range so these data can be fit to Eq. (4.34).
log J = −4.91(0.09) − 0.88 (0.02 ) pH (4.39)
This equation can be transformed to give the Ca2+ flux as a function of the activ-
ity of H+.
(0.020)
J = 10 −9.91(0.09) aH0.88
+ (4.40)
–7
–8
log J
–9
–10
–11
3 4 5 6 7
pH
Figure 4.7. Logarithm of the Ca2+ release flux versus pH for fluorapatite

dissolution. The equation for the best-fit line is (4.39).
Non-ideal chemical reactors

There are many cases when chemical reactors behave non-ideally so that
reactor models must be adjusted to account for the processes that cause the
non-ideal behavior. Following are a few examples of how non-ideal cases

can be modeled.
Sampled batch reactor

Most batch reactor experiments that are used to measure the dissolution
of solids are designed so that the mass of solution declines as samples are
removed but the amount of solid, which settles to the bottom of the reac-
tor during the sampling, remains constant. This means that the A/M ratio
changes from sample to sample. To correct the concentration of each sam-
ple to the concentration that would have been found if the A/M ratio had
remained constant, the concentration of each subsequent sample must be
adjusted downward to account for the declining mass of solution. The
amount of a substance dissolved into the solution during an interval between
samples equals the change in the concentration over that interval multiplied
by the mass of solution in the reactor during that interval. The total amount
of substance (n, moles) that has been released to the solution at any sample
time is the sum of the mass of solution (M, kg) times the change in concen-
tration between samples (∆m, mol/kg).
sample n
n= ∑ ( Δm) M (4.41)
sample 1
Dividing n at each sample time by the initial mass of solution in the reactor
corrects the concentration to what it would have been if no solution had
been withdrawn by previous samples. This corrects the concentration versus
time data to match that expected for an ideal batch reactor.
Alternatively, using the chord or polynomial fit method to find the initial
rate approximately corrects for the changing A/M ratio in a sampled batch
reactor provided the sample size is small relative to the total amount of solu-
tion in the reactor.
Mixing in reactors
Modeling incomplete mixing in chemical reactors can be quite challenging.
Most models use a distribution function of the residence time for fluid elem-
ents in the reactor (Danckwerts, 1953; Denbigh and Turner, 1984). Nauman
(2008) provides a concise history and a useful review of this theory.
Many reaction-transport models treat parcels of fluid as if they are slugs
flowing in a plug flow reactor. In real systems, hydrodynamic mixing and
diffusion (Aris, 1956; Taylor, 1953) cause dispersion of species among the
slugs. Gelhar et al. (1992) review the factors that affect dispersion in aquifer
systems.
Non-steady state mixed flow reactors 75
Non-steady state mixed flow reactors

Reaction rates can be determined from mixed flow reactors operating under
non-steady state conditions using the following procedure. Material balance
as defined by Eq. (4.1) requires that the rate of generation of a species by
a reaction (rgen, mol/sec) equals the rate of output (rout, mol/sec) minus the
rate of input (rin, mol/sec) plus the rate of accumulation (racc, mol/sec) in the
reactor.
rgen = (rout − rin ) + racc (4.42)
The (rout − rin) term is the species concentration in the feed solution (min, mol/
kg) minus its concentration in the effluent (mout, mol/kg) multiplied by the
flow rate of the effluent stream (Q, kg/sec).
rout − rin = ( mout − min )Q (4.43)
The rate of accumulation of the species in the reactor is the time rate of
change of its concentration multiplied by the mass of solution in the reactor
(M, kg).
 dm 
racc =  out  M (4.44)
 dt 
Equations (4.42), (4.43), and (4.44) are combined to find the rate of reaction
for non-steady state conditions.
 dm 
rgen = ( mout − min )Q +  out  M (4.45)
 dt 
The first term in this equation is simple to evaluate using the concentration
and flow rate data that are typically collected for MFR experiments. The
second term is evaluated by fitting the effluent concentration to a power law
function and then taking its derivative.
mout = at b (4.46)
 dmout 
 = abt
b −1
 (4.47)
dt 
The a and b parameters are then used to find the rate of change in the efflu-
ent concentration. The rate of generation at each sample time is found by
combining Eqs (4.45) and (4.47).
rgen = ( mout − min )Q + (abt b −1 ) M (4.48)

Equation (4.44) shows that the larger the mass of solution in the reactor
the larger the rate of accumulation relative to the rate of generation and the
longer it takes for the reactor to reach steady state. This means that a reactor
designed to contain a very small amount of solution will reach steady state
relatively quickly.
Example 4.7. Ca2+ flux from dissolving wollastonite in a non-steady

state MFR
Weissbart and Rimstidt (2000) measured the concentration of Ca2+ in the efflu-
ent of a mixed flow reactor in which wollastonite was dissolving. In their experi-
ments, the Ca2+ concentration in the effluent decreased continuously over the
Table 4.3. Calculation of non-steady state rates of Ca2+ release from dissolving wollastonite. The
mass of solution in the reactor 0.047 kg. Data fromWeissbart and Rimstidt (2000).
Q t mCa (Q)(mCa) (dmCa/dt)(M) rCa

kg/sec sec mol/kg mol/sec mol/sec mol/sec
U1 4.25 × 10−5 1632 1.43 × 10−3 6.08 × 10−8 −8.65 × 10−8 −2.57 × 10−8
U2 2.47 × 10−5 15960 8.13 × 10−4 2.01 × 10−8 −1.02 × 10−9 1.91 × 10−8
U3 1.70 × 10−5 83220 2.16 × 10−4 3.67 × 10−9 −4.08 × 10−11 3.63 × 10−9
U4 1.17 × 10−5 97200 1.99 × 10−4 2.32 × 10−9 −3.02 × 10−11 2.30 × 10−9
U5 4.02 × 10−5 99600 3.40 × 10−5 1.37 × 10−9 −2.88 × 10−11 1.34 × 10−9
U6 3.10 × 10−5 159900 5.79 × 10−5 1.79 × 10−9 −1.14 × 10−11 1.78 × 10−9
U7 9.67 × 10−5 161700 2.20 × 10−5 2.13 × 10−9 −1.12 × 10−11 2.12 × 10−9
U8 1.05 × 10−4 177060 1.85 × 10−5 1.95 × 10−9 −9.38 × 10−12 1.93 × 10−9
U9 8.43 × 10−5 346020 1.94 × 10−5 1.64 × 10−9 −2.54 × 10−12 1.63 × 10−9
U10 7.95 × 10−5 509700 1.94 × 10−5 1.54 × 10−9 −1.20 × 10−12 1.54 × 10−9
U11 2.25 × 10−4 600000 4.80 × 10−6 1.08 × 10−9 −8.70 × 10−13 1.08 × 10−9
course of the experiment and appears to not reach a steady state. The Ca2+ flux
from the dissolving wollastonite can be found using Eq. (4.48) and the data in
Table 4.3.
The (Q)(mCa) column of Table 4.3 is the rate that the effluent removes Ca from
the reactor. It is the first term of Eq. (4.48). The second term in Eq. (4.48) is
found by fitting the concentration of Ca2+ in the effluent versus time to a power
law (Figure 4.8a).
mCa2+ = 3.52t −0.948 (4.49)

Non-steady state mixed flow reactors 77
(a) 10×10–4
8×10–4
6×10–4
mCa, mol/kg
4×10–4
2×10–4
0
0 200000 400000 600000
time, sec
(b) 2.0 ×10–8
1.5 ×10–8
rCa, mol/sec
1.0 ×10–8
5.0 ×10–9
0
0 200000 400000 600000
time, sec
Figure 4.8. (a) Ca concentration in the MFR effluent solution as a function of

time. (b) Rate of Ca release from dissolving wollastonite as a function of time.
The equation for the line is rCa = 7.18 × 10−6t-0.675. A combination of errors caused
the method to overcorrect the rate for sample U1 so that datum is not shown
on the graphs.
The time rate of change of Ca concentration in the reactor solution is the deriva-
tive of Eq. (4.49).
 dmCa 
 = (3.52 ) ( −0.948) t = −3.34t −1.948
−1.948
 (4.50)
dt 
Equation (4.50) is evaluated for each sample time and the (dmCa/dt)(M) column
reports the dissolved Ca that is retained in the reactor causing the Ca concentra-
tion to increase. Adding the (Q)(mCa) and (dmCa/dt)(M) columns gives the rate of
calcium release from the wollastonite at each sampling time (Figure 4.8b).
Minerals usually dissolve incongruently during the initial stages of the dis-
solution process. In this case, Ca is released from wollastonite faster than Si. In
addition, the rate of both Ca and Si declines with increasing extent of reaction.
Eventually the difference between release rates, as well as the change in the rates,
from one sample time to the next becomes smaller than the resolution of the
rate measurements. When both of these conditions exist the rates are assumed
to represent the dissolution rate that is appropriate for long-term geochemical
processes.
Chapter 5
Molecular kinetics
Rate equations are quantitative models of the time course of chemical reac-
tions. Although rate equations are based on macroscopic observations, they
reflect processes that occur at the molecular scale. This chapter reviews some
of the important models that link these two scales. These models are espe-
cially useful because they constrain the mathematical form of rate equa-
tions and they provide a conceptual basis for thinking about the reactions.
Because water is so important in geochemical systems, this chapter focuses
on models for reaction rates in the aqueous phase.
Transition-state theory
Chemical reactions break and reform bonds within and between molecules
so that the bonding arrangement of the reactants gives way to the bond-
ing arrangement of the products. For a reaction to occur: (1) the molecules
must “collide” with each other to form a cluster; (2) the atoms in that cluster
must be configured in the approximate geometry of the products; and (3)
the cluster must contain sufficient energy to allow the rearrangement of the
electrons from the breaking bonds to the developing bonds.
The idea that molecules must collide to form a cluster correctly predicts
that the reaction rate increases with concentration (or pressure) because
increasing the number of molecules per unit volume will result in more col-
lisions. This is the reason that rate equations stipulate that rates increase
with increasing concentration, pressure, or activity.
The Franck–Condon principle states that the rearrangement of electrons
to form new bonds occurs so quickly that the nuclei of the reacting atoms do
not change position during the bond breaking–bond forming process. This
means that for a reaction to occur, the reacting species must collide in such a
way that they form a cluster with the atoms arranged with the geometry of
the product molecules. This requirement means that most collisions do not
lead to a reaction. A reactive cluster that meets these geometric constraints
is called an activated complex.
79
80 Molecular kinetics
cool
Figure 5.1. The Maxwell–

Boltzmann distribution of number of molecules
molecular energies shows hot
that at low temperature
there are fewer high-
energy molecules available
to form clusters that have
energies that exceed Ea
the activation energy.
Increasing the temperature
increases the number
of molecules that have
enough energy to form
activated complexes. energy
To form an activated complex, the colliding molecules must have enough

energy to overcome the repulsive forces that would otherwise keep them
from aligning into the geometry of the product molecules. The Maxwell–
Boltzmann energy distribution portrays the sharing of energy among a col-
lection of interacting molecules (Figure 5.1). This distribution shows that
only a relatively few molecules have enough energy to form a cluster with
the critical amount of energy needed for a reaction to occur. This critical
amount of energy is called the activation energy (Ea, kJ/mol). Increasing the
temperature distributes more thermal energy among the molecules so that
there are more clusters with sufficient energy to form an activated complex.
This explains why reaction rates increase with increasing temperature.
Transition-state theory, invented by Eyring (1935) and Evans and Polanyi
(1935), is a statistical mechanics model that quantitatively accounts for the
three factors that control the reaction process. Eyring used the term “acti-
vated complex” and Evans and Polanyi used the term “transition state”, to
describe the cluster of molecules that is transitional between the reactants
and products. These terms continue to be used interchangeably.
Transition-state theory arose from statistical mechanics and describes
the distribution of energy among molecules using partition functions. This
means that a firm grasp of the theoretical underpinnings of transition-state
theory requires an understanding of statistical mechanics. Houston (2001)
describes many of the concepts discussed here from a statistical mechanics
point of view and Fueno (1999) explains partition functions. However, the
quasi-equilibrium model presented here is a simpler approach that illustrates
Quasi-equilibrium model 81
the conceptual basis of transition-state theory without resorting to the use

of partition functions.
Quasi-equilibrium model
The quasi-equilibrium model considers a simple reaction where atom A
reacts with molecule BC and displaces atom C to produce molecule AB.
A + BC = AB + C (5.1)
For this reaction to occur, molecule A must approach molecule BC from the
correct direction with enough translational energy to form a transition-state
cluster.
A + BC = ABC* (5.2)
Figure 5.2 shows how the energy of A + BC changes as a function of the

distance between the molecules. A transition state can only form when the
translational energy of A and BC equals or exceeds the energy needed to
reach the saddle point that lies between the two valleys that parallel the
A↔BC and the AB↔C axes. Once a transition-state cluster has formed it
will break down to form AB +C.
The quasi-equilibrium model assumes that the transition-state cluster is
in equilibrium with the A and BC species so that their activities are related
by an equilibrium constant.
aABC*
K* = (5.3)
aA aBC
The activity of the transition-state cluster, ABC*, is found from the product
of the equilibrium constant and the activity of each reactant.
aABC* = K *aA aBC (5.4)
The concentration of ABC* is its activity divided by its activity coefficient.
 1  *
mABC* =  K aA aBC (5.5)
 γ ABC* 
In this case, the concentration of the activated complex is expressed in mol/

kg (molal) but other concentration units could be used. The rate that ABC*
dissociates to form products, AB + C, is controlled by the vibrational fre-
quency (kBT/h) of the A–B bond.
k T  k T   K* 
R =  B  mABC * =  B   a a (5.6)
 h   h   γ ABC*  A BC
(a)
A B- C Reactants
BC distance
A
Transition state Products

A-B-C
A- B C
AB C distance
(b)
A-B-C
Ea
Energy
Figure 5.2. (a) Schematic

contour plot showing
the energy of the A-B-C
cluster as a function of
the distances between
the atoms. ABC is the A B-C
transition state. (b)
Graph of the energy of
the cluster as a function
of its location along the A- B C
reaction coordinate,
which is the dashed line
in (a). Reaction coordinate
The Boltzmann constant (kB) has a value of 1.381 × 10−23 J/K and the Plank
constant (h) is 6.626 × 10−34 J sec. This equation is further modified by intro-
ducing a transmission coefficient (κ, no units) to account for the fact that a
few of the transition-state clusters fail to form products and instead revert
Quasi-equilibrium model 83
back to being reactants. The transmission coefficient often has a value near
one and is usually ignored.
 κ   kBT  *
R=   K aA aBC (5.7)
 γ ABC*   h 
Unit analysis of this equation shows that it has units of molal/sec. The first
three terms are combined into a rate constant.
 κ   kBT  *
k=  K (5.8)
 γ ABC*   h 
The resulting equation is a rate law for a second-order reaction.
R = kaA aBC (5.9)
Example 5.1. Concentration of the activated complex for the ferric

thiosulfate reaction
Williamson and Rimstidt (1993) reported the rate of decomposition of ferric

thiosulfate to form ferrous iron and tetrathionate ions. The reaction brings two
ferric thiosulfate molecules together into an activated complex, which decom-
poses to form two ferrous ions and a tetrathionate ion.
2 FeS2O3+ = [(FeS2O3)22+]* → 2 Fe2+ + S4O62− (5.10)
At 25°C the rate equation for this reaction is
R = −100 mFeS
2
+
2 O3
(5.11)
If the concentration of FeS2O3+ is 0.001 molal, the rate of disappearance of

ferric thiosulfate is 0.0001 mol/kg sec. Equation (5.6) can be used to find the
concentration of the postulated activated complex, [(FeS2O3)22+]*. At 298 K,
(kBT/h) = 6.214 × 1012 sec−1.
R 1 × 10 −4 mol
m (FeS2O3 )22+  * = = = 1.61 × 10 −19 (5.12)
 
( B )
k T h 6 .21 × 10 12
kgg
This concentration is well below detection limits for aqueous species and any
[(FeS2O3)22+]* cluster that forms has a lifetime of ~1.6 × 10−13 sec, which makes
this exercise entirely hypothetical but hopefully instructive.
Effect of temperature on rates

The quasi-equilibrium model can be further developed to show how tem-
perature affects rates. The first step is to relate the free energy of activation
(∆Gr*) to ln K*.
∆Gr* = − RT ln K * (5.13)
Because ∆Gr = ∆Hr − T∆Sr, Eq. (5.13) can be expanded to relate the enthalpy
and entropy of activation to K*.
∆H r* − T ∆Sr* = − RT ln K * (5.14)
Combining Eqs (5.8) and (5.14) gives a relationship between the enthalpy
and entropy of activation and the rate constant (k).
 − ∆H r*  1  ∆Sr*  κ kBT  
ln k =  + + ln   (5.15)
 R  T  R  γ ABC* h  
For an ideal gas, the activation energy (Ea kJ/mol) is related to ∆H*.
∆H * = ∆ ( E + PV ) = ∆E a + ∆nRT (5.16)
In Eq. (5.2), 2 moles of reactants combine to form 1 mole of transition-state

molecules, so ∆n = −1 which makes ∆H* = Ea − RT. That means for an ideal
gas at 25°C, ∆H* is smaller than Ea by about 2.5 kJ/mol. This is a relatively
small amount considering that Ea for most reactions are on the order of
tens of kJ/mol. This energy difference can also be formulated in terms of
the volume change associated with the formation of the activated complex.
For an ideal gas, the volume change (∆V*) associated with the formation of
a 1 molecule transition state from 2 molecules of reactants is approximately
2500 × 10−5 m3/mol. For aqueous solutions, which are featured in this chap-
ter, the activation volume for reactions seldom exceeds 25 × 10−5 m3/mol
(Drljaca et al., 1988), which is so much smaller than for the gas phase that
the difference between Ea and ∆H* for aqueous phase reactions is negligi-
ble. Because κ and γ are both approximately equal to one, they are typically
ignored and the terms in brackets in Eq. (5.15) are combined into the pre-
exponential constant, A.
 ∆S *  k T 
ln A =  r + ln  B   (5.17)
 R  h 
These adjustments result in the Arrhenius equation, which is used to correl-

ate rate constants over a range of temperatures.
Effect of temperature on rates 85
(a) 30
25
20
∆T, °C
10 °
15
0°
75 0°
5 °
25
10 Q10
0
20 40 60 80 100 120
Ea, kJ/mol
(b) 5
4 Figure 5.3. (a) The change

25° in temperature that is
needed to double the
reaction rate as a function
3 50° of the activation energy for
k/ko
the reaction. The contours

75° show that ∆T changes as
the temperature increases.
100°
(b) The relative change in
2 Q10
the rate constant caused
by a 10°C increase in
temperature as a function
of activation energy.
1 The contours show that
20 40 60 80 100 120 k/ko decreases as the
Ea, kJ/mol temperature increases.
 − Ea  1
ln k =  + ln A (5.18)
 R  T
The Arrhenius equation is frequently recast to log base 10.
 − Ea  1
log k =  + log A (5.19)
 2.303R  T
It can be converted into another convenient form by integrating its deriva-

tive between two temperatures.
k  − Ea  1 1
log  2  = − (5.20)
 k1  2.303R  T2 T1 
This equation is the basis for the Q10 rule of thumb, which states that for
many reactions the rate doubles for every 10°C increase in temperature. This
approximation seems to work well for biochemical reactions near room
temperature but is somewhat problematic for geochemical reactions that
occur over a wider range of temperatures and have a wider range of activa-
tion energies. The effect of temperature and activation energy on doubling
is shown in Figure 5.3a. Figure 5.3b shows how many times the rate will
increase for every 10°C increase in temperature. These figures show that the
Q10 rule applies for reactions that have an activation energy near 50 kJ/
mol and occur near 25°C. Note that for a reaction with a low activation
energy, the temperature must increase by a large amount to double the rate
(Figure 5.3a); and for a reaction with a high activation energy only a small
temperature increase will double the rate.
Example 5.2. Finding the activation energy for the ferric thiosulfate

reaction using the Arrhenius equation
Williamson and Rimstidt (1993) determined the rate of the ferric thiosulfate
reaction at temperatures ranging from 12° to 34°C. When those data are plotted
on a graph of log k versus 1/T (Figure 5.4) the equation for the best-fit straight
line is
log k = 23.9 − 6426/T
The activation energy for this reaction can be calculated from the slope of
this line.
E a = 2.303R(6426 ) = 123, 040 J/mol = 123 kJ/mol
The relatively large value of this activation energy is explained by the fact that
the two positively charged ions must have enough energy to overcome their elec-
trostatic repulsion. A postulated geometry for the activated complex is shown in
Figure 5.5. Note that the reactants approach each other in a way that brings the
negatively charged thiosulfate adjacent to the positively charged ferric iron ion
of the other reactant molecule and vice versa. This geometry keeps the highly
charged ferric iron ions separated as far as possible.
The very large activation energy of the ferric thiosulfate reaction means that
a 5°C temperature increase, from 25° to ~30°C, will cause the rate to double
(Figure 5.3).
Compensation effect 87
• •
• •
2.5
•
log k
2
•
1.5
•
1
0.0032 0.0033 0.0034 0.0035 0.0036
1/T, K–1
Figure 5.4. Arrhenius plot for the ferric thiosulfate reaction. The slope of the log k
versus 1/T graph is 6426 so the activation energy for the reaction is 123 kJ/mol.
(a) O (b) (c)

O Fe
Fe S S O O
Fe S S O
O O S S O
O O
O S S O
O O S S Fe
O
O S S Fe O Fe
O
Figure 5.5. Schematic diagram showing the steps in the ferric thiosulfate

reaction. (a) Two FeS2O3+ molecules approaching each other. (b) The
approximate geometry of the activated complex. (c) The reaction products,
which are two ferrous iron ions and a tetrathionate ion.
Compensation effect
Several studies have reported that for a series of related reactions the pre-
exponential in the Arrhenius equation correlates with the activation energy.
This correlation has a physical significance for homogeneous, gas phase
reactions but there is no theoretical justification for reactions involving con-
densed phases (Garn, 1975). Compensation effect models should be viewed
with caution because errors in real data are known to cause a correlation
between the slope and intercept of regression models (Liu and Guo, 2001;
Rimstidt et al., 2012). Comparing the log k values at 298 K to the activation
energy can minimize the correlation due to these errors (Ohlin et al., 2010).
Effect of ionic strength on rates (Brönsted–Bjerrum equation)

The quasi-equilibrium model provides a simple way to account for the
effect of ionic strength on the rates of aqueous reactions. In this model the
ions A and B combine to form an activated complex that breaks down to
products.
A + B = AB* → products (5.21)
At infinite dilution (I = 0) where the activity coefficient for the activated
complex equals one (and assuming that κ = 1) the rate constant for this
reaction is related to the equilibrium constant for the formation of the acti-
vated complex.
k T
ko =  B  K * (5.22)
 h 
If the rate equation is written in terms of concentration rather than activity,

the activity coefficients are incorporated into the rate constant so the rate
constant changes with changing ionic strength.
γ γ  k T γ γ 
k =  A B   B  K * =  A B  ko (5.23)

 γ AB*  h   γ AB* 
The Debye–Hückel limiting law relates the activity coefficients of all the spe-
cies, including the activated complex, to the ionic strength.
− log γ i = AZ 2 I (5.24)
Zi is the charge on the ion; I is the ionic strength of the solution; and A =
0.5092 (at 25°C). Equations (5.22) and (5.23) can be combined to create a
relationship between log k and ionic strength.
( 2
)
log k = log ko + ZA2 + ZB2 − (ZA + ZB ) A I (5.25)
ZA + ZB is the charge on the transition state. This equation can be

simplified.
log k = log ko + 2ZAZB A I = log ko + 1.02ZAZB I (5.26)
This equation predicts that a graph of log k versus √I will be a straight line
with a slope of 2AZAZB. At 25°C, 2A = 1.02 ≈ 1, so the slope of the line is
Effect of ionic strength on rates 89
approximately equal to the product of the charges on the A and B ions. This
relationship is not only an effective way to predict the effect of ionic strength
on rates but it can also be used to confirm the stoichiometry of postulated
transition states.
Example 5.3. Testing the postulated stoichiometry of the ferric

thiosulfate transition state using rate versus ionic strength data
In the previous two examples, the activated complex for the ferric thiosulfate
reaction was postulated to be [(FeS2O3)22+]*. This molecule is a combination of
two FeS2O3+ ions, each with a +1 charge, so the product ZA ZB is +1. If this pos-
tulated stoichiometry for the activated complex is correct, the slope of a graph
of log k versus √I (Figure 5.6) should be +1. The Williamson and Rimstidt
(1993) data for the rate as a function of ionic strength produces a straight line
with a slope of 0.928.
log k = 0.928 I + 2.17
Multiplying the slope of the line by 1.02 gives a value of 0.945 for the product of
the charges on the reacting ions, which is very close to the expected value of +1.
Note that this equation can be used to predict that log ko = 2.17. This test gives
further support to the postulated composition of the activated complex.
2.6
•
• •
2.5 27°C •
•
2.4
log k
2.3
2.2
2.1
2
0 0.1 0.2 0.3 0.4 0.5
I
Figure 5.6. The graph of log k versus √I for the ferric thiosulfate reaction. This
graph has a slope of 0.928 (≈ 1), which is consistent with each of the interacting
molecules having a charge of +1.
Effect of pressure on rates

In addition to the effects of temperature and ionic strength, rates are also
affected by changing pressure. By analogy to the effect of pressure on equi-
librium constants, the expected effect of pressure on rate constants (Asano
and le Nobel, 1978; Drljaca et al., 1988; Eckert, 1972) is related to the change
in volume that results from the conversion of the reactants to the activated
complex. This activation volume (∆V*) is the difference between the partial
molal volumes of the reactants and the transition state.
∆V * = VAB* − VA − VB (5.27)
van Dldik et al. (1989) have compiled a large amount of data pertaining to
the effect of pressure on reactions in aqueous media. For the reaction, A +
B = AB* → products, the rate constant change with pressure is proportional
to the volume of activation.
 ∂ ln k  ∆V *  ∂ log k  ∆V *
  = − or 
 ∂P  T  =− (5.28)
∂P T RT 2.303RT
Two kinds of volume of change occur when dissolved species react to form
an activated complex. The volume occupied by the species themselves
changes and the associated water molecules undergo rearrangements that
change their volume.
When Eq. (5.28) is integrated between 1 bar where the rate constant is
k1 and P where the rate constant is k2, log (k2/k1) is predicted to be a linear
function of pressure.
k  ∆V *
log  2  = − P (5.29)
 k1  2.303RT
Actual log k versus P relationships usually display some curvature over large
ranges of pressure so the log k values are often fitted to a parabolic function
of pressure.
log k = a + bP + cP 2 (5.30)
The activation volume is calculated from the slope of this fit at P ≈ 0 (usually
at 0.1 MPa)
 ∂ log k 
  =b (5.31)
∂P  T
A warning is appropriate here. If the rates are expressed using volume units
such as molarity, the compressibility of the solvent must be considered when
Effect of pressure on rates 91
calculating the activation volume. This problem can be avoided by using

pressure-independent units such as mole fraction or molality.
Example 5.4. The activation volume for Mn(II)-catalyzed SO2 oxidation
Huss Jr. et al. (1982) measured the rate of Mn(II)-catalyzed oxidation of SO2 by
dissolved oxygen at pressures up to about 140 MPa (1400 bars). A fit of log R
versus P for the data shown in Figure 5.7 gives the equation
log k = 7.76 × 10 −7 P 2 − 1.95 × 10 −3 P − 5.47 (5.32)
d log k
= 2 (7.76 × 10 −7 ) P − 1.95 × 10 −3 (5.33)
dP
The slope of this line when P = 0 is −1.95 × 10−3 MPa−1.
 m 3MPa   1 
∆V * = − (2.303)(298.15K )  8.314 × 10 −6   −1.95 × 10
−3


 K mol  MPa  (5.34)
m 3
∆V * = 1.11 × 10 −5
mol
–5.4
•
–5.5
•
•
•
log R
–5.6
•
•
–5.7
•
–5.8
0 25 50 75 100 125 150
P, MPa
Figure 5.7. The effect of pressure on the Mn-catalyzed rate of oxidation of

dissolved SO2. The slope of the log R versus P graph is used to calculate the
activation volume, ∆V*, for the reaction.
Electron transfer reactions

Oxidation–reduction reactions, which transfer electrons from a donor spe-
cies to an acceptor species, play a central role in many biological and geo-
logical processes. Unlike strongly coupled reactions where the reactants
form a structurally defined activated complex, as described by transition-
state theory, the species that participate in electron transfer reactions are
weakly coupled and retain their individuality. R.A. Marcus (Marcus, 1964,
1968, 1985) developed the foundation of electron transfer theory.
Electron transfer reactions involve five steps (Astruc, 1995).
1. The donor and acceptor species come together to form a precursor complex.
2. Thermal activation allows solvent molecules in the precursor complex to attain
an optimized arrangement for electron transfer to occur, i.e. an activated complex
forms.
3. An electron transfers from the donor to the acceptor to form a successor
complex.
4. The successor complex relaxes to a ground state.
5. The products separate by diffusion.
The overall rate of an electron transfer reaction is controlled by the rearrange-

ment of the solvent molecules in the precursor complex. This can be appre-
ciated by considering the self-exchange of an electron from a donor ferrous
iron ion to an acceptor ferric iron ion. The ions in self-exchange reactions
like this one can be distinguished using radioactive tracers (Silverman and
Dodson, 1952).
Fe*(H2O)62+ + Fe(H2O)63+ = Fe*(H2O)63+ + Fe(H2O)62+ (5.35)
For this reaction to occur, the water molecules coordinating the ferrous ion
at an average distance of 2.21Å must move closer to the ion because the aver-
age distance of water molecules coordinating a ferric iron ion is 2.05Å. The
electrostatic force that keeps each water molecule near the ion is constantly
perturbed by thermal vibrations that cause each water molecule to move
toward and away from the ion as if it was attached by a spring. According
to Hook’s law, the potential energy associated with each bond is a parabolic
function of the vibrational displacement, x.
1 2
Em = kx (5.36)
2
Figure 5.8 shows the parabolic relationships between the energy (E) of the
donor and acceptor species and the average displacement (x) of the solvent
molecules from their most stable configuration. The intersection of these
two parabolas is the energy barrier to the electron transfer step (i.e. free
energy of activation, ∆G*). The free energy of activation is a function of the
Photochemical reactions 93
Figure 5.8. Graph of the

energy of donor (D) and
acceptor (A) species,
Energy
D and their waters of

A hydration, as a function
of the displacement of the
water molecules along
Em
the reaction coordinate
∆G* (x). ∆G° is free energy
driving the reaction,
0
–∆G° shown here as a negative
value so the reaction is
spontaneous. ∆G* is the
energy barrier that must
be crossed for the electron
to transfer from the donor
xD x* xA
to the acceptor. Em is the
Reaction coordinate reorganization energy.
reorganization energy (Em or λ in many derivations, e.g. Marcus (2000)), and

the standard state free energy of reaction (∆G°) (Houston, 2001).
∆G * =
( E m + ∆G °)2 (5.37)
4E m
The rate constant for the electron exchange reaction is the product of a fre-
quency term and ∆G*.
kT  ( E m + ∆G °)2 
k= exp  −  (5.38)
h  4 E m kT 
This simplified discussion of electron transfer for outer sphere interactions

where the electron is transferred through solvent molecules is given to provide
a conceptual basis for understanding this kind of reaction. Many electron
transfer reactions are more complicated due to quantum mechanics effects
(electron tunneling) and inner sphere interactions (Astruc, 1995). Basolo
and Pearson (1967) give more details about electron transfer reactions.
Photochemical reactions
Solar energy delivers enormous amounts of energy to the Earth’s sur-
face, making photochemical reactions very important in geochemical and
biological processes. The global average solar energy incident at the top of
the atmosphere is 342 W/m2. Some 102 W/m2 is reflected back into space
and between 65 and 98 W/m2 is absorbed by the atmosphere (Arking, 1996).
That leaves between 142 and 175 W/m2 as the average solar energy flux at
the Earth’s surface, so that over a year sunlight delivers between 4.5 and
5.5 million kJ/m2 of energy to the Earth’s surface. Much of that radiant
energy is converted to heat but a significant fraction drives photochemical
reactions, the most important of which is photosynthesis. Many other pho-
tochemical reactions are important to geochemistry. For example, photo-
oxidation of Fe2+ (Southworth, 1995) and Mn2+ (Nico et al., 2002) exert an
important control on the redox chemistry of stream water. Most of those
photochemical reactions occur in surface waters and in films of water on
mineral surfaces.
Planck’s law states that the amount of energy that each photon carries
depends upon its frequency (ν, sec−1), multiplied by Planck’s constant (6.63
× 10−34 J/sec) and the frequency depends on the speed of light (c, 2.99 × 108
m/sec) and the wavelength (λ, m).
hc
E ( J/photon ) = hν = (5.39)
λ
This means that light with shorter wavelengths delivers a larger amount of
energy per photon. Blue green light (λ = 480 to 500 nm), the most intense
sunlight at the Earth’s surface, delivers 240 to 250 kJ per Einstein. An
Einstein is 1 mole of photons (NA = 6.02 × 1023) regardless of wavelength.
N Ahc
E ( J/Einstein ) = (5.40)
λ
Figure 5.9 shows the relative intensity of sunlight and the amount of
energy delivered by one Einstein of light as a function of wavelength.
When a molecule or ion absorbs a photon, that photon’s energy can be
dissipated in several different ways, but one way is for that energy to cause
a chemical reaction to occur. The first law of photochemistry is that a com-
pound must absorb light for a photochemical reaction to occur (Grotthuss–
Draper law). The second law of photochemistry is that each photon that
is absorbed activates only one molecule for a subsequent reaction (Stark–
Einstein law). The quantum yield (φ) is defined as the number of molecules
that react divided by the number of photons absorbed. It can also be defined
in terms of moles.
number of moles that photoreact

φ= (5.41)
bed
number of Einsteins absorb
Radiolysis 95
(a) relative intensity
ultraviolet
infrared
orange
yellow
green
violet
blue
red
400
(b)
energy, kJ/E
300
200
Figure 5.9. (a) Graph of the
relative intensity of solar
radiation at the Earth’s
surface. (b) Graph of the
100 energy contained in 1 mole
800 700 600 500 400 300 of photons as a function of
wavelength, nm wavelength of light.
Leifer (1988) provides useful theoretical and practical knowledge about

aquatic photochemistry.
Radiolysis
Radiolysis reactions are brought about by the absorption of ionizing radi-
ation. The radiation may belong to the light group, i.e. X-rays, γ-rays, elec-
trons–positrons, muons; or the heavy group, i.e. protons, α-particles, fission
fragments. When these particles pass through a medium, some of their
energy is transferred to the molecules of that medium raising their electronic
states. This can lead to neutral dissociation or to ionization if the energy
transfer is sufficiently large.
Radiolysis of water produces hydroxyl free radicals (•OH) (Le Caër, 2011)
along with hydrated electrons and •H atoms. Subsequent reactions produce
H2O2, H2, O2, and, depending upon the solution composition, various other
species. Because the H2 can rapidly diffuse away from the site of genera-
tion, radiolysis reactions often alter the local redox conditions. For example,
uranium deposits frequently show post-depositional oxidation because of

radiolysis (Debessy et al., 1988; Derome et al., 2003; Smetannikov, 2011).
Free radicals and hydrogen peroxide produced by the radiolysis of water
in the subsurface can oxidize pyrite in the absence of O2 (Lefticariu et al.,
2010). Products of water radiolysis increase the leaching rate of basalt
(Yoneawa et al., 1996). Radiolysis may have been the first important source
of atmospheric O2 (Draganic, 2005) and Lin et al. (2005) postulated that the
natural radiolysis of water could produce enough H2 to sustain microbial
communities in the deep subsurface.
The chemical effect of radiolysis is controlled by a combination of
absorbed dose and the radiation chemical yield value (G). Absorbed dose
is measured in grays (1 Gy = 1 J/kg) and the G value indicates how many
molecules of a species are produced per 100 ev (1 ev = 1.602 × 10−19 J) of
absorbed energy. Ershov and Gordeev (2008) provide a model for the yield
of H2, H2O2, and O2 from the radiolysis of water and aqueous solutions and
Spinks and Woods (1990) give details about the connection between reac-
tion kinetics and radiolysis processes.
Free radicals
Free radicals are chemical species that have one or more unpaired electrons.
They can have a positive, negative, or zero charge. Although a few radi-
cals are stable (e.g. O2 and NO) or highly persistent, most are highly reac-
tive. Thermally activated, electron transfer, photochemical, and radiolysis
processes can all produce free radicals. Free radicals are common reaction
intermediates in organic and biochemical reactions and they are no doubt
important in many organic geochemical processes (Dominé et al., 2002).
Free radicals play important roles in many inorganic geochemical proc-
esses. For example, hydroxyl free radicals are produced during pyrite oxida-
tion by O2. Presumably they form as the result of a modified Fenton reaction
(Rimstidt and Vaughan, 2003). Free radicals are further involved as various
sulfoxy species are oxidized to sulfate (Druschel et al., 2003). Oxidation of Fe2+
is another example of a reaction that involves free radical intermediates.
Breaking Si-O bonds by fracturing quartz produces O3–Si• and Si–O• free
radicals that can subsequently react with water to produce •OH radicals
(Narayanasamy and Kubicki, 2005). The •OH species is quite toxic to cells
and may be responsible for the lung scarring that is symptomatic of silicosis.
•OH radicals can also alter DNA and may be responsible for the develop-
ment of lung cancer (Fubini, 1998).
Free radicals are studied by generating them using radiolysis. These stud-
ies have produced extensive tabulations of rate constants for free radical
reaction rates (Bielski et al., 1985; Buxton et al., 1988; Buxton et al., 1995;
Neta et al., 1988).
Chemical bonding and reaction mechanisms 97
Reaction mechanisms
A reaction mechanism is a model that explains, elementary step by elemen-
tary step, how reactants become products. Reaction mechanism models also
explain how the rate of each elementary step is influenced by the structure
of its transition state. Because several different reaction paths may be possi-
ble for a particular reaction, there is no way to a priori determine a reaction
mechanism. Transition states cannot be observed directly because they are
present at very low concentrations and have a fleeting existence. This means
that reaction mechanisms must be inferred using a combination of forward
models based on fundamental physical principles and reverse models cal-
ibrated by the outcomes of experimental tests. Detailed studies of many
kinds of reactions have generated some guiding principles for working out
reaction mechanisms (Ašperger, 2003; Moore and Pearson, 1981). Casey
(2001) and Casey and Swaddle (2003) give some guidelines that are specific
to the dissolution of oxide and hydroxide compounds, which are common
in geochemical environments.
Chemical bonding and reaction mechanisms

Chemical reactions occur when electrons move from the set of chemical
bonds that comprise the reactants to form a new set of chemical bonds that
comprise the products. This means that the nature of chemical bonding
must be understood in order to construct models of reaction mechanisms.
Even though a rigorous definition of a chemical bond does not exist,
the idea that electrostatic forces hold together the atoms in a molecule is
well established. When atoms approach closely enough to form a bond, the
valence electrons are attracted by the positive charge in the space between
the nuclei and they accumulate between the atoms. The buildup of negative
charge between the atoms attracts the atoms’ nuclei causing them to become
bonded. G.N. Lewis developed a simple model whereby a chemical bond is
regarded as a pair of electrons, with opposite spins, residing between the
bonded nuclei (Lewis, 1916). This simple idea has far-ranging implications
(Jensen, 1980) and formed the foundation for many models of chemical
structures developed throughout the twentieth century. Valence bond the-
ory, which grew out of the Lewis model, is the basis of Linus Pauling’s fam-
ous book The Nature of the Chemical Bond (Pauling, 1960). Valence bond
theory assumes that electrons are highly localized in chemical bonds as well
as in non-bonded sites as lone pairs. A contrasting model, molecular orbital
theory, assumes that electrons are delocalized into molecular orbitals, which
have geometries that are described by quantum mechanics (Gimarc, 1974).
Both models are useful and complementary (Gillespie and Robinson, 2006).
Quantum mechanics models of electron distribution in molecules now pro-
vide a powerful visualization of the localized bond and lone pair electron
distributions in the form of the electron localization function (ELF). ELF

maps for a number of important earth materials are shown in Gibbs et al.
(2005).
Lewis acid–base theory (Jensen, 1980) is an outgrowth of the Lewis
model of chemical bonds. A Lewis acid is a chemical species that can accept
an electron pair. Lewis acids can be cations like Fe3+ or Cu2+ or they can be
species with empty or partially empty valence orbitals such as CO2 or SO2.
Lewis bases can donate an electron pair. Lewis bases are anions like OH− or
S2− or they can be species with lone pairs such as H2O or NH3. The trans-
fer of cations from a solid, such as szomolnokite (FeSO4•H2O), to form a
hydrated ferrous ion in solution is a typical Lewis acid–base reaction.
FeSO4•H2O+ 5 H2O = Fe(H2O)62+ + SO42− (aq)
In this reaction, ferrous iron, a Lewis acid, reacts with the lone pair electrons
on water molecules, which are Lewis bases, to form a hydrated cation, a
Lewis adduct. Although the sulfate ion also interacts with water molecules,
this interaction is much weaker than the Lewis acid–base interaction of the
ferrous iron. Many, perhaps all, chemical reactions can be viewed as Lewis
acid–base interactions, making this theory very useful for developing mod-
els of reaction mechanisms.
Organic chemists refer to Lewis acids as electrophiles because they are
attracted to electron-rich sites on other molecules, and they refer to Lewis
bases as nucleophiles because they are attracted to electron-deficient molecu-
lar sites. Ingold (1969) developed a classification of reaction mechanisms
based on this idea and this classification is the foundation for modeling the
mechanisms of organic reactions (Bruckner, 2002; Grossman, 1999). Casey
(2001) and Casey and Swaddle (2003) adapted some of these principles to
apply to the dissolution of oxides. The electron-rich, and therefore nucleo-
philic, sites on molecules are nicely visualized using the electron localization
function (Gibbs et al., 2005).
Donor–acceptor theory (Gutmann, 1978) further extends the Lewis bond-
ing theory to explain how bonds in reacting molecules lengthen (become
weaker) or shorten (become stronger) as a result of the Lewis base (electron
donor) and Lewis acid (electron acceptor) interactions. Gutmann proposed
three rules for predicting bond length variations during donor–acceptor
interactions:
1. As donor and acceptor atoms approach each other, the induced rearrangement
of electrons in both the donor and the acceptor causes the adjacent bonds in each
to lengthen.
2. Donor–acceptor interactions cause electron density shifts throughout the entire
molecule and those shifts produce changes in bond lengths.
3. As the coordination number of an atom in an adduct increases so do the bond
lengths of any bonds originating from the coordination center.
Chemical bonding and reaction mechanisms 99
Example 5.5. Hydrolysis of a Si–O bond
The donor–acceptor rules are consistent with quantum mechanical models that
show how electrons in a bond respond as a donor molecule approaches the elec-
trophilic site on an acceptor molecule. A well-known quantum mechanics model
by Lasaga and Gibbs (1990) illustrates the attack of a H2O molecule on a Si–O
bond in a H6Si2O5 molecule. Their model shows that the electron pair on the
water molecule’s oxygen atom mounts a nucleophilic attack on one of the Si
atoms leading to a 5-coordinated Si atom. At the same time, one of the water
molecule’s hydrogen atoms mounts an electrophilic attack on the bridging oxy-
gen in the H6Si2O5 molecule. The simultaneous alignment of the oxygen end of
the water molecule with the Si atom and the electrophilic attack of the hydrogen
atom on the bridging oxygen atom creates the geometry of the transition state
(Figure 5.10). Once that transition state has formed, the curly arrows show the
shifts in electron density from preexisting bonds to the developing bonds. This
shift divides the H6Si2O5 molecule into two H4SiO4 molecules.
(a) (b)
(c)
Figure 5.10. Schematic illustration showing how nucleophilic and electrophilic interactions cause (a) a water
molecule to align with a Si–O bond to form (b) a transition state that results in electron rearrangement,
which disassociates the Si–O bond and forms (c) two Si–OH moieties.
Rate equations and reaction mechanisms

A postulated reaction mechanism must be consistent with the rate data as
well as the general principles of chemistry and physics described in this chap-
ter. This congruency between the rate equation, general physicochemical
principles, and the reaction mechanism is a foundation of chemical kinetics.
Although many different guidelines might be formulated for creating this
congruency, the ones listed below are often useful (Brezonik, 1994; Edwards
et al., 1968).
1. For elementary reactions, the composition of the transition state is reflected by
the rate equation so that the number of molecules of each species found in the
transition state is given by the reaction order for that species, i.e. the molecu-
larity of the transition state is inferred from the reaction order for each species
found in the rate equation. If the rate law is r = kmAa mBb , we infer that the reaction
is aA + bB → [AaBb]*→ products.
2. Solvent molecules may participate in the activated complex but unless the con-
centration (activity) of the solvent can be varied widely this situation is difficult
to recognize. Therefore, the activity of the solvent is usually set to one in the rate
law. For the reaction aA + bH2O = cC, the rate law would be r = kmAaaHb 2O = kmAa .
Note that the composition of the solvent can affect the rate even if solvent mol-
ecules do not participate in the transition state (Amis, 1966).
3. Most reactions are bimolecular and only a few are either monomolecular or ter-
molecular. When the overall reaction order is greater than three, it is likely that
it is a chain reaction with one or more equilibrated intermediates prior to the
rate-determining step. For the reactions 2A + 2B = 2AB (fast) and 2AB → 2C
(slow), the empirical rate equation might be written as r = kmA2 mB2 and appear to
be an overall fourth-order reaction but it would be more meaningful to write the
rate law in terms of the intermediate AB, i.e. r = kmAB 2
.
4. Rate equations that display non-integer (fractional) orders typically arise from
chain (multistep) reactions, often involving free radical intermediates.
5. Rate equations that display non-integer (fractional) orders sometimes arise from
rapid equilibria (sometimes competitive) prior to the rate-determining step.
6. The rate-determining step for the forward reaction sequence must be the same as
the rate-determining step for the reverse reaction sequence. This is required by
the principle of microscopic reversibility.
7. All intermediates produced in elementary steps must be consumed in subsequent
steps otherwise they will show up as reaction products.
8. If the stoichiometric coefficient for species i is greater than the reaction order for
i, one or more reaction steps occur after the rate-determining step.
9. If the reaction order increases as the concentration of one (or more) of the reac-
tants increases, the overall reaction may consist of two or more parallel paths.
10. Highly reactive intermediates, which typically remain at a low concentration,
are more likely to react with stable species, which occur at high concentrations,
rather than reacting with each other.
Catalysts and inhibitors 101
11. Elementary reactions must be feasible from a bond energy and geometry stand-
point. Extensive bond and atom rearrangement does not occur in a single
elementary step.
12. The charge on the transition state of the rate-determining step can be deduced
from the effect of ionic strength on the rate.
Catalysts and inhibitors

The previous section describes several strategies for interpreting experimen-
tal results to infer the nature of a reaction mechanism. Identifying species
that catalyze or inhibit the reaction rate can provide additional valuable
evidence about the reaction. A catalyst is a species that participates in the
reaction and increases the reaction rate but emerges unchanged after the
reaction is complete. For example, Dove (1994) has shown that the presence
of Na+ ions in solution increases the rate of quartz dissolution but the Na+
is not consumed by the reaction. A catalyst lowers the activation energy for
the reaction’s rate-determining step. An inhibitor is a negative catalyst so it
is a species that participates in a reaction and decreases its rate without being
consumed by the reaction. Catalysts and inhibitors are very useful chemical
probes that can be used to test hypothesized reaction mechanisms.
Example 5.6. Cu2+ catalysis of the ferric thiosulfate reaction
The model of the ferric thiosulfate reaction mechanism developed in Examples 5.3,

5.4, and 5.5 can be further tested using catalysts and inhibitors. The slow rate and
high activation energy of this reaction is the result of the work required to bring
the two positively charged FeS2O3+ species together to form the transition state.
This reaction would be much faster if the reacting species had zero charge.
Adding a small amount of Cu2+ to the reaction dramatically increases the rate.
The Cu2+ forms a thiosulfate complex and because it is uncharged the CuS2O30
species can react more easily with a FeS2O3+ so that the solution contains a
higher concentration of activated complexes.
CuS2O30 + FeS2O3+ = (CuS2O3•FeS2O3)+ → Cu+ + Fe2+ + S4O62 (5.42)
The Cu+ is rapidly re-oxidized by a very fast reaction with ferric iron, which
converts it back to Cu2+.
Cu+ + Fe3+ = Cu2+ + Fe2+ (5.43)
This meets the definition of a catalyst, which is a species or site that lowers the
activation energy of the reaction and is not consumed by the reaction.
The ferric thiosulfate reaction can be inhibited by the addition of Ag+, which
forms a very strong complex with thiosulfate and that complex is unreactive
because Ag+ cannot be reduced to Ag0 by reaction with thiosulfate or oxidized
by reaction with Fe3+.
Chapter 6
Surface kinetics
Many important reactions happen at interfaces. Because reactions between

aqueous species and mineral surfaces are so important in low temperature
geochemistry, this chapter focuses on reactions at solid/solution interfaces.
Reaction rates at solid/solution interfaces are controlled by the area of
the interface as well as by the chemical and physical conditions that occur
there. Surface reactions are approximately confined to a two-dimensional
region, so their rates are expressed in terms of how fast species are created
per unit of surface area, and this means that the rates have units of flux
(J, mol/m2sec). The flux notation (J) and terminology is used throughout
this book. The environment at the solid/solution interface is a hybrid of the
bulk solid and bulk solution, so models of the chemical and physical condi-
tions controlling the reaction rates must account for this transitional char-
acter. Equilibrium thermodynamics provides a powerful starting point for
constraining the surface conditions. At equilibrium the chemical potential
of each component must be the same throughout the system, so the chem-
ical potential of the components in the surface are equal to their chemical
potentials in the solid and solution phases. At low temperatures, the slow
rate of equilibration between the bulk solid and the surface may void this
requirement for the solid but it should apply for the components in the bulk
solution. Also, at equilibrium the principle of detailed balance requires that
the rates of forward reactions in the interface must equal the rates of the
reverse reactions. In addition, the forward and reverse reaction steps must
be the same. Models of reaction rates at equilibrium are well constrained by
these principles but as the system departs from equilibrium these require-
ments fall away and we must search for other principles to model interfacial
reaction rates.
Creating a new interfacial area requires work to be done and the amount
of work needed to create a unit area of new surface is the surface free energy
(σ, J/m2).
 dG 
σ = (6.1)
 dA  T , P ,etc.
102
Geometric surface area models 103
The surface free energy term is always positive; this means that the equi-
librium surface should acquire a geometry that minimizes its surface free
energy. An atomically smooth sphere has the lowest overall surface free
energy for solids with isotropic bonding. For example, amorphous silica
tends to develop a spherical form. For crystalline solids, the faces with the
highest atomic packing densities and shortest bond lengths have the lowest
σ values. These “low index” faces tend to occur more frequently in crystal
forms, although the growing crystals are not at equilibrium with the sur-
rounding solution so most crystals actually display growth forms rather
than true equilibrium forms. A crystal that is at perfect equilibrium with
the surrounding solution would have atomically smooth faces, but actual
crystals usually display growth hummocks and other surface irregularities
related to the growth process.
The value of σ for a crystal face relative to the other crystal faces can be
found using Wulff’s theorem, which states that there exists an interior point
such that the surface free energy of a crystal face is proportional to the perpen-
dicular d
istance from that point to the crystal face (Wolff and Gualtieri, 1962).
Geometric surface area models

Because reaction rates between minerals and solutions are directly propor-
tional to the interfacial area between the phases, it is necessary to quantify
this area for rate models. There are various methods to measure the inter-
facial area, but a useful first step for model building is to develop idealized
reference models for reacting surfaces. Idealized surface area models neglect
important surface features in a trade-off for simplicity. As such they provide
a handy approximation of the relationship between surface geometry and
reaction rates.
As a first approximation, the surface areas of solid particles can be mod-
eled as if the grains are smooth spheres with an effective diameter equal
to the real grain’s smallest dimension. There are several methods to deter-
mine particle size, including sieving, sedimentation rates, light scattering,
and optical measurements. All these methods will find a range of grain sizes.
If this range is large, the surface area models must account for the particle
size distribution. If the range is small, a weighted average of the maximum
diameter (Dmax, m) and minimum diameter (Dmin, m) is used to determine
an effective diameter (De, m). The averaging method of Tester et al. (1994),
which assumes a flat particle size distribution between the maximum and
minimum diameters, seems to work well.
Dmax − Dmin
De = (6.2)
 Dmax 
ln 
 Dmin 
104 Surface kinetics
The effective diameter given by this model is very close to the arithmetic
average of the maximum and minimum grain diameter for a narrow grain
size range, but for a wider size range it is lower because the smaller particles
contribute more surface area.
The simplest model that relates the specific surface area of grains to their
diameter assumes that the grains are spherical. This model expresses the
specific geometric surface area (Ageo, m2/g) as a function of the molar vol-
ume (Vm, m3/mol) of the substance, the diameter of the grains (De, m), and
the molecular weight of the substance (Wm, g/mol).
6Vm  m 2 
Ageo = (6.3)
DWm  g 
Equation (6.3) can be transformed to give the specific surface area as a func-
tion of the density (ρ, g/cm3) and diameter (D, m) of the grains.
6 × 10 −6  m 2 
Ageo = (6.4)
Dρ  g 
Example 6.1. Geometric specific surface area of quartz grains
If a sample of quartz sand is sieved to recover the 1 to 2 mm diameter size frac-
tion, the effective diameter of the grains in the sample can be found using Eq.
(6.2).
2 −1
De = = 1.44 mm = 1.44 × 10 −3 m
 2
ln  
 1
The density of quartz is 2.66 g/cm3, so the geometric specific surface area of the
sample is found using Eq. (6.4).
6 × 10 −6  m2 
Ageo = = 0.00157 
(1.44 × 10 )(2.66)
−3
 g 
Surface site density

Each surface reaction takes place at a discrete site where an attacking reagent
docks and participates in an elementary reaction. The exact nature of the
site varies from case to case but the overall reaction rate is proportional to
the number of docking sites. If these sites are nearly homogeneously distrib-
uted over the surface, the number of reaction sites, and therefore the overall
Surface site density 105
reaction rate, is directly proportional to the surface area. The number of

potential sites per unit area is approximately equal to the number of formula
units that crop out on the surface. The area occupied by one formula unit
(Afu, m2/fu) is a function of the molar volume (m3/mol) of the substance and
Avogadro’s number (NA).
2
V  3
Afu =  m  (6.5)
 NA 
The area occupied by 1 mole of formula units (Amol, m2/mol) is the area per
formula unit multiplied by Avogadro’s number.
Amol = Afu N A (6.6)
The formula unit density (Dfu, mol/m2) is the reciprocal of Amol.
1
Dfu = (6.7)
Amol
For a surface that is dissolving homogeneously and congruently, the average

lifetime of a single formula unit (tfu, sec) is the reciprocal of the dissolution
flux (J, mol/m2sec) and the area occupied by 1 mole of formula units.
1
t fu = (6.8)
JAmol
Example 6.2. Average lifetime of a Mg2SiO4 unit on a dissolving for-

sterite surface
Forsterite (Mg2SiO4) has a molar volume of 4.365 × 10−5 m3/mol. The average
area occupied by one formula unit of forsterite is found using Eq. (6.5).
2
 m3  3
4.365 × 10 −5
 mol  m2
Afu =  = 1.738 × 10 −19 (6.9)
fu  fu
 6.023 × 10 23 
 mol 
The surface area occupied by 1 mole of forsterite is found using Eq. (6.6).
 m2   fu  m2
Amol = 1.738 × 10 −19   6.023 × 10 23  = 1.047 × 105 (6.10)
 fu   mol  mol
The site density is found using Eq. (6.7).

1 mol
Dfu = = 9.553 × 10 −6 2 (6.11)
1.047 × 105 m
The dissolution rate of forsterite at 25°C and pH 3 is 1.99 × 10−8 mol/m2sec so

the average lifetime of a surface formula unit is found using Eq. (6.8).
1
t fu = = 480 sec ≈ 8 min (6.12)
 mol   m2 
 1.99 × 10 −8
 1.047 × 10 5

m 2sec   mol 
Surface area/volume models

Rate models that express the rate of change of concentration (R, molal/sec)
due to the dissolution or precipitation of a solid must account for the sur-
face area to volume (A/V) or surface area to mass of solution (A/M) ratio of
the reacting system. Table 6.1 gives A/V and A/M ratios for some common
geometries.
Example 6.3. A/M ratio for packed beds
The porosity of packed beds ranges from about 50% for loosely packed grains
to about 25% for close packed grains. The A/M for a packed bed of 1 mm diam-
eter grains with 30% porosity filled with a solution with a density of 1 g/cm3 is
about 14 m2/kg.
A 1− φ  6   0.7   6  m2
M =  φ   1000 ρD  =  0.3   (1000 )(0.001) 
= 14
kg
Figure 6.1 shows how the A/M ratios for packed beds filled with a solution with a
density near 1 g/cm3 varies with grain diameter and porosity. Real grains are not
perfect spheres so these values should be increased by 5 to 10 times to account
for their surface roughness.
Reactions at surfaces
Surface reactions comprise four steps.
1. Adsorption of a fluid phase species (A) to a surface site (S): A(aq or v) +
S = A…S
2. Surface diffusion of the reacting species to a reactive site (S#): A…S → A…S#
3. Reaction between the reactive site and the adsorbed species to produce a product
species (B): A…S# = A–S#→ B–S
Adsorption (Langmuir model) 107
Table 6.1. A/V and A/M ratios for some typical solid/solution geometries.
The A/V ratios are used for models where the solution concentrations are
expressed in volume units. The A/M ratios are used in models where the
solution concentrations are expressed in mass units. The units for density, ρ,
are g/cm3.
Geometry A/V (m2/m3) and A/M (m2/kg) ratio
Cylindrical pipe A = 2R and A = 2
radius (R, m) V M 1000 ρR
Rectangular pipe A = 2 ( H + W ) and A = 2 ( H + W )
height (H, m), width (W, m) V HW M 1000 ρHW
Fracture with parallel walls A = 2 and A = 2
width (W, m) V W M 1000 ρW
Packed bed of spheres
A =  1 − φ   6  and A =  1 − φ   6 
diameter (D, m), porosity (φ) V  φ   D  M  φ   1000 ρD 
4. Desorption of the product species: B–S = B…S → B(aq or v) + S
Any of these steps can be rate limiting.
Adsorption (Langmuir model)

The adsorption step is often modeled using the Langmuir isotherm, which
is developed from two primary assumptions: (1) only monolayer adsorp-
tion occurs and (2) the sorption enthalpy is the same for all sites regardless
of the amount of surface coverage. This model was originally derived for
adsorption of a gas onto a solid surface but it is developed here to model
adsorption of a solution species to a solid surface. For the sorption process,
the rate of change of fractional surface coverage (θ = Γa/Γ, no units), which
is the ratio of the surface concentration of occupied sites (Γa, mol/m2) to the
surface concentration of all adsorption sites (Γ, mol/m2), is directly propor-
tional to the adsorption rate constant (k+, molal/sec), the concentration of
adsorbate (m, molal), the number of surface sites, and the proportion of the
surface that is not covered by adsorbate (1 − θ). The rate of desorption is
directly proportional to the desorption rate constant (k−, sec−1), the number
of surface sites, and the fraction of those sites that are covered by adsor-
bate. The overall rate is the difference between the adsorption rate and the
desorption rate.
(a) 1500
1000
A/M, m2/kg
500
= 0.25
= 0. 5
0
0 0.0005 0.001
grain diameter, m
(b) 3000
1000
= 0.25
A/M, m2/kg
100
= 0.5
Figure 6.1. (a) A/M ratio
for packed beds filled with
a solution with a density
of 1.0 g/cm3 as a function 10
of grain diameter and bed
porosity. (b) Same data
presented on a graph with 3
logarithmic scales. 0.00001 0.0001 0.001
grain diameter, m
dθ
= k+ mΓ (1 − θ ) − k− Γθ (6.13)
dt
At equilibrium, the principle of detailed balance requires that the adsorp-

tion rate equals the desorption rate (dθ/dt = 0).
k+ mΓ (1 − θ ) = k− Γθ (6.14)
Equation (6.14) can be rearranged to define an apparent equilibrium con-

stant (Kb, molal) for the sorption reaction.
Adsorption (Langmuir model) 109
(6.15)
k+ Γ θ
m =
k− Γ 1 − θ

 θ 
Kbm =  (6.16)
 1 − θ 
The fractional surface coverage is found by solving Eq. (6.16) for θ.
Kbm
θ= (6.17)
1 + Kbm
If θ is decomposed into Γa and Γ, Eq. (6.17) can be written as an isotherm

that expresses surface concentration in terms of solution concentration.
Kbm
Γa = Γ (6.18)
1 + Kbm
Figure 6.2 illustrates the relationship between the surface concentration

and the solution concentration for some values of Kb.
Isotherm experiments, which determine Γa for a wide range of m, can
be analyzed to find Kb and Γt. Equation (6.18) can be linearized in several
different ways to avoid the need for using a nonlinear regression for data
analysis. The double reciprocal of the Langmuir equation produces the
Lineweaver–Burke equation, which is often used to fit adsorption data.
1 1 1
= + (6.19)
Γ a Γ ΓK b m
A graph of 1/Γa versus 1/m has a slope of 1/ΓKb and an intercept of 1/Γ.
This fit gives excessive weight to data in the low concentration range and is
sensitive to data error. The Scatchard equation gives more weight to data in
the high concentration range.
Γa
= ΓK b + K b Γ a (6.20)
m
According to this equation, a graph of Γa/m versus Γa has a slope of Kb and

an intercept of ΓKb. Another linear form proposed by Langmuir is less sen-
sitive to data error and gives more weight to the middle and high concen-
tration ranges.
m m 1
= + (6.21)
Γ a Γ ΓK b
A graph of m/Γa versus m has a slope of 1/Γ and an intercept of 1/ΓKb.

1× 10–5
30
8× 10–6
10
6× 10–6
, mol/m2 3
4× 10–6
a
Figure 6.2. Langmuir 2× 10–6

sorption isotherms 0. 3
contoured for values of
Kb ranging from 0.3 to 0
30. Γ is 1 × 10−5 mol/m2 0 0.2 0.4 0.6 0.8 1
for each isotherm. m, mol/kg
Distribution coefficient
The distribution coefficient is a simplification of the Langmuir equation,
Eq. (6.16). For small amounts of surface coverage, the 1 − θ term in that
equation is approximately equal to 1, so it simplifies to a linear relationship
between the fractional surface coverage and the solution concentration.
Kbm = θ (6.22)
This equation can be rearranged and combined with the definition of

fractional surface coverage (θ = Γa/Γ) to find the surface concentration (Γa,
mol/m2).
Γ a = Γθ = ΓK b m (6.23)
The specific amount adsorbed (nsp, mol/g-solid) is the surface concentration

(Γa, mol/m2) multiplied by the specific surface area of the solid (Asp, m2/g-
solid).
nsp = K b mΓAsp (6.24)
The distribution coefficient (KD, (mol/g-solid)/molal) is the specific amount

sorbed divided by the solution concentration.
nsp
KD = (6.25)
m
Surface catalysis: unimolecular decomposition 111
K b mΓAsp
KD = = K b ΓAsp (6.26)
m
In this derivation, Γa and Γt are multiplied by the specific surface area of the
solid in order to convert the concentration units from mol/m2 to mol/g-solid.
Defining distribution coefficients this way makes them functions of the spe-
cific surface area so KD values vary with grain size.
BET surface area
Reaction rates between fluid species and solid surfaces are often normalized
using a surface area value determined by the BET (Brunauer–Emmett–Teller)
method. The theory behind this method was developed by Brunauer et al.
(1938) and is thoroughly described and evaluated in Lowell and Shields
(1991). This model is based on the idea that as the pressure and temperature
of a gas approaches the pressure and temperature where liquid and vapor are
in equilibrium, the molecules will deposit on the surface as a thin, multilayer
film. In most cases N2 is used as the gas and a powdered sample is placed into
an evacuated chamber that is immersed into boiling liquid nitrogen to fix the
temperature. Increasing amounts of N2 gas are introduced into the chamber
and the amount adsorbed at pressures ranging from 5 to 35 kPa is measured.
Usually three to five data points are determined. These data are fit to the BET
isotherm and the amount of gas that comprises a monolayer on the solid sur-
face is computed from the slope and intercept of the isotherm. The surface
area is then calculated assuming a sorption cross-section for an adsorbed N2
molecule. Prior to the BET surface area determination, the surface of the sam-
ple is usually “cleaned” by heating in a vacuum, often to temperatures near
300°C. This treatment has little effect on most minerals but can drastically
alter hydrous phases and sulfide minerals that decompose releasing water or
sulfur vapor into the vacuum system. BET-determined surface areas are usu-
ally five to ten times larger than surface areas calculated using Eq. (6.3). For
example, White and Peterson (1990) report that the average BET surface area
for a relatively large number of samples is about seven times the surface area
estimated from the geometric surface area estimated using Eq. (6.4).
Surface catalysis: unimolecular decomposition

Surfaces often catalyze chemical reactions. The simplest model of surface
catalysis of a decomposition reaction involves two steps. First the reactant
adsorbs to a surface to form a surface complex and then the surface com-
plex decomposes to products.

A + S← 
→ AS k
k+1
+2
→ products (6.27)
k−1
After a short time, the rate that A adsorbs to the surface equals the rate
that AS breaks down to products. The rate of disappearance of A can be
expressed in terms of the rate of conversion of AS to products.
dmA
= k+2 Γ AS = k+2θΓ (6.28)
dt
Once the reaction reaches steady state, the rate of change of the AS concen-
tration is zero.
d Γ AS
= 0 = k+1AΓ (1 − θ ) − k−1Γθ − k+2 Γ (6.29)
dt
This equation can be rearranged to find the fraction of surface sites covered
by adsorbed A.
k+1mA
θ= (6.30)
k+1mA + k−1 + k+2
Equation (6.30) can be combined with Eq. (6.28) to find the rate of
consumption of A.
dmA k+1k+2 ΓmA

= (6.31)
dt k+1mA + k−1 + k+2
This result is analogous to the Michaelis–Menton equation that is used to

describe enzyme-catalyzed decomposition rates.
If k+2 >> k+1mA and k+2 >> k−1, the rate-limiting step is adsorption.
dmA k+1k+2 ΓmA

= = k+1ΓmA (6.32)
dt small + small + k+2
If k+2 << k+1mA and k+2 << k−1, the fraction of surface coverage is described
by the Langmuir isotherm.
k+1mA
θ= (6.33)
k+1mA + k−1
This means that the rate depends upon the concentration of AS.
dmA K b k+2 ΓmA

= (6.34)
dt K b mA + 1
Surface catalysis: unimolecular decomposition 113
When mA is small, so the fraction of surface coverage is small, the rate is

first order in A.
dmA
= K b k+2 ΓmA (6.35)
dt
When mA is large, so the fraction of surface coverage is large, the rate is zer-
oth order in A.
dmA
= k+2 Γ (6.36)
dt
Example 6.4. Oxidation rate of UO2
In a detailed study of the dissolution rate of synthetic UO2, de Pablo et al. (1999)
found that instead of being linear, as is typical of many mineral dissolution reac-
tions, the log J versus log MHCO3 graph shows a distinct curvature. They inter-
preted this to mean that bicarbonate ions sorb to a surface site as a dissolution
reaction step and when bicarbonate concentrations are high these sites become
nearly saturated with adsorbed bicarbonate. This means that the reaction order
transitions from first order at low bicarbonate concentration to zeroth order at
high bicarbonate concentration. To model this behavior, they fit the rate data to
–6
–6
–8
–8
–10
log J
–10
–12
Figure 6.3. Graph of the
0
–12 uranium dissolution flux
–1 from oxidizing UO2 as a
lo –2 function of bicarbonate
gm
H –3 0 and dissolved oxygen
CO –1
3 – –4 –2 concentrations at 25°C,
–5 –3 based on data from de
–4 log m O 2
–5 Pablo et al. (1999).
an equation that assumes that both bicarbonate and oxygen interact with the
dissolving surface via Langmuir isotherms.
k+1k+2 Γ UO2 mO2 mHCO3−

J= (6.37)
k−1 + k+2 mHCO3− + k+1mO2
A graph of log J as a function of log M HCO3 and log M O2 (Figure 6.3) shows that
for high values of log M O2 there is a linear relationship between log J and log
M HCO3 , but at low values of log M O2 and high values log M HCO3 , the log J values
become independent of log M HCO3 .
Surface retreat rate
For a dissolving flat surface, the linear rate of surface retreat (dz/dt, m/sec)
is the dissolution flux (J, mol/m2sec) multiplied by the molar volume of the
solid (Vm, m3/mol).
dz
= − JVm (6.38)
dt
Example 6.5. Surface retreat rate for dissolving forsterite
The rate of forsterite dissolution in Example 6.2 is 1.99 × 10−8 mol/m2sec and the
molar volume of forsterite is 4.365 × 10−5 m3/mol. At pH 3 and 25°C the surface
retreat rate is calculated using Eq. (6.38).
dz  mol   m3  m
= − 1.99 × 10 −8 2   4.365 × 10 −5  = −8.69 × 10
−13
(6.39)
dt  m sec   mol  sec
The dissolving surface retreats at a rate of 27.4 µm/yr.
Shrinking particle model
Many hydrometallurgy and chemical engineering processes involve the dis-
solution of solid particles. These processes are typically modeled using a
shrinking particle model, which accounts for the change in dissolution rate
due to decreasing grain surface area as they dissolve away (Burkin, 2001;
Levenspiel, 1972a). The classical derivation of the shrinking particle rate
equation given here shows how the dissolution rate constant, k+(mol/m2sec),
can be calculated from the particle rate constant (kp) that is derived from a fit
of the extent of reaction versus time. The shrinking particle model assumes
Shrinking particle model 115
that the dissolution flux (J = k+, mol/m2sec) is constant over the full extent
of the reaction, either because the solution composition is constant, making
the reaction pseudo-zeroth order, or because the reaction is actually zeroth
order. The first step is to express the rate of reaction of a spherical particle
(r, mol/sec) in terms of the particle’s radius (R, m).
r = − Ak+ = −4π R 2 k+ (6.40)
The rate of volume loss for the particle (dV/dt, m3/sec) is the rate of reaction
multiplied by the molar volume (Vm, m3/mol).
dV
= rVm = −4π R 2 k+Vm (6.41)
dt
The rate of volume loss of the particle is also given by the time derivative of
the volume formula for a sphere.
dV dR
= 4π R 2 (6.42)
dt dt
Setting Eq. (6.41) equal to Eq. (6.42) gives the rate of change of the radius.
dR
= − k+Vm (6.43)
dt
The fraction of material that is reacted away (α) at any time can be expressed
in terms of the radius at that time (R, m) and the initial radius (Ro, m).
 R3 
α = 1−  3 (6.44)
 Ro 
R = Ro (1 − α ) 3
1
(6.45)
The time derivative of Eq. (6.45) gives the rate of change of the fraction
reacted in terms of the rate of change in the particle radius.
dα  R 2  dR
= −3  3  (6.46)
dt  Ro  dt
The right-hand side of this equation can be recast in terms of α by substi-

tuting Eq. (6.43) for dR/dt and Eq. (6.45) for R.
d α 3k+Vm
(1 − α ) 3
2
= (6.47)
dt Ro
It is convenient to gather the constants on the right-hand side of this equa-

tion into a particle rate constant (kp, sec−1).
k+Vm
kp = (6.48)
Ro
This makes (6.47) into a simple function of α.
dα
= 3k p (1 − α ) 3
2
(6.49)
dt
Integrating Eq. (6.49) and evaluating the integral gives a relationship

between the fraction reacted and time.
α t
dα
∫ = 3k p ∫ dt (6.50)
(1 − α ) 3
2
o 0
1 − (1 − α )
1
3 = k pt (6.51)
A graph of 1 − (1 − α ) 3 versus t has the slope of kp. The rate constant (k+) for
1
the dissolution flux is found by rearranging Eq. (6.48).
Ro k p
k+ = (6.52)
Vm
Example 6.6. Forsterite dissolution flux determined using the shrinking

particle model
Van Herk et al. (1989) proposed using forsterite (Mg2SiO4) to neutralize waste
industrial acids. In order to provide rate data to support their idea, they per-
formed a series of laboratory tests that determined the fraction of forsterite
reacted away (α) by contact with acidic solutions for various times. Their data
can be used, along with Eqs (6.51) and (6.52), to find the rate constant for for-
sterite dissolution.
The experiment dissolved 105–125 µm forsterite grains in pH 1 HCl solution
at 40°C. The results are shown in Figure 6.4. When 1 − (1 − α)⅓ is graphed versus
t, the slope of the resulting straight line is 1.23 × 10−6 sec−1. The effective diam
eter of the particles can be estimated using Eq. (6.2).
Dmax − Dmin 125 × 10 −6 − 105 × 10 −6 20 × 10 −6

De = = = = 115 × 10 −6 m (6.53)
 Dmax   125 × 10 −6  0.174
ln  ln  
 Dmin  105 × 10 −6 
Shrinking particle model 117
(a) 1
0.8
0.6
0.4
0.2
0
0 2000 4000 6000
time, min
(b) 0.5
0.4
0.3
1 – (1 – )1/3
0.2
0.1
0
0 1× 105 2× 105 3× 105 4 ×105
time, sec
Figure 6.4. (a) Fraction of forsterite reacted away as a function of time. (b) The
reacted fraction data fit to the shrinking particle model. The slope of the line is kp.
So Ro = 57.5 × 10−6 m and Vm = 4.365 × 10−5 m3/mol. Equation (6.52) can be

used to find k+.
Ro k p (57.5 × 10 −6 ) (1.23 × 10 −6 ) mol

k+ = = = 1.62 × 10 −6 2 (6.54)
Vm ( 4.365 × 10 −5 ) m sec
Particle lifetime model
Geochemists are more familiar with the closely related particle lifetime
model of Lasaga (1998b), which is based on the same assumptions as the
shrinking particle model. This section shows how these models are related.
When the temperature, solution composition, and other rate-controlling
variables remain nearly constant, the dissolution flux (J = k+, mol/m2sec) for
a dissolving particle is constant and the release rate of the dissolving sub-
stance (r, mol/sec) is proportional to the particle’s surface area (A, m2).
r = Ak+ (6.55)
If the particle is a sphere with a diameter D (m) its area A = πD2, m2. The
rate of volume loss of the particle is the release rate multiplied by the molar
volume of the dissolving substance (Vm, m3/mol).
dV
= − rVm = −π D 2 k+Vm (6.56)
dt
The rate of volume loss of the particle is also given by the time derivative of
the volume change for a sphere (V = πD3/6).
dV π D 2 dD
= (6.57)
dt 2 dt
Setting these two equations equal to each other gives the rate of reduction
of the particle’s diameter.
dD
= −2 k+Vm (6.58)
dt
This equation is integrated and evaluated between the initial diameter (Do,
m) and the diameter (Dt, m) when time = t.
Dt = Do − 2 k+Vmt (6.59)
If the particle dissolves until its diameter reaches zero, it has dissolved away
completely and the particle lifetime (tl, sec) is found by setting Dt = 0 in Eq.
(6.59).
Do
t = (6.60)
2 k+Vm
Because Do/2 = Ro, the lifetime of the grain can also be expressed in terms
of kp as defined by Eq. (6.48). This relationship links the shrinking particle
and particle lifetime models.
Particle lifetime model 119
Ro 1
t = = (6.61)
k+Vm k p
Equation (6.60) can be rearranged to find the ratio of the original diameter
(Do) to the lifetime (tl).
Do
2k+Vm = (6.62)
t
Substituting Eq. (6.62) into Eq. (6.59) gives a relationship between dimen-
sionless diameter and dimensionless time.
Dt  t
= 1− (6.63)
Do  t 
This equation shows that the particle diameter shrinks as a linear function
of elapsed time as shown in Figure 6.5.
The surface area of the particle is πD2, so the diameter can be
expressed in terms of the surface area and that relationship can be sub-
stituted into Eq. (6.63) to find the reduced surface area as a function of
reduced time.
1
 A 2
D=  (6.64)
π
2
At  t
= 1 −  (6.65)
Ao  t 
Equation (6.55) states that the rate of transfer of material (r, mol/sec) from
the particle to the solution is directly proportional to the surface area, so it
can be combined with Eq. (6.65) to give the dimensionless release rate as a
function of dimensionless time.
2
r  t
= 1− (6.66)
ro  t 
The particle diameter can be expressed in terms of the particle volume (V

= πD3/6) and that relationship can be substituted into Eq. (6.63) to find the
reduced particle volume as a function of reduced time.
1
 6V  3
D= (6.67)
 π 
0.8
0.6
reduced dimension D/Do
0.4 A/Ao
Figure 6.5. Dimensionless V/Vo

0.2
diameter, surface area,
and volume of a dissolving
spherical particle as a
function of dimensionless 0
time based on Eqs (6.1), 0 0.2 0.4 0.6 0.8 1
(6.3), and (6.6). reduced time
3
Vt  t
= 1 −  (6.68)
Vo  t 
Multiplying the volume of the particle by the molar volume of the sub-
stance gives the number of moles of substance in the particle and the ratio,
nt/no, is the fraction of material remaining (p) at any time and p equals one
minus the fraction of material that has dissolved away (α).
3
nt  t
= p = (1 − α ) = 1 −  (6.69)
no  t 
Combining Eqs (6.61) and (6.69) gives a relationship between α and t.
1 − α = (1 − k pt )
3
(6.70)
Taking the cube root of both sides of this equation followed by rearrange-
ment produces the shrinking particle model derived earlier as Eq. (6.51).
Equation (6.63) can also be rearranged and expressed in terms of kp.
 D t
1 − t = = k pt (6.71)
 Do  t
This relationship can be used for dissolving grains that are sampled and their
diameters measured over the course of an experiment. The slope of a graph
Particle lifetime model 121
of (1 − Dt/Do) versus time is kp (= 1/tl = 2k+Vm/Do), which can be used to find

the dissolution rate constant. Liu et al. (2008) used this approach to deter-
mine the rate of nonoxidative dissolution of galena in an acid solution.
Equation (6.66) can be rearranged and expressed in terms of kp.
1
r 2
t
1−  t  = = k pt (6.72)
 ro  t
If the rate of change of the solution concentration is measured at the onset

of the experiment and again at several later times, the slope of a graph of
the left-hand side of this equation versus time is kp (= 1/tl = 2rVm/Do), which
can be used to find kp and the dissolution rate constant (k+). This approach
would be especially suitable for mixed flow reactor experiments where the
rates are determined directly from the solution composition and flow rate.
Example 6.7. Mineral grain lifetimes
Goldich (1938) proposed a mineral weathering stability series that is based on

Bowen’s reaction series. According to the Goldich series, olivine is very unstable
in the weathering environment and quartz is very stable. Goldich’s qualitative
model seems reasonable, but we now have extensive mineral dissolution data
that can be used to quantify this model by calculating the lifetime of 1 mm diam-
eter mineral grains.
The dissolution rate constant for forsterite at pH = 4.5 is 2.75 × 10−9 mol/
m sec (Palandri and Kharaka, 2004) and molar volume is 4.365 × 10−5 m3/mol
2
(Robie and Hemmingway, 1995). The lifetime of a 1 mm diameter grain is cal-
culated using Eq. (6.60).
Do 1 × 10 −3 m
t = = = 8.33 × 109 sec (6.73)
2 k+Vm  mol   m3 
2  2.75 × 10 −9 2   4.365 × 10 −5 
 m sec   mol 
The lifetime of a 1 mm diameter forsterite grain is 131 years.

The dissolution rate constant for quartz is 2.75 × 10−13 mol/m2sec (Palandri and
Kharaka, 2004) and molar volume is 2.269 × 10−5 m3/mol (Robie and Hemmingway,
1995). The lifetime of a 1 mm diameter grain is calculated using Eq. (6.60).
Do 1 × 10 −3 m
t = = = 8.01 × 1013 sec (6.74)
2 k+Vm  mol  m3 
2  2.75 × 10 −13
  2.269 × 10
−5

 m 2sec   mol 
The lifetime of a 1 mm diameter quartz grain is 2.54 million years, which means
that the quartz grain will persist 19,000 times longer than the forsterite grain.
Precipitation and dissolution surfaces

Precipitation and dissolution reactions at mineral surfaces are quite com-
plex because both the surface structure and the reaction mechanisms change
as a function of the driving force for the reaction. A number of conceptual
and quantitative models capture one or another aspect of the growth and/
or dissolution process but, like the blind men and elephant proverb, each
model explains only selected aspects of these processes so we lack an overall
picture of the beast. The point of this section is to caution the reader against
over-interpreting any of these models.
Equilibrium condition
At equilibrium the chemical potential of each component in the solid equals
its chemical potential in the solution so ∆µr = 0 and the concentration of
each component in solution remains unchanged over time. However, dis-
solution and precipitation reactions continue to occur and their rates are
exactly equal because of the principle of detailed balance. This means that
although there is no net mass transfer between the solid and solution, these
reactions constantly reconstruct the surface. For most solids at room tem-
perature this effect is small but, at the high temperatures or for the long
times characteristic of many geological settings, surface reconstruction can
transfer significant quantities of trace element or isotopic species between
the surface and the solution. At low temperatures the rate of exchange of
components between the surface and the interior of the crystal is limited by
slow solid-state diffusion, so the composition of crystal interiors typically
remains unchanged over geologically significant times. The constant inter-
change of components between the solution and the surface allows the sur-
face defects and heterogeneities to anneal away over time so that the solid
surface evolves toward its equilibrium structure and composition.
The Wulff theorem states that the surface free energy of a crystal face is
directly proportional to the distance from the Wulff point, which lies in the
interior of the crystal (Mutaftschiev, 2001; Wulff, 1977). For amorphous
solids, the equilibrium shape is a sphere. For crystals, the equilibrium shape
can be predicted using the Wulff construction, which relates the surface free
energy of a crystal face to the length of a vector that is perpendicular to
that crystal face and passes through the Wulff point in the interior of the
crystal. Wolff and Gualtieri (1962) describe the methods used to predict
the equilibrium shape of a crystal. They also point out that crystals sel-
dom achieve a perfect equilibrium shape because they grow from a non-
equilibrium, supersaturated solution. Dissolution rate experiments typically
use broken grains produced by crushing. The surface free energy of some
sites on these grains can be quite high and these sites dissolve quickly when
Slightly undersaturated/slightly supersaturated conditions 123
first exposed to the solution (Petrovich, 1981a, 1981b). As dissolution pro-

ceeds, these highly reactive surface sites dissolve away and the dissolution
rate decreases. However, dissolution continues to deplete the most reactive
sites on the surface leading to a surface structure and composition that is
unlike the equilibrium surface.
The Kossel crystal model is another useful way to visualize the relationship
between crystal structure and crystal shape. This model visualizes a crystal
as an ordered assemblage of cube-shaped growth units (Figure 6.6). The
Kossel crystal model provides the conceptual foundation for the periodic
bond chain theory (Hartman and Perdok, 1955a, 1955b, 1955c). Periodic
bond chains (PBC) are chains of strong bonds that run through the crystal
and define the direction of a PBC vector. Flat faces (F-faces) on a Kossel
crystal are parallel to at least two PBC vectors. Stepped faces (S-faces) are
parallel to at least one PBC vector and kinked faces (K-faces) are not par-
allel to any PBC vector.
Whether at equilibrium or not, the surface atoms of a solid are under-
bonded so they must complete their coordination by forming bonds with
H2O, OH, or a ligand from the solution. The surface species can ionize in
response to changing pH to produce a surface charge that is responsible
for further interactions with solution species. Stumm (1992) summarizes the
most important models of the chemistry of the solid/solution interface.
Slightly undersaturated/slightly supersaturated conditions

Departure from equilibrium produces a growing or dissolving surface
with a composition and structure that reflects competing kinetic processes.
Growing crystal surfaces develop convex surface features (hillocks), which
because of surface free energy effects have a chemical potential that is
greater than atomically flat, equilibrium crystal faces. Dissolving crystals
develop concave surface structures such as etch pits that have a lower chem-
ical potential than atomically flat, equilibrium faces. The shape of growing
crystals is controlled by a complex interplay of growth rate and growth site
inhibition by adsorbed species (Sunagawa, 2005). Although less well stud-
ied, crystal dissolution shapes appear to be controlled by similar processes.
Multicomponent minerals dissolve incongruently as weakly bonded com-
ponents are preferentially released to the solution. This enriches the surface
in strongly bonded components, which makes the dissolving surface behave
like a metastable phase that is less soluble than the bulk phase.
The dissolution or precipitation rate for near-equilibrium conditions can
be modeled as a function of reaction affinity (A, kJ/mol = −∆µr).
J = J + (1 − e − ∆µr RT
) = J + (1 − e A RT ) (6.75)
ledge
F-face terrace
kink
S-face
K-face
Figure 6.6. Schematic dislocation

diagram of a Kossel crystal
showing F-, S-, and K-faces 2D
predicted by the PBC nucleus
theory along with features
predicted by various
growth models. Based on
growth
Figure 1 in Sleutel et al. unit
(2012). spiral hillock
The affinity model is based on the principle of detailed balance (see the
derivation in Chapter 3). According to this model, when A = 0, J = 0 and
when A > 5RT, J ≈ J+. This model seems to work for simple compounds, such
as quartz, that dissolve congruently and reversibly. It is easier to visualize
the goodness of fit by converting Eq. (6.75) to a linear form (Figure 6.7).
(J J + ) = 1 − e A RT (6.76)
Although this model has considerable conceptual value and provides

modelers with a simple way to force reaction rates to zero at equilibrium, it
is not well suited to express the complexities associated with the dissolution
of many minerals. Changes in the reaction mechanism with increasing affin-
ity (Arvidson and Luttgbe, 2010) reset the far-from-equilibrium dissolution
flux, J+, to different values so the J versus A graph does not approach a plat-
eau at A = 5RT. Incongruent dissolution changes the surface composition
away from that of the bulk solid (Schott and Oelkers, 1995) so the chemical
potential of the surface components are different from the bulk solid. This
means that the A = 0 reference point based on the unreacted mineral does
not apply to the leached surface and the J versus A graph does not cross
the affinity axis at zero. Both of these effects invalidate the fundamental
assumption of the principle of detailed balance, which requires that the for-
ward and reverse reactions remain the same over the entire range of the
model.
Slightly undersaturated/slightly supersaturated conditions 125
(a) A/RT
0 2 3 4 5 6
4 10–7
3 10–7
J, mol/m2sec
2 10–7
1 10–7
0
0 5 10 15 20 30
A, kJ/mol
(b) A/R
0 1 2 3
1. 2
(b )
1
0. 8
Figure 6.7. (a) Quartz

0. 6
J/J+
dissolution flux at 300°C, J,

as a function of the affinity,
0. 4 A, driving the reaction. (b)
Quartz dissolution flux,
J/J+, as a function of the
0. 2 linearized affinity, 1 −
eA/RT. The slope of the best-
fit solid line is 0.88 and
0 the theoretical slope is 1.0.
0 0.2 0.4 0.6 0.8 1 Data from Berger et al.
1 – eA/RT (1994).
The Kossel crystal model shown in Figure 6.6 provides the conceptual
basis for many models of crystal growth and dissolution. Most of those
models also incorporate the concepts of terrace, ledge, and kink (TLK),
which describe the free energy of attachment of a growth unit as being pro-
portional to the number of bonds (1 through 6) between it and the crystal.
According to the TLK model, the energy needed to form a 2D nucleus,
which initiates the growth of a new layer, is high, making the predicted rates
of crystal growth much lower than is observed. This problem was resolved
by the Burton–Cabrera–Frank (BCF) theory (Burton and Cabrera, 1949;
Burton et al., 1949, 1951; Cabrera and Burton, 1949) that postulates that
new ledges are the result of line dislocations that grow into spiral hil
locks (Figure 6.6). Zhang and Nancollas (1990) and Teng et al. (2000) give
examples of the application of these concepts to the growth of minerals.
Dissolution processes can be visualized as the inverse of the Kossel crystal
growth processes. For example, line defects at the spiral hillocks provide
steps that enhance dissolution rates leading to etch pit formation (Brantley
et al., 1986a; Brantley et al., 1986b). Etch pits and vacancy islands (Dove
et al., 2005) exert a profound control on the surface structure of dissolving
solids (Lasaga and Blum, 1986; MacInnis and Brantley, 1993). Etch pits
and the opening of vacancy islands cause weathered mineral surfaces to be
extremely rough (Berner et al., 1980; Velbel, 1989). Monte Carlo models of
dissolving Kossel crystal surfaces show that surface roughness increases with
the extent of the reaction and the departure from equilibrium (Bandstra and
Brantley, 2008; Wehrli, 1989). Because ledge retreat can initiate at crystal
edges, the edge regions dissolve more quickly than the centers of the faces,
so the dissolving crystal becomes rounded. Eventually the dissolving crystal
becomes bounded by rough and curved surfaces from which growth units
can be removed at any location (Cabrera and Vermiyea, 1958).
Incongruent dissolution changes the surface chemistry of dissolving min-
erals. Incongruent dissolution occurs when there are significant differences
among the bond strengths of the mineral’s components. This means that in
the initial stages of the dissolution process, weakly bonded components are
released quickly and strongly bonded components are retained in the sur-
face. The release rate of the weakly bonded components diminishes as their
concentration declines until their release rate matches the release rate of the
strongly bonded components. Eventually the dissolution becomes congruent
as the mineral surface becomes paved by a “leached layer” that is enriched
in the slowest dissolving components (Hellmann et al., 1990; Schott et al.,
2012). The leached layer can undergo reconstruction reactions that cross-
link the components, which further slows its dissolution rate (Casey et al.,
1993).
Very undersaturated/very supersaturated conditions 127
Very undersaturated/very supersaturated conditions

With increasing supersaturation, dissolved species interact to form oligo
mers that increase in size in proportion to the degree of supersaturation.
These oligomers become the most important growth units for precipitating
solids (Cölfen and Antonietti, 2008). The oligomers undergo self-assembly
and their shape has a profound effect on the internal structure of the grow-
ing solid (Damasceno et al., 2012). Oligomer attachment causes microscopic
surface roughening and their slow diffusion rates promote the formation of
dendrites. Dendritic ice crystals, e.g. snowflakes, are a familiar example of
this growth habit (Furukawa and Shimada, 1993). There are many examples
of mineral dendrite formation.
Incongruent dissolution and leached layer thickness becomes more mani-
fest at very low pH (Weissbart and Rimstidt, 2000). Under these conditions
the leached layer can dissolve by shedding oligomers as well as monomers
(Dietzel, 2000; Weissbart and Rimstidt, 2000). These oligomers subsequently
depolymerize after they move away from the dissolving surface.
Chapter 7
Diffusion and advection
Diffusion and fluid advection are the most important mass transfer pro-
cesses in low temperature geochemical systems. Whenever mass transfer
processes are slow relative to chemical reaction rates, diffusion and advec-
tion processes will control the overall rate of a geochemical process. This
chapter introduces some simple models that link transport rates to chemical
reaction rates.
Advection
Advection rates are expressed in terms of volume discharge, specific dis-
charge, or surficial velocity depending upon the requirements of the model.
Discharge (also called volumetric flow rate) (Q = dV/dt, m3/sec) is the vol-
ume of fluid passing through a cross-sectional area per unit time. Specific
discharge (also called Darcy velocity) (q = Q/A, m/sec) is the volumetric flow
rate divided by the surface area through which the fluid is flowing. Surficial
velocity (v, m/sec) is the fluid velocity relative to an adjacent solid surface.
For a packed bed, the surficial velocity is the specific discharge divided by
the porosity (v = q/ϕ).
Viscosity is a fluid’s resistance to flow, so advection models must account
for viscosity either as an explicit term or in a scaling constant. The dynamic
viscosity (η, kg/m sec) equals the kinematic viscosity (ν, m2/sec) divided by
the density (ρ, kg/m3). The dynamic viscosity of liquid water in equilibrium
with the vapor phase for temperatures between the freezing and critical
points can be computed from an equation given by Watson et al. (1980).
ln η = −10.1357 + 1.5064τ 1 3 − 0.1189τ + 1.7604τ 2 − 0.53367τ 7 3 + 4.9643 × 10 −3 τ 10

647.073
τ = −1 (7.1)
T (K )
The density of liquid water in equilibrium with the vapor phase for tem-
peratures between the freezing and critical points can be computed from an
equation given by Wagner and Pruß (2002).
128
Dimensionless numbers 129
1 + 1.9927ϑ 1 3 + 1.0997ϑ 2 3 − 0.51089ϑ 5 3 

ρ = 322 
 − 1.7549ϑ 16 3 − 45.517ϑ 43 3 − 6.7469 × 105 ϑ 110 3 
(7.2)
T (K )
ϑ =1−
647.096
Fluids flow in response to a pressure difference. Buoyancy forces, due to

density differences related to differences in the temperature or salinity, can
cause fluid flow. Buoyancy forces are considered in models of convective
flow. Fluids also flow because of differences in the hydrostatic head between
a source and discharge region. Hydrostatic head is the difference in elevation
(∆z, m), which produces a pressure difference because of gravitational accel-
eration (P = ρg∆z). The Manning equation and Darcy’s law are examples of
equations that model flow driven by hydrostatic forces. The Manning equa-
tion predicts flow velocity (v, m/sec) in open channels (Chaudhry, 2008) as a
function of the channel’s cross-sectional area (A, m2), wetted perimeter (P,
m), and the slope of the water surface (s, m/m).
k
v= ( A P )2 3 s1 2 (7.3)
n
In this model, k is a flow constant (= 1 m1/3/sec for SI units) that makes the
units work out and n (no units) is an empirically determined channel rough-
ness factor that ranges from ~0.03 for clean and straight natural streams
to ~0.05 for natural streams lined with cobbles and large boulders (Barnes,
1967; Limerinos, 1970). More sophisticated models have been developed
and calibrated (Bjerklie and Dingman, 2005). Darcy’s law predicts the spe-
cific discharge (q, m/sec) in porous media (packed beds) as a function of the
pressure drop across the bed (∆P, Pa) and the length of the bed (L, m).
 k   ∆P 
q=   (7.4)
 η  L 
In this model, k (m2) is the permeability of the packed bed and η (kg/m sec)
is the dynamic viscosity of the fluid.
Dimensionless numbers
Dimensionless numbers are commonly used in diffusion and advection
models because they simplify the scaling of the models from laboratory
experiments to practical dimensions. This approach also takes advantage of
the Buckingham π theorem (Barenblatt, 2003), which states that n indepen-
dent variables with k independent dimensions can be expressed in terms of
p independent dimensionless numbers.
130 Diffusion and advection
p=n−k (7.5)
Dimensionless numbers are widely used in engineering disciplines but are

less common in geochemical models. The notation for dimensionless num-
bers consists of a capitalized first letter followed by a lower case letter.
Methods used to estimate the numerical values of dimensionless numbers
range from purely theoretical to purely empirical. It is advisable to carefully
review each estimation method before incorporating a dimensionless num-
ber into a model.
Dimensionless numbers are scale independent and contain a character-
istic length term (L) that adjusts their magnitude to match the scale of the
model. The velocity, v (m/sec), in these definitions is the superficial velocity
of the fluid.
The Reynolds number, Re, is defined as the ratio of inertial forces, which
resist a change in the flow velocity, to the viscosity, which slows the flow
velocity (Denny, 1993; Vogel, 1994). Purcell (1977) discusses several applica-
tions of the Reynolds number to real situations. For flow past a flat plate the
Reynolds number is
ρvL
Re = (7.6)
η
For this geometry, when Re < ~2300 the flow is laminar and when Re >
~4000 the flow is turbulent.
For flow through a packed bed with a porosity (φ, no units) and consisting
of grains with a diameter (D, m), the Reynolds number is
ρvD
Re = (7.7)
η (1 − φ )
When Re < ~10 the flow is laminar and when Re > ~2000 the flow is fully
turbulent.
For flow in a smooth-walled fracture (Qian et al., 2005) with an aperture
width of L, the Reynolds number is
ρvL
Re = (7.8)
2η
When Re > ~2300 the flow is turbulent.

For a vessel stirred (Lamberto et al., 1999) with an agitator of diameter (D,
m) with a rotational speed of N (revolutions/sec), the Reynolds number is
ρND 2
Re = (7.9)
η
Dimensionless numbers 131
When Re > ~10,000 the flow is fully turbulent.

The Schmidt number, Sc, is defined as the ratio of viscosity to diffusivity
(Di, m2/sec).
η ν
Sc = = (7.10)
ρDi Di
The Schmidt number and Reynolds number can be combined to find the
Péclet number.
Pe = Re Sc (7.11)
The Péclet number is defined as the ratio of mass transfer by advection (v,
m/sec) to mass transfer by dispersion (DL, m2/sec) and diffusion (Di, m2/
sec).
vL
Pe = (7.12)
Di + DL
For Pe < 50, diffusion/dispersion dominates the transfer of material between

the solution and surface; and for Pe > 100, advection dominates. For many
geochemical scenarios, Di << DL (Knapp, 1989) and only longitudinal dis-
persion needs to be considered in models. For Taylor dispersion in a fracture
with a half-width of L (m), the longitudinal dispersion coefficient is (Horne
and Rodriguez, 1983)
2 L2v 2
DL = (7.13)
105 Di
The longitudinal dispersivity, αL (m), is the longitudinal dispersion coeffi-

cient (DL, m2/sec) plus the diffusion coefficient (Di, m2/sec) divided by the
fluid velocity (v, m/sec).
DL + Di
αL = (7.14)
v
This makes the Péclet number inversely related to the longitudinal dispersiv-
ity (Knapp, 1989).
1
Pe = (7.15)
αL
The 1 in the numerator of this fraction has units of meters. Empirical stud-
ies of flow in geological media have demonstrated that αL is related to the
scale of the study, L (m) (Table 7.1).
Table 7.1. Values of c and m

(Schulze-Makuch, 2005) that can
be used to calculate longitudinal
dispersivity using Eq. (7.16).
Medium c m
Unconsolidated 0.085 0.81

Sandstones 0.01 0.92
Carbonates 0.80 0.40
Basalts 0.15 0.61
Granites 0.21 0.51
α L = cLm (7.16)
The Damköhler numbers (Da) compare the rate of consumption or pro-

duction of a dissolved substance to the rate of delivery or removal of that
substance at the reaction site (Steefel, 2008). The first and second Damköhler
numbers are most useful for geochemical models. DaI compares the reaction
rate to the advection rate.
( A / M )kL
Da I = (7.17)
meqv
If DaI >1, the reaction produces or consumes aqueous species faster than
they are delivered or removed by advection, so the overall rate is limited by
fluid flow. DaII compares the reaction rate to the rate of diffusive mass trans-
fer across the solid/solution interface.
Da II = Da I Pe =
( A M ) kL2 (7.18)
Di meq
If DaII >1, the reaction produces or consumes aqueous species faster than
they are delivered or removed by diffusion so the reaction rate is limited by
diffusion. A/M (m2/kg) is the ratio of the interfacial area between the solu-
tion and solid to the mass of solution; k (mol/m2sec) is the dissolution flux
rate constant; meq (mol/kg) is the equilibrium concentration; v (m/sec) is the
linear flow velocity; and Di (m2/sec) is the effective diffusion coefficient.
Fick’s laws
Diffusion is caused by the displacement of molecules due to random molecu-
lar motions. If the molecules are not homogeneously distributed in a region,
Fick’s laws 133
x+∆x
x ∆x
Jout
A
Figure 7.1. Volume element

Jin
used to develop Fick’s
second law.
more of them will be displaced away from areas of high concentration into
areas of low concentration than vice versa. This causes net transport down a
concentration gradient. Erdey-Grúz (1974) and Robinson and Stokes (1959)
provide theoretical details about how and why diffusion occurs in aqueous
solutions. Cussler (2009), Denny (1993), Levich (1962), Probstein (1989),
and Vogel (1994) are useful resources for models that link diffusion and
advection. Berner (1980) and Lerman (1979) present models that are appli-
cable to sediments.
Fick’s laws are the basis for diffusive transport models. Fick’s first law
states that diffusive flux (J, mol/m2sec), is the product of a concentration
gradient (dc/dx, mol/m4) and a diffusion coefficient (D, m2/sec). Note that
concentration (c) is expressed with units of mol/m3.
dc
J = −D (7.19)
dx
Fick’s second law is related to the first by a simple derivation, which consid-
ers the volume element shown in Figure 7.1. As molecules diffuse into and
out of the volume element, the rate of change in the number of molecules
in the volume is the difference between the flux into and out of the volume
across the A areas at each end of the volume.
∆n
= − A( J x + ∆x − J x ) (7.20)
∆t
The change in the number of molecules divided by the volume of the ele-
ment is the concentration change in the volume.
∆n
∆c = so ∆n = ∆c ( A∆x ) (7.21)
A∆x
Substituting Eq. (7.21) into Eq. (7.20) gives the rate of change of concen-
tration in the volume.
∆c ( A∆x )
= − A( J x + ∆x − J x ) (7.22)
∆t
∆c  A ( J − Jx ) (J − Jx )
= −   x + ∆x = − x + ∆x (7.23)
∆t  A  ∆x ∆x
The fluxes are defined in terms of the concentration gradients using Fick’s
first law.
  ∂c   ∂c     ∂c   ∂c  
 D  − D   D  − 
∆c  ∂x  x + ∆x  ∂x  x    ∂x  x + ∆x  ∂x  x 
=− =− (7.24)
∆t ∆x ∆x
When ∆x → 0, Eq. (7.24) becomes Fick’s second law.
dc  d 2c 
= −D  2  (7.25)
dt  dt 
Using Fick’s laws to model geochemical situations requires the modeler to

define boundary conditions based on the geometric and chemical relation-
ships between the solids and the solution. The resulting differential equa-
tions can be quite challenging to solve but many analytical solutions have
already been developed. Crank (1975) gives detailed solutions for many rel-
evant problems and because Fourier’s law is the analog of Fick’s first law,
with suitable replacement of variables, the solutions for heat flow problems
in Carslaw and Jaeger (1959) can be applied to diffusion models.
Diffusion coefficients
Several experimental designs are used to measure diffusion coefficients
(Robinson and Stokes, 1959). Tracer diffusion coefficients for ions are
typically calculated from the limiting equivalent conductivity of the ions
(Lerman, 1979; Oelkers and Helgeson, 1988; Robinson and Stokes, 1959).
At 25°C, the diffusion coefficients for most ions range between 0.5 × 10−9 and
10 × 10−9 m2/sec (Li and Gregory, 1974; Miller, 1982; Oelkers and Helgeson,
1988). The diffusion coefficients for uncharged organic species are similar to
those for most ions (Oelkers, 1991).
The diffusion coefficients of H+ and OH− ions are much larger than
other ions because their movement in solution involves a kind of hop-
ping mechanism (Grotthuss mechanism) related to the very fast dissocia-
tion and re-association of water molecules (Cukierman, 2006; Tuckerman
et al., 2002).
Temperature, solution composition, and tortuosity 135
Table 7.2. Diffusion coefficients for selected aqueous species (Miller, 1982;

Oelkers and Helgeson, 1988). The a, b, and c values are coefficients for
Eq. (7.27).
D (25°C) Ea (25°C) Ea (125°C)

Species m2/sec a b c kJ/mol kJ/mol
H+ 9.3 × 10−9 −7.330 194.7 −119613 11.6 4.3

OH− 5.5 × 10−9 −7.418 278.3 −158688 15.1 5.3
Na+ 1.3 × 10−9 −6.990 −196.5 −107366 17.5 10.4
Ca2+ 0.9 × 10−9 −7.585 177.9 −185795 20.5 9.6
Cl− 2.0 × 10−9 −7.049 −96.55 −115838 16.7 9.0
SO42− 1.1 × 10−9 −6.959 −256.2 −100951 17.9 11.7
methane 1.7 × 10−9 −6.849 −203.5 −109402 18.0 11.2
octane 0.7 × 10−9 −7.212 −223.7 −104675 17.7 11.3
methanol 1.6 × 10−9 −6.984 −70.83 −140820 19.5 10.8
octanol 0.7 × 10−9 −7.335 −41.29 −147114 19.7 10.6
benzene 1.1 × 10−9 −7.262 63.43 −171300 20.8 10.2
acetic 1.2 × 10−9 −7.026 −150.8 −122463 18.6 11.1
acid
Effect of temperature, solution composition,

and tortuosity on D
Diffusion is a thermally activated process so diffusion rates increase with
increasing temperature. For many types of diffusion, this temperature effect
is modeled by fitting the diffusion coefficient to an Arrhenius type law.
D = Ae − Ea RT
(7.26)
For aqueous species, graphs of log D versus 1/T are curved because Ea
decreases with increasing temperature (Figure 7.2). Values of Ea range from
around 20 kJ/mol at 25°C to near 10 kJ/mol at 300°C (Table 7.2). Ea for H+
and OH− are lower and their diffusion coefficients change somewhat less
with increasing temperature, but they are always larger than for other spe-
cies. Diffusion coefficients for aqueous species along the water liquid/vapor
curve are usually fit by a polynomial in 1/T.
b c
log D = a + + 2 (7.27)
T T
This makes the activation energy a function of temperature.

 d log D 
E a = 2.303R  (
2c
 = 2.303R b + T
 d (1 T ) 
) (7.28)
(a) 300° 200° 100° 25°C

–7.0
–7.5
H+
–8.0
Cl – OH –
logD
–8.5 Na +
Ca
–9.0 2+ SO
2
4 –
9.5
0.0017 0.0022 0.0027 0.0032 0.0037
1/T, K–1
(b) 300° 200° 100° 25°C

–7.5
–8.0 oc
tan oc
e tan
ol
–8.5 methane
logD
methanol
–9.0
acetic acid
be
nz
Figure 7.2. Log D versus

en
e
1/T behavior for selected

aqueous species (Miller, –9.5
1982; Oelkers, 1991;
0.0017 0.0022 0.0027 0.0032 0.0037
Oelkers and Helgeson,
1988). 1/T, K–1
According to the Stokes–Einstein equation, diffusion coefficients have a

reciprocal relationship to the dynamic viscosity of the host fluid (η) and the
hydrodynamic radius of the diffusing species (R).
K BT (7.29)
D=
6πηR
Temperature, solution composition, and tortuosity 137
This equation can be recast to create a function that can be used to extrapo-
late diffusion coefficients to other temperatures.
 ηT   T 
DT2 =  1   2  DT1 (7.30)
 T1   ηT2 
(a) 2.5 ×10–8
2.0 ×10–8
D, m2/sec
1.5 ×10–8
1.0 ×10–8
5.0 ×10–9
0
0 50 100 150 200 250 300
T, °C
(b) 300° 200° 100° 25°C

–7.5
–8.0
log D
–8.5
–9.0
–9.5
0.0017 0.0022 0.0027 0.0032 0.0037
1/T
Figure 7.3. (a) Diffusion coefficient of H4SiO4 as a function of temperature

estimated using the Stokes–Einstein equation. (b) Log D for H4SiO4 versus 1/T
estimated using the Stokes–Einstein equation.
Example 7.1. Diffusion coefficient of silica (H4SiO4) as a function of

temperature
The diffusion coefficient of H4SiO4 in pure water at 25°C is 1.17 × 10−9 m2/sec
(Rebreanu et al., 2008). This value can be extrapolated to hydrothermal
temperatures using Eq. (7.30). The easiest way to perform this extrapolation is
to log transform Eq. (7.30).
 ηT  T 
log DT2 = log  1  + log  2  + log DT1 (7.31)
 T1   ηT2 
Values of η are calculated from Eq. (7.1) and used to find the first and second
terms of Eq. (7.31). The values of log (T2/ηT2) can be fitted to a temperature
function.
2
T   1  1
log  2  = −100039   − 289.64   + 7.6142 (7.32)
 ηT2   T2   T2 
At T1 = 298°C, log (η/T1) = −5.524 and log DT1 = −8.93. These values along
with Eq. (7.32) are inserted into Eq. (7.31) to produce a temperature function
for log DT2.
2
 1  1
log DT2 = −100039   − 289.64   − 6.840 (7.33)
 T2   T2 
The values of log D and D predicted by Eq. (7.33) are shown in a graph in
Figure 7.3.
The effect of the solution composition on D is complex. To maintain

charge neutrality, the diffusion rates of cations and anions must be linked.
The mutual diffusion coefficient for combinations of cations and anions is
approximately the harmonic mean of their tracer diffusion coefficients.
2DcationDanion
Dmutual = (7.34)
Dcation + Danion
In addition, ions can interact with each other to form ion pairs and
complexes. The diffusion coefficients of ion pairs can be estimated using a
method described by Applin and Lasaga (1984).
Diffusion rates in porous media are slower than in bulk solution because
the species must diffuse around the solid particles, which increases their dif-
fusion path length. This increase in diffusion path length is expressed as
Mineral dissolution without flow 139
tortuosity (τ), which is defined as the ratio of the actual distance traveled
by the species to the linear distance between the initial and final location of
the species. Tortuosity increases with decreasing porosity (φ) and there are
many theoretical and empirical models for this relationship (Lerman, 1979;
Shen and Chen, 2007). Three examples of empirical fits taken from Shen
and Chen (2007) illustrate the diversity of tortuosity models.
τ 2 = φ −1.14 (7.35)
τ 2 = φ + 3.79 (1 − φ ) (7.36)
τ 2 = 1 − 2.02 ln φ (7.37)
Diffusion coefficients can be adjusted for use in porous media by dividing

them by τ2.
D
Dτ = (7.38)
τ2
Mineral dissolution without flow

Fick’s second law can be used to model the one-dimensional concentration
gradient in a static solution in contact with a dissolving mineral. This model
solves Eq. (7.25) for boundary conditions specifying that when x = 0, c = cs
and when x = ∞, c = 0. This model predicts concentration as a function of
position and time (Crank, 1975).
 x 
c( x, t ) = cs erfc  (7.39)
 2 Dt 
In this equation, erfc(z) is the complementary error function: erfc(z) = 1 −

erf(z).
Example 7.2. Gypsum (CaSO4•2H2O) dissolution in an unstirred

solution
If gypsum is immersed in an unstirred aqueous solution it will dissolve more

rapidly than the Ca2+ and SO42− ions can diffuse away from the surface. This
means that the solution very near the surface will be nearly in equilibrium with
the gypsum while the solution distant from the surface will contain no Ca2+ or
SO42− ions. Equation (7.39) can be used to predict the concentration of these
ions as a function of distance from the surface at a chosen time.
The mutual diffusion coefficient of CaSO4 is found using Eq. (7.34).
15 15 ×10 –3
100 sec
10 ×10 –3
10 10 sec
concentration, mol/m3
1.0 sec
5 5 ×10 –3
0.1 sec
0 0
0 20 40 60 80 100
distance, µm
Figure 7.4. Concentration of Ca2+ or SO42− as a function of distance and elapsed

time from a gypsum surface. The gypsum is dissolving into a static solution.
2DCa2+ DSO24− 2 (0.9 × 10 −9 ) (1.1 × 10 −9 )

DCaSO4 = = = 0.95 × 10 −9 (7.40)
DCa2+ + DSO24− 0.9 × 10 −9 + 1.1 × 10 −9
DCaSO4 is 0.95 × 10−9 m2/sec and gypsum solubility is 15 × 10−3 mol/kg (cs = 15
mol/m3) (Dutrizac, 2002). Using these values, Eq. (7.39) predicts the concentra-
tion versus distance profiles shown in Figure 7.4.
Transport-limited dissolution
Reactions are limited by the slowest step in the overall process. If the reac-
tion flux (JR, mol/m2sec) for a dissolving solid is fast compared to the dif-
fusion flux (JD, mol/m2sec), the dissolution process is limited by the rate of
transfer of reactants or products between the solution and the solid’s sur-
face. When JR ≈ JD, the overall rate of dissolution is controlled by mixed
kinetics, which is modeled by convolving the models for both the dissolution
and transport process.
The second Damköhler number, Eq. (7.18), is often used to test whether
a process is limited by diffusive transport. DaII is the ratio of the rate of
reaction at the surface (numerator) to the rate of transfer between the sur-
face and the solution (denominator). If DaII >> 1, the surface reaction is
fast relative to the diffusion rate. If DaII << 1, the surface reaction is slow
relative to the transport rate, so the overall rate is reaction limited. If DaII
Transport-limited dissolution 141
≈ 1, the surface reaction rate is nearly the same as the transport rate, so the
overall rate is controlled by mixed kinetics. The DaII test applies only for no
flow situations.
Example 7.3. Gypsum dissolution into a static fluid
At Earth surface conditions, gypsum (CaSO4•2H2O) dissolution appears to be

controlled partly by the rate of Ca2+ and SO42− release at the mineral surface and
partly by the rate of transport of these species away from the dissolving surface.
The second Damköhler number, Eq. (7.18), can be used to determine the relative
importance of reaction rate versus diffusion rate for gypsum dissolving into a
static solution in a fracture with a width, L, of 1.0 × 10−3 m.
A/V = 2/L = 2.0 × 103 m2/m3 (Table 6.1)

k+ = 7.0 × 10−5 mol/m2sec (Colombani, 2008)
L = 1.0 × 10−3 m
DCaSO4 = 9.54 × 10−9 m2/sec (Example 7.2)
ceq = 15 mol/m3 (Dutrizac, 2002)
( A V ) k+ L2 = (2.0 × 103 )(7.0 × 10 −5 )(1.0 × 10 −3 )

2
Da II = = 0.99
Di ceq (9.45 × 10 −9 )(15)
This means that for a 1 mm (0.001 m) wide fracture, the rate of gypsum dis-
solution nearly equals the rate of diffusive transport away from the dissolving
surface. For narrower fractures, DaII < 1, which means that the rate of diffusive
transfer is fast compared to the dissolution rate. DaII > 1 for wider fractures,
which means that the reaction rate is faster than the diffusive transfer rate.
Mass transfer coefficients are the basis for models where the dissolved
species are transported by a combination of diffusive and advective pro-
cesses. The diffusive mass transfer coefficient (kD, m/sec) is based on bound-
ary layer theory. The basic premise of boundary layer theory is that, for
laminar flow, the fluid velocity adjacent to a solid surface is zero (the “no
slip condition”) and the velocity increases as a parabolic function of dis-
tance away from the surface until it matches the velocity of the bulk fluid
(Figure 7.5). This means that there is a thin layer of fluid with a thickness
of δD (m) adjacent to the surface that is effectively static. The rate of mass
transport through this layer is limited by the diffusion rate of the dissolved
species. The diffusional boundary layer is much thinner than the velocity
boundary layer. For laminar flow past a flat surface, the thickness of the dif-
fusional boundary layer is related to the thickness of the velocity boundary
layer (δV) by the Schmidt number, which compares the fluid viscosity to the
diffusivity (Probstein, 1989).
vf
flow velocity
Figure 7.5. The velocity solid

boundary layer thickness 0
(δV) and the diffusional
ceq
boundary layer thickness
(δD) for laminar flow cs
past a flat surface. The
concentration
diffusional boundary layer

is approximated as a thin
solid
layer of static fluid at the

solid surface where only
diffusional mass transport
c
occurs. See discussions
in Probstein (1989), Denny
(1993), and Vogel (1994) for 0
0 V
more details. D
distance
0.6
δD = δV (7.41)
Sc1 3
The Schmidt number appears in most formulations of diffusion mass trans-

fer coefficients (Table 7.3).
The Nernst film model is used to quantify diffusional transport through
the “static” boundary layer. This model approximates the low velocity
boundary layer as a thin, static film between the surface and the free flowing
solution. Fick’s first law gives the diffusional flux (JD, mol/m2sec) through
this film (Figure 7.6).
Di
JD = (cs − c ) (7.42)
δD
The concentration gradient is the surface concentration (cs, mol/m3) minus

the bulk solution concentration (c, mol/m3) divided by the thickness of
the diffusion boundary layer (δD, m). Di (m2/sec) is the diffusion coefficient
and the D/δD term in Eq. (7.42) is recast as a mass transfer coefficient, kD
(m/sec).
Table 7.3. Models used to estimate kD for commonly encountered liquid/solid

interface geometries adapted from Table 8.3–3 in Cussler (2009). D (m2/sec) is
the diffusion coefficient; q (m/sec) is the Darcy velocity of the fluid; and ν
(m2/sec) is the kinematic viscosity.
Interface Equation Notes

1 1
Laminar flow along flat  D   Lq  2
ν 3 L = plate length
kD = 0.646      
plate  L  ν  D
−0.42 23
Packed bed  Lq  ν L = grain diameter
kD = 1.17q    
 ν  D
1
Turbulent flow through  D   Lq 
0.8
ν 3 L = pipe diameter
circular pipe kD = 0.026      
 L  ν  D
0.8
Laminar flow through  D   L2 q  L = pipe diameter
kD = 1.62   
circular pipe  L   L p D  Lp = pipe length
 D   Lq  2  ν  3 
1 1
Forced convection kD =    2.0 + 0.6      L = sphere diameter
around a sphere  L    ν   D  
Free convection around   L3 g ρ  4  ν  3 

1 1
L = sphere diameter
 D

kD =   2.0 + 0.6    
a sphere  L 
  ρν 2   D   g = gravitational
acceleration
Turbulent flow through 0.8 1 L = (2/π)(slit width)
 D   Lq  ν 3
a horizontal slit kD = 0.026      
 L  ν  D
Transport through a  D L = membrane
kD =  
membrane  L thickness
Spinning disk 1 1 L = disk diameter
 D   L2ω  2
ν 3
kD = 0.62      ω = rotation rate
 L   ν  D
(radians/sec)
J D = kD (cs − c ) (7.43)
The flux of material into or out of the boundary layer due to reactions at the
surface (JR, mol/m2sec) is driven by the difference between the surface con-
centration (cs, mol/m3) and the equilibrium concentration (ceq, mol/m3) and
is proportional to the dissolution rate constant (k+, mol/m2sec).
 ceq − cs 
J R = k+   (7.44)
 ceq 
The right-hand side of this relationship is equivalent to k+(1 − Q/K), so this

equation is valid for both dissolution and precipitation at near-equilibrium
conditions. Equation (7.44) can also be written in terms of a reaction mass
transfer coefficient (kR, m/sec).
k+
JR =
ceq
(ceq − cs ) (7.45)
J R = kR (ceq − cs ) (7.46)
When a solid is immersed in a solution it will eventually reach a steady-state

dissolution rate where the rate of reaction at the surface equals the rate of
transport through the static film; JD = JR. This means that Eq. (7.43) can be
set equal to Eq. (7.46).
kR (ceq − cs ) = kD (cs − c ) (7.47)
The surface concentration can be found by rearranging Eq. (7.47).
kRceq + kDc
cs = (7.48)
kR + kD
This definition of surface concentration can be substituted into either Eq.

(7.43) or Eq. (7.45) to find the combined diffusion and reaction flux (J, mol/
m2sec) at steady state.
kR kD
J=
kR + kD
(ceq − c ) (7.49)
According to this equation, if kR >>kD, then kR + kD ≃ kD so J = kD(ceq − c),

which means that the flux to or from the solution is controlled by diffusion.
If kD >> kR, then kR + kD ≃ kR so J = ks(ceq − c) and the flux to the solution
is controlled by the surface reaction rate.
Values of kD are influenced by the interface geometry and the fluid flow
velocity. Table 7.3 lists some kD models for geometries that might be useful
to geochemists. The estimated accuracy of most of these models is on the
order of ±10% but much larger uncertainties are possible (Cussler, 2009).
Example 7.4. Gypsum dissolution as a function of fluid velocity
At Earth surface conditions, gypsum (CaSO4•2H2O) dissolution rates

are controlled by a combination of dissolution rate and mass transfer (see
Example 7.3). The relative control by each of these processes depends upon
the fluid flow velocity. Equation (7.49) can be used to model the overall rate
–3
JD
–4
JR
log J
–5
–6
10–8 10–6 10–4 10–2
q, m/sec
Figure 7.6. Dissolution flux of gypsum into pure water flowing past a flat plate.
The line labeled JR is the dissolution flux due to surface reaction only and
the line labeled JD is for the diffusion flux only.
in terms of the combined effects of dissolution and transport control. This

model assumes laminar flow past a flat plate with L = 0.001 m and q = 0.0001
to 0.1 m/sec.
The data needed for this model are given below.
c = 0 mol/m3
ceq = 15 mol/m3 (Dutrizac, 2002)
k+ = 7.0 × 10−5 mol/m2sec (Colombani, 2008)
so kR = k+/ceq = 4.67 × 10−6 m/sec
DCaSO4 = 9.54 × 10−9 m2/sec (Example 7.2)
ν = 8.94 × 10−7 m2/sec (Watson et al., 1980) (ρ = 1000 kg/m3)
1 1
 D   Lq  2 ν 3
kD = 0.646  i     D 
 L ν  i
1 1
 9.45 × 10 −9   1 × 10 −3 q  2
 8.94 × 10 −7  3
kD = 0.646  (7.50)
 1 × 10 −3   8.94 × 10 −7   
9.45 × 10 −9 
1
kD = 9.52 × 10 −4 q 2
Inserting these values into Eq. (7.49) produces the results shown in Figure 7.6.
It is relatively easy to appreciate the mixed kinetics of this example. When
kR >>kD, kR + kD ≃ kD and J = kDceq and when kD >> kR, ks + kD ≃ ks and
J = kRceq.
Effect of temperature on mixed kinetics

The diffusion coefficient, the dynamic viscosity, and the mineral disso-
lution rate all change with temperature. Mineral dissolution reactions
have higher activation energies than diffusion processes. This means that
increasing the temperature increases the mineral dissolution rate more
than the diffusion flux so that mineral dissolution and precipitation pro-
cesses become limited by the relatively slower diffusion fluxes at high
temperatures.
Example 7.5. Gypsum dissolution flux as a function of temperature
Modeling the effect of temperature on gypsum dissolution requires temperature

functions for the variables in Eq. (7.49).
ceq = 14.4 + 5.88 × 10 −2 t − 8.84 × 10 −4 t 2 (7.51)
where t = temperature, °C, Dutrizac (2002).
temperature, °C
100 50 25 0
–2
–3
log J, mol/m2 sec
JR
–4
JD
–5
0.0025 0.003 0.0035

1/T, K–1
Figure 7.7. Comparison of the CaSO4 fluxes predicted for reaction limited

dissolution, kRceq, transport limited dissolution, kDceq, and mixed kinetics
(kRkD/ kRkD)ceq.
Coating growth model 147
The 25°C dissolution rate constant, 7 × 10−5 mol/m2 sec (Colombani, 2008), can
be extrapolated to other temperatures using the Arrhenius equation and an acti-
vation energy of 41.8 kJ/mol (Liu and Nancollas, 1971).
Ea  1 1 2183 2183
log kT2 = log kT1 +  −  = −4.15 + 7.32 − = 3.17 − (7.52)
2.303R  T1 T2  T2 T2
The diffusion coefficient for CaSO4 (DCaSO4 ) as a function of temperature is the

harmonic mean of DCa and DSO4 calculated from the temperature functions in
Table 7.1.
The kinematic viscosity of water (ν) is calculated from the temperature func-
tion for the dynamic viscosity given by Eq. (7.1) and the temperature function
for the density given by Eq. (7.2).
This model assumes laminar flow past a flat plate with L = 0.001 m and q =
0.001 m/sec (Table 7.3). Figure 7.7 shows that the rate of gypsum dissolution
changes from reaction limited at T < 0°C to transport limited for T > 100° C.
The apparent activation energy for the mixed kinetics model ranges from ~22 kJ/
mol at 0°C to ~5 kJ/mol at 100°C.
Examples 7.4 and 7.5 show that diffusive transport tends to become a
limiting process for geochemical reactions as flow rates decrease and tem-
peratures increase. This means that although chemical reaction rates often
limit geochemical processes at or near the Earth’s surface, diffusional trans-
port becomes more rate limiting at depth because of higher temperatures
and slower flow rates due to lower rock permeability.
Coating growth model
The rate of interaction between minerals and solution species is sometimes
limited by the diffusion of a reactant from the solution through a coating
of insoluble reaction products (Figure 7.8). The diffusion rate through this
coating can be many orders of magnitude lower than the solution diffusion
rate, so that the rate of delivery of the reactant to the mineral surface is
much slower than its rate of reaction with the mineral. The thickness of the
coating (x, m) is a function of the number of moles of coating precipitated
per square meter (np, mol/m2), the molar volume of the coating (Vm, m3/mol),
the surface area of the grains (A, m2), and the porosity of the coating (φ).
The coating is likely to be porous if the molar volume of the coating is less
than the molar volume of the mineral or if some of the dissolving constitu-
ents are lost to the solution. This model is only applicable when x is much
smaller than the grain diameter.
cr
concentration
coating
solution
solid
Figure 7.8. Conceptual
model showing how the
rate of reaction between
a mineral with a coating
of reaction products is
controlled by the rate of
diffusion of a reactant from
the surrounding solution 0
through the coating to the 0 x
surface. distance
n pVm
x= (7.53)
( − φ)
1
The flux of reactant, Jr (mol/m2sec), through the coating is a function of

the diffusion coefficient of the reactant in the coating, D (m2/sec), and its
concentration in the surrounding solution, cr (mol/m3).
D
Jr = cr (7.54)
x
Although the reactant concentration at the mineral’s surface is finite, we will

assume that it is so small relative to the solution concentration that it can be
set to zero. The flux of constituents from the mineral into the coating is Jp =
νr Jr (mol/m2sec), where νr = moles of coating formed per moles of mineral
dissolved. Combining this relationship with Eqs (7.53) and (7.54) gives an
expression for the rate of precipitate formation.
 ν Dc (1 − φ )  1
Jp =  r r  n (7.55)
 Vm p
The terms in the parentheses can be combined into a rate constant (kp,
mol2/m4sec).
ν r Dcr (1 − φ )
kp = (7.56)
Vm
Equation (7.55) can be rewritten in terms of kp and np.
dn p k p
= (7.57)
dt np
This equation can be integrated with the boundary conditions that np = 0

when t = 0 and np = np when t = t.
np t
∫ n dnp p = k p ∫ dt (7.58)
0 0
n 2p
= k pt (7.59)
2
This equation can be solved for np.
np = ( )
2 k p t1 2 (7.60)
The thickness of the coating increases with the square root of time.
x = Vm n p = Vm ( )
2 k p t1 2 (7.61)
The rate of coating formation is the time derivative of Eq. (7.60).
 kp 

dn p 1
dt
=
2
( )
2 k p t −1 2 = 
 2 
−1 2
t (7.62)
This means that the thickness of the coating increases as a linear function
of t−½.
dx  kp 
= Vm  −1 2
t (7.63)
dt  2 
As the coating grows, the rate of consumption of the reactant declines as a

linear function of the square root of time.
dnr  1   k p  −1 2
=  t (7.64)
dt  ν r   2 
This means that a graph of the rate of reactant consumption versus t−½ is
a straight line with a slope of (1/νr)(kp/2)½. The diffusion coefficient for the
reactant in the coating can be calculated from this slope by rearranging Eq.
(7.56). This equation is only valid after enough time has passed to allow a
sufficiently thick coating to develop so that the flux of the reactant to the
mineral surface is smaller than its rate of consumption.
Example 7.6. Iron oxyhydroxide coating on pyrite for AMD mitigation
Huminicki and Rimstidt (2009) suggested that treating pyritic mine wastes with
bicarbonate solutions would produce an iron oxyhydroxide coating on the pyrite
surfaces, which would inhibit the development of acid mine drainage. The over-
all reaction converts iron from the pyrite into an iron oxyhydroxide coating.
FeS2(pyrite) + 3.75 O2 + 3.5 H2O + 4 HCO3−
= Fe(OH)3 (iron oxyhydroxide) + 4 H2CO3 + 2 SO42−
The iron oxyhydroxide coating acts as a barrier to the transport of dissolved

oxygen from the solution to the pyrite surface and thereby reduces the rate of
pyrite oxidation. They report, based on the experiments of Nicholson et al.
(1990), that the dissolved oxygen diffusion coefficient through the iron oxyhy-
droxide coating is ~2 × 10−17 m2/sec, which is about 8 orders of magnitude slower
than for pure water. Equation (7.61) can be used to predict the coating thickness
as a function of time and Eq. (7.64) can be used to find the rate of O2 con-
sumption as the coating becomes thicker. The following information is needed
to compute kp.
νr = number of moles iron oxyhydroxide/number of moles of O2 reacted =

1/3.75 = 0.267
D = 2 × 10−17 m2/sec (Huminicki and Rimstidt, 2009)
φ = 0.10 (assuming that the coating has a porosity of 10%)
Vm = 2.72 × 10−5 m3/mol
cr = 0.27 mol/m3 (dissolved O2 concentration in air-saturated water)
ν r Dcr (1 − φ ) (0.267 ) (2 × 10 −17 ) (0.27 )(0.9) mol2

kp = = = 4.77 × 10 −14 4
Vm 2.72 × 10 −5
m sec
The coating thickness as a function of time shown in Figure 7.9a is calculated

using Eq. (7.61).
x = Vm ( )
2 k p t1 2 = (2.72 × 10 −5 ) ( )
2 ( 4.77 × 10 −14 ) t1 2 = 8.40 × 10 −12 t1 2
The oxygen consumption rate as a function of time shown in Figure 7.9b is cal-

culated using Eq. (7.64).
dnr  1   k p  −1 2  4.77 × 10 −14  1 2

=   t = (0.267 )   t = 4.12 × 10 t
−8 −1 2

dt  ν r   2   2 
(a)
3.0×10–6
2.0×10–6
thickness, m
1.0 ×10–6
0
0 10 20 30 40 50
time, weeks
(b)
1.5×10–11
(b)
o×ygen consumption rate, mol/m2 sec
1.0×10–11
0.5 ×10–11
0
0 10 20 30 40 50
time, weeks
Figure 7.9. (a) Thickness of an iron oxyhydroxide coating on pyrite undergoing

oxidation in a solution buffered to a near-neutral pH. (b) Rate of oxygen
consumption by the pyrite oxidation reaction as a coating of iron oxyhydroxide
grows in thickness.
The rate of dissolved oxygen consumption by pyrite oxidation in air-saturated

water at pH = 7 is 4.80 × 10−10 mol/m2sec based on the rate equation from
Williamson and Rimstidt (1994). The model predicts that oxygen flux through
the coating after 1 week, 8.83 × 10−13 mol/m2sec, is only about 0.2% of this rate.
Shrinking core models

As the coating on a reacting particle grows thicker, the diameter of the
unreacted core decreases and the resulting decrease in the interfacial area
between the core and the coating further reduces the rate of delivery of reac-
tant. Figure 7.10 illustrates the geometry of a particle with a shrinking core.
The rate of conversion of the core into coating (dn/dt, mol/sec) is a function
of the radius of the core (R, m), the stoichiometric coefficient (∆νi = moles
coating formed per moles of core destroyed), the diffusion coefficient of the
reactant through the coating (D, m2/sec), and the concentration of the reac-
tant in the surrounding solution (c, mol/m3).
dn 4π R 2  dc 
=− D (7.65)
dt νr  dR 
If the diffusion rate of reactant through the coating is fast compared to the
motion of the core/coating interface, a steady-state concentration gradient
will develop so that Eq. (7.64) can be integrated across R to find the rate of
reaction in terms of R and Ro.
dn 4π Dc  RRo 
= (7.66)
dt ν r  Ro − R 
coating
unreacted
core
Ro
Figure 7.10. Schematic
diagram showing the
conceptual basis for the
shrinking core model.
Shrinking core models 153
The number of moles of substance in the particle is n = 4πR3/3Vm. The

fraction reacted (α) is related to the ratio of the radii of the core and par-
ticle cubed.
R3
α = 1− (7.67)
Ro3
This equation can be differentiated with respect to time.
d α 3R 2  dR 
= 3   (7.68)
dt Ro  dt 
Substituting Eqs (7.67) and (7.68) into Eq. (7.66) gives the rate of trans-
formation in terms of the fraction of the particle reacted.
d α 3VmDc (1 − α 1 3 )
= (7.69)
dt ν r Ro2 1 − (1 − α 1 3 )
This equation can be integrated using the boundary conditions that α = 0

when t = 0.
2  2V Dc   2VmDc 
1 − α − (1 − α ) =  m 2  t = k pt
23

3  νr Ro   k p = ν R 2  (7.70)
r o
A graph of 1 − 23 α− (1 − α)⅔ versus t has a slope of kp. Typically kp is found

from experimental data and then used to find D. There are other shrinking
core models derived for different geometries and with different boundary
conditions (Heizmann et al., 1986; Wen, 1968).
The complexity of Eq. (7.70) creates difficulties for using it in forward
models to find α as a function of t. It is easier to graph t as a function of α
and then use the graph to find α at the time of interest.
Example 7.7. Development of a limonite pseudomorph after pyrite
Pyrite crystals that oxidize in an environment with a near-neutral pH and with

enough alkalinity to neutralize the acid produced by the oxidation reaction typ-
ically become replaced by a mixture of iron oxyhydroxide minerals, mostly goe-
thite (FeOOH). This mixture is sometimes referred to as “limonite”.
FeS2(pyrite) + 3.75 O2 + 2.5 H2O + 4 HCO3−
= FeOOH(goethite) + 4 H2CO3 + 2 SO42−
The rate of conversion of pyrite to limonite is controlled by the diffusive rate of

transfer of dissolved oxygen through the limonite coating. The fraction of pyrite
0.8
0.6
0.4
0.2
0
0 20 40 60 80 100
time, yr
Figure 7.11. Fraction of pyrite converted to goethite as a function of time for a

1 cm diameter (Ro = 0.5 cm) pyrite grain.
converted to goethite, α, can be modeled as a function of time using Eq. (7.70)

with the following information.
Vm (py) = 2.08 × 10−5 m3/mol; Vm(goe) = 2.39 × 10−5 m3/mol
Vm ( goe ) 2.08 × 10 −5
φ = 1− = 1− = 0.13
Vm ( py ) 2.39 × 10 −5
τ 2 = 1 − 2.02 ln φ = 1 − 2.02 ln (0.13) = 9.24
DO2 = 2.3 × 10 −5 m 2 /sec (Lerman, 1979)
DO2 2.3 × 10 −9
Dτ O2 = = = 2.49 × 10 −10 m 2 /sec
τ2 9.24
cr = 0.27 mol/m 3
(cr = dissolved O2 concentration in air-saturated water)

moles goethite formed 1
νr = = = 0.267
moles O2 reacted 3.75
Ro = 0.005 m = 0.5 cm
Shrinking core models 155
2VmDc 2 (2.08 × 10 −5 ) (2.49 × 10 −10 ) (0.27 )

kp = = = 1.05 × 10 −10 sec −1
ν r Ro2 (0.267)(0.01)2
Using these values we can find the time needed to convert various fractions of a
pyrite grain into the goethite coating (Figure 7.11).
2 2
1 − α − (1 − α ) 1 − α − (1 − α )
23 23
t= 3 = 3
kp 1.05 × 10 −10
This model shows that the rate of conversion of pyrite to goethite is quite fast
during the initial stages of the process so that 50% of the pyrite is converted dur-
ing the first 11 years of exposure to oxidizing conditions. As the layer of goethite
grows thicker, the diffusion distance becomes long and the rate of conversion
slows so that it takes an additional 89 years to convert the remaining pyrite to
goethite.
Chapter 8
Quasi-kinetics
Quasi-kinetic models deal with processes that are controlled by mass trans-
fer rates rather than by chemical reaction rates. These models assume nearly
instantaneous attainment of equilibrium within the region of interest, so
changes in the species distribution are controlled by the rate of transfer of
substances into or out of that region. These models are constrained by con-
tinuity equations making them similar to the chemical reactors models in
Chapter 4.
Local equilibrium assumption

Most of the models considered in this chapter rely on the local equilibrium
assumption. This assumption requires that the rates of chemical reaction
and local mass transfer within the model’s spatial domain are fast relative to
the residence time of a slug of solution within that domain. Knapp (1989)
and Bahr and Rubin (1987) have evaluated conditions where the local equi-
librium assumption is valid and the Knapp treatment is summarized by Zhu
and Anderson (2002).
For the local equilibrium assumption to be valid, both the mineral and
solution reaction rates and the transport rate to and from the minerals’ sur-
faces must be fast. These constraints are best tested using the first Damköhler
number, DaI, and the Péclet number, Pe. DaI compares the rate of consump-
tion (or production) of a species by chemical reaction to the rate of delivery
(or removal) of that species by advection.
Da I =
RL
=
{(A M ) J } L (8.1)
meqv meqv
R (molal/sec) is the reaction rate and meq (molal) is the equilibrium con-
centration of the reactant. v (m/sec) is the surficial velocity of the fluid and L
is the characteristic length of the domain. A rule of thumb is that when DaI
< 0.1, less than 10% of the aqueous species will react as the solution transits
L; and when DaI > 10, more than 90% of the aqueous species will react over
156
Local equilibrium assumption 157
this distance. Even if the reaction rate is fast, the rate of reaction between
solution and the mineral might still be limited by the rate of diffusive and
dispersive transport to or from the minerals’ surfaces. The Péclet number,
Pe, compares the rate of species transport to or from the minerals’ surfaces
by diffusion or dispersion to the rate that they are added to or removed from
the domain by advection. The Péclet number is the ratio of mass transfer
by advection to the rate of mass transfer to or from the minerals’ surfaces
expressed by the sum of the diffusion coefficient (Di, m2/sec) and the longi-
tudinal dispersion coefficient (DL, m2/sec).
vL
Pe = (8.2)
Di + DL
When Pe < 50, diffusion and dispersion dominate and aqueous species are
delivered to or removed from the surface faster than they are added to or
removed from the domain by advection. When Pe > 100, the rate of advec-
tion is so fast that the reacting species are transported into or out of the
domain before they can migrate to or from minerals’ surfaces. Figure 8.1
shows a map of Da and Pe. The local equilibrium assumption is valid in the
upper right corner of the map where Da > 10 and Pe > 100 but reactions
will be unlikely to reach equilibrium for the other conditions shown because
they are either transport or reaction limited, or both.
Example 8.1. Equilibration of dissolved silica with quartz in a fracture
The quartz geothermometer (Fournier, 1977) requires that the dissolved silica
concentration in an ascending hydrothermal solution becomes “quenched in” as
the fluid approaches the surface. On the other hand, the phase transfer model for
the rate of quartz deposition, described in this chapter, requires that the LEA be
valid. The conditions for which each of these models is appropriate can be found
by calculating values of DaI and Pe. Figure 8.1 shows the temperature range for
the transition from LEA to quenched conditions.
The rate of quartz precipitation, Rp (molal/sec), (Rimstidt and Barnes, 1980)
is needed to calculate DaI.
Rp =
A
M (
k+ 1 − Q K ) (8.3)
For a fracture, the surface area to mass of solution ratio (A/M = 2/1000ρW)
depends on the fracture width (W, m) (see Table 6.1). This model considers a
500 µm wide fracture so (A/M) = 4. To simplify the calculations, the solution
density (ρ) is set equal to one over the entire temperature range and the degree
of supersaturation (Q/K) is set equal to two, which makes Rp = −(A/M)k+. The
158 Quasi-kinetics
8
10
6 LEA Valid
10
Fast reaction Fast reaction
4
Slow transport Fast transport
10
2
10 300°C
200°
1
Da
100°
–2
10 25°
–4 Slow reaction
10 Slow transport
Slow reaction
–6 Fast transport
10
–8
10
5
0.01 0.1 11 0 100 1000 10 4 10
Pe
Figure 8.1. Map of DaI versus Pe that shows that the Local Equilibrium
Assumption (LEA) is valid for DaI >~10 and Pe > ~100. The dots are values of the
DaI and Pe values for quartz precipitation in a fracture at different temperatures
calculated in Example 8.1. They show that the LEA is valid for T > ~250°C.
rate constant, k+, for each temperature is computed from Rimstidt and Barnes
(1980) and Rimstidt (1997b).
log k+ = − (3705 T ) − 0.372 (8.4)
Rp = ( 4 )(0.185) (10 −3705 T ) (8.5)
The equilibrium quartz concentration, meq, is calculated from Rimstidt

(1997b).
log m = − (1107 T ) − 0.025 (8.6)
We will consider a characteristic length, L, of 10 m and an ascension rate, v, of

1 × 10−5 m/sec, which is approximately 1 m/day.
It is useful to find DaI at 25°C to illustrate the calculation procedure.
Rp = 2.77 × 10−13
L = 10 m
Local equilibrium assumption 159
meq = 1.83 × 10−4 mol/kg

v = 1 × 10−5 m/sec
Da I =
RL
=
(2.77 × 10 −13 )(10) = 1.51 × 10 −3 (8.7)
meqv (1.83 × 10 −4 ) (1 × 10 −5 )
Note that if L = 100 m, DaI = 1.5 × 10−2, then DaI scales directly with L. This
value of DaI falls in the “slow reaction” part of Figure 8.1, indicating that the
solution does not reach equilibrium within a 10 or even 100 m fracture length
because of the very slow precipitation rate of quartz.
The diffusion coefficient for dissolved silica, Di (m2/sec), and the longitudinal
dispersion coefficient, DL (m2/sec), are needed to compute the Péclet number.
Di is calculated from equation (5) in Rebreanu et al. (2008) and DL is calculated
from the fracture width, using the Horne and Rodriguez (1983) model.
DL =
2  2 ( )
 W 2 v2 
 (8.8)
105  Di 


vL
Pe = (8.9)
DL + Di
Calculating Pe at 25°C requires the following information.

L = 10 m
v = 1 × 10−5 m/sec
Di = 1.1 × 10−9 m2/sec
DL =
2  ( )
 W 2 v2 
2  =
2  ( )
 0.0005 2 1 × 10 −5 2 
2 ( )
= 4.3 × 10 −9 (8.10)

105  Di  105  1.1 × 10 −9 
   
vL
Pe = = 2.3 × 10 4 (8.11)
DL + Di
This result means that the transport of silica to the quartz surface by dispersion
and diffusion is very fast and is not rate limiting.
Values of DaI and Pe calculated for other temperatures are plotted on
Figure 8.1. The figure shows that the LEA holds for high temperatures (>~250°C)
but fails for T <~200°C. This is consistent with using the quartz geothermom-
eter to find the temperature of geothermal reservoirs. The quartz geothermom-
eter is based on the assumption that the dissolved silica concentration remains
unchanged as the solution flows to the surface (Fournier, 1977).
160 Quasi-kinetics
Box models
Some models assume that a system reaches a steady state rather than equi-
librium. Equilibrium is defined by the principle of detailed balance, which
requires that the forward and reverse rates are equal and that each step along
the reaction path is reversible. The forward and reverse rates of steady-
state processes are equal but the process steps that produce the forward
rate are different from those that produce the reverse rate. At steady state,
the state variables of an open system remain constant even though there is
mass and/or energy flow through the system. The steady-state assumption
is especially useful for processes that occur in a series, because the concen-
trations of intermediates that are formed and subsequently destroyed are
constant. Perturbation of a steady-state system produces a transient state
where the state variables evolve over time and approach a new steady state
asymptotically.
Box models are steady-state models that are used to follow the flow of
a conserved substance between reservoirs in the geological environment
(Lasaga, 1980b; Lasaga and Berner, 1998; Lerman and Wu, 2008). Box
models divide the environment into various reservoirs, represented by boxes
(Figure 8.2), each of which represents a major part of the overall system. If
the model considers cycling of a substance throughout the entire Earth, the
reservoirs might be the oceans, the atmosphere, the crust, etc. For example,
Figure 8.3 shows the main reservoirs and fluxes for the global water cycle
(Berner and Berner, 1987). It is presumed that each of these reservoirs is
well mixed, so the concentration of the substance is approximately the same
throughout. Material is transferred from one reservoir to another via fluxes
(F) of the substance, which are represented by arrows. Simply constructing a
diagram shows how a substance flows through the system (Figures 8.2a and
8.3). In addition, box models are used to quantify how the overall system
might respond to a perturbation.
The simplest box model consists of one reservoir that is fed by a con-
stant flux (F0) (Figure 8.2). The flux out of a reservoir is the product of the
mass of the substance (M) in the reservoir and a mass transfer constant
(k). Box model reservoirs are simply gigantic ideal mixed flow reactors (see
Chapter 4) where there is no net generation or consumption of the sub-
stance. Over geologic time spans these reactors tend to attain a steady state
so that the flux into a reservoir is matched by the flux out, which means that
the rate of accumulation in the reservoir is zero and M is constant (Mss). The
flux out of the reservoir equals a mass transfer constant (k, yr−1) times the
mass of the substance in the reservoir.
F0 = F = kM ss (8.12)
Box models 161
(a)
reservoir
flux in flux out
F0 F1 = k 1 M 1
M1
(b) 12
Figure 8.2. (a) Schematic
illustration of a box model
Mss with a constant flow (Fo) of
10
material into the reservoir,
which contains M amount
8 of material. The flux out
of the reservoir (F1) is
a function of the mass
6 transfer constant (k1) and
M
tc = 3 of the amount of material

in the reservoir (M1). (b)
4 Illustration of the transient
behavior of a single,
initially empty, reservoir
2 as it returns to a steady
tc = 1 state M of 10. The reservoir
refills to 63% of the steady-
0 state value when one time
0 2 4 6 8 10 constant (tc= 1/k = 1) has
t passed.
F0 F
k= = (8.13)
M ss M ss
Once the mass transfer constant is known, it is possible to model the effects
of a perturbation on the reservoir contents and output. The rate of change
of mass of material in the reservoir is the difference between the flux into
and the flux out of the reservoir.
dM
= F0 − kM (8.14)
dt
This equation can be rearranged and integrated from an initial condition

where M = Mo when t = 0.
dM
dt = (8.15)
F0 − kM
162 Quasi-kinetics
atmosphere
13 10 9 kg
vapor transport
110 10 9 kg/yr
386 10 9 kg/yr
423 10 9 kg/yr
precipitation
73 10 9 kg/yr
precipitation
evaporation
37 10 9 kg/yr
evaporation
ice
29000 10 9 kg
lakes & rivers runoff ocean

Figure 8.3. Box model of
130 10 9 kg 1370000 10 9 kg
the water cycle (Berner 37 10 9 kg/yr
and Berner, 1987) showing
groundwater
the major reservoirs and
9500 10 9 kg
fluxes.
t M
dM
∫ dt = ∫ F0 − kM
(8.16)
0 M0
 1   F − kM 
t =  −  ln  0 (8.17)
 k   F0 − kM 0 
This equation can be solved for M as a function of t.
F0 − kM
e − kt = (8.18)
F0 − kM 0
F0  F0 
M= −  − M 0  e − kt (8.19)
k  k 
Figure 8.2b, which is based on Eq. (8.19), shows how an empty reservoir
returns to the steady state. During the initial stage of the process the res-
ervoir contains very little of the substance, so the rate of removal by kM
is low and M grows quickly. However, increasing M over time causes kM
to increase and that causes the rate of growth of M to slow. Eventually, M
becomes large enough to cause kM to equal F0 and M has returned to Mss.
Because M approaches Mss as an asymptote, it is not possible to mathem-
atically define a time for the return to steady-state conditions. However, it
is possible to find a practical time by defining a time constant (tc) for the
system.
Box models 163
1
tc = (8.20)
k
If 1tc is substituted into Eq. (8.19), the calculated value of M is 63% of Mss.
At 3tc, M is 95% of Mss and at 5tc, M is 99% of Mss. Somewhere between 3tc
and 5tc the difference between M and Mss becomes smaller than the uncer-
tainty in determining M so that there is no longer a way to distinguish M
from Mss and the system has effectively reached steady state.
Although the box models treat substances as a continuum, they actually
consist of discrete particles (atoms, molecules, clusters, etc.) and it is useful
to consider the average time that a single particle might remain in a reser-
voir. Some particles will persist in the reservoir for longer times and some
for shorter but the average time spent in the reservoir is defined by the resi-
dence time (tres).
M ss
tres = (8.21)
F0
For a one-reservoir model, tres = tc; see Eq. (8.13). Residence-time analysis is
a useful way to understand the behavior of tracers, such as tritiated water, in
geochemical cycles. Residence-time analysis methods for more complicated
situations are described in many chemical engineering textbooks (Hill Jr.,
1977).
Although one-box models are informative, most geochemical cycles con-
sist of multiple reservoirs coupled by many fluxes; see Figure 8.3 for exam-
ple. Models for multiple reservoir–multiple flux situations and for coupled
geochemical cycles are somewhat more complicated but are well under-
stood (Lasaga, 1980a, 1980b; Lasaga and Berner, 1998; Lerman and Wu,
2008). One important insight that comes from these more complex models
is that increasing complexity seems to increase the stability of the system.
This means that environmental perturbations caused by natural processes or
human activities tend to be damped by complex natural systems.
Example 8.2. Perturbation of the global water cycle
Figure 8.3 shows that for the global water cycle, the sum of the water evaporated
from the oceans and the continents is 496 × 109 kg/yr. At steady state the pre-
cipitation flux must have the same value. The water content of the atmosphere
can be modeled as a single reservoir with a water content of 13 × 109 kg, so k =
(496 × 109 kg/yr/13 × 109 kg) = 38.2 yr−1. For this simple one-box model, tc = tres
= 0.026 yr = 9.6 days. This means that the average residence time of a water mol-
ecule in the atmosphere is slightly less than 10 days, so we expect that the rainfall
flux should respond quickly to a change in the evaporation flux. For example, if
164 Quasi-kinetics
time, days
0 20 40 60 80
9
13.0 10
water in atmosphere, kg 12.5 10

9 evaporation atmosphere
precipitation
9
12.0 10
9
11.5 10
0 0.05 0.1 0.15 0.2 0.25
time, yr
Figure 8.4. Change in the water content of the Earth’s atmosphere in response

to a sudden 10% decrease in the evaporation rate.
atmospheric particulates from a volcanic eruption were to reduce insolation at

the Earth’s surface so that the evaporation flux decreases by 10%, F0 in Eq. (8.19)
would drop to 446 × 109 kg/yr. Figure 8.4 shows that this would cause the water
content of the atmosphere to decline over a period of about 50 days to a new
steady-state value of 11.7 × 109 kg (90% of the previous value).
Phase transfer models

Phase transfer models account for the transfer of material from one phase
to another in response to changing intensive or extensive variables. This dis-
cussion will be limited to the dissolution or precipitation of minerals in con-
tact with an aqueous solution, but the approach could be used for other sce-
narios, such as a mineral growing in an igneous melt in response to declining
temperature.
Mineral solubility is a function of temperature and pressure (Rimstidt,
1997a) as well as extensive variables such as solution composition. In this
model we assume that the extensive variables remain constant and express
the solubility of a mineral only in terms of temperature and pressure. The
amount of mineral that will dissolve or precipitate in response to changing
Phase transfer models 165
the temperature and pressure can be found by taking the derivative of the
mineral’s temperature and pressure solubility function.
 ∂m   ∂m 
dm =  dT +  dP (8.22)
 ∂T  P  ∂P  T
The partial derivatives in this equation are phase transfer coefficients,

αT and αP (not to be confused with mass transfer coefficients used in
Chapter 7).
 ∂m   ∂m 
αT =  and α P =  (8.23)
 ∂T  P  ∂P  T
If a solution is traversing a temperature gradient (Z′T = dT/dZ) and pressure

gradient (Z′P = dP/dZ), the change in solubility (dm/dZ, molal/m) along the
flow path is the sum of the solubility changes caused by the changing tem-
perature and pressure.
dm
= αT ZT′ + α P ZP′ (8.24)
dZ
If the solution has a Darcy velocity (q = dZ/dt) along the direction of the
pressure and temperature gradients, the rate of removal of the dissolved
constituent from the solution (molal/sec) is
dm
R= = q (αT ZT′ + α P ZP′ ) (8.25)
dt
Dividing R by (A/M) gives the flux (J, mol/m2sec) of the dissolved compo-
nent onto a depositional surface and multiplying J by the molar volume
(Vm, m3/mol) of the depositing mineral gives the rate of increase in thickness
(dz/dt, m/sec) of a growing layer of that mineral (see Chapter 6).
dz M
= qVm   (αT ZT′ + α P ZP′ ) (8.26)
dt  A
Example 8.3. Rate of growth of quartz veins
Most quartz veins are the result of the cooling and decompression of flowing
hydrothermal solutions. Because quartz solubility increases with increasing tem-
perature and pressure, the presence of quartz veins in an outcrop indicates that
silica-bearing solutions were flowing along a fracture system away from a hot,
high-pressure source toward a cooler, lower pressure discharge point.
166 Quasi-kinetics
0.3
0.25
0. 2
dz/dt, cm/yr
0.15 LEA valid
0. 1
0.05
0
0 50 100 150 200 250 300
T, °C
Figure 8.5. Growth rate of a layer of quartz on the wall of a fracture with a 1 cm
aperture predicted by the phase transfer model. The quartz is precipitating from
an aqueous solution flowing with a Darcy velocity of 1 m/day along a geothermal
gradient of 25°/km.
This example estimates the rate of growth of quartz on the wall of a 10 cm
wide (W = 0.1 m) fracture as the result of the upward flow of a hydrothermal
solution at the rate of 4.3 m/day along a geothermal gradient of 25°/km.
A 2
= (Table 6.1) ρ for liquid water is from Wagner and Pruß (2002)
M ρW
q = 4.3 m/day = 5 × 10 −5 m/sec
ZT′ = 25 K/km ( ~ average geothermal gradient )
m = 10(−1107.12 T − 0.0254) = e (−2549.7 T − 0.0585) ( Rimstidt,1997b )
 dm   2549.7  (−2549.7 T − 0.0585)

αT =  =
 dT   T 2 
(e )
For relatively shallow earth conditions, we can assume that pressure has little
or no effect on solubility so that αP = 0.
When these parameters are inserted into Eq. (8.26) they predict that the rate of
growth of quartz on the walls of the fracture (Figure 8.5) ranges from 0.25 cm/yr
Reaction path models 167
at 300°C (Z ≈ 11 km) to 0.02 cm/yr at 25°C (surface). This model requires that

local equilibrium is maintained (see Example 8.1) so it is only valid for high tem-
peratures (T > ~250°C). The rates for lower temperatures are too high because
the model does not account for the slow precipitation kinetics.
Quartz has prograde solubility so it precipitates as the solution cools. Many
minerals (e.g. calcite, barite, gypsum) have retrograde solubility so they precipi-
tate as the temperature increases (Rimstidt, 1997a).
Reaction path models

Geochemical systems typically contain many interacting chemical com-
ponents distributed among multiple phases, and changing the amount of
one or more components reshuffles component distribution among all the
phases. Reaction path models predict how the amounts and compositions
of phases change in response to the addition or removal of a component or
substance from the system. They can be envisioned as a titration experiment
in which the extent of reaction, ξ, is advanced by adding or removing a com-
ponent or substance in a stepwise fashion. The number of moles, ni, of each
substances is linked to the extent of reaction via the reaction coefficient, so
that each increment in ξ changes ni.
ni = nio + ν i ξ (8.27)
Reaction path models are local equilibrium models that assume that all the
reaction rates are so fast that the system is completely equilibrated at each
step. This means that the rates of formation or destruction of all the species
in the model are tied by the extent of reaction to the rate of addition, or
reaction, of the titrant.
dni dξ
= νi (8.28)
dt dt
If the rate of addition of the titrant is specified, then the rates of formation
or destruction of all the other phases can be calculated from Eq. (8.28); for
example, see Zhu and Lu (2009).
Reaction path models are unreliable whenever the local equilibrium
assumption fails. This occurs whenever slow reaction kinetics does not allow
a phase to nucleate and grow or when a phase dissolves away too slowly.
This means that predictions of reaction path models must be evaluated by
a knowledgeable geochemist to identify these refractory phases. The typ-
ical remedy for this problem is to remove the uncooperative phase from the
model and allow a more reactive, but metastable, phase to take its place.
168 Quasi-kinetics
Most reaction path models are implemented using sophisticated com-

puter programs that can predict the equilibrium redistribution of all species
as a substance is added. These computer codes can deal with very large and
complex arrays of reactions. The mathematical basis of their algorithms
is explained in Helgeson (1968), Helgeson et al. (1969), and Helgeson
(1979) and their implementation is described by Bethke (2007) and Zhu and
Anderson (2002).
Simple reaction path models can be constructed using spreadsheet pro-
grams. These models require simplifications such as assuming that the solu-
tion is dilute so the activity coefficients are unity and that only the predom-
inant species, which are used to balance the reactions, need be considered
in the model.
Example 8.4. Reaction of potassium feldspar with rainwater (from

Steinmann et al., 1994)
When rainwater interacts with a rock that contains potassium feldspar, the feld-
spar dissolves and new minerals grow until the aqueous solution comes to equi-
librium with the potassium feldspar. The reactions add K+ and H4SiO4 to the
solution and consume H+. This reaction path model tracks the changing solu-
tion composition on an activity diagram (Figure 8.6) to show how the solution
composition traverses the stability fields of gibbsite (gib), kaolinite (kaol), and
muscovite (mu) until it reaches equilibrium with potassium feldspar (Kf). The
phase boundaries are defined by five reactions.
1. kaol + 5 H2O = 2 gib +2 H4SiO4 log K1 = −8.50
2. mu + 9 H2O + H+ = gib + K+ + 3 H4SiO4 log K2 = −9.38
3. 2 mu + 3 H2O + 2H = 3 kaol + 2 K
+ +
log K3 = 6.74
4. 3 Kf + 12 H2O + 2H+ = mu + 2 K+ + 6 H4SiO4 log K4 = −14.41
5. 2 Kf + 8 H2O + 2H+ = kaol + 2 K+ + 4 H4SiO4 log K5 = −7.36
This model tracks the changes in the concentrations of H+, K+, and H4SiO4 as
small amounts of potassium feldspar are added to 1 kg of rainwater solution.
The feldspar first alters to gibbsite, then to kaolinite and then to muscovite. The
reaction for each path segment is:
νK νH νSi νgib νkaol νmu

A. Kf + H+ + 7 H2O = gib + K+ + 3 H4SiO4 1 −1 3 1 0 0
B. Kf + 2 gib +H+ = 1.5 kaol + K+ + 0.5 H2O 1 −1 0 −2 1.5 0
C. Kf + H+ 4.5 H2O = 0.5 kaol + K+ + 2 H4SiO4 1 −1 2 0 0.5 0
D. Kf + kaol + 3 H2O = mu + 2 H4SiO4 0 0 2 0 −1 1
Reaction path models 169
mu Kf
4
D
3
2
log aK /aH
1 gib kaol
0 B
A
–1
–2
–5.5 –5 –4.5 –4 –3.5 –3
log aH4SiO4
Figure 8.6. Reaction path model of solution composition as potassium feldspar is

titrated into rainwater.
The stoichiometric coefficients, νi, are moles of each species produced or con-
sumed per mole of potassium feldspar reacted.
Slightly acidic rainwater has a H+ concentration of about 3.2 × 10−5 mol/kg
(pH ~4.5) and a K+ concentration of about 3.2 × 10−6 mol/kg (0.12 ppm) and
contains about 5.8 × 10−6 mol/kg (0.1 ppm) dissolved silica.
The reaction path begins at log aK/aH = −1.0 and log aH4SiO4 = −5.2. As the
amount of K and Si in the system is incremented by the addition of potassium
feldspar, reaction A occurs followed by reactions B, C, and D. The stepwise
increase in the extent of reaction, ξ, is reflected by changes in the concentration
(activity) of the aqueous species and by changes in the number of moles of each
mineral.
mH+ = ( mH+ )o + ν H+ ξ
mK + = ( mK + )o + ν K + ξ
mH4 SiO4 = ( mH4 SiO4 )o + ν H4 SiO4 ξ
ngib = ( ngib )o + ν gibξ

170 Quasi-kinetics
nkaol = ( nkaol )o + ν kaolξ
nmu = ( nmu )o + ν muξ
The size of the ξ step must be slightly smaller than the change in hydrogen ion
concentration at the end of the reaction path otherwise its concentration will
become negative. This choice can be made by trial and error.
The resulting diagram shows that the first increments of potassium feldspar
are converted to gibbsite as the solution tracks along path A. Then the gibbsite
formed along path A dissolves and its components along with those from the
added potassium feldspar are converted to kaolinite along path B. After the
gibbsite is consumed, potassium feldspar is converted to kaolinite along path
C until the solution composition reaches saturation with respect to muscovite.
Finally, potassium feldspar is converted to kaolinite and muscovite until the
solution composition reaches the muscovite–kaolinite–potassium feldspar triple
point. At that point the solution is in equilibrium with potassium feldspar and
no further reaction takes place.
Mass balance models

Mass balance models deduce the extent of reaction between a solution
and a group of minerals in a batch reactor from the difference in the ele-
ment concentrations in the initial and final solutions. Mass balance mod-
els were developed to interpret mineral weathering reactions (Garrels
and MacKenzie, 1967) but the approach is general and the method could
be extended to many other situations. Mass balance models are often
applied to understand changes in groundwater composition due to reac-
tions that take place along the flow path. In this case, the models must
assume that the mineralogy is the same from place to place along the
flow path.
The first step in developing a mass balance model is to compute the diffe-
rence in the concentration of each element between the final and initial
solution. In an ideal batch reactor this difference is the result of reactions
between the solution and the solid phases. The next step is to postulate reac-
tions that might have caused those changes in the solution composition and
to construct a set of linear equations to represent the effect of each reaction.
Finally, the linear equations are solved to determine the extent of each of
the reactions that must have occurred. If a reasonable solution is not found,
the postulated reactions are revised until a fit is found between the reactions
and the changes in solution chemistry. This means that the model may not
produce a unique fit to the data.
Mass balance models 171
Although mass balance and reaction path models share a conceptual

basis, they are quite different in their implementation. Reaction path mod-
els begin with an initial solution and rock composition and use equilibrium
reactions to predict a final solution and rock mineralogy. Mass balance
models use initial and final solution compositions and rock mineralogy to
determine the extent of mineral solution reactions. Those reactions need not
have reached equilibrium.
Mass balance models can be extended to account for solution specia-
tion and stable isotope exchange but the computations become unwieldy
(Plummer et al., 1983). Computer codes are needed to deal with these more
complicated situations (El-Kadi et al., 2011; Plummer et al., 1991, 1992).
Example 8.5. Weathering of the Sierra Nevada batholith
Garrels and MacKenzie (1967) published the classic example of the application of
a mass balance model to account for chemical weathering in the granitic rocks of
the Sierra Nevada batholith. The model, which uses water analyses from Feth et al.
(1964), takes the chemical analysis for snowmelt as an initial water composition
and the average ephemeral spring water as a final water composition. It is presumed
that the infiltrating snowmelt dissolved CO2 from soil gases to make H2CO3 that
reacted with plagioclase, potassium feldspar, biotite, and quartz to form kaolinite.
The difference between the snowmelt and spring water compositions (∆m) is the
result of the reaction of the water with the minerals in the batholith.
Species Spring water Snow melt ∆m

µmol/kg µmol/kg µmol/kg
Na+ 134 24 110
Ca2+ 78 10 68
Mg2+ 29 07 22
K 28 08 20
HCO3− 328 18 310
SiO2(aq) 273 3 270
The minerals in the batholith that are available for weathering reactions are pla-
gioclase, which consists of albite (NaAlSi3O8) and anorthite (CaAl2Si2O8); bio-
tite (KMg3AlSi3O10(OH)2); potassium feldspar (KAlSi3O8); and quartz (SiO2).
The reactions produce kaolinite (Al2Si2O5(OH)4).
There are five postulated weathering reactions. They are written as a series of
linear equations. This is the key feature of the mass balance approach.
(albite–ab) 2 ab + 9 H2O + 2 H2CO3 − kaol − 2 Na+ − 2 HCO3−

− 4 H4SiO4 = 0
(anorthite–an) an + H2O + 2 H2CO3 − kaol – Ca2+ − 2 HCO3− = 0
172 Quasi-kinetics
(biotite–bio) 2 bio+ H2O + 14H2CO3 − kaol − 2K+ − 6Mg2+ − 14HCO3−

− 4H4SiO4 = 0
(K-feldspar–Kf) 2 Kf + 9 H2O + 2 H2CO3 − kaol − 2 K+ − 2 HCO3−
− 4 H4SiO4 = 0
(quartz–qz) qz + 2 H2O − H4SiO4 = 0
The objective is to determine the extent of each reaction, ξ, by distributing

the change in water composition over these five reactions. This can be done by
accounting for how much of each aqueous species is consumed or produced by
each of the reactions.
species (ab) (an) (bio) (Kf) (qz) ∆m, µmol/kg

Na −2 ab + 0 an + 0 bio + 0 Kf + 0 qz = 110 = ∆mNa
Ca 0 ab − 1 an + 0 bio + 0 Kf + 0 qz = 68 = ∆mCa
Mg 0 ab − 0 an − 6 bio + 0 Kf + 0 qz = 22 = ∆mMg
K 0 ab + 0 an − 2 bio − 2 Kf + 0 qz = 20 = ∆mK
HCO3 −2 ab – 2 an − 14 bio − 2 Kf + 0 qz = 310 = ∆mHCO3
H4SiO4 −4 ab + 0 an − 4 bio − 4 Kf − 1 qz = 270 = ∆mH4SiO4
There are six equations and five unknowns. The extent of each reaction can be
estimated using multiple linear regression. Multiplying the estimated extent of
reaction, ∆ξi (values in parentheses are the standard error of the estimate) by the
reaction coefficient for the mineral gives the number of moles of that mineral
consumed per kilogram of water. If the contact time, ∆t, is known, the rate of
each reaction can be estimated from ∆ξi /∆t.
reaction ξi, µmol/kg νi ∆ξiνi, µmol/kg

(ab) −55(0.5) 2 −110
(an) −68(0.8) 1 −68
(bio) −3.7(0.1) 2 −7.3
(Kf) −6.3(0.6) 2 −12.6
(qz) −10(3) 2 −20
Garrels and MacKenzie (1967) solved for these values using a sequential sub-
traction process. The more general multiple linear regression method used here
distributes the rounding and analytical errors over the ξi estimates and is amen-
able to expansion to a much larger set of reactions.
Partition coefficients
The trace element concentrations in geologic materials are strongly linked to
major element behaviors. This means that trace element distributions offer
useful clues about processes that control the distribution and redistribution
of major elements. The distribution of a trace element between a solid and
Mixed flow reactors 173
solution is expressed by a distribution coefficient (McIntire, 1963; Rimstidt

et al., 1998), which is the ratio of the mole fraction of the trace (XTr) and
major (XM) element in the solid to the ratio of the concentration of the trace
(mTr) to major (mM) element in the solution.
X   mTr 
λ =  Tr   m  (8.29)
 XM  M
This definition incorporates the activity coefficients for the trace element
in the solid and solution into λ; see McIntire (1963) for an approach that
explicitly accounts for activity coefficients.
The model assumes that during the reaction process the solid’s surface
is always in equilibrium with the solution. Then it becomes completely iso-
lated when it is overgrown by the next increment of precipitating solid. At
low temperatures, the diffusion rate of trace elements in most solids is very
slow, so the trace element concentration in each layer remains unchanged
once that layer is overgrown.
If λ > 1, the trace element is enriched in the solid relative to the solution
during precipitation, so mTr/mM is reduced as precipitation proceeds. If λ <
1, the trace element is enriched in the solution relative to the solid during
precipitation, so mTr/mM increases as precipitation progresses. In a mixed
flow reactor operating at a steady state, mTr/mM remains constant and the
solid has a constant XTr/XM ratio throughout. If the solid precipitates in a
batch reactor, mTr/mM changes during the precipitation process causing the
XTr/XM ratio to change from the center to the surface of the precipitating
solid.
Mixed flow reactors

In a mixed flow reactor operating at steady state, the concentrations of Tr
and M in the solution are constant. This means that the composition of the
solid is also constant. XTr/XM reflects the relative rates of removal of the
trace and major elements from the solution. These rates are the difference
between the concentration of the element in the feed and effluent solutions
multiplied by the flow rate (Q, kg/sec) through the reactor.

X Tr RTr
= =
(( m ) − ( mTr )out )Q ( mTr )in − ( mTr )out
Tr in
= (8.30)
X M RM (( m
M )in − ( mM )out )Q ( mM )in − ( mM )out
Combining Eqs (8.29) and (8.30) and realizing that the effluent solution is a
sample of the solution in the reactor leads to
174 Quasi-kinetics
λ=
( mTr )in − ( mTr )out ( mTr )out
=
(( m ) ( mTr )out ) − 1
Tr in
(8.31)
( mM )in − ( mM )out ( mM )out (( m
M )in ( mM )out ) − 1
This means that λ can be computed from the difference between the con-
centration of Tr and M in the feed and effluent solutions regardless of the
flow or precipitation rate. These same data along with the flow rate can be
used to find the precipitation rate of the major element, so that a mixed flow
reactor experiment is well suited to quantify the effect of the precipitation
rate on λ.
Example 8.6. Trapping toxic trace elements from solutions using a

mixed flow reactor
Trace elements can be removed from dilute solutions by co-precipitation with

a major element. For example, a longstanding method used to collect radium
from very dilute solution is to co-precipitate it with barium sulfate (Gordon and
Rowley, 1957).
Because calcium and bicarbonate are non-toxic and widely available, co-pre-
cipitation of radioactive or toxic trace elements with calcite is a reasonable strat-
egy to clean up industrial or radioactive waste solutions (Curti, 1999). Because
mixed flow reactors can process a continuous stream of solution they might be
employed to remove radioactive Co or radioactive Sr from a solution by co-pre-
cipitation with calcite. The reactor would mix a bicarbonate-rich solution with
a calcium-rich solution containing the trace element, which would be trapped
in the precipitating calcite. Let p (= mout/min) be the fraction of M or Tr that
remains in the effluent solution so that α (= 1 − p) is the fraction that is pre-
cipitated in the reactor. The effectiveness of such a process can be evaluated by
rewriting Eq. (8.31) first in terms of pTr and pM.
λ=
(1 pTr ) − 1 (8.32)
(1 pM ) − 1
Equation (8.32) is then written in terms of αTr and αM.
λ=
(1 (1 − α )) − 1
Tr
(8.33)
(1 (1 − α )) − 1
M
Further manipulation moves αTr to the left-hand side.
1  1 
= λ − 1 + 1 (8.34)
1 − αTr  1 − αM 
Taking the reciprocal of both sides of Eq. (8.34) allows it to be solved for αTr.
Batch reactors 175
0.8
0.6 Co ( =3.7 )
Tr
=1
0.4
0.2
Sr ( =0.04)
0
0 0.2 0.4 0.6 0.8 1
M
Figure 8.7. Fraction of trace element removed from solution as a function of the

fraction of major element precipitated in a mixed flow reactor.
   1  
αTr = 1 −  1  λ  − 1 + 1  (8.35)
   1 − αM  
This equation and λ values from Curti (1999) can be used to model the incorpor-
ation of Co and Sr into calcite precipitating in a mixed flow reactor.
λCo = 3.7
λSr = 0.04
Graphs of Eq. (8.35) shown in Figure 8.7 demonstrate that this scheme could
work for Co because precipitating 20% of the Ca would remove nearly 50% of
the Co. Using a series of mixed flow reactors would further improve the process
because removal of 50% of the remaining Co in a second reactor would leave
25% and a third reactor would further reduce the Co to 12.5%. On the other
hand, this process would be inefficient for Sr because precipitating 80% of the
Ca would only remove about 10% of the Sr. Hence, co-precipitation in a mixed
flow reactor is only effective for trapping trace elements if λ > 1.
Batch reactors
Co-precipitation in batch reactors is modeled using the Doerner–Hoskins
approach (Doerner and Hoskins, 1925). Because batch reactors are closed
176 Quasi-kinetics
systems, differences in the removal rate of the trace and major elements from
the solution cause the solution to become relatively enriched or depleted
in the trace element as the precipitation proceeds. This simple model from
Curti (1997) considers how removing a small amount of Tr and M by pre-
cipitating a small mass of solid affects the composition of the solution.
 X Tr  − ∆mTr  m + ∆mTr 
 X  = = λ  Tr (8.36)
M surface − ∆mM  mM + ∆mM 
As the amount of solid precipitated tends to zero, ∆mTr → dmTr and ∆mM
→ dmM so
dmTr m
= λ Tr (8.37)
dmM mM
Integrating this relation from initial concentrations gives a relationship

that links the concentration of Tr to the concentration of M.
m  m 
ln  Tr
o 
= λ ln  Mo 
(8.38)
 mTr   mM 
This expression allows the calculation of λ from the initial and final con-
centrations of the trace and major element.
m  m 
λ = ln  Tr
o 
ln  Mo 
(8.39)
 mTr   mM 
Applying the antilog transform to Eq. (8.38) gives a relationship that pre-
dicts the concentration of the trace element in the solution as a function of
the amount of major element precipitated.
λ
mTr  mM 
= o  (8.40)
mTr
o
 mM 
Example 8.7. Trace elements in calcite cements
The concentrations of Fe, Mn, and Mg in calcite cements are often used to inter-
pret their depositional environment. These interpretations use distribution coef-
ficients to estimate the composition of the solution based on an analysis of the
calcite cement. Laboratory determinations of these distribution coefficients are
typically carried out in batch reactors. Creating a model of the precipitation pro-
cess in these experiments is helpful for understanding the uncertainties associated
with the measurements. If the calcite is precipitated in a batch reactor, Eq. (8.40)
Batch reactors 177
Mn ( =14.8)
0.8
Fe ( =2.7)
0.6
Tr
=1.0
0.4
0.2
Mg ( =0.097)
0
0 0.2 0.4 0.6 0.8 1
M
Figure 8.8. Fraction of trace element removed from solution as a function of the

fraction of major element precipitated in a batch reactor.
can be modified to model the co-precipitation of the trace elements. Referring to

the notation used in Example 8.3, m/mo is the fraction of element that remains in
solution (= p) and α (= 1 − p) is the fraction that is precipitated.
Equation (8.40) can be recast in terms of the fraction of element remaining
in solution.
pTr = ( pM )
λ
(8.41)
This equation can be further recast in terms of the fraction of the element
precipitated.
1 − αTr = (1 − α M )
λ
(8.42)
Equation (8.42) can be rearranged to find the fraction of Tr that has precipitated
as a function of the fraction of M that has precipitated using the λ values listed
below.
αTr = 1 − (1 − α M )
λ
(8.43)
λMn = 14.8 (Curti, 1999)

λFe = 2.7 (Curti, 1999)
178 Quasi-kinetics
λMg = 0.097 (Curti, 1999)
Figure 8.8 shows that within analytical resolution all the Mn is removed from
solution by the time 30% of the Ca has precipitated. This means that to deter-
mine λMn, only small amounts of Ca can be allowed to precipitate and the value
of λMn will be strongly affected by the Ca analysis. On the other hand, deter-
mining λMg requires the precipitation of most of the Ca and the accuracy of
λMg depends upon accurate Mg determinations. Values of λ near 1 are easiest to
determine and most accurate. Note that these same constraints apply to deter-
mining distribution coefficients using mixed flow reactors (see Figure 8.7).
Chromatography models and retardation factors

Chromatographic separation occurs when the movement of dissolved spe-
cies is retarded by interaction with a stationary phase along a flow path.
Chromatography is a well-established way to separate chemical species in
the laboratory. Approaches used to model laboratory chromatography pro-
cesses are described by Laub (1985), Cazes and Scott (2002), and Arnold
et al. (1986).
Chromatographic effects occur in natural settings when fluids percolate
through and interact with rock or soil materials. There are several one-
dimensional models of chromatographic processes for high-temperature
processes (Appelo, 1996; Guy, 1993; Hofmann, 1972; Korzhinskii, 1970). All
chromatography models assume that a chemical component in the moving
fluid equilibrates instantly with a mineral phase as a slug of the fluid moves
through the rock, i.e. local equilibrium is maintained. This assumption is
combined with a continuity relationship to produce a differential equation
that can be used to model the rate of transport of one or more components
along the flow path. Chromatography models have been applied to metaso-
matic infiltration processes such as those that produce skarns. When viewed
at a point along the flow path, the moving fluid transports reactants to that
point where they react with the minerals until the mineralogy is transformed
to an assemblage that is in equilibrium with the fluid. This means that the
extent of the reaction and mineralogy at each point along the flow path
could be modeled using a reaction path-type model. These infiltration pro-
cesses tend to produce sharp reaction fronts (Guy, 1993; Helfferich, 1989;
Korzhinskii, 1970; Schechter et al., 1987) that are related to phase boundary
crossings (see Example 8.4). Advancing reaction fronts can interact with
each other in two and three dimensions to produce interesting patterns
(Dria et al., 1987). These chromatography models for infiltration processes
at metamorphic conditions are successful because the reaction rates are high
and the infiltration rates are low so local equilibrium is maintained.
Chromatography models and retardation factors 179
For near surface conditions, mineral reaction rates slow and the flow rates
are relatively faster so the local equilibrium assumption often fails and chro-
matography models are less useful for modeling mineral transformations.
However, many adsorption–desorption processes are fast enough to main-
tain local equilibrium so that chromatography models are useful for mod-
eling the retardation of species that adsorb to aquifer minerals. The central
feature of these models is the retardation factor Rf (no units), which is the
ratio of the velocity of transport of the adsorbing species to the velocity of
a non-adsorbing, unreactive tracer (Higgins, 1959).
ρb
Rf = 1 + Kd (8.44)
φ
ρb (g/cm3) is the bulk density of the packed bed, φ (no units) is the por-
osity, and Kd (cm3/g) is the distribution coefficient. See Zheng and Bennett
(1995) for a derivation of this equation. It is related to the KD (mol/g-solid/
mol/kg-water) derived in Chapter 6 by a units transformation involving the
molecular weight of the adsorbing species and the density of water. Because
Kd is related to KD, its value is also a function of the specific surface area
of the solid.
 mol i 
g solid 1  = 1000 K = K  cm soln 
3
KD  × (8.45)
d 
 mol i   g solid 
D
kg soln
 kg soln 1000 cm 3soln 
Example 8.8. Zn2+ migration in a lateritic soil
Chotantarat et al. (2011) performed a column experiment using a pH 5 Zn2+

solution (co = 4.61 × 10−3 mol/L), which was pumped through a 10 cm long col-
umn packed with a lateritic soil from the Akara mine area in Thailand. The bulk
density of the soil in the packed column was 0.92 g/cm3 and the porosity was
0.66. The columns were calibrated using an inert Br− tracer (Figure 8.9a), which
showed that the breakthrough, where c/co = 0.5, occurred when 1.0 pore volumes
were eluted. The Zn2+ experiment (Figure 8.9b) showed that c/co = 0.5 occurred
when 26 pore volumes were eluted so Rf = 26. The simplest way to analyze these
data is to find Kd from the rearranged Eq. (8.44).
Rf − 1 26 − 1
Kd = = = 17.9 (8.46)
ρb φ 0.92 0.66
More sophisticated methods are typically used to find Rf (Nkedl-Kizza et al.,

1987) and the adsorption behavior is often modeled using more complex iso-
therms (Chotantarat et al., 2011) but the principles are the same.
180 Quasi-kinetics
(a)
c/co
0.5
0
0 1 2 3 4 5
pore volumes
(b)
1
c/co
0.5
0
0 20 40 60 80 100
pore volumes
Figure 8.9. (a) Breakthrough curve for three Br− tracer experiments showing that
the inert tracer breakthrough occurs at one pore volume. The dashed line shows
the pattern expected for ideal plug flow and the deviation of concentration from
this line is the result of dispersion and diffusion. (b) Breakthrough curve for the
Zn2+ experiment showing that c/co occurs after 26 pore volumes have eluted.
Chromatography models and retardation factors 181
If a mixture of aqueous species with different Kd values was injected into a

plug flow reactor as a spike (small slug), adsorption processes would cause some
species to be retarded more than others. This would cause each species to elute
from the reactor at a different time. This is the basis for chromatographic separa-
tions in the laboratory.
Chapter 9
Accretion and transformation kinetics
Under near-equilibrium conditions, solids form from aqueous solutions

via the addition of monomer growth units. The terrace, ledge, and kink
(TLK) model and the Burton–Cabrera–Frank (BCF) theory give reasonable
descriptions of this process. At higher degrees of supersaturation, mono-
mers addition is joined by the accretion of polymer and larger growth units,
up to and including stable crystallites. The solids formed by accretion of
these larger units are often metastable and eventually transform to more
stable forms. This means that the formation of a stable crystalline solid from
a very supersaturated solution involves several steps. This chapter presents
models that describe the formation and accretion of larger growth units and
transformation of the resulting metastable solids.
Polymerization rates
Polymers such as DNA, proteins, polysaccharides, and polyphenols are
essential components of organisms; and synthetic polymers such as Nylon,
polyethylene, styrene, and Teflon are basic raw materials for modern tech-
nology, so it is not surprising that there is a rich literature about the for-
mation of polymers. Most of that literature focuses on the formation of
biopolymers and plastic but the polymerization models (Dotson et al., 1996)
for those cases are potentially useful to geochemists.
Many biopolymers and synthetic polymers are linear, which means
that their constituent monomers are linked up in a head to tail fashion.
For example, cellulose, (C6H10O5)n, is a linear array of glucose molecules,
and polyethylene, (C2H2)nH2, is a linear grouping of ethylene molecules.
However, polymers linked in three dimensions play a more important role
in geochemical situations. For example, humic acids are randomly linked
three-dimensional polymers consisting of various degradation products of
biochemical substances.
Models for the formation of linear polymers are relatively simple (Dotson
et al., 1996). The chemical reaction for the growth of an (AB)n linear poly-
mer involves hooking up additional (AB)m molecules head to tail.
182
Polymerization rates 183
(AB)n + (AB)m = (AB)n+m (9.1)
The overall rate of this reaction depends upon the rate of interaction between
the A and B functional groups with concentrations A and B.
dA dB
= = − kAB (9.2)
dt dt
Because the monomer AB contains equal amounts of A and B, A = B. At the

initial conditions only the AB monomer exists. This species can be treated as
a polymer (P) with a chain length of one so A = B = Po. Equation (9.2) can
be rewritten in terms of the concentration of polymer units (P).
dP
= − kP 2 (9.3)
dt
Equation (9.3) can be integrated with the initial conditions of P = Po when

t = 0 to find the concentration of P as a function of time.
Po
P= (9.4)
1 + ktPo
The rate of decrease of monomer concentration (dP1/dt) is proportional to

the concentration of P1 and P.
dP1
= −2 kPP
1 (9.5)
dt
The factor of two in this equation arises because there are two distinguishable
reactions, one for each end of the polymer. Additional equations can be writ-
ten to find the rate of change in the concentration of polymers of each chain
length (Dotson et al., 1996). This model has didactic value, but most geo-
chemical polymers are three-dimensional so they have more than two attach-
ment points. Models of three-dimensional polymerization networks, such as
occur with silica polymerization (Jin et al., 2011), can become quite complex.
Example 9.1. Silica scaling
Deposition of amorphous silica (silica scale) onto working surfaces in geother-

mal power plants (Gunnarson and Arnórsson, 2003) and water desalinization
plants (Weng, 1995) reduces rates of heat and mass transfer thereby reducing
the efficiency of these enterprises. Silica scaling occurs when solutions become
supersaturated with respect to amorphous silica because of cooling, evaporation
of water, or a chemical reaction. The reaction between two monomers joins two
Si atoms and eliminates a water molecule.
184 Accretion and transformation kinetics
K

H 4SiO 4 + H 4SiO 4 
1

 H6Si2 O7 + H2 O (9.6)
The next polymerization step follows the same pattern.
K

H6Si2 O7 + H 4SiO 4 
2

 H8Si3O10 + H2 O (9.7)
The rate of monomer loss from solution (R, molal/sec) tends to be proportional
to the fourth power of monomer concentration (Icopini et al., 2005; Rothbaum
and Rohde, 1979). Icopini et al. (2005) explain this by assuming that the rate-
determining step is a reaction converting the trimer, formed by reaction (9.7),
into a cyclical tetramer.
H8Si3O10 + H 4SiO 4 k

3
→ H8Si 4 O12 + 2H2 O (9.8)
If reactions (9.6) and (9.7) reach equilibrium quickly, the overall rate is propor-
tional to the concentration of H8Si3O10, which is related to the monomer con-
centration by the K1 and K2 equilibrium constants. The f(γ) term is the product
of activity coefficients needed to convert the activity of the silica species into
concentrations.
mH4 4 SiO4
R = f ( g )k3 mH8Si3 O10 mH4 SiO4 = f ( g )k3 K1K 2
aH2 2O (9.9)
Equation (9.9) can be simplified by representing the silica concentration as m

and combining all the other terms into an apparent rate constant (k´).
dm
− = k ′m 4 (9.10)
dt
Equation (9.10) can be rearranged and integrated for the boundary condition
that when t = 0, m = mo.
m t
dm
∫m 4
= − k ∫ dt (9.11)
mo 0
1 −3
− k ′t = ( m − mo−3 ) (9.12)
3
The concentration of monomer as a function of time is found by solving Eq.

(9.12) for m.
Nucleation rates 185
25
monomer concentration, millimolal
20
pH 11
pH 3
15
10
5
pH 7
0
0 1000 2000 3000
time, hr
Figure 9.1. Silica monomer concentration as a function of time due to monomer

incorporation into polymers. The formation of silica polymers is fast at near-
neutral pH and slow at both high and low pH.
3
 1 
m= −1 / 3 
(9.13)
 3k ′t + mo 
This equation, along with values of mo and k´ (Table 2 in Icopini et al., 2005),
can be used to forward model the monomer concentration (m, millimolal) as a
function of time (Figure 9.1) for an ionic strength of 0.01.
mo = 20.8 mm (millimolal) = 2.08 × 10−2 m (same as for experiments in Icopini
et al., 2005)
k´(pH = 3) = 1.17 × 10−9 mm−3sec−1

k´(pH = 7) = 4.17 × 10−8 mm−3sec−1
k´(pH = 11) = 3.53 × 10−10 mm−3sec−1
Nucleation rates
When a homogeneous aqueous phase becomes supersaturated with respect
to a solid phase, the rate of appearance of the new solid phase is con-
trolled by nucleation kinetics. Nucleation occurs whenever enough of the
solute species come together to create a particle that can grow spontan-
eously. Classical models of nucleation kinetics are based on a combination
of statistical mechanics and thermodynamic principles (Dunning, 1969;

Ford, 2004; Leubner, 2000). All these models assume that the rate of
appearance of nuclei is proportional to the equilibrium concentration of
molecular clusters of a critical size. These models contrast with spinodal
decomposition models that assume that fluctuations in the solution lead
to the aggregation of solute molecules into a disordered cluster and the
rate of formation of the solid depends upon the rate of rearrangement of
molecules in the cluster to form a crystal (Gebauer et al., 2008; Meldrum
and Sear, 2008).
Solubility as a function of cluster size

Classical nucleation models are based on the concept that work is required
to create the new interface bounding a growing molecular cluster. The
amount of work (∆Gc, J/cluster) needed to produce the cluster’s interface
is the product of the cluster’s interfacial area (A, m2) and the surface free
energy of the interface (σ, mJ/m2) (Adamson and Gast, 1997). This means
that a cluster’s surface free energy becomes more positive as it grows in size,
but this increase is offset by a decrease in the cluster’s bulk free energy of
formation (∆Gb, J/cluster), which is directly proportional to the number of
molecules (n) in the cluster and the degree of supersaturation (S = Q/K).
These two effects add together to give the overall free energy of formation
of a cluster.
∆Gc = ∆Gb + ∆Gs = − nkBT ln S + σA (9.14)
This relationship can be recast to find a relationship between the solubility

product of a cluster (Kc) with a diameter (D, m) and the solubility product
of the bulk solid (Kb) (Berner, 1980).
 (2 3) bσVm   2 
log K c = log K b +    (9.15)
 2.303RT   D 
The surface free energy of a cluster plays a critical role in determining its
stability. Figure 9.2a shows that surface free energy of ionic solids correlates
with the substance’s bulk solubility (Söhnel, 1982) and Figure 9.2b shows
how the solubility product of a cluster decreases with increasing cluster
diameter until it approaches the solubility product of the bulk phase. A
larger value of σ results in a higher solubility product for a particular cluster
diameter. For clusters that are larger than about 1 µm the contribution of
∆Gs to the overall free energy is negligible, so their solubility is effectively the
same as the bulk phase.
Nucleation 187
(a) 160
120
, mJ/m 2
80
40
0
–6 –4 –2 0 2
log s
(b) 25
20
Figure 9.2. (a) Correlation
barite between the interfacial free
energy (σ, mJ/m2) of ionic
15 solids and the logarithm of
their molar solubility (Ms,
Kc/Kb
mol/L) at 25°C. The linear

10 relationship is σ = −17.8 log
Ms + 34.8 (Söhnel, 1982).
gypsum (b) Solubility product of
gypsum (CaSO4•2H2O)(σ
5
= 88 mJ/mol) and barite
(BaSO4) (σ = 136 mJ/mol)
clusters relative to the
0 solubility product of the
10–9 10–8 10–7 10–6 bulk solid as a function of
D, m cluster diameter.
Nucleation
Nucleation models postulate that clusters must attain a critical size before
they can irreversibly grow to a macroscopic size. Although details differ,
all the classical nucleation models develop the same general relationship
between the nucleation barrier, surface free energy, and degree of saturation.
The models mostly differ in how the rate of addition of a monomer to the
critical-sized cluster is derived from statistical mechanics (Dunning, 1955,

1969; Nielsen, 1964; Turnbull and Fisher, 1949; Uhlmann and Chalmers,
1966; Walton, 1969). The critical size barrier idea is cast in terms of the
free energy of formation of the clusters making it analogous to the quasi-
thermodynamic transition-state model presented in Chapter 5. Classical
nucleation models all assume that the surface free energy is independent of
cluster size. It is more likely that the surface free energy should be adjusted
to account for its variation with cluster size (Fokin et al., 2010; Larson and
Garside, 1986).
With increasing solution concentration there is a tendency for dissolved
species to interact with each other to form dimers, trimers, and larger clus-
ters that are in dynamic equilibrium with the monomers. The concentration
of each cluster size can be expressed by an equilibrium constant for each
reaction step.
A + A = A2 K 2 = aA2 aA2
A 2 + A = A3 K 3 = aA3 aA3
(9.16)

A n −1+ A = A n K n = aAn aAn
This derivation uses a molar (mol/L) concentration scale to maintain con-

sistency with previous derivations. The model could be constructed using a
molal (mol/kg) concentration scale to avoid complications that arise with
the large changes in solution density associated with changing temperature
or solute concentrations. The free energy of reaction for assembling n mole-
cules into a cluster is found from the equilibrium constant for the reaction.
∆Gr° = ∆G °f ( An ) − n∆G °f ( A) = − RT ln K n (9.17)
The free energy of formation of the clusters is related to the cluster

concentration.
∆G °f ( An ) = − nRT ln aA − RT ln K n = − RT ln M An (9.18)
Dividing ∆Gr°(An) by Avogadro’s number (clusters/mol) gives the free energy

of formation of a single cluster (∆Gc, J/cluster). This gives a relationship
between ∆Gc and the number of clusters per liter (cAn, clusters/L).
∆Gc = − kBT ln cAn (9.19)
The equilibrium cluster concentration must be known in order to evaluate

the right-hand side of Eq. (9.19), but that presents a significant analytical
difficulty because the cluster concentration is very low. This problem is over-
come by realizing that the amount of work (∆Gc) needed to accumulate n
Nucleation 189
molecules of A into a single cluster is the difference between the decrease

in the bulk free energy (∆Gb, J/cluster) due to adding more molecules to the
cluster and the increase in the surface free energy (∆Gs, J/cluster) due to the
cluster’s increasing surface area (Adamson and Gast, 1997). This means that
∆Gc can be expressed in terms of the number of molecules in each cluster
(nA) and the Boltzmann constant (kB).
∆Gc = ∆Gb + ∆Gs = − nkBT ln S + Ac σ (9.20)
S (= Q/K, no units) is the degree of saturation of the solution. The surface

area of a cluster (Ac, m2/cluster) can be expressed in terms of the cluster’s
volume (Vc, m3/cluster) using the relationship: Ac = (36π)⅓Vc⅔. The number
of molecules in the cluster is Vc divided by the volume of 1 molecule (V1 =
Vm/NA, m3/molecule).
∆Gc = − nkBT ln S + (36 π ) n 2 3V12 3σ

13
(9.21)
A graph of ∆Gc versus n displays a maximum (n*), which corresponds to

the number of molecules in a critical cluster that is in equilibrium with the
supersaturated solution. The addition of one more molecule to this cluster
will cause it to spontaneously grow to macroscopic size. The value of n* is
found from the derivative of Eq. (9.21).
d ∆Gc
= − kBT ln S + (2 3)(36 π ) nA−1 3V12 3σ
13
(9.22)
dn
When d∆Gc/dn = 0, n = n* so Eq. (9.22) can be set equal to zero and solved
for n*.
kBT ln S = (2 3)(36π )
13
( n* )−1 3 V12 3σ (9.23)
n* =
(32π 3)V12σ 3 (9.24)
( kBT ln S )3
The free energy of formation of the critical nucleus (∆Gc*, J/cluster) is found
by substituting Eq. (9.24) into Eq. (9.21).
23
(32π 3)V12σ3 k T ln S + 36π 1 3  (32π 3)V12σ3 
∆Gc* = − ( B ) ( )  3 
V12 3σ (9.25)
( kBT ln S )3  ( kBT ln S ) 
Equation (9.25) can be simplified using the following relationship.

23 23
48π
(36π )1 3 (32π 3)2 3 = (6)2 3 
32   192 
π = (64 ) π = 16π =
23
 π = (9.26)
3  3  3
This allows the right-hand side of Eq. (9.25) to be simplified to a single

term.
∆Gc* = −
(32π 3)V12σ3 + ( 48π 3)V12σ3 = (16π 3)V12σ3 (9.27)
( kBT ln S )2 ( kBT ln S )2 ( kBT ln S )2
The nucleation rate (N, nuclei/L sec) is the product of the flux of monomers
(JA, molecules/m2 sec) to each critical nucleus and the surface area of each
critical nucleus (Ac*, m2/cluster) multiplied by the number of critical nuclei
per unit volume of solution (c*, nuclei/L).
N = J AAc*c* (9.28)
Determining the flux of monomers to the critical nuclei (JA, molecules/L sec)
is a challenge. This flux has been modeled using various statistical mechan-
ics approaches. These models estimate the flux of molecules from the solu-
tion to the surface of the critical nuclei but they include various adjustments
such as the Zeldovich factor (Markov, 2003), which accounts for the possi-
bility that some critical-sized clusters will dissolve to a smaller size instead
of growing into crystallites. Presently there is no widely accepted model for
JA, so Eq. (9.28) is usually recast into a semi-empirical form, which com-
bines the monomer flux and cluster surface area into a pre-exponential term
(Ω, crystallites/L sec).
N = Ωc* (9.29)
Rearranging Eq. (9.19) gives the concentration of critical-sized clusters.
c* = e − ∆Gc (9.30)
* kB T
Combining Eqs (9.27), (9.29), and (9.30) produces an equation that expresses
the nucleation rate in terms of the degree of saturation.
 (16 π 3)V 2 σ3   (16 π 3)V12 σ3 

1
− kB T  − 
N = Ωe (− ∆Gc kB T )  (k T ln S )2   k T (k T ln S )2 
= Ωe = Ωe
*
B B B
(9.31)
This equation can be log-transformed to give log N in terms of log S.
(16π 3)V12σ 3 log S −2

log N = log Ω − ( ) (9.32)
( ln10kBT )3
Nucleation 191
Experimental results suggest that log Ω is on the order of 33 ± 3 for nucle-

ation in aqueous solution at 25°C (Nielsen, 1964; Pina and Putnis, 2002).
During their initial growth stages, the clusters are likely to have structures
that are more disordered than that of the bulk solid (Garten and Head,
1970). This idea has been incorporated into models that posit the formation
of a disordered cluster as an initial step followed by the formation of a crys-
tallite in a second step (Erdemir et al., 2009). These models are more com-
plicated conceptually and quantitatively than the classical model presented
here. Alternatively, Söhnel and Garside (1988) deal with this possibility by
making the surface free energy of the clusters a function of their size.
Example 9.2. Barite (BaSO4) nucleation rate
Barite scales can clog water treatment, petroleum production, and geothermal
facilities (Boerlage et al., 2002; He et al., 1994). An additional problem is that
the barite that precipitates in these and other systems often incorporates signifi-
cant amounts of radioactive elements, especially radium. Removal and disposal
of these radioactive barite precipitates is complicated and costly (Ceccarello
et al., 2004). Schemes to mitigate barite scale formation typically use additives
to inhibit barite nucleation. The first step toward evaluating the efficacy of these
additives is to develop a model for barite nucleation rates for untreated condi-
tions. Most of these additives are chosen because they increase the surface free
energy of the barite clusters thus slowing the nucleation rate.
The following data are needed to construct a model of barite nucleation.
σ = 136 mJ/m2 = 0.136 J/m2

Vm = 5.21 × 10−5 m3/mol so V1 = 8.65 × 10−29 m3/molecule
kB = 1.38 × 10−23 J/molecule K
(36π)⅓ = 4.84
32π/3 = 33.5
16π/3 = 16.8
T = 298 K
The first step toward developing a conceptual model of the nucleation process
uses Eq. (9.21) to map out ∆Gc as a function of log S. It is convenient to recast
this equation by inserting the appropriate numerical constants.
∆Gc = − n (1.38 × 10 −23 ) (298) 2.303 log S + 4.84 n 2 3 (8.65 × 10 −29 ) (136 × 10 −3 )
23
= − n (9.47 × 10 −21 ) log S + n 2 3 (1.29 × 10 −19 )

(9.33)
The maximum on each graph of ∆Gc versus n (Figure 9.3a) occurs where n = n*.
These graphs show that the ∆Gc* values associated with each maximum decrease
(a) (b)
2×10 –18 1000
1.5
800
1×10–18
Gc, J/cluster
2.0 600
0
n*
2.5 400
–1×10–18 3.0
200
–2×10 –18 0
0 100 200 300 400 0 1 2 3 4 5
n, molecules log S
(c)
10
4
log N
–2
0 1 2 3 4 5 6 7
log S
Figure 9.3. (a) Free energy of formation of BaSO4 clusters as a function of the number of BaSO4 molecules in
the cluster contoured for values of log S ranging from 1.5 to 3.0. (b) Number of BaSO4 molecules in the critical
cluster as a function of log S. (c) Nucleation rate (nuclei/L sec) of BaSO4 as a function of log S.
as log S increases, which means that the barrier to nucleation decreases as the
solution becomes more supersaturated.
The size of the critical nucleus as a function of the degree of saturation is
found using Eq. (9.34).
Aggregation rates 193
n* =
(33.5)(8.65 × 10 −29 )2 (0.136)3 =
742.7
(9.34)
(1.38 × 10 ) (298) (2.303) ( log S ) ( S )3
−23 3 3 3 3
log
This equation shows that the size of the critical nucleus grows very large when
the degree of saturation drops below 10 (log S < 1) and becomes very small
when the degree of saturation exceeds 10,000 (log S > 4) (Figure 9.3b).
The barite nucleation rate as a function of the degree of saturation is calcu-
lated using Eq. (9.32).
(16.8)(8.65 × 10 −29 )2 (0.136)3 log S −2 = 33 − 373 log S −2

3 ( ) ( )
log N = 33 − (9.35)
((2.303)(1.38 × 10 −23 )(298))
Figure 9.3c shows that the log N versus log S curve is very steep, so that homo-
geneous nucleation rates are very low when log S is less than 3 but they become
very fast as log S increases from 3 to 4. When log S is less than about 3, barite
nucleates so slowly that it is unlikely to cause a technological problem. Barite
formation for log S < 3 will involve heterogeneous nucleation (Kashchiev and
van Rosmalen, 2003), which is modeled using a modified geometry and surface
free energy for the critical cluster.
Aggregation rates
If the concentration of polymers or crystallites is high enough they can
aggregate into larger clusters. For the simplest case, each cluster (A) moves
around in the solution as the result of Brownian motion (diffusive behav-
ior) and whenever two clusters come into contact they stick together.
This is called diffusion-limited aggregation (DLA). For clusters of equal
diameter, the concentration of A decreases following a second-order rate
equation.
dcA k
= − S cA2 (9.36)
dt 2
The rate constant for this equation can be calculated from the Smoluchowski
model.
8kBT
kS = (9.37)
3η
The Boltzmann constant (kB) equals 1.38 × 10−23 J/molecule K and at 25°C
the dynamic viscosity of water (η) equals 8.91 × 104 Pa sec (1 Pa = 1 J/m3).
8kBT 8 (1.38 × 10 −23 ) (298) m3

kS = = = 1.23 × 10 −17 (9.38)
3η 3 (8.91 × 10 )−4
cluster sec
This constant can be inserted into Eq. (9.36) to predict the rate of disap-
pearance of A due to diffusion-limited aggregation at 25°C.
dcA  clusters   1  1
 = −   kS cA = −   1.23 × 10 cA
−17 2
 3
2
(9.39)
dt m sec 2 2
Equation (9.36) can be integrated from the initial cluster concentration (cA°)
to the time when cA = ½cA°.
2 2 2 2
− + o =− + o = − kS t1 2 (9.40)
cA cA 1 2 cA cA
Equation (9.40) can be rearranged to find the time needed to reduce the
cluster concentration to half its original value.
2
t1/ 2 = (9.41)
kS cAo
Equation (9.41) shows that cluster aggregation is very sensitive to the cluster
concentration, so that for low concentrations very long times are required
for significant aggregation to occur (Figure 9.4a) but aggregation times are
very short for suspensions with high cluster concentrations. Because dif-
fusion rates increase with increasing temperature, the aggregation times
decrease significantly with increasing temperature (Figure 9.4b). When log
kS is graphed versus 1/T (Figure 9.4c), the activation energy for aggregation
is found to be ~15.5 kJ/mol, which is consistent with a diffusion-limited
process.
Diffusion-limited aggregation (DLA) is often referred to as a fast process
because the above model predicts the maximum aggregation rate. Reaction-
limited aggregation (RLA) rates are slower than predicted by the DLA
model. The effect of the slow attachment reaction rate on the aggregation
rate is expressed as a stability ratio (W) and is defined as the ratio of the
Smoluchowski rate constant (kS) to the observed rate constant (kO).
kS
W= (9.42)
kO
Figure 9.5 shows that W declines dramatically with increasing ionic

strength because the ions cause the double layer to collapse, which decreases
the electrostatic repulsion between the colloid particles.
Aggregation rates 195
(a) (b)
10
8 7000
10
7 6000
1 yr
10 6 5000
1 wk
10 5 1 day 4000
t1/2, sec
t1/2, sec
10
4 3000
1 hr
10 3 2000
2
10 1000
10 1 0
10 –8 10 –6 10 –4 10 –2 0 50 100 150 200 250 300
c°, mol/L temperature, °C
(c)
–15.5
–16.0
log kS
–16.5
–17.0
–17.5
0.002 0.003 0.004
1/T, K–1
Figure 9.4. (a) DLA model prediction of the time needed for half the clusters to aggregate as a function of clus-
ime needed for half the clusters to aggregate as a function of temperature for c° = 1 × 10−4
ter concentration. (b) T
mol/L. (c) Arrhenius plot showing log kS as a function of 1/T for a DLA process.
Example 9.3. Aggregation time for silica clusters
Axford (1997) showed that the W value for silica cluster aggregation is affected
by the ionic strength and cluster size. This information can be used to model the
effect of ionic strength and cluster size on the aggregation time.
The first step in building this model uses multiple linear regression to fit the W
data in Figure 9.5 as a function of ionic strength (I) and cluster size (N).
7
N
6 1
5 10
Figure 9.5. Logarithm
of experimentally
determined stability 4 100
log W
ratios for colloidal silica

(cA° = 6 × 1016 clusters/ 1000
3
m3) as a function of the
logarithm of the ionic
10000
strength (I = molarity 2
of KCl) contoured in
aggregate size expressed
1
as the number (N) of
12 nm diameter unit
clusters in the aggregate 0
(pH = 9.4) (Axford, 1997). –0.4 –0.3 –0.2 –0.1
log I, mol/L
logW = 1.85 − 11.2 log I − 0.872 log N (9.43)
The next step substitutes log kS into the logarithmic transform of Eq. (9.42) to
find an empirical rate constant (ko).
log ko = log kS − logW = −16.9 − logW (9.44)
Then Eq. (9.41) is recast in terms of the empirical rate constant (ko).
log t1 2 = 0.301 − log cA° − log ko (9.45)
Combining Eqs (9.43), (9.44), and (9.45) gives a function for t½ as a function of
cluster size, ionic strength, and initial cluster concentration.
log t1 2 = 19.1 − log cA° − 11.2 log I − 0.872 log N (9.46)
Figure 9.6 shows that t½ is very small for large clusters at high ionic strength and
very large for small clusters at low ionic strength. The silica cluster aggregation
rate is also strongly affected by pH, which controls the surface charge of the
clusters.
Solid–solid transformation rates 197
–0.1
log t = 1
2
–0.2
3
log I
4
–0.3
5
–0.4
0 1 2 3 4
log N
Figure 9.6. Logarithm of the time needed for one half of the silica clusters to
aggregate (log t½, time in seconds) contoured in terms of the logarithm of
the ionic strength (I) and logarithm of the cluster size (N = number of 12 nm
diameter units in the cluster).
The previous discussion assumed perikinetic conditions where the con-

tact rate of the clusters is limited to collisions due to their Brownian motion.
Orthokinetic aggregation refers to situations where fluid motion in the sus-
pension due to stirring or fluid flow increases the contact rate, see for exam-
ple Edzwald et al. (1974). There are many good books that cover this, and
related topics, in more detail, e.g. Meyer (1999). Burd and Jackson (2009)
review the principles that apply to aggregation in marine settings. Further
complexity in the aggregation process arises when the clusters interact via
asymmetric force fields (Bishop et al., 2009), which causes oriented aggre-
gation. Oriented aggregation appears to be a common geochemical process
that plays a role in the formation of many minerals (Burrows et al., 2010;
Penn, 2004; Penn et al., 2007).
Solid–solid transformation rates
The rate of transformation of a metastable solid (parent) phase (A) to form
a more stable solid (product) phase (B) is usually modeled using the Avrami
equation (Avrami, 1939, 1940), which is also known as the Johnson–Mehl–
Avrami–Kolmogorov (JMAK) equation. This equation is based on a model
that assumes that the transformation involves the nucleation of the prod-
uct phase followed by its growth until the parent phase is replaced by the
product phase. This model is based on three assumptions: (1) nucleation

occurs homogeneously throughout the parent phase, (2) the growth rate is
constant over the entire time of the transformation, and (3) the growth rate
of the new phase is the same in all directions.
The JMAK equation predicts the fraction of parent phase transformed
(α = V/Vt, m3/m3), which is the volume of parent transformed divided by the
total volume, as a function of time. In the initial stages of the transform-
ation, the product phase nucleates and grows at random places throughout
the parent phase. The product phase has a spherical shape because it grows
at the same rate in all directions. Eventually the growing spheres overlap and
there is no nucleation or growth in the overlapped volume. If this overlap
did not occur, the extended volume of transformed material would be Ve
and the fraction of material transformed would be αe (= Ve/V, m3/m3). Once
the product phases overlap, the rate of change of α must be adjusted for the
overlap of the product phase regions.
d α = d α e (1 − α ) (9.47)
For the case where abundant nuclei are present in the initial stages of the
transformation, which is termed site saturation, the extended volume of one
sphere of product phase is modeled by changing the sign for the rate con-
stant in the shrinking particle model (Chapter 6) to make it a growing par-
ticle model. The linear growth rate of each sphere of the product phase is G
(m/sec) and its volume is V (m3).
4π
Vi = (Gt )3 (9.48)
3
The extended fraction of transformed material is found by multiplying Eq.

(9.48) by the number of nuclei in the total volume N (= nuclei/Vt).
αe =
∑V i
=N
4π
(Gt )3 (9.49)
Vt 3
The rate of change of the extended fraction of transformed material is

found by taking the time derivative of Eq. (9.49).
d α e = N 4π (Gt ) dt
2
(9.50)
Equation (9.50) is substituted into Eq. (9.47) to eliminate the αe term.
d α = (1 − α ) N 4π (Gt ) dt
2
(9.51)
Equation (9.51) can be rearranged and integrated from t = 0 to t.

Solid–solid transformation rates 199
α t
dα
∫ (1 − α ) = N 4πG ∫ t dt
2 2
(9.52)
0 0
N 4πG 2 3
− ln (1 − α ) = t = kt3 (9.53)
3
Equation (9.53) is transformed and rearranged to find α as a

function of t.
α = 1 − e − kt3 (9.54)
A similar derivation, which assumes that the transformation rate is con-

trolled by the nucleation rate, gives an equation where t is raised to the fourth
power. The classical Avrami equation generalizes these results by assuming
that α is related to t by an equation with an arbitrary value for n.
α = 1 − e − ktn (9.55)
This equation is transformed to a linear form to extract values of k and n

from experimental data.
ln ( − ln (1 − α )) = ln k + n ln t (9.56)
Values of n found from experimental results are sometimes interpreted to

identify the predominant reaction process. For example, if n ≈ 4 the trans-
formation rate is said to be limited by nucleation rates and if n ≈ 3 the
transformation rate is said to be limited by the growth rate of the product
phase.
A modified version of the Avrami raises both k and t to the power of
n, which gives k more conventional units of sec−1 rather than the units of
sec−n that are required for the classical Avrami equation (Maffezzoli et al.,
1995).
α = 1 − e − (kt)
n
(9.57)
The linearized version of the modified Avrami model is slightly different

from Eq. (9.56).
ln ( − ln (1 − α )) = n ln k + n ln t (9.58)
Although criticized (Marangoni, 1998), this modified version is widely used

and it is important to be aware that the values of n and k are not inter-
changeable between the two forms (see Figure 9.7). Both forms predict “s”-
shaped curves for graphs of α versus t. There is no theoretical justification
for interpreting reaction processes from values of n found using the modi-
fied Avrami equation.
It is often useful to express the conversion in terms of a characteristic
time. The classical Avrami equation (9.55) can be solved to express t as a
function of α.
 ln (1 − α ) 
1n
t = −  (9.59)
 k
Substituting 0.5 for α in Eq. (9.59) gives the time to 50% conversion.
 ln (1 − 0.5) 
1n 1n
 0.693 
t1 2 =  − = (9.60)
 k   k 
The modified Avrami equation (9.57) can be solved to express t as a

function of α.
1
(− ln (1 − α ))
1n
t= (9.61)
k
Substituting 0.5 for α in Eq. (9.61) gives the time to 50% conversion.
1 1
( − ln (1 − α )) = (0.693)
1n 1n
t= (9.62)
k k
TTT diagrams
Time–temperature–transformation (TTT) diagrams are extensively used in
materials sciences to map out the characteristic time for solid–solid trans-
formations. They are sometimes called kinetic phase diagrams. These dia-
grams show the extent of transformation, α, contours on graphs with log
t on the x-axis and either T or 1/T on the y-axis. They are often prepared
by direct plotting of experimental data but they can be constructed for any
type of reaction using the integrated rate equation along with a temperature
function (Arrhenius equation) for the rate constant.
The simplest case is a TTT diagram based on a zeroth-order rate
equation.
dn
= −k (9.63)
dt
The integrated form of this equation gives the number of moles remaining
at a given time.
TTT diagrams 201
(a) 1.00
k = 10–4
n=3
0.80
k = 10–5
n=3
0.60
k = 10–4
n=2
0.40
k = 10–5
0.20 n=2
0.00 Figure 9.7. (a) Examples

0 40 80 120 160 200 of the effect of changing k
time, sec and n on the extent of the
transformation versus time
(b) 1 pattern predicted by the
k = 10–4 classical Avrami equation.
n=3 Increasing k or decreasing
0.8 n decreases the slope of
k = 10–4
the curve. (b) Examples
n=2
of the effect of changing
k and n on the extent of
0.6
the transformation versus
time pattern predicted
by the modified Avrami
0.4 k = 10–5
equation. Increasing k
n=2
decreases the slope of the
curve but changing n has
0.2 k = 10–5 very little effect on the
n=3 slope. Comparison of the
t½ values for k = 1 × 10−5
0 and n = 3 between models
0 50000 100000 150000 illustrates their significant
time, sec quantitative difference.
n − no = − kt (9.64)
Equation (9.64) can be expressed in terms of the fraction remaining (p), or

better yet in terms of the fraction transformed (α).
n  k
1− = 1− p = α =   t (9.65)
no  no 
Equation (9.65) is transformed to give ln α as a function of ln t.
ln α = ln k − ln no + ln t (9.66)
The log-transformed Arrhenius equation is substituted into Eq. (9.66) to get

a function that expresses α as a function of T and t.
E  1
ln α = ln A −  a  − ln no + ln t
 RT
 A  E  1
= ln   −  a  + ln t (9.67)
 no   R  T
 A   Ea  1
logα = log   −   + log t
 no   2.303R  T
Although the TTT diagrams typically show T as a function of log t, it is

more convenient to rearrange this equation to solve for log t as a function
of 1/T.
 A  Ea  1
log t = − log   + logα +  (9.68)
 no   2.303R  T
In this case, the fraction-transformed boundary will be a straight line on a

graph of 1/T versus log t. A graph of T versus log t can be constructed by
taking the reciprocal of Eq. (9.68). That graph will display a curved line.
The TTT equation for the classical Avrami model is developed from Eq.
(9.56).
E  1
ln ( − ln (1 − α )) = ln A −  a  + n ln t (9.69)
 RT
Equation (9.69) is rearranged so log t is a function of 1/T.
 1  E  1
ln t =    − ln A + ln ( − ln (1 − α )) +  a  
 n   R T
(9.70)
 1   Ea  1 
log t =    − log A + log ( −2.303 log (1 − α )) + 
 n   2.303R  T 
The TTT equation for the modified Avrami model (Eq. (9.58)) is slightly
different.
(1 n) ln ( − ln (1 − α )) = ln A − 
Ea  1
 + ln t (9.71)
RT
Equation (9.71) is rearranged so log t is a function of 1/T.

TTT diagrams 203
E  1
ln t = − ln A + (1 n ) ln ( − ln (1 − α )) +  a 
 RT
(9.72)
 Ea  1
log t = − log A + (1 n ) log ( −2.303 log (1 − α )) + 
 2.303R  T
(a) 200
150
goethite
= 0. 1
0.5
temperature, °C
0.9
100
50
schwertmannite
0
10 100 1000 10000 100000
time, sec
(b) 0.0022
0.0024
0.0026 goethite
= 0. 1
100° C
0. 5
0.0028 0. 9
1/T, K–1
75°
0.0030
50°
0.0032 schwertmannite
25°
0.0034
0.0036
10 100 1000 10000 100000
time, sec
Figure 9.8. TTT diagrams contoured in the extent of transformation of

schwertmannite to goethite based on data from Davidson et al. (2008). (a) When
a linear temperature axis is used the contours are curved lines. (b) When a
reciprocal temperature axis is used the contours are straight lines.
Example 9.4. Schwertmannite to goethite transformation
Schwertmannite (sch) is a fairly common mineral in acid mine drainage settings.

It is metastable and eventually transforms into goethite (goe).
Fe8O8(OH)8 − 2x(SO4)x (sch) + 2xH2O = 8FeOOH(goe) + 2xH+ + xSO42−
Davidson et al. (2008) measured the rates of this transformation over the tem-
perature range of 60 to 190°C and fit the results to the modified Avrami equa-
tion. The average value of n from their fits was 2.1(0.25). Fitting their reported
rate constants to the Arrhenius equation gives an estimated value of Ea of 32.4
kJ/mol with log A = 1.31. These values are substituted into Eq. (9.72) to create
the TTT diagram shown in Figure 9.8.
Chapter 10
Pattern formation
So far this book has offered appetizers. This chapter deals with the prepa-
ration of the main course by introducing concepts for linking the simple
models discussed in this book to understand the complex processes that
lead to pattern formation in geological settings. Geoscientists spend much
of their time and effort identifying and explaining naturally occurring spa-
tial and temporal patterns with the goal of interpreting those patterns to
understand the processes and conditions that formed them. They are chal-
lenged by the need to identify meaningful patterns in situations that often
appear to be chaotic (Crutchfield, 2012). Meaningful patterns are frequently
subtle and recognizing them often requires clues provided by process mod-
els. Most pattern-forming systems are too complex to interpret in a holis-
tic way so our strategy is to first parse them into simple processes, which
can be accurately modeled using methods like those explained in this book.
The resulting models of discrete processes are then linked to simulate the
overall pattern-forming scenario. This strategy is widely used in science and
technology. For example, engineers design processing plants by dividing the
overall process into unit operations, each of which is responsible for a single
chemical or physical transformation of a feedstock (Gupta and Yan, 2006;
Hendricks, 2006; McCabe et al., 1993). These unit operations are modeled
separately and the models are combined to simulate the entire processing
plant. This strategy is especially effective when the unit operations occur in
a linear array of steps so that the product of one step is the feed for the next.
Natural processes are often more complicated because they can switch from
one path to another in a stochastic manner or because there is one or more
feedback loops in the overall process. Regardless of the complexity of the
situation, the divide and analyze strategy is the most effective way to under-
stand how observed patterns are related to unit processes. The challenge of
interpreting pattern-forming processes is a very exciting scientific frontier
(Ball, 1999; Nicolis and Prigogine, 1989).
Geochemists observe patterns of element and mineral distribution
in nature and then use forward and inverse models of unit processes to
understand how these patterns developed. The patterns result from the
205
206 Pattern formation
interplay between energy flow and feedback in a network of unit proc-

esses, which means that the element and mineral distribution patterns
preserve valuable information about their conditions of formation. The
reductionist strategy of dividing the overall process into simple units is the
first step in interpreting the overall process. The second step is recogniz-
ing and understanding emergent behaviors caused by the feedback loops.
Non-equilibrium thermodynamics and the Turing model offer powerful
insights into the behavior of linked energy and mass flow processes with
feedback.
Non-equilibrium thermodynamics
Non-equilibrium thermodynamics (de Groot and Mazur, 1984; Prigogine,
1977) affords an abstract, and therefore very general, foundation for under-
standing pattern formation. The abstract nature of this approach makes
direct application to geochemical problems difficult, but non-equilibrium
thermodynamics does provide a powerful conceptual basis for thinking
about pattern formation.
Non-equilibrium thermodynamics uses the notion of rate of entropy pro-
duction (Kondepudi and Prigogine, 1998) to show that physical and chemi-
cal transformations can all be described in terms of forces and fluxes. For
example, the force driving a chemical reaction is the difference between the
chemical potential of the reactants and products. The greater this difference,
the faster the flux (rate) of conversion of reactants into products. The force
driving diffusion is the difference between a species’ chemical potentials
in two spatial regions. The diffusion flux from the region of high chemical
potential to the region of low chemical potential is directly proportional to
this force. Similar relationships exist between the rate of flow of electrical
current and voltage differences and the rate of heat flow and temperature
differences. In all these cases, a resistance limits the rate of transformation
or mass transfer. This resistance shows up in rate equations as the rate con-
stant and in diffusion equations as the diffusion coefficient divided by dis-
tance. This means that all these entropy-producing processes are analogous
at a very fundamental level.
An even more potent concept comes from the Onsager reciprocal relations,
which states that there is a coupling between conjugate force–flux pairs. For
example, mass transfer of species by diffusion in an aqueous solution causes
a change in concentration, which is accompanied by heat consumption or
release due to the heat of dilution. This sets up a thermal gradient, which
causes heat flow. The resulting link between heat flux and isothermal diffu-
sion is the Dufour effect. Its conjugate is the Soret effect, which is the diffu-
sional mass flux linked to heat flow. The Soret effect has been coupled with
Turing model 207
gravitational effects to separate chemical species and isotopes in chemical

technology (Platten et al., 2003; Powers and Wilke, 1957). It may be impor-
tant in certain geotechnical applications (Vidal and Murphy, 1999) and it
appears to cause separation of chemical species and isotopes in some geo-
logical settings (Cygan and Carrigan, 1992).
The relationship between forces and fluxes is simple only when the driving
force is small. As the driving force grows, this relationship becomes increas-
ingly nonlinear. This effect, combined with the coupling of processes, leads
to some surprising behaviors. The most important of these is spontaneous
self-organization (Nicolis and Prigogine, 1977; Prigogine, 1980; Prigogine
and Stengers, 1984). This means that many of the patterns observed in geo-
logical settings are the result of feedback in complex networks of nonlin-
early interacting forces and fluxes. Even though we can usually model the
unit processes involved, linking these models to predict the pattern-forming
behavior is challenging. Nonetheless, non-equilibrium thermodynamics
provides a quantitative and conceptual foundation for understanding pat-
tern formation.
Turing model
The coupling of individual processes, either due to Onsager reciprocal type
relationships or due to continuity requirements, produces feedbacks between
unit processes. Feedback effects control the location and structure of many
patterns. Feedback occurs when the output of a succeeding unit process
somehow affects the behavior of an antecedent unit process. Turing (1953)
quantified the relationship between feedback and pattern formation in bio-
logical systems and the Turing model has proved useful for explaining many
kinds of biological patterns (Kondo and Miura, 2010). It has great potential
for explaining many other kinds of pattern formation such as oscillating
chemical reactions and Liesegang bands. The Briggs–Rauscher reaction,
the Bray–Liebhafsky reaction, and the Belousov–Zhabotinsky reaction are
well-known examples of a growing number of recognized oscillating chemi-
cal reactions (Epstein et al., 1983; Field and Schneider, 1989) that exhibit
cyclical temporal patterns. These same reactions can also produce combined
temporal and spatial patterns, which appear as waves that move through the
reaction medium (Ross et al., 1988; Winfree, 1972). Liesegang (1913) rec-
ognized the occurrence of spatial patterns caused by precipitation linked to
diffusion and used this idea to explain band formation in agates (Liesegang,
1915). Although agate bands are probably not a good example of this phe-
nomenon (Heaney, 1993), Liesegang bands do occur in many geological set-
tings (Augustithis et al., 1980; Fu et al., 1994) and they are relatively easy
to produce in laboratory settings (Henisch, 1988; Krug et al., 1996; Müller
208 Pattern formation
et al., 1982). Feedback processes that link mass and energy fluxes are a key
feature of pattern formation.
Geological examples
Porphyry copper ore deposits (John, 2010) are a good example of how
energy and mass flux linked by feedback produce recognizable geologic pat-
terns. Heat from the mantle produces copper-enriched magma by partial
melting in the upper mantle or at the base of the crust. Because the magma
has a lower density than the surrounding rocks, it rises carrying along this
thermal energy until it is emplaced near the Earth’s surface where cooling
and crystallization release heat to the surroundings. The crystallizing magma
also releases copper-rich hydrothermal fluids with sufficient pressure to
hydrofracture a well-defined region near the crystallizing magma. The shape
of this region is controlled by a dynamic interaction between the magmatic
fluids and the surrounding meteoric water (Weiss et al., 2012). The cool-
ing hydrothermal solutions permeate the fractures where changes in pres-
sure and temperature cause the precipitation of copper-bearing minerals
in enough abundance to form an economic deposit. The porphyry copper
process is driven by energy transport from the upper mantle to the Earth’s
surface and this energy flow does work, some of which concentrates copper.
The location of the copper-enriched zone is fixed by feedback between the
fluid flow, permeability generation, and chemical precipitation. These linked
processes produce a distinct pattern of igneous features, rock alteration, iso-
topic distribution, etc., which form the basis for the exploration models used
to discover new deposits (John, 2010). Geothermal energy drives this aspect
of the porphyry copper pattern.
A subsequent phase of porphyry copper deposit pattern formation is
driven by solar energy, via the hydrologic cycle. The hydrologic cycle trans-
ports water from the oceans to the location of a porphyry copper deposit.
The resulting erosion exposes the sulfide-rich porphyry copper ore to the
atmosphere allowing the copper sulfides to oxidize and release copper into
solution. Some of the soluble copper is dispersed by flowing surface and
groundwater to produce copper dispersal patterns that are used to locate
deposits using geochemical exploration techniques (Rose et al., 1979). Some
of the copper-bearing groundwater infiltrates to the water table, where the
redox conditions change from oxidizing to reducing so the dissolved copper
reacts with pyrite and other sulfides preserved by the reducing conditions to
precipitate minerals that form an enrichment blanket. The enrichment blan-
ket often contains very high copper values and is a key exploration target.
Similar energy + mass flow with feedback processes are responsible for
the formation of other types of ore deposits. Ore deposits are a subset of
Geological examples 209
the large number of geological and geochemical processes that are driven
by energy and mass fluxes leading to pattern formation. Examples of a few
that have been modeled include patterns in stylolites (Merino, 1992), agates
(Wang and Merino, 1995), terra rosa (Merino and Banerjee, 2008), and bur-
ial dolomites (Merino and Canals, 2011) with many more examples given in
Jamtveit and Hammer (2012), Jamtveit and Meakin (1999), and Ortoleva
(1994). Creating general models that link pattern formation and unit proc-
esses like those described in this book is an exciting scientific frontier.
References
Adamson, A.W., Gast, A.P. (1997). Physical Chemistry of Surfaces. Wiley, New York.
Alkattan, M., Oelkers, E.H., Dandurand, J.L., Schott, J. (1998). An experimental study
of calcite and limestone dissolution rates as a function of pH from −1 to 3 and
temperature from 25 to 80°C. Chemical Geology, 151, 199–214.
Amis, E.S. (1966). Solvent Effects on Reaction Rates and Mechanisms. Academic Press,
New York.
Appelo, C.A.J. (1996). Multicomponent ion exchange and chromatography in natural
systems. In Reactive Transport in Porous Media, eds. Lichtner, P.C., Steefel, C.I.
Mineralogical Society of America, Washington D.C., pp. 193–227.
Applin, K.R., Lasaga, A.C. (1984). The determination of SO42−, NaSO4−, and MgSO4o
tracer diffusion coefficients and their application to diagenetic flux calculations.
Geochimica et Cosmochimica Acta, 48, 2151–2162.
Aris, R. (1956). On the dispersion of a solute in a fluid flowing through a tube.
Proceedings of the Royal Society A, 235, 67–77.
Aris, R. (1989). Elementary Chemical Reactor Analysis. Dover Publications,
New York.
Aris, R. (1994). Mathematical Modeling Techniques. Dover Publications, New York.
Arking, A. (1996). Absorption of solar energy in the atmosphere: Discrepancy between
model and observations. Science, 273, 779–782.
Arnold, F.H., Schofield, S.A., Blanch, H.W. (1986). Analytical affinity chromatography
I. Local equilibrium and the measurement of association and inhibition constants.
Journal of Chromatography, 355, 1–12.
Arvidson, R.S., Luttgbe, A. (2010). Mineral dissolution kinetics as a function of
distance from equilibrium: New experimental results. Chemical Geology, 269,
79–88.
Asano, T., le Nobel, W.J. (1978). Activation and reaction volumes in solution. Chemical
Reviews, 78, 407–489.
Ašperger, S. (2003). Chemical Kinetics and Inorganic Reaction Mechanisms, 2nd edn.
Kluwer Academic, New York.
Astruc, D. (1995). Electron transfer and radical processes in transition-metal chemistry.
VCH Publishers, Inc., New York.
Augustithis, S.S., Mposkos, E., Vgenopoulos, A. (1980). Diffusion rings (sphaeroids) in
bauxite. Chemical Geology, 30, 351–362.
Avrahami, M., Golding, R.M. (1968). The oxidation of the sulphide ion at very low
concentrations in aqueous solutions. Journal of the Chemical Society A, 647–651.
Avrami, M. (1939). Kinetics of phase change. I. Journal of Chemical Physics, 7,
1103–1112.
210
References 211
Avrami, M. (1940). Kinetics of phase change. II. Journal of Chemical Physics, 8,

212–224.
Axford, S.D.T. (1997). Aggregation of colloidal silica: Reaction-limited kernel,
stability ratio and distribution moments. Journal of the Chemical Society, Faraday
Transactions, 93, 303–311.
Bahr, J.M., Rubin, J. (1987). Direct comparison of kinetic and local equilibrium
formulations for solute transport affected by surface reactions. Water Resources
Research, 23, 438–452.
Ball, P. (1999). The Self-Made Tapestry. Oxford University Press, Oxford.
Bandstra, J.Z., Brantley, S.L. (2008). Surface evolution of dissolving minerals investigated
with a kinetic Ising model. Geochimica et Cosmochimica Acta, 72, 2587–2600.
Barenblatt, G.I. (2003). Scaling. Cambridge University Press, Cambridge, U.K.
Barnes, H.H. (1967). Roughness Characteristics of Natural Channels. U.S. Geological
Survey, Washington D.C., p. 213.
Basolo, F., Pearson, R.G. (1967). Mechanisms of Inorganic Reactions, 2nd edn. John
Wiley & Sons, New York.
Bender, E.A. (1978). An Introduction to Mathematical Modeling. Wiley-Interscience,
New York.
Berger, G., Cadore, E., Schott, J., Dove, P.M. (1994). Dissolution rate of quartz in
lead and sodium electrolyte solutions between 25 and 300°C: Effect of the nature
of surface complexes and reaction affinity. Geochimica et Cosmochimica Acta, 58,
541–551.
Berner, E.K., Berner, R.A. (1987). The Global Water Cycle. Prentice-Hall, Inc.,
Englewood Cliffs, N.J.
Berner, R.A. (1980). Early Diagenesis. Princeton University Press, Princeton.
Berner, R.A., Sjöberg, E.L., Velbel, M.A., Krom, M.D. (1980). Dissolution of
pyroxenes and amphiboles during weathering. Science, 207, 1205–1206.
Bethke, C.M. (2007). Geochemical and Biogeochemical Reaction Modeling, 2nd edn.
Oxford University Press, New York.
Bevington, P.R. (1969). Data Reduction and Error Analysis for the Physical Sciences.
McGraw-Hill, New York.
Bielski, B.H.J., Cabelli, D.E., Arudi, R.L. (1985). Reactivity of HO2/O2− radicals in
aqueous solution. Journal of Chemical and Physical Reference Data, 14, 1041–1100.
Bishop, K.J.M., Wilmer, C.E., Soh, S., Grzybowski, B.A. (2009). Nanoscale forces and
their uses in self-assembly. Small, 5, 1600–1630.
Bjerklie, D.M., Dingman, S.L. (2005). Comparison of constitutive flow resistance
equations based on the Manning and Chezy equations applied to natural rivers.
Water Resources Research, 41, W11502.
Boerlage, S.F.E., Kennedy, M.D., Bremere, I., Witkamp, G.J., Van der Hoek, J.P.,
Schippers, J.C. (2002). The scaling potential of barium sulphate in reverse osmosis
systems. Journal of Membrane Science, 197, 251–268.
Brantley, S.L., Crane, S.R., Crerar, D.A., Hellmann, R., Stallard, R. (1986a).
Dissolution at dislocation etch pits in quartz. Geochimica et Cosmochimica Acta, 50,
2349–2361.
Brantley, S.L., Crane, S.R., Crerar, D.A., Hellmann, R., Stallard, R. (1986b).
Dislocation etch pits in quartz. In Geochemical Processes at Mineral Surfaces, eds.
Davis, J.A., Hayes, K.F. American Chemical Society, Washington D.C., pp. 635–649.
212 References
Brezonik, P.L. (1994). Chemical Kinetics and Process Dynamics in Aquatic Systems.
Lewis Publishers, Boca Raton F.L.
Bridgman, P.W. (1931). Dimensional Analysis. Yale University Press, New Haven C.T.
Bruckner, R. (2002). Advanced Organic Chemistry Reaction Mechanisms. Harcourt
Academic Press, San Diego C.A.
Brunauer, S., Emmett, P.H., Teller, E. (1938). Adsorption of gases in multimolecular
layers. Journal of the American Chemical Society, 60, 309–319.
Bunge, H.J. (1997). Some remarks on modelling and simulation of physical phenomena.
Textures and Microstructures, 28, 151–165.
Burd, A.B., Jackson, G.A. (2009). Particle aggregation. Annual Reviews of Marine
Science, 1, 65–90.
Burkin, A.R. (2001). Chemical Hydrometallurgy. Imperial College Press, London.
Burrows, N.D., Yuwono, V.M., Penn, R.L. (2010). Quantifying the kinetics of crystal
growth by oriented aggregation. MRS Bulletin, 35, 133–137.
Burton, W.K., Cabrera, N. (1949). Crystal growth and surface structure: Part 1.
Discussions of the Faraday Society, 5, 33–39.
Burton, W.K., Cabrera, N., Frank, F.C. (1949). Role of dislocations in crystal growth.
Nature, 163, 398–399.
Burton, W.K., Cabrera, N., Frank, F.C. (1951). The growth of crystals and the
equilibrium structure of their surfaces. Philosophical Transactions Royal Society of
London, 243, 299–358.
Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, A.B. (1988). Critical review of rate
constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals
(•OH/•O−) in aqueous solution. Journal of Physical and Chemical Reference Data, 17,
513–886.
Buxton, G.V., Mulazzani, Q.G., Ross, A.B. (1995). Critical review of rate constants for
reactions of transients from metal ions and metal complexes in aqueous solution.
Journal of Chemical and Physical Reference Data, 24, 1055–1349.
Cabrera, N., Burton, W.K. (1949). Crystal growth and surface structure: Part 2.
Discussions of the Faraday Society, 5, 40–48.
Cabrera, N., Vermiyea, D.A. (1958). The growth of crystals from solution. In Growth
and Perfection of Crystals, eds. Doremus, R.H., Roberts, B.W., Turnbull, D. John
Wiley & Sons, New York, pp. 393–410.
Campanario, J.M. (1995). Automatic ‘balancing’ of chemical equations. Computers in
Chemistry, 19, 85–90.
Capellos, C., Beielski, B.H.J. (1972). Kinetic Systems. Wiley-Interscience, New York.
Carslaw, H.S., Jaeger, J.C. (1959). Conduction of Heat in Solids. Oxford University Press,
New York.
Casado, J., López-Quintela, M.A., Lorenzo-Barral, F.M. (1986). The initial rate method
in chemical kinetics. Journal of Chemical Education, 63, 450–452.
Casey, W.H. (2001). A view of reactions at mineral surfaces from the aqueous phase.
Mineralogical Magazine, 65, 323–337.
Casey, W.H., Swaddle, T.W. (2003). Why small? The use of small inorganic clusters to
understand mineral surface and dissolution reactions in geochemistry. Reviews in
Geophysics, 41, 1008.
References 213
Casey, W.H., Westrich, H.R., Banfield, J.F., Ferruzzi, G., Arnold, G.W. (1993). Leaching
and reconstruction at the surface of dissolving chain-silicate minerals. Nature, 366,
253–256.
Cazes, J., Scott, R.P.W. (2002). Chromatography Theory. Marcel Dekker, Inc.,
New York.
Ceccarello, S., Black, S., Read, D., Hodson, M.E. (2004). Industrial radioactive barite
scale: Suppression of radium uptake by introduction of competing ions. Minerals
Engineering, 17, 323–330.
Chaïrat, C., Schott, J., Oelkers, E.H., Lartigue, J.-E., Harouiya, N. (2007). Kinetics
and mechanism of natural fluroapatite dissolution at 25°C and pH from 3 to 12.
Chaudhry, M.H. (2008). Open-Channel Flow. Springer, New York.
Chermak, J.A., Rimstidt, J.D. (1990). Hydrothermal transformation rate of kaolinite to
muscovite/illite. Geochimica et Cosmochimica Acta, 54, 2979–2990.
Chotantarat, S., Ong, S.K., Sutthirat, C., Osathaphan, K. (2011). Effect of pH on
transport of Pb2+, Mn2+, Zn2+, and Ni2+ through lateritic soil: Column experiments
and transport modeling. Journal of Environmental Sciences, 23, 640–648.
Cölfen, H., Antonietti, M. (2008). Mesocrystals and Nonclassical Crystallization. John
Wiley & Sons, Chichester, U.K.
Colombani, J. (2008). Measurement of the pure dissolution rate constant of a mineral in
water. Geochimica et Cosmochimica Acta, 72, 5634–5640.
Crank, J. (1975). The Mathematics of Diffusion. Oxford University Press, New York.
Crutchfield, J.P. (2012). Between order and chaos. Nature Physics, 8, 17–24.
Cukierman, S. (2006). Et tu, Grotthuss! and other unfinished stories. Biochimica et
Biophysica Acta, 1725, 876–885.
Curti, E. (1997). Coprecipitation of Radionucleides: basic concepts, literature review, and
first applications. Wettingen, Switzerland, p. 107.
Curti, E. (1999). Coprecipitation of radionuclides with calcite: Estimation of partition
coefficients based on a review of laboratory investigations and geochemical data.
Applied Geochemistry, 14, 433–445.
Cussler, E.L. (2009). Diffusion, 3rd edn. Cambridge University Press, Cambridge.
Cygan, R.T., Carrigan, C.R. (1992). Time-dependent Soret transport: Applications to
brine and magma. Chemical Geology, 95, 201–212.
Damasceno, P.F., Engel, M., Glotzer, S.C. (2012). Predictive self-assembly of polyhedra
into complex structures. Science, 337, 453–457.
Danckwerts, P.V. (1953). Continuous flow systems. Distribution of residence times.
Chemical Engineering Science, 2, 1–13.
Davidson, L.E., Shaw, S., Benning, L.G. (2008). The kinetics and mechanism of
schwertmannite transformation to goethite and hematite under alkaline conditions.
American Mineralogist, 93, 1326–1337.
de Groot, S.R., Mazur, P. (1984). Non-equilibrium Thermodynamics. Dover Publications,
New York.
de Pablo, J., Casas, I., Giménez, J., Molera, M., Rovira, M., Duro, L., Bruno, J. (1999).
The oxidative dissolution mechanism of uranium dioxide. I. The effect of temperature
in hydrogen carbonate medium. Geochimica et Cosmochimica Acta, 63, 3079–3103.
214 References
Debessy, J., Pagel, M., Beny, J.-M., Christensen, H., Hickel, B., Kosztolanyi, C., Poty,
B. (1988). Radiolysis evidenced by H2–O2 and H2-bearing fluid inclusions in three
uranium deposits. Geochimica et Cosmochimica Acta, 52, 1155–1167.
Deming, W.E. (1943). Statistical Adjustment of Data. Dover Publications, New York.
Denbigh, K.G., Turner, J.C.R. (1984). Residence-time distributions, mixing, and
dispersion. In Chemical Reactor Theory: An Introduction, 3rd edn. Cambridge
University Press, New York, pp. 81–110.
Denny, M.W. (1993). Air and Water. Princeton University Press, Princeton.
Derome, D., Cathelineau, M., Lhomme, T., Cuney, M. (2003). Fluid inclusion evidence
of the differential migration of H2 and O2 in the McArthur River unconformity-type
uranium deposit (Saskatchewan, Canada). Possible role on post-ore modifications of
the host rocks. Journal of Geochemical Exploration, 78–79, 525–530.
Dietzel, M. (2000). Dissolution of silicates and the stability of polysilicic acid.
Doerner, H.A., Hoskins, W.M. (1925). Co-precipitation of radium and barium sulfates.
Journal of the American Chemical Society, 47, 662–675.
Dominé, F., Dounaceur, R., Scacchi, G., Marquaire, P.-M., Dessort, D., Pradier, B.,
Brevart, O. (2002). Up to what temperature is petroleum stable? New insights from a
5200 free radical reactions model. Organic Geochemistry, 33, 1487–1499.
Dotson, N.A., Galván, R., Laurence, R.L., Tirrel, M. (1996). Polymerization Process
Modeling. VCH Publishers, Inc., New York.
Douglas, J.F. (1969). Dimensional Analysis for Engineers. Sir Isaac Pitman & Sons Ltd.,
London.
Dove, P.M. (1994). The dissolution kinetics of quartz in sodium chloride solutions at
25°C to 300°C. American Journal of Science, 294, 665–712.
Dove, P.M., Han, N., De Yoreo, J.J. (2005). Mechanisms of classical crystal growth
theory explain quartz and silicate dissolution behavior. Proceedings of the National
Academy of Sciences, 102, 15357–15362.
Draganic, I.G. (2005). Radiolysis of water: A look at its origin and occurrence in the
nature. Radiation Physics and Chemistry, 72, 181–186.
Dria, M.A., Bryant, S.L., Schechter, R.S., Lake, L.W. (1987). Interacting precipitation/
dissolution waves: The movement of inorganic contaminants in groundwater. Water
Resources Research, 23, 2076–2090.
Drljaca, A., Hubbard, C.D., van Dldik, R., Asano, T., Basilevsky, M.V., le Nobel,
W.J. (1988). Activation and reaction volumes in solution. 3. Chemical Reviews, 98,
2167–2298.
Druschel, G.K., Hamers, R.J., Luthe IIIr, G.W., Banfield, J.F. (2003). Kinetics and
mechanism of trithionate and tetrathionate oxidation at low pH by hydroxyl
radicals. Aquatic Geochemistry, 9, 145–164.
Dunning, W.J. (1955). Theory of crystal nucleation from vapor, liquid, and solid
systems. In Chemistry of the Solid State, ed. Garner, W.E. Academic Press, Inc., New
York, pp. 159–183.
Dunning, W.J. (1969). General and theoretical introduction. In Nucleation, ed.
Zettlemoyer, A.C. Marcel Dekker, Inc., New York, pp. 1–67.
Dutrizac, J.E. (2002). Calcium sulphate solubilities in simulated zinc processing
solutions. Hydrometallurgy, 2002, 109–135.
References 215
Eckert, C.A. (1972). High pressure kinetics in solution. Annual Review of Physical
Chemistry, 239–264.
Edwards, J.O., Green, E.F., Ross, J. (1968). From stoichiometry and rate law to
mechanism. Journal of Chemical Education, 45, 381–385.
Edzwald, J.K., Upchurch, J.B., O’Melia, C.R. (1974). Coagulation in estuaries.
Environmental Science and Technology, 8, 58–63.
El-Kadi, A., Plummer, L.N., Aggarwal, P. (2011). NETPATH-WIN: An interactive user
version of the mass-balance model, NETPATH. Groundwater, 49, 593–599.
Epstein, I.R., Kustin, K., DeKepper, P., Orbán, M. (1983). Oscillating chemical
reactions. Scientific American, 248, 112–123.
Erdemir, D., Lee, A.Y., Myerson, A.S. (2009). Nucleation of crystals from solution:
Classical and two-step models. Accounts of Chemical Research, 42, 621–629.
Erdey-Grúz, T. (1974). Transport Phenomena in Aqueous Solutions. John Wiley & Sons,
New York.
Ershov, B.G., Gordeev, A.V. (2008). A model for radiolysis of water and aqueous
solutions. Radiation Physics and Chemistry, 77, 928–935.
Evans, M.G., Polanyi, M. (1935). Some applications of the transition state method
to the calculation of reaction velocities, especially in solution. Transactions of the
Faraday Society, 31 875–894.
Eyring, H. (1935). The activated complex in chemical reactions. Journal of Chemical
Physics, 3, 107–115.
Feth, J.H., Robertson, C.E., Polzer, W.L. (1964). Sources of mineral constituents in water
from granitic rocks, Sierra Nevada, California and Nevada. U.S. Geological Survey.
Field, R.J., Schneider, F.W. (1989). Oscillating chemical reactions and nonlinear
dynamics. Journal of Chemical Education, 66, 195–204.
Fokin, V.M., Zanotto, E.D., Schmelzer, J.W.P. (2010). On the thermodynamic driving
force for interpretation of nucleation experiments. Journal of Non-Crystalline Solids,
356, 2185–2191.
Ford, I.J. (2004). Statistical mechanics of nucleation: A review. Journal of Mechanical
Engineering Science, 218, 883–899.
Fournier, R.O. (1977). Chemical geothermometers and mixing models for geothermal
systems. Geothermics, 5, 41–50.
Fu, L., Milliken, K.L., Sharp Jr., J.M. (1994). Porosity and permeability variations
in fractured and Liesegang-banded Breathitt sandstones (Middle Pennsylvanian),
eastern Kentucky: Diagenetic controls and implications for modeling dual porosity
systems. Journal of Hydrology, 154, 351–381.
Fubini, B. (1998). Surface chemistry and quartz hazard. Annals of Occupational
Hygiene, 42, 521–530.
Fueno, T. (1999). The Transition State. Gordon & Breach, Amsterdam, p. 329.
Furukawa, Y., Shimada, W. (1993). Three-dimensional pattern formation during growth
of ice dendrites – its relation to universal law of dendritic growth. Journal of Crystal
Growth, 128, 234–239.
Garg, L.C., Maren, T.H. (1972). The rates of hydration of carbon dioxide and
dehydration of carbonic acid at 37°C. Biochimica et Biophysical Acta, 261, 70–76.
Garn, P.D. (1975). An examination of the kinetic compensation effect. Journal of
Thermal Analysis, 7, 475–478.
216 References
Garrels, R.M., MacKenzie, F.T. (1967). Origin of the chemical composition of some
springs and lakes. In Equilibrium Concept in Natural Water Systems, ed. Gould, R.F.
American Chemical Society, Washington D.C., pp. 222–242.
Garten, V.A., Head, R.B. (1970). Homogeneous nucleation in aqueous solution. Journal
of Crystal Growth, 6, 349–351.
Gauch Jr, H.G. (1993). Prediction, parsimony and noise. American Scientist, 81,
468–478.
Gebauer, D., Völkel, A., Cölfen, H. (2008). Stable prenucleation calcium carbonate
clusters. Science, 322, 1819–1822.
Gelhar, L.W., Welty, C., Rehfeldt, K.R. (1992). A critical review of data on field-scale
dispersion in aquifers. Water Resources Research, 28, 1955–1974.
Gibbs, G.V., Cox, D.F., Ross, N.L., Crawford, T.D., Burt, J.B., Rosso, K.M. (2005).
A mapping of the electron localization function for earth materials. Physics and
Chemistry of Minerals, 32, 208–221.
Gillespie, R.J., Robinson, E.A. (2006). Gilbert N. Lewis and the chemical bond:
The electron pair and the octet rule from 1916 to the present day. Journal of
Computational Chemistry, 28, 87–97.
Gimarc, B.M. (1974). Applications of qualitative molecular orbital theory. Accounts of
Chemical Research, 7, 384–392.
Goldich, S.S. (1938). A study in rock weathering. Geology, 46, 17–58.
Goldschmidt, V.M. (1958). Geochemistry. Clarendon Press, Oxford.
Gordon, L., Rowley, K. (1957). Coprecipitation of radium with barium sulfate.
Analytical Chemistry, 29, 34–37.
Greenberg, A.E., Clesceri, L.S., Eaton, A.D. (1992). Standard Methods for the
Examination of Water and Wastewater, 18th edn. American Public Health
Association, Washington D.C.
Grossman, R.B. (1999). The Art of Writing Reasonable Reaction Mechanisms. Springer,
New York.
Gunnarson, I., Arnórsson, S. (2003). Silica scaling: The main obstacle in efficient use of
high-temperature geothermal fluids. International Geothermal Conference, Reykjavík,
Iceland, pp. 30–36.
Gupta, A., Yan, D.S. (2006). Mineral Processing Design and Operation. Elsevier,
Amsterdam.
Gutmann, V. (1978). The Donor–Acceptor Approach to Molecular Interactions. Plenum
Press, New York.
Guy, B. (1993). Mathematical revision of Korzhinskii’s theory of infiltration
metasomatic zoning. European Journal of Mineralogy, 5, 317–339.
Hall, C., Day, J. (1977). Systems and Models: terms and basic principles. In Ecosystem
Modeling in Theory and Practice, eds Hall, C., Day, J. Wiley-Interscience, New York,
pp. 6–36.
Hall, C., Day, J., Odum, H. (1977). A circuit language for energy and matter. In
Ecosystem Modeling in Theory and Practice, eds Hall, C., Day, J. Wiley-Interscience,
New York, pp. 37–49.
Harris, R.L. (1999). Information Graphics. Oxford University Press, Oxford.
Harte, J. (1988). Consider a Spherical Cow. University Science Books, Mill
Valley, C.A.
References 217
Hartman, P., Perdok, W.G. (1955a). On the relations between structure and morphology
of crystals. I. Acta Crystallographica, 8, 49–52.
Hartman, P., Perdok, W.G. (1955b). On the relations between structure and morphology
of crystals. II. Acta Crystallographica, 8, 521–524.
Hartman, P., Perdok, W.G. (1955c). On the relations between structure and morphology
of crystals. III. Acta Crystallographica, 8, 525–529.
Harvey, M.C., Schreiber, M.E., Rimstidt, J.D., Griffith, M.A. (2006). Scorodite
dissolution kinetics: Implications for arsenic release. Environmental Science and
Technology, 40, 6709–6714.
He, S., Oddo, J.E., Tomson, M.B. (1994). The inhibition of gypsum and barite nucleation
in NaCl brines at temperatures from 25 to 90°C. Applied Geochemistry, 9, 561–567.
Heaney, P.J. (1993). A proposed mechanism for the growth of chalcedony. Contributions
to Mineralogy and Petrology, 115, 66–74.
Heizmann, J.J., Bessieres, J., Bessieres, A. (1986). Advances in kinetic models. Journal de
chimie physique, 83, 725–732.
Helfferich, F.G. (1989). The theory of precipitation/dissolution waves. AIChE Journal,
35, 75–87.
Helgeson, H.C. (1968). Evaluation of irreversible reactions in geochemical processes
involving minerals and aqueous solutions – I. Thermodynamic relations. Geochimica
et Cosmochimica Acta, 32, 853–877.
Helgeson, H.C. (1979). Mass transfer among minerals and hydrothermal solution. In
Geochemistry of Hydrothermal Ore Deposits, second edn., ed. Barnes, H.L. John
Wiley & Sons, New York, pp. 568–610.
Helgeson, H.C., Garrels, R.M., MacKenzie, F.T. (1969). Evaluation of irreversible
reactions in geochemical processes involving minerals and aqueous solutions – II.
Applications. Geochimica et Cosmochimica Acta, 33, 455–481.
Hellmann, R., Eggleston, C.M., Hochella Jr., M.F., Crerar, D.A. (1990). The formation
of leached layers on albite surfaces during dissolution under hydrothermal
conditions. Geochimica et Cosmochimica Acta, 54, 1267–1281.
Hem, J.D. (1985). Study and Interpretation of the Chemical Characteristics of Natural
Water. U.S. Geological Survey, Washington D.C.
Hendricks, D. (2006). Water Treatment Unit Processes. Taylor & Francis/CRC Press,
Boca Raton, F.L.
Henisch, H.K. (1988). Crystals in Gels and Liesegang Rings. Cambridge University
Press, Cambridge.
Higgins, G.H. (1959). Evaluation of the ground-water contamination hazard from
underground nuclear explosions. Journal of Geophysical Research, 64,
1509–1519.
Hill Jr., C.G. (1977). An Introduction to Chemical Engineering Kinetics and Reactor
Design. John Wiley & Sons, New York.
Hofmann, A. (1972). Chromatographic theory of infiltration metasomatism and its
application to feldspars. American Journal of Science, 272, 69–90.
Horne, R.L., Rodriguez, F. (1983). Dispersion in tracer flow in fractured geothermal
systems. Geophysical Research Letters, 10, 289–292.
Houston, P.L. (2001). Chemical Kinetics and Reaction Dynamics. Dover Publications,
New York.
218 References
Huminicki, D.M.C., Rimstidt, J.D. (2009). Iron oxyhydroxide coating of pyrite for acid
mine drainage control. Applied Geochemistry, 24, 1626–1634.
Huntley, H.E. (1967). Dimensional Analysis. Dover Publications, New York.
Huss Jr., A., Lim, P.K., Eckert, C.A. (1982). Oxidation of aqueous sulfur dioxide.
2. High-pressure studies and proposed reaction mechanisms. Journal of Physical
Chemistry, 86, 4229–4233.
Icopini, G.A., Brantley, S.L., Heaney, P.J. (2005). Kinetics of silica oligomerization and
nanocolloid formation as a function of pH and ionic strength at 25°C. Geochimica et
Cosmochimica Acta, 69, 293–303.
Ingold, C.K. (1969). Structure and Mechanism in Organic Chemistry. Cornell University
Press, Ithaca.
Jamtveit, B., Hammer, Ø. (2012). Sculpting of rocks by reactive fluids. Geochemical
Perspectives, 1, 341–480.
Jamtveit, B., Meakin, P. (1999). Growth, Dissolution and Pattern Formation in
Geosystems. Kluwer Academic Publishers, Dordrecht.
Jensen, W.B. (1980). The Lewis Acid-Base Concepts. John Wiley & Sons, New York.
Jin, L., Auerbach, S.M., Monson, P.A. (2011). Modeling three-dimensional network
formation with an atomic lattice model: Application to silicic acid polymerization.
Journal of Chemical Physics, 134, 134703.
John, D.A. (2010). Porphyry Copper Deposit Model. U.S. Geological Survey,
Washington D.C., p. 169.
Kashchiev, D., van Rosmalen, G.M. (2003). Review: Nucleation in solution revisited.
Crystal Research and Technology, 38, 555–574.
Kirby, C.S., Cravotta III, C.A. (2005a). Net alkalinity and net acidity 1: Theoretical
considerations. Applied Geochemistry, 20, 1920–1940.
Kirby, C.S., Cravotta III, C.A. (2005b). Net alkalinity and net acidity 2: Practical
considerations. Applied Geochemistry, 20, 1941–1964.
Kline, S.J. (1965). Similitude and Approximation Theory. McGraw-Hill Book Company,
New York.
Knapp, R.B. (1989). Spatial and temporal scales of local equilibrium in dynamic fluid-
rock systems. Geochimica et Cosmochimica Acta, 53, 1955–1964.
Kondepudi, D., Prigogine, I. (1998). Modern Thermodynamics. John Wiley & Sons,
Chichester, U.K.
Kondo, S., Miura, T. (2010). Reaction–diffusion model as a framework for
understanding biological pattern formation. Science, 329, 1616–1620.
Korzhinskii, D.S. (1970). Theory of Metasomatic Zoning. Clarendon Press, Oxford, London.
Krishnamurthy, E.V. (1978). Generalized matrix inverse approach for automatic
balancing of chemical equations. International Journal of Mathematical Education in
Science and Technology, 9, 323–328.
Krug, H.-J., Brandstädter, H., Jacob, K.H. (1996). Morphological instabilities in pattern
formation by precipitation and crystallization processes. Geologische rundschau, 85, 19–28.
Laidler, K.J. (1987a). Theories of Reaction Rates. Harper & Row, New York, pp. 80–99.
Laidler, K.J. (1987b). Chemical Kinetics, 3rd edn. Harper & Row, New York.
Lamberto, D.J., Alverez, M.M., Muzzio, F.J. (1999). Experimental and computational
investigation of the laminar flow structure in a stirred tank. Chemical Engineering
Science, 54, 919–942.
References 219
Larson, M.A., Garside, J. (1986). Solute clustering and interfacial tension. Journal of
Crystal Growth, 76, 88–92.
Lasaga, A.C. (1980a). Dynamic treatment of geochemical cycles: global kinetics. In:
Kinetics of Geochemical Processes, eds Lasaga, A.C., Kirkpatrick, R.J. Mineralogical
Society of America, Washington D.C., pp. 69–105.
Lasaga, A.C. (1980b). The kinetic treatment of geochemical cycles. Geochimica et
Lasaga, A.C. (1998a). Geochemical Kinetics. Princeton University Press,
Princeton N.J.
Lasaga, A.C. (1998b). Kinetic Theory in Earth Sciences. Princeton, University Press,
Princeton, N.J.
Lasaga, A.C., Berner, R.A. (1998). Fundamental aspects of quantitative models for
geochemical cycles. Chemical Geology, 145, 161–175.
Lasaga, A.C., Blum, A.E. (1986). Surface chemistry, etch pits and mineral-water
reactions. Geochimica et Cosmochimica Acta, 50, 2363–2379.
Lasaga, A.C., Gibbs, G. (1990). Ab-initio quantum mechanical calculations of water-
rock interactions: Adsorbtion and hydrolysis reactions. American Journal of Science,
290, 263–295.
Laub, R.J. (1985). Theory of chromatography. In Inorganic Chromatographic Analysis,
ed. Macdonald, J.C. John Wiley & Sons, New York, pp. 13–186.
Le Caër, S. (2011). Water radiolysis: Influence of oxide surfaces on H2 production under
ionizing radiation. Water, 3, 235–253.
Lefticariu, L., Pratt, L.A., LaVern, J.A., Schimmelmann, A. (2010). Anoxic pyrite
oxidation by water radiolysis products – A potential source of biosustaining energy.
Earth and Planetary Science Letters, 292, 57–67.
Leifer, A. (1988). The Kinetics of Envrionmental Aquatic Photochemistry. American
Chemical Society, Washington D.C.
Lerman, A. (1979). Geochemical Processes: Water and Sediment Environments. Wiley-
Interscience, New York.
Lerman, A., Wu, L. (2008). Kinetics of global geochemical cycles. In Kinetics of Water-
Rock Interaction, eds Brantley, S.L., Kubicki, J.D., White, A.F. Springer, New York,
pp. 655–736.
Leubner, I.H. (2000). Particle nucleation and growth models. Current Opinion in Colloid
and Interface Science, 5, 151–159.
Levenspiel, O. (1972a). Chemical Reaction Engineering. John Wiley & Sons,
New York.
Levenspiel, O. (1972b). Single Ideal Reactors, Chemical Reaction Engineering, 2nd edn.
John Wiley & Sons, New York, pp. 91–117.
Levich, V.G. (1962). Physicochemical Hydrodynamics. Prentice-Hall, Inc., Englewood
Cliffs, N.J.
Lewis, G.N. (1916). The atom and the molecule. Journal of the American Chemical
Society, 38, 762–785.
Li, Y.-H., Gregory, S. (1974). Diffusion of ions in sea water and in deep-sea sediments.
Liesegang, R.E. (1913). Geologische Diffusionen. Verlag von Theodor Steinkopff,
Dresden.
220 References
Liesegang, R.E. (1915). Die Achate. Verlag von Theodor Steinkopff, Dresden.
Limerinos, J.T. (1970). Manning Coefficient from Measured Bed Roughness in Natural
Channels, U.S. Geological Survey Water Supply Paper 1898-B.
Lin, L.-H., Slater, G.F., Lollar, B.S., Lacrampe-Couloume, G., Onstott, T.C. (2005). The
yield and isotopic composition of radiolytic H2, a potential energy source for the
deep subsurface biosphere. Geochimica et Cosmochimica Acta, 69, 893–903.
Liu, J., Aruguete, D.M., Jinschek, J.R., Rimstidt, J.D., Hochella Jr., M.F. (2008). The
non-oxidative dissolution of galena nanocrystals: Insights into mineral dissolution
rates as a function of grain size, shape, and aggregation state. Geochimica et
Liu, L., Guo, Q.-X. (2001). Isokinetic relationship, isoequilibrium relationship, and
enthalpy–entropy compensation. Chemical Reviews, 101, 673–695.
Liu, S.-T., Nancollas, G.H. (1971). The kinetics of dissolution of calcium sulfate
dihydrate. Journal of Inorganic and Nuclear Chemistry, 33, 2311–2316.
Lowell, S., Shields, J.E. (1991). Powder Surface Area and Porosity, 3rd edn. Chapman &
Hall, London.
MacInnis, I.N., Brantley, S.L. (1993). Development of etch pit size distributions on
dissolving minerals. Chemical Geology, 105, 31–49.
Maffezzoli, A., Kenny, J.M., Torre, L. (1995). On the physical dimensions of the Avrami
constant. Thermochimica Acta, 269/270, 185–190.
Mandel, J. (1964). The Statistical Analysis of Experimental Data. Dover Publications,
New York.
Marangoni, A.G. (1998). On the use and misuse of the Avrami equation in
characterization of the kinetics of fat crystallization. Journal of the American Oil
Chemists’ Society, 75, 1465–1467.
Marcus, R.A. (1964). Chemical and electrochemical electron-transfer theory. Annual
Reviews of Physical Chemistry, 155–196.
Marcus, R.A. (1968). Theoretical relations among rate constants, barriers, and Brønsted
slopes of chemical reactions. Journal of Physical Chemistry, 72, 891–898.
Marcus, R.A. (1985). Electron transfers in chemistry and biology. Biochimica et
Biophysica Acta, 811, 265–322.
Marcus, R.A. (2000). Tutorial on rate constants and reorganization energies. Journal of
Electroanalytical Chemistry, 483, 2–6.
Marin, G.B., Yablonsky, G.S. (2011). Kinetics of Chemical Reactions. Wiley-VCH,
Weinheim, Germany.
Markov, I.V. (2003). Crystal Growth for Beginners. World Scientific, London.
McCabe, W.L., Smith, J.C., Harriott, P. (1993). Unit Operations of Chemical
Engineering. McGraw-Hill, Inc., New York.
McClamroch, N.H. (1980). State Models of Dynamic Systems. Springer-Verlag,
New York.
McIntire, W.L. (1963). Trace element partition coefficients: A review of theory and
applications to geology. Geochimica et Cosmochimica Acta, 27, 1209–1264.
Meldrum, F.C., Sear, R.P. (2008). Now you see them. Science, 322, 1802–1103.
Merino, E. (1992). Self-organization in stylolites. American Scientist, 80, 466–473.
Merino, E., Banerjee, A. (2008). Terra rosa genesis, implications for karst, and eolian
dust: A geodynamic thread. Journal of Geology, 116, 62–75.
References 221
Merino, E., Canals, A. (2011). Self-accelerating dolomite-for-calcite replacement:

Self-organized dynamics of burial dolomitization and associated mineralization.
American Journal of Science, 311, 573–607.
Meyer, D. (1999). Surfaces, Interfaces, and Colloids, 2nd edn. John Wiley & Sons,
New York.
Miller, D.G. (1982). Estimation of Tracer Diffusion Coefficients of Ions in Aqueous
Solution. Lawrence Livermore Laboratory, Livermore, C.A.
Millero, F.J. (2001). The oxidation of hydrogen sulfide in natural waters. In The Physical
Chemistry of Natural Waters. John Wiley & Sons, Inc., New York, pp. 582–632.
Moore, J.W., Pearson, R.G. (1981). Kinetics and Mechanism. John Wiley & Sons,
New York.
Müller, S.C., Kai, S., Ross, J. (1982). Curiosities in periodic precipitation patterns.
Science, 216, 635–637.
Murphy, G. (1950). Similitude in Engineering. Ronald Press Company, New York.
Mutaftschiev, B. (2001). The Atomistic Nature of Crystal Growth. Springer, Berlin.
Narayanasamy, J., Kubicki, J.D. (2005). Mechanism of hydroxyl radical generation from
a silica surface: Molecular orbital calculations. Journal of Physical Chemistry B, 109,
21796–21807.
Nauman, E.B. (2008). Residence time theory. Industrial & Engineering Chemistry
Research, 47, 3752–3766.
Neta, P., Huie, R.E., Ross, A.B. (1988). Rate constants for reactions of inorganic
radicals in aqueous solution. Journal of Chemical and Physical Reference Data, 17,
1027–1284.
Nicholson, R.V., Gillham, R.W., Reardon, E.J. (1990). Pyrite oxidation in carbonate-
buffered solution: 2. Rate control by oxide coatings. Geochemica et Cosmochimica
Acta, 54, 395–402.
Nico, P., Anastasio, C., Zasoski, R.J. (2002). Rapid photo-oxidation of Mn(II) mediated
by humic substances. Geochimica et Cosmochimica Acta, 66, 4047–4056.
Nicolis, G., Prigogine, I. (1977). Self-Organization in Nonequilibrium Systems. John
Wiley & Sons, New York.
Nicolis, G., Prigogine, I. (1989). Exploring Complexity. W.H. Freeman & Company,
New York.
Nielsen, A.E. (1964). Kinetics of Precipitation. Macmillan Company, New York.
Nkedl-Kizza, P., Rao, P.S.C., Hornsby, A.G. (1987). Influence of organic cosolvents on
leaching of hydrophobic organic chemicals through soils. Environmental Science and
Technology, 21, 1107–1111.
Nordstrom, D.K. (1977). Thermochemical redox equilibria of Zo Bell’s solution.
Nordstrom, D.K. (2012). Models, validation, and applied geochemistry: Issues in
science, communication, and philosophy. Applied Geochemistry, 27, 1899–1919.
Oelkers, E.H. (1991). Calculation of diffusion coefficients for aqueous organic species at
temperatures from 0 to 350°C. Geochimica et Cosmochimica Acta, 55, 3515–3529.
Oelkers, E.H., Helgeson, H.C. (1988). Calculation of the thermodynamic and transport
properties of aqueous species at high pressures and temperatures: Aqueous tracer
diffusion coefficients of ions to 1000°C and 5 kb. Geochimica et Cosmochimica Acta,
52, 63–85.
222 References
Ohlin, C.A., Villa, E.M., Rustad, J.R., Casey, W.H. (2010). Dissolution of insulating
oxide materials at the molecular scale. Nature Materials, 9, 11–19.
Oreskes, N., Shrader-Frechette, K., Belitz, K. (1994). Verification, validation, and
confirmation of numerical models in the Earth Sciences. Science, 263, 641–646.
Ortoleva, P.J. (1994). Geochemical Self-Organization. Oxford University Press, New York.
Overton, W. (1977). A strategy of model construction. In Ecosystem Modeling in
Theory and Practice: An introduction with case histories, eds, Hall, C., Day, J. Wiley-
Interscience, New York, pp. 50–73.
Palandri, J.L., Kharaka, Y.K. (2004). A compilation of rate parameters of water-mineral
interaction kinetics for application to geochemical modeling. U.S. Geological Survey,
Menlo Park, C.A., p. 64.
Pauling, L. (1960). The Nature of the Chemical Bond and the Structure of Molecules and
Crystals: An Introduction to Modern Structural Chemistry. Cornell University Press,
Ithaca, N.Y. (3rd edition).
Penn, R.L. (2004). Kinetics of oriented aggregation. Journal of Physical Chemistry B,
108, 12707–12712.
Penn, R.L., Tanaka, K., Erbs, J. (2007). Size dependent kinetics of oriented aggregation.
Journal of Crystal Growth, 309, 97–102.
Petrovich, R. (1981a). Kinetics of dissolution of mechanically comminuted rock-
forming oxides and silicates – I. Deformation and dissolution of quartz under
laboratory conditions. Geochimica et Cosmochimica Acta, 45, 1665–1674.
Petrovich, R. (1981b). Kinetics of dissolution of mechanically comminuted rock-
forming oxides and silicates – II. Deformation and dissolution of oxides and silicates
in the laboratory and at the Earth’s surface. Geochimica et Cosmochimica Acta, 45,
1675–1686.
Pina, C.M., Putnis, A. (2002). The kinetics of nucleation of solid solutions from
aqueous solutions: A new model for calculating non-equilibrium distribution
coefficients. Geochimica et Cosmochimica Acta, 66, 185–192.
Piscitelle, L.J. (1990). Determination of initial rates for a general class of chemical
reactions: A methodology. International Journal of Chemical Kinetics, 22, 683–688.
Platten, J.K., Bou-Ali, M.M., Dutrieux, J.F. (2003). Enhanced molecular separation
in inclined thermogravitational columns. Journal of Physical Chemistry B, 107,
11763–11767.
Plummer, L.N., Parkhurst, D.L., Thorstenson, D.C. (1983). Development of reaction
models for ground-water systems. Geochimica et Cosmochimica Acta, 47, 665–686.
Plummer, L.N., Prestemon, E.C., Parkhurst, D.L. (1991). An interactive code
(NETPATH) for modeling net geochemical reactions along a flow path. U.S.
Geological Survey, Reston, V.A., p. 227.
Plummer, L.N., Prestemon, E.C., Parkhurst, D.L. (1992). NETPATH: An interactive
code for interpreting NET geochemical reactions from chemical and isotopic data
along a flow PATH. In 7th International Symposium on Water-Rock Interaction–
WRI-7, eds Kharaka, Y.F., Maest, A.S. A.A. Balkema, Park City, U.T., pp.
239–242.
Pollard, J.H. (1977). A Handbook of Numerical and Statistical Techniques. Cambridge
University Press, Cambridge.
Powers, J.E., Wilke, C.R. (1957). Separation of liquids by thermal diffusion. A.I.Ch.E.
Journal, 3, 231–222.
References 223
Presnall, D.C. (1986). An algebraic method for determining equilibrium crystallization

and fusion paths in multicomponent systems. American Mineralogist, 71, 1061–1070.
Prigogine, I. (1977). Time, structure and fluctuations. Nobel Lecture December 8, 1977,
p. 23.
Prigogine, I. (1980). From Being to Becoming. W.H. Freeman & Company, San Francisco.
Prigogine, I., Stengers, I. (1984). Order out of Chaos. Bantam Books, Toronto.
Probstein, R.F. (1989). Physicochemical Hydrodynamics: An Introduction. Butterworths,
Boston.
Purcell, E.M. (1977). Life at low Reynolds number. American Journal of Physics, 45, 3–11.
Qian, J., Zhan, H., Zhao, W., Sun, F. (2005). Experimental study of turbulent
unconfined groundwater flow in a single fracture. Journal of Hydrology, 311,
134–142.
Rebreanu, L., Vanderborght, J.-P., Chou, L. (2008). The diffusion coefficient of
dissolved silica revisited. Marine Chemistry, 112, 230–233.
Rescigno, A., Thakur, A.K. (1988). Development of compartmental concepts. In
Pharmacokinetics: Mathematical and statistical approaches to metabolism and
distribution of chemicals and drugs, eds Pecile, A., Rescigno, A. Plenum Press, New
York, pp. 19–26.
Rimstidt, J.D. (1997a). Gangue mineral transport and deposition. In The Geochemistry
of Hydrothermal Ore Deposits, 3rd edn., ed. Barnes, H.L. John Wiley & Sons, New
York, pp. 487–515.
Rimstidt, J.D. (1997b). Quartz solubility at low temperatures. Geochimica et
Rimstidt, J.D., Balog, A., Webb, J. (1998). Distribution of trace elements between
carbonate minerals and aqueous solutions. Geochimica et Cosmochimica Acta, 62,
1851–1863.
Rimstidt, J.D., Barnes, H.L. (1980). The kinetics of silica-water reactions. Geochimica et
Rimstidt, J.D., Brantley, S.L., Olsen, A.A. (2012). Systematic review of forsterite
dissolution data. Geochimica et Cosmochimica Acta, 99, 159–178.
Rimstidt, J.D., Newcomb, W.D. (1993). Measurement and analysis of rate data: The rate of
reaction of ferric iron with pyrite. Geochimica et Cosmochimica Acta, 57, 1919–1934.
Rimstidt, J.D., Vaughan, D.J. (2003). Pyrite oxidation: A state-of-the-art assessment of
the reaction mechanism. Geochimica et Cosmochimica Acta, 67, 873–880.
Robie, R.A., Hemmingway, B.S. (1995). Thermodynamic Properties of Minerals and
Related Substances at 298.15 K and 1 Bar (105 Pascals) Pressure and at Higher
Temperatures. U.S. Geological Survey, Washington D.C., p. 461.
Robinson, R.A., Stokes, R.H. (1959). Electrolyte Solutions, 2nd edn. Dover
Publications, New York.
Rose, A.W., Hawkes, H.E., Webb, J.S. (1979). Geochemistry in Mineral Exploration.
Academic Press, London.
Ross, J., Müller, S.C., Vidal, C. (1988). Chemical waves. Science, 240, 460–465.
Rothbaum, H.P., Rohde, A.G. (1979). Kinetics of silica polymerization and deposition
from dilute solutions between 5 and 180°C. Journal of Colloid and Interface Science,
71, 533–559.
Rubinow, S.I. (1975). Tracers in physiological systems. In Introduction to Mathematical
Biology. John Wiley & Sons, New York, pp. 104–155.
224 References
Schechter, R.S., Bryant, S.L., Lake, L.W. (1987). Isotherm-free chromatography:

Propagation of precipitation/dissolution waves. Chemical Engineering
Communications, 58, 353–376.
Schepartz, B. (1980). Dimensional Analysis in the Biomedical Sciences. Charles C
Thomas, Springfield, I.L.
Schmidt, L.D. (1998). The Engineering of Chemical Reactions. Oxford University Press,
New York.
Schott, J., Oelkers, E.H. (1995). Dissolution and crystallization rates of silicate minerals
as a function of chemical affinity. Pure & Applied Chemistry, 67, 903–910.
Schott, J., Pokrovsky, O.S., Spalla, O., Devreux, F., Gloter, A., Mielczarski, J.A. (2012).
Formation, growth and transformation of leached layers during silicate minerals
dissolution: The example of wollastonite. Geochimica et Cosmochimica Acta, 98,
259–281.
Schulze-Makuch, D. (2005). Longitudinal dispersivity data and implications for scaling
behavior. Ground Water, 43, 443–456.
Shen, L., Chen, Z. (2007). Critical review of the impact of tortuosity on diffusion.
Chemical Engineering Science, 62, 3748–3755.
Shoemaker, C. (1977). Mathematical construction of ecological models. In Ecosystem
Modeling in Theory and Practice, eds Hall, C., Day, J. Wiley-Interscience, New York,
pp. 76–113.
Silverman, J., Dodson, R.W. (1952). The exchange reaction between the two oxidation
states of iron in acid solution. Journal of Physical Chemistry, 56, 846–852.
Sjöberg, E.L., Rickard, D. (1983). The influence of experimental design on the rate of
calcite dissolution. Geochimica et Cosmochimica Acta, 47, 2281–2285.
Sleutel, M., Maes, D., Van Driessche, A. (2012). What can mesoscopic level in situ
observations teach us about kinetics and thermodynamics of crystallization?
In: Kinetics and Thermodynamics of Multistep Nucleation and Self-Assembly in
Nanoscale Materials, eds Nicolis, G., Maes, D. John Wiley & Sons, Inc. Hoboken,
N.J.
Smetannikov, A.F. (2011). Hydrogen generation during the radiolysis of crystallization
water in carnallite and possible consequences of this process. Geochemistry
International, 49, 916–924.
Söhnel, O. (1982). Electrolyte crystal-aqueous solution interfacial tensions from
crystallization data. Journal of Crystal Growth, 57, 101–108.
Söhnel, O., Garside, J. (1988). Solute clustering and nucleation. Journal of Crystal
Growth, 89, 202–208.
Southworth, B.A. (1995). Hydroxyl radical production via the photo-Fenton reaction in
natural waters. Department of Civil and Environmental Engineering. Massachusetts
Institute of Technology, p. 188.
Sparks, D.L. (1989). Kinetics of Soil Chemical Processes. Academic Press, Inc., San
Diego.
Spinks, J.W.T., Woods, R.J. (1990). An Introduction to Radiation Chemistry, 3rd edn.
John Wiley & Sons, New York.
Staicu, C.I. (1982). Restricted and General Dimensional Analysis. Abacus Press,
Tunbridge Wells, Kent, U.K.
References 225
Steefel, C.I. (2008). Geochemical kinetics and transport. In Kinetics of Water-Rock

Interaction, eds Brantley, S.L., Kubicki, J.D., White, A.F. Springer, New York, pp.
545–589.
Steinmann, P., Lichtner, P.C., Shotyk, W. (1994). Reaction path approach to mineral
weathering reactions. Clays and Clay Minerals, 42, 197–206.
Stumm, W. (1990). Aquatic Chemical Kinetics. John Wiley & Sons, New York, p. 545.
Stumm, W. (1992). Chemistry of the Solid-Water Interface. John Wiley & Sons,
New York.
Sunagawa, I. (2005). Crystals: Growth, Morphology, and Perfections. Cambridge
University Press, Cambridge.
Taylor, G. (1953). Dispersion of soluble matter in solvent flowing slowly through a tube.
Proceedings of the Royal Society A, 219, 186–203.
Taylor, J.R. (1982). An Introduction to Error Analysis. University Science Books, Mill
Valley, C.A.
Teng, H.H., Dove, P.M., DeYoreo, J.J. (2000). Kinetics of calcite growth: Surface
processes and relationships to macroscopic rate laws. Geochimica et Cosmochimica
Acta, 64, 2255–2266.
Tester, J.W., Worley, W.G., Robinson, B.A., Grigsby, C.O., Feerer, J.L. (1994).
Correlating quartz dissolution kinetics in pure water from 25 to 625°C. Geochimica
et Cosmochimica Acta, 58, 2407–2420.
Tuckerman, M.E., Marx, D., Parrinello, M. (2002). The nature and transport
mechanism of hydrated hydroxide ions in aqueous solution. Nature, 417, 925–929.
Tufte, E.R. (2001). The Visual Display of Quantitative Information. Graphics Press,
Cheshire, CT.
Turing, A.M. (1953). The chemical basis of morphogenesis. Philosophical Transactions
of the Royal Society B, 237, 37–72.
Turnbull, D., Fisher, J.C. (1949). Rate of nucleation in condensed systems. Journal of
Chemical Physics, 17, 71–73.
Uhlmann, D.R., Chalmers, B. (1966). The energetics of nucleation. In Nucleation
Phenomena, eds Michaels, A.S. American Chemical Society, Washington D.C.,
pp. 1–13.
van Boekel, M.A.J.S. (2009). Kinetic Modeling of Reactions in Foods. CRC Press, Boca
Raton, F.L.
van Dldik, R., Asano, T., le Nobel, W.J. (1989). Activation and reaction volumes in
solution. 2. Chemical Reviews, 89, 549–688.
Van Herk, J., Pietersen, H.S., Schuiling, R.D. (1989). Neutralization of industrial
waste acids with olivine – the dissolution of forsteritic olivine at 40–70°C. Chemical
Geology, 76, 341–352.
Velbel, M.A. (1989). Weathering of hornblende to ferruginous products by a
dissolution–reprecipitation mechanism: Petrography and stoichiometry. Clays and
Clay Minerals, 37, 515–524.
Vidal, O., Murphy, W.M. (1999). Calculation of the effect of gaseous thermodiffusion
and thermogravitation processes on the relative humidity surrounding a high level
nuclear waste canister. Waste Management, 19, 189–198.
Vogel, S. (1994). Life in Moving Fluids, 2nd edn. Princeton University Press, Princeton, N.J.
226 References
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.
Wainer, H. (2005). Graphic Discovery. Princeton University Press, Princeton, NJ.
Waley, S.G. (1981). An easy method for the determination of initial rates. Biochemistry
Journal, 193, 1009–1012.
Walton, A.G. (1969). Nucleation in liquids and solutions. In Nucleation, ed.
Zettlemoyer, A.C. Marcel Dekker, Inc., New York, pp. 225–307.
Wang, Y., Merino, E. (1995). Origin of fibrosity and banding in agates from flood
basalts. American Journal of Science, 295, 49–77.
Watson, J.T.R., Basu, R.S., Sengers, J.V. (1980). An improved representation equation
for the dynamic viscosity of water substance. Journal of Physical and Chemical
Reference Data, 9, 1255–1290.
Weber Jr., W.J., DiGiano, F.A. (1996). Process Dynamics in Environmental Systems. John
Wiley & Sons, Inc., New York.
Wechsler, J. (1988). On aesthetics in science. In Design Science Collection, ed. Loeb, A.L.
Birkhäuser, Boston, p. 180.
Wehrli, B. (1989). Monte Carlo simulations of surface morphologies during mineral
dissolution. Journal of Colloid and Interface Science, 132, 230–242.
Weinstein, L., Adam, J.A. (2008). Guesstimation. Princeton University Press, Princeton, N.J.
Weiss, P., Driesner, T., Heinrich, C.A. (2012). Porphyry-copper ore shells form at stable
pressure-temperature fronts within dynamic fluid plumes. Science, 338, 1613–1616.
Weissbart, E.J., Rimstidt, J.D. (2000). Wollastonite: Incongruent dissolution and leached
layer formation. Geochimica et Cosmochimica Acta, 64, 4007–4016.
Wen, C.Y. (1968). Noncatalytic heterogeneous solid fluid reaction models. Industrial and
Engineering Chemistry, 60, 34–54.
Weng, P.F. (1995). Silica scale inhibition and colloidal silica dispersion for reverse
osmosis systems. Desalination, 103, 59–67.
White, A.F., Peterson, M.L. (1990). Role of reactive-surface area characterization in
geochemical kinetic models. In ACS Symposium Series 416, Chemical Modeling of
Aqueous Systems II, eds Melchior, D.C., Bassett, R.L. American Chemical Society,
Los Angeles, CA, pp. 461–475.
Williamson, M.A., Rimstidt, J.D. (1993). The rate of decomposition of the ferric-
thiosulfate complex in acidic aqueous solutions. Geochimica et Cosmochimica Acta,
57, 3555–3561.
Williamson, M.A., Rimstidt, J.D. (1994). The kinetics and electrochemical rate-
determining step of aqueous pyrite oxidation. Geochimica et Cosmochimica Acta, 58,
5443–5454.
Winfree, A.T. (1972). Spiral waves of chemical activity. Science, 175, 634–636.
Wolff, G.A., Gualtieri, J.G. (1962). PBC vector, critical bond energy ratio and crystal
equilibrium form. American Mineralogist, 47, 562–584.
Wulff, G. (1977). On the question of the rate of growth and dissolution of crystal faces.
In Crystal Form and Structure, ed. Schneer, C.J. Dowden, Hutchinson & Ross, Inc,
Stroudsburg, P.A, pp. 43–52.
Yoneawa, C., Tanaka, Y., Kamioka, H. (1996). Water-rock reactions during gamma-ray
irradiation. Applied Geochemistry, 11, 461–469.
References 227
Zhang, J.-W., Nancollas, G.H. (1990). Mechanisms of growth and dissolution of sparing
soluble salts. In Mineral-Water Interface Geochemistry, eds Hochella, M.F., White,
A.F. Mineralogical Society of America, Washington D.C., pp. 365–396.
Zhang, J.-Z., Millero, F.J. (1993). The products from the oxidation of H2S in seawater.
Zheng, C., Bennett, G.D. (1995). Applied Contaminant Transport Modeling. Van
Nostrand Reinhold, New York.
Zhu, C., Anderson, G. (2002). Environmental Applications of Geochemical Modeling.
Cambridge University Press, Cambridge.
Zhu, C., Lu, P. (2009). Alkali feldspar dissolution and secondary mineral precipitation
in batch systems: 3. Saturation states of product minerals and reaction paths.
Zlokarnik, M. (1991). Dimensional Analysis and Scale-up in Chemical Engineering.
Springer-Verlag, New York.
Index
absorbed dose, 96 box model, 160

acceptor, 92 Bray–Liebhafsky reaction, 207
acid mine drainage, 10 breakthrough, 179
acidity, 16 Briggs–Rauscher reaction, 207
activated complex, 37, 79, 83, 86, 88, 89, 90, Brönsted–Bjerrum equation, 88
92, 100, 101 Brownian motion, 193
activation Brunauer–Emmett–Teller (BET)
energy, 80, 86 method, 111
volume, 90, 91 Buckingham π theorem, 20, 129
activity buoyancy force, 129
coefficient, 88 Burton–Cabrera–Frank (BCF) theory,
ratio, 46 126, 182
adsorbate, 107
advection, 128 calcite, 174, 176
affinity, 46, 123 carbon dioxide, 15, 49
agate, 207 catalyst, 101
aggregation rate, 193 characteristic
alkalinity, 16 length, 130, 158
antilog transformation, 27 time, 42, 53
apparent rate constant, 38 Chauvenet’s criterion, 22
Arrhenius equation, 30, 72, 84, 135, 147 chemical
Avrami equation, 197 bond, 97
classical, 199 potential, 46, 122
modified, 199 reactor, 56
weathering, 6
balanced chemical reactions, 9 chord method, 65, 70
barite, 174, 191 chromatography model, 178
batch reactor, 56 coating
continuous tracer, 63 growth model, 147
rate, 64, 66 thickness, 149
sampled, 74 coefficient of determination, 29
spike tracer, 60 compensation effect, 87
BCF theory, see Burton–Cabrera–Frank complimentary error function, 139
(BCF) theory composite reaction, 37
Belousov–Zhabotinsky reaction, 207 computational error, 4
BET surface area, 111 conductivity, 17
blunders, 21 conservation of mass, 47, 56
Boltzmann constant, 82, 189, 193 continuity equation, 59, 61, 63, 156
boundary layer, 141 critical nucleus, 189, 192
228
Index 229
Damköhler number, 132, 140, 141, 156 forsterite, 105, 114, 116, 121

Darcy velocity, 128, 165 fraction reacted, 40, 42, 115, 116, 120, 153,
Darcy’s law, 129 174, 177, 198
Debye–Hückel limiting law, 88 fraction remaining, 40, 42, 44, 120, 174, 177
degree of saturation, 189 fractional surface coverage, 109, 110
dendrite, 127 Franck–Condon principle, 79
denudation, 6 free energy of activation, 84, 92
detailed balance, 45, 48, 50, 122 free radical, 96
diffusion, 62, 74, 132
diffusion coefficient, 134, 147 geometric surface area, 104
ion pair, 138 Grotthuss mechanism, 134
mutual, 138, 139 Grotthuss–Draper law, 94
tracer, 134 growth units, 127
diffusion-limited aggregation, 193 guess method, 64
dimension, 17 gypsum, 139, 141, 144, 146
dimensional analysis, 17
dimensionless numbers, 18, 20, 119, 129 half-life, 42, 44, 53, 55
discharge, 128 hardness, 17
dispersion, 62, 74 hillock, 123
distribution coefficient, 110, 173 homogeneous
Doerner–Hoskins model, 175 dimensions, 18
donor, 92 units, 18
donor–acceptor theory, 98, 99 Hook’s law, 92
double log method, 72 hot spring, 60, 61
Dufour effect, 206 hydrated electron, 95
dynamic viscosity, 128 hydrogen sulfide, 43
hydrostatic head, 129
effective diameter, 103, 104, 116 hydroxyl free radicals, 96
Eh, 17
Einstein, 94 illite, 30
electron incongruent dissolution, 78
localization function, 98 Ingold mechanism classification, 98
transfer, 92, 96 inhibitor, 101
tunneling, 93 initial rate method, 28, 65
electrophile, 98 International System of Units, 13
elementary reaction, 37, 97, 101 ionic strength, 88, 101
enrichment blanket, 208 iron oxyhydroxide coating, 150
entropy production, 206
error, 21 JMAK equation, 197
error propagation, 24 Johnson–Mehl–Avrami–Kolmogorov
etch pit, 123, 126 (JMAK) equation, 197
extent of reaction, 38, 169
kaolinite, 30
ferric thiosulfate, 83, 86, 89, 101 kinematic viscosity, 128, 147
Fick’s laws, 132, 142 kink, 126
fluorapatite, 73 Kossel crystal model, 123
flux, 37, 102, 124, 144, 148
diffusion, 140 Langmuir
reaction, 140 isotherm, 112
formal error, 4 model, 107
230 Index
leached layer, 126, 127 oriented aggregation, 197

ledge, 126 orthokinetic aggregation, 197
Lewis oscillating chemical reaction, 207
acid–base theory, 98 oxidation–reduction, 92
bond model, 97
Liesegang bands, 207 packed bed, 106
limonite pseudomorph, 153 particle lifetime model, 118
Lineweaver–Burke equation, 109 partition
local equilibrium assumption, 156, 167, 178 coefficient, 173
logarithm, 20 function, 80
logarithmic transformation, 25, 27 pattern formation, 205
longitudinal Péclet number, 131, 156
dispersion coefficient, 131, 159 perikinetic aggregation, 197
dispersivity, 131 periodic bond chain theory, 123
phase transfer
Manning equation, 129 coefficient, 165
Marcus theory, 92 model, 164
mass balance models, 170 photochemical, 93, 96
mass transfer coefficient, 141, 144 photochemistry
diffusion, 141 first law, 94
reaction, 144 second law, 94
Maxwell–Boltzmann distribution, 80 photo-oxidation, 94
mean, 21 photosynthesis, 94
metasomatic infiltration, 178 Planck’s law, 94
metastable persistence, 36 Plank constant, 82
Michaelis–Menton equation, 112 plug flow reactor, 28, 56
microscopic reversibility, 45 continuous tracer, 64
mineral stability, 121 rate, 70
mixed flow reactor, 56 spike tracer, 62
continuous tracer, 63 polymerization rate, 182
non-steady state, 75, 76 polynomial method, 70
rate, 69 porphyry copper deposit, 208
spike tracer, 61 potassium feldspar, 168
mixing, 57, 74 precursor complex, 92
molecular orbital theory, 97 pre-exponential, 72, 84, 87, 190
molecularity, 37, 100 pressure, 90
pX notation, 13
Nernst film model, 142 pyrite, 10, 28, 33, 39, 40, 69, 70, 96, 150, 153
no slip condition, 141
non-equilibrium thermodynamics, 206 Q10 rule, 86
notation, 12 quartz, 53, 96, 121, 124, 157, 165
nucleation, 198 quasi-equilibrium model, 81
rate, 185, 190, 191 quasi-kinetic model, 156
nucleophile, 98 R2, 29
numerical derivative, 32, 65
radiation chemical yield, 96
oligiomer, 127 radiolysis, 95, 96
Onsager reciprocal relations, 206 radium, 174
opposed reaction, 47 random error, 21
Index 231
rate standard error, 22

constant, 36 Stark–Einstein law, 94
equation, 36, 79 steady state, 160
rate-determining step, 100 stoichiometric coefficient, 38, 100
reaction Stokes–Einstein equation, 136
intermediates, 100 structural error, 4
mechanism, 97, 100, 101 successor complex, 92
order, 36, 72, 100 sulfur dioxide, 91
progress variable, 39 surface area/mass, 106
rate, 36 surface area/volume, 106
reaction-limited aggregation, 194 surface
reactive intermediates, 100 catalysis, 111
reconstruction reaction, 126 concentration, 109
regression model, 27 free energy, 102, 186
reorganization energy, 93 reactions, 106
reservoir, 160 retreat rate, 114
residence time, 163 site, 105
retardation factor, 179 surficial velocity, 128
reversible reaction, 52 systematic error, 21
Reynolds number, 130
rounding numbers, 23 TDS, 17
terrace, ledge, and kink (TLK) model,
salinity, 17 126, 182
Schmidt number, 131, 141 tetrathionate, 83
schwertmannite, 204 time constant, 43, 162
scorodite, 66 time–temperature–transformation (TTT)
self-assembly, 127 diagram, 200
self-exchange reaction, 92 TLK model, 126
shrinking core model, 152 tortuosity, 139
shrinking particle model, 114, 118, 198 total dissolved solids, see TDS
SI units, 13 tracer, 179
Sierra Nevada batholith, 171 continuous, 59
significant figures, 23 spike, 59
silica, 53, 99, 138, 157, 183, 195 tracer kinetics, 59
similitude, 19 transition state, 37, 46, 80, 81, 90, 97,
site density, 105 100, 101
slug, 56, 62, 74 transition-state theory, 79
Smoluchowski model, 193 transmission coefficient, 82
solid–solid transformation model, 197 TTT diagram, 200
Soret effect, 206 Turing model, 207
specific discharge, 128
specific surface area unit analysis, 18
BET, 111 unit operation, 205
geometric, 104 units, 13, 17
spinodal, 186 unopposed reaction, 39
spiral hillock, 126 uranium dioxide, 113
stability ratio, 194
staged reactors, 56 vacancy island, 126
standard deviation, 22 valence bond theory, 97
232 Index
validation, 5 weathering, 168, 170

velocity boundary layer, 141 wollastonite, 76
verification, 4, 5, 9 Wulff
vibrational frequency, 81 construction, 122
viscosity, 128 theorem, 103,
volumetric flow rate, 128 122
water cycle, 163 Zeldovich factor, 190

Geochemical Rate Models

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Geochemical Rate Models

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Geochemical Rate Models

Uploaded by

Copyright:

Available Formats

Geochemical Rate Models

An Introduction to Geochemical Kinetics

Rate models (kinetic models) are used to predict rates of geochemical

J. Donald Rimstidt is a Professor Emeritus of Geochemistry at Virginia

“Don Rimstidt brings 30 years of teaching experience at a world class

David J. Vaughan, Research Professor, University of Manchester, and

Virginia Polytechnic Institute and State University

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of

This publication is in copyright. Subject to statutory exception

Library of Congress Cataloguing in Publication data

ISBN 978-1-107-02997-2 Hardback

Additional resources for this publication at www.cambridge.org/rimstidt

Cambridge University Press has no responsibility for the persistence or accuracy of

This book is intended for use as a textbook for an upper-level undergradu-

models. The thermodynamic models describe the equilibrium endpoint for

Kinetics and dynamics

possible range of behaviors of the system of interest. Finally, good models

Table 1.1. Checklist of recommended validation and verification procedures

Are the equations used dimensionally homogeneous?

a particular model depends upon whether it predicts the behavior of the

than an untested one. Finally, collecting new observations to test a model’s

Example 1.1. Estimation method for validation of a chemical

Denudation is the lowering of the elevation of the Earth’s surface by chemical

Balancing chemical reactions

Although balancing equations by inspection is the most convenient method,

Example 2.1. Pyrite oxidation reaction

Acid mine drainage is a common environmental problem associated with disposed

2. Inspection of this unbalanced “equation” shows that an equal number of

or the equation can be multiplied by 4 to get

A more systematic alternative to this approach begins by expressing the reaction

These coefficients produce a balanced equation.

Kinetics variable notation is quite different from discipline to discipline

Table 2.1. Base units and important related units in the SI

Quantity Name Symbol

Table 2.2. Decimal multiples of SI units

Factor Prefix Symbol Factor Prefix Symbol

1024 yotta Y 10–1 deci d

milligrams per liter, mg/L milligrams solute/1 L solution

means that a complex transformation is needed to make molar concentra-

A convention that is based on historical practice is reporting dissolved

Conductivity is a measure of the ability of an aqueous solution to carry

Salinity is a measure of the mass of dissolved salts in the solution. This

a football field to a yardstick gives the ratio of 100:1 allowing us to say

8.314 x10-2 dm 3 bar 1000 cm 3 1 J/bar 8.314 J

A fundamental principle of dimensional analysis is that after all unit conver-

This approach assumes that the behavior of the system is unrelated to

Example 2.2. Surface area to volume ratio of a sphere

Equation (2.8) can be solved to find b.

For cubes, b = 6. Other regular convex solids have intermediate values of b.

The concept of similitude is further formalized by the Buckingham π the-

Adding logarithms is equivalent to multiplying numbers so it is often con-

245 × 63 = (10 2.39 )(101.80 ) = 10 2.39 +1.80 = 10 4.19 = 1.6 × 10 4

Likewise division is accomplished by subtracting powers of ten.

245 63 = 10 2.39 101.80 = 10 2.39 −1.80 = 10 0.59 = 3.9

Base 10 logarithms are so often used in geochemical calculations that it is

log 1 = 0.0 100

The standard deviation of the measurements, σx, is a measure of the amount

The best estimate of the uncertainty in x is the standard error of the

δ q = 0.0171 (0.01) = 0.000171 (2.35)

δ k = (2.303) (6.61 × 10 −5 ) (0.27 ) = 4.11 × 10 −5 (2.36)

δ (log k ) = (0.434 )(0.622 ) = 0.270 (2.37)

mFe3+ = a + bt + ct 2 + ... (2.38)

mFe3+ = 1.73 × 10 −3 − 1.23 × 10 −4 t + 3.80 × 10 −6 t 2 (2.40)

δ log k = (0.18)2 + (992 T )2 = 2.67 (2.45)