Deformation Of Earth Materials An Introduction

To The Rheology Of Solid Earth Shunichiro Karato

download

https://ebookbell.com/product/deformation-of-earth-materials-an-
introduction-to-the-rheology-of-solid-earth-shunichiro-
karato-4408448

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Earth Dynamics Deformations And Oscillations Of The Rotating Earth

Smylie De

https://ebookbell.com/product/earth-dynamics-deformations-and-
oscillations-of-the-rotating-earth-smylie-de-4584288

Deformation Of The Continental Crust The Legacy Of Mike Coward

Geological Society Special Publication No 272 A C Ries R W H Butler

https://ebookbell.com/product/deformation-of-the-continental-crust-
the-legacy-of-mike-coward-geological-society-special-publication-
no-272-a-c-ries-r-w-h-butler-2142134

Testing Of The Plastic Deformation Of Metals T W Clyne J E Campbell

https://ebookbell.com/product/testing-of-the-plastic-deformation-of-
metals-t-w-clyne-j-e-campbell-49138750

Plastic Deformation Of Minerals And Rocks Shunichiro Karato

https://ebookbell.com/product/plastic-deformation-of-minerals-and-
rocks-shunichiro-karato-50924138
Plastic Deformation Of Minerals And Rocks Shunichir Karato

https://ebookbell.com/product/plastic-deformation-of-minerals-and-
rocks-shunichir-karato-2347650

Constrained Deformation Of Materials Devices Heterogeneous Structures

And Thermomechanical Modeling 1st Edition Yl Shen Auth

https://ebookbell.com/product/constrained-deformation-of-materials-
devices-heterogeneous-structures-and-thermomechanical-modeling-1st-
edition-yl-shen-auth-2530346

Continuum Deformation Of Multiagent Systems 1st Edition Hossein

Rastgoftar Auth

https://ebookbell.com/product/continuum-deformation-of-multiagent-
systems-1st-edition-hossein-rastgoftar-auth-5696284

Plastic Deformation Of Nanocrystalline Materials 1st Edition Glezer

https://ebookbell.com/product/plastic-deformation-of-nanocrystalline-
materials-1st-edition-glezer-6750314

The Physics Of The Deformation Of Densely Packed Granular Materials

Maarten Anton Koenders

https://ebookbell.com/product/the-physics-of-the-deformation-of-
densely-packed-granular-materials-maarten-anton-koenders-34725598
This page intentionally left blank
Deformation of Earth Materials

Much of the recent progress in the solid Earth sciences is based

on the interpretation of a range of geophysical and geological
observations in terms of the properties and deformation of
Earth materials. One of the greatest challenges facing geo-
scientists in achieving this lies in finding a link between phys-
ical processes operating in minerals at the smallest length
scales to geodynamic phenomena and geophysical observa-
tions across thousands of kilometers.
This graduate textbook presents a comprehensive and
unified treatment of the materials science of deformation as
applied to solid Earth geophysics and geology. Materials
science and geophysics are integrated to help explain
important recent developments, including the discovery of
detailed structure in the Earth’s interior by high-resolution
seismic imaging, and the discovery of the unexpectedly
large effects of high pressure on material properties, such
as the high solubility of water in some minerals. Starting
from fundamentals such as continuum mechanics and
thermodynamics, the materials science of deformation of Earth
materials is presented in a systematic way that covers elastic,
anelastic, and viscous deformation. Although emphasis is
placed on the fundamental underlying theory, advanced
discussions on current debates are also included to bring read-
ers to the cutting edge of science in this interdisciplinary area.
Deformation of Earth Materials is a textbook for graduate
courses on the rheology and dynamics of the solid Earth, and
will also provide a much-needed reference for geoscientists in
many fields, including geology, geophysics, geochemistry,
materials science, mineralogy, and ceramics. It includes review
questions with solutions, which allow readers to monitor their
understanding of the material presented.

S H U N - I C H I R O K A R A T O is a Professor in the Department of

Geology and Geophysics at Yale University. His research
interests include experimental and theoretical studies of the
physics and chemistry of minerals, and their applications to
geophysical and geological problems. Professor Karato is
a Fellow of the American Geophysical Union and a recipient
of the Alexander von Humboldt Prize (1995), the Japan
Academy Award (1999), and the Vening Meinesz medal
from the Vening Meinesz School of Geodynamics in The
Netherlands (2006). He is the author of more than 160 journal
articles and has written/edited seven other books.
Deformation
of Earth
Materials
An Introduction to the Rheology of Solid Earth

Shun-ichiro Karato
Yale University, Department of Geology &
Geophysics, New Haven, CT, USA
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521844048

© S. Karato 2008

This publication is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.
First published in print format 2008

ISBN-13 978-0-511-39478-2 eBook (NetLibrary)

ISBN-13 978-0-521-84404-8 hardback

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.
Contents

Preface page ix

Part I General background 1

1 Stress and strain 3

1.1 Stress 3
1.2 Deformation, strain 7

2 Thermodynamics 13
2.1 Thermodynamics of reversible processes 13
2.2 Some comments on the thermodynamics of a stressed system 28
2.3 Thermodynamics of irreversible processes 29
2.4 Thermally activated processes 32

3 Phenomenological theory of deformation 34

3.1 Classification of deformation 34
3.2 Some general features of plastic deformation 35
3.3 Constitutive relationships for non-linear rheology 36
3.4 Constitutive relation for transient creep 38
3.5 Linear time-dependent deformation 39

Part II Materials science of deformation 49

4 Elasticity 51
4.1 Introduction 51
4.2 Elastic constants 52
4.3 Isothermal versus adiabatic elastic constants 55
4.4 Experimental techniques 57
4.5 Some general trends in elasticity: Birch’s law 59
4.6 Effects of chemical composition 67
4.7 Elastic constants in several crystal structures 70
4.8 Effects of phase transformations 72

5 Crystalline defects 75
5.1 Defects and plastic deformation: general introduction 75
5.2 Point defects 76
5.3 Dislocations 82
5.4 Grain boundaries 94

6 Experimental techniques for study of plastic deformation 99

6.1 Introduction 99
6.2 Sample preparation and characterization 99 v
vi Contents

6.3 Control of thermochemical environment and its characterization 102

6.4 Generation and measurements of stress and strain 104
6.5 Methods of mechanical tests 108
6.6 Various deformation geometries 112

7 Brittle deformation, brittle–plastic and brittle–ductile transition 114

7.1 Brittle fracture and plastic flow: a general introduction 114
7.2 Brittle fracture 115
7.3 Transitions between different regimes of deformation 118

8 Diffusion and diffusional creep 123

8.1 Fick’s law 123
8.2 Diffusion and point defects 125
8.3 High-diffusivity paths 126
8.4 Self-diffusion, chemical diffusion 127
8.5 Grain-size sensitive creep (diffusional creep, superplasticity) 129

9 Dislocation creep 143

9.1 General experimental observations on dislocation creep 143
9.2 The Orowan equation 145
9.3 Dynamics of dislocation motion 145
9.4 Dislocation multiplication, annihilation 154
9.5 Models for steady-state dislocation creep 157
9.6 Low-temperature plasticity (power-law breakdown) 161
9.7 Deformation of a polycrystalline aggregate by dislocation creep 162
9.8 How to identify the microscopic mechanisms of creep 164
9.9 Summary of dislocation creep models and a deformation mechanism map 164

10 Effects of pressure and water 168

10.1 Introduction 168
10.2 Intrinsic effects of pressure 169
10.3 Effects of water 181

11 Physical mechanisms of seismic wave attenuation 199

11.1 Introduction 199
11.2 Experimental techniques of anelasticity measurements 199
11.3 Solid-state mechanisms of anelasticity 202
11.4 Anelasticity in a partially molten material 210

12 Deformation of multi-phase materials 214

12.1 Introduction 214
12.2 Some simple examples 215
12.3 More general considerations 216
12.4 Percolation 222
12.5 Chemical effects 225
12.6 Deformation of a single-phase polycrystalline material 225
12.7 Experimental observations 225
12.8 Structure and plastic deformation of a partially molten material 227

13 Grain size 232

13.1 Introduction 232
13.2 Grain-boundary migration 233
13.3 Grain growth 236
13.4 Dynamic recrystallization 243
Contents vii

13.5 Effects of phase transformations 249

13.6 Grain size in Earth’s interior 253

14 Lattice-preferred orientation 255

14.1 Introduction 255
14.2 Lattice-preferred orientation: definition, measurement and representation 256
14.3 Mechanisms of lattice-preferred orientation 262
14.4 A fabric diagram 268
14.5 Summary 269

15 Effects of phase transformations 271

15.1 Introduction 271
15.2 Effects of crystal structure and chemical bonding: isomechanical groups 271
15.3 Effects of transformation-induced stress–strain: transformation plasticity 280
15.4 Effects of grain-size reduction 286
15.5 Anomalous rheology associated with a second-order phase transformation 286
15.6 Other effects 287

16 Stability and localization of deformation 288

16.1 Introduction 288
16.2 General principles of instability and localization 289
16.3 Mechanisms of shear instability and localization 293
16.4 Long-term behavior of a shear zone 300
16.5 Localization of deformation in Earth 300

Part III Geological and geophysical applications 303

17 Composition and structure of Earth’s interior 305

17.1 Gross structure of Earth and other terrestrial planets 305
17.2 Physical conditions of Earth’s interior 306
17.3 Composition of Earth and other terrestrial planets 314
17.4 Summary: Earth structure related to rheological properties 322

18 Inference of rheological structure of Earth from time-dependent deformation 323

18.1 Time-dependent deformation and rheology of Earth’s interior 323
18.2 Seismic wave attenuation 324
18.3 Time-dependent deformation caused by a surface load: post-glacial isostatic
crustal rebound 326
18.4 Time-dependent deformation caused by an internal load and its
gravitational signature 331
18.5 Summary 337

19 Inference of rheological structure of Earth from mineral physics 338

19.1 Introduction 338
19.2 General notes on inferring the rheological properties in Earth’s interior
from mineral physics 339
19.3 Strength profile of the crust and the upper mantle 342
19.4 Rheological properties of the deep mantle 358
19.5 Rheological properties of the core 361

20 Heterogeneity of Earth structure and its geodynamic implications 363

20.1 Introduction 363
20.2 High-resolution seismology 364
20.3 Geodynamical interpretation of velocity (and attenuation) tomography 370
viii Contents

21 Seismic anisotropy and its geodynamic implications 391

21.1 Introduction 391
21.2 Some fundamentals of elastic wave propagation in anisotropic media 392
21.3 Seismological methods for detecting anisotropic structures 398
21.4 Major seismological observations 401
21.5 Mineral physics bases of geodynamic interpretation of seismic anisotropy 402
21.6 Geodynamic interpretation of seismic anisotropy 407

References 412
Materials index 452
Subject index 454

The colour plates are between pages 118 and 119.

Preface

Understanding the microscopic physics of deformation materials science of deformation of minerals and
is critical in many branches of solid Earth science. rocks over various time-scales are described in addition
Long-term geological processes such as plate tectonics to the applications of these results to important geo-
and mantle convection involve plastic deformation of logical and geophysical problems. Properties of materi-
Earth materials, and hence understanding the plastic als discussed include elastic, anelastic (viscoelastic),
properties of Earth materials is key to the study of and plastic properties. The emphasis is on an interdis-
these geological processes. Interpretation of seismolog- ciplinary approach, and, consequently, I have included
ical observations such as tomographic images or seis- discussions on some advanced, controversial issues
mic anisotropy requires knowledge of elastic, anelastic where they are highly relevant to Earth science prob-
properties of Earth materials and the processes of plas- lems. They include the role of hydrogen, effects of
tic deformation that cause anisotropic structures. pressure, deformation of two-phase materials, local-
Therefore there is an obvious need for understanding ization of deformation and the link between viscoelas-
a range of deformation-related properties of Earth tic deformation and plastic flow. This book is intended
materials in solid Earth science. However, learning to serve as a textbook for a course at a graduate level in
about deformation-related properties is challenging an Earth science program, but it may also be useful for
because deformation in various geological processes students in materials science as well as researchers
involves a variety of microscopic processes. Owing to in both areas. No previous knowledge of geology/
the presence of multiple deformation mechanisms, geophysics or of materials science is assumed. The
the results obtained under some conditions may not basics of continuum mechanics and thermodynamics
necessarily be applicable to a geological problem that are presented as far as they are relevant to the main
involves deformation under different conditions. There- topics of this book.
fore in order to conduct experimental or theoretical Significant progress has occurred in the study of
research on deformation, one needs to have a broad deformation of Earth materials during the last 30
knowledge of various mechanisms to define conditions years, mainly through experimental studies. Experi-
under which a study is to be conducted. Similarly, mental studies on synthetic samples under well-defined
when one attempts to use results of experimental or chemical conditions and the theoretical interpretation
theoretical studies to understand a geological problem, of these results have played an important role in under-
one needs to evaluate the validity of applying partic- standing the microscopic mechanisms of deformation.
ular results to a given geological problem. However, Important progress has also been made to expand
there was no single book available in which a broad the pressure range over which plastic deformation can
range of the physics of deformation of materials was be investigated, and the first low-strain anelasticity
treated in a systematic manner that would be useful for measurements have been conducted. In addition,
a student (or a scientist) in solid Earth science. The some large-strain deformation experiments have been
motivation of writing this book was to fulfill this need. performed that have provided important new insights
In this book, I have attempted to provide a unified, into the microstructural evolution during deformation.
interdisciplinary treatment of the science of deforma- However, experimental data are always obtained under
tion of Earth with an emphasis on the materials limited conditions and their applications to the Earth
science (microscopic) approach. Fundamentals of the involve large extrapolation. It is critical to understand ix
x Preface

the scaling laws based on the physics and chemistry of and a geodynamic interpretation of anomalies in seis-
deformation of materials in order to properly apply mic wave propagation. Some of the representative
experimental data to Earth. A number of examples of experimental data are summarized in tables.
such scaling laws are discussed in this book. However, the emphasis of this book is on presenting
This book consists of three parts: Part I basic theoretical concepts and consequently references
(Chapters 1–3) provides a general background includ- to the data are not exhaustive. Many problems (with
ing basic continuum mechanics, thermodynamics and solutions) are provided to make sure a reader under-
phenomenological theory of deformation. Most of this stands the content of this book. Some of them are
part, particularly Chapters 1 and 2 contain material advanced and these are shown by an asterisk.
that can be found in many other textbooks. Therefore The content of this book is largely based on lectures
those who are familiar with basic continuum mechan- that I have given at the University of Minnesota and
ics and thermodynamics can skip this part. Part II Yale University as well as at other institutions. I thank
(Chapters 4–16) presents a detailed account of materi- students and my colleagues at these institutions who
als science of time-dependent deformation, including have given me opportunities to improve my under-
elastic, anelastic and plastic deformation with an standing of the subjects discussed in this book through
emphasis on anelastic and plastic deformation. They inspiring questions. Some parts of this book have
include, not only the basics of properties of materials been read/reviewed by A. S. Argon, D. Bercovici,
characterizing deformation (i.e., elasticity and viscos- H. W. Green, S. Hier-Majumder, G. Hirth, I. Jackson,
ity (creep strength)), but also the physical princi- D. L. Kohlstedt, J. Korenaga, R. C. Liebermann,
ples controlling the microstructural developments J.-P. Montagner, M. Nakada, C. J. Spiers, J. A. Tullis
(grain size and lattice-preferred orientation). Part III and J. A. Van Orman. However, they do not always
(Chapters 17–21) provides some applications of the agree with the ideas presented in this book and any
materials science of deformation to important geolog- mistakes are obviously my own. W. Landuyt, Z. Jiang
ical and geophysical problems, including the rheolog- and P. Skemer helped to prepare the figures. I should
ical structure of solid Earth and the interpretation of also thank the editors at Cambridge University Press
the pattern of material circulation in the mantle and for their patience. Last but not least, I thank my family,
core from geophysical observations. Specific topics particularly my wife, Yoko, for her understanding, for-
covered include the lithosphere–asthenosphere struc- bearance and support during the long gestation of this
ture, rheological stratification of Earth’s deep mantle monograph. Thank you all.
Part I
General background
1 Stress and strain

The concept of stress and strain is key to the understanding of deformation. When a force is applied to
a continuum medium, stress is developed inside it. Stress is the force per unit area acting on a given
plane along a certain direction. For a given applied force, the stress developed in a material depends
on the orientation of the plane considered. Stress can be decomposed into hydrostatic stress (pressure)
and deviatoric stress. Plastic deformation (in non-porous materials) occurs due to deviatoric stress.
Deformation is characterized by the deformation gradient tensor, which can be decomposed into
rigid body rotation and strain. Deformation such as simple shear involves both strain and rigid body
rotation and hence is referred to as rotational deformation whereas pure shear or tri-axial compression
involves only strain and has no rigid body rotation and hence is referred to as irrotational deformation.
In rotational deformation, the principal axes of strain rotate with respect to those of stress whereas
they remain parallel in irrotational deformation. Strain can be decomposed into dilatational
(volumetric) strain and shear strain. Plastic deformation (in a non-porous material) causes shear strain
and not dilatational strain. Both stress and strain are second-rank tensors, and can be characterized by
the orientation of the principal axes and the magnitude of the principal stress and strain and both have
three invariants that do not depend on the coordinate system chosen.

Key words stress, strain, deformation gradient, vorticity, principal strain, principal stress, invariants
of stress, invariants of strain, normal stress, shear stress, Mohr’s circle, the Flinn diagram, foliation,
lineation, coaxial deformation, non-coaxial deformation.

force due to an external field such as the gravity

1.1. Stress
field. Therefore the forces that act on this small
1.1.1. Definition of stress block include (1) short-range forces due to the dis-
placement of atoms within this block, (2) long-range
This chapter provides a brief summary of the basic forces such as gravity that act equally on each atom
concept of stress and strain that is relevant to under- and (3) the forces that act on this block through the
standing plastic deformation. For a more comprehen- surface from the neighboring materials. The (small)
sive treatment of stress and strain, the reader may displacements of each atom inside this region cause
consult M ALVERN (1969), M ASE (1970), M EANS (1976). forces to act on surrounding atoms, but by assump-
In any deformed or deforming continuum material tion these forces are short range. Therefore one
there must be a force inside it. Consider a small block can consider them as forces between a pair of atoms
of a deformed material. Forces acting on the material A and B. However, because of Newton’s law of action
can be classified into two categories, i.e., a short-range and counter-action, the forces acting between two
force due to atomic interactions and the long-range atoms are anti-symmetric: fAB ¼ fBA where fAB (BA) 3
4 Deformation of Earth Materials

x3 X
3
x~i ¼ aij xj (1:4)
j¼1
T1
T
where aij is the transformation matrix that satisfies the
orthonormality relation,
T2 x2
X
3
aij ajm ¼ im (1:5)
T3 j¼1
x1
where im is the Kronecker delta (im ¼ 1 for i ¼ m,
FIGURE 1.1 Forces acting on a small pyramid.
im ¼ 0 otherwise). Now in this new coordinate system,
are the force exerted by atom A (B) to B (A). we may write a relation similar to equation (1.2) as,
Consequently these forces caused by atomic displace- X
3
ment within a body must cancel. The long-range force T~i ¼ ~ij n~j : (1:6)
is called a body force, but if one takes this region as j¼1

small, then the magnitude of this body force will Noting that the traction (T) transforms as a vector in
become negligible compared to the surface force (i.e., the same way as the coordinate system, equation (1.4),
the third class of force above). Therefore the net force we have,
acting on the small region must be the forces across
the surface of that region from the neighboring mate- X
3
T~i ¼ aij Tj : (1:7)
rials. To characterize this force, let us consider a small j¼1
piece of block that contains a plane with the area of dS
and whose normal is n (n is the unit vector). Let T be Inserting equation (1.2), the relation (1.7) becomes,
the force (per unit area) acting on the surface dS from
X
3
outside this block (positive when the force is compres- T~i ¼ jk aij nk : (1:8)
sive) and consider the force balance (Fig. 1.1). The j;k¼1
force balance should be attained among the force T
Now using the orthonormality relation (1.5), one has,
as well as the forces T1,2,3 that act on the surface
dS1,2,3 respectively (dS1,2,3 are the projected area of X
3
dS on the plane normal to the x1,2,3 axis). Then the ni ¼ aji n~i : (1:9)
j¼1
force balance relation for the block yields,
X
3 Inserting this relation into equation (1.8) and compar-
T dS ¼ T j dSj : (1:1) ing the result with equation (1.6), one obtains,1
j¼1
X
3
Now using the relation dSj ¼ nj dS, one obtains, ~ij ¼ kl aik ajl : (1:10)
k;l¼1
X
3 X
3
Ti ¼ T ij nj ¼ ij nj (1:2)
j¼1 j¼1
The quantity that follows this transformation law is
referred to as a second rank tensor.
where Ti is the ith component of the force T and ij is
the ith component of the traction Tj, namely the ith
component of force acting on a plane whose normal is 1.1.2. Principal stress, stress invariants
the jth direction ðnij ¼ T ij Þ. This is the definition of
stress. From the balance of torque, one can also show, In any material, there must be a certain orientation of a
plane on which the direction of traction (T) is normal
ij ¼ ji : (1:3) to it. For that direction of n, one can write,
The values of stress thus defined depend on the Ti ¼ ni (1:11)
coordinate system chosen. Let us denote quantities in
a new coordinate system by a tilda, then the new coor-
dinate and the old coordinate system are related to
   
each other by, 1
In the matrix notation, ~ ¼ A    AT where A ¼ aij and AT ¼ aji .
Stress and strain 5

where  is a scalar quantity to be determined. From σ1

equations (1.11) and (1.2),

X
3
σ2
ðij  ij Þnj ¼ 0: (1:12)
j¼1 σn
x1
For this equation to have a non-trivial solution other σ3 θ σ3
than n ¼ 0, one must have, τ
x3
  x2
 ij  ij  ¼ 0 (1:13)
  σ2
where  Xij  is the determinant of a matrix Xij. Writing
equation (1.13) explicitly, one obtains,
 
 11  
 12 13  σ1
 21 22   23  ¼ 3 þ I 2 þ II  þ III ¼ 0

 31 32 33    FIGURE 1.2 Geometry of normal and shear stress on a plane.

(1:14)
1.1.3. Normal stress, shear stress,
with
Mohr’s circle
I ¼ 11 þ 22 þ 33 (1:15a)
Now let us consider the normal and shear stress on a
given plane subjected to an external force (Fig. 1.2).
II ¼ 11 22  11 33  33 22 þ 212 þ 213 þ 223
Let x1 be the axis parallel to the maximum compres-
(1:15b) sional stress 1 and x2 and x3 be the axes perpendicular
to x1. Consider a plane whose normal is at the angle 
III ¼ 11 22 33 þ 212 23 31  11 223 from x3 (positive counterclockwise). Now, we define a
(1:15c)
 22 213  33 212 : new coordinate system whose x01 axis is normal to the
plane, but the x02 axis is the same as the x2 axis. Then
Therefore, there are three solutions to equation (1.14), the transformation matrix is,
1 ; 2 ; 3 ð1 42 43 Þ.These are referred to as the
2 3
principal stresses. The corresponding n is the orienta- cos  0 sin 
tion of principal stress. If the stress tensor is written ½aij  ¼ 4 0 1 0 5 (1:18)
using the coordinate whose orientation coincides with sin  0 cos 
the orientation of principal stress, then,
2 3 and hence,
1 0 0
2 3
½ij  ¼ 4 0 2 0 5: (1:16) 1 þ 3 1  3
þ cos 2 0
1  3
sin 2
0 0 3 6 2 2 2 7
6 7
6 7
½~
ij  ¼ 6 0 2 0 7:
6 7
It is also seen that because equation (1.14) is a scalar 4 1  3 1 þ 3 1  3 5
sin 2 0  cos 2
equation, the values of I, II and III are independ- 2 2 2
ent of the coordinate. These quantities are called the (1:19)
invariants of stress tensor. These quantities play
important roles in the formal theory of plasticity (see
Section 3.3). Equations (1.15a–c) can also be written Problem 1.1
in terms of the principal stress as,
Derive equation (1.19).
I ¼  1 þ  2 þ  3 (1:17a)
Solution
II ¼ 1 2  2 3  3 1 (1:17b)

and The stress tensor (1.16) can be rotated through the

operation of the transformation matrix (1.18) using
III ¼ 1 2 3 : (1:17c) equation (1.10),
6 Deformation of Earth Materials

2 32 32 3
cos  0 sin  1 0 0 cos  0 sin  A = ( 0 , σ1)
6 76 76 7
6 76 76 7 B = ( 0 , σ3)
½~
ij  ¼ 6 0 1 0 76 0 2 0 76 0 1 0 7
4 54 54 5 τ C = ( 0 , (σ1 + σ3) / 2 )
sin  0 cos  0 0 3 sin  0 cos  R = (σ1 − σ3) / 2
2 1 þ 3 1  3 1  3 3
þ cos 2 0 sin 2
6 2 2 2 7 R
6 7
6 7
¼6 0 2 0 7:
6 7 σn
4 5 A C B
1  3 1 þ 3 1  3
sin 2 0  cos 2
2 2 2

Therefore the shear stress  and normal stress n on this FIGURE 1.3 A Mohr circle corresponding to two-dimensional stress
plane are showing the variation of normal, n , and shear stress, , on a plane.

 1  3
~13   ¼ sin 2 (1:20)
2
and Problem 1.2
 1 þ 3  1  3
~33  n ¼  cos 2 (1:21) Show that the second invariant of deviatoric stress
2 2 1h
can be written as II0 ¼ ð1  2 Þ2 þ ð2  3 Þ2 þ
respectively. It follows that the maximum shear stress i 6
is on the two conjugate planes that are inclined by ð3  1 Þ2 :
p=4 with respect to the x1 axis and its absolute mag-
nitude is ð1  3 Þ=2. Similarly, the maximum com-
Solution
pressional stress is on a plane that is normal to the x1
axis and its value is 1. It is customary to use 1  3 as If one uses a coordinate system parallel to the
(differential (or deviatoric)) stress in rock deformation principal axes of stress, from equation (1.15), one
literature, but the shear stress,   ð1  3 Þ=2, is also has II0 ¼ 01 02  01 03  03 02 . Using I0 ¼ 01 þ 02 þ
often used. Eliminating  from equations (1.20) and
03 ¼ 0; one finds I2 ¼ 02 02 02
1 þ 2 þ 3 þ 2ð1 2 þ
0 0
(1.21), one has,
2 3 þ 3 1 Þ ¼ 0. Therefore II0 ¼ 2 ð1 þ 2 þ02
0 0 0 0 1 02 02
3 Þ:
 1 þ 3 2 1
 2 þ n  ¼ ð1  3 Þ2 : (1:22) Now, inserting 01 ¼ 1  13 ð1 þ 2 þ 3 Þ etc., one
2 4 h i
obtains II0 ¼ 16 ð1  2 Þ2 þ ð2  3 Þ2 þð3  1 Þ2 :
Thus, the normal and shear stress on planes with var-
ious orientations can be visualized on a two-dimensional
plane (–n space) as a circle whose center is located Problem 1.3
at ð0; ð1 þ 3 Þ=2Þ and the radius ð1  3 Þ=2
(Fig. 1.3). This is called a Mohr’s circle and plays an Show that when the stress has axial symmetry with
important role in studying the brittle fracture that is respect to the x1 axis (i.e., 2 ¼ 3 ), then n ¼ Pþ
controlled by the stress state (shear–normal stress ratio; ð1  3 Þðcos2   13Þ.
see Section 7.3).
When 1 ¼ 2 ¼ 3 ð¼ PÞ, then the stress is isotro- Solution
pic (hydrostatic). The hydrostatic component of stress
does not cause plastic flow (this is not true for porous From (1.21), one obtains, n ¼ ð1 þ 3 Þ=2 þ
materials, but we do not discuss porous materials ðð1  3 Þ=2Þ cos 2. Now cos 2 ¼ 2 cos2   1 and
here), so it is useful to define deviatoric stress P ¼ 13 ð1 þ 2 þ 3 Þ ¼ 13 ð1 þ 23 Þ ¼ 1  23 ð1  3 Þ.
Therefore n ¼ P þ ð1  3 Þðcos2   13Þ.
0ij  ij  ij P: (1:23)

When we discuss plastic deformation in this book, we

use ij (without prime) to mean deviatoric stress for Equations similar to (1.15)–(1.17) apply to the
simplicity. deviatoric stress.
Stress and strain 7

1.2. Deformation, strain

1.2.1. Definition of strain

Deformation refers to a change in the shape of a mate- dx Q(x + dx)

rial. Since homogeneous displacement of material points
P(x)
does not cause deformation, deformation must be ~ ~
u+ du
related to spatial variation or gradient of displacement. ~
u
Therefore, deformation is characterized by a displace- dX Qo(X + dX)
ment gradient tensor,
Po(X)
@ui
dij  : (1:24)
@xj

where ui is the displacement and xj is the spatial coor-

FIGURE 1.4 Deformation causes the change in relative positions
dinate (after deformation). However, this displacement of material points.
gradient includes the rigid-body rotation that has noth-
ing to do with deformation. In order to focus on defor- Inserting equation (1.30) into (1.26) one finds,
mation, let us consider two adjacent material points
!
P0(X) and Q0(X þ dX), which will be moved to P(x) 1 @ui @uj X 3
@uk @uk
and Q(x þ dx) after deformation (Fig. 1.4). A small "ij ¼ þ  : (1:31)
2 @xj @xi k¼1 @xi @xj
vector connecting P0 and Q0, dX, changes to dx after
deformation. Let us consider how the length of these This definition of strain uses the deformed state as a
two segments changes. The difference in the squares of reference frame and is called the Eulerian strain. One
the length of these small elements is given by, can also define strain using the initial, undeformed
reference state. This is referred to as the Lagrangian
X
3 X
3
ðdxÞ2  ðdXÞ2 ¼ ðdxi Þ2  ðdXi Þ2 strain. For small strain, there is no difference between
i¼1 i¼1 the Eulerian and Lagrangian strain and both are
X3  
@Xk @Xk reduced to2
¼ ij  dxi dxj : (1:25)
i;j;k¼1
@xi @xj  
1 @ui @uj
"ij ¼ þ : (1:32)
Therefore deformation is characterized by a quantity, 2 @xj @xi
!
1 X3
@Xk @Xk
"ij  ij  (1:26) 1.2.2. Meaning of strain tensor
2 k¼1
@xi @xj
The interpretation of strain is easier in this linearized
which is the definition of strain, "ij . With this defini-
form. The displacement gradient can be decomposed
tion, the equation (1.25) can be written as,
into two components,
X
ðdxÞ2  ðdXÞ2  2 "ij dxi dxj : (1:27)    
@ui 1 @ui @uj 1 @ui @uj
i;j ¼ þ þ  : (1:33)
@xj 2 @xj @xi 2 @xj @xi
From the definition of strain, it immediately follows
The first component is a symmetric part,
that the strain is a symmetric tensor, namely,
 
1 @ui @uj
"ij ¼ "ji : (1:28) "ij ¼ þ ¼ "ji (1:34)
2 @xj @xi
Now, from Fig. 1.4, one obtains, which represents the strain (as will be shown later in
this chapter).
dui ¼ dxi  dXi (1:29)

hence 2
Note that in some literature, another definition of shear strain is used in
which "ij ¼ @ ui =@xj þ @ uj =@xi for i 6¼ j and "ii ¼ @ ui =@xi ; e.g., Hobbs
@ ui @Xi
¼ ij  : (1:30) et al. (1976). In such a case, the symbol ij is often used for the non-
@ xj @xj diagonal (i 6¼ j) strain component instead of "ij .
8 Deformation of Earth Materials

Let us first consider the physical meaning of the Obviously, normal strain can be present in defor-
@u
second part, 12ð@@ xuji  @ xji Þ. The second part is an anti- mation without a volume change. For example,
0 1
symmetric tensor, namely, " 0 0
  @
"ij ¼ 0  2 " 1
0 A represents an elongation
1 @ ui @ uj
oij ¼  ¼ oji ðoii ¼ 0Þ: (1:35) 0 0  12 "
2 @ xj @ xi
along the 1-axis and contraction along the 2 and
The displacement of a small vector duj due to the 3 axes without volume change.
operation of this matrix is given by, Now let us consider the off-diagonal components
X
3 of strain tensor. From equation (1.39), it is clear that
uo
d~i ¼ oij duj : (1:36) when all the diagonal components are zero, then all the
j¼1
displacement vectors must be normal to the direction
Since oii ¼ 0, the displacement occurs only to the direc- of the initial vector. Therefore, there is no change in
tions that are normal to the initial orientation. Therefore length due to the off-diagonal component of strain.
the operation of this matrix causes the rotation of mate- Note, also, that since strain is a symmetric tensor,
rial points with the axis that is normal to both ith and jth "ij ¼ "ji , the directions of rotation of two orthogonal
directions with the magnitude (positive clockwise), axes are toward the opposite direction with the same
uo
d~ magnitude (Fig. 1.5). Consequently, the angle of two
i
tan ij ¼  ¼ oji ¼ oij : (1:37) orthogonal axes change from p=2 to (see Problem 1.4),
dui
p
(Again this rotation tensor is defined using the defor-  tan1 2"ij : (1:43)
med state. So it is referred to as the Eulerian rotation 2
tensor.) To represent this, a rotation vector is often Therefore, the off-diagonal components of strain ten-
used that is defined as, sor (i.e., "ij with i 6¼ j) represent the shape change with-
wð¼ ðo1 ; o2 ; o3 ÞÞ  ðo23 ; o31 ; o12 Þ: (1:38) out volume change, namely shear strain.

Thus oi represents a rotation with respect to the ith

axis. The anti-symmetric tensor, oij , is often referred to Problem 1.4*
as a vorticity tensor.
Now we turn to the symmetric part of displacement Derive equation (1.43). (Assume a small strain for
gradient tensor, "ij . The displacement due to the oper- simplicity. The result also works for a finite strain, see
ation of "ij is, M ASE (1970).)

X
3
u"i ¼
d~ "ij duj : (1:39) Solution
j¼1
Let the small angle of rotation of the i axis to the j axis
From equation (1.39), it follows that the length of a
due to the operation of strain tensor be ij (positive
component of vector u0i changes to,
clockwise), then (Fig. 1.5),
u~i ¼ ð1 þ "ii Þu0i : (1:40)
du~j
Therefore the diagonal component of strain tensor tan ij ¼   ij ¼ ð"ji þ oji Þ ¼ "ij þ oij :
dui
represents the change in length, so that this component
of strain, "ii , is called normal strain. Consequently, Similarly, if the rotation of the j axis relative to the i
axis is ji , one obtains,
V
¼ ð1 þ "11 Þð1 þ "22 Þð1 þ "33 Þ  1 þ "11 þ "22 þ "33
V0 du~i
tan ji ¼   ji ¼ ð"ij þ oij Þ ¼ "ij  oij :
(1:41) duj

where V0 is initial volume and V is the final volume and (Note that the rigid-body rotations of the two axes are
the strain is assumed to be small (this assumption can opposite with the same magnitude.) Therefore, the net
be relaxed and the same argument can be applied to a change in the angle between i and j axes is given by
finite strain, see e.g., M ASE (1970)). Thus,
4ij ¼ ij þ ji ¼ 2"ij  tan 4ij :
X
3
4V
"kk ¼ : (1:42) Hence 4ij ¼  tan1 2"ij .
k¼1
V
Stress and strain 9

 
x2′  "11  " "12 "13 

x2  "21 "22  " "32  ¼ "3 þ I" "2 þ II" " þ III"

 "31 "32 "33  " 
¼0
(1:48)

with

I" ¼ "11 þ "22 þ "33 (1:49a)

x1′ II" ¼ "11 "22  "11 "33  "33 "22 þ "212 þ "213 þ "223
(1:49b)

x1 III" ¼ "11 "22 "33 þ 2"12 "23 "31  "11 "223  "22 "213
FIGURE 1.5 Geometry of shear deformation.  "33 "212 : (1:49c)

1.2.3. Principal strain, strain ellipsoid Therefore, there are three solutions of equation (1.48),
"1 ; "2 ; "3 ð"1 4"2 4"3 Þ. These are referred to as the prin-
We have seen two different cases for strain, one in which cipal strain. The corresponding u0 are the orientations
the displacement caused by the strain tensor is normal to of principal strain. If the strain tensor is written using
the original direction of the material line and another the coordinate whose orientation coincides with the
where the displacement is normal to the original mate- orientation of principal strain, then,
rial line. In this section, we will learn that in any material 2 3
and in any geometry of strain, there are three directions "1 0 0
½"ij  ¼ 0 "2 0 5:
4 (1:50)
along which the displacement is normal to the direction
0 0 "3
of original line segment. These are referred to as the
orientation of principal strain, and the magnitude of A strain ellipsoid is a useful way to visualize the
strain along these orientations are called principal strain. geometry of strain. Let us consider a spherical body
One can define the principal strains ð"1 ; "2 ; "3 ; in a space and deform it. The shape of a sphere is
"1 4"2 4"3 Þ in the following way. Recall that the nor- described by,
mal displacement along the direction i,  u~i , along the
vector u is given by, ðu1 Þ2 þ ðu2 Þ2 þ ðu3 Þ2 ¼ 1: (1:51)

X
3 The shape of the sphere will change due to deforma-
u~i ¼ "ij uj : (1:44) tion. Let us choose a coordinate system such that the
j¼1
directions of 1, 2 and 3 axes coincide with the directions
Now, let u be the direction in space along which the of principal strain. Then the length of each axis of the
displacement is parallel to the direction u. Then, original sphere along each direction of the coordinate
system should change to u~i ¼ ð1 þ "ii Þui , and therefore
u~i ¼ "ui (1:45)
the sphere will change to an ellipsoid,
where " is a scalar quantity to be determined. From
u 1 Þ2
ð~ u2 Þ2
ð~ u3 Þ2
ð~
equations (1.44) and (1.45), 2
þ 2
þ ¼ 1: (1:52)
ð1 þ "1 Þ ð1 þ "2 Þ ð1 þ "3 Þ2
X
3
ð"ij  "ij Þuj ¼ 0: (1:46) A three-dimensional ellipsoid defined by this equa-
j¼1
tion is called a strain ellipsoid. For example, if the
For this equation to have a non-trivial solution other shape of grains is initially spherical, then the shape of
than u ¼ 0, one must have, grains after deformation represents the strain ellip-
soid. The strain of a rock specimen can be deter-
j"ij  "ij j ¼ 0 (1:47)
mined by the measurements of the shape of grains
 
where  Xij  is the determinant of a matrix Xij. Writing or some objects whose initial shape is inferred to be
equation (1.47) explicitly, one gets, nearly spherical.
10 Deformation of Earth Materials

k=∞
Problem 1.5*

Consider a simple shear deformation in which the k=1

displacement of material occurs only in one direction
(the displacement vector is given by u ¼ (y, 0, 0)).

a = (ε1 + 1) / (ε2 + 1)
Calculate the strain ellipsoid, and find how the
principal axes of the strain ellipsoid rotate with strain.
Also find the relation between the angle of tilt of the
initially vertical line and the angle of the maximum
elongation direction relative to the horizontal axis.

Solution
1 k=0
For simplicity, let us analyze the geometry in the x–y plane
1 b = (ε2 + 1) / (ε3 + 1)
(normal to the shear plane) where shear occurs. Consider
a circle defined by x2 þ y2 ¼ 1: By deformation, this FIGURE 1.6 The Flinn diagram (after H OBBS et al., 1976).
circle changes to an ellipsoid, ðx þ yÞ2 þ y2 ¼ 1, i.e.,
as  ! 1 and "2 ¼ A1=2
yy  1 changes from 0 at  ¼ 0
2 2 2 to –1 at  ! 1.
x þ 2xy þ ð þ 1Þy ¼ 1: (1)

Now let us find a new coordinate system that is tilted

from the original one by an angle  (positive counter-
1.2.4. The Flinn diagram
clockwise). With this new coordinate system, ðx; yÞ !
ðX; YÞ with The three principal strains define the geometry of the
     strain ellipsoid. Consequently, the shape of the strain
x cos  sin  X
¼ : (2) ellipsoid is completely characterized by two ratios,
y sin  cos  Y
a  ð"1 þ 1Þ=ð"2 þ 1Þ and b  ð"2 þ 1Þ=ð"3 þ 1Þ. A
By inserting this relation into (1), one finds, diagram showing strain geometry on an a–b plane is
called the Flinn diagram (Fig. 1.6) (F LINN , 1962). In
AXX X2 þ AXY XY þ AYY Y2 ¼ 1 (3) this diagram, for points along the horizontal axis,
with k  ða  1Þ=ðb  1Þ ¼ 0, and they correspond to the
0 1 0 1 flattening strain ð"1 ¼ "2 4"3 ða ¼ 1; b41ÞÞ. For points
AXX 1 þ 12  2  12  2 cos 2   sin 2 along the vertical axis, k ¼ 1, and they correspond to
@ AXY A ¼ @ 2ðcos 2  12  sin 2Þ A (4) the extensional strain ð"1 4"2 ¼ "3 ðb ¼ 1; a41ÞÞ. For
AYY 1 þ 12  2 þ 12  2 cos 2 þ  sin 2 points along the central line, k ¼ 1 (a ¼ b, i.e.,
Now, in order to obtain the orientation in which ð"1 þ 1Þ=ð"2 þ 1Þ ¼ ð"2 þ 1Þ=ð"3 þ 1ÞÞ and deforma-
the X–Y directions coincide with the orientations of tion is plane strain (two-dimensional strain where
principal strain, we set AXY ¼ 0, and get tan 2 ¼ "2 ¼ 0), when there is no volume change during defor-
2=: AXX 5AYY and therefore X is the direction of mation (see Problem 1.6).
maximum elongation. Because the change in the angle
(’) of the initially vertical line from the vertical direction
Problem 1.6
is determined by the strain as tan ’ ¼ , we find,
1 pffiffiffiffiffiffiffiffiffiffiffiffiffi Show that the deformation of materials represented by
tan  ¼ ð þ 4 þ  2 Þ the points on the line for k ¼ 1 in the Flinn diagram is
2 (5)
1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : plane strain (two-dimensional strain) if the volume is
¼ ð tan ’ þ 4 þ tan2 ’Þ
2 conserved.
At  ¼ 0,  ¼ p=4. As strain goes to infinity,  ! 1, Solution
i.e., ’ ! p=2, and tan  ! 0 hence  ! 0: the direction
of maximum elongation approaches the direction If the volume is conserved by deformation, then
1=2
of shear. "1 ¼ AXX  1 changes from 0 at  ¼ 0 to 1 ð"1 þ 1Þð"2 þ 1Þð"3 þ 1Þ ¼ 1 (see equation (1.41)).
Stress and strain 11

foliation lineation FIGURE 1.7 Typical cases of (a)

foliation and (b) lineation.

Combined with the relation ð"1 þ 1Þ=ð"2 þ 1Þ ¼ other possible causes for lineation including the pref-
ð"2 þ 1Þ=ð"3 þ 1Þ, we obtain ð"2 þ 1Þ3 ¼ 1 and hence erential growth of minerals (e.g., H OBBS et al., 1976).
"2 ¼ 0. Therefore deformation is plane strain. Consequently, the interpretation of the significance
of these reference frames (foliation/lineation) in natu-
ral rocks is not always unique. In particular, the ques-
tion of growth origin versus deformation origin, and
1.2.5. Foliation, lineation (Fig. 1.7)
the strain ellipsoid versus the shear plane/shear direc-
When the anisotropic microstructure of a rock is tion can be elusive in some cases. Interpretation and
studied, it is critical to define the reference frame of identification of foliation/lineation become more diffi-
the coordinate. Once one identifies a plane of reference cult if the deformation geometry is not constant with
and the reference direction on that plane, then the three time. Consequently, it is important to state clearly how
orthogonal axes (parallel to lineation (X direction), one defines foliation/lineation in the structural analysis
normal to lineation on the foliation plane (Y direction), of a deformed rock. For more details on foliation and
normal to foliation (Z direction)) define the reference lineation, a reader is referred to a structural geology
frame. textbook such as H OBBS et al. (1976).
Foliation is usually used to define a reference plane
and lineation is used define a reference direction on the 1.2.6. Various deformation geometries
foliation plane. Foliation is a planar feature in a given
rock, but its origin can be various (H OBBS et al., 1976). The geometry of strain is completely characterized by the
The foliation plane may be defined by a plane normal principal strain, and therefore a diagram such as the Flinn
to the maximum shortening strain (Fig. 1.7). Foliation diagram (Fig. 1.6) can be used to define strain. However,
can also be caused by compositional layering, grain-size in order to characterize the geometry of deformation
variation and the orientation of platy minerals such as completely, it is necessary to characterize the deformation
mica. When deformation is heterogeneous, such as the gradient tensor ðdij ð¼ "ij þ oij ÞÞ. Therefore the rota-
case for S-C mylonite (L ISTER and S NOKE , 1984), one tional component (vorticity tensor), oij, must also be
can identify two planar structures, one corresponds to characterized. In this connection, it is important to dis-
the strain ellipsoid (a plane normal to maximum short- tinguish between irrotational and rotational deformation
ening, "3 ) and another to the shear plane. geometry. Rotational deformation geometry refers to
Lineation is a linear feature that occurs repetitively deformation in which oij ¼ 6 0, and irrotational deforma-
in a rock. In most cases, the lineation is found on the tion geometry corresponds to oij ¼ 0. The distinction
foliation plane, although there are some exceptions. between them is important at finite strain. To illustrate
The most common is mineral lineation, which is defined this point, let us consider two-dimensional deformation
by the alignment of non-spherical minerals such as (Fig. 1.8). For irrotational deformation, the orientations
clay minerals. The alignment of spinel grains in a spinel of the principal axes of strain are always parallel to those
lherzolite and recrystallized orthopyroxene in a garnet of principal stress. Therefore such a deformation is called
lherzolite are often used to define the lineation in peri- coaxial deformation. In contrast, when deformation is
dotites. One cause of lineation is strain, and in this case, rotational, such as simple shear, the orientations of
the direction of lineation is parallel to the maximum principal axes of strain rotate progressively with respect
elongation direction. However, there are a number of to those of the stress (see Problem 1.5). This type of
12 Deformation of Earth Materials

irrotational deformation et al., 1983; S IMPSON and S CHMID , 1983). In most

of them, the nature of anisotropic microstructures,
such as lattice-preferred orientation (Chapter 14), is
used to infer the rotational component of deformation.
However, the physical basis for inferring the rotational
component is not always well established.
Some details of deformation geometries in typical
experimental studies are discussed in Chapter 6.

rotational deformation 1.2.7. Macroscopic, and microscopic stress

and strain

Stress and strain in a material can be heterogeneous.

Let us consider a material to which a macroscopically
homogeneous stress (strain) is applied. At any point in
FIGURE 1.8 Irrotational and rotational deformation. a material, one can define a microscopic, local stress
(strain). The magnitude and orientation of microscopic
deformation is called non-coaxial deformation. (When stress (strain) can be different from that of a macro-
deformation is infinitesimal, this distinction is not impor- scopic (imposed) stress (strain). This is caused by the
tant: the principal axes of instantaneous strain are always heterogeneity of a material such as the grain-to-grain
parallel to the principal axis of stress as far as the property heterogeneity and the presence of defects. In particu-
of the material is isotropic.) lar, the grain-scale heterogeneity in stress (strain) is
Various methods of identifying the rotational com- critical to the understanding of deformation of a poly-
ponent of deformation have been proposed (B OUCHEZ crystalline material (see Chapters 12 and 14).
2 Thermodynamics

The nature of the deformation of materials depends on the physical and chemical state of the materials.
Thermodynamics provides a rigorous way by which the physical and chemical state of materials can
be characterized. A brief account is made of the concepts of thermodynamics of reversible as well
as irreversible processes that are needed to understand the plastic deformation of materials and
related processes. The principles governing the chemical equilibrium are outlined including the
concept of chemical potential, the law of mass action, and the Clapeyron slope (i.e., the slope of a
phase boundary in the pressure-temperature space). When a system is out of equilibrium, a flow of
materials and/or energy occurs. The principles governing the irreversible processes are outlined.
Irreversible processes often occur through thermally activated processes. The basic concepts of
thermally activated processes are summarized based on the statistical physics.

Key words entropy, chemical potential, Gibbs free energy, fugacity, activity, Clapeyron slope,
phase diagrams, rate theory, generalized force, the Onsager reciprocal relation.

2.1. Thermodynamics of reversible 2.1.1. The first and the second principles
processes of thermodynamics

Thermodynamics provides a framework by which the The first principle of thermodynamics is the law of conser-
nature of thermochemical equilibrium is defined, and, vation of energy, which states that the change in the inter-
in cases where a system is out of equilibrium, it defines nal energy, dE, is the sum of the mechanical work done to
the direction to which a given material will change. It the system, the change in the energy due to the addition of
gives a basis for analyzing the composition and struc- materials and the heat added to the system, namely,
ture of geological materials, experimental data and the
dE ¼ W þ Z þ Q (2:1)
way in which the experimental results should be
extrapolated to Earth’s interior where necessary. This where W ¼ P dV (the symbol  is used to indicate a
chapter provides a succinct review of some of the change in some quantity that depends on the path) is
important concepts in thermodynamics that play sig- the mechanical work done to the system where P is the
nificant roles in understanding the deformation of pressure, dV is the volume change, Z is the change in
materials in Earth’s interior. More complete discus- internal energy due to the change in the number of
sions on thermodynamics can be found in the text- atomic species, i.e.,
books such as C ALLEN (1960), DE G ROOT and M AZUR X @E 
(1962), L ANDAU and L IFSHITZ (1964) and P RIGOGINE Z ¼ dni (2:2)
i
@ni S;V;nj
and D EFAY (1950).
13
14 Deformation of Earth Materials

where ni is the molar amount of the ith species and Q is F ¼ E  TS (2:10b)

the change in ‘‘heat.’’ Thus
and
X @E 
dE ¼ PdV þ dni þ Q: (2:3) G ¼ E  TS þ PV (2:10c)
i
@ni S;V;nj
respectively and therefore,
Note that ‘‘heat’’ is the change in energy other than
X@U
the mechanical work and energy caused by the dH ¼ T dS þ V dP þ dni  Q0
exchange of material. These two quantities (mechan- i
@ni S;V;nj
ical work and the energy associated with the transport (2:11a)
of matter) are related to the average motion of atoms.
In contrast, the third term, Q, is related to the proper- X@U
dF ¼ S dT  P dV þ dni  Q0
ties of materials that involve random motion or the @ni S;V;nj
i
random arrangement of atoms. The second principle of
(2:11b)
thermodynamics is concerned with the nature of pro-
cesses related to this third term. This principle states and
that there exists a quantity called entropy that is deter- X@U
mined by the amount of heat introduced to the system dG ¼ S dT þ V dP þ dni  Q0 :
i
@n i S;V;nj
divided by temperature, namely,
(2:11c)
Q
dS ¼ (2:4) It follows from (2.8), (2.11a)–(2.11c) that for a
T
closed system and for constant S and V (S and P, T
and that the entropy increases during any natural pro-
and V, T and P), dE ¼ Q0 ðdH ¼ Q0 ; dF ¼ Q0 ;
cesses. When the process is reversible (i.e., the system is
dG ¼ Q0 Þ so that E (H, F, G) is minimum at equili-
in equilibrium), the entropy will be the maximum, i.e.,
brium. Also from (2.8), (2.11a)–(2.11c), one obtains
dS ¼ 0 (2:5)      
@E @H @F
¼ ¼
whereas @ni S;V;nj @ni S;P;nj @ni T;V;nj
 
dS40 (2:6) @G
¼  i : (2:12)
@ni T;P;nj
for irreversible processes. Equation (2.6) may be writ-
ten as This is the definition of the chemical potential. Thus at
thermochemical equilibrium,
Q Q0
dS ¼ de S þ di S ¼ þ (2:7) X
T T dE ¼ T dS  P dV þ i dni (2:13a)
i
where de S ¼ Q=T is the entropy coming from the
exterior of the system and di S ¼ Q0 =T is the entropy X
production inside the system. For reversible processes dH ¼ T dS þ V dP þ i dni (2:13b)
i
Q0 ¼ 0 and for irreversible processes, Q0 40. From
(2.3) and (2.7), one finds, X
dF ¼ S dT  P dV þ i dni (2:13c)
X @E  i
dE ¼ T dS  P dV þ dni  Q0 : (2:8)
i
@ni S;V;ni X
dG ¼ S dT þ V dP þ i dni : (2:13d)
For equilibrium, i

X @E  From (2.13), one has

dE ¼ T dS  P dV þ dni (2:9)
i
@ni S;V;ni    
@E @H
and E ¼ EðS; V; ni Þ. T¼ ¼ (2:14a)
@S V;ni @S P;ni
The enthalpy (H), Helmholtz free energy (F ), and
the Gibbs free energy (G) can be defined as,    
@F @G
S¼ ¼ (2:14b)
H ¼ E þ PV (2:10a) @T V;ni @T P;ni
Thermodynamics 15

   
@E @F If we consider a process in which the quantity Z is kept
P¼ ¼ (2:14c)
@V S;ni @V T;ni constant, then, dZ ¼ 0 and
     
    @Z @Y @Z
@H @G ¼ : (2:17)
V¼ ¼ : (2:14d) @X Y;ni @X Z;ni @Y X;ni
@P S;ni @P T;ni

It can be seen that the thermodynamic quantities such Examples of such a relation include
as T, P, S and V (and i) can be derived from E, H, F      
@S @P @S
and G. Therefore these quantities (E, H, F and G) ¼ (2:18a)
@T P;ni @T S;ni @P T;ni
are called the thermodynamic potentials. The thermo-
dynamic potentials assume the minimum value at      
@T @S @T
thermochemical equilibrium. Because we will mostly ¼ (2:18b)
@V S;ni @V T;ni @S V;ni
consider a system at constant temperature and pres-
sure, the most frequently used thermodynamic and
potential is the Gibbs free energy. i is the thermo-
     
dynamic potential of the ith species (per unit mole). To @V @V @P
¼ : (2:18c)
emphasize the fact that i is the thermodynamic poten- @T P;ni @P T;ni @T V;ni
tial of the ith species per mole, it is often called the
partial molar thermodynamic potential (partial molar These thermodynamic identities (the Maxwell relations
Gibbs free energy when the independent variables are T and the relations (2.18)) are often used in manipulating
and P). thermodynamic relationships (e.g., Chapter 4).
Using the rule of calculus, it follows from (2.13) Now let us rewrite (2.13d) as,
and (2.14), X i 
1 P
 2   2      dS ¼ dE þ dV þ  dni : (2:19)
@ E @ E @T @P T T i
T
¼ ) ¼
@S @V ni @V @S ni @V S;ni @S V;ni
At equilibrium, the entropy is a maximum, i.e., dS ¼ 0.
(2:15a)
Consider a case where two systems (1 and 2) are in
   2      contact. In this case the condition for equilibrium can
@2 H @ H @T @V
¼ ) ¼ be written as
@S @P ni @P @S ni @P S;ni @S V;ni
(2:15b) 1 1 P1 P2
dS ¼ dE1 þ dE2 þ dV1 þ dV2
T1 T2 T1 T2
   2      X      (2:20)
@2 F @ F @S @P i i
¼ ) ¼ þ  1 dni1 þ  2 dni2 ¼ 0:
@T @V ni @V @T ni @V T;ni @T V;ni i
T1 T2

(2:15c) Because we consider a system with a constant energy

and (dE1 þ dE2 ¼ 0), volume (dV1 þ dV2 ¼ 0) and matter
 2   2      ðdni1 þ dni2 ¼ 0Þ, (2.20) becomes,
@ G @ G @S @V    
¼ ) ¼ : 1 1 P1 P2
@T @P ni @P @T ni @P T;ni @T P;ni dS ¼  dE1 þ  dV1
T1 T2 T1 T2
(2:15d)   (2:21)
X i i
These relations (2.15) are called the Maxwell relations. þ  1 þ 2 dni1 ¼ 0:
i
T1 T2
Similar relations among thermodynamic variables
can also be derived. Consider a quantity such as This must occur for any arbitrary changes of internal
entropy that is a function of two parameters (such as energy (dE1), volume (dV1) and chemical composition
temperature and pressure; this is a case for a closed ðdni1 Þ. Therefore when two systems (1 and 2) are in con-
system, i.e., ni is kept constant), i.e., Z ¼ Z(X, Y; ni), tact and in equilibrium, 1=T1 ¼ 1=T2 ; P1 =T1 ¼ P2 =T2
then, and i1 =T1 ¼ i2 =T2 and hence the conditions of equi-
    librium are
@Z @Z
dZ ¼ dX þ dY: (2:16)
@X Y;ni @Y X;ni T1 ¼ T2 (2:22a)
16 Deformation of Earth Materials

P1 ¼ P2 (2:22b) 1.0

and

i1 ¼ i2 : (2:22c) 0.8

The variables such as temperature, pressure and the

concentration of ith species do not depend on the size 0.6

Sconfig /R
of the system. These variables are called intensive quan-
tities. In contrast, quantities such as entropy, internal
energy and Gibbs free energy increase linearly with the 0.4
size of the system. They are called extensive quantities.
It follows that,
0.2
SðlE; lV; lni Þ ¼ lSðE; V; ni Þ (2:23)

where l is an arbitrary parameter. Differentiating

(2.23) with l, and putting l ¼ 1, one obtains, 0.0
0.0 0.2 0.4 0.6 0.8 1.0
X X
TS ¼ E þ PV  i ni : (2:24)
i FIGURE 2.1 A plot of configurational entropy
Sconfig ðxÞ ¼ R½ x log x þ ð1  xÞ logð1  xÞ.
Differentiating this equation, and comparing the
results with equation (2.19), one finds,
X N!
S dT  V dP þ ni di ¼ 0: (2:25) S ¼ kB  log
ðN  nÞ!n!
i (2:27)
ffi RNmol ½x log x þ ð1  xÞ logð1  xÞ
This is the Gibbs–Duhem relation, which shows that the
intensive variables are not all independent. where x ¼ n=N and N ¼ Na Nmol (Na is the Avogadro
The concept of entropy is closely related to the number, and Nmol is the molar abundance of the rele-
atomistic nature of matter, namely the fact that vant species) where the Stirling formula, N!  N
matter is made of a large number of atoms. A system log N  N for N  1 was used. The entropy correspond-
composed of a large number of atoms may assume a ing to this case may be called configurational entropy
large number of possible micro-states. All micro-states Sconfig and is plotted as a function of concentration x in
with the same macro-state (temperature, volume etc.) Fig. 2.1. The configurational entropy is proportional
are equally probable. Consequently, a system most to the amount of material, and for unit mole of mate-
likely assumes a macro-state for which the number of rial, it is given by
corresponding micro-states is the maximum (i.e., the
maximum entropy). Thus the concept of entropy must Sconfig ¼ R½x log x þ ð1  xÞ logð1  xÞ: (2:28)
be closely related to the number of the micro-state, W, The micro-state of matter may also be characterized
as (for the derivation of this relation see e.g., L ANDAU by the nature of lattice vibration; that is, matter with
and L IFSHITZ (1964)), different frequencies of lattice vibration is considered
S ¼ kB  log W (2:26) to be in different states. The vibrational entropy
defined by this is related to the frequencies of atomic
where kB is the Boltzmann constant.1 The number of vibration as (e.g., A NDERSON , 1996; B ORN and H UANG ,
micro-states may be defined by the number of ways in 1954, see Box 2.1),
which atoms can be distributed. When n atoms are X  hoi 
distributed on N sites, then, W ¼ N Cn ¼ N!=ðN  nÞ!n!, Svib kB log (2:29)
i
2pkB T
and,
where h is the Planck constant, kB is the Boltzmann
constant, oi is the (angular) frequency of lattice vibra-
1
When log is used in a theoretical equation in this book, the base is e (this is
tion of mode i (for a crystal that contains N atoms in
often written as ln). In contrast, when experimental data are plotted, the the unit cell, there are 3N modes of lattice vibration). It
base is 10 (unless specified otherwise). can be seen that a system with a higher frequency of
Thermodynamics 17

lattice sites (lattice vibration) or by large displacements

Box 2.1 Lattice vibration and the that result in an exchange of atoms among various
vibrational entropy sites. Therefore the entropy may be written as,
Small random motion of atoms around their S ¼ Svib þ Sconfig : (2:31)
stable positions causes ‘‘disorder’’ in a material
that contributes to the entropy. To calculate the
contribution to entropy from lattice vibration, we 2.1.2. Activity, fugacity
note that the internal energy due to lattice vibration
is given by (e.g., B ORN and H UANG , 1954) Using the equations (2.10), (2.12) and (2.28), we can
write the chemical potential of a component as a func-
P
1
tion of the concentration x (for x 1),
ðni hoi =2pÞ expðni hoi =2pkB TÞ
X
3N
ni ¼0
E¼ P1 ðT; P; xÞ ¼ 0 ðT; PÞ þ RT log x (2:32)
i¼1 expðni hoi =2pkB TÞ
ni ¼0
where 0 is the chemical potential for a pure phase
X
3N
hoi =2p
¼ (x ¼ 1). In a system that contains several components,
i¼1
expðhoi =2pkB TÞ  1 (2.32) can be generalized to,
where ni is the number of phonons of the ith i ðT; P; xi Þ ¼ 0i ðT; PÞ þ RT log xi (2:33)
mode of lattice vibration and oi is its (angular)
frequency. Using the thermodynamic relation where the suffix i indicates a quantity for the ith
R component.
F ¼ T E=T2 dT, one obtains
X  
hoi

F ¼ U þ kB T log 1  exp 
i
2pkB T Problem 2.1

where U is the energy of a static lattice (at T ¼ 0 K). Derive equation (2.32).
Therefore from S ¼ ð@F=@TÞ,
(
X  
hoi
 Solution
Svib ¼ kB log 1  exp 
i
2pkB T From (2.10) and (2.12), noting that E, V and S are the
) extensive variables, one obtains,
X hoi 1
       
2pkB T expðhoi =2pkB TÞ  1 @G @E @V @S
i ¼ ¼ þP T
@nmol
T;P @nmol T;P @nmol T;P @nmol T;P
X hoi  
kB log : @Sconfig
2pkB T ¼ e þ P  Tsvib  T
i @nmol
 
0 @Sconfig
The approximation is for high temperature, i.e.,  T :
@nmol
hoi =2pkB T 1.
where 0  e þ P  Tsvib and e,  and svib are molar
internal energy, molar volume and molar (vibrational)
vibration has a lower entropy. When the vibrational entropy respectively.
frequency changes between two phases (A and B), then Now, noting that dx ¼ dnmol =Nmol ðfrom
the change in entropy is given by, x ¼ n=N ¼ nmol =Nmol Þ, it follows from (2.28),

X @Sconfig =@nmol T;P ¼ R logðx=ð1  xÞÞ R log x.
oBi oBD
Svib  SA B
vib  Svib ¼ kB log R log : Therefore one obtains ðT; P; xÞ ¼ 0 ðT; PÞþ
i
oA
i oAD
RT log x.
(2:30)
where oA;B
D is a characteristic frequency of lattice vibra-
tion (the Debye frequency; see Box 4.3 in Chapter 4) of
a phase A or B. Activity
In a solid, the micro-state may be defined either by In deriving (2.32), we made an assumption that the
small displacements of atomic positions from their component under consideration has a small quantity
18 Deformation of Earth Materials

(dilute solution) so that atoms in the component do not increases logarithmically with pressure. For a non-
interact with each other or with other species. Such a ideal gas, one can assume a similar relation, i.e.,
material is called an ideal solution. In a real material
fðP; TÞ
where the interaction of atoms of a given component is ðP; TÞ ¼  ðP0 ; TÞ þ RT log (2:41)
P0
not negligible, a modification of these relations is
needed. A useful way to do this is to introduce the where  ðT; P0 Þ is identical to the ideal gas. This is the
concept of activity (of the ith component), ai, which is definition of fugacity, f. The fugacity coefficient, , is
defined by, often used to characterize the deviation from ideal gas,

i ðT; P; ai Þ ¼ 0i ðT; PÞ þ RT log ai : (2:34) f  P: (2:42)

If 0i ðT; PÞ is the chemical potential of a pure phase, then Obviously, f ! P ( ! 1) as P ! 0.

by definition, for a pure system, the activity is 1 (for The fugacity of a given fluid can be calculated from
example, if pure Ni is present in a system, then the activity the equation of state. Let us integrate @=@P ¼  ( is
of Ni is aNi ¼ 1). Now we can relate (2.34) to (2.33) by the molar volume) to obtain
introducing the activity coefficient,  i, defined by, Z P
ðP; T Þ ¼ ðP0 ; T Þ þ ð; TÞ d: (2:43)
ai  i xi (2:35) P0

to get Now for an ideal gas,

Z P
i ðT; P; xi Þ ¼ 0i ðT; PÞ þ RT log i xi : (2:36) id ðP; T Þ ¼ id ðP0 ; T Þ þ id ð; T Þ d: (2:44)
P0
The activity coefficient can be either  i > 1 or  i < 1.
Subtracting (2.44) from (2.43), one has,
Fugacity ðP; TÞ  id ðP; T Þ ¼ ðP0 ; T Þ  id ðP0 ; T Þ
Z P
For an ideal gas, the (molar) internal energy (e) is a 
function only of temperature (Joule’s law), i.e., þ ð; T Þ  id ð; TÞ d:
P0
e ¼ e(T ). And the enthalpy is h ¼ e þ P. Therefore (2:45)
using the equation of state (P ¼ RT), one finds that
enthalpy is also a function only of temperature, Noting that any fluid becomes an ideal gas at zero

namely, h ¼ h(T ). To get an equation for (molar) pressure, i.e., Plim ðP0 ; T Þ  id ðP0 ; T Þ ¼ 0 and
0!0
entropy, recall the relation (2.19) for a closed system,
from (2.40) and (2.41), ðP; T Þ  id ðP; T Þ ¼
1 P RT logðf ðP; TÞ=PÞ. Therefore one obtains
ds ¼ de þ d: (2:37)
T T Z P
fðP; T Þ 1 lim
Using the definition of specific heat (cv), this equation log ¼ ð; T Þ  id ð; T Þ d:
P RT P0!0 P0
is translated into,
(2:46)
1 de P cv P
ds ¼ dT þ d ¼ dT þ d: (2:38) The fugacity of a gaseous species at any T and P can be
T dT T T T calculated from the equation of state (i.e.,  ¼  (P,T))
Integrating this equation, one obtains, using equation (2.46).
Z Non-ideal gas behavior occurs when the mutual
T
cv  distance of molecules becomes comparable to the
sðT; Þ ¼ sðT0 ; 0 Þ þ dT þ R log : (2:39)
T0 T 0 molecular size, lm. The mean distance of molecules in
Inserting the relation P ¼ RT, and using  ¼ h  Ts, a fluid is given by l ¼ ð=NA Þ1=3 ¼ ðRT=PNA Þ1=3 where
one has,  is the molar volume. When l=lm  1, then a gas
behaves like an ideal gas, whereas when l=lm  1, it
P
id ðP; TÞ ¼  ðP0 ; TÞ þ RT log (2:40) becomes a non-ideal gas. For water, lm  0.3 nm and
P0
l=lm  1 at a pressure of 0.5 GPa (at 1673 K),
RT
where  ðP0 ; TÞ ¼ hðTÞ  TsðP0 ; T0 ÞT T0 ðcv =TÞdT. whereas for hydrogen, lm  0:1 nm and one needs
This equation indicates that the chemical potential 15 GPa to see non-ideal behavior (at 1673 K)
(partial molar Gibbs free energy) of an ideal gas (Fig. 2.2b).
Thermodynamics 19

(a) FIGURE 2.2 (a) Fugacity of water

1018 and carbon dioxide as a function of
temperature (T) and pressure (P).
1016 Note the large deviation from ideal
gas behavior at higher pressures
1014 (data for the equation of state are
from F ROST and W OOD (1997a)).
(b) A comparison of the fugacity
1012
of water (thin curve) with ideal gas
behavior (thick curve). Significant
1010 2000
1600 1800
Fugacity (GPa)

deviation from the ideal gas

1400 behavior is seen when the mean
T = 1000 K 1200
108 distance of water molecules, l,
is close to lm (where lm is the
106 molecular size).
1600 1800 2000
CO2 1000 1200 1400
104

102
H2O
100

10–2

10–4
0 2 4 6 8 10 12 14
Pressure (GPa)

(b) 20

H2O

15 3

1573 K
Fugacity (GPa)

10 2 l/lm
ity
ac
fug

l/lm
5 1

s)
(ideal ga
fugacity

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Pressure (GPa)
20 Deformation of Earth Materials

When a fluid behaves like an ideal gas whose equa- TABLE 2.1 Equation of state parameters, a(T), b(T), c(T) and d(T)
tion of state is P ¼ RT, then its fugacity defined by for water and carbon dioxide.
equation (2.41) is equal to its (partial) pressure. All parameters are assumed to be parabolic function of
However, as fluids are compressed, their resistance temperature: m ¼ m0 þ m1 T þ m2 T 2 (where m is a, b, c or d).
to compression increases and the molar volume Units listed in the table are for m0. The unit for m1 is
does not change with pressure as much as an equation ½m1  ¼ ½m0 =T, and for m2 is ½m2  ¼ ½m0 =T 2 . Units:
of state of an ideal gas would imply. If the molar a (m6 Pa K1/2 mol1), b (m3), c (m3 Pa1/2), d (m3 Pa1).
volume does not change with pressure, for example,
then the fugacity will be an exponential function of CO2 H2O
pressure, a a0 ¼ 5.373 a0 ¼ 5.395 10
 
ðTÞ  P a1¼ 5.6829 103 a1¼ 6.362 102
fðP; T Þ P0 exp : (2:47) a2¼ 4.045 106 a2¼ 2.368 105
RT
b b0¼ 4.288 105 b0¼ 2.7732 105
Important examples are water and carbon dioxide. – b1¼ 2.0179 108
The fugacities of water and carbon dioxide can be – b2 ¼ 9.2125 1012
calculated from the equations of state (Fig. 2.2). c c0 ¼ 7.526 1010 c0 ¼ 3.934 1010
Water behaves like a nearly ideal gas up to 0.3 GPa c1 ¼ 1.1440 1013 c1 ¼ 5.66 1013
(at T 41000 K), but its property starts to deviate – c2 ¼ 2.485 1016
from ideal gas behavior above 0.5 GPa. At
d d0 ¼ 3.707 1015 d0 ¼ 2.186 1015
P ¼ 2 GPa ðT ¼ 1500 KÞ, for example, the fugacity d1 ¼ 1.198 1020 d1 ¼ 3.6836 1018
of water is 13 GPa and at P ¼ 3 GPa ðT ¼ 1500 KÞ, d2 ¼ 1.0464 1022 d2 ¼ 1.6127 1021
it is 55 GPa. The large fugacity of water under high
pressures means that water is chemically highly reac-
tive under deep Earth conditions. The behavior of
carbon dioxide is similar. When extrapolating labo-
Solution
ratory data involving these fluids obtained at low Using equation (2.46), one obtains
pressures to higher pressures, one must take into
account the non-ideal gas behavior of these fluids Z P pffiffiffiffi 
f 1 aR T pffiffiffi
(see Chapter 10). log ¼  þbþc  þ d d
P RT 0 ðRTþbÞðRTþ2bÞ
Z P  
1 a 1 1
¼ pffiffiffiffi 
Problem 2.2
RT 0 T ðRTþbÞ ðRT=2þbÞ
pffiffiffi o
þbþc  þ d d
The equations of state of water and carbon dioxide are
approximately given by the following formula (F ROST
and performing elementary integration and remember-
and W OOD , 1997b),
ing that the parameters a, b, c and d are functions of
pffiffiffiffi temperature, T, one obtains
RT aðTÞR T
ðP; TÞ ¼  þ bðTÞ
P ðRT
pffiffiffiffi þ bðTÞPÞðRT þ 2bðTÞPÞ fðP; TÞ aðTÞ RTþbðTÞP bðTÞP
þ cðTÞ P þ dðTÞP: log ¼ pffiffiffiffi log þ
P bðTÞRT T RTþ2bðTÞP RT
pffiffiffiffi 2
Where parameters (a, b, c and d) are functions of 2 cðTÞP P dðTÞP
þ þ :
temperature, but not of pressure (see Table 2.1). 3 RT 2RT
Show that the fugacity of these fluids is given by
Note that these gases behave like an ideal gas (i.e.,
fðP; TÞ aðTÞ RT þ bðTÞP bðTÞP f ! P) as P ! 0 as they should. At intermediate
log ¼ pffiffiffiffi log þ
P bðTÞRT T RT þ 2bðTÞP RT pressures (P  5–20 GPa for water or carbon dioxide),
pffiffiffiffi 2
2 cðTÞP P dðTÞP the third term (bðTÞP=RT) dominates and f=P
þ þ
3 RT 2RT expðbðTÞP=RTÞ whereas at extreme pressures (i.e.,

and using the parameter values shown below calculate P ! 1), f=P exp d ðTÞP2 =2RT . The results of the
the fugacities of water and carbon dioxide for the con- fugacity calculation are shown in Fig. 2.2.
ditions 0 5 P 5 20 GPa and 1000 5 T 52000 K.
Thermodynamics 21

Problem 2.3 x1 2

A1 xA2    B11 B22   
¼
Derive equation (2.47). xB11 xB22    1 2
A1 A2   
 
1 0A1 þ  1 0B1 
 exp  RT
Solution
where  i is the activity coefficient for the ith species
Inserting the equation of state for an ideal gas, defined by (2.35). These relations are frequently used in
P ¼ RT, into (2.46) and assuming ðP; TÞ ¼  is calculating the concentration of defects in minerals
constant, one has including point defects and trace elements.
 
f 1 lim P
log ¼   ðP  P 0 Þ  RT log
P RT P0!0 P0
P lim P0 Problem 2.4
¼ þ log :
RT P0!0 P
Consider a chemical reaction Ni þ 12 O2 ¼ NiO. The
Hence fðP; TÞ P0 expððTÞ  P=RTÞ. molar volumes, molar entropies and molar enthal-pies
of each phase are given in Table 2.1. Calculate the oxygen
fugacity for the temperature of T ¼ 1000  1600 K and
P ¼ 0:1MPa  10 GPa when both Ni and NiO co-exist.
2.1.3. Chemical equilibrium: the law of mass action
Solution
Consider a chemical reaction,
The law of mass action gives, ðfO2 =P0 Þ1=2 ¼ KðT; PÞ 
1 A1 þ 2 A2 þ    ¼ 1 B1 þ 2 B2 þ  (2:48) aNiO =aNi . When both Ni and NiO exist, then
aNi ¼ aNiO ¼ 1, so that
where Ai, Bi are chemical species and i, i are the
stoichiometric coefficients (e.g., H2 O ¼ H2 þ 12 O2 ). 1 .
At equilibrium for given T and P, the Gibbs free
ðfO2 =P0 Þ1=2 ¼ exp ðð 0
NiO  0Ni  0O2
2
Þ
RT : Þ
energy of the system must be a minimum with respect Now writing the temperature and pressure dependence
to the chemical reaction. When a chemical reaction of chemical potential explicitly and remembering that
described by (2.48) proceeds by a small amount, l, the pressure dependence of chemical potential is
the concentration of each species will change as included in the fugacity, one has
ni ¼ i l. The condition for chemical equilibrium
 1=2
reads fO2
! P0
X@G   
X eNiO  eNi  12 eO2 þ PðNiO  Ni Þ  T sNiO  sNi  12 sO2
G ¼ 0 ¼ ni ¼ i i l (2:49) ¼ exp :
RT
i
@ni i
Note that the enthalpy of the formation of NiO from
and hence
the elements at zero pressure (0.1 MPa) is h0NiO ¼
X
 i i ¼ 0 (2:50) eNiO  eNi  12 eO2 . Inserting the values of thermody-
i namic parameters from Table 2.2, one obtains the fugac-
where (2.12) is used. Inserting the relation (2.32) into ity versus T–P relations shown in Fig. 2.3 (for simplicity,
this equation, one finds, we assumed constant molar volumes for solid phases).
Note that the oxygen fugacity increases with pressure.
 
x1 2
A1 xA2   1 0A1 þ   1 0B1  
¼ exp  :
x 1 x 2  
B1 B2
RT
(2:51a) Problem 2.5*

This is called the law of mass action that relates the Discuss how the fugacity of water is controlled in a
concentration of chemical species with their chemical system that contains water as well as other materials such
potential. When the solution is not ideal (a case where as olivine, a metal that modifies the oxygen fugacity
solute atoms have a strong interaction with others), (H OBBS , 1984; K ARATO et al., 1986). For simplicity,
then equation (2.51a) must be modified to, assume that all the gaseous phases behave like an ideal gas.
22 Deformation of Earth Materials

TABLE 2.2 Thermodynamic properties of various oxides and fH2 O þ fH2 þ fO2 ¼ P: (3)
metals relevant to the oxygen fugacity buffer.
3
In the case where only water is present, then the dis-
 ( 106 m /mol): molar volume, h0 ðkJ/molÞ: molar enthalpy sociation of one mole of water produces one mole of
of formation from elements, s (J/mol K): molar entropy.
All quantities are at room pressure and T ¼ 298 K. Molar
hydrogen and 1/2 mole of oxygen, so fH2 ¼ 2fO2 .
volumes of some materials change with temperature and Inserting this into the equation for the law of mass
pressure as well as with phase transformations. However, action, and noting that one has
these changes are small relative to the difference in molar
volume of metals and their oxides. 3 2=3 1=3 2=3
fH2 O þ f P K ðT; PÞ ¼ P (4)
22=3 H2 O 0 1
 h0 s where for simplicity, we assume that all the gasses are
O2 24 798 0 205.15 ideal, so that all the fugacity coefficients are 1.This
Fe 7.09 0 27.28 equation gives the fugacity of water when only water
FeO 12.00 272.04 59.80 is present. At high pressures, exceeding 1 GPa, the
Ni 6.59 0 29.87 second term in this equation is small (confirm this
NiO 10.97 239.74 37.99 yourself), so that fH2 O P, but when significant disso-
Mo 9.39 0 28.66 ciation occurs (at lower pressures), then the water
MoO2 19.58 587.85 50.02 fugacity will be lower.
Now consider a case where some other species are
present that also react with oxygen, hydrogen etc. For
example, let us consider a case where material A (e.g.,
1010 Fe) reacts with oxygen to form another mineral Ax Oy
(e.g., Fe2O3), namely,
T = 1373 K
y
105 xA þ O2 ¼ Ax Oy : (5)
2
Oxygen Fugacity (Pa)

Re-ReO2 The corresponding law of mass action gives

100 y=2
axA fO2
¼ K2 ðT; PÞ: (6)
Ni-NiO aAx Oy Py=2
0
10–5
This means that when a substance A is present, oxygen
Mo-MoO2 atoms provided by the dissociation reaction of water
10–10 react with A to form Ax Oy. Then (if enough A is present)
0 5 10 15 20 25 30 the oxygen fugacity is controlled by reaction (5) and
Pressure (GPa) 2=y
should be fO2 =P0 ¼ K2 ðT; PÞ. Inserting this into (2)
FIGURE 2.3 Oxygen fugacities corresponding to several metal-
oxide buffers.
and with (3) (assuming ideal gas behavior), one finds,
2=y 2=y
PP0 K2 PP0 K2
fH2 O 1=y
; fH2 1=y
: (7)
1þK1
1 K2 1þK1 K2
Solution
It follows that, when the oxygen fugacity buffered by the
If water alone exists, then the chemical reaction that 1=y
reaction xA þ y2 O2 ¼ Ax Oy is low, i.e., K2 K1 1,
will occur is then fH2 P  fH2 O , whereas when the oxygen fugacity
1=y 2=y
1 is high ðK2 K1  1Þ, then fH2 O PP0 K2  fH2 .
H2 O ¼ H2 þ O2 : (1)
2
The law of mass action demands
2.1.4. Phase transformations: the Clapeyron
1=2 1=2
fH2 O P0 =fH2 fO2 ¼ K1 ðT; PÞ: (2) slope, the Ehrenfest slope

Now the total pressure of the gas must be the same as the For a given chemical composition, a stable phase at a
given pressure, P, so that (assuming ideal gas behavior) given pressure and temperature is the phase for which
Thermodynamics 23

the Gibbs free energy is the minimum. When a material This type of phase transformation does not involve
with a given chemical composition can assume several changes in density or in entropy (hence no latent
phases, then as the P, T conditions change, the phase heat). Note that although there is no change in density
with the minimum Gibbs free energy may change from in these types of transformation, there is a change in
one to another. In these cases, the stable phase for a the elastic constants and thermal expansion (the
material changes with these variables, and a phase second derivatives of Gibbs free energy), and therefore
transformation occurs. They include  to transforma- there must be a change in seismic wave velocities asso-
tion in quartz, order–disorder transformation in pla- ciated with a second-order phase transformation.
gioclase,  (olivine) to (wadsleyite) transformation Schematic diagrams showing the change in free
in (Mg, Fe)2SiO4 and  (bcc) to " (hcp) transformation energy associated with a first- and a second-order
in iron. transformation are shown in Fig. 2.4. In the case
A phase transformation may be classified into two where a first-order transformation is considered, a
groups. In some cases, a phase transformation involves material can assume two possible states. When the
a change in the first derivatives of Gibbs free energy free energy of one phase is lower than the other, then
(e.g., ð@G=@TÞP;ni ¼ S or ð@G=@PÞT;ni ¼ V, where S is a phase with lower free energy is more stable. Therefore
entropy and V is volume). This type of phase trans- if the transition from one state to the other is kineti-
formations is called the first order phase transforma- cally possible, then all the materials will transform to a
tion. Many phase transformations in silicates and phase with the lowest free energy. Note, however, that
metals are of this type. In these cases, there is a change this transition involves kinetic processes over a local
in density (molar volume) and heat is either released or maximum of free energy, and therefore the transfor-
absorbed upon the phase transformation (due to the mation takes a certain time to be completed.
change in molar entropy; recall that T dS is the latent Consequently, a metastable phase can exist in the case
heat). Another is the case where there is no change in of a first-order transformation when the kinetics
the molar volume or entropy (the first derivatives of involved are sluggish for a given time-scale. Examples
Gibbs free energy), but changes occur only in the include the presence of diamond at the Earth’s surface
second derivatives. This type of phase transformations (the stable phase for carbon at the Earth’s surface is
is referred to as the second order phase transformation. graphite, so we would not have diamond if the presence
Many of the structural phase transformations belong of everything on Earth were controlled by thermody-
to this class. The  to transformation of quartz is namic stability), and the possible presence of metasta-
close to this type and many structural transformations ble olivine in cold regions of subducting slabs (see
of perovskite belong to this type (e.g., G HOSE , 1985). Chapters 17 and 20). The situation is different for a

(a) (P, T ) = (P1, T1) (P, T ) = (Pc, Tc) (P, T ) = (P2, T2)
G G G
phase 2 phase 1
phase 1 phase 2 phase 2
phase 1

(b)
(P, T ) = (P1, T1) (P, T ) = (Pc, Tc) (P, T ) = (P2, T2)
G G G

FIGURE 2.4 Free energy of a system having (a) first-order and (b) second-order phase transformations. The horizontal axis represents atomic
configuration (atomic positions, crystal structure).
24 Deformation of Earth Materials

The situation is different for a multi-component

Box 2.2 The Gibbs phase rule system. Let us consider a two-component system.
The state of a system containing c-components If two phases co-exist in such a system, there are
and p-phases can be specified by (c  1)p þ 2 two degrees of freedom f ¼ 2 þ 2  2 ¼ 2. That is T
variables. Two are T and P, and for each p-phase, and P can be modified independently. This means
one needs to specify the fraction of phases that that there is a space in a T–P plane where the two
requires c1 variables. Now these p-phases are in phases co-exist. For this reason, a multi-component
chemical equilibrium, and therefore the chemical system usually shows a gradual phase transition: one
potential of each component in p-phases must be phase changes to another within a certain T–P range.
equal, For example, at a given pressure, melting in a multi-
component system begins at a certain temperature (the
11 ¼ 12 ¼    ¼ 1p solidus) and completes at another temperature (the
liquidus). At a temperature between the solidus and
21 ¼ 22 ¼    ¼ 2p
the liquidus, solid and melt co-exist. This is why melt-
        ing in Earth occurs usually as partial melting.
c1 ¼ c2 ¼    ¼ cp Let us derive an equation for the Clapeyron slope in
terms of other thermodynamic parameters. Consider a
where ij is the chemical potential of ith component
boundary between two phases for a single-component
in the jth phase. This means that there are c(p  1)
system (univariant transformation). Along the boun-
constraints. Therefore the degree of freedom of the
dary the Gibbs free energy of two phases must be
system, f, is
identical, namely,
f ¼ ðc  1Þp þ 2  cðp  1Þ ¼ c  p þ 2
G1 ¼ G2 : (2:52)
This relation is referred to as the Gibbs phase rule.
Now take the derivative along the boundary (the suffix ni
is omitted because we consider a single-component sys-
tem) to find dG1 ¼ dG2 along the boundary)
second-order transformation that occurs due to the        
instability of one phase. For a second order transfor- @G1 @G1 @G2 @G2
dT þ dP ¼ dT þ dP:
mation, no metastable phase can exist. @T P @P T @T P @P T
A couple of points may be noted. According to the (2:53)
Gibbs phase rule (Box 2.2), for a material with c
Using the identities ð@G=@TÞP ¼ S and ð@G=@PÞT ¼ V,
components, there exist f ¼ c  p þ 2 (c, the number
one obtains
of components; p, the number of phases) degrees of
 
freedom at given P and T. For example, for a single- dP S1 S2
component system (c ¼ 1), if three phases co-exist ¼ (2:54)
dT eq V1 V2
(p ¼ 3), then there are zero degrees of freedom
(f ¼ 0). That is, there is only one set of T and P at where the suffix ‘‘eq’’ is used to clearly indicate that
which three phases co-exist. Similarly, when two the derivative is taken along the equilibrium boun-
phases co-exist in a single-component system, then dary. This relation is called the Clapeyron–Clausius
there is one degree of freedom ðf ¼ 1  2 þ 2 ¼ 1Þ: relation.
that is, if T is changed then so is P. Therefore when Similar relations can be derived for a second-order
two phases co-exist in a single-component system, the transformation (Problem 2.6; e.g., C ALLEN , 1960),
temperature and pressure must be related, P ¼ PðTÞ.  
dP 1 2 C1 C2
The slope of this curve, ðdP=dTÞeq , for the first-order ¼ ¼ (2:55)
phase transformations is referred to as the Clapeyron dT eq 1=K1 1=K2 Tð1 2 Þ
slope.2
where 1,2 is the thermal expansion of 1, 2 phase, K1,2 is
the (isothermal) bulk modulus of 1, 2 phase, and C1,2 is
the specific heat (at constant pressure) of 1, 2 phase.
2
In some literature, ðdT=dPÞeq , is called the Clapeyron slope. It does not This relation was derived by E HRENFEST (1933) and
matter which definition one uses as far as one defines it clearly. hence should be called the Ehrenfest relation.
Thermodynamics 25

 
dP @V1 =@T  @V2 =@T @S1 =@T  @S2 =@T
¼ ¼
Problem 2.6* dT eq @V1 =@P  @V2 =@P @S1 =@P  @S2 =@P
@S1 =@T  @S2 =@T 1 2
Derive equation (2.55). ¼ ¼
@V1 =@P  @V2 =@T 1=K1 1=K2
C1 C2
¼ : (6)
Solution Tð1 2 Þ

For a second-order transformation, the first derivatives

of the Gibbs free energy are identical for the two
co-existing phases, V1 ¼ V2, S1 ¼ S2. Therefore, along
the boundary, the following relations must be 2.1.5. Phase diagrams
satisfied,
For a given chemical composition with temperature
        (T) and pressure (P), there exist a certain number of
@V1 @V1 @V2 @V2
dT þ dP ¼ dT þ dP phases determined by the Gibbs phase rule. A diagram
@T P @P T @T P @P T
showing the stable phases on a certain parameter space
(1a)
is referred to as a phase diagram. In constructing a
        phase diagram, one usually fixes the chemical compo-
@S1 @S1 @S2 @S2
dT þ dP ¼ dT þ dP: sition, i.e., the system is assumed to be closed. For
@T P @P T @T P @P T
a closed system, the stability of each phase is solely
(1b) determined by temperature and pressure. Fig. 2.5 illus-
trates some of the phase diagrams for binary (two-
These two equations can be combined to give,
component) systems.
2 3 A phase diagram is usually constructed based on
@V1 @V2 @V1 @V2
    direct experimental studies. However, because the
6 @T @P 7
6 @T @P 7 dT ¼ 0: (2)
4 @S @S2 @S1 @S2 5 dP stability of each phase is determined by the chemical
1
  potential, a phase diagram can be constructed theo-
@T @T @P @P
retically if the dependence of the chemical potential
In order for this equation to have a non-trivial solu- of each phase on T, P and composition (ni) is
tion, the following relation must be satisfied, known, i.e.,

@V1 @V2 @V1 @V2 iA ðT; P; ni Þ ¼ iB ðT; P; ni Þ (2:56)

 
@T @T @P @P
¼ 0: (3)
@S1 @S2 @S1 @S2 where iA;B is the chemical potential of the ith species in
 
@T @T @P @P phase A or B. Consider a single-component system,
where a material (with a fixed composition) can assume
Therefore, two phases (A or B). Then the equilibrium temperature
     and pressure are determined by
@V1 @V2 2 @V1 @V2 @S1 @S2
 þ   ¼ 0: (4)
@T @T @P @P @T @T
eA  TsA þ PA ¼ eB  TsB þ PB (2:57)
where the Maxwell relation (equation (2.15),
where eA,B, sA,B and A,B are molar internal energy,
@S=@P ¼ @V=@T) was used. Using the definitions
entropy and volume of phase A and B respectively.
of thermal expansion (th  ð1=VÞð@V=@TÞP ), (iso-
When all the parameters (eA,B, sA,B and A,B) are
thermal) bulk modulus (K  Vð@P=@VÞT ) and the
independent of temperature and pressure (this is a
specific heat at constant pressure (C  Tð@S=@TÞP )
good approximation for liquids and solids for a
and the fact that V1 ¼ V2, one finds,
small range of temperature and pressure), the
 
1 1 1 phase boundary can be calculated from eA,B, sA,B
ð1  2 Þ2   ðC1  C2 Þ ¼ 0: (5)
T K1 K2 and A,B as,

Now solving equations (1a) and (1b), the slope of the eA eB sA sB
P¼ þ T: (2:58)
phase boundary in the T–P space is given by, A B A B
26 Deformation of Earth Materials

(a) (b) FIGURE 2.5 Typical phase

T T diagrams for a binary system
involving melting reactions. (a) A
case for a solid solution, (b) a case
Liquid Liquid for eutectic behavior, (c) a case for
eutectic behavior with limited
mutual solubility, (d) a case for two
Liquid + A Liquid + B phases that have limited solubility
Liquid + Solid
below a critical temperature,
Tc, but show complete mixing
above Tc.
Solid
A+B

XA XB XA XB

(c) (d)
T T

Liquid

Tc
T1 A B

A+B A+B
solvus

XA X XB XA X XB

TABLE 2.3 Some thermodynamic parameters related to phase

Problem 2.7 transformations (from N AVROTSKY (1994)).
3
Units: e (kJ/mol), s (J/mol K),  ( 106 m /mol), dP/dT
Calculate the phase boundaries as a function of (MPa/K)
temperature and pressure for the olivine !
wadsleyite ! ringwoodite (in Mg2SiO4) phase e s  dP/dT
transformation using the values of thermodynamic
parameters listed in the Table 2.3. Mg2SiO4
olivine ! wadsleyite 27.1 9.0 3.16 2.8
olivine ! ringwoodite 39.1 15.0 4.14 3.6
Solution Fe2SiO4
olivine ! wadsleyite 9.6 10.9 3.20 3.4
The phase boundaries for these two reactions can
olivine ! ringwoodite 3.8 14.0 4.24 3.3
be calculated from (2.58). To calculate the thermo-
MgSiO3
dynamic parameters for wadsleyite ! ringwoodite, use
pyroxene ! garnet 35.1 2.0 2.83 0.71
the relation Xwad!ring ¼ Xoli!ring  Xoli!wad .
pyroxene ! ilmenite 59.4 15.5 4.94 3.3
Poli!wad ðGPaÞ ¼ 8:57 þ 0:004 27 T ðKÞ and
ilmenite ! perovskite 51.1 6.0 1.89 3.2
Pwad!ring ðGPaÞ ¼ 12:24 þ 0:006 12 T ðKÞ.
garnet ! perovskite 75.0 7.5 –
Thermodynamics 27

Solid-solution, eutectic melting a modification to the phase diagram toward the end-
When there are two or more components in the system, member component representing the effects of finite
there are additional degrees of freedom by which the solubility (Fig. 2.5c). A phase diagram for a silicate and
chemical potential is controlled. Consequently, the water system at high T is an important example.
phase diagram depends on how the chemical potential Consider the equilibrium at temperature T1 below the
of each phase varies with the composition. For sim- eutectic point. When the amount of B is small, then the
plicity let us consider a two-component system. The only phase that exists is a phase A that contains a small
component i ¼ 1 and 2 may assume various phases amount of B. According to the Gibbs phase rule, in
such as solid and liquid. Two cases may be distin- such a case we have f ¼ c  p þ 2 ¼ 3, that is this
guished. One is the case in which the two components phase, i.e., phase A with a small amount of B can
mix well in both the solid and liquid phases. In this exist for a range of T, P and composition. When the
case, the contribution from the configurational amount of B in the system increases, then at a certain
entropy is similar for both the solid and liquid phases, point, the phase A can no longer dissolve all the com-
and the free energy of each phase changes with compo- ponent B and there will be two phases ðX2 4X 4X1 Þ.
sition similarly following the compositional depend- The same thing happens from another side, namely the
ence of internal energy, entropy and the molar B-phase side. Consequently the domain is divided into
volume. The phase diagram corresponding to this one-phase domains in each side of the phase diagram
case is shown in Fig. 2.5a. In such a case, solid A and (A-rich or B-rich, X5X1 ; X 4X2 ) and a two-phase
B are said to form a solid-solution. Another is the case domain ðX2 4X 4X1 Þ. In the latter domain, there are
where mixing occurs only in the liquid phase. In this two phases that co-exist, and therefore the degree of
case, the contribution from the configurational freedom is f ¼ 2. Consequently, if temperature and
entropy is important only in the liquid phase. pressure are prescribed, then the chemical composi-
Consequently, the free energy of the liquid becomes tions of a material must be fixed. The boundaries
low in the intermediate concentration of a given spe- between the one- and two-phase regions correspond
cies, and therefore the solidus of the system is reduced to the solubility of each species into another.
significantly at intermediate compositions (Fig. 2.5b). Usually the solubility of another phase into a given
Melting behavior due to this type of mixing property is phase increases with temperature, so the boundaries
called eutectic melting. separating two one-phase domains will become closer
The solid-solution type behavior is observed when as temperature rises. These boundaries are often
the solid phases involved have similar properties (crys- referred to as a solvus. When mutual solubility is
tal structure and chemical bonding). The examples large, then at a certain temperature below the melting
include magnesiowüstite (MgO and FeO), olivine temperature, the two solvus curves merge. Above this
(fayalite Fe2SiO4 and forsterite Mg2SiO4), plagioclase critical temperature (Tc) the two phases mix com-
feldspar (albite NaAlSi3O8 and anorthite CaAl2Si2O8). pletely. Above this temperature mixing occurs both in
In all of these cases, ions that have similar ionic radii solid-state and liquid-state, and therefore the phase
are incorporated as a solid-solution in the solid phase. diagram above this temperature should look like that
If the ionic radii are largely different then the solubility of a solid-solution (Fig. 2.5d). Obviously the solvus
in the solid phase is limited and the eutectic behavior curves or any of the boundaries on a phase diagram
will occur. This is the case for the MgO–CaO, also depend on pressure. The temperature and pressure
MgSiO3–Mg2SiO4 systems. dependence of solvus curves for various combinations
of minerals is used as petrological barometers and/or
thermometers (e.g., W OOD and F RASER , 1976).
Solvus
Let us now consider a two-component system in which Effects of non-stoichiometry: a phase diagram
there is a finite solubility of each phase into another in for an open system
the solid state as well as in the liquid state. First, con- The phase diagram considered above assumes that the
sider a system in which mixing is complete in the liquid chemical composition of each phase is independent of
state and a small degree of mixing also occurs in the T and P except in cases where finite solubility of one
solid state. In such a case, a phase diagram needs to be component occurs in each phase. For example, a phase
modified. A solid phase always contains, in this case, a diagram for (Mg, Fe)O is usually constructed assuming
finite amount of secondary component so that there is that this is a two-component system (MgO and FeO)
28 Deformation of Earth Materials

assuming that the material exchange occurs only 7.5

through Mg , Fe keeping the number of atoms in
hematite
the (solid) system constant. This is not strictly true
when the system under consideration is open (i.e.,
when the system exchanges materials with the sur-
rounding system). In a system like XO (X ¼ Mg etc.; 7.0

10 000 / T(K)
O, oxygen), the ratio of the number of atoms of X and
O (stoichiometry) can deviate from what the chemical magnetite
formula would indicate. The deviation from the formal
chemical formula is referred to as non-stoichiometry. γ-iron wüstite
When non-stoichiometry occurs in an ionic crystal, 6.5
then charge balance must be maintained by creating
another type of charged species. This is usually done by
creating point defects or by incorporating another spe-
cies. One example is an Fe-bearing mineral such as
6.0
olivine ((Mg, Fe)2SiO4) that can have non-stoichiometry –10 –5 0 5
caused by a change of valence state of iron log f O (MPa)
2

ðFe2þ , Fe3þ Þ. In this case the charge balance is main- FIGURE 2.6 A phase diagram of Fe-O at 0.1 MPa.
tained by the change in the concentration of M-site
vacancies that have a negative effective charge (see at the intermediate oxygen fugacity and hematite
Chapter 5). Another example is a combined substitu- (Fe2O3) at high oxygen fugacity. Iron in wüstite is
tion such as Al3þ þ Fe3þ , Si4þ þ Fe2þ . In these mostly ferrous iron (Fe2 þ), whereas in magnetite
cases, an additional variable such as oxygen fugacity there are both ferrous iron (Fe2 þ) and ferric iron
or the activity of Al2 O3 is needed to specify the degree (Fe3 þ) and finally at high oxygen fugacity all iron
of non-stoichiometry. The degree of non-stoichiometry changes to ferric iron (Fe3 þ). The stability of iron-
in the former type of processes is usually small (10 4 bearing olivine can be analyzed in a similar way.
or less in olivine) but can be large in an Fe-rich Olivine accepts ferrous iron but not ferric iron (ferric
compound such as FeO (in FeO the non-stoichiometry iron is present in olivine but only with a very small
is 8%, i.e., Fe0.92O). Even in cases where the degree of amount, 1–10 ppm, as point defects) and therefore it
non-stoichiometry is small, its effects on physical prop- is stable only within a certain range of oxygen fugacity
erties can be important. In a binary material (such as that is determined by the stability of wüstite (FeO).
XO), the oxygen fugacity is used as an additional var- A somewhat different phase diagram applies when a
iable in constructing a phase diagram (N ITSAN , 1974). given mineral favors ferric iron more than ferrous iron.
(In a ternary system such as Mg2SiO4, the stoichiom- In such a case, even at an oxygen fugacity in which iron
etry is defined by two ratios (i.e., Mg=O; Mg=Si), and would occur as FeO, iron in that mineral can be ferric
hence one needs two additional parameters to completely iron. In some cases, the stability field of the ferric iron-
describe the chemical state of the system. Both oxygen bearing phase expands to a much lower oxygen fugac-
fugacity and the oxide activity must be specified in such ity, and in such a case a mineral containing ferric iron
a case.) could co-exist with metallic iron. An important case is
To illustrate this point, let us consider a phase dia- silicate perovskite that favors ferric iron, and the for-
gram of Fe–O. Iron (Fe) can assume three different mation of silicate perovskite from ringwoodite leads to
valence states dependent on oxygen fugacity, fO2: met- the formation of metallic iron (e.g., F ROST et al., 2004).
allic iron Fe0 at low oxygen fugacity, ferrous iron Fe2 þ
at intermediate oxygen fugacity and ferric iron Fe3 þ at 2.2. Some comments on the
high oxygen fugacity, see Fig. 2.6. Each species (Fe0, thermodynamics of a stressed system
Fe2 þ and Fe3 þ) has a different chemical character and
therefore the stable phases at different conditions will In the usual treatment of thermodynamics, the energy
depend on the oxygen fugacity. Consequently in an change of a system due to mechanical work is treated
Fe–O system, four compounds may be present depen- assuming hydrostatic stress. That is dW ¼ P dV, i.e.,
dent upon the oxygen fugacity, i.e., metallic iron at low the work done against pressure. An extension of such a
oxygen fugacity, wüstite (FeO) and magnetite (Fe3O4) treatment to non-hydrostatic stress conditions is
Thermodynamics 29

σn into play because the displacement is normal to the

boundary), thus
0
G ðxÞ ¼ E  TS þ n ðxÞ  O ¼ G0 þ n ðxÞ O (2:60)
where E* is the internal energy including the strain
grain energy (e.g., (2.59)), S* is the entropy (vibrational
entropy), n(x) is the normal stress on the grain boun-
liquid/grain boundary
dary that depends on the orientation of the boundary
grain
with respect to the applied stress, G0 ¼ E  TS þ PO
Ω and 0n ðxÞ is the deviatoric component of the normal
vacancy stress ð 0n ðxÞ ¼ n ðxÞ  PÞ. Equation (2.60) defines
the Gibbs free energy of a solid under stress near a
FIGURE 2.7 The process of formation of a vacancy (or dissolution boundary. Note that G ðxÞ is heterogeneous and there
of atoms) in a stressed solid at a grain boundary (or crystal–liquid
interface).
is no absolute equilibrium state under the deviatoric
stress. The free energy defined by (2.60) represents
only the local equilibrium. Also note that although
needed when the deformation of materials is consid- the internal energy (E*) depends on stress, its depend-
ered. The strain energy of stressed solid (per unit vol- ence is of second order (equation (2.59)) and the domi-
ume) is usually written as (e.g., L ANDAU and L IFSHITZ , nant role of stress is through the n ðxÞ  O (or 0n ðxÞ  OÞ
1959; S CHMALZRIED , 1995) term.
X There have been extensive debates concerning the
dW ¼ ij d"ij (2:59) exact formula for the free energy of stressed solids (see
ij
e.g., J OHNSON and S CHMALZRIED , 1992; K AMB , 1961;
where ij is stress and d"ij is the strain increment. P ATERSON , 1973; S HIMIZU , 1992). However, except for
However, this expression does not capture the energy the above well-defined case, many controversies are
difference that plays an important role in diffusional around the strain energy term. The contribution to
(or pressure-solution) creep (Chapter 8). According to this term in an actual stressed solid is usually domi-
this equation, the energy of the system is independent nated by the energy due to heterogeneous dislocation
of the orientation of the interface at which the reaction density rather than the strain energy of dislocation-free
occurs, but the free energy change associated with a solids (e.g., P OIRIER and G UILLOPÉ , 1979), and the
particular reaction under non-hydrostatic stress relevance of such a controversy to real geological issues
depends on the process that occurs at the crystal sur- (or physical processes) is questionable.
face (or grain boundary). Consequently, the free
energy change may depend on the orientation of the 2.3. Thermodynamics of irreversible
interface with respect to the applied stress at which the processes
reaction occurs. To illustrate this point, let us consider
the formation/destruction of a vacancy (dissolution/ A brief account of the thermodynamics of irreversible
precipitation of a stressed solid) into a grain boundary processes is given in this section. From the microscopic
(or a liquid) (Fig. 2.7). Consider the formation of a point of view, elastic deformation occurs when atomic
vacancy in a crystal. The formation of a vacancy is displacement from stable positions is limited to the
made by the removal of an atom from inside a crystal small vicinity of stable positions at which the material
to the surface. Therefore, there is a volume expansion, assumes minimum free energy. When the force that
and hence the formation free energy of a vacancy is causes deformation is removed, atoms move back to
often written as G ¼ E  TS þ PO, where O is the their original positions. Therefore elastic deformation
volume of a vacancy (atom). However, to be more is reversible (and instantaneous). However, statistical
specific, the energy associated with vacancy formation mechanics shows that, at finite temperatures, there is
depends on where it occurs. To see this, let us recall that a non-zero probability that an atom can assume a
vacancy creation (destruction) occurs mostly at grain position away from the equilibrium position. This ran-
boundaries. When an atom is removed from the inte- dom motion of atoms is referred to as fluctuation.
rior of a crystal to its grain boundary, one needs to do Consequently, an atom could sometimes move to a
work against normal stress (only normal stress comes new stable position over the potential barrier with a
30 Deformation of Earth Materials

certain probability (see section 2.3). This process can natural to define the gradients of these quantities as
be assisted by the applied stress. Once this happens, generalized forces that drive the motion of correspond-
then upon the removal of the stress, the atomic config- ing (extensive) quantities. Therefore we define a gener-
uration of the material would not revert to the original alized force by
configuration instantaneously: thus the deformation
X  r (2:62)
becomes irreversible (and time dependent because the
process now depends on the probability of atomic where  ¼ 1=T; P=T; =T and define a linear rela-
jumps). tionship between the force and the flux, J, of an exten-
Plastic deformation occurs due to irreversible pro- sive quantity,
cesses. Therefore an understanding of irreversible pro-
cesses is critical in the study of plastic deformation. The J ¼ L  X ¼ L  r (2:63)
important issues here are the relationship between where L is a material constant that is often referred to
mass flux and various driving forces for mass flux as a phenomenological coefficient. For instance, the
and the interaction of various (generalized) forces. extensive quantity corresponding to an intensive
More detailed discussions of the thermodynamics of parameter 1=T is the internal energy, so the flux corre-
irreversible processes can be found in C ALLEN , 1960, sponding to rð1=TÞ will be the flux of energy, i.e., heat
DE G ROOT and M AZUR (1962) and G LANSDORFF and
flow. In this case, the empirical law is Fourier’s law,
P RIGOGINE (1971).
JT ¼ krT (2:64)
2.3.1. Flux and the generalized forces
where JT is the energy flux and k is thermal conductiv-
Irreversible motion or the flux of atoms (or energy) ity. Comparing Fourier’s law with equations (2.62) and
occurs when a system is out of equilibrium. One can (2.63) ( ¼ 1=T), one obtains
surmise that some forces cause such a flux.
L ¼ kT2 . (2:65)
Experimental observations show that in most cases
the rate of motion of atoms or the flux of atoms (or Similarly, the extensive quantity corresponding to an
energy) is linearly proportional to the force. An exam- intensive parameter =T is the number of atoms, n,
ple of such an empirical law is Fick’s law of diffusion so the corresponding flux is the flux of atoms. The
(of atoms) (also see Chapter 8), empirical relation for this case is therefore Fick’s law
J ¼ Drc: (2:61) of atomic diffusion,

where J is the flux of atoms, D is diffusion coefficient J ¼ Drc (2:66)

and c is the concentration of atoms (number of atoms
where D is the diffusion coefficient and c is the concen-
per unit volume). Fick’s law, (2.61), is given for a case
tration of atoms (number of atoms per unit volume)
where only one type of force is present (concentration
and we have,
gradient) and only single atomic species move. The
generalization of this type of flow law is important in Dc
L¼ . (2:67)
considering atomic transport (and plastic flow) in an R
ionic crystal.
In generalizing an equation such as (2.61), it is
useful to examine the physical basis for atomic flow. Problem 2.8
In the previous section (section 2.1), we learned that in
Derive equation (2.67).
a system at thermochemical equilibrium, the entropy
must be maximum and hence the following quantities
(intensive parameters) must be homogeneous, 1=T; Solution
P=T and =T (equation 2.22). When the system is
Using the relation (2.32),  ¼ 0 þ RT log x ¼ 0 þ
out of equilibrium, a flow of atoms (or energy) will
RT log ðc=c0 Þ, one obtains X ¼ rð=TÞ ¼ Rðrc=cÞ.
occur due to the heterogeneity of these quantities. Such
Inserting this into (2.64), one gets J ¼ Drc ¼
a flow produces entropy. Since entropy is given by the
ðDc=RÞX  LX, and therefore L ¼ Dc=R.
product of extensive quantities (such as energy) and
intensive parameters (such as temperature), it is
Thermodynamics 31

From the above discussion, two methods of general- the fact that the entropy production rate (equation 2.66)
ization of the above linear relationship are obvious. First, is a scalar, so that upon the transformation of the coor-
Fick’s law describes the flux of atoms due to the concen- dinate this quantity must not change.
tration gradient, but more generally, atomic flux due to Among the off-diagonal components that represent
the gradient of chemical potential X ¼ rð=TÞ can be the coupling of different fluxes, the following symme-
considered. The chemical potential may include not only try relation, the Onsager reciprocal relation, must be
the gradient of concentration (gradient of configura- satisfied,
tional entropy) but also the gradient of free energy due
Lij ¼ Lji (2:70)
to other effects such as electrostatic field. Electrostatic
interaction among different diffusing species plays an if the independent fluxes are written as linear functions
important role in ionic crystals (see Chapter 8 for more of the independent generalized forces. The Onsager
details). In such a case, the electrostatic energy must be reciprocal relation is a consequence of the symmetry
included in the chemical potential thus of fluctuation with respect to time (for details see e.g.,
C ALLEN , 1960; DE G ROOT and M AZUR , 1962; L ANDAU
 ð0 þ RT log ðc=c0 Þ þ q Þ
X ¼ r ¼ r and L IFSHITZ , 1964).
T T
q rc q rc
¼ r R ¼ ER (2:68)
T c T c 2.3.2. Some notes on the driving forces for
plastic deformation
where q is the electrostatic charge of the species, is the
electrostatic potential and E is the electrostatic field The driving forces considered in the previous section
(E ¼ r ). may be called thermodynamic forces (such as a force
Second, in writing equation (2.62), we considered due to the gradient of concentration of a given species).
only one force and one flux. A natural generalization However, under a deviatoric stress, there are forces that
of this relation is directly arise from the applied macroscopic (stress) field.
X In a solid under deviatoric stress, what forces drive
Ji ¼ Lij  Xj (2:69)
j
large-scale atomic motion leading to plastic flow? Two
cases can be distinguished. First, for an atomic species
where we include various types of forces and corre- such as individual atoms or isolated point defects, there
sponding fluxes. The phenomenological coefficients dis- is no direct interaction between applied stress and these
cussed above correspond to the diagonal components, atomic species. In a perfect crystal, atoms are in their
Lkk, namely a material parameter describing the flow of lattice sites without any stress or strain. Similarly for
intensive variables (matter (or energy)) due to the corre- point defects, most of them are also isotropic and are
sponding gradient of extensive parameters (generalized not associated with deviatoric strain so that they do not
force). Off-diagonal components of phenomenological interact with deviatoric stress (exceptions are defect
coefficients (Lij with i ¼
6 j) express the coupling of differ- complexes, which can have anisotropic strain field and
ent fluxes. Physically these terms represent the flux of the interact with shear stress causing anelasticity; see
ith variable caused by the generalized force conjugate to Chapter 11). Consequently applied stress provides no
the jth variable. For example, the temperature gradient direct driving forces for atomic species. The most impor-
can cause the diffusion flux of atoms (Soret effect), and tant driving force for the motion of these atomic species
conversely the concentration gradient can cause the (e.g., point defects) under deviatoric stress is the ther-
temperature gradient (Dufour effect). Diffusion in a modynamic force, X ¼ rð=TÞ ¼ Rðrc=cÞ, caused
multi-component system is an important example by the heterogeneity of defect concentration due to the
where coupling among different species can cause heterogeneous microstructure of a polycrystalline mate-
important effects on mass transport (e.g., L ASAGA , rial (see Chapter 8). Heterogeneity of defect concentra-
1997, see also Chapter 8). The entropy production rate tion can also occur due to the local stress heterogeneity
contains various terms including the scalar (PV term), due to the presence of a dislocation or a grain boundary
vector (heat flow, diffusional mass flow) as well as tensor (see Chapters 8 and 9). In these cases, dislocation
(energy dissipation due to plastic flow). A coupling of motion or motion of atoms across a grain boundary
flux of various variables can occur only when these occurs due to (indirect) thermodynamic force. Second
variables have the same transformation with respect to is the case for dislocation motion, which is different
the change of the coordinate system. This comes from from the case for point defects. Due to the long-range
32 Deformation of Earth Materials

X
displacement field associated with it, a dislocation has a ¼ Lij Xi Xj (2:73)
direct mechanical interaction with the applied stress i;j
(Peach–Koehler force: see Chapter 5). A dislocation
where we used the Onsager reciprocal relation. Now con-
moves both by the direct effect of applied stress as well
sider a case where one force (Xk) is fixed but others are free
as by the thermodynamic forces caused by the concen-
to vary. The entropy production rate is minimum with
tration gradient of point defects around it.
respect to Xi ði ¼
6 kÞ when the following condition is met,
In some literature, the gradient in chemical potential
that drives processes such as diffusion under stress, , is @ X ij j
¼ L X ¼ Ji ¼ 0 ði 6¼ kÞ: (2:74)
written as (e.g., S CHMALZRIED , 1995) @Xi j

 O
r ¼ Rr log a   r (2:71) At the stationary state, all the fluxes other than the one
T T
corresponding to a fixed force (Xk) should vanish. Now
where a is the activity and O is the molar volume of the from the conservation equation for an extensive
species involved. This expression is misleading and parameter i ; ð@i =@tÞ ¼ r Ji , it is seen that the
should be used with care. First, this equation would state of minimum entropy production rate (i.e.,
imply that the applied stress has a direct effect on defect Ji ¼ 0) corresponds to the stationary state
motion. This is not true in many cases. For example, ð@i =@t ¼ 0Þ. This is referred to as the principle of
diffusion flux in a polycrystalline material under devia- minimum entropy production rate.
toric stress is caused by the indirect thermodynamic
force due to the local (grain scale) variation in the
normal component of stress at the place where crea- 2.4. Thermally activated processes
tion/destruction of defects occur (Chapter 8). In other
2.4.1. Absolute rate theory
words, stress in this equation should not be the macro-
scopic stress.3 The use of equation (2.71) will obscure Basic theory
this basic physics of stress-induced kinetic processes An irreversible process involves an atomic jump from
and is not recommended. Second, in the case of diffu- one stable position to the next. The atomic jump is
sion flux in a polycrystalline material, the two terms in often caused by a large fluctuation of atomic positions
equation (2.71) are redundant if the stress dependence due to lattice vibration. Such a process is referred to as a
of concentration is included in the activity term thermally activated process. The energy of an atom in a
because in this case, the driving force is a thermody- crystal has a minimum value at the lattice sites. As one
namic driving force, rð=TÞ ¼ Rr log a, caused by measures the energy of an atom from one lattice site to
the grain-scale variation in stress, and is given the neighboring site, the energy has a local maximum
by rð=TÞ ¼ Rr log a ¼ ðO=TÞ  ð =LÞ ðO=TÞ somewhere in the middle of the two lattice sites. The
r ða ¼ a0 expð O=RTÞÞ. value of the energy maximum depends on the pass
along which one moves an atom. The position at which
2.3.3. Stationary (steady-state) state: the principle the energy at the peak is minimum is called a saddle point
of minimum entropy production rate (Fig. 2.8). Through thermal vibration, atomic positions
fluctuate. Therefore, with some non-zero probability, a
The flow of materials (or energy) causes entropy pro- system can assume an atomic configuration in which an
duction (energy dissipation). The rate of local entropy
production is given by H
dS X i i X i S
 ¼ JX ¼ J  ri : (2:72)
dt i i
H*
Inserting the relation (2.69) into (2.72), one gets
A
A′
FIGURE 2.8 A plot of enthalpy as a function of atomic position
3
If equation (2.71) is used, with being the macroscopic stress, then there associated with an atomic jump from position A to position A0 . S is the
would be no diffusion or deformation for a homogeneous deviatoric saddle point. H ¼ E þ PV is the activation enthalpy (E* is the
stress. activation energy and V* is the activation volume).
Thermodynamics 33

atom goes to a saddle point. At a saddle point, the ES  EA is always positive whereas VS  VA can be
potential energy of an atom has a maximum with respect positive or negative (see Chapter 10). Consequently,
to the spatial coordinate, and consequently, this config- an atomic jump is always enhanced by higher temper-
uration is unstable and an atom moves to the next stable ature, whereas higher pressure usually suppresses an
position (or goes back to the original position). When the atomic jump (Chapter 10). Also note that the temper-
velocity of an atom at that saddle position is ‘‘positive,’’ ature dependence of thermally activated processes
then the atom goes to the next stable position (a lattice results in the time dependence of the process because
site). Since this new configuration is stable, that atom these processes occur as a result of fluctuation.
does not go back to the original position instantaneously.
Therefore the motion over a saddle point is irreversible. Effects of stress on thermal activation
The rate at which this occurs may be calculated by The above formula shows that the rate at which a ther-
applying the principles of statistical physics and the fol- mally activated process occurs is strongly dependent on
lowing theory is called the absolute rate theory, which temperature and pressure through activation enthalpy,
was developed by E YRING (1935). The application of H*. When an atom (or defect) has an interaction with
this theory to atomic migration in a solid was made by stress, the activation enthalpy becomes dependent on the
F LYNN (1968) and V INEYARD (1957) (see also F LYNN stress. This is a case for dislocations (see Chapters 5 and
(1972)). Noting that the activated state has different 9) which have an anisotropic strain field and hence have
chemical bonding and hence the frequencies of lattice strong interactions with applied (deviatoric) stress. In
vibration are different, one obtains (e.g., F LYNN (1972)) such a case, the activation enthalpy becomes stress
  dependent, H*( ). In these cases, applied stress changes
H
w ¼ ~  exp  (2:75) the energy of a dislocation when it moves from the
RT ground state to the saddle point configuration. The
where stress works on a piece of dislocation when a piece of
dislocation moves to the direction of the Peach–Koehler
Q
3N force (Chapter 5) and hence the enthalpy will be
oi
i¼1 reduced. On the other hand, when a dislocation moves
~ ¼ (2:76)
Q
3N in the opposite direction, the enthalpy for the barrier will
oi0
i¼2 increase. The net rate of motion will then be given by

is a constant with a dimension of frequency (oi and o0i w ¼ wþ  w



  
are frequencies of lattice vibration with the ith mode at H þð Þ H ð Þ
¼ ~ exp   exp  (2:78)
the ground state (A) and at the saddle point configu- RT RT
ration (S) respectively) and although the backward activation is not very impor-
H ¼ HS  HA ¼ ðES  EA Þ þ PðVS  VA Þ (2:77) tant at high stresses.
Because the activation enthalpy in these cases
is the activation enthalpy (ES,A (VS,A): internal energy decreases with stress, there is a threshold stress, c, at
(volume) of a system with state S and A). An interpre- which activation enthalpy vanishes,
tation of equation (2.75) is that ~ is the ‘‘trial fre-
H ð c Þ ¼ 0: (2:79)
quency,’’ and expðH =RTÞ is the success rate of a
jump. ~ is of the order of the Debye frequency (char- When stress reaches this level, then motion of defects is
acteristic frequency of lattice vibration, see Chapter 4) possible without the help of thermal activation. In
and does not change very much with pressure or tem- other words, there is a finite probability of defect
perature. The most important parameter to control the motion even at T ¼ 0 K. Such a motion of defects is
rate of process is H*, the activation enthalpy. This term referred to as athermal motion. For athermal motion,
(i.e., expðH =RTÞ) can change by orders of magni- fluctuation does not play an important role and hence
tude as temperature and/or pressure (or other chemical the process becomes (nearly) time independent. A phe-
factors such as water fugacity) change. nomenon often referred to as yield corresponds to this
The rate of atomic jump depends exponentially on athermal motion of defects (dislocations) (the stress at
temperature and pressure and hence the effects of tem- which plastic deformation occurs at low temperatures
perature and pressure on an atomic jump are large. is often referred to as a yield stress).
 3 Phenomenological theory
of deformation

The formal theory of deformation plays an important role in formulating energy dissipation
(i.e., seismic wave attenuation) and non-linear rheological relationships. This chapter presents a
brief summary of the phenomenological theory of plastic deformation. This includes the classification
of deformation (elastic, viscous, plastic etc.), the mathematical formula for constitutive relations,
formulation of transient creep and the mathematical formula appropriate for non-linear rheology.

Key words elasticity, visco-elasticity, anelasticity, constitutive relations, Levy–von Mises equation,
mechanical equation of state, transient creep, creep response function, creep compliance function,
Kramers–Kronig relation, Maxwell model, Voigt model, Zener model, Burgers model.

3.1. Classification of deformation dependent deformation and a fraction of the strain is

not recoverable.
The response of a material to applied stress can occur Deformation that has time-dependent and non-
in a variety of ways. Under some conditions, an recoverable strain is called non-elastic deformation.
equilibrium state can be achieved upon applying a Microscopically, this type of deformation occurs when
stress, whereas time-dependent or steady-state defor- the atomic motion is so large that atoms move, over the
mation can also be achieved. To understand the potential hill, to the next stable positions (Fig. 3.1b).
microscopic basis for this difference, it is useful to Once atoms move over the potential hill, then even if
consider the nature of atomic displacement caused the stress is removed, they will not go back to their
by an applied stress. At static equilibrium, each original positions immediately. Note, however, that
atom occupies a position corresponding to the mini- under some conditions, recoverable but time-dependent
mum potential energy (Fig. 3.1a). Upon applying a (delayed) deformation could occur. This is the case when
stress (a force per unit area), atoms move their posi- deformation involves atomic motion over the
tions from their stable positions. If the stress is small, potential hill, but deformation causes elastic strain inside
or the temperature is low (or time is short), then only a material (back stress). In this case, after the removal of
small, instantaneous displacement will occur. external stress, atomic motion occurs in such a way as to
Consequently, when the stress is removed, atoms go reduce the internal (back) stress associated with elastic
back to their original positions. This type of instanta- strain so that the final equilibrium state will have no
neous and recoverable deformation is called elastic permanent strain: strain is recoverable but time-depend-
deformation. In contrast, when a large stress is applied ent. This type of deformation is often referred to as
or stress is applied for a long time (at high temper- anelastic deformation. In contrast, deformation can be
atures), then a material responds to the stress not time-dependent and strain is non-recoverable. This
only instantaneously but also through delayed, time- type of deformation is referred to as viscous (or
34
Phenomenological theory of deformation 35

(a) (b) FIGURE 3.1 Atomistic view of

elastic non-elastic (a) elastic (recoverable) and (b)
non-elastic (non-recoverable)
deformation.

plastic1) deformation. However, both anelastic and vis- σ = constant

coelastic deformation are often collectively referred to as
anelastic deformation.
Deformation can occur nearly instantaneously and

Strain
transient steady state
be non-recoverable. Fracture or low-temperature
yielding at high stress is nearly time-independent yet
deformation is non-recoverable. This type of behavior
is, however, better considered as an end-member εe
behavior of plastic deformation (e.g., H ART (1970)).
The mode of non-elastic deformation is conveniently Time
classified into brittle and plastic (ductile) deformation.
FIGURE 3.2 A schematic diagram of a strain versus time relation for
In more casual terms, brittle deformation is fracture and a constant stress test showing various stages of deformation, where "e
ductile deformation is flow. These two processes are is the elastic strain. This figure corresponds to a case where work-
nearly independent and therefore the non-elastic hardening occurs in the transient stage.
response of a material to the external force is either
brittle fracture or ductile flow whichever is easier.
Fracture involves the macroscopic breaking of chemical elastic response yielding elastic strain, but we focus on
bonds (at the scale of cracks), which occurs in most the non-elastic component. Non-elastic deformation
cases in a localized fashion. In most cases, ductile flow usually starts with a transient period in which the strain
involves the microscopic motion of atoms and in many rate (at constant stress) or stress (at constant strain
cases occurs homogeneously. Consequently, micro- rate) changes with time. Eventually a material will
scopic, thermally activated motion of atoms (defects) usually assume steady-state deformation in which the
controls ductile deformation, but thermally activated strain rate (for constant stress) or stress (for constant
processes play a less important role in brittle fracture. strain rate) remains constant with time. A few points
In this book, we focus on plastic flow, but a brief should be noted. First, the steady state is only approx-
summary of brittle deformation is given in Chapter 7. imately defined, and in fact, there are several quasi-
Extensive reviews of brittle deformation are given by steady-states of deformation for a given material.
P ATERSON and W ONG (2005) and S CHOLZ (2002). This is due to the fact that plastic deformation is sensi-
tive to microstructures and there is a hierarchy
3.2. Some general features of plastic of microstructures. As microstructures evolve, plastic
deformation properties also evolve. Consequently, when different
microstructures assume quasi-steady-state at different
Fig. 3.2 illustrates a typical result of mechanical tests in time-scales, there will be several quasi-steady-states of
the plastic regime. In all cases, there is an instantaneous deformation for a given material. One example is the
deformation of a polycrystalline material in the dislo-
cation creep regime (see Chapter 9). Dislocation micro-
1
In the literature, the term ‘‘plastic deformation’’ is often used to describe
structures at each grain level will evolve during
nearly time-independent yielding, and the other type of strongly time-
dependent deformation is called viscous flow. However, in this book I use
deformation, and quasi-steady-state deformation will
‘‘plastic deformation’’ for any time-dependent non-recoverable be reached when dislocation microstructures achieve
deformation. steady state (constant dislocation density, constant
36 Deformation of Earth Materials

subgrain structures etc.). At the same time, orienta- steady state, the plastic properties of a material are
tions of individual grains will evolve with deformation characterized by a single scalar parameter, viscosity,2
to result in a lattice-preferred orientation (see , defined by,
Chapter 14) that will also affect plastic deformation.
ij  2"_ ij : (3:2)
Consequently, another quasi-steady-state will be
achieved when steady-state lattice-preferred orienta- Consequently, the plastic properties in such a case
tion is established. Grain size may also evolve during can be determined by measuring the one component
deformation that affects the plastic flow (see of strain rate corresponding to one type of stress.
Chapter 13). In most cases, these grain-scale micro- For example, one can apply normal stress on the x1
structures evolve with longer time constants than plane, 1 , and observe the corresponding strain
dislocation microstructures. Therefore quasi-steady- rate "_ 1 . The viscosity can be determined by
states corresponding to steady-state lattice-preferred  ¼ ð1  3 Þ=2ð"_ 1  "_ 3 Þ ¼ ð1  3 Þ=3"_ 1 (note that
orientation and steady-state grain size will be achieved "_ 3 ¼ 12"_ 1 ) where 3 is the lateral stress (confining
in later stages than the quasi-steady-state correspond- pressure). Given this effective viscosity, strain rates
ing to steady-state dislocation microstructures. It is corresponding to any other stress components can
important to define the scale that one is talking about be calculated from (3.2). For a more general aniso-
when one discusses the time-dependent deformation of tropic rheology, one has to determine the rheological
materials. Second, under some conditions, steady-state properties for a large number of orientations.
deformation becomes unstable and localized unstable In the case of non-linear rheology, one may deter-
deformation occurs (Chapter 16). mine a relationship such as "_ 1 ¼ Að1  3 Þn (n > 1).
In this chapter we will discuss some general (phe- But this equation cannot be generalized to "_ ij ¼ Anij ,
nomenological) aspects of the mathematical descrip- because such a relation violates the rule of transforma-
tion of mechanical behavior in the plastic regime. The tion of tensors (if one rotates the coordinate system,
issues that we discuss in this chapter include: (i) the the left- and the right-hand sides would change differ-
formulation of non-linear flow laws (section 3.3), ently). If a material is isotropic, stress and strain rate
(ii) the formulation of transient deformation behavior must be related by a scalar quantity. Therefore, for
(section 3.4) and (iii) the influence of ductile deforma- non-linear rheology, one must have,
tion on energy loss and on apparent elastic moduli
"_ ij ¼ Bðij Þ  ij (3:3)
(section 3.5).
where B is a scalar function of stress. Because B is a
3.3. Constitutive relationships for scalar, it must contain only the invariants of (devia-
non-linear rheology toric) stress tensor (see Chapter 1). Because the first
invariant of stress tensor is hydrostatic pressure, the
The plastic properties of a material are expressed in first invariant is usually assumed not to contribute to
terms of a relationship between stress and strain or deformation (other than its effect on a constant
strain-rate. Therefore a material parameter (or para- term related to the rate of thermal activation, see
meters) that characterizes plastic deformation must Chapter 10). The third invariant is also assumed not
express a relation between two second rank tensors to contribute to assure that energy dissipation is always
(deviatoric stress ij and deviatoric strain rate "_ ij ). In positive (see equation (3.11)). In this case, the simplest
the simplest case of linear rheology at steady state, the form will be,
relation should be
X B ¼ BðII Þ ¼ B  IIðn1Þ=2 : (3:4)
ij  2 ijkl "_ kl (3:1)
k;l Therefore equation (3.3) becomes

where ijkl is the viscosity tensor. It follows that a large "_ ij ¼ B  IIðn1Þ=2
 ij : (3:5)
number of measurements along various orientations
are needed to fully characterize the plastic properties
2
Similar to an elastic material (see Chapter 4), there are two viscosities for
of a material. However, in many cases, a material is
an isotropic material, one for shear and another for bulk deformation.
approximately isotropic. This would be the case where However, for a perfectly dense solid, the bulk viscosity is infinite.
one considers deformation of a polycrystal whose Therefore we consider only shear viscosity here (for the bulk viscosity, see
microstructure is nearly isotropic. In such a case, at also Chapter 12).
Phenomenological theory of deformation 37

This is referred to as the Levy–von Mises equation. Therefore II ¼ 13ð1  3 Þ2 (note that only deviatoric
In this case (i.e., isotropic rheology), one can trans- stress contributes to plastic flow in the ductile regime).
late the results from one type of deformation geom- The flow law for the tri-axial test assumes the form,
etry to another deformation geometry. Note that
when large internal stress exists in addition to the "_ 1 ¼ "_ ¼ Cð1  3 Þn
applied stress, the BðII Þ term will be largely modi-
fied by the internal stress. Consequently, the plastic Comparing this relation with equation (3.5),
flow of a material will be affected by the presence of B ¼ ðC=2Þ3ðnþ1Þ=2 .
internal stress. An application of this notion to plas- Now for simple shear,
tic deformation during a phase transformation is 2 3
1
discussed in Chapter 15. 0 2_ 0
"_ ij ¼ 4 12_ 0 05
0 0 0
Problem 3.1*
and
Let us consider two cases. In uni-axial (tri-axial) 2 3
compression tests, one applies uni-axial stress, and 0  0
0ij ¼ 4 0 0 5:
measures uni-axial strain rate. Usually, the results of
0 0 0
such tests are presented as a relationship between
1  3 and "_ 1 as,
Therefore II ¼  2 and using (3.5), one gets
C ðnþ1Þ=2 n
1
2_ ¼ 2 3  . Hence D ¼ 3ðnþ1Þ=2 C.
"_ 1 ¼ Cð1  3 Þn (1)

and C and n are the constants that are determined by

experiments. Equation (3.5) can be rewritten using the equivalent
The second is a case of simple shear tests. In this case, stress (a scalar)
the results are usually presented as a relationship
between shear stress  and shear strain rate ,_ e  ð3II Þ1=2 (3:6)
n
_ ¼ D : (2)
to obtain
Show that,
"_ ij ¼ 3ðn1Þ=2  B  n1
e  ij (3:7)
D ¼ 3ðnþ1Þ=2 C: (3)
We can also define the equivalent strain rate (a scalar)
Solution  1=2
4
"_ e  II"_ (3:8)
Under the assumption of isotropic plasticity, the 3
rheological equation for both cases must satisfy
then
equation (3.5). Now, the strain-rate tensor for tri-
axial compression is
"_ e ¼ 2  3ðnþ1Þ=2  B  ne  (3:9)
2 3
"_ 0 0
"_ ij ¼ 4 0 2"_
1
0 5 These definitions, i.e., equations (3.6) and (3.8), are
0 0 12"_ made to recover the relations e ¼ 1  3 and "_ e ¼ "_ 1
for tri-axial compression.
and the deviatoric stress tensor is It is often useful to define effective viscosity, eff ,
2 3 2 3
1 0 0 P 0 0 by ij  2eff "_ ij , then
6 7 6 7
0ij ¼ 4 0 3 0 5  4 0 P 05 1
 eff ¼ 21 3ðn1Þ=2 B 1ne
0 0 3 0 0 P 1=n ð1nÞ=n
22 3 ¼ 21=n 3ðn1Þ=2n B "_ e : (3:10)
3 ð 1   3 Þ 0 0
6 7
¼4 0 13ð1  3 Þ 0 5: The effective viscosity decreases with the increase of
1
0 0 3ð1  3 Þ stress and increases with strain rate.
38 Deformation of Earth Materials

Plastic deformation dissipates energy. The energy pressure are attained). If the mechanical equation of
dissipation per unit volume due to plastic deformation state indeed exists, then one can predict the mechanical
is given by properties of a material including transient behavior
X without knowing its detailed history. H ART (1970)
¼ ij "_ ij  (3:11) showed that if a mechanical equation of state exists,
i;j
then y must be a unique function of strain, yð"Þ, and
Using the equations (3.6) and (3.7), one obtains should not explicitly depend on time (see also
1=n ðnþ1Þ=n M C C ARTNEY (1976)). Constitutive relations consistent
 ¼ B1 nþ1
e ¼ B "_ e : (3:12) with this requirement include
Note that the energy dissipation by plastic flow is a
 ¼ A"q "_ r þ 0 (3:14)
strong function of stress (or strain rate).
(e.g., H ART , 1970; P OIRIER , 1980) or
3.4. Constitutive relation for transient    s 
"
creep  ¼ B 1  exp  "_ r þ 0  (3:15)
"0
Transient creep is a phenomenon in which the rate of (e.g., C HINH et al., 2004; V OCE , 1948) (q, r, s > 0) which
deformation at a given stress (temperature and pres- have been used in the literature to describe strain
sure) changes with time (or strain). Since all of these (work) hardening behavior. Note that equation (3.14)
variables (stress, temperature, pressure) are kept con- does not describe steady-state flow (at infinite strain,
stant, this temporal variation of mechanical properties strength would become infinite which is physically
must be attributed to the evolution of internal structure meaningless), whereas (3.15) does.
(or internal mechanical state such as stress distribu- However, there is no reason to believe that a
tion). Therefore a general equation for transient creep mechanical equation of state exists for a range of
is given by materials or conditions, and in fact, in many cases,
 ¼ Fð";
_ T; P; yÞ (3:13) mechanical properties of materials indeed depend on
their history and hence a mechanical equation of state
where y is a parameter that defines the internal struc- does not exist. For example, when deformation occurs
ture or state of a material such as dislocation density or by grain-size sensitive creep and when grain size is
grain size, or the distribution of internal stress. As controlled by grain growth, then the strength of the
deformation proceeds, y evolves that leads to transient material depends on the entire history of that material
behavior. Annealing may also cause a change in y (the initial grain size and the temperature–pressure
without a change in strain that leads to time-dependent history). Similarly, deformation superposed with
(transient) behavior. The specific form of such a func- pre-existing deformation depends on the pre-existing
tion depends on the microscopic mechanisms of defor- conditions such as the initial stress (this is the case for
mation that are discussed in Chapters 8 and 9. In this post-glacial rebound, in which deformation occurs
section I will summarize some of the general aspects of due to stress caused by the variation of surface load
the formulation of time-dependent deformation. in addition to the pre-existing stress due to long-term
There have been some discussions as to how to convection, see Chapter 18).
formulate time-dependent deformation. The question In these cases, the transient creep behavior can be
is closely related to the issue of whether a mechanical described explicitly in terms of time variation of strain
equation of state exists or not (H ART , 1970; H OLLOMON , (at a constant stress). In many cases the experimental
1947; Z ENER and H OLLOMON , 1946). When the stress data can be fitted to the logarithmic constitutive equa-
required for plastic deformation depends on the instan- tion, namely,
taneous values of the strain, strain rate, temperature
"ðtÞ ¼ "ela þ B logð1 þ Atn Þ þ "_ s t ð05 n  1Þ (3:16)
and pressure and not on their past histories, it is said
that a mechanical equation of state exists (similar to where "ela is elastic strain A and B are constants, and "_ s is
the equation of state of material, by which the density the steady-state strain-rate. This equation is reduced to
of a given material is defined only by the state variables the following ‘‘parabolic’’ flow law for Atn  1,
such as temperature and pressure without any depend-
ence on the path through which the temperature and "ðtÞ ¼ "ela þ ABtn þ "_ s t  (3:17)
 Another Random Document on
Scribd Without Any Related Topics
The next step (F) consists in laying out the form of the propeller blade upon
all four sides and ends of the block, and Step G is the final one of cutting out
the propeller, scooping out its blades concave on one side, and carving them
convex on the opposite side. A very sharp knife must be used for cutting; and
the work must be done slowly and carefully, because the least slip is likely to
ruin the propeller. The entering-edge of each blade is the almost straight
edge, and should be cut very thin. The ends of the blades should also be cut
thin, while the hub should be cut away as much as can safely be done without
weakening the propeller.
When you have completed cutting the propellers, place them at their centers
across the edge of a knife-blade, and if they do not balance perfectly, locate
the trouble and correct it. Finish the work with fine emery-paper, and then
shellac it. Some boys glue silk over the ends of their propeller blades, for a
distance of ½ inch or so, to reinforce them and make them less likely to split.
The Propeller-shafts are made of heavy piano-wire, bent into a hook at one
end (Fig. 38) to receive the rubber strands of the motor, and cut of the right
length to extend through the hole in the bearing, through a glass bead,
through the propeller, and then to bend over the side of the hub (Figs. 37 and
38). By bending over the end of the shaft against the hub, it is held securely
in place.
The Motors consist of twelve strands of 1/8-inch flat rubber, each, and as
these are 1 yard in length, exactly 24 yards of rubber are required. The
rubber is not connected direct to the hooks on the bow and propeller-shafts,
as the wire would quickly cut through the strands. Instead, small rings are
bent out of wire, with pieces of small rubber-tubing slipped over the wire, and
the ends of the rubber strands are looped through these rings and bound in
place with thread (Fig. 39). The wire rings are then slipped on and off the
hooks quickly. As light and heat cause rubber to deteriorate, you must remove
the motors from the machine after use, pack away in a covered box, and keep
in a cool place, in order to get the longest life possible out of the rubber.
It has been found that rubber motors can be wound much farther by
lubricating them with glycerine. It is only necessary to put a few drops of the
glycerine upon a clean cloth, and rub it over the outside strands; then wind
the motors, and it will work over the surface of the inner strands until all parts
are covered.
 Fig. 45.—A Home-made Motor Winder.
Fig. 46.—The Kind of Egg-beater to Use.
Fig. 47.—How the Motors are Connected to Winder for Winding.

Of course the rubber motors must be twisted an equal number of turns, in

order to make the propellers work the same, and this is usually done with an
ingenious winder made from an egg-beater, which winds both motors
simultaneously.
The Home-made Motor-winder shown in Fig. 45 is made from a Dover
egg-beater (Fig. 46). To convert the egg-beater into a winder, it is necessary
to cut off the loop ends and the center pivot wires on which the loops turn.
Then bend the cut-off ends of the loops into hooks, and punch them to fit
over the pivot wire ends, as before (Fig. 45). The ends of the pivot wires must
be riveted to keep the hooks in position.
Figure 47 shows
How the Egg-beater Winds the Motors. While an assistant supports the
model by the propeller end, you remove the motor rings from the hooks on
the bow of the fuselage, and slip them on to the hooks of the egg-beater.
Then you turn the crank of the winder, counting the turns as you do so, and
when you have wound the motors as far as you wish, slip off the motor rings,
and slip them back on to the bow hooks of the model aeroplane. Motors of
models like that shown in this chapter are wound one-thousand turns or more
for each flight.
Wind the Motors Slowly, especially after the first row of knots begin, as it
puts the rubber to the least amount of strain by doing this. Quick winding not
only strains the rubber but makes the knots form in bunches, and uneven
winding, of course, produces an uneven unwinding.
The propellers must be held after the motors have been wound, to keep them
in check. Figure 34 shows
The Position to Take for Launching a Model from the hand. The machine
should not be thrown forward, as the movement would cause too great a
disturbance of the air, resulting in the machine losing its stability, and
probably upsetting. The best method is to give the model a slight push that
will start it off at a speed a trifle under that produced by its propellers.
 CHAPTER IV
A HOME-MADE TOY MOTOR-BOAT
The toy motor-boat shown in Figs. 48 and 49 is propelled by a tin propeller
run by a rubber-band motor. A handful of rubber-bands will cost only a few
cents, and the rest of the working material can be picked up at home.

Fig. 49.—The Completed Motor-boat.

Prepare the Bottom of the Hull out of
a piece of wood 1 inch thick, making it of
the shape and dimensions shown in Fig.
51. Be careful to curve the side edges the
same. Use a saw for cutting out the piece,
then smooth up the edges with a plane
and sandpaper. The stern should be
sawed off on a bevel as shown in Fig. 52.

Fig. 50.—Stern, with Motor in Place.

The Sides of the hull (B, Figs. 52

and 53) are thin strips 2½ inches
Fig. 51.—Diagram of Hull. wide. Nail one to one edge of the
bottom block, then saw off the bow
end on a line with the bow of the
bottom block, and the stern end on the same slant as the bevel cut on the
stern of the bottom block. With one piece in position, nail on the second side
and trim off its ends. If you have any difficulty in making a neat joint between
the bow ends of sides B, take a piece of tin from a can, bend it around the
bow, and tack it in place as shown in Fig. 48. The stern piece (C, Figs. 53 and
54) should be cut next, to fit the slanted ends of the sides.
The Deck (D) extends from
the bow almost to the center
of the boat. Its top surface
should taper in its length and
curve from side to side. The
piece may be whittled or
planed to this shape. Fasten it
with brads to the top edges of
the sides of the boat.

Figs. 52 and 53.—How the Hull, Sides, Stern and

Deck Pieces are Assembled.
 Fig. 48.—Launching the Toy Motor-boat.

To Complete the Boat, go over the work carefully, trim off all projecting
edges, drive nail heads beneath the surfaces, putty nail holes and cracks, and
give the wood two coats of paint of whatever color you want to have the
motor-boat.
The Propeller (E, Fig. 54) is cut from the side of a tin can. Cut a piece 3
inches long and ¾ inch wide, round its ends, and with the point of a nail
pierce a hole through it each side of the center of the length of the piece (Fig.
55). To finish the propeller, it is only necessary to take hold of the two ends
and twist the piece into the shape shown in Fig. 56.
The Propeller-shaft requires a short piece of wire with one end bent into a
hook (F, Fig. 56). Stick the straight end of this shaft through one hole in the
propeller, and the hooked end through the other hole, then twist the hooked
end over on to the main part of the shaft, as shown in Fig. 57. Make a tight
twist so the propeller will be held perfectly rigid on the shaft.
The Bearing Plate G (Figs. 54 and 58) supports the propeller. Cut it out of a
piece of tin 1½ inches wide by 3 inches long, bend it in half crosswise to give
it stiffness, and then bend it lengthwise to the angle shown so it will fit over
the slanted stern of the boat. Punch two holes through the upper end for
nailing the plate to the stern, and a hole at the lower end for the propeller-
shaft to run through.
For a Thrust Bearing, slip a couple of beads over the propeller-shaft,
between the propeller and bearing plate G. Probably you can find glass beads
in your mother's button bag.
 Fig. 54.—Longitudinal Section of Assembled Motor-boat.
Figs. 55-59.—Details of Propeller.
Fig. 60.—Rubber-band Motor.

After slipping the beads on to the shaft, and sticking the shaft through the
hole in bearing plate G, bend the end of the shaft into a hook; then screw a
small screw-hook into the bottom of the hull, at the bow end (I, Fig. 54), and
you will be ready for
The Rubber-band Motor. Rubber-bands about 1½ inches in length are best
for the purpose. Loop these together end to end (Fig. 60) to form a strand
that will reach from hook I to the hook on the propeller-shaft; then form three
more strands of this same length, and slip the end loops of all four strands
over the hooks.
To Wind the Motor, give the propeller about one hundred turns with your
finger; then, keep hold of the propeller until you launch the boat.
There are many ways of elaborating upon the design and construction of this
toy motor-boat, but, having given the necessary instructions for building a
simple model, I am going to leave further development for you to work out.
Here is an opportunity for you to use your ingenuity. Devise an adjustable
rudder, add a keel, finish off the cockpit with a coaming, install a headlight
made from a pocket flashlight—in fact, see just how complete a motor-boat
model you can build.
 CHAPTER V
HOME-MADE TOY WATER-MOTORS
You can own a water-motor like the one shown in Fig. 61, because its
construction requires nothing but easily obtained materials.

Fig. 61.—A Varnish-can Water-motor in Operation.

The Case of this water-motor is made of an empty varnish can—preferably
one of gallon capacity. Nothing better could be desired. The tin can makes a
lightweight compact case; the spout in the top is in just the right place and of
the right size to receive the water power from a faucet; and as the water
connections can be made tight there is no possibility of water splashing on to
the floor—a big argument in your favor when seeking permission to use the
motor in the bath-tub, wash-basin, or kitchen sink.
You can get an empty varnish can from
any painter, or at a paint store. The first
step in converting the can into the motor
case consists in removing the bottom. You
will find this soldered in place, in all
probability, and it can be removed quickly
by holding the can over the flame of a
gas burner until the solder melts, when a
few taps upon the edges will cause the
piece of tin to drop off.
The Water-motor Wheel is shown in
the cross-sections of the water-motor
(Figs. 63 and 64), and Figs. 65 to 67
show its details. The diameter of the
wheel should be about ½ inch less than
the inside width of the can. In the model
from which the drawings were made, this
measurement is 5½ inches. Cut the two
side pieces of the wheel out of a piece of Fig. 62.—The Completed Varnish-
cigar-box wood, and bore a ¼-inch hole can Water-motor.
through the center of each for the wheel
axle. Fasten a spool to the center of one side piece for a pulley-wheel (Fig.
66).
 Figs. 63 and 64.—Sections through Water-motor Case.

Prepare Eight Paddles 1¾ inches wide and 2½ inches long, out of cigar-
box wood. Locate the positions for the ends of the paddles, upon the side
pieces, by drawing a horizontal line, a vertical line, and two diagonal lines at
angles of 45 degrees, through their centers. This will simplify the matter of
spacing the paddles equidistant from one another (Fig. 67). Use brads for
fastening the side pieces to the paddle ends. Those removed from the cigar
boxes will do.
The Wheel Shaft should be a trifle shorter than the inside width of the can,
and enough smaller than the ¼-inch hole in the wheel side pieces so the
wheel will turn freely. Locate the centers for the axle upon the two sides of
the can, in the proper position so there will be the same margin above and at
the ends of the wheel. Drive a nail through each side of the can into the axle
end.
 Fig. 65.—The Completed Water-motor Wheel.
Figs. 66 and 67.—Details of Water-motor Wheel.

An Outlet for the water after it has passed over the wheel paddles must be
provided, and the best way is to fasten a strip to two opposite sides of the
can so as to raise the bottom about an inch, as shown in Figs. 62, 63, and 64.
For a Pulley-belt use a piece of heavy cord. Cut a slot through the front of
the can for the belt to run through, and make this slot large enough so the
cord will not rub against the sides (Fig. 63).
Pulley-wheels for attaining different speeds can be made of spools of
various sizes. A bicycle wheel with the tire removed, mounted in a frame, is
excellent for a large wheel.
Connecting up the Water-motor. If
you operate the water-motor in the
kitchen sink, you can either build a
platform as shown in Fig. 61, to bring the
spout of the varnish-can case up to the
level of the faucet, or you can set the
water-motor in the sink and lead a piece
of rubber tubing from the spout to the
faucet, as shown in Fig. 68. If you use the
latter arrangement, slip the lower end of
the rubber tubing over a short piece of
glass, brass, or tin tubing, and stick the
short tubing through a hole in a cork
large enough to fit the spout of the
Figs. 68-69.—How to Make a Water-
tight Connection between Faucet varnish-can case (Fig. 69). If you raise
and Water-motor. the water-motor high enough so the
faucet will set down into the spout, you
can cut a large enough hole for the faucet, through a cork, and then fit the
cork in the spout as shown in Fig. 64.
Another Water-motor. The little water-motor in Fig. 70 will furnish
sufficient power to operate simple mechanical toys.
The Water-motor Wheel. Procure two baking-powder can covers for the
ends of the water-motor wheel (A, Fig. 72), a cigar-box out of which to make
the wheel paddles, and a stick ¼ inch square and 5 inches long for the wheel
axle (B, Fig. 72).
Cut eight paddles from the
cigar-box wood 1 inch wide
and 5 inches long. Take a pair
of these strips and fasten
them to one can cover, in line
with each other, and close
against the sides of the cover
(C, Fig. 73). Fasten with tacks
or brads driven through the
cover into the ends of the
strips. Take another pair of
strips and fasten them to the
same cover, in a similar
manner, at right angles to pair
C (D, Fig. 72). Then tack the
pairs of strips E and F to the
cover halfway between pairs C
and D. With the paddles in
position, locate the exact Fig. 70.—A Small Water-motor that can be
Operated in a Wash-Basin.
center of the end of the can
cover, and drive a nail through
at this point into the end of axle B. Slip the free ends of the paddles into the
other can cover, and carefully drive tacks or brads through the cover into
them. Drive a nail through the center of the cover into the end of axle B.
The Wheel Supports. Figure 74 shows the supports for the wheel. Cut the
end pieces G 4 inches wide and 6 inches high, and the cross strips H 1¾
inches wide and 5½ inches long. Nail pieces G to H, as shown, allowing the
lower ends of G to extend ½ inch below strips H, and leaving a space of ½
inch between strips H. The axle holes in pieces G (Fig. 74) should be located
in the center of the width of these pieces, and halfway between their tops and
strips H. Bore the holes with a
gimlet, or make them by
driving a large nail through the
pieces, and then withdrawing
it.
To Mount the Wheel upon
the supports, withdraw the
nails driven into the ends of
axle B, slip the wheel between
uprights G, and drive the nails
through the holes in G back
into the holes in the axle ends
(Fig. 71).
The Pulley Wheel. One can
cover should be converted into
a pulley by winding several
turns of string around it, near
each edge, leaving a groove
between the string. Coat the Fig. 71.—The Water-motor Wheel.
string with glue to make it
stick fast to the cover.
The Water-motor Case.
Figure 70 shows how the
water-motor case is
constructed by fastening
boards N, I, J, K, L, and M to
the wheel supports G. There
must be a slot through I and
another through J, for the
string belt to pass through,
and a hole through K for the
intake of water from a faucet. These can be cut out of the edges of the
boards, as shown, before they are nailed in place. Leave an opening between
boards N and M, and the bottom of ends G, for an outlet for waste water.
Figs. 72 and 73.—Details of Water-motor Wheel.

Fig. 74.—Support for Water-motor Wheel.

 CHAPTER VI
A HOME-MADE TOY RAILWAY
It is often thought that a toy railway is beyond a boy's
ingenuity to construct, whereas, in reality, it is one of the
simplest toys he can make. This applies to the tracks,
stations, and cars of every description, all of which can be
made with a few strips of wood, some spools, nails,
cardboard, and a bottle of glue, for materials. If you have
passed the age of caring for such toys as this, you will, no
doubt, enjoy the making of one for your younger brother,
or for one of your boy relatives.
Figure 76 shows a railway set up and in running order. As
shown in the illustration,

Fig. 75.—Upright.

Fig. 76.—The Toy Railway in Operation.

The Trolley-line, or overhead cable, runs around the wheels of two
supports, one at either end of the track. Prepare four pieces of wood the
shape and size of that shown in Fig. 75 for the uprights of these supports,
and make two wheels three inches in diameter. The wheels may be marked
out with a home-made compass—a pencil tied to the end of a piece of string,
if you haven't a compass. When the wheels have been cut out, place them in
 your bench-vise, one at a time, and with
a file make a groove around the edge as
shown at C, Fig. 77. Bore a three-eighths-
inch hole through each upright at F, Fig.
75, and another through the center of
each wheel. Now fasten two of the
uprights six inches apart upon a block of
wood, as shown at A and B, Fig. 77.
Whittle a shaft to fit loosely in the holes
of the uprights, and, after slipping it into
them, fasten one of the wheels upon one
Fig. 77.—Support for Trolley-line. end and a small spool upon the other
(see C and D in Fig. 77). A weight of
some sort should be fastened to the base,
as shown at E. The uprights for the other support should be similarly mounted
upon another block of wood. Fasten the remaining wheel to an axle run
through the holes in the uprights, and, as it is unnecessary to have a spool
upon the other end of the axle, cut it off short and drive a nail through it to
prevent it from slipping through the holes. Having thus prepared the supports,
place them as far apart as you wish to extend the railway, and run a cord
around the two wheels and tie it. Then set the supports a little farther apart,
if necessary, to tighten the cord. Run another cord from spool D to

Fig. 76.—The Toy Railway in Operation (2).

A Water-motor, steam engine, or whatever power you can get with which
to operate the railway. A bicycle inverted with the tire removed from its rear
wheel has been used satisfactorily, as has also a sewing-machine with the belt
slipped off and the cord from the spool put in its place.
A good substitute for the tin tracks ordinarily sold in shops for toy railways will
be found in those shown in Fig. 78. These
 Fig. 78.—The Tracks.

Tracks consist of quarter-inch strips mounted upon pieces of cardboard.

Make a small gimlet-hole in one end of each stick, and drive a short finishing
nail in the opposite end (see Fig. 78). Cut the cardboard strips the length of
the sticks, and tack them to the sticks as shown in the illustration. If inch and
one-half spools are used for the car wheels, the inside gauge of the tracks
should be an inch and three-quarters. By lapping the cardboard strips over
the ends of the sticks, and the sticks over the ends of the cardboard strips,
and placing the nail dowels in the ends of the sticks as in the drawing, a
strong track is formed when the pieces are fitted together. This may be
extended to any desired length by adding more sections to it.

Fig. 79.—A Top View of Car Truck.

The Cars for this railway will have their trucks constructed alike, and it is a
simple matter to transform a car from one style into another. Figure 79 shows
a top view of a truck. For the bed of this cut a three-eighths-inch board
twelve inches long by two and one-quarter inches wide, and, after rounding
the ends as shown in the drawing, cut a mortise at A and B two and three-
eighths inches from either end.
Procure two one and one-half inch spools for wheels,
and drive a wooden peg through the hole in each,
cutting off the ends so they project a little beyond
the hole, as shown in Fig. 80. Then bore four holes in
the edges of the truck-bed with a gimlet at C, D, E,
and F (see Fig. 79), and, after setting the spools in
mortises A and B, pivot them in place with small
finishing nails driven into the wooden pegs. These
nails should fit loosely in the gimlet holes. In order to
drive them into the exact centers of the spools, it is
best to locate these points upon the ends of the pegs
before placing the spools in the frame. A quarter-inch Fig. 80.—Spool Wheels.
hole should be bored in the top of the truck-bed at G
and H (Fig. 79) in which to fasten the two uprights I
and J (see Fig. 81). Make the uprights four inches long and whittle a peg
upon the lower ends to fit holes G and H (see Fig. 82). Bore a hole with a
gimlet in the top of each and run a piece of heavy wire from one to the other,
bending it as shown in Fig. 81. Fasten K between I and J, as shown. Place a
small brass ring upon the wire before you fasten it in place. A small hook
should be screwed into one end of the truck and a screw-eye into the other
end, for couplings, should you wish to hitch two or more cars together.

Fig. 81.—The Completed Car Truck.

Fig. 83.—A Gondola Car.

A Gondola Car, such as shown in Fig. 83, should have its truck made similar
to Fig. 79, with the exception that it should be two inches shorter, in order
that cigar-box strips can be used for the side pieces. Cut the strips an inch
and one-half high and fasten them to the bed of the car with brads. This car
may be used as a trailer.
The car shown in Fig. 81 is a rather crude affair, but with a little more work
may be transformed into a better looking car—
 A
Stree
t Car
such
as is
show
n in
Figs.
84
Fig. 84.mdash;Side View. and
85
being
an example of what can be made. The sides, ends,
and roof of this car are made of cardboard, the
patterns for the cutting of which are shown on page
55. Figure 86 shows a cross-section taken through Fig. 85.—End View.
the center of the car. The two side pieces A should
be prepared first, as shown in Fig. 87. With a ruler
and lead-pencil draw in the windows about as shown in the drawing, using
double lines to indicate the sash. Then, with a sharp knife, cut out the center
of each just inside of the inner line. These windows may be left open or may
be covered on the inside with tissue-paper. If tissue-paper is used, oil it to
make it more transparent. When the two sides have been prepared, bend
each along the dotted lines (see Fig. 87) and tack one to each side of your car
truck as shown in Fig. 86. When properly bent, the distance between the
upper part of the sides should be two and three-quarters inches. Cut the two
inner ends of the car the shape of Fig. 88, using a compass with a radius of
two and one-half inches with which to describe the curve at the top. Draw in
the panels and sash lines as you did those upon the side pieces, being careful
to get them on the same level, and cut out the door and window openings.
Fasten these end pieces between the sides with glue, and also tack them to
the uprights of the car (I and J, Fig. 81), which will come just inside of them.
The roof is made in two sections (B and C, Fig. 86). For B cut a piece of
cardboard twelve and one-quarter by three and three-quarter inches (Fig. 89),
draw the curved end with a compass, using the radius shown on the drawing,
and slit the corners as indicated by the dotted lines. When this piece has thus
been prepared, remove the wire from the top of the truck (see Fig. 81). Bend
the cardboard over the sides and ends of the car, and lap corners D and E
over F and G, and H and I over J and K, tacking them with thread to hold
them in place. To fasten this part of the roof to the top of the car, cut a
number of small strips of linen, and glue them to the under side of the roof
and to the inside face of the sides and ends of the car (see Fig. 86). The
upper portion of the roof C should be made out of a piece of cardboard bent
into the shape of Fig. 90, and cut at the ends so the upper portion of C
projects a little beyond its sides. Draw the ventilation lights upon the sides of
C as shown on the drawings, and then fasten the piece upon the top of B with
strips of linen in the same manner as you fastened B in place. C should now
have the same curve to its top as B. Cut and glue a piece of cardboard in
each end of C to complete the roof. The shape of this piece is shown in Fig.
91. The outer ends of the car should be made as shown in Fig. 92, and tacked
around the ends of the wooden truck platform, and also fastened to the under
side of the roof with strips of linen.
 Figs. 86-94.—Details of Toy Street Car.

The window openings may be cut in each end, but it will make a stronger car
if they are simply drawn upon it. Cut four cardboard steps similar to Fig. 93
and tack them to the sides of the front and rear platforms. When the car has
been put together, replace the wire in the tops of uprights I and J (Fig. 81),
running the ends through the roof (see Fig. 84). Paint the sides and ends of
the car yellow with brown trimmings, and paint the roof a light gray. Water
colors can be used for the purpose. Letter the name of your car-line upon the
sides and the number of the car upon each end and side. The route should be
lettered upon strips of cardboard with pins run through them as shown in Fig.
94, these strips to stick in the roof of the car (see Figs. 84 and 85).
Having seen how the car is made, you will find it a simple matter to make
designs for
Other Cars, using the same scheme for the trucks, and altering the patterns
for the sides, ends, and roof, to suit the design.
Nothing has, as yet, been said about the
Operation of the Railway, and though Fig. 76 probably shows sufficiently
clear how it is run, a few words may be helpful. The car or cars are placed
between the wooden tracks, and the trolley (or cord attached to the ring on
top of the car) is tied to the trolley-line as in the illustration. Upon starting
your engine, water-motor, or whatever motive-power you have, the car will
run from one end of the track to the other. When it has reached the support
of the trolley-line, it will stop long enough for the cord trolley to pass around
the wooden wheel, and then run in the opposite direction until the other
support is reached. It will thus be seen that the trolley hangs to the upper
part of the cable, or trolley-line, in running one way, and to the lower part on
the return run. In changing the direction of the run, the ring to which the
trolley is attached slides to the other end of the car.

Fig. 95.—The Railway Depot.

A Station such as is illustrated in Fig. 95 is made out of cardboard and
mounted upon a seven-eighths-inch board large enough to form a railway
platform. After cutting out the side and end pieces, with door and window
openings placed as shown in the illustration, fasten them together with strips
of linen glued in the corners. Make the roof low and extend it over the
platform upon each side and over the gable-ends, as shown in the illustration.
Paint the sides of the depot the regulation depot red, and the roof a shingle or
slate color. Paint the door and window-sash black, letter the name of the
station upon the gable-ends, and with a ruler and lead-pencil rule off the
boards upon the sides, and the slate or shingles upon the roof. As this is a
typical railway station, two may be made of the same pattern, one for either
end of your car line.
 CHAPTER VII
HOME-MADE TOY ELEVATORS
The elevator shown in Fig. 96 is a unique mechanical toy well worth one's
making. Release the little car at the top floor, and it will descend to the
ground floor, and then return to the starting point, without you having to
touch it a second time. A magical elevator? Perhaps so. A little mechanical
device performs the trick.
The same plan may be followed for installing the doll-house elevator in
Chapter XIII, but the more stories there are the more fun there is in operating
the elevator. This is why I have adapted the scheme to
A Toy Office Building. Six stories are shown in Fig. 96, but you can make a
modern sky-scraper with as many stories as you like. A packing-case 3 feet 6
inches long, stood on end, was used for the model. Another box or two can
be added to the top for additional stories. Besides the box, or boxes, get
enough box boards for floors and partitions.
Make the Floors in two pieces (A and B, Fig. 98), so the opening for the
elevator shaft can be cut out of the end of one piece in the manner shown.
This opening should be about 5 inches square. Mark out and cut the boards
for all of the floors at one time, and be careful to get the shaft opening the
same in each floor. Cut the notch C in board A about 1 inch square.
Fasten the floor boards in place with nails driven through the sides of the box.
The Partitions, a pattern for which is shown in Fig. 99, can be made quicker
by omitting the doorway, but this is easy to cut by sawing along the sides and
then splitting out the piece between the saw cuts.
The Elevator Car should be built up of cigar-box wood, as shown in Figs.
101 and 102. The front portion (D) should be about 3 inches wide, 2½ inches
deep, and 4 inches high, and the rear portion (E) should be of the same
width, 2 inches deep, and 2½ inches high. Fasten these upon the base piece
F as shown.
The Elevator Guides. Bore the holes G through the top and bottom of the
car, close to the sides, for guide wires H to run through (Figs. 101 and 102).
These holes may be bored with a screw-eye if you haven't a gimlet or drill.
 Bell-wire, or almost any wire that
you have on hand, will do for the
guides. Fasten two screw-eyes
into the under side of the top of
the shaft, the same distance
apart as holes G, and in the
proper position so they will come
exactly over them (I, Fig. 100).
Use the car for determining these
measurements. Then bore two
holes through the bottom of the
shaft directly below the screw-
eyes (J, Fig. 100). Attach the wire
to one screw-eye, run it down
through holes G in the car,
through one of the holes J, then
across to and up through the
other hole J, up through the other
set of holes G in the car, and
attach to the second screw-eye I.
The Cables. The elevator is lifted
by means of cord L (Figs. 97 and
101). Fasten this cord to a tack
driven into the top of the car,
then run it up and over spool M
(Figs. 97 and 101), over spool N
(Fig. 97), and tie to weight K.
The Counter-balance. A bottle,
filled with sand to make it weigh
more than twice as much as the
Fig. 96.—A Toy Office Building with Elevator.
car, should be used for this.
Screw a small screw-eye into the
cork to tie the cord to.
The counter-balance runs up and down in
The Smoke-stack, which is fastened to the back of the building (Fig. 97).
Make the stack of cardboard mailing-tubes, joining them end to end with
bands of paper pasted around them. Fasten the stack to the back of the
building with wire straps, and brace the top as shown in Fig. 96, but leave it
unattached until you have adjusted
 The Overhead Pulleys, or sheaves. These
are spools. You will see by looking at Fig. 100
that spool M turns on the axle O, and the ends
of this axle are cut to fit snugly in screw-eyes I.
Fasten pulley spool N in the smoke-stack by
means of a wooden axle pushed through holes
pierced in the side of the stack, as is shown in
the small drawing above, Fig. 97. Bore a hole
through the back of the building for the cable
cord L to run through (P, Figs. 97 and 100),
and cut another through the smoke-stack.
How the Car Operates. When the weight
and cord have been adjusted and the smoke-
stack erected, the elevator will run from the
ground floor up to the roof of its own accord,
because the counter-balance is much heavier
than the car. To make it descend it is necessary
to add weight to the car, to make it enough
heavier than the counter-balance so it will drop
of its own accord. This is done with
Ballast consisting of a bottle of sand or salt of
twice the combined weight of counter-balance
K and the car. After filling the bottle, cork it up,
and screw a screw-eye into the cork. Then
screw the eye of a 2-inch hook-and-eye into
the roof of the building, directly over the center
of box E of the elevator (R, Figs. 97 and 101),
and attach one end of a rubber-band to the
hook and tack the other end to the top of the
Fig. 97.—Section through elevator-shaft (Fig. 101).
Elevator Shaft.
With the hook and rubber-band properly
adjusted, this is what happens when the car
ascends to the top of the shaft. The bottom of the rear portion of the car
strikes bottle Q, lifts it enough to release the end of the hook (R), and the
rubber-band springs the hook out of the way (Fig. 97). The bottle remains
upon the rear portion of the car, and its weight carries the car to the bottom
of the shaft.
Fig. 98.—Floors.

Fig. 99.—
Partitions.
 Fig. 100.—Front View of Elevator Shaft.
Figs. 101 and 102.—Elevator Car Details.

To Make the Car Rise to the top of the shaft again, remove bottle Q.
Replace the bottle upon the end of hook R, and it will be in position for the
next trip downwards.
Cut the holes Y and Z (Fig. 100) through the outside wall of the shaft for hand
holes through which to reach bottle Q and hook R.
Figures 97, 100, and 103 show
 A Simple Control for stopping the car at the
different floor levels. Stick S may be a piece of
broom-handle, curtain-pole, or flagstaff. Bore a
hole through the bottom of the shaft, directly
below holes C in the floors (T, Fig. 100), and
slip the stick through hole T and into slots C.
Then locate on one side of stick S points just
below the under face of each floor, and upon
the opposite side locate points just above
where the back edge of the elevator will come
when the car is raised to each floor level (Fig.
97). Remove the stick, and drill or bore a small
hole at each point marked; then replace it, nail
a small block (U, Fig. 100) across the top end
to hold it in place, and drive a nail, with its
head filed off, into each of the holes.
When stick S is turned to the position shown in
Fig. 97, while the car is going down, the first
nail below the car will project beneath it and
bring it to a stop; and if the stick is turned in
the opposite direction while the elevator is
going up, the first nail above the car will project
over the back edge of portion E and bring the
car to a stop.
Two Levers operate the brakes (W, Figs. 97
Fig. 103.—Detail of Brake and
Controlling Levers. and 100). Cut these of the shape shown in Fig.
103, and screw one to each side wall. Then
tack a piece of cord to stick S, wrap the ends of
the cord once around, slip them through screw-eyes V screwed into the side
walls, and tie to tacks driven into levers W.
One series of brakes can now be set by pulling forward one lever, and the
other series by pulling forward the other lever. By driving a nail into stick S at
X (Fig. 100), and a nail into the bottom of the shaft, each side of stick S, the
levers will turn the stick just far enough in either direction to bring the brakes
into operation.
If there is a kitchen porch to your house, construct
The Outdoor Elevator shown in Fig. 104 to run from the ground up to that
porch. If you live in an upper story of an apartment building, your elevator
 can be made to run to a greater
height, which, of course, will make
more fun.
It will save considerable work to
use the porch, because for one
thing you will not have to build an
upper platform to stand upon to
reach the elevator car when it runs
to the top, and for another thing
the supports for the guides and
cable can be fastened directly to
one of the porch posts.
Figure 105 shows a large detail of
The Guide Supports. Cross strips
A, B, and C should be 18 or 20
inches long, about 2 inches wide,
and 1 inch thick. At a distance of
about 1 inch from one end of strips
A and B screw a screw-eye into one
edge, and 8 inches from those eyes
screw a second screw-eye (D, Fig.
105). Screw-eyes with ½ inch eyes
are large enough. A dozen will cost
about 5 cents at the hardward
store. The elevator guides are
fastened to these.
Besides the screw-eyes there must
be two clothes-line pulleys for the
cable to run over. These cost 5
Fig. 104.—An Outdoor Elevator. cents apiece. Screw one pulley into
the edge of strip B, halfway
between the two screw-eyes D (E, Fig. 105), the other into an edge of strip C
at the same distance from the end that you have placed the pulley in strip B
(F, Fig. 105).
Nail strip A to the porch post as close to the ground as you can get it, strip B
to the same face of the same post, about 18 inches above the porch railing,
and strip C to the opposite face of the post at the same height as strip B. Nail
these strips securely in place.
If you cannot find a starch-box
or other small box out of
which to make
The Car, go to a grocery
store. You will be sure to find
just what you want there. It is
not likely that the grocer will
charge you anything for a
small box like this. If you have
placed screw-eyes D 8 inches
apart, as directed, the width of
the box should be a trifle less
than this measurement, but if
the box you pick up is wider
the screw-eyes can be spaced
as much farther apart as is
necessary to accommodate it.
Figure 106 shows how the box
is converted into the car.
Screw two screw-eyes into
each side of the box, one
above the other, as shown at
G, for the elevator guides to
run through, screw another Fig. 105.—Supports for Elevator Guides and
into the exact center of the Cables.
top (H) to tie the hoisting
cable to, and screw another
into the exact center of the bottom to tie the lowering cable to. Nail a narrow
strip across the open front of the car, at the bottom, to keep things from
falling out.
Get heavy wrapping-twine or stovepipe wire for
The Guides. Attach these guides to screw-eyes D in strip B, first, drop them
through screw-eyes G in the sides of the car, and then fasten to screw-eyes D
in strip A.
The Counter-balance is a large can filled with earth, sand, or small stones.
Its weight must be equal to about three times that of the empty car. Fasten
the lifting cable through holes punched in opposite sides of the can (Fig. 107).
Use a strong wrapping-twine for
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com