Rheology of Rocks

Brian Evans and David L. Kohlstedt

1. LABORATORY MEASUREMENTS producing accelerated deformation kinetics may be

mutually exclusive goals. This conflict can be mitigated by
1.1. Strategy and Techniques testing single phaserocks or by fabricating synthetic rocks
For a rock of given mineralogy and microstructure, the with speciallydesignedphasecompositions. The chemical
variables important in determining strength are pressure, fugacities of the mineral components and any volatiles,
temperature, strain, strain history, strain rate, pore fluid particularly water, must also be controlled. Unless the
pressure,grain size, fugacities of water and other volatiles, experimenter aims to investigate the properties of partial
and chemical activities of the mineral components. melts, eutectic melting needsto be avoided.
Although earth scientistsmay now duplicate pressuresand Deformation at low strain rates can also be studied by
temperaturesappropriate to the mantle and core in modern reducing a length scale.For mechanismsprimarily limited
high pressureapparatus,they still cannot study mechanical by diffusion, the most important length scale is the grain
properties under truly natural conditions. Time scales in size. Natural rocks with grain sizesof a few microns may
the Earth are too long, and length scalestoo large. be found, and synthetic aggregates of olivinc, calcite,
Sinceexact deformation conditions cannot be duplicated feldspars, and quartz have been produced in a range of
in the laboratory, the experimenter’sstrategy must involve very small grain sizes. Such synthetic rocks can bc usedto
determining the kinetic parameters of the appropriate understand the effect of variations of grain size, second
processesat laboratory conditions and extrapolating to phase abundance, or dissolved hydroxyl content on the
much lower strain rates [58]. Two convenient techniques physical properties of rocks.
are available to aid laboratory studies. Temperature and,
hence,kinetic rates may be increased,or processesmay be 1.2. Apparatus
studiedat smaller length scales[60]. Modem apparatus use a variety of loading schemes,
Testing at high temperatures also imposes constraints. including rotary shear, double block shear, conventional
For example,maintaining chemicaland phasestability and triaxial, full triaxial, diamond anvil, and large volume,
opposedmulti anvil devices [60, 901. The most common
loading geometry is the conventionaltriaxial configuration
B. Evans,Massachusetts Institute of Technology, Department (Figure 1). Such machines have cylindrical loading
ofEarth, Atmospheric and Planetary Sciences, Cambridge, MA symmetry with coaxial stress and strain. If the axial
02139-4307 compressive stress, oa, is larger than the confining
pressure,P,, then loading is triaxial compression;if o,<P,
D. L. Kohlstedt, University of Minnesota, Minneapolis,
Department of Geology and Geophysics, Minneapolis, MN
then the sampleextends. When pore fluids are present, the
55455 effective stressesmust be considered: the effective axial
stress CJ,’is a,-P, the effective pressure is P,‘=P,-P, By
Rock Physics and Phase Relations using forcing blocks with an inclined cut, non-coaxial
A Handbook of Physical Constants experiments can be carried out in plane strain or pure
AGU Reference Shelf 3 shear.Displacementsin this geometry are limited by jacket
Copyright 1995 by the American Geophysical Union. .48
 EVANS AND KOHLSTEDT 149

Conventional Triaxial Conventional Triaxial Confined Torsion graphite, tungsten, molybdenum, or platinum elements,
(Simple Shear) (Axial Compression)
0.
placed either inside or outside the pressure vessel. In the
a. 0.
solid medium assembly, samplesare about 10 mm long,
while those in the gas apparatusare about 25-50 mm. At
higher temperatures, the chemical fugacities of volatiles
are significant in determining the mechanicalproperties of
the matrix minerals and must be controlled by a buffer
system.

2. DEFORMATION MECHANISMS

In general, inelastic deformation occurs by defect

motion. Brittle failure results from crack growth or
frictional slip [48]; plastic flow results from motion of
point defects, dislocations, and twins, or from grain
Fig. 1. Schematics of several different loading boundary sliding [64]. To predict mechanicalbehavior the
configurations in conventional triaxial machines; below production, motion, and annihilation of strain-carrying
these are Mohr’s circles for each configuration. The defects must be related to parameters such as stress,
experimenter independentlycontrols the axial load, (J,, the temperature, fugacity, and pressure [63]. Plastic
pressureof the confining medium, P, and the pore fluid constitutive laws can be categorized by identifying the
pressure, P, Primed quantities are effective stresses,for rate-controlling steps of the dominant mechanism [18]:
example, a,‘=a,-P, In the conventional triaxial Nabarro-Herring and Coble creep are point defect
configuration, when on’ is greater than PC’the sample processes.Glide, climb, cross slip, and Harper-Dom creep
shortens along its axis; when 0: is less than PC’, the occur by dislocation motion. Twinning and grain boundary
sample extends. The simple shear configuration allows sliding involve planar defects. In pressure solution, ions
plane strain experiments to be done, but slip distancesare are transportedthrough a fluid.
limited. During confined torsion an additional shear stress
is applied; the principal stressesare no longer parallel to 3. DIFFUSION FLOW
the sampleaxis. This configuration allows very large shear
strains, but, thus far, only room temperature experiments Parts of the crust and mantle may deform by diffusion
havebeendone [90]. creep [5]. Rocks thought to be deformed in this regime are
composed of small, equant grains lacking dislocation
structures and lattice preferred orientation. Such
flexibility, and are typically a few millimeters. New rotary
microstructures are found in sheared olivine xenoliths,
shear devices overcome this limitation, enabling tests to
quartz mylonites, and calcite-dominated thrusts. Since
large strains.
minerals are composed of several ions, diffusion of the
The confining medium may be either a weak solid or a
true fluid [60, 901. Solid medium machines are capableof slowest ion along its fastest path (either grain matrix or
grain boundary) will determine creep rate. Theory and
higher pressuresand longer time spans,but, owing to the
experiments have identified three diffusion creep
strength of the solid medium, uncertainties in stress and
mechanisms.In the first, diffusion through matrix grains
pressure exist; but those decreaseas the strength of the
pressure medium decreases.Media used are AlSiMag controls creeprate; in the second,grain boundary diffusion
dominates;and in the third, reactions at the grain interface
ceramic, pyrophyllite, magnesium oxide, calcite, and salt
(listed in order of decreasing strength). The recent are governing.
introduction of liquid salt cells significantly improved Flow laws describing diffusion creep indicate strain
stress resolution [22]. Gas medium devicesalso have low rate is directly related to stress and inversely related to
uncertainty in stress; additionally, they provide grain size [18]. If matrix diffusion dominates (Nabarro-
independent control of pore fluid pressure. Whether Herring creep), the steadystate creeprate is
porosity is drained or undrained is an important
consideration for tectonic processes in the crust and . /A?\ l-4
lithosphere. &=] vy u
Heating is provided by a cylindrical furnace, made with 4 RT I d2
150 RHEOLOGYOF ROCKS

where CJis the differential stress, l’ is the molar volume, Despite the importance of defect chemistry for strength,
# is an effective matrix diffusion coeficicnt, d is the details of the point defect chemistry have been explored
averagegrain size, R is the gas constant, and T is absolute only for Mg-Fe olivine [53]. The majority defects for that
temperature. If grain boundary diffusion is rate-controlling mineral are metal vacancies, V”Mr and electron holes, h’,
(Coble creep), the creep rate is localized near the octahedral metal cations, Me,. Charge
neutrality requires that the mole fraction of holes, Nk, be
twice that of the metal vacancies, NV,, : Nh. = 2 NY,, .
(2) Metal vacancies, oxygen interstitia% and silic&
interstitials are formed in the following reactions:
where 6 is the grain boundary width, and DGB is the
boundary ditTusion coefficient. For grains to remain in + 0, (gl + MeSiO, (g) + 3MeLe =
(3)
contact, grain boundary sliding must also occur. Analysis 2Meke + Vie + Me,SiO,(g)
of diffusion creep accompaniedby boundary sliding yields
constitutive equations that differ from (2) only in the
G 0, (’ + 2Me$ = 0; + 2MeL6
numerical prefactor.

3.1. Point Defect Chemistry

The magnitudes of lattice and grain boundary diffusion VLe + MeSiO,(g) = Met + Sir’ + 30; (5)
constants in (1) and (2) are affected by the number and
mobility of the defects involving the slowest ion. Both The Gibbs phase rule requires that point defect
constantsfollow Arrhenius laws characteristic of thermally concentrationsin a four-component compound dependon
activated processes[64, 181.The diffusivity of a given ion three activities, P, and T. For fixed Mg-to-Fe ratio, P and
depends on the concentration of other defects as well. T, then point defect concentrationswill be determined by
Point defects in minerals are usually charged; the oxygen and orthopyroxene activities. Using the law of
consequently, the concentration of each point defect is mass action, combined with the charge neutrality
coupled to that of others by the condition that the crystal condition, the concentrations of Me vacancies, 0
be electrically neutral [76]. interstitials and Si interstitials are given by
In addition to intrinsic defects involving
stoichiometric components, extrinsic point defects (i.e.
impurities) may also exist. Becauseof charge coupling,
extrinsic defects can influence the number and mobility of
intrinsic defects and, hence, the creep rate. Chief among
the extrinsic lattice defects are those involving dissolved
hydrogen, hydroxyl, or water. One extremely important
conclusion of recent research is that diffusion flow is
strongly affectedby point defect chemistry [3 11.
In ternary or quatemary minerals, many point defectsare
possible, and special notations have been developed to
describethem. The most common, called the Kriiger-Vink where the K:s are functions of T and P only.
notation, indicates the charge (superscript), site Now, the diffusivity of a component ion, D, is related to
(subscript), and nature of the defect. For example, V “M. the concentration, Ni, and diffusivity, D,, of its associated
denotes a vacancy on the cation (metal) sublattice with point defectsby
double negative charge, relative to a normal occupiedsite.
The concentrationsof the point defectswill dependon the D=N,D, (9)
fugacities of other chemical speciesthrough point defect
reactions. In actual crystals, the concentrationsof one pair Thus, D is a function of the component activities, e.g., (6-
of oppositely charged defects, the majority point defects, 8). Defect concentrations are small, Ni < la3, and,
greatly exceed those of all others over a wide range of therefore, the ion diffusivity will be much smaller than that
thermodynamic conditions, so that the point defect of the point defects, Thus, point defect concentrationscan
chemistry can be described by only a few point defect adjust rapidly to changes in thermodynamic conditions,
reactions. while bulk composition changesmuch more slowly [51].
 EVANS AND KOHLSTEDT 151

TABLE 1. Di&sion Flow in Rocks

Sample A’ n m Q Remarks Source
(Wmol-‘)
Solnhofen 4.2-4.9 ’ - 2 or 214 Grain size <lO pm; boundary 77
Limestone 3 sliding ; equiaxedgrains; but some
flattening; Stressesbelow 100 MPa;
Grain growth during experiment.

Synthetic 1.7 1.9 190 Synthetic hot pressedsamples 92

Calcite Grain size sensitivecreepbelow 25
MPa; Grain sizes2-40 pm; Some
grain growth.

Synthetic 1.1 3.0 380 Synthetic hot pressedsamples; IO

Dunite (Dry) Grain sizesfrom 3-13 pm;

Porosityfrom 5-15%
Synthetic 1.5x10-3 1.4 3.4 250b 0.3 ml water added; 38
Dunite (wet) (Upa+) (mm)” Somegrain growth; Very small
amount of melt;

Synthetic 7.7x10-2 1.4 1.7 290” Somegrain growth; trace melt; 38

Dunite (dry) (MPa-“)(mm)”

Synthetic Diffusional flow identified by creep Tullis and Yund,

Albite microstructure pet-s. comm., 1993
“mol =moleof themineral(quartz,calcite,olivine,etc.)
bActivationenergyassumedto be 2/3 of that for volumediffusionof O/Si
‘Activationenergyassumedto besameasMg-Felatticediffusion

3.2. Diffusion Creep Experiments and extensivegrain boundary sliding.

The equationsfor Nabarro-Herring and Coble creep, (1) Limited data exist for diffusion creep in limestone,
and (2), may be generalizedto the form olivine, and feldspar (Table 1). At stressesbelow 100 MPa,
creep in Solnhofen limestone and fine-grained, synthetic
marbles is sensitive to grain size. Grain boundary sliding
contributes as much as two-thirds of the total strain [77,
where D’ is the diffusion coefficient for the s owest ion 921. In Solnhofen, dislocations were present, and grains
along its fastest path given by (9); m is eith r 2 or 3 were somewhatflattened after deformation. A quantitative
depending on whether diffusion is faster th ough the comparison of diffusion coefficients is not possible,
matrix or along the boundary; and n should ei ual 1. By because boundary diffusivities are not available for
measuringcreeprate as a function of grain size, stress,and limestone.
temperature, the parameters m, n, and the activation In fine-grained synthetic dunite, Coble creep dominates
energy of D’ may be determined. Comparison of the when water is added, (m=3.4); but Nabarro-Herring creep
experimental values with those expectedtheoretically may may prevail when the sampleis dry (m = 1.7) [38]. In both
be usedto identify diffusion flow. Substantialdeviationsof cases,values for n are greater than unity and may reflect
n from unity probably indicate that other processes,e.g. some contribution by dislocation flow. The effective
dislocation flow, are contributing to bulk deformation. diffusivity for dry creep agreeswith the value reported for
Microstructural evidence for diffusion creep includes the interdiesion of Mg and Fe in olivine, but is much larger
maintenanceof fine, equi-axedgrains, lack of dislocations, than values for self-diffusion of oxygen or silicon [35, 36,
152 RHEOLOGY OF ROCKS

741. This result suggests that flux of oxygen and silicon are
greater along boundaries than through the matrix, while
the opposite is true for octahedral cations.

4. DISLOCATION FLOW

Dislocations are linear defects characterized by two

vectors and a plane: the line direction, I, which is the
tangent to the defect at any point; the Burgers vector, b,
which is the closure failure of a circuit made around the
A dislocation; and the glide plane, which is represented by its
normal vector, n. The glide plane contains both I and b
(Figure 2). The Burgers vector, usually a low index
direction of the crystal structure, gives the displacement
induced by the passage of the dislocation. Dislocations
with common n and b comprise a slip system. For plastic
flow to occur homogeneously by dislocation glide alone,
five independent slip systems must operate (the Von Mises

=c
criterion); but if strain is heterogeneous, this restriction is

t
relaxed [57].
As rocks creep, dislocations are generated, migrate, and
are annihilated, the applied stresses doing work during all
three steps. In silicates, only small portions of the
dislocation, called jogs and kinks, are mobile at a given
instant. Dislocation motion in the glide plane occurs by
kink migration; while climb out of the plane occurs by jog
motion. The crystal structure offers intrinsic resistance to
the motion of both; for kink migration, the resistance is
Fig. 2a. Schematic drawing of an edge dislocation in a called lattice friction. Each dislocation segment also
simple cubic material. The edge of the extra half plane of interacts with the local stress field. When the force of that
atoms in the center of the crystal is the dislocation core. interaction, combined with thermal activation, is large
Fig. 26. In this schematic of an edge dislocation, the core enough, the segment migrates to cause strain.
region is drawn as a line. The normal to the glide plane is The net local stress is the superposition of all externally
n, the Burgers vector is shown as b, and the line direction applied stresses and any internal stresses arising from
is I. Kinks are abrupt changes in the line direction which other dislocations, point defects, precipitates, or interfaces.
lie in the glide plane. Jogs are abrupt changes in direction The internal stress, which typically resists migration, may
which take the dislocation out of the glide plane. The size increase with strain (hardening) or decrease with time
of the jog is greatly exaggerated here. Motion along the (recovery), depending on changes in the internal
glide plane is conservative, requires no atomic diffusion, microstructure. The change of the internal stress field, dot
and proceeds by the migration of kinks in the plane. may be written as
Dislocation climb occurs as jogs migrate and atoms diffuse
into or out of the extra half plane. Diffusion transport can do‘ = h’d&-r’dt (11)
be along the dislocation core or through the lattice. The
mobilities and numbers of jogs and kinks can be affected
by the concentration and mobility of point defects in the where h’ and r’ are the strain-hardening coefficient and the
recovery rate, respectively. When softening owing to
crystal.
recovery balances hardening, the steady state creep rate is
Fig. 2c. Screw dislocations can slip along any plane
cozonal with I. Since b is parallel to I for all screw .
E=L
dislocations, and because b lies in any plane that is (12)
cozonal with I, slip in any of those planes is conservative. h’ ’
 EVANS AND KOHLSTEDT 153

Alternatively and equivalently, the creep rate may be typical constitutive models are given in Table 2; the list is
written far from exhaustive.
Given accurate constitutive equations and appropriate
E =jwb materials properties, deformation maps may be constructed
(13)
which indicate the area of dominance of each mechanism
[ 181.The independentvariables of the map may be chosen
where p is the density of the mobile defects,b is the length as any two of the pertinent variables in (14), provided that
of the Burgers vector, and v is the velocity of the mobile the remaining parameters are fixed. The boundaries
defect [ 181. plotted on the map represent the conditions where the
strain rates of two mechanismsare equal.
4.1. Dislocation Flow Laws The mechanismsfall into several classes:(1) plasticity,
To formulate a constitutive law one must identify the controlled by glide, occurring at high stress or low
rate-limiting step in the sequenceof generation, motion, temperature; (2) creep, controlled by climb or cross slip,
and recovery of the pertinent defects. For example, occurring at high temperatures; or (3) Harper-Dorn creep,
dislocations may be generatedat boundaries, intersections occurring at very low stresses, or high temperatures.
with other dislocations, or precipitates. Motion may occur Changes in the flow law are also accompanied by
by climb, glide, or cross-slip. Recovery may occur by microstructural changes.For example, metals creeping at
cross-slip or climb allowing reactions with dislocationsof high temperature may be divided into two classes [85],
oppositeBurgers vector (recovery), or by interaction of a called pure metal type, where nG.5,and alloy type, where
dislocation with a static or moving boundary
nG3. Microstructural observations indicate that
(recrystallization).
Chemical activities can profoundly affect creep rate. In dislocations organize to form subgrains in the first type,
minerals, kinks and jogs may be charged. As with point but are scattered homogeneouslythroughout the crystal
defects, line defects may be surrounded by charge- even after large strains in the second.In metal types, creep
compensating point defects which are dragged along is presumably controlled by the dislocation multiplication
during dislocation motion. Thus, the mobility and number rate; while for alloy types, creep may be controlled by
of jogs and kinks probably dependson concentration and viscous migration. In metals, simple halides and oxides, it
mobility of certain point defects [3 11.Similarly, if recovery is often found that the activation energy for high-
occursby climb, which requires diffusion, the recoveryrate temperature creep is the sameas the activation energy for
will respondto difisivity changescausedby variations in self-diffusion of the slowestspecies.
chemical activity.
4.2. Dislocation Creep Experiments
The rate-limiting step for creep will, in general, be
thermally activated, but may change with variations of A common feature of many of the high-temperature
stress, CT,temperature, T, elements of the microstructure, constitutive laws is that creep rate is related to differential
S, or chemical activities, ai of all but one of the p stress,(T,by a power law of the form,
components[3, 311. In completely general form, the creep
rate is (15)

E= I(a,S,T, P,a ,,..., ai ,..., apml),exp(-g) (14)

where ai is the activity of the ith component, q,, n, and
Qz are constants,and AZ is a weak function of T. Often
where AG, the free energy of activation (often called Q), experimentalists simply adopt (15) and determine n
may be a function of stressor pressure. experimentally (Table 3) 18, 15, 411. At lower
Among the possible rate-limiting steps for creep are temperatures and higher stresses,creep is controlled by
glide migration, climb migration, recovery by climb, glide (glide or barrier models), leading to an exponential
recovery by cross slip, or dislocation generation. The dependence of creep rate on stress. Scant data are
deformation rate of the aggregate will be determined by available,but for olivine, this transition may occur at 200-
the rate-limiting step of the fastest mechanism. If 600 MF’a.
appropriate assumptions can be made as to which With more detailed examination, creep experiments in
multiplication, migration, or recovery process controls rocks show great complexity, belying naive trust in (15).
strain and which step in that processcontrols strain rate, Substantialdifferences may exist from one rock to another,
then theoretical constitutive laws may be developed.Some even when they are dominantly monomineralic (Figures 3
154 RHEOLOGY OF ROCKS

TABLE 2. Constitutive Laws for Dislocation Creep

Mechanism Constitutive Law Remarks Reference

Harper-Dorn Low stress,low dislocation 67

densities

Glide controlled; Dislocations

Viscous Glide dragging solute atmosphere. 85
High temperature, Intermediate
stress

High and Low Effective diffusion dominatedby 18

Temperature lattice diffusion (high temp.) or
pipe diffusion (low temp.)

Creep rate controlled by escape

Recovery- rate of pile ups 94
controlled High temperature,Intermediate
stress

Localized obstaclesrequire cross

Crossslip slip; high temperature, 63
Controlled intermediate stress

Glide Stress-activated; 95
controlled Low temperature; high stress

Barrier Obstacledominated; stress- 18

controlled activated;
Low temperature; high stress

and 4). For example, for Yule marble [28] and Carrara up in the range 440-540 kJ/mol. In
the range 3.3-4.5 and Qd,*
marble [78], the empirically determinedflow law constants both polycrystals and single crystals, strain rate depends
are, respectively, n=7.7 and 7.6 and Qx=255 and 420 on intrinsic variables including temperature, oxygen
kJ/mole (for deformation at strain rates of 10e3to low6s-’ fugacity and pyroxene activity, and on extrinsic dopants
and temperaturesof 500 to 1OOO’C).Thus, although there including water (or hydrogen) fugacity [3, 381 and,
is general agreement concerning the stress exponent, possibly, carbon activity [21]. Both single crystal and
activation energies for the two rocks differ significantly. polycrystalline olivine are weakened at high temperature
Further, cross slip might be important [12], which would by a factor of 2.5 when water is added. The diffusion rate
lead to an entirely different flow law (Table 2). for hydrogen is fast -- 10mgto lo-l2 mzsl at SOO-lOOO’C,
The largest data set currently availableis for olivine-rich suggesting an extremely mobile defect, e.g., interstitial
rocks (Figure 5). Coarse-grainedolivine undergoespower protons, charge-compensatedby electron holes 138,501.
law creep at stressesbelow 100 MPa with values of n in In olivine single crystals, no one dislocation climb model
 EVANS AND KOHLSTEDT 155

SvmbolsUsed in Table 2 plastic flow strength is influenced greatly by extrinsic

a Distance between Peierls hills
defects [59]. The strength of natural quartzite rocks
Cross sectional area of dislocation core
deformed with water present is intermediate betweenthat
of natural single crystals and synthetic single crystals [42]
Pre exponential factors, may be a
function of temperature, etc.
(Figure 4). At high temperatures, the stress and
temperature sensitivity of strain rate are the samefor both
b Length of Burgers vector
“wet” and “dry” experiments.
bo Width of unsplit dislocation
The kinetics of equilibration of extrinsic point defects
Concentration of solute in crystal matrix
=o are apparently quite slow, and the defect concentrations
D Constant for self-diffusion through lattice are quite low [59]. Differences in chemical environment
DC Constant for core diffusion and purity probably account for much of the scatter in
D so/ Diffusion coefficient of solute in crystal strength measurements. In nature and in experiments,
matrix water-related defects may be introduced through cracking,
Width of split dislocation a processwhich may be important for strain localization in
Activation energy of pertinent process shearzones [44].
Elastic misfit of solute ion Water may affect dislocation creep through changes in
Preexponential constant, possibly a function multiplication rates [40, 521changesin subgrain boundary
of temperature migration rates, or changes in climb recovery rates [59,
Stacking fault energy 871. Multiplication effects might explain the upper and
Boltzman’s constant
lower yield points often observed in synthetic single
Adjustable parameter crystals and natural polycrystals [40, 42, 521. Prismatic
Density of dislocation sources dislocationsare apparently generatedduring the relaxation
Shear modulus
of pressure in the tiny water clusters which are found in
weak crystals. Two mechanisms for climb enhancement
Exponent - often equal to 2
have been proposed; both assume changes in the
Freq. of vibration of dislocation segments
dislocationcore region and the general impurity level [59].
Constants which depend on barrier geometry Water acceleratesself-diffusion of oxygen in quartz [17],
and strength
but its effect on silicon transport is unknown. If hydrolytic
Pm Dislocation density, stress independent weakening involves local changesof the dislocation core,
cl Differential stress leading to acceleratedkink migration, for example, then
T Absolute temperature pipe diffusion rates may also be important [59].
2 Peierls stress (i.e. Lattice friction stress at
absolute zero)
i-2 Molecular volume 5. PRESSURE SOLUTION

Field observationsand microscopic examinationssuggest

that rocks at temperaturesand pressuresup to greenschist
facies are deformed by solution transport [831. An essential
element of the processis that deformation occurs at grain
or interphase boundaries, not in the grain interior. Such
pressure-solution surfaces may be distributed throughout
the rock on the scaleof millimeters or meters. The former
can account for the creep behavior over a wide set of distribution, called intergranular pressuresolution, results
conditions, nor is there a simple correspondencebetween in a fitted grain fabric, while the latter results in widely
any of the self-diffusion coefficients and that of creep rate. spaced features of localized strain, called stylolites or
Both single crystals and polycrystals typically have solution seams.
activation energiesfor creep that are larger than those for
self-diffusion. Thus, the data suggestthat self-diffusion is 5.1. Constitutive Equations
not the rate-limiting step [3] or that the jog concentration Although Gibbs [19] considereda solid in contact with
is undersaturated[321. fluid reservoirsof differing pressures,uncertainties remain
Unlike olivine, quartz has very low concentrations of in the complete thermodynamic description of a porous,
intrinsic point defects at relevant temperatures. Thus, its fluid-filled solid, loaded nonhydrostatically.Two classesof
156 RHEOLOGYOF ROCKS

TABLE 3a. Power Law Creep Constants: Calcite Rocks

Rock Name log A n e Remarks Source
(MPa-W) @AlOP)

Yule (T cyl.) -3.6 7.7 256 28,71

Yule (1cyl.) -3.9 8.3 260 11
Carrara 3.1 7.6 418 Creep Regime “2”; Dry 78
Carrara 8.0 4.2 427 Creep Regime “3”; Dry 78
Solnhofen 3.4 4.7 298 Flow regime “2”; Dry 77
Synthetic 2.00’ 3.3 190 Grain size dependent:Flow law 92
includesa factor d’.3

TABLE 3b. Power Law Creep Constants:Quartz Rocks (Dry)

Rock Name IogA n Remarks Source
(iLlPa)-“.+ @X2&)
Simpson 11.0 377 27
II -11.17 6.5 266 56
11 -10.10 5.7 243 26
Heavitree -2.90 2.0 168 cl-quartz; 79
-4.5 1.9 123 a-quartz 24
195 o-quartz; transient strains to 0.8% 69
51 /3-quartz;transient strains to 0.8% 69
Heavitree -5.0 2.4 163 a-quartz; NaCl medium 37
11 -3.507 2.3 171 calcite medium 37
11 -5.463 2.8 184 NaCl medium; vacuum dried 37
Heavitree -5.30 2.9 170 43
II -5.398 4.0 300 vacuum dried 43
Simpson -5.936 2.7 134 cx-quartz; Pp=l.O-1.25 42

TABLE 3c. Power Law Creep Constants: Quartz Rocks (Wet)

Rock Name IogA n e Remarks Source
(upa)-+ (kJtnol-1)
Canyon -1.35 2.6 230 56
Creek
Simpson -1.7 1.8 167 a-quartz; 0.4% water added 24
Heavitree -5.66 2.7 120 a-quartz; 0.4% water added 43
Heavitree -2.28 1.4 146 a-quartz; 0.28% water added 37
8, -2.54 1.8 151 0.39% water added 37
Simpson -5.94 2.7 134 a-quartz; water from medium; 42
Pc=l.O-1.25
Synthetic 2.3 148 hot-pressedfrom gel; p-quartz 49
II 4.0 152 ” from silicic acid; pquartz 49
Average -7.18 3.1 135 Best est. basedon gel precursor 49
Average -9.40 4.0 135 Est. for silicic acid precursor; 49
 EVANS AND KOHLSTEDT 157

TABLE 3d. Power Law Creep Constants.Olivine Rocks

Rock Name log A n Q Remarks Source
(iMPa”.+) (A&01-1)

Mt. Burnett 3.1 3.3 465 7

0%)
Mt Burnett -1.0 2.1 226 7
(Wet)
Mt. Burnett 3.0 419 39
0-Y)
Mt Burnett 2.5 5. I 392 65
(Wet)
CollectedData 4.8 3.0 502 Analysis of single and 20
WY) polycrystalline creep data
Anita Bay (Dry) 4.5 3.6 535 9
Anita Bay (Wet) 4.0 3.4 444 9
Aheim 2.6 4.5 498 9
(Wet)
Synthetic (Dry) 5.4 3.5 540 Dry; Q from 9 38

Synthetic (Wet) 3.3 3 420 Wet; Q from 9 38

mechanisms have been suggested: enhanced solution- Less work has been done on undercutting mechanisms,
transport along a high-diffusivity boundary phase (Table but some experiments have examined undercutting
4) and undercutting of the contact [4]. For the enhanced involving brittle or plastic processes[62, 841.
transport mechanisms,two possiblerate-limiting stepsare
dissolution [66] and diffusion [14, 72, 961. Most workers 5.2. Pressure Solution Experiments
assume quasistatic equilibrium, but Lehner and Bataille Of the many experiments designedto explore pressure
[46, 471haveused non-equilibrium thermodynamics. solution in the laboratory, the great majority have
The properties of the fluid-laden contact zone are critical measureddensification of powdered aggregatesor porous,
for kinetics. Estimates of boundary diffusivity vary by five natural rocks, although some have measured dissolution
orders of magnitude, depending on whether the fluid is rates of stressedsingle crystals or the shear strength of a
supposedto be a continuous film or an impurity on a nearly dense polycrystalline aggregate. The densification
semicoherent boundary [6]. Some theories require an configuration is identical to liquid phasesintering and hot
adsorbedwater layer capableof supporting normal stresses pressing [S11.
[14, 72, 961. If such a layer exists, it may have Microstructural evidence for solution transfer is often
thermodynamic properties different from bulk water. very convincing [ll, 911, but identifying the rate-
Supporting this supposition are measurementsof nonzero controlling step can be problematic. Experiments on KC1
wetting angles in quartz, olivine, [33, 45, 931 and calcite [62] and NaCl [82] suggestthat failure of contactsbetween
[25, 341. Raj [66] postulates a hybrid, island/channel grains may occur via dislocation flow. Other experiments
structure allowing rapid transport in the channels,but still with brine-saturated,NaCl aggregatesat 3%45”C, indicate
providing mechanical support at the contact. All the that densitication rate is inversely proportional to grain
transport models have two aspectsin common. First, they size, suggestinginterface control [66]. Pharr and Ashby’s
are linear in driving force (either effective stress, CI,, [62] undercutting model, derived from experimentson KC1
effective pressure, P,, or the difference between effective and sucrose, differs from the transport models in several
pressureand the ratio of the interfacial energy, y, divided important respects,but also assumesa low wetting angle.
by the radius of curvature of the interface, r). Secondly, The nature of the interface is clearly quite important:
strain rate is proportional to l/d when deformation is deformation occurs more quickly when dissimilar
limited by interface transport, or to l/d3 when it is limited materials are pressed together than when grains of the
by boundary diffusion. samematerial are pressed[29].
158 RHEOLOGY OF ROCKS

- I 6. BRITTLE-DUCTILE AND BRITTLE-PLASTIC

/ Cal&e Rock TRANSITIONS

At the Earth’s surface, most rocks fail by cataclastic

mechanisms (481. At greater depth, flow proceeds
exclusivelyby plastic processes.The transition from brittle
to plastic deformation occurs in two stages [16, 231. The
first, called the brittle-ductile transition, is a change in
failure mode from localized to distributed failure. The
second, brittle-plastic transition is a change from brittle
cracking to plastic flow alone (Table 5) [73]. When brittle
I I
0 I I I and plastic mechanismsoccur simultaneously,deformation
600 1000 1400 is called semibrittle.
Temperature (K) The relative contribution of each mechanism is
determined by mineral properties, phase chemistry, pore
Fig. 3. Differential stressnecessaryto causecreep at a rate geometry, stress conditions, temperature, lithostatic
of 1O-5s-r in several calcite rocks at various temperatures pressure, fluid pressure, and chemical fugacities. Failure
(see Table 3a for sources). The transition between power mode is determined by the above, the mechanical
law creepand stress-activated,glide-controlled creep is the characteristics of the loading system, and the geometry of
upper horizontal line (see Table 2). The transition to the load-bearing member. Localized failure is possible
diffusional creep (lower horizontal line) has beenobserved with plastic flow or brittle cracking, acting alone or in
for Solnhofenlimestone. combination.
For conventional triaxial loading of rocks in the gas
medium apparatus, the pressure of the brittle-ductile
transition is predicted empirically by the intersection of the
Mohr-Coulomb criterion with Byerlee’s friction law [48]

Power-Law
L_ -I- Breakdown

Power-Law
Creep 1

c <SyntheticI, wateradded”
SyntheticII, qoter odded”
-

2
1000 1100 1200
Temperature (K) [ Diffusional Creep 1
I I I I I

Fig. 4. Creep strengths for quartz rocks as a function of 1300 1500 1700
temperature at a strain rate of 1O-5s-r. Rocks dried in Temperature (K)
vacuum prior to testing tend to be stronger than rocks
tested under room “dry” conditions. Thicker lines show Fig. 5. Differential stressesnecessaryto causecreep at 10e5
creep strengths for quartz rocks with added water. s-l in olivine rocks in the interval 1350-175OY; foZ is
Equilibration is apparently very slow, leading to strengths buffered by the iron jackets at Fe-FeO. The presenceof
which vary considerably, depending on the actual water water causesa reduction in flow strength by a factor of
concentration in the sample. No studies have definitively about 2-5. Breakdown of the power-law relation occurs at
demonstrateddiffusional flow in quartz-rich rocks. about 500-1000MPa.
 EVANS AND KOHLSTEDT 159

TABLE 4. Interface Transport Models for PressureSolution

Rate-Limiting Deviatoric Constitutive Law Remarks Ref.
Mechanism Volumetric
Grain boundary (gb)
diffusion rate-limiting. 14, 72
Grain . 32R C,D,s Q
Boundary Deviatoric &= Diffusion in pore space
RTpS d3 rapid. Pore fluid.
Diffusion

Grain Cubic Packing

Deviatoric Constant Volume 66
Boundary
Diffusion No increasein internal
energy

Interface Cubic packing

Deviatoric Constant Volume 66
Reaction
Dissolution and
precipitation rate-
controlling and equal

Grain Density of aggregate= l/r2

Volumetric Diss. and precip. fast; gb 66
Boundary
Diffusion diff. rate-controlling

Cubic packing 66
Interface Volumetric Solution nearly inviscid

Fluid Phase Fluid phasealong

2-Q c,,,(l- cpsr)QJP, (I- A> p, boundary; n =2 for 80
Boundary Volumetric 6, =
Diffusion kT Vs,ps (A)” d3 sphericalgrains, 4 for
cubic grains

Grain Rate of diss. along

Boundary Both circular contact, radius a. 46
Diffusion Assumesthat diss. rate is
Island Mode1 constant along the contact

Fluid Diff. in pore fluid rate-

. RCoD,Omd ((0~ -p/)+3b, -03)]
limiting. Effective bulk 61
Diffusion Deviatoric E, =
RT d’ diff. in porous solid
derived using Archie’s law

Reaction at solid-liquid
. R2k’ 6 {(03-pf)+3h -(s3)}
interface rate-limiting.
Source/Sink Deviatoric
Reaction EI =RTd, d Island/channelboundary 61
mode1assumed.

Diffusion along grain

n aDgb {(~3-p,)+3h-a3)) contact rate-limiting. 61
Source/Sink Deviatoric ;, =--
Diffusion RT diz d Island/channel mode1
160 RHEOLOGY OF ROCKS

The brittle-plastic transition is bounded in differential

Symbols used in Table 4
stress-pressure space by the criterion
a Typical dimension of the grain-grain contact.
a Product of area fraction of solid islands in contact and q-q=P, (16)
step height, divided by a lattice dimension:
numerically.of order 1. suggested by C. Goetze on the basis of Edmond and
Paterson’s data [13]. At pressures greater than the
& /jE Equivalent volumetric strain and strain rate.
strength, deformation proceeds without cracking. Under
Cgbf Solute concentration in boundary fluid of loaded those conditions, the relaxed von Mises criterion is
interface.
satisfied. Semibrittle behavior occurs at lower prcssurcs
Concentration of solute in contact with unstressed
(for fixed temperature and strain rate), or at lower
planar surface.
temperature (for fixed pressure and strain rate). Using
d Grain size
(16) even with the current limited data, failure mode maps
6 Width of the grain boundary.
4 Width of the island in the island-channel model
Df Diffusion coefficient of the solute in the fluid
Db
Grain boundary diffusion coefficient.
.
E, Deviatoric strain rate.
.
El Rate of change of greatest principal strain
Y Specific surface energy of solid/liquid boundary
7 Viscosity of fluid.
k, ‘CO
Lineal growth rate of crystal in a solution with
concentration CO.
k’ Solution or precipitation rate constant in mol me2 s-l
Non-localized
“‘a/ Archie’s law exponent relating porosity and electrical
resistivity.
pe Applied lithostatic pressure, P,, minus pore fluid -Law
pressure, Pf
p,, pc p%Density of ihe solid, fluid, and aggregate, respectively.
r Curvature of the solid-liquid interface. Pressure
R,k Gas constant and Boltzmann’s constant.
0 Applied differential stress, o,-(J~ / I
1 Brittle / , CarraraMarble , 1~
oe , Equivalent stress,
CJe={+T, -c72)2 f(& -03)2+(L73 -cY1)2]}~1
-
on Effective normal stress on the grain boundary region
T Absolute temperature.
V Component of barycentric velocity normal to interface
s Volume fraction of the island-channel region occupied
by solid
W
” Speed of the interface along its normal.
x Area fraction of the boundary actually bearing the load. 0 L’ - 1
Z Numerical constant = 216 &%
400 800
R Molar volume of the solid.
B ’ Pressure (MPa)
Fig. 6a. Schematic trend of strength versus pressure for a
rock undergoing the brittle-plastic transition. Nonlocalized
plastic flow occurs when the differential stress is less than
the pressure. Nonlocalized brittle fracture occurs when the
(Figure 6). Abnormally high fluid pressure gradients or the strength is greater than the pressure, but less than the
presence of partial melts can induce semibrittle or strength to cause frictional sliding.
completely brittle behavior, even at high temperatures or Fig. 6b. Similar curves based on data for Carrara marble
low strain rates. deformed at a variety of strain rates and temperatures.
 EVANS AND KOHLSTEDT 161

TABLE 5. Phenomenologyof Brittle-Ductile Transition in Intact Rocks [16]

Failure Mode < _____-__-_-__-______--------- Ductile _-___--_--___-__-___----- ---- ->
< _-_---Localized ________-___-__~ >
Mechanism Class < ____________ Brittle------------><-------- SemiBrittle------ >< ___________ plastic __-________ >

Strain to Failure < _____________ <3% ____________- >< _-------- --- >3%------------ >< ------------ >5% ------------ >
Work Softening < - >
StressDrops < >
Loss in Cohesion < ---->
Microcracking <- ------ m-w-- ---- >

Dilatancy <
--m-w-- ---- --I_ >

Acoustic Emission
Press.Dep. Strength < -I-___ >

Temp. Dep. Strength < ------------- low ______________ ________________ >< ______-____ high ______________ _______________ >

Deformation Mech. Distributed and Localized Distributed Microcracking Fully Plastic

Microcracking Local Plasticity
Macroscopic Appearance

can be constructed in differential stress - pressurespaceor With increasing confining pressure and decreasing
in pressure-temperaturespace. dilatancy factor, h,, is predicted to increase. Experiments
In calcite and halite at elevated pressure, fully plastic on a variety of rocks agree with the mechanical analyses,
flow is possibleeven at room temperature [ 161.In contrast, except that the hardening modulus critical for incipient
feldspathic rocks deform by localized brittle fracture or localization is predicted to be overly negative as compared
distributed cataclasis at all pressures, for temperatures with the experimentalvalues.
below 750°C [30, 86, 88). High temperatures are also During cataclastic failure of rocks in compression,
necessaryfor plastic flow in quartzites [89] and peridotites, dilatant microcracks nucleate,grow, and coalesce[48]. For
but those rocks do not show a distributed cataclastic flow dilatant wing cracks growing from an inclined preexisting
regime as do feldspar rocks. flaw, fracture mechanics models [55, 751 indicate that
strain hardening occurs until the dilatant cracks interact
6.1. Mechanics of Brittle-Ductile Transition elastically; at which point strain softening and localization
From a continuum mechanicspoint of view, localization occur. Although rigorous tests of the fracture mechanics
in a pressure sensitive, dilatant material under models do not exist, the theories do provide rational
axisymmetric loading is determined by three parameters explanations for several observations, including, for
[70]: an internal friction coefficient p, a dilatancy factor 0, example, dependenceof fracture strength on the inverse of
and a hardening coefficient h. The normalized critical the square root of grain size. The models do not explain
hardening modulus, h,, for the inception of strain the empirical observationthat the friction law bounds the
localization is brittle-ductile transition, unless fracture toughness and
plastic flow strength scalewith eachother.
(17)
6.2. Mechanics of Semibrittle Deformation
Plastic flow mechanisms and brittle cracking can
whereg is the shear modulus, v is Poisson’sratio, and N is interact in a variety of ways. Cracks may be nucleated at
l/J3 for axisymmetric compression.For most stress states, dislocation pileups [97], intersecting twins, rigid second
the model predicts negative h, for shear band formation, phases, or incoherent boundaries. Cavities may form
both for a yield vertex model and for isotropic hardening. during creepat sliding grain boundaries.Cracks may grow
 162 RHEOLOGY OF ROCKS

or blunt by creep processes [54, 681. Based on tensile crack mechanism [97] shows that cracks nucleated by
failure experiments several broad classes of fracture dislocation pileups propagate to a length which scales with
mechanisms can be defined [2] depending on the the number of dislocations in the pileup, and with the
partitioning of strain into rate independent plasticity, difference between the resolved applied stress and the
creep, and fracture processes: cleavage, intergranular Peierls stress.
brittle fracture, plastic void growth, and plastic rupture are Despite progress in understanding semibrittle failure, a
low-temperature processes; intergranular creep fracture, satisfactov constitutive law does not exist. Because of the
creep void growth; and creep rupture are high-temperature potential complexity of mixed deformation mechanisms, it
processes. is naive to expect one theory to represent semibrittle
As with brittle fracture, semibrittle deformation may be deformation over a range of conditions. In the same way
treated by prescribing a nucleation criterion and predicting that deformation maps are necessary to represent plastic
growth to failure according to a separate failure criterion, flow, multimechanism maps will surely be needed to
often involving a critical damage state. Then, contours of describe semibrittle deformation.
time or strain to failure may be plotted in differential
Acknowledgments: Funding for this work was provided by NSF
stress-temperature space. Most work has concentrated on
GeosciencesDivision by grants EAR91 18969 (BE), EAR901823
tensile loading, but some attention has been paid to (DLK), and OCE920041 (DLK). We thank the editor and the
compressive, multiaxial loading [1, 55, 751. In much the reviewer for comments and forbearance, and apologize to the
same way that confining pressure stabilizes the authors of many excellent papers which could not be cited, owing
propagation of brittle wing cracks, an analysis of the Stroh to space restrictions.

REFERENCES

1. Ashby, M. F., J. D. Embury, S. H. Tectonophysics, 136,27-63, 1987. Deformation of earth materials: Six
Cooksley, and D. Teirlinck, Fracture 9. Chopra, and M. S. Paterson, The easy pieces, Reviews of Geophysics,
maps with pressure as a variable, experimental deformation of dunite, Supplement, 823-843, 1991.
ScriptaMetall. 19, 385-390, 1985. Tectonophysics, 78, 453473, 1981. 16. Evans, B., Fredrich, J. T., and Wong,
2. Ashby, M. F., C. Ghandi, and D. M. 10. Cooper, R. F., and D. L. Kohlstedt, T.-f., The brittle to ductile transition
R. Taplin, Fracture mechanism maps Rheology and structure of olivine- in rocks: Recent experimental and
and their construction for fee metals basalt partial melts, J. Geophys. Rex, theoretical progress, in The Brittle-
and alloys, Acta Metal. 27, 699, 91, 9315-9323, 1986. Ductile Transition - The Heard
1979. Il. DeBoer, R. B., P. J. C. Nategaal, and Volume, Geophys. Monogr. Ser. vol
3. Bai, Q., S. J. Mackwell, and D. L. E. M., Duyvis, Pressure solution 56, edited by A. G. Duba, W. B.
Kohlstedt, High-temperature creep of experiments on quartz sand, Durham, J. W. Handin, H. F. Wang,
olivine single crystals: 1: Mechanical Geochimica Cosmochim. Acta, 41, pp. l-20, AGU, Washington, D. C.,
results for buffered samples, J. 257-64, 1977. 1990.
Geophys. Rex, 96,244 l-2464, 199 1. 12. DeBresser, J. H. P., and C. J. Spiers, 17. Farver, J. R., and R. A. Yund,
4. Bathurst, R. G. C., Diagenetic fabrics High-temperature deformation of Measurement of oxygen grain
in some British Dinantian calcite single crystals by r+ and ft boundary diffusion in natural tine-
limestones, Liverpool Manchester slip, in, Dexormation Mechanisms, grained quartz aggregates, Geochim.
Geol. J., 2, 11-36, 1958. Rheology and Tectonics, Geol. So. Cosmochim. Acta, 5.5, 1597-1607,
5. Boullier, A.M. and Gueguen, Y., Spec. Publ. No. 54, edited by R. J. 1991.
Origin of some mylonites by super Knipe, and E. H. Rutter, pp. 285- 18. Frost, H. J., and M. F. Ashby,
plastic flow, Co&rib. Mineral. 298, Geol. Sot. London, UK., 1990. Deformation-mechanism Maps,
Petrol. 50, 93-104 1975. 13. Edmond, J. M., and M. S. Paterson, Pergamon Press, Oxford, Great
6. Brady, J. B., Intragranular diffusion Volume changes during the Britain, 166 pp., 1982.
in metamorphic rocks, Amer. J. Sci., deformation of rocks at high 19. Gibbs, J. W., On the equilibrium of
283-A, 540-88, 1983. pressures, Znt. J. Rock Mech. Min. heterogeneous systems, Trans. Conn.
Carter, N. L., and H. G. Sci. Geomech. Abstr., 9, 161-182, Acad., Vol. III, 108-248, 1876.
AveLallemant, High temperature 1972. 20. Goetze, C., The mechanisms of creep
flow of dunite and peridotite, Geol. 14. Elliot, D., Diffusion flow law in in olivine, Phil. Trans. R. Sot.
Sot Am. Bull., 81,2181-2202, 1970. metamorphic rocks, Geol. Sot. Amer. London, A288, 59-l 19,1978.
Carter, N. L., and M. C. Tsenn, Flow Bull., 84,2645-2664, 1973. 21. Green, H. W., II, Rheological
properties of continental lithosphere, 15. Evans, B., and G. Dresen, implications of the dissolution of
 EVANS AND KOHLSTEDT 163

volatiles in mantle olivine, Phys. angles in the system quartz-CO,- Res., 89,42814297, 1984.
Earth Planet. Inter., 51, 123-124, H,O-NaCI at 800°C and I-15 kbar: 44. Kronenberg, A. K., S. H. Kirby, R.
1988. the effects of pressure and fluid D. Aines, and G. R. Rossman,
22. Green, H. W., II, and R. S. Borch, A composition on the permeability of Solubility and diffusional uptake of
new molten salt cell for precise stress quartzite, Earth Planet. Sci. Let., hydrogen in quartz at high water
measurement at high pressure, Eur. 114, 171-184, 1992. pressures: Implications for hydrolytic
J. Mineral., I, 213-219, 1989. 34. Holness, M. B., and C. M. Graham, weakening, J. Geophys. Res., 91,
23. Griggs, D. and J. Handin, Equilibrium dihedral angles in the 12,723-12,744, 1986.
Observations on fracture and a system CO,-%O-NaCl-calcite, and 45. Lee, V. W., S. J. Mackwell, and S. L.
hypothesis of earthquakes, in Rock implications for fluid flow during Brantley, The effect of fluid
Deformation, Mem. 79, edited by D. metamorphism, Contrib. Mineral. chemistry on wetting textures in
Griggs and J. Handin, pp. 347-364, Petrol., 108, 368-383, 1991. Novaculite, J. Geophys. Res., 96,
Geo. Sot. Am., N.Y. 1960. 35. Houlier, B., M. Cheraghmakani, and 10,023-10,037, 1991.
24. Hansen, F. D. and N. L., Carter, 0. Jaoul, Silicon diffusion in San 46. Lehner, F. K., Thermodynamics of
Creep of selected crustal rocks at Carlos olivine, Phys. Earth Planet. rock deformation by pressure
1000 MPa, EOS, Trans. Am. Inter., 62, 329-340, 1991. solution, in Deformation Process in
Geophys. Sot., 63, 437, 1982. 36. Houlier, B., 0. Jaoul, F. Abel, and R. Minerals, Ceramics and Rocks,
25. Hay, R. S., and B. Evans, C. Liebermann, Oxygen and silicon edited by D. J. Barber and P. G.
Intergranular distribution of pore self-diffusion in natural olivine at T= Meredith, pp. 296-333, Unwin-
fluid and the nature of high-angle 13OO”C., Phys. Earth Planet. Inter., Hyman, Boston, 1990.
grain boundaries in limestone and 50,240-250, 1988. 47. Lehner, F. K., and J. Bataille, Non
marble, J. Geophys. Res., 93, 8959- 37. Jaoul, O., J. Tullis, and A. equilibrium thermodynamics of
8974,198s. Kronenberg, The effect of varying pressure solution, Pure Appl.
26. Heard, H. C., Comparisons of the water contents on the creep behavior Geophys., 122,53-85, 1985.
flow properties of rocks at crustal of Heavitree quartzite, J. Geophys. 48. Lockner, D. A., Rock failure, in this
conditions, Philos. Trans. R. Sot. Res.. 89, 42974312, 1984. volume, 1994.
Lond, Ser. A, 283, 173-186, 1976. 38. Karato, S.-I., M. S. Paterson, and J. 49. Luan F.-c. and M. S. Paterson,
27. Heard, H. C. and N. L. Carter, D. F&Gerald, Rheology of synthetic Preparation and deformation of
Experimentally induced “natural” olivine aggregates: Influence of grain synthetic aggregates of quartz, J.
intergranular flow in quartz and size and water, J. Geophys. Res., 91, Geophys. Res., 97,301-344, 1992.
quartzite, Am. J. Sci., 266, l-42, 8151-8176, 1986. 50. Ma&well, S. J., and D. L. Kohlstedt,
1968. 39. Kirby, S. H., and C. B. Raleigh, Diffusion of hydrogen in olivine:
28. Heard, H. C. and, C. B. Raleigh, Mechanisms of high-temperature, Implications for water in the mantle,
Steady-state flow in marble at 500- solid-state flow in minerals and J. Geophys. Res., 95, 5079-5088,
SOOY!, Geol. Sot. Am. BuIl. 83, 935- ceramics and their bearing on the 1990.
956,1972. creep behavior of the mantle, 51. Mackwell, S. J., D. Dimos, and D. L.
29. Hickman, S. H., and B. Evans, Tectonophysics, 19, 165-194, 1973. Kohlstedt, Transient creep of olivine:
Experimental Pressure Solution in 40. Kirby, S. H., and J. W. McCormick, Point-defect relaxation times, Philos.
Halite, 1: The effect of Creep of hydrolytically weakened Mag., 57, 779-789, 1988.
grain/interphase boundary structure, synthetic quartz crystals oriented to 52. McLaren, A. C., J. D. Fitz Gerald,
J. Geological Sot. London, 148, promote {2110} cOOI> slip: A brief and J. Gerretsen, Dislocation
549-560, 1991. summary of work to date, Bull. nucleation and multiplication in
30. Hirth, G., and J. Tullis, The effects Mineral., 102, 124-137, 1979. synthetic quartz: Relevance to water
of pressure and porosity on the 41. Kirby, S. H., and A. K. Kronenberg, weakening, Phys. Chem. Minerals,
micromechanics of the brittle-ductile Rheology of the lithosphere: Selected 16,465482,1989.
transition in quartzite, J. Geophys. topics, Rev. Geophys., 25, 1219- 53. Nakamura, A., and H. Schmalzried,
Rex, 94, 17825-17838, 1989. 1244,1987. On the nonstoichiometry and point
31. Hobbs, B. E., The influence of 42. Koch, P. S., J. M. Christie, A. Ord, defects of olivine, Phys. Chem.
metamorphic environment upon the and R. P. George Jr., Effect of water Miner., 10,27-37, 1983.
deformation of minerals, on the rheology of experimentally 54. Needleman, A. and J. R. Rice, Plastic
Tectonophysics, 78, 335-383, 1981. deformed quartzite, J. Geophys. Res., creep flow effects in the diffusive
32. Hobbs, B. E., Constraints on the 94, 13975-13,996, 1989. cavitation of grain boundaries, Acta
mechanism of deformation of olivine 43. Kronenberg, A., and J. Tullis, Flow Metall., 28, 1315-1332, 1980.
imposed by defect chemistry, strengths of quartz aggregates: Grain 55. Nemat-Nasser, S., and H. Hot-ii,
Tectonophysics, 92, 35-69, 1983. size and pressure effects due to Compression-induced non planar
33. Holness. M. B.. Eauilibrium
, 1
dihedra1 hydrolytic weakening, J. Geophys. crack extension with application to
164 RHEOLOGY OF ROCKS

splitting, exfoliation, and rock burst, dimensional constitutive relations Schutjens, Densitication of
J. Geophys. Rex, 87, 68056822, and ductile fracture, edited by S. crystalline aggregates by fluid-phase
1982. Nemat-Nasser, pp. 173-I 84, North- diffusional creep, in Deformation
56. Parrish, D. K., A. L. Krivz, and N. L. Holland Publishing, 198 1. Processes in Minerals, Ceramics and
Carter, Finite element folds of 69. Ross, J. V., S. J. Bauer, and N. L. Rocks, edited by D. J. Barber and P.
similar geometry, Tectonophysics, Carter, Effect of the a-P quartz G. Meredith, pp. 334-353, Unwin-
32, 183-207, 1976. transition on the creep properties of Hyman, Boston, 1990.
57. Paterson, M. S., The ductility of quartzite and granite, Geophys. Res. 81. Swinkels, F. B., and M. F. Ashby, A
rocks, in Physics of Strength and Letters, IO, 1129-l 132, 1983. second report on sintering diagrams,
Plasticity, edited by A. S. Argon, pp. 70. Rudnicki, J. W., and J. R. Rice, Acta. metall., 29,259-281, 1981.
199-208, MlT Press, Cambridge, Conditions for the localization of 82. Tada, R., and R. Siever,
MA, 1969. deformation in pressure-sensitive Experimental knife-edge pressure
58. Paterson, M. S., Problems in the dilatant materials, .I. Mech. Phys. solution of halite, Geochim.
extrapolation of laboratory Solids, 23,371-394, 1975. Cosmochim. Acta, 50,29-36, 1986.
rheological data, Tectonophysics, 71. Rutter, E. H., The influence of 83. Tada, R., and R. Siever, Pressure
133,3343, 1987. temperature, strain-rate and solution during diagenesis: A review,
59. Paterson, M. S., Interaction of water interstitial water in the experimental Ann. Rev, Earth Planet. Sci., 17, 89-
with quartz and its influence in deformation of calcite rocks, 118,1989.
dislocation flow: Ah overview, in Tectonophysics, 31, T21-T28, 1974. 84. Tada, R., R. Maliva, and R. Siever,
Rheology of Solids and of the Earth, 72. Rutter, E. H., The kinetics of rock A new mechanism for pressure
edited by S.-i. Karato, and M. deformation by pressure solution, solution in porous quartzose
Toriumi, pp. 107-142, Oxford Phil. Trans. R. Sot. Lond., A, 283, sandstone, Geochim. Cosmochim.
University Press, 1989. 203-219,1976. Acta, SI, 2295-2301,1987.
60. Paterson, M. S., Rock deformation 73. Rutter, E. H., On the nomenclature of 85. Takeuchi, S. and A. S. Argon,
experimentation, in The Brittle- mode of failure transitions in rocks, Review: Steady state creep of single-
Ductile Transition in Rocks, The Tectonophysics, 122, 381-387, 1986. phase crystalline matter of high-
Heard Volume, Geophys. Monogr. 74. Ryerson, F. J., W. B. Durham, D. J. temperatures., J. Mater. Sci., II,
Ser., vol. 56, edited by A. G. Duba et Chemiak, and W. A. Lanford, 1542-1566, 1976.
al., pp. 187-194, AGU, Washington, Oxygen diffusion in olivine: Effect of 86. Tullis, J., Experimental studies of
1990. oxygen fugacity and implications for deformation mechanisms and
61. Paterson, M. S., A theory for creep, J. Geophys. Res., 94, 4105- microstructures in quartzo-
granular flow accommodated by 4118,1989. feldspathic rocks, in Deformation
material transfer via an intergranular 75. Sammis, C. G., and M. F. Ashby, Processes in Minerals, Ceramics,
fluid, Tectonophysics, in press, 1994. The failure of brittle porous solids and Rocks, edited by D. Barber and
62. Pharr, G. M., and M. F. Ashby, On under compressive stress states, Acta P. Meredith, pp. 190-227, Unwin and
creep enhanced by a liquid phase, Metall., 34, 51 l-526, 1986. Hyman, Cambridge, 1990.
Acta. metall., 31, 129-138, 1983. 76. Schmalzried, H., Solid State 87. Tullis, J., and R. A. Yund, Hydrolytic
63. Poirier, J.-P., Creep of crystals: Reactions, Mongr. Modem Chem. weakening of experimentally
High-temperature deformation Ser., vol. 12, 254 pp., Academic deformed Westerly granite and Hale
processes in metal, ceramics and Press, NY, 198 1. albite rock, J. Strut. Geol., 2, 439-
minerals, Cambridge University 77. S&mid, S., J. N. Boland, and M. S. 451,198O.
Press, Cambridge, 260 pp. 1985. Paterson, Super-plastic flow in 88. Tullis, J., and R. A. Yund, Transition
64. Poirier, J.-P., Plastic rheology of tinegrained limestone, from cataclastic flow to dislocation
crystals, companion volume 2, 1994. Tectonophysics. 43,257-291, 1977. creep of feldspar: Mechanisms and
65. Post, R. L. Jr., High-temperature 78. S&mid, S. M., M. S. Paterson and J. microstructures, Geology, 1.5, 606-
creep of Mt. Bumet dunite, N. Boland, High-temperature flow 609, 1987.
Tectonophysics. 42,75-l 10, 1977. and dynamic recrystallization in 89. Tullis, J., and R. A. Yund, The
66. Raj, R., Creep in polycrystalline Carrara marble, Tectonophysics, 65, Brittle-ductile transition in feldspar
aggregates by matter transport 245-280, 1980. aggregates: An experimental study,
through a liquid phase, J. Geophys. 79. Shelton, G. L., Experimental in Fault Mechanics and Transport
Res., 87,4731-4739, 1982. deformation of single phase and Properties of Rocks: A Festchrijii for
67. Ranalli, G., Rheology of the Earth, polyphase crustal rocks at high W. F. Brace, edited by B. Evans and
Allen & Unwin, Boston, MA, pp. pressures and temperatures, Ph.D. T.-f. Wang, pp. 89-118, Academic
235-317, 1986. Diss., Brown Univ. Providence, R. I., Press, London, U.K., 1992.
68. Rice, James R., Creep cavitation of 146 pp., 1981. 90. Tullis, T. E., and J. Tullis,
grain interfaces, in Three- 80. Spiers, C. J., and P. M. T. M. Experimental rock deformation
 EVANS AND KOHLSTEDT 165

techniques, in Mineral and Rock Deformation Mechanisms, Rheology crystals, J. Appl. Phys., 28, 1185
Deformation: Luboratov Studies, and Tectonics, Geo. Sot. Spec. Pub. 1189,1957.
The Paterson Volume, Geophys. 54, pp. 259-284, 1990. 95. Weertman, J., Dislocation climb
Monogr. Ser., vol. 36., edited by B. 93. Watson, E. B., and J. M. Brenan, theory of steady-state creep, Trans.
E. Hobbs and H. C. Heard, pp. 297- Fluids in the lithosphere, part 1: Am. Sot. Metal., 61,681-694, 1968.
324, AGU, Washington, D. C., 1986. Experimentallydetermined wetting 96. Weyl, P. K., Pressure solution and
91. Urai, J. L., C. J. Spiers, H. J. Zwart, characteristics of C02-H2p fluids the force of crystallization - a
and G. S. Lister, Weakening of rock and their implications tor fluid phenomenological theory, J.
salt by water during long-term creep, transport, host-rock physical Geophys. Res., 64,2001-2025, 1959.
Nature, 324, 554-557, 1986. properties, and fluid inclusion 97. Wong, T.-f., A note on the
92. Walker, A. N., E. H. Rutter, and K. formation, submitted to Earth and propagation behavior of a crack
H. Brodie, Experimental study of Planetaly Sci. Left., 85, 497-515, nucleated by a dislocation pile-up, J
grain-size sensitive flow of synthetic, 1987. Geophys. Res., 95,8639-8646, 1990.
hot-pressed calcite rocks, in 94. Weertman, J., Steady-state creep of