UCRL-JRNL-202061
New Developments in
Deformation Experiments at
High Pressure
W. B. Durham, D. J. Weidner, S.-I. Karato, Y.
Wang
January 28, 2004
Reviews in Mineralogy
Disclaimer
This document was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor the University of California nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for
the accuracy, completeness, or usefulness of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any
specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise,
does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United
States Government or the University of California. The views and opinions of authors expressed herein
do not necessarily state or reflect those of the United States Government or the University of California,
and shall not be used for advertising or product endorsement purposes.
2
New Developments in Deformation Experiments
at High Pressure
William B. Durham
University of California,
Lawrence Livermore National Laboratory, P.O. Box 808
Livermore, California 94550
Donald J. Weidner
Department??
State University of New York
Stony Brook, New York ZIP??
Shun-ichiro Karato
Departmentof Geology and Geophysics
319 Kline Geology Laboratory, Yale University
New Haven, Connecticut 06520
Yanbin Wang
GeoSoilEnviroCARS
University of Chicago
Chicago, Illinois ZIP??
INTRODUCTION
Although the importance of rheological properties in controlling the dynamics and
evolution of the whole mantle of Earth is well-recognized, experimental studies of
rheological properties and deformation-induced microstructures have mostly been limited
to low-pressure conditions. This is mainly a result of technical limitations in conducting
quantitative rheological experiments under high-pressure conditions. A combination of
factors is changing this situation. Increased resolution of composition and configuration
of Earth’s interior has created a greater demand for well-resolved laboratory
measurement of the effects of pressure on the behavior of materials. Higher-strength
materials have become readily available for containing high-pressure research devices,
and new analytical capabilities—in particular very bright synchrotron X-ray sources—are
now readily available to high-pressure researchers.
One of the biggest issues in global geodynamics is the style of mantle convection
and the nature of chemical differentiation associated with convectional mass transport.
Although evidence for deep mantle circulation has recently been found through seismic
tomography (e.g., van der Hilst et al. (1997)), complications in convection style have also
been noted. They include (1) significant modifications of flow geometry across the
mantle transition zone as seen from high resolution tomographic studies (Fukao et al.
1992; Masters et al. 2000; van der Hilst et al. 1991) and (2) complicated patterns of flow
in the deep lower mantle (~1500-2500 km), perhaps caused by chemical heterogeneity
(Kellogg et al. 1999; van der Hilst and Karason 1999).
These studies indicate that while large-scale circulation involving the whole mantle
no doubt occurs, significant deviations from simple flow geometry are also present. Two
mineral properties have strong influence on convection: (1) density and (2) viscosity
(rheology) contrasts. In the past, the effects of density contrast have been emphasized
1529-6466/00/0051-0002$05.00
2
Durham, Weidner, Karato & Wang
(Honda et al. 1993; Kellogg et al. 1999; Tackley et al. 1993), and the influence of
rheology has been demonstrated by geodynamic modeling (Davies 1995; Karato et al.
2001).
Rheological properties of the Earth’s mantle can be inferred from analyses of
geodynamic data such as post-glacial crustal movement and gravity anomalies
(Mitrovioca and Forte 1997; Peltier 1998). While such geodynamic inference provides an
important data set on rheological structures of the Earth, there are serious limitations for
these approaches. First, the resolution of post-glacial rebound data to infer mantle
viscosity is limited to ~1200 km. Below that depth, there is little constraint for viscosity
from post-glacial rebound (Mitrovioca and Peltier 1991). Furthermore, strain magnitude
involved in post-glacial rebound is much smaller than strain magnitude involved in
convection, and this raises the issue of effects of transient rheology (Karato 1998).
Gravity data (e.g., the geoid) have better sensitivity to radial variation in viscosity in the
deep mantle (Hager 1984), but they suffer from non-uniqueness (King 1995; Thoraval
and Richards 1997). Therefore, experimental studies on mineral rheology remain a vital
component in inferring mantle rheology.
High-pressure rheology may also hold some of the keys for understanding deep
earthquakes. Four classes of mechanism have been suggested over the years, including
thermal runaway instabilities (Hobbs and Ord 1988; Ogawa 1987), instabilities
accompanying recrystallization (Post 1977), dehydration embrittlement (Jiao et al. 2000;
Raleigh and Paterson 1965), and instabilities associated with polymorphic phase
transformations (see also reviews by Green and Houston (1995) and Kirby et al. (1996)).
Again, the main reason uncertainty still exists is the poor knowledge of materials
properties at high pressures. Experimental testing of the thermal runaway involved in
plastic instability requires high rates of deformation, which are difficult to obtain in small
samples; modeling would be a useful tool for this mechanism, but critical parameters for
models (rheological properties of deep Earth materials) have been missing so far (Hobbs
and Ord 1988; Karato et al. 2001). Well-controlled deformation experiments under deep
mantle conditions are needed to provide greater understanding of the shearing instability
associated with the phase transformations of olivine to wadsleyite and ringwoodite
(Burnley et al. 1991; Green et al. 1990) and the possibility of faulting associated with
dehydration of the dense hydrous magnesium silicates (Green 2001).
Why is pressure important?
Pressure is one of the most important variables in defining the ductile rheology of
mantle minerals. Where the deformation mechanism is fixed, the effect of pressure on
creep strength is given by:
σ ( P ) /σ (0) ∝ exp( PV * / nRT )
(1)
where σ(P) is the creep strength at pressure P, V* is activation volume, n is the stress
exponent, R is the gas constant, and T is temperature. For olivine, assuming an activation
volume of V* = 15 cm3/mol, T = 1600 K and n = 3, σ(10 GPa)/σ(0) = 43. This is to be
compared with the effect of water, which will change the creep strength of olivine by a
factor of ~10 from dry to water-saturated (Hirth and Kohlstedt 1996; Karato and Jung
2002 in press) or partial melting that will affect creep strength by no more than a factor of
2 (Kohlstedt and Chopra 1994). However, the pressure effect is small at the low pressures
where studies are made, e.g. σ(0.5 GPa)/σ(0) = 1.2. Therefore, high-pressure studies are
necessary in order to determine the flow properties in the deep Earth.
Such large effects of pressure on the flow laws can result in changes of dominant
flow mechanisms with depth in the Earth. Karato and Wu (1993) suggest that dislocation
flow is overtaken by diffusion flow as the creep mechanism for olivine within the upper
Developments in Deformation Experiments at High Pressure
3
mantle. Their reasoning is based on the differences in the pressure dependence, i.e., in the
value of V* in Equation (1), of the two flow laws.
Pressure is an agent for phase transformation within the Earth. These high-pressure
phases, with different crystal structures, will also differ in their mechanical properties.
Rheology of high-pressure phases needs to be investigated within the stability field of the
mineral. Because of equipment limitations, the flow properties of minerals such as
ringwoodite or majorite, which set the strength characteristics of the transition zone,
remain virtually unknown.
The effects of water can be well characterized only by high-pressure experiments. At
low pressures (P<0.5 GPa), the creep strength of olivine decreases with pressure under
water-saturated conditions (Mei and Kohlstedt 2000b; Mei and Kohlstedt 2000a).
However, at higher pressures (P>1 GPa), the creep strength of olivine increases with
pressure even at water-saturated conditions (Karato and Jung 2002 in press). Therefore
results that can be extrapolated to infer creep strength (viscosity) at deep portions
(>30 km) of Earth can be obtained only by quantitative experiments at pressures higher
than ~1 GPa.
Terminology related to strength and deformation.
Usage sometimes varies among research disciplines, so let us define a few key terms,
taking the language of rock mechanics as a basis (e.g., Jaeger and Cook (1976)). It is
assumed that the meaning of the stress (σ ij) and strain (ε ij) tensors are understood.
Deformation (used already in the title without definition) refers loosely to any process
that results in strain. We take positive values of stress and strain to indicate compression,
following geologic convention. Pressure, hydrostatic pressure, and mean normal stress
are all equivalently the trace P of the stress tensor, namely,
σij = σ ′ij + P
(2)
where σ ′ij is the deviatoric stress tensor, i, j = 1, 2, 3, representing orthogonal directions.
Confining pressure Pc is the hydrostatic stress generated by laboratory devices to
simulate geologic overburden. Although P and Pc are different whenσ ′ij is non-zero, the
difference in most geologic settings is small and the distinction is very often ignored.
Different terms for the same physical quantity are sometimes used to connote different
situations. Strain can be divided into two categories: elastic, the recoverable portion,
which owes its existence to the state of non-zero σij, and inelastic, the permanent strain
that remains forever after stress has been removed. Classic rock mechanics usage
recognizes two categories of deformation that result in inelastic strain: brittle and ductile.
Brittle deformation is associated with discontinuities of displacement and sudden loss of
strength (a term to be defined shortly) and is generally not volume conservative. Ductile
deformation is associated with finite displacement gradients, retention of strength, and
volume conservation at scales above the nm. Note that there are important strainproducing phenomena that do not fit well into either category, in particular those that
involve polymorphism (phase change) or chemical reaction. They may be ignored for the
purposes of this review. The strength of a material (in specific circumstances called the
yield strength, ultimate strength, etc.) is the maximum stress the material can support
under the stated environmental conditions. Flow and failure are closely related terms for
the state that exists the moment that applied stress matches strength for ductile and brittle
materials, respectively.
The term plasticity requires special note. It was originally defined for metals as
essentially equivalent to the phenomenon of flow, just defined, but was defined in terms
of the yield envelope (Hill 1950) without qualification as to ductile or brittle yield. While
metallurgists and most rock rheologists have retained this meaning for the term plasticity
4
Durham, Weidner, Karato & Wang
(e.g., Evans and Kohlstedt [1995] and Poirier [1985], many in the computational
branches of geology have taken plasticity to mean any yield process, whether volume
conservative or not. In this review, we follow tradition and use the terms ductile and
plastic interchangeably.
Here we focus exclusively on ductile deformation, so we use the terms rheology
(flow of materials under stress), deformation, and flow interchangeably. The rheology of
a material is described by a constitutive law of the type
ε˙ij = Af (σ ij′ ) , A = A(P, T, …)
(3)
where all relevant state variables are included in the term A. Engineering the containment of very high pressures in the laboratory generally requires high symmetry in the
sample assembly. As will be seen below, two basic types of high-symmetry strain
environments exist: axisymmetric and rotational. A shorthand form of Equation (3) has
evolved for each of these geometries and we will use that shorthand here. For
axisymmetric deformation (whose two-dimensional analog is called pure shear), the
values of the three principal stresses are σ1 ≠ σ 2 = σ3, and we simplify the strain rate
tensor to a scalar ε˙ , which we call the strain rate or shortening rate. The deviatoric
stress tensor can be characterized by the scalar σ = σ1 – σ3, which we call the differential
stress or simply stress. For rotational deformation, the strain is approximately simple
shear, and Equation (3) becomes a scalar relationship between the shear strain rate, γ˙ ,
and the shear stress, τ.
Despite the apparent simplicity of this language of strength and deformation, one
should always keep two things in mind: (1) the terminology is not universal, and (2) the
simple language belies the flow of rocks in nature, which can be counted on to be both
inhomogeneous and low symmetry.
A BRIEF HISTORY OF HIGH-PRESSURE APPARATUS
To 5 GPa: Cylindrical devices
Confining pressure in rock mechanics testing serves several purposes: (1) simulating
geologic overburden, (2) activating mechanisms of flow and fracture that depend on both
normal and shear components of stress, and (3) suppressing brittle deformation where
one is interested in gathering information on purely ductile phenomena. The scientific
needs for doing measurements under pressure are illustrated in detail in several other
contributions to this volume. These needs have driven the development of experimental
pressure systems since the early days of rock mechanics.
The earliest experimental machines for deforming rocks under elevated confining
pressure were some version of a piston pushing on a cylindrical rock held fast in a
cylindrical container. A solid or fluid medium within the container, or even walls of the
container itself, provided confining pressure to the rock sample. The reader may find an
early history of such machines in Tullis and Tullis (1986), for example. The direction of
apparatus development after 1900 was influenced primarily by P.W. Bridgman, who
made such rapid advances in the design of hydrostatic equipment that the only reasonable
path for rock mechanics to follow was to adapt deformation to Bridgman’s vessels.
Bridgman, who conceived the dynamic piston seal still commonly used, was involved in
that adaptation (Bridgman 1952c), although he was no longer alone in his efforts.
Continuing to this day, virtually every new deformation rig is some adaptation of a
hydrostatic design. The backbone of the history of deformation machines is mostly a
history of pressure vessels.
The technological breakthrough behind modern gas and liquid pressure vessels was
Developments in Deformation Experiments at High Pressure
5
the Bridgman unsupported area packing (Bridgman 1914), a geometry that intensified the
vessel’s own pressure at the point of seal and made it possible to “reach without leak any
pressure allowed by the mechanical strength of the walls of the containing vessels”
(Bridgman 1952b). Bridgman regularly worked at pressures in excess of 1 GPa and
occasionally reached 2 GPa with his fluid vessels (Bridgman 1935b). The application of
the Bridgman-seal vessel to rock deformation led to the creation of what is commonly
called in rock mechanics the “triaxial” apparatus (which is somewhat inappropriate,
because the stress environment in the fluid pressure medium can be no lower than
axisymmetric). Griggs, who learned the techniques of high pressure from Bridgman and
shared his “gift for gadgets” (Rubey 1972), initiated this application by developing a
version of the unsupported area packing around a solid moving piston (Fig. 1), thereby
allowing sample shortening to be equated (with appropriate corrections for elasticity) to
piston displacement (Griggs 1936). Griggs (1936) added a second innovation to the
design, a pressure-compensating double piston that automatically kept pressure constant
even as the sample was shortened (Fig. 1). Further development of the triaxial rig was
aided by advances such as the controlled-clearance (Newhall) packing to reduce piston
friction (Handin 1953; Turner et al. 1956); internal heating to achieve higher sample
temperatures (Griggs et al. 1960); and the internal force gauge, which removed piston
friction from the measurement of differential force on the sample (Heard and Carter
1968; Paterson 1970). This list is by no
means exhaustive. A notable recent
development is the proliferation in many
of the world’s most renown rock
mechanics laboratories of one particular
triaxial apparatus, the Paterson rig, which
is a 0.3-GPa ready-to-use system
complete with actuation and internal
force gauging for both axial and highdisplacement torsional deformation
(Paterson 1990; Paterson and Olgaard
2000). As of this writing, there are ten
such vessels in use in North America and
Europe.
Practical matters related to sealing a
high-pressure fluid have generally
limited triaxial work to about P = 1 GPa,
with most triaxial machines, in fact,
having maximum design pressures of
0.3-0.5 GPa (Paterson 1978). At higher
pressures, the solid pressure medium
piston-cylinder apparatus is used. The
solid-medium press used most widely for
geological applications is the Griggs
apparatus (Griggs 1967), adapted from
the hydrostatic solid-medium pistoncylinder apparatus (Boyd and England
1960), and subsequently improved to
provide a stress resolution of a few MPa
(Borch and Green 1989; Green and
Borch 1989; Tingle et al. 1993). Since a
solid medium has a finite shear strength,
measurements are limited to materials
Figure 1. The Griggs triaxial deformation apparatus, including the yoked pistons to hold pressure
constant during deformation. As piston P1 moves
against the sample, piston P2 automatically withdraws from the vessel at the same rate, holding the
pressurized volume constant. After Griggs (1936),
Fig. 2; reproduced with permission.
6
Durham, Weidner, Karato & Wang
that are stronger than the medium, so very low strain rates and temperatures near the
sample melting point must be avoided, but pressures of several GPa are now easily
accessible. The Griggs apparatus has been regularly used to 1.5-1.8 GPa, and to a
maximum pressure of 4 GPa.
A recent development is an apparatus built by Getting (Getting 1998; Getting and
Spetzler 1993) that combines aspects of the solid-medium and fluid-triaxial deformation
machines. Getting’s is a piston-cylinder device that features precision packings that are
sufficiently tight to contain solid argon (an exceedingly weak solid) but sufficiently
reproducible in their frictional behavior that much of the uncertainty in stress introduced
by piston friction can be accounted for by calibration. Getting (1998) claims pressure
uncertainty at P = 3 GPa at the extraordinary level of a few MPa. If this resolution applies
also to the sum of σ1 + P, as expected, it may mean that rock strengths can be measured
with a resolution that is comparable to that of many triaxial gas rigs that operate at far
lower pressures.
The practical pressure limit for all cylindrical machines is around 5 GPa. While the
pressure capacity of a hollow cylinder (or any shape, for that matter) is a function of wall
thickness and therefore virtually unlimited (Eremets 1996), the same is not true for the
piston. As one approaches P = 5 GPa, the stress required to drive the piston into the
pressure cylinder (which must be 5 GPa plus the stress to overcome packing friction plus
the strength of the sample) causes pistons constructed of even the strongest of steels to
creep. Pistons will swell and seize in a short amount of time. Note that this 5-GPa limit
exists for hydrostatic piston-cylinder vessels as well. Bridgman reached higher pressures
in hydrostatic vessels by confining the entire piston and cylinder in a larger vessel, thus
raising σ3 on the piston to the pressure in the outer vessel and allowing the piston to
apply correspondingly more stress to the inner vessel without exceeding its strength limit.
His vessel-inside-vessel multi-staged devices have been used to reach pressures of at least
10 GPa (e.g., Bridgman (1942; 1948)). Because of the complexity of those vessels,
however, they are not well suited to being instrumented for deformation studies.
5 GPa and above: Anvil devices
A much simpler method for achieving an elevated σ3 is to give the pistons a tapered
shape. (The change from rectangular to trapezoidal section also merits a name change:
from piston to anvil.) Now a component of the force applied in the σ1 direction at the
larger end of the piston/anvil is directed in the σ3 direction at the smaller end (platen
face), making σ3 > 0 at the platen face. The drawback of this solution is, of course, that
one cannot expect to contain the pressure medium over large displacements in a
conventional cylindrical pressure vessel if the sides of the (moving) anvil are tapered.
Sealing against a moving tapered surface can be achieved for small displacements
through use of a crushable gasket material; so if one is satisfied with anvil displacements
that are less than gasket thickness, one can reach high values of σ1 at the platen faces
without exceeding the failure criterion of the anvils. The first application of this principle
is generally credited to Bridgman’s (Bridgman 1952a) opposed-anvil device, which is
capable of 10 GPa, although Bridgman used the same principle many years earlier in
confining samples in a 5-GPa shear apparatus (Bridgman 1935a). It is also the underlying
principle of the Drickamer cell (see below). Note that the original purpose of anvil
displacement in these designs was pressure generation rather than sample deformation,
although as we will see below, the possibility of the latter was not ignored.
A slight variation, which allows for larger volumes at a given pressure, is to combine
the function of pressure container and pressure generator by clustering several anvils in a
quasi-spherically symmetric pattern. These are called multianvil devices. Four-anvil
Developments in Deformation Experiments at High Pressure
7
tetrahedral (Hall 1958; Houck and Hutton 1963; Lloyd et al. 1959), six-anvil cubic
(Carter et al. 1964; Houck and Hutton 1963; Osugi et al. 1964), and eight-anvil
octahedral (Kawai and Endo 1970; Walker et al. 1990) have been constructed. The eightanvil devices are multistage, with the eight anvils of the final stage contained within the
working volume of an intermediate cubic stage. They are often referred to, therefore, as
6/8 devices. Hydraulic actuation of the several anvils was sometimes independent or in
opposed pairs (Carter et al. 1964; Hall 1958), but better symmetry of anvil displacement
was achieved, along with higher pressures, when the anvils were driven by wedge-shaped
guide blocks that then required only one hydraulic actuator (Lloyd et al. 1959). There are
at least two versions of the wedge-type cubic multianvil device: a split-sphere device
(Houck and Hutton 1963) and the DIA (Osugi et al. 1964; Shimomura et al. 1985); the
latter will be describe below in detail.
Wedged guide block machines achieve high pressures more reliably than machines
with multiple actuators, and for that reason they have proliferated, especially the DIA and
6/8 devices (e.g., see review by Onodera (1987)). Note, however, that their design
function is exclusively hydrostatic pressure generation. Advances in high-pressure
rheology required that the quasi-isotropy within the pressurized volume be broken; those
advances are the main subjects of this review.
Another reason for the recent proliferation of multianvil devices is a serendipitous
advantage they have over pressure cylinders: optical and/or X-ray transparency along a
line of sight leading directly to and from the sample (Fig. 2). In the case of diamond (or
sapphire) anvil cells, the anvils themselves are transparent. For devices where anvils are
opaque, lines of sight are always available in the narrow gaps between anvils, and the
gaskets compressed in those gaps can be made of X-ray-transparent weak material (e.g.,
pyrophyllite). The analytical potential of high-energy X-rays and their recent availability
at very bright synchrotron sources has been the motivation for development of a new
generation of deformation machines that have the pressure capacity of anvil devices and
the lowered symmetry of deformation machines.
MEASUREMENT METHODS AT HIGH PRESSURE
Stress measurement
Measurement of deviatoric stress is key to quantifying the rheological properties of a
material. Stress provides information about strength relative to the laboratory time scale.
Stress, measured in conjunction with ductile strain, can yield flow laws within the context
of the state of the sample. The crucial ingredient to facilitate such studies is the
development of a stress meter, or piezometer.
In experiments operating below 5 GPa, the deforming force is applied directly to the
sample and various strategies are designed to measure the magnitude of this deforming
force. Stress is then given as the ratio of force to sample area. Typically, a load cell
placed in series with the deforming piston defines this force along the deforming column.
Combinations of friction between the force gauge and the sample and forces that support
the piston, such as in the multianvil system, continually degrade the ability to relate a
measured force to sample stress. These problems begin to interfere with the resulting
accuracy at 0.3 GPa and completely overwhelm the measurements by 5 GPa.
A revolutionary technique for stress and strain measurements under high-pressure
(and temperature) conditions in situ has now become feasible using high-energy X-rays
generated by a synchrotron radiation facility. These new tools are just now being
explored and their limitations defined. The exceptional quality of these tools rests in
the fact that they are directly monitoring the sample. Stresses are measured in the sample.
8
Durham, Weidner, Karato & Wang
Figure 2. Sample assembly in the DIA cubic apparatus, shown with three of the six anvils removed
and with the sample assembly itself, i.e., the cube, shown in cross section. Analytical synchrotron
X-rays enter and exit along a horizontal line through the sample and normal to the section as shown
here. The X-rays pass through the vertical gaps between anvils. Drawing by M. Vaughan.
Different positions in the sample can be isolated to test for stress uniformity.
Measurements of stress and strain can be time resolved with a precision of about one
minute. Strain is also obtained by images of the sample. Distribution of strain with
position and time can be defined. All of this is done with a current accuracy of a few tens
of MPa in differential stress and 10-4 in strain.
Strategies for measuring stress are found through use of the X-ray diffraction or
pressure-sensitive probes (such as ruby fluorescence), and are being made available by
exploiting powerful synchrotron X-ray sources. The basic principle of X-ray piezometers
is that lattice strain is sensitive to elastic strains and not to plastic strains. Changes in
distances between atoms are expected to balance stress fields that are present. These
stress fields may be created by dislocations or external forces, but departures from the
lattice spacing that the material experiences at zero pressure and stress are sensibly
related to the stress field within the sample. Furthermore, the coupling between the strain
and stress is expressed by the elastic moduli. There are two styles of X-ray piezometers,
one based on stress gradient and the other based on strain anisotropy.
Developments in Deformation Experiments at High Pressure
9
Stress gradient. Where the inertia term can be neglected, variation in the stress field
is restricted to obey the mechanical equilibrium relations:
3
∂σ ij
∑ ∂x
j=1
=0
j
Three equations, one for each value of i, represent the vector force balance equations.
The first of these equations is:
∂σ 11 ∂σ 12 ∂σ 13
+
+
=0
∂x1
∂x 2
∂x 3
Thus, if there is a variation of the normal stress, σ11 , with x1 , there must also be a
variation of the shear stresses, σ12, in the x2 direction and/or a variation of σ13 in the x3
direction. The peak magnitude of the shear stress is related to the variation of the normal
stress (or pressure) and the length scales of two types of stress. This has been exploited in
the diamond anvil cell geometry by measuring the radial gradient of pressure to deduce
the axial (along the axis of the diamonds) variation of shear stress. This is illustrated in
Figure 3. The shear stress, τ, vanishes along the center plane of the sample and along the
central axis of the cell where ∂P/∂r is zero. The maximum shear stress is at the samplediamond interface and is given by
τmax=(∂P/∂r)max d/2
where d is the thickness of the sample. This technique was developed by Sung et al.
(1977) and was further used by Meade and Jeanloz (1988a,b; 1990). These experiments
relied on ruby dust distributed across the sample as a pressure marker for delineating the
Figure 3. Illustration of stress field in a diamond-anvil cell. The radial distribution of
pressure, illustrated at the top, is coupled to the shear stress distribution along the axial
direction. Force couples indicate the sense of the shear stress distribution within the cell.
10
Durham, Weidner, Karato & Wang
pressure gradient. As Wu and Bassett (1993) illustrate, the pressure marker can also be an
X-ray diffraction standard such as gold. The objective is to measure the radial pressure
profile to define the shear-stress distribution. Often, the peak shear stress is taken as the
yield point of the solid.
Stress gradients coupled with deviatoric stress affect the diffraction peak shape as
described in detail by Weidner(1998) and Weidner et al. (1998). In this case,
heterogeneities in the stress field broaden the diffraction peaks. Since the length scale of
stress variation is the same for normal and shear stress, then the magnitude of the
heterogeneity for all stresses is similar. Weidner et al. (1998) model the relationship
between the observed strain (that creates the broadened diffraction peak) and the stress
magnitude and conclude that Young’s modulus is the appropriate coupling elastic
modulus, or:
σdifferential = E εdifferential
where the differential strain is deduced as the “strain broadening” contribution to the
peak profile. Since for the peak-broadening measurement, the full-width, half-height of
the diffraction peak is used to define the strain broadening and represents roughly twothirds of the scattering diffraction planes, then most of the sample must be experiencing
the inferred shear stress. Deviatoric stress can originate from dislocations, from
heterogeneities in material or properties, or any number of sources. Peak-broadening
analysis is often used in commercial applications to determine stresses induced by
processing. For example, internal stress generated by welding is often analyzed by this
method. In high-pressure studies, the stress is most commonly generated by compressing
powdered samples (e.g., Chen et al. 1998). As a rheological tool, this approach resembles
an indentation experiment where the indenter is the same material as the sample. Weidner
et al. (1994) studied the high-temperature and -pressure strength of diamond with this
method. Chen et al. (1998) studied plastic deformation of olivine, wadsleyite, and
ringwoodite. Flow can also be studied with this piezometer, with the constraint that this
becomes a relaxation experiment (Weidner et al. 2001).
Strain anisotropy. A uniaxial stress will introduce elastic strains in the sample. The strain
parallel to the axis of compression will generally be larger than the strain perpendicular.
The material’s elastic moduli quantitatively define the relationship between these strains
and the imposed stresses. X-rays sample the distance between lattice planes (called dspacing) whose normals are parallel to the diffraction vector, which is the bisector
between the incident X-ray and the detector. The lattice spacing is insensitive to strain
history (plastic strain) but reflects the elastic strain field. Figure 4 illustrates a Debye ring
that would be observed from a powder sample in a stress field using a monochromatic Xray beam and an area detector. The lattice spacing is related to the distance from the
center of the image to the Debye ring through Bragg’s law. For the angle Ψ of zero, the
lattice spacing reflects that measured parallel to the x1 direction. For Ψ of ninety degrees,
the X-rays are sampling grains aligned parallel to the x2 direction. The strains deduced
from these measurements compared with ambient conditions are related to the stresses
by:
3
3
ε ij = ∑ ∑ sijklσ kl
k =1 l =1
where s is the elastic compliance tensor.
In the actual case, more than one set of diffraction planes will produce Debye rings
similar to the one illustrated here. Each will be approximately circular in shape, but
distorted by the stress field. The amount of distortion may be different if the elastic
Developments in Deformation Experiments at High Pressure
11
Figure 4. X-ray diffraction line on a two-dimensional detector from a sample under
stress. The distance from the center to the ring defines the scattering angle, 2θ , and
hence the lattice spacing through Bragg’s law. The presence of the stress field
distorts the Debye ring from its normally circular shape. The usual stress field is
axisymmetric, with σ2 = σ3. The diffraction vector at ψ = 0 lies at an angle θ from
the direction of σ1. θ is typically only 3 to 4 degrees.
compliance tensor is not isotropic. To evaluate this in detail, we need to be careful to
account for the crystal orientation when defining the compliance tensor. Thus, in a nonhydrostatic stress field, crystallographic planes with different orientations change their
spacing by different amounts. For cubic crystals, the relationship between the Ψ
dependence of strain and the differential stress, ∆σ, is given by:
εhkl(Ψ) = ∆σ /3 {1 – 3cos2 Ψ} {(S11 – S12) [1 – Γ(hkl)] + S44 Γ(hkl) / 2}
Each diffraction peak corresponds to a set of lattice planes defined by the Miller
index, hkl. In an anisotropic crystal, the elastic moduli will depend on the particular set of
planes that are considered. The function, Γ(hkl), varies from 0 to 1 and accounts for this
variation as it multiplies the different elastic compliance moduli, Sij. Thus, for a particular
diffraction plane, hkl, the right hand bracket contains a constant value. Then ε varies in a
fixed manner with the angle, Ψ.
The above expression is based on the assumption that each grain experiences the
same stress field, or the Reuss condition. A similar relation can be derived by assuming
that each grain satisfies the Voigt condition of iso-strain. In this case the equation is
altered by replacing the elastic shear compliances, S44 and (S11 – S12) by their Voigt
equivalent. This relation for cubic symmetry is generalized by Singh et al. (1998) for all
crystal symmetries.
This phenomenon gives rise to a couple of strategies for a piezometer. Stress has a
manifestation even for a single value of Ψ. For an elastically anisotropic material, the
strain for a given crystallographic plane will depend on the orientation of crystal with
respect to the applied stress and the stress magnitude. Therefore by measuring the strain
for several diffraction peaks corresponding to different orientations, one can determine
the stress magnitude using the known elastic constants of the crystal. Such a
measurement can now be done in a multianvil high-pressure apparatus using a
synchrotron X-ray source (Weidner et al. 1992). The accuracy relies on knowing the
elastic anisotropy at the pressure/temperature conditions of the experiment.
A more robust strategy comes from mapping out the d spacing as a function of Ψ.
The values of the elastic moduli are still required, but the details of anisotropy are less
critical to yielding precise measurements. With a precision of 10-4 for the lattice spacing
12
Durham, Weidner, Karato & Wang
and a typical elastic modulus of 200 GPa, it should be possible to resolve differential
stresses of ~20 MPa. Singh has reduced the elasticity equations to a closed-form solution
for several crystal symmetries and has given a concise summary (Singh et al. 1998).
This methodology was first applied in a diamond-anvil cell by Kinsland and Bassett
(1977) by passing the X-ray beam through a Be gasket, perpendicular to the axis of the
diamonds. The multianvil system has not been accessible by this technique because the
anvils themselves cast a shadow and limit the range of Ψ where the diffracted signal can
be observed. However, we have recently experimented with cubic BN anvils. These are
nearly as hard as diamond but are about an order of magnitude less expensive.
Furthermore, they are transparent to X-rays. Thus, by using these anvils it is now possible
to observe diffraction spectra in any plane relative to the incident beam.
Here we have illustrated this tool with monochromatic X-rays and angle dispersive
measurements. Energy dispersive methods are equally valid for these measurements and
are sometimes advantageous. Spectra need to be collected for different values of Ψ
through collimators that fix the two-theta value. Weidner’s group has designed and used a
conical slit system for this purpose as illustrated in Figure 5. Two concentric cones,
whose angle is the desired diffraction angle for white, energy dispersive diffraction,
create the slit itself. Solid-state detectors with energy discrimination are placed at specific
values of Ψ behind the slit system. Each detector is calibrated independently. Lattice
spacings are used as illustrated above to define the differential stress.
Figure 5. Conical slit assembly in use at National Synchrotron Light Source. The cone defines the
two-theta angle and is constructed by two concentric cones with this angle. The X-rays that pass
between the two cones originate from the sample at that angle as all other X-rays are blocked by the
cones. Multiple energy discriminating detectors, located behind the slit assembly, record the X-ray
signals. A YAG crystal located inside the slit assembly fluoresces in the X-ray beam, emitting
visible light where an X-ray photon passes. The visible light is magnified by a microscope assembly
and recorded by a digital camera.
Strain rate measurement
Synchrotron X-rays can also be used to measure sample length. Direct images of the
sample can be obtained using an incident beam whose dimensions are larger than the
sample. Platinum or gold foils above and below the sample are easily viewed. The X-ray
image is obtained by projecting the X-ray onto a fluorescent screen that is viewed with a
CCD camera though a magnifying system. A typical image is shown in Figure 6. The
Developments in Deformation Experiments at High Pressure
13
horizontal lines in this image are gold discs placed at the ends of the sample. The black
edges are the shadows of the opaque tungsten carbide anvils. Cubic boron nitride anvils
are transparent and can be used to view the entire sample because they do not cast such a
shadow. Figure 5 illustrates the position of the microscope for these measurements. The
fluorescent screen is located on-axis inside the conical slit system. We determine that, by
comparing two images, strains of 10-4 can be measured. For images taken 100 seconds
apart, this allows resolution of strain rates of 10-6 s-1.
Figure 6. X-ray shadowgraph of MgO sample during compression. The horizontal
dark lines are images of gold foil that bound the sample whose length is ~1 mm. The
horizontal dimension is defined by the opening between the high-pressure anvils. The
shadow across the top is caused by a platinum foil wrapped around the end-plug.
The X-ray image yields the total strain of the sample, which is the sum of the elastic
and plastic strain or:
εob = εel + εpl
In an experiment both the elastic and plastic components may vary with time. Thus,
to deduce the plastic strain rate, it is necessary to account for the variations of all strain
components. The elastic strain, however, is reflected in the diffraction pattern and can be
represented by a unit cell length, a. This cell-edge length is obtained from analysis of all
of the diffraction peaks recorded with a diffraction vector parallel to the sample axis. In
elastically isotropic samples, there is no ambiguity in defining this value. Anisotropic
samples in a differential stress field will exhibit lattice spacings that reflect the aggregate
properties. In this case, an average cell dimension based on several diffraction peaks
should be obtained. We define the Kung ratio, Rk, as:
Rk = (d/d0) / (a/a0)
where d is the length of the sample measured with an image and a is a unit cell length
measured from diffraction; the zero subscript refers to a reference value such as ambient
conditions. This ratio, which is normalized to one, represents the length of the sample in
the units of the cell dimension. This number is proportional to the number of unit cells
that make up the length. The Kung ratio remains constant if all deformations are elastic.
Thermal expansion, pressure, and stress will not change the Kung ratio if the sample
responds elastically. Plastic strain changes the Kung ratio. In this case the number of unit
cells that define the length changes. Thus, the Kung ratio can be used as the proxy for
plastic strain, or:
εpl = [Rk(2) – Rk(1)] / Rk(1)
where (1) and (2) represent two states of the material.
Strain is not always uniform in the sample. Visual tools as described here provide
opportunity to map the strain field as a function of space and time during the experiment.
Vaughan et al. (2000) demonstrate that one can embed strain markers in a sample during
preparation and view the position of these markers during the experiment. Figure 7, from
this work, shows a grid made from a TEM gold sample grid, in the sample at different P,
T conditions. The sample was a sintered olivine cylinder. The cylinder was sliced into
14
Durham, Weidner, Karato & Wang
two halves, the grid inserted, and the sample halves placed back together. The sample
assembly included hard end-plugs in a DIA apparatus, so as to produce shortening during
compression. The grid size is initially about 70 microns. With these images, it is possible
to map the strain as a function of position in the sample and as a function of time by
comparing successive images.
Figure 7. A TEM gold grid embedded on the central plane of a cylindrical sample.
Images at different conditions illustrate different amounts of strain. The initial
pressurization distorted the grid because surfaces of sample assembly parts were not
perfectly flat. Subsequent strain can be mapped by following specific strain markers.
MODERN TECHNIQUES FOR DEFORMATION AT HIGH PRESSURES
In current technology, deformation at P > 10 GPa is the realm of anvil devices. We
describe in detail four methods currently in use or in active development: (1) sample
assembly modification to hydrostatic multianvil devices, (2) sample assembly
modification to the DAC, (3) the D-DIA, and (4) the RDA.
Sample assemblies
Sample assemblies used in these machines share several common features of those
used in hydrostatic devices, serving common requirements such as high pressure, high
temperature, and control of chemical activity in the sample. With the exception of the
specialized environment of the DAC, these common features include a pressure medium,
such as soft fired pyrophyllite, ZrO2, MgO/5%Cr2O3, castable ceramic or epoxy; a
furnace, for example, metal foil, LaCrO3, or graphite; and a soft medium surrounding the
sample itself, for example, BN, MgO, NaCl, or glass (the latter two occasionally as
liquids). Descriptions of such assemblies can be found in, e.g., Kawai and Endo (1970),
Liebermann and Wang (1992), Walker(1991), and Weidner et al. (1992). For control of
chemical environment the sample can be encapsulated along with buffering compounds
(see e.g., Rubie et al. 1993).
Non-hydrostaticity is introduced by introducing hard platens that impart shear or
Developments in Deformation Experiments at High Pressure
15
normal stresses to the sample. Several of the methods for achieving this are compared in
Figure 8. They provide axisymmetric deformation in the multianvil devices (Figs. 8a-8c),
simple shear in the multianvil (Fig. 8d), axisymmetric deformation in the DAC (Fig. 8e),
and rotational (approximately simple shear) deformation in the RDA (Fig. 8f). We
compare these assemblies briefly here and in more detail below.
Figure 8. Starting (P = 0) configurations of sample columns used for deformation experiments at
very high pressures. For simplicity, details such as sample encapsulation, thermocouples, and
furnace electrodes are not shown. (a) Stress relaxation. The sample is loaded elastically during
pressurization. The furnace and outer and inner pressure media shown here also apply to (b)-(d). (b)
Constant stress/strain rate tests. The porosity within the crushable ends of the column takes up most
of the column shortening during pressurization. (c) Specialized version of (b), with a fluid in the
column assuring strictly hydrostatic loading of the sample during pressurization. (d) Simple shear,
high strain. (e) Column for deformation in the diamond-anvil cell. Use of a separate piston is
optional, i.e., the diamond anvils can also act as pistons. In some applications, a ruby piston can also
serve as a differential stress gage. (f) Assembly for the rotational Drickamer apparatus, also shown
in Figure 13.
The adaptations to the DIA and 6/8 discussed in the next section use a column such
as shown in Figure 8a, in which the source of deviatoric stress in the sample is the closure
of the anvils during initial pressurization. At the scale of Figure 8a, the shortening of the
column (elastic plus inelastic) is approximately 20% of the column length (Durham and
Rubie 1998), so the “crushable” material at the ends of the column, present to protect
thermocouple wires and furnace electrodes, has a porosity somewhat less than 20% of the
column length. To avoid deforming the sample during pressurization in devices capable
of deforming at fixed pressure (currently only the D-DIA) one uses a column such as in
Figure 8b, where the length of crushable material is sufficient to absorb all anvil
displacement during pressurization When a strictly hydrostatic environment is required
during pressurization, such as when the sample is a fragile single crystal, it is also
possible to create a liquid cell around the sample before the pressurizing load is applied
(Fig. 8c) (Durham, unpublished results).
16
Durham, Weidner, Karato & Wang
Figure 8d shows the method for converting axial displacement to high-strain, simpleshear deformation in the multianvil cell (Karato and Rubie 1997). As discussed below,
this design was first introduced as a method for imposing larger strains in the stressrelaxation mode of the hydrostatic multianvil device. However, it can also be used in the
D-DIA, presumably with slightly more crushable material at the ends, to carry out
constant stress or constant strain-rate tests to very high simple-shear strains.
The very simple deformation assembly for the DAC (Fig. 8e) reflects the greatly
reduced available volume and the fact that laser heating can be used in place of a
resistance heater to achieve local heating of the sample. Finally, the assembly for the
RDA (Fig. 8f) is the rotational analog to the axisymmetric columns in Figures 8a-8c.
Because rotational deformation and pressurization are entirely decoupled in the RDA,
there is no need for crushable material in the cell.
Modifications to the sample assembly in a multianvil press
Fujimura et al. (1981) developed one of the first techniques for converting the
environment of the 6/8 device from hydrostatic to non-hydrostatic, thus making the 6/8
device a sort of deformation apparatus. By making the elastic character and/or
mechanical strength of a sample assembly anisotropic, one can create a deviatoric stress
in the assembly (e.g., Fig. 8a). This deviatoric stress can be relaxed by plastic flow upon
heating. This technique has been used in numerous studies (Bussod et al. 1993; Durham
and Rubie 1998; Fujimura 1989; Green et al. 1990; Karato et al. 1998; Karato and Rubie
1997; Weidner et al. 2001; Weidner et al. 1998). One typically achieves the anisotropic
strain by embedding a rigid sample column in the compliant pressure medium. As the
assembly is pressurized at room temperature (where most relevant geologic materials are
very strong), a large deviatoric stress is generated on the sample. When temperature is
increased, the sample softens and begins to flow, and the stress gradually relaxes. The
analysis by Karato and Rubie (1997) (see also Durham and Rubie [1998]) showed that
the mode of deformation in most cases is this “stress relaxation,” i.e., the magnitude of
deviatoric stress changes significantly within a single experiment.
This technique has several limitations. First, pressurization and plastic deformation
are not completely separated; therefore, deformation during pressurization likely occurs
unless special care is taken in sample assembly to absorb the initial-stage shape change of
sample assembly (Durham and Rubie 1998; Karato and Rubie 1997). Second,
deformation by this method is “stress relaxation,” and the interpretation is complicated
because of the change in stress that could cause the change in deformation mechanisms.
Third, the amount of inelastic displacement is limited to the amount that can be imposed
elastically during pressurization, usually well under 100 µm. By using a thin sample
sandwiched between two pistons cut at 45°, Karato and Rubie (1997) were able to
convert this very small displacement to relatively large strains (γ ≈ 1-2) (Fig. 8d), yet the
maximum strain is not large enough to obtain steady-state fabrics. Quasi-constant
displacement rate tests can also be made with multianvil apparatus through the
continuous movement of the hydraulic ram (Bussod et al. 1993; Green et al. 1990),
although one must realize that pressure is increasing steadily and significantly in such
tests.
The multianvil deformation technique has been applied to ~25 GPa, ~2000 K with a
sample dimension of ~1-2 mm diameter and ~2-4 mm long, or a thin (~0.2 mm) disk
sample with a similar diameter (Chen et al. 1998; Cordier and Rubie 2001). Sacrificing
pressure for sample size, Green et al. (1990) deformed 3-mm diameter x 6-mm length
samples in a 6/8 device. The major advantage of this technique as compared with a
diamond-anvil cell technique is that rheological properties can be investigated under
more homogeneous high temperature and pressure conditions with a better-controlled
Developments in Deformation Experiments at High Pressure
17
chemical environment. Also, because relatively large samples can be used, microstructural evolution during deformation and its effects on rheology can be investigated
(Green et al. 1990, 1992; Karato 1998a; Karato et al. 1998).
The limited plastic deformation is a serious shortcoming of the method. The
displacement available is generally not sufficient to probe important rheological
questions about lattice preferred orientation and strength in the steady state. Second,
because the stress magnitude changes during an experiment, dominant mechanisms of
deformation may change in a single run, making interpretation of mechanical data and
microstructures difficult. This last point, however, may turn out to be an advantage for
the study of rheology at a very small strain-rate (small stress) that is relevant to Earth.
Diamond-anvil cell
The highest pressures obtainable in the laboratory are in the diamond-anvil cell. This
device loads the sample uniaxially and, thus, naturally produces a deviatoric stress
environment. The first examinations of differential stresses in the diamond cell were
motivated by the need to produce a hydrostatic environment. Use of the diamond cell as a
deformation device was pioneered by Kinsland and Bassett (1977) and Sung et al. (1977).
The former used X-ray transparent gaskets and passed the X-ray beam through the
gaskets, perpendicular to the diamond-cell axis. Stress was measured from the shape of
the Debye rings as discussed above. The measurements provided the first estimates of the
room temperature strength of MgO at high pressures. Sung et al. (1977) developed the
methodology for using pressure gradients to deduce shear stress. In this case, a sample is
sandwiched between two single crystals of diamond, and radial distribution of pressure is
determined by measuring the shift of fluorescence lines of ruby crystals located at various
points in the sample space. Then, assuming that the pressure gradient is supported by the
sample strength, one can estimate the sample’s strength from the equation for force
balance. This technique has been used extensively by Meade and Jeanloz (1988a,b; 1990)
at room temperature to the pressure of ~40 GPa.
Chai and Brown (1996) developed a diamond-cell piezometer based on the splitting
of the ruby fluorescence lines. They find that the R2 line reflects the average stress, that
the splitting between the R1 and R2 lines is sensitive to the differential stress, and that the
character of the sensitivity depends on whether the load is applied parallel to the c-axis or
the a-axis. This phenomenon suggests experiments that can be done with a diamond-anvil
cell at room temperatures using single crystal ruby as the pressure/stress calibrant.
The major advantage of diamond cell studies is the high pressures that can be used
under which a sample can be plastically deformed Wenk et al. (2000) recently deformed
Fe at ~220 GPa). However, there are major limitations with this approach: homogeneous
heating is difficult in a diamond-anvil cell, and almost all previous results were obtained
at room temperature. In addition, the strain rate is not well constrained even though
deformation is likely to be time-dependent in these tests. Furthermore, the sample space
is so small that some important effects such as grain-size sensitivity of strength are
difficult to measure.
Poirier et al. (1981) and Sotin and Poirier (1990) report on a sapphire cell that
enables larger sample volumes than a diamond cell and is more versatile for higher
temperature studies. They calibrate the stress field from the characteristics of the loading
system and observe the movement of strain markers in the sample. This enables them to
define strain rate along with the stress determinations.
Deformation-DIA
Samples of MgO and polycrystalline tantalum were deformed in the deformationDIA (D-DIA) at the National Synchrotron Light Source (NSLS) in February 2002. This
18
Durham, Weidner, Karato & Wang
marked the first materials test ever above P = 10 GPa in which virtually all relevant
rheological independent variables were under full and independent operator control. We
provide here a detailed description of this new apparatus.
The D-DIA (Wang et al. 2002) is a modification of the cubic anvil device known as
the DIA (e.g., Osugi et al. (1964) and Shimomura et al. (1985)) a single-stage, wedged
guide block machine. The hydrostatic DIA usually operates to ~15 GPa and ~2000 K.
The D-DIA modification gives independent displacement control to one pair of opposing
anvils, reducing the cubic symmetry of the DIA to tetragonal, thus allowing high-strain
deformation experiments to 15 GPa.
The original DIA consists of
symmetric upper and lower guide
blocks, four wedge-shaped thrust
blocks, and six anvils, as indicated in
Figure 9. Four of the anvils are
mounted on the inside faces of the
thrust blocks and the other two are
mounted on the inside central faces of
the guide blocks, the square fronts of
the anvils thus defining a cubic
volume at the center of the apparatus.
The operation of the DIA can be
visualized by recognizing that the
eight inner inclined surfaces of the
guide blocks define a virtual regular
octahedron whose dimension changes
with the separation distance between
the guide blocks. The six anvils are
aligned with the six apices of this
virtual octahedron, and parts are
machined such that the (fixed)
distance from anvil face to associated
apex of the virtual octahedron is
precisely the same for each anvil. As
the guide blocks close or open, all Figure 9. The original hydrostatic DIA apparatus. A
displacements of the anvils are single hydraulic ram (large arrows) drives 2 wedgeguide blocks toward one another. The 6 anvils
symmetric about the center. The force type
(truncated pyramids on the guide blocks and wedge
of a single hydraulic actuator (or ram) blocks; only 5 are visible here) are thus driven toward
applied along the vertical axis in one another, symmetrically compressing a cubic
Figure 9 closes the guide blocks and, sample assembly (not shown) at the very center.
thus, compresses the cell hydrostatically. Ram forces of 100 T are typically required with various truncation sizes
(typically 3 to 6 mm) to reach maximum pressures near 20 GPa (see e.g., Shimomura et
al. (1992) and Utsumi et al. (1992)). By the cubic symmetry of the DIA assemby, this
force is divided into three equal portions along each of the three orthogonal directions
defined by opposing pairs of anvils.
Independent control of one anvil pair in D-DIA is provided by two additional
hydraulic actuators—called differential rams—within each guide block (Fig. 10). These
rams may be thought of not as parts that are added, but rather as a central portion of each
guide block that has been cut free so that it can move vertically. A small hole is
introduced into the side of each guide block to allow access for hydraulic fluid to each
ram, and seals for the pressurized plenum below each differential ram are provided. The
Developments in Deformation Experiments at High Pressure
19
conical shape of the differential rams is rather unimportant; it leaves more steel on the
guide blocks and makes them slightly stronger. Note that in the D-DIA modification, the
guide blocks not only support the forces confining the thrust blocks, they now have also
become thick-wall pressure containers for the hydraulic fluid driving the differential
rams. For this reason, the guide blocks of the D-DIA are complete discs rather than
crossed steel members as in the original DIA (Fig. 9).
Figure 10. Computer-generated 3/4 section through the D-DIA. Comparing to Figure 9, the
guide blocks are now full cylindrical shapes, but the four wedge blocks and six anvils are
identical to those in Figure 9. The two differential rams, the large conical pieces within the
top and bottom guide blocks, move independently of the guide blocks themselves, and can
thus generate a deviatoric state of stress in the assembly at the center (not shown). In actual
operation, as the differential rams displace inward toward the center, the guide blocks move
apart to allow the four wedge blocks to displace outward. The displacement rates are
independently controllable and are usually chosen such that the net volume of the sample
assembly, and therefore the confining pressure, remains constant.
The differential rams have a (pressurized) diameter of 89 mm and are therefore
capable of generating a force of 125 T at a hydraulic pressure of 0.2 GPa, sufficient to
overcome the force of the main ram (up to 33% of the main ram maximum of 300 T) and
apply differential force even at the highest confining pressure. Finite element analysis of
the D-DIA guide blocks show that the maximum octahedral stress occurs on the inner
bore in a “typical” operation where the differential load is 67% of the main ram load
(presuming some force is needed to overcome packing friction and sample strength)
(Wang et al. 2002). The stress at the base of the inclined surface of the guide block, the
next highest point of stress concentration in the guide block, is roughly 90% of the stress
in the bore. Note that in the original DIA with very large notches cut in the guide blocks,
the stress concentration at the base of the inclined surfaces is twice that of the D-DIA
under a similar main ram load.
The key point of independent control of the differential rams is that deformation can
be imposed without the necessity of increasing confining pressure (which has been the
20
Durham, Weidner, Karato & Wang
long-standing limitation of deformation studies in 6/8 multianvil and diamond-anvil
assemblies). In normal operation, the D-DIA sample is brought to run conditions of P and
T with differential rams fully withdrawn in the same manner as the hydrostatic DIA.
Advancing the differential rams then introduces a non-cubic shape change to the
assembly. Furthermore, by simultaneously draining hydraulic fluid from the main ram at
an appropriate rate, the four side anvils retract, and the total force of the main ram as well
as the volume of the sample cell can be held constant. In the synchrotron X-ray beamline)
pressure itself can be monitored and, accordingly, serves as the process variable. The
operation at NSLS in February 2002, showed that this procedure gave very satisfactory
control of pressure over very large displacements of the differential rams (Fig. 11).
The differential rams are driven by high-precision syringe pumps, and their
velocities are controllable from approximately 10-7 to 10-2 mm/s. When both rams are
driven symmetrically, this translates to a strain rate on a typical 1-mm-length, 1-mm3volume sample of 2 x 10-2 ≥ ε̇ ≥ 2 x 10 -7 s-1. These rates are much faster than most
relevant rates in geology, but as with all experimental rock mechanics work—for which
these rates are typical—it is the human time scale and not the geologic time scale that
governs the duration of an experiment. As with most experimental studies of rock
deformation, appropriate scaling analysis is critical to apply these results to Earth.
Differential ram displacements of >1 mm are possible, so strains in pure shear
compression can approach 1. Deformation in simple shear, using 45˚-cut pistons (see
“Sample Assemblies,” above) is also possible, allowing for much higher strains, at some
cost to sample volume. Finally, the D-DIA is capable of extensional, as well
compressional, deformation, because the sense of motion of the six anvils can be
reversed.
Rotational Drickamer apparatus (RDA)
Large strain deformation experiments can be conducted in the torsion mode,
allowing a detailed study of deformation microstructures as well as rheology. Paterson
and Olgaard (2000) described such an apparatus in which a torsion actuator is attached to
a gas-medium high-pressure apparatus. This apparatus contains an internal load-cell,
making possible a precise measurement of stress. However, the maximum pressure of
operation is limited to ~0.5 GPa. Deformation experiments can be conducted at higher
pressures by twisting a thin sample between anvils. Bridgman (1935a; 1937) was the first
to apply the technique to 5 GPa with an apparatus that consisted of two fixed anvils
bearing on a flat, rotating anvil. The design was later modified by Griggs et al. (1960) to
remove the intermediate anvil and rotate one of the anvils instead. Similar attempts were
made in the late 1960s at room temperatures and at pressures to 7 GPa (Abey and
Stromberg 1969; Riecker and Seifert 1964) in which a sample was sheared, and the
strength was determined by the measurements of torque on an anvil needed to deform the
sample.
Figure 11 (next page). Results plotted vs. run time from a D-DIA experiment on a stacked sample
of tantalum (starting length 1 mm) and MgO (starting length 1.5 mm), both polycrystalline. (a) Five
traces showing, from top to bottom: measured hydraulic oil pressure in the top and bottom
differential rams (scale on right-hand axis); combined centerward displacement of the two rams as
measured by displacement transducers outside the sample assembly; shortening of the MgO portion
of the sample; and shortening of the Ta component (scale for last three on left-hand axis). Sample
length vs. time was measured directly using X-radiography. (b) Detail of the sample shortening, also
indicating the strain rates (displacement rates normalized by sample length). The run was conducted
by pumping hydraulic fluid to the differential rams at two different constant rate (hence deformation
steps (1) and (2)), while draining fluid from the main ram in order to keep pressure at a constant
level of 10±1 GPa. Note that shortening rates of the samples were also approximately constant. The
steady increase in differential ram pressure during both deformation steps is the result of steadily
increasing friction on the gaskets squeezed between the differential and side anvils.
Developments in Deformation Experiments at High Pressure
21
Yamazaki and Karato (2001) have improved these techniques by modifying the
Drickamer-type high-pressure apparatus (Fig. 12). This apparatus allows large rotational
shear deformation of a sample at higher pressures and temperatures. A Drickamer
apparatus consists of a pair of anvils in a cylinder with a gasket material between
the anvils and a thin disk of sample squeezed between the anvils to reach high pressure
22
Durham, Weidner, Karato & Wang
Figure 12. Sketch of a rotational Drickamer apparatus (Yamazaki and Karato 2001).
Figure 13. Sample assembly for a rotational Drickamer apparatus. A thin
disk of sample is sandwiched between two anvils. A sample is cut into two
pieces and W5%Re and W25%Re foils are inserted vertically; these act
both as strain markers and as a thermocouple. The change in shape of these
foils can be measured after an experiment or during an experiment through
X-ray absorption. A small cylindrical W3%Re at the center acts as an
electrode for the heater and as one component of the thermocouple with a
W25%Re foil.
(Fig. 13). The gasket provides an extra support for the anvils and, as a result, this
apparatus yields significantly higher pressures than does a Bridgman apparatus without a
gasket (Perez-Albuerne et al. 1964; Prins 1984). Because much of the force is supported
by the gasket and the sample itself, the cylinder does not support a large load.
Consequently, it is possible to make holes in the cylinder to provide a path for X-rays; the
Developments in Deformation Experiments at High Pressure
23
thermocouple leads can be taken through these holes. The X-ray path feature is important
for high-resolution stress and temperature measurements, both of which are critical for
rheological studies. The Drickamer apparatus has recently been used for the in situ
measurements of static properties using a synchrotron X-ray facility under conditions up
to ~35 GPa and ~2000 K (Funamori and Yagi 1993).
In the Rotational Drickamer Apparatus (RDA) modification, one of the anvils is
fixed with the frame, whereas another is attached to a rotational actuator. A gearbox with
an ac-servo motor provides controlled rotation of one of the anvils. A sample is first
pressurized (and heated), and then the rotational actuator is started. The sample is twisted
between the two anvils, and the geometry of deformation is rotational shear. The strain
(and strain rate) increases from zero to the maximum value, which is determined by the
motor rotation speed and sample thickness. With the current motor and gear combination,
a shear strain rate from ~0 to ~10-3 s-1 can be realized. This design has three advantages.
First, the motion of the anvil for deformation is orthogonal to the motion of the anvil for
pressurization. Therefore, pressurization and deformation can be clearly separated.
Second, unlike axisymmetric deformation, the shape of the sample does not change
appreciably during rotational deformation, allowing very large strain deformation and
making possible studies of microstructural evolution. Third, because of the radial
gradient in strain (strain rate), microstructural evolution and/or rheology at different
strain-rates can be investigated in a single run.
The diameter of the anvil tip in the current RDA is 4 mm (sample diameter is ~1-2
mm, thickness ~0.2-0.4 mm). With a load of ~30 T, a pressure of ~15 GPa can be
generated using tungsten carbide anvils. The maximum pressure is determined primarily
by the strength of anvil materials (and the diameter of sample space) and is expected to
be more than ~30 GPa when a harder material such as sintered diamond or cubic-BN is
used. The rotational Drickamer apparatus constructed at Yale University has been tested
at P = 12 GPa, T = 1473 K. A polycrystalline sample of MgO was deformed homogeneously to the maximum shear strain of ~3. Two methods can be employed to estimate the
stress magnitude. First, X-ray diffraction techniques can be used for deformation
experiments with the RDA in a synchrotron radiation facility in the manner discussed
previously. At this writing, such experiments are planned but have not yet been carried
out. To adopt the diffraction technique for rotation geometry, a conical shape window is
made in a cylinder to collect diffracted X-rays for a range of angles (Fig. 14b). One
complication with this apparatus is the radial variation of stress. This is inevitable in the
torsion test (e.g., Paterson and Olgaard 2000); Fig. 14a). In measuring the stress
magnitude by X-ray diffraction, one must consider the spatial resolution. Important
factors that control the spatial resolution are the cross-sectional area of the X-ray beam
and the 2θ value used for diffraction (with energy dispersive mode). A smaller beam size
and a larger 2θ are preferred for a better spatial resolution of stress measurements.
However, a smaller beam size results in a weaker signal. Also the 2θ values are limited
by the energy of X-ray and the d-spacings as well as the geometry of the apparatus (we
will use ~10-50 × 10-50 µm2 beam [slit] size and 2θ ≈ 8-10°). To allow for control of
2θ, the apparatus is set on a mobile stage (with a remote control through a stepping
motor) whose angle and position with respect to the beam line can be adjusted to
optimize the conditions. The resolution of the stress measurement is also dependent on
the detector. We anticipate that a spatial resolution of 100-200 µm and the stress
resolution better than 0.1 GPa will be obtained by this technique. The use of a hollow
specimen could help solve the problem of stress heterogeneity (Fig. 14d).
We note that this technique not only allows us to estimate the stress magnitude, but
also provides information as to the grain-scale stress (strain) distribution in a
polycrystalline material (either homogeneous stress or homogeneous strain (Funamori et
24
Durham, Weidner, Karato & Wang
..
Figure 14. (A) Stress distribution in a sample in a torsion test (n is stress exponent, γ = Aτ n ).
(B) X-ray diffraction geometry. Stress magnitude can be estimated from the difference in d-spacing
collected from two different diffracted beams through two windows that are cut at 45o with respect
to the vertical axis. (C) A top view of a sample and the X-ray path. In order to determine the stress,
X-ray diffraction must occur in a region of a sample where the shear direction is normal to the X-ray
beam. (D) The geometry of a hollow sample, where stress heterogeneity is less.
al. 1994). Such information is useful in understanding the physics of deformation of
polycrystals (Kocks 1970).
Another method is to use dislocation density as a stress indicator. This technique can
be used when the deformation mechanism is dislocation creep and when stress has been
held at a constant value for at least a few percent strain. The method does not work well
if stress is inhomogeneous. We have recently used this technique for olivine and are able
to estimate the stress magnitude with an uncertainty of ~10-15% (Jung and Karato 2001;
Karato and Jung 2002 in press). The reliability of such measurements for high-pressure
phases is not as high as that for olivine because the calibration curve for dislocation
density vs. stress relationship is not available. However, the relation between dislocation
density and stress is nearly universal,
ρ = α b −2 ( σ / µ ) 2
(ρ: dislocation density, α: a constant of order unity, b: the length of the Burgers vector, σ:
stress, µ: shear modulus (e.g., Kohlstedt and Weathers 1980) We consider that such
measurements should provide at least semi-quantitative data on the stress magnitude.
We will perform both single-layer deformation experiments and two-layer deformation experiments. A single-layer deformation experiment is simpler to interpret, but the
big advantage of a two-layer design is that one can determine the relative strength of two
samples very accurately. Our experience with this apparatus showed that we can deform
our samples to strains up to at least γ ≈ 10 at ~15 GPa. Such a large strain is essential
for the study of deformation fabric.
SUMMARY AND PERSPECTIVES
We have summarized some of the recent developments in quantitative
characterization of stress and strain and the controlled generation of deviatoric stresses
under high pressure (and temperature). These developments will allow us to explore
Developments in Deformation Experiments at High Pressure
25
rheological behavior of Earth materials under much deeper conditions than heretofore
possible. Such studies will, for the first time, provide the critical data sets necessary for
understanding whole mantle dynamics through modeling and/or through the analyses of
observations such as seismic anisotropy. Further technical developments need to be made
to make these techniques robust. First, stress measurements using different techniques
must be benchmarked. Three techniques are currently used to measure stress under high
pressure: (1) Measurement of force using an external load-cell with a low-friction sample
design (Tingle et al. 1993), (2) synchrotron X-ray measurement of orientation
dependence of d spacing (Singh et al. 1998), and (3) inference from dislocation densities
(Karato and Jung 2002 in press). Below ~4 GPa, all of these techniques can be employed
to determine the strength of a standard materials such as olivine. A comparison of results
of these measurements will provide a measure of reliability of each technique. Second, a
number of issues still remain regarding the sample assembly. They include a proper
control of thermodynamic environment, particularly water fugacity (or water content) and
oxygen fugacity. Third, in almost all of these techniques under high pressure, it is almost
inevitable to have deformation during pressurization. An experimental procedure (e.g.,
annealing; a very low strength medium surrounding the sample) must be established to
minimize the effects of initial stage deformation.
It must also be emphasized that the sample preparation and characterization are as
critical as careful mechanical tests. Samples with negligible porosity must be used and
grain size and water content must be measured both before and after each experiment
(control of other thermodynamic variables such as oxygen fugacity is also important; see
Rubie et al. (1993)). In almost all laboratory deformation experiments, conditions are
often close to the boundary between dislocation and diffusion creep (Karato et al. 1998),
and, therefore, precise measurements of grain size are needed to interpret the data. Also,
water content must be measured both before and after each experiment. Water is known
to have significant effects on rheology, but its content in samples is difficult to control
during high-pressure experiments.
Using the results of mechanical tests and sample characterization, dominant
microscopic mechanisms of deformation must be identified. A meaningful comparison of
strength of different materials (or for the same material for different conditions) can only
be made when the deformation mechanism is the same, and the comparison must be
made based on an appropriate scaling. For instance, if the dominant mechanism of
deformation is diffusion creep, then the data must be compared by normalizing with
respect to the grain size.
With the use of these new techniques (combined with careful characterization of
samples and the analyses of deformation mechanisms), we will be able to cast first light
on the rheology of the more than 90% of Earth’s mantle that has not been accessible by
quantitative studies. Results of such studies will be critical to our better understanding of
dynamics and evolution of this planet.
ACKNOWLEDGMENTS
We are most grateful to H. W. Green, II, for reviewing the manuscript. Funding was
provided by the National Science Foundation under award EAR-0135551. Work by
WBD performed under the auspices of the U.S. Department of Energy by the Lawrence
Livermore National Laboratory under contract W-7405-ENG-48. Additional portions of
this work were performed at GeoSoilEnviroCARS (GSECARS), Sector 13, Advanced
Photon Source at Argonne National Laboratory. GSECARS is supported by the National
Science Foundation - Earth Sciences, Department of Energy - Geosciences, W.M. Keck
Foundation, and the U.S. Department of Agriculture. Use of the Advanced Photon Source
26
Durham, Weidner, Karato & Wang
was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of
Energy Research, under Contract No. W-31-109-Eng-38.
REFERENCES
Abey AE, Stromberg HD (1969) 70 kilobar shear apparatus. Rev Sci Instrum 40:557
Borch RS, Green HW II (1989) Deformation of peridotite at high pressure in a new molten cell:
Comparison of traditional and homologous temperature treatments. Phys Earth Planet Inter 55:
269-276
Boyd FR, England JL (1960) Apparatus for phase equilibrium measurements at pressures up to 50 kilobars
and temperatures up to 1750˚C. J Geophys Res 65:741-748
Bridgman PW (1914) The technique of high pressure experimenting. Proc Am Acad Arts Sci 49:627-643
Bridgman PW (1935a) Effects of high shearing stress combined with high hydrostatic pressure. Phys Rev
48:825-847
Bridgman PW (1935b) Polymorphism, principally of the elements, up to 50,000 kg/cm2. Phys Rev 48:
893-906
Bridgman PW (1937) Shearing phenomena at high pressures, particularly in inorganic compounds. Proc
Am Acad Arts Sci 72:45-136
Bridgman PW (1942) Pressure-volume relations for seventeen elements to 100,000 kg/cm2. Proc Am Acad
Arts Sci 74:425-440
Bridgman PW (1948) The compression of 39 substances to 100,000 kg/cm2. Proc Am Acad Arts Sci 76:
55-70
Bridgman PW (1952a) The resistance of 72 elements, alloys, and compounds to 100,000 kg/cm2. Proc Am
Acad Arts Sci 81:65-NN
Bridgman PW (1952b) The Physics of High Pressure. G Bell and Sons, London
Bridgman PW (1952c) Studies in Large Plastic Flow and Fracture with Special Emphasis on the Effects of
Hydrostatic Pressure. McGraw-Hill, New York
Burnley PC, Green HW II, Prior D (1991) Faulting associated with the olivine to spinel transformation in
Mg2GeO4 and its implications for deep-focus earthquakes. J Geophys Res 96:425-443
Bussod GY, Katsura T, Rubie DC (1993) The large volume multi-anvil press as a high P-T deformation
apparatus. Pure Appl Geophys 141:579-599
Carter N.L, Christie J.M, Griggs D. T (1964) Experimental deformation and recrystallization of quartz.
J Geol 72:687-733
Chai M, Brown JM (1996) Effects of static non-hydrostatic stress on the R lines of ruby single crystals.
Geophys Res Lett 23:3539-3542
Chen J, Inoue T, Weidner DJ, Wu Y, Vaughan MT (1998) Strength and water weakening of mantle
minerals, olivine, wadsleyite and ringwoodite. Geophys Res Lett 25:575-578
Cordier P, Rubie DC (2001) Plastic deformation of minerals under extreme pressure using a multi-anvil
apparatus. Mater Sci Eng A309-310:38-43
Davies GF (1995) Penetration of plates and plumes through the mantle transition zone. Earth Planet Sci
Lett 133, 507-516
Durham WB, Rubie DC (1998) Can the multianvil apparatus really be used for high-pressure deformation
experiments? In Properties of Earth and Planetary Materials at High Pressure and Temperature,
Geophysical Monograph 101. M Manghnani, Y Yagi (eds) p 63-70. American Geophysical Union
Eremets M. I (1996) High Pressure Experimental Methods. Oxford University Press
Fujimura A (1989) Preferred orientation of mantle minerals. In Rheology of Solids and of the Earth.
S-i Karato, M Toriumi (eds) p 263-283. Oxford University Press
Fujimura A, Endo S, Kato M, Kumazawa M (1981) Preferred orientation of β−Mn2GeO4. In Programme
and Abstracts, The Japan Seismological Society, p 185
Fukao Y, Obayashi M, Inoue H, Nenbai M (1992) Subducting albs stagnant in the mantle transition zone.
J Geophys Res 97:4809-4822
Funamori N, Yagi T (1993) High pressure and high temperature in situ X-ray observation of MgSiO3
perovskite under lower mantle conditions. Geophys Res Lett 20:387-390
Funamori N, Yagi T, Uchida T (1994) Deviatoric stress measurement under uniaxial compression by a
powder X-ray diffraction method. J Appl Phys 75:4327-4331
Getting IC (1998) New determination of the bismuth I-II equilibrium pressure: A proposed modification to
the practical pressure scale. Metrologia 35:119-132
Getting IC, Spetzler HA (1993) Gas-charged piston-cylinder apparatus for pressures to 4 GPa. In HighPressure Science and Technology. SC Schmidt, JW Shaner, GA Samara, M Ross (eds) p 1581-1584.
American Institute of Physics
Developments in Deformation Experiments at High Pressure
27
Green HW II (2001) Earthquakes at depth and their enabling mineral reactions. Eleventh Annual V.M.
Goldschmidt Conference (abstr)
Green HW II, Borch RS (1989) A new molten salt cell for precision stress measurement at high pressure.
European J Mineral 1:213-219
Green HW II, Houston H (1995) The mechanics of deep earthquakes. Ann Rev Earth Planet Sci 23:169-213
Green HW II, Scholz CH, Tingle TN, Young TE, Koczynski TA (1992) Acoustic emissions produced by
anticrack faulting during the olivine-spinel transformation. Geophys Res Lett 19:789-792
Green HW II, Young TE, Walker D, Scholz CH (1990) Anticrack-associated faulting at very high pressure
in natural olivine. Nature 348:720-722
Griggs DT (1936) Deformation of rocks under high confining pressure. J Geol 44:541-577
Griggs DT (1967) Hydrolytic weakening of quartz and other silicates. Geophys J R Astron Soc 14:19-31
Griggs DT, Turner FJ, Heard HC (1960) Deformation of rocks at 500˚ to 800˚C. In Rock Deformation,
Geol Soc Am Memoir 79. DT Griggs, JW Handin (eds) p 39-104. Geol Soc Am, Boulder, Colorado
Hager BH (1984) Subducted slabs and the geoid: constraints on mantle rheology and flow. J Geophys Res
89:6003-6015
Hall HT (1958) Some high-pressure, high-temperature apparatus design considerations: Equipment for use
at 100,000 atmospheres and 3000˚C. Rev Sci Instrum 29:267-275
Handin JW (1953) An application of high pressure in geophysics: experimental rock deformation. Am Soc
Mech Eng Trans 75:315-325
Heard HC, Carter NL (1968) Experimentally induced "natural" intergranular flow in quartz and quartzite.
Am J Sci 266:1-42
Hill R (1950) The Mathematical Theory of Plasticity. Oxford University Press
Hirth G, Kohlstedt DL (1996) Water in the oceanic upper mantle; implications for rheology, melt extraction
and the evolution of the lithosphere. Earth Planet Sci Lett 144:93-108
Hobbs BE, Ord A (1988) Plastic instabilities: implications for the origin of intermediate and deep focus
earthquakes. J Geophys Res 89:10521-10540
Honda S, Yuen DA, Balachandar S, Reuteler D (1993) Three-dimensional instabilities of mantle convection with multiple phase transitions. Science 259:1308-1311
Houck JC, Hutton UO (1963) Correlation of factors influencing the pressures generated in multi-anvil
devices. In High-Pressure Measurement. AA Giardini, EC Lloyd (eds). Butterworths, London
Jaeger JC, Cook NGW (1976) Fundamentals of Rock Mechanics, 2nd edition. Chapman and Hall
Jiao W, Siliver PG, Fei Y, Prewitt CT (2000) Do deep earthquakes occur on preexisting weak zones? An
examination of the Tonga subduction zone. J Geophys Res 105:28125-28138
Jung H, Karato S (2001) Effect of water on the size of dynamically recrystallized grains in olivine. J Struct
Geol 23:1337-1344
Karato S (1998) Micro-physics of post-glacial rebound. In Ice Age Dynamics. P Wu (ed) p 351-364.
Trans Tech. ???
Karato S (1998a) Effects of pressure on plastic deformation of polycrystalline solids: some geological
applications. In High Pressure Research in Materials Sciences. PY Yu (ed) p 3-14. Materials Research
Society
Karato S, Dupas-Bruzek C, Rubie DC (1998) Plastic deformation of silicate spinel under the transition zone
conditions of the Earth. Nature 395:266-269
Karato S, Jung H (2002 in press) Effects of pressure on high-temperature dislocation creep in olivine
polycrystals. Phil Mag (in press)
Karato S, Riedel MR, Yuen DA (2001) Rheological structure and deformation of subducted slabs in the
mantle transition zone: implications for mantle circulation and deep earthquakes. Phys Earth Planet
Inter 127:83-108
Karato S, Rubie DC (1997) Toward experimental study of plastic deformation under deep mantle
conditions: a new multianvil sample assembly for deformation studies under high pressures and
temperatures. J Geophys Res 102:20111-20122
Karato S-i, Wu P (1993) Rheology of the upper mantle; a synthesis. Science 260:771-778
Kawai N, Endo S (1970) The generation of ultrahigh hydrostatic pressure by a split sphere apparatus. Rev
Sci Instrum 41:1178-1181
Kellogg LH, Hager BH, van der Hilst RD (1999) Compositional stratification in the deep mantle. Science
283:1881-1884
King SD (1995) Radial models of mantle viscosity: results from a generic algorithm. Geophys J Intl
122:725-734
Kinsland GL, Bassett W (1977) Strength of MgO and NaCl polycrystals to confining pressures of 250 kbar
at 25 C. J Appl Phys 48:978-985
Kirby SH, Stein S, Okal EA, Rubie DC (1996) Metastable mantle phase transformations and deep
earthquakes in subducting oceanic lithosphere. Rev Geophys 34:261-306
28
Durham, Weidner, Karato & Wang
Kocks UF (1970) The relation between polycrystal deformation and single-crystal deformation. Metall
Trans 1:1121-1143
Kohlstedt DL, Chopra PN (1994) Influence of basaltic melt on the creep of polycrystalline olivine under
hydrous conditions. In Magmatic Systems. Intl Geophysics Series, Vol. 57. MP Ryan (ed) p 37-53
Kohlstedt DL, Weathers MS (1980) Deformation-induced microstructures, paleopiezometers, and
differential stresses in deeply eroded fault zones. J Geophys Res 85:6269-6285
Liebermann RC, Wang Y (1992) Characterization of sample environment in a uniaxial split-sphere
apparatus. In High-Pressure Research: Application to Earth and Planetary Sciences. Y Syono, MH
Manghnani (eds) p 19-31. American Geophysical Union, Washington, DC
Lloyd EC, Hutton UO, Johnson DP (1959) Compact multi-anvil wedge-type high pressure apparatus. J Res
Nat Bur Stand 59:63C
Masters G, Laske G, Bolton H, Dziewonski AM (2000) The relative behavior of shear velocity, bulk sound
speed, and compressional velocity in the mantle: Implications for chemical and thermal structure. In
Earth's Deep Interior. S Karato, AM Forte, RC Liebermann, G Masters, L Stixrude (eds) p 63-87.
American Geophysical Union, Washington, DC
Meade C, Jeanloz R (1988a) The yield strength of B1 sand B2 phases of NaCl. J Geophys Res 93:
3270-3274
Meade C, Jeanloz R (1988b) The yield strength of MgO to 40 GPa. J Geophys Res 93:3261-3269
Meade C, Jeanloz R (1990) The strength of mantle silicates at high pressures and room temperature:
implications for the viscosity of the mantle. Nature 348:533-535
Mei S, Kohlstedt DL (2000a) Influence of water on plastic deformation of olivine aggregates, 1. Diffusion
creep regime. J Geophys Res 105:21457-21469
Mei S, Kohlstedt DL (2000b) Influence of water on plastic deformation of olivine aggregates, 2.
Dislocation creep regime. J Geophys Res 105:21471-21481
Mitrovioca JX, Forte AM (1997) Radial profile of mantle viscosity: results from the joint inversion of
convection and postglacial rebound observables. J Geophys Res 102:2751-2769
Mitrovioca JX, Peltier WR (1991) A complete formalism for the inversion of postglacial rebound data:
resolving power analysis. Geophys. J Intl 104:267-288
Ogawa M (1987) Shear instability in a viscoelastic material as the cause for deep earthquakes. J Geophys.
Res 92:13801-13810
Onodera A (1987) Octahedral-anvil high-pressure devices. High Temp-High Press 19:579-609
Osugi J, Shimizu K, Inoue T, Yasunami K (1964) A compact cubic anvil high pressure apparatus. Rev Phys
Chem Japan. 34:1-6
Paterson M. S (1970) A high-pressure, high-temperature apparatus for rock deformation. Intl J Rock Mech
Min Sci 7:517-526
Paterson MS (1978) Experimental Rock Deformation: the Brittle Field. Springer-Verlag
Paterson MS (1990) Rock deformation experimentation. In The Brittle-Ductile Transition in Rocks: The
Heard Volume. AG Duba, WB Durham, JW Handin, HF Wang (eds) p 187-194. American
Geophysical Union, Washington, DC
Paterson MS, Olgaard DL (2000) Rock deformation tests to large shear strains in torsion. J Struct Geol
22:1341-1358
Peltier WR (1998) Postglacial variation in the level of the sea: implications for climate dynamics and solidEarth geophysics. Rev Geophys 36:603-689
Perez-Albuerne EA, Forgsgren KF, Drickamer HG (1964) Apparatus for X-ray measurements at very high
pressure. Rev .Sci Instrum 35, 29-33
Poirier J-P, Sotin C, Peyronneau J (1981) Viscosity of high-pressure ice VI and evolution and dynamics of
Ganymede. Nature 292:225-227
Post RL (1977) High-temperature creep of Mt. Burnett dunite. Tectonophysics 42:75-110
Prins JF (1984) A semiempirical description of pressure generation between Bridgman anvils. High TempHigh Press 16:657-664
Raleigh CB, Paterson MS (1965) Experimental deformation of serpentinite and its tectonic implications.
J Geophys. Res 70:3965-3985
Riecker RE, Seifert KE (1964) Olivine shear strength at high pressure and room temperature. Geol Soc Am
Bull 75:571-574
Rubey WW (1972) Foreword. In Flow and Fracture of Rocks, Vol. 16, Geophysical Monograph. HC
Heard, IY Borg, NL Carter, CB Raleigh (eds) p ix-x. American Geophysical Union, Washington, DC
Rubie DC, Karato S-i, Yan H, O'Neill HSC (1993) Low differential stress and controlled chemical
environment in multianvil high-pressure experiments. Phys Chem Minerals 20:315-322
Shimomura O, Utsumi W, Taniguchi T, Kikegawa T, Nagashima T (1992) A new high pressure and high
temperature apparatus with sintered diamond anvils for synchrotron radiation use. In High-pressure
research: Application to Earth and planetary sciences. Y Syono, MH Manghnani (eds) p 3-11. Terra
Scientific Publishing, Tokyo
Developments in Deformation Experiments at High Pressure
29
Shimomura O, Yamaoka S, Yagi T, Wakatsuki M, Tsuji K, Kawamura H, Hamaya N, Fukuoga O, Aoki K,
Akimoto S (1985) Multi-anvil type X-ray system for synchrotron radiation. In Solid State Physics
Under Pressure. S Minomura (ed), p 351-356. Terra Scientific Publishing, Tokyo
Singh A. K, Balasingh C, Mao H.-k, Hemley R, Shu J (1998) Analysis of lattice strains measured under
nonhydrostatic pressure. J Appl Phys 83:7567-7578
Sotin C, Poirier J-P (1990) The sapphire anvil cell as a creep apparatus. In The Brittle-Ductile Transition in
Rocks: The Heard Volume, Vol. Geophys. Monogr. No. 56. AG Duba, WB Durham, JW Handin, HF
Wang (eds) p 219-223. American Geophysical Union, Washington, DC
Sung CM, Goetze C, Mao HK (1977) Pressure distribution in the diamond anvil press and the shear
strength of fayalite. Rev Sci Instrum 48:1386-1391
Tackley PJ, Stevenson DJ, Glatzmaier GA, Schubert G (1993) Effects of endothermic phase transition at
670 km depth in a spherical model of mantle convection in the Earth's mantle. Nature 361:699-704
Thoraval C, Richards MA (1997) The geoid constraint in global geodynamics: viscosity structure, mantle
heterogeneity models and boundary conditions. Geophys J Intl 131:1-8
Tingle TN, Green HW II, Young TE, Koczynski TA (1993) Improvements to Griggs-type apparatus for
mechanical testing at high pressures and temperatures. Pure Appl Geophys 141:523-543
Tullis TE, Tullis J (1986) Experimental rock deformation techniques. In Mineral and Rock Deformation:
Laboratory Studies, The Paterson Volume. BE Hobbs, HC Heard (eds) p 297-324. American
Geophysical Union, Washington, DC
Turner FJ, Griggs DT, Clark RH, Dixon RH (1956) Deformation of Yule marble. Part VII: Development of
oriented fabrics at 300˚-500˚C. Geol Soc Am Bull 67:1259-1294
Utsumi W, Yagi T, Leinenweber K, Shimomura O, Taniguchi T (1992) High pressure and high temperature
generation using sintered diamond anvils. In High-pressure research: Application to Earth and
planetary sciences. Y. Syono and M. H. Manghnani), p 37-42. American Geophysical Union,
Washington, DC
van der Hilst RD, Engdahl R, Spakman W, Nolet G (1991) Tomographic imaging of subducted lithosphere
below northwest Pacific island arcs. Nature 353:37-43
van der Hilst RD, Karason H (1999) Compositional heterogeneity in the bottom 1000 kilometers of Earth's
mantle: toward a hybrid convection model. Science 283:1885-1888
van der Hilst RD, Widiyantoro RDS, Engdahl ER (1997) Evidence for deep mantle circulation from global
tomography. Nature 386:578-584
Vaughan M, Chen J, Li L, Weidner DJ, Li B (2000) Use of X-ray imaging techniques at high pressure and
temperature for strain measurements. In Science and Technology of High Pressure, Proceedings of
AIRAPT-17. MH Manghnani, WJ Nellis, M. F. Nicol (eds) p 1097-1098. Universities Press
Walker D (1991) Lubrication, gasketing, and precision in multianvil experiments. Am Mineral 76:
1092-1100
Walker D, Carpenter MA, Hitch CM (1990) Some simplifications to multianvil devices for high-pressure
experiments. Am Mineral 75:1020-1028
Wang Y, Durham W, Getting I (2002) D-DIA: A new apparatus for high-temperature deformation at
pressures up to 15 GPa. Rev Sci Instrum (submitted)
Weidner DJ (1998) Rheological studies at high pressure. Rev Mineral 3:493-524
Weidner DJ, Chen J, Xu Y, Wu Y, Vaughan MT, Li L (2001) Subduction Zone Rheology. Phys Earth
Planet Inter 127:67-81
Weidner DJ, Vaughan MT, Ko J, Wang Y, Liu X, Yeganeh-Haeri A, Pacalo RE, Zhao Y (1992)
Characterization of stress, pressure, and temperature in SAM85, A DIA type high pressure apparatus.
In High-Pressure Research: Application to Earth and Planetary Sciences, Vol. Geophysical Monograph
67. Y Syono, MH Manghnani (eds) p 13-17. Terra Scientific Publishing Company, Tokyo, and
American Geophysical Union, Washington, DC
Weidner DJ, Wang Y, Chen G, Ando J, Vaughan MT (1998) Rheology measurements at high pressure and
temperature. In Properties of Earth and Planetary Materials at High Pressure and Temperature,
Geophysical Monograph 101. M Manghnani, Y Yagi (eds) p 473-482. American Geophysical Union
Weidner DJ, Wang Y, Vaughan MT (1994) Strength of diamond. Science 266:419-422
Wenk H-R, Matthies S, Hemley RJ, Mao HK, Shu J (2000) The plastic deformation of iron at pressure of
the Earth's inner core. Nature 405:1044-1047
Wu TC, Bassett WA (1993) Deviatoric stress in a diamond anvil cell using synchrotron radiation with two
diffraction geometries. Pure Appl Geophys 141:509-519
Yamazaki D, Karato S (2001) High pressure rotational deformation apparatus to 15 GPa. Rev Sci Instrum
72:4207-4211
30
Durham, Weidner, Karato & Wang
contents
INTRODUCTION..................................................................................................................................1
Why is pressure important?.............................................................................................................2
Terminology related to strength and deformation. .........................................................................3
A BRIEF HISTORY OF HIGH-PRESSURE APPARATUS...............................................................4
To 5 GPa: Cylindrical devices ........................................................................................................4
5 GPa and above: Anvil devices .....................................................................................................6
MEASUREMENT METHODS AT HIGH PRESSURE......................................................................7
Stress measurement .........................................................................................................................7
Strain rate measurement ................................................................................................................12
MODERN TECHNIQUES FOR DEFORMATION AT HIGH PRESSURES .................................14
Sample assemblies .........................................................................................................................14
Modifications to the sample assembly in a multianvil press........................................................16
Diamond-anvil cell ........................................................................................................................17
Deformation-DIA ..........................................................................................................................17
Rotational Drickamer apparatus (RDA) .......................................................................................20
SUMMARY AND PERSPECTIVES..................................................................................................24
ACKNOWLEDGMENTS....................................................................................................................25
REFERENCES .....................................................................................................................................26
S
y; y; y; y; y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;
;yy;y;y;y;y;y;y;y;y;y;y;y;
B
y;
y; y; y; y; y;
y; y;
y; y; y; y; y; y; y; y; y; y; y; y; y; y; y; y;y;y;y;y;y;y; y;y; y;y;y; y;y; y; y; y; y; y; y; y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;y;y;y;yy;;y;y;y;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;yy;;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;yy;;y;y;yy;;y;y;y;yy;;y;yy;;y;y;yy;;y;y;y;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;y;yy;; y;y;y;y;y;y;y;y;y;y;yy;;y;yy;;y;y;y;yy;;y;y;yy;;y;y;y;yy;;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y; y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;;yy;y;;yy;y;y;y;y;;y;yy;y;y;y;yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;;yy;y;y;y;;yy;y;y;y;y;y;y;y;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y; y;y;y; y;y; y;y;y; y;y;y; y;y; y;y;y; y;y; y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;;yy;y;y;y;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
;
y
y; y; y; y; y; y; y; y;y;y;y;y;y; y; y; y; y; y; y; y; y; y; y; y; y; y; y;y;y;y;y;y; y; y; y; y; y; y; y; y; y; y; y; y; y; y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
γ
K
y; y; y; y; y; y; y;
;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
;y;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;;y;yy;y;y;y;y;y;y;y;y;y;y;y;y;
y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;
P1
γ
P2
H.P.
Figure 1
b o ro n e p o xy
p
re
zirc o n ia e n d p lu g s s s u re m e d iu m
Na Cl p re s s u re c a lib ra n t
s a m p le (p o wd e re d d ia m o n d )
a lu m in a s le e ve
g ra p h ite h e a te r
P t h e a te r le a d
a lu m in a
th e rm o c o u p le TC tu b in g
le a d s
th e rm o c o u p le
Figure 2
b o ro n n itrid e
s a m p le c h a m b e r
Mo c u rre n t d is k
Pressure
r
Diamond Loading
Z
d
t
Diamond Loading
Figure 3
2q
Figure 4
1 3 e le m e n t S S d e te c to r
c o n ic a l s lit
(o u te r c o n e
s h o w n in o u tlin e )
m ir r o r a n d
Y A G c r y sta l
lo n g w o r k in g d is ta n c e
(in s id e c o n ic a l s lit)
m ic r o s c o p e (1 0 X ) 2 n d m ir r o r
C C D c a m e r a
Figure 5
m in ia tu r e 6 -8
h ig h -p r e s s u r e
m o d u le (T c u p )
X -r a y s fr o m
sy n c h r o tr o n
p a th s o f h o r iz o n ta lly
a n d v e r tic a lly
d iffr a c te d X -r a y s
Figure 6
Figure 7
;yy; ;y yy
y;y;
;;
;
y
yy
;;
y; y; yy
;; ;y
;
y
;;;;;;
yyyyyy
;;;;;
yyyyy
;
yy yyyyy
;;
;;;;;; y
yyyyyy
;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;
;;;;;;;;;;;;;
;;;;;;;;;;;;;
(b)
(c)
soft
;;;;;;;;;;
;;;;;;;;;;;;
(a)
;;;;;;;;;;;;;
;;;;;;;;;;;;;
(d)
sample
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;
ZrO2
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(e)
Figure 8
piston
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(f)
crushable
;;;;;;;;;
;;;;;;;;;
;;;;;;; anvil
;;;;;;;;;
furnace
;;;;;
;;;;;
;;;;;
soft P medium
outer P medium
3
60
Top ram P
Ta
Displacement (mm)
2.5
MgO
40
Oil P (MPa)
Bot ram P
20
2
P to 10 GPa
T to 500˚C
0
1.5
1
Total
ram disp
Def
step
(1)
MgO
Def
step
(2)
-20
-40
Ta
0.5
-60
5:33:20
11:06:40
16:40:00
22:13:20
H:M:S
Figure 11a
1.1
Displacement (mm)
1
P = 10 GPa
T = 500˚C
0.9
.
-5 -1
ε =1.2 x 10 s
MgO
-5
0.45 x 10
s
-1
0.8
Ta
0.7
MgO
Def
step
(1)
0.6
0.8 x 10
0.5
0.4
13:53:20
Def step (2)
-5
s
-1
-5
Ta
0.43 x 10
16:40:00
19:26:40
s
-1
22:13:20
H:M:S
Figure 11b
5 cm
Bearing
y; y;
;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;
y;y; y;y;y; y;y;y; y;
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
y
y
;
y
;
y
;
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
;
;yy;y;y;y;y;y;y;y;y;y;y;;yy;;y;y;yy;y;y;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
y
y
y
;
y; ;y y; y;
;y;y;y;y;y;y;yy;;yy;;yy;y;y;y;y;y;y;y;;yy;;yy;;yy;y;y;;yy;y;y;y;y;y;y;y;y;y;y;yy;y;y;;y;yy;y;;y;yyy;y;;;y;y;yyy;;;y;yyyy;;;;y;yyy;;;y;y;yyy;;;;yy;yy;;;yy;y;yy;;;;yyy;;;yyy;y;y;;;yyy;y;;;yyy;y;y;;;yyyy;;y;;yyy;;;yyyy;;;yyy;;;yyy;;;;yyy;;;yyy;;;;yyy;;;yyy;;;;yyy;;;yyy;;;yyyy;;;yyy;;;yyy;y;;yy;y;;yy;y;y;y;y;y;;y;y;y;y;y;y;y;;y;y;yy;;y;yy;;;y;yyy;;;yyy;;;y;yyy;;y;yy;;yy;yy;;yy;y;;yyy;y;;yyy;;;yyy;;;;yyy;;;yyy;;;y;yy;y;;;yyy;;y;yy;;;yyy;;y;yyy;;y;;yyy;;y;yyy;;;yyy;;;y;yyy;;;y;yy;;;yyy;;;y;yy;;;yyy;;y;yy;
;y ;y y; y;y; y; y;y; y;y;y; y;y;y; y; y;;y ;y y; y; y; ;y ;y ;y y; y; y; ;y ;y ;y y; y; y; ;y ;y y; y;
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;yy;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;;y;yy;y;;y;y;yyy;;;y;y;yyy;;;y;yyy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yy;;yyy;;;yyy;;yyy;;;yyy;;yyy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yyy;;yyy;;;yyy;;yy;y;;yy;y;y;y;y;y;y;y;;y;y;y;y;y;y;;y;yy;;y;yy;;;yyy;;;yyy;;y;yy;;y;yy;yy;y;;yy;y;yyy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yy;;yyy;;;yyy;;yyy;;
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yyy;;;yy;y;;yyy;;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;yy;;yy;;yy;;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;yy;;yy;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;;yy;;yy;yy;;yy;y;y;y;y;y;y;y;y;;y;y;y;y;;yy;;yy;;yy;;yy;y;y;y;yyy;;yy;;yy;;yy;;y;;y
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;yy;;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;yy;;yy;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;;yy;;yy;yy;;yy;y;y;y;y;y;y;y;y;;y;y;y;y;;yy;;yy;;yy;;yy;y;y;;yy;yy;
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;;yy;yy;;y;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;y;y;yy;;y;y;y;y;yy;;y;y;y;yy;;y;y;y;yy;;y;y;y;yy;;y;y;yyy;;;yy;;yyy;;;yy;;yyy;;;yy;;yy;;;yy;;yy;;;yy;;yy;;yyy;;yy;;yyy;;yy;;yyy;;;yy;;yy;;;yy;;yy;;;yy;;yy;;;yyy;;yy;y;yy;y;y;y;y;y;y;y;y;y;y;y;y;;y;y;y;y;y;;y;yy;;yy;;;yyy
;y ;y ;y ;yy; y; y;y; y;y;y; y;y;y; y; y;y; y; y; y; y; y; y; y; y; y; ;y ;y y; y; y; ;y ;y y; y; y; y;
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;y;;y;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;;yy;y;;yyy;;;yyy;;;yyy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yy;;yyy;;;yyy;;yyy;;;yyy;;yyy;;;yyy;;;yy;;;yyy;;;yy;;;yyy;;;yy;yy;;y;yy;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;y;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;yy;;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;yy;;yy;yy;;yy;;yy;;yy;;y;;yy;;y;y;y;;yy;y;y
;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;y;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;yy;;yy;;yy;;yy;;yy;;y;;yy;;y;;yy;;yy;yy;;yy;yy;;yy;;yy;;yy;;y;;y
y;y;y;;y;y;y;y;y;y;y;y;y;y;y ;yy;;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;yy;;y;y;y;yy;;y;y;y;yy;;y;y;y;;yy;;yy;;;yyy;;y;yy;y;y;
y;y;y; y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y; y;y;y; y;y;y; y;
yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
yy;;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y; y;y; y;y; y;y; y;y; y;y; y;y; y;y;
y; y;
;yy;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;y;
y; y;y; y;y; y;y; y; y; y;y; y;y; y;
(for confining pressure)
Load
Guide sleeve
Gasket
(for deformation)
Rotation
Anvil
Sample
Guide block
yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy
Harmonic gear
Timing
belt
Motor
Figure 12
Diffracted
X-ray
σ
5
X-ray
3
Window
n=1
x
Sample
x
X-ray
1-2 mm
(top view)
Figure 14