SANDIA REPORT

SAND94-3266 • UC-704
Unlimited Release fi?P?/^#*
Printed February 1995 ^C/X/^,^

OST,
Dynamic Properties of Ceramic Materials

D. E. Grady

Prepared by
Sandia National Laboratories
Albuquerque, New Mexico 87185 and Livermore, California 94550
for ttie United States Department of Energy
under Contract DE-AC04-94AL85000

Approved for public release; distribution is unlimited.

D\STR\BUT!ON OF THIS DOCUMENT IS UNLIMITED

Issued by Sandia National Laboratories, operated for the United States
Department of Energy by Sandia Corporation.
NOTICE: This report was prepared as an account of work sponsored by an
agency of the United States Government. Neither the United States Govern-
ment nor any agency thereof, nor any of their employees, nor any of their
contractors, subcontractors, or 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, prod-
uct, or process disclosed, or represents that its use would not infringe pri-
vately 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, any agency thereof or any of
their contractors or subcontractors. The views and opinions expressed
herein do not necessarily state or reflect those of the United States Govern-
ment, any agency thereof or any of their contractors.

Printed in the United States of America. This report has been reproduced
directly from the best available copy.

Available to DOE and DOE contractors from

Office of Scientific and Ttechnical Information
PO Box 62
Oak Ridge, TN 37831

Prices availablefi-om(615) 576-8401, FTS 626-8401

Available to the public from

National Tfechnical Information Service
US Department of Commerce
5285 Port Royal Rd
Springfield, VA 22161

NTIS price codes

Printed copy: A05
Microfiche copy: AOl
 DISCLAIMER

This report was prepared as an account of work sponsored by an

agency of the United States Government. Neither the United States
Government nor any agency Thereof, 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 any
agency thereof. The views and opinions of authors expressed herein
do not necessarily state or reflect those of the United States
Government or any agency thereof.
 DISCLAIMER

Portions of this document may be illegible

in electronic image products. Images are
produced from the best available original
document.
 SANfD94-3266 Distribution
Unlimited Release Category UC-704
Printed February 1995

Dynamic Properties of Ceramic Materials

D. E. Grady
Experimental Impact Physics Department 1433
Sandia National Laboratories
Albuquerque, NM 87185

Abstract

The present study offers new data and analysis on the transient shock strength and equa-
tion-of-state properties of ceramics. Various dynamic data on nine high strength ceramics
are provided with wave profile measurements, through velocity interferometry techniques,
the principal observable. Compressive failure in the shock wave front, with emphasis on
brittle versus ductile mechanisms of deformation, is examined in some detail. Extensive
spall strength data are provided and related to the theoretical spall strength, and to energy-
based theories of the spall process. Failure waves, as a mechanism of deformation in the
transient shock process, are examined. Strength and equation-of-state analysis of shock
data on silicon carbide, boron carbide, tungsten carbide, silicon dioxide and aluminum ni-
tride is presented with particular emphasis on phase transition properties for the latter two.
Wave profile measurements on selected ceramics are investigated for evidence of rate sen-
sitive elastic precursor decay in the shock front failure process.

Sponsored by the U.S Deparlment of Energy and conducted under the auspices of the U S Department of Energy under
Contract DE-ACO4-94AL850O0

DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED MASTER

4
 Acknowledgments

The author would like to thank Marlin Kipp for providing most of the supporting compu-
tational analysis and Ron Moody For conducting the shock compression experiments.
6
 Contents

Acknowledgments 5
Contents 7

1. Introduction and Summary 9

2. Compressiive Failure Mechanisms 12
3. Failure Wave Processes 21
4. Dynamic Spall in Ceramics 25
5. Failure Criteria 32
6. Silicon Carbide and Boron Carbide 35
7. Aluminum Nitride 44
8. Tungsten Carbide 54
9 Silicon Dioxide 65
10 Precursor Decay 73
11. Conclusions 86
12 References 87

7
8
 Introduction and Summary

1 Introduction and Summary

The current program focuses on the measurement of response properties of high-strength
ceramic materials to support modeling activities facilitating computational and engineer-
ing analyses of ceramic structures under dynamic loads. As material response models and
calculational methods mature, computational analysis will play an increasingly important
role as cost-effective and time-efficient tools for the design, optimization, and evaluation
of such structures.

The material response models for high-strength brittle solids are currently in a state of
rapid development. A number of such models are currently available. A clear consensus
as to the most appropriate modelling framework, or the critical deformation and equation-
of-state features which must be implemented in these models, has not yet emerged.

Thus the present experimental study, in which controlled impact experiments and velocity
interferometry diagnostics are used to acquire high-resolution shock compression data, is
timely and performs several important functions critical to the development of adequate
computational models for ceramic materials. First, a base of engineering dynamic strength
and equation-of-state data is being acquired which establishes material constants neces-
sary to all of the models. Second, time-resolved shock-wave profile measurements are be-
ing provided which are used to benchmark and validate existing computational models.
Finally, and perhaps most importantly, the present shock-wave measurements provide a
window into the physical mechanisms of dynamic deformation which are active in brittie
solids subjected to transient impact loading. This latter contribution is critical to the
thoughtful development of the most appropriate models for describing the dynamic be-
havior ceramic materials.

A range of activities has been completed since the last reporting period [Grady and Wise,
1993] which are included in the present report. Some were released earlier as Technical
Memoranda while others have been prepared for recent technical conferences. References
will be cited where relevant open literature is available.

In Section 2 previously unpublished shock compression and release profiles for aluminum
oxide — Coors AD995 — are reported. The VISAR experiment used to measure the ve-
locity histories is described and the data are used to examine the mechanisms of failure
and deformation within the rise of the shock compression front. Although convincing ar-
guments for neither ductile (dislocation processes) nor brittle (fracture processes with lo-
cal loss of cohesion) failiu-e in the shock front are available, the latter (brittie) mechanism
is assumed. Microstructural features of the dynamic fracture process consistent with ener-
gy and time constraints are explored. Characteristic microfracture spacings and crack-
opening displacements are developed.

In Section 3 the concept of failure waves, recentiy put forth and investigated experimen-
tally be several authors, is considered within the context of uniaxial shock waves. The
failure wave is examined within the context of both a subsonic wave and a delayed kinetic
process. Previously reported stress (both axial and transverse) and particle velocity mea-
surements associated with passage of the failure wave, along with kinematic constraints of

9
the uniaxial loading conditions, are used to reach the conclusion that dilatancy (a transfer
of lattice strain to void volume) must accompany the failure wave process. Estimates of
dilatant void volume for failure waves in glass are made.

Section 4 provides a discussion of the dynamic tensile (spall) strength of brittie solids. The
planar-impact spall experiment is described and data for eight ceramics are provided. The
spall data are compared with calculated theoretical spall strength values and a theory of
the dynamic spall process based on both inherent iiaw and fracture energy concepts is pre-
sented. Strain rate dependent relations between static tensile strength and spall strength for
brittie solids is developed.

In Section 5 brittie fracture in the shock wave front is again assumed and the theory of
Griffith for failure of a brittie solid under a general state of stress is examined. In particular
the ratio of the Hugoniot elastic limit to the spall strength predicted by the Griffith failure
criterion is calculated and compared with the experimental data for the ceramics reported
in Section 4. The comparison is not inconsistent with brittle failure in the shock process.

Parts of Section 2, 4 and 5 are included in a paper prepared for a Workshop on Fracture
and Damage in Quasibrittle Structures, Prague, Czech Republic, September 21-23, 1994,
and was published in the Proceedings of that Workshop [Grady, 1994a].

Shock wave profile data for silicon carbide and boron carbide are presented in Section 6
and are used to examine the starkly different dynamic strength and shock equation-of-state
properties of these two carbide ceramics. The wave profile signature of silicon carbide is
characteristic of high strength metal behavior and is readily modeled with standard work-
hardening elastic-plastic models. In contrast, post-yield dynamic flow properties of boron
carbide suggest a catastrophic near or total loss of strength, and details of the VISAR mea-
surement indicate a heterogeneous deformation process. The wave-profile measurements
for boron carbide are open to other interpretations, however, and the possibility of an un-
usual shock-induced lattice compression (phase transition) needs to be considered further.
The content of Section 6 was prepared for publication in the International DYMAT Con-
ference proceedings, Oxford, England, September 26-30, 1994 [Grady, 1994b].

In Section 7 new shock-wave data for aluminum nitride ceramic, principally targeting dy-
namic effects of the hexagonal-to-cubic phase transformation under shock compression in
this material, are presented. The principal effort in this study is concentrated on the model-
ling issues necessary to the presentation of this large Av phase change in computational
simulations. Section 7 has also been published in the 1994 International DYMAT Confer-
ence [Kipp and Grady, 1994].

New shock equation-of-state and spall data for several liquid-phase sintered tungsten car-
bide ceramics are provided in Section 8. One material was extracted from the core of the
BS-41 armor piercing round. The relatively low Hugoniot elastic limit, strong hardening
characteristics, and large spall strength presumably reflect the metal-ceramic properties of
this material.

10
 Introduction and Summary

In Section 9 several tests on silicon dioxide (quartz rock) are presented which focus on ex-
ploring a critical stress level in the shock-induced a-quartz-to-stishovite phase transition
in this material. Features of this transition were previously reported by workers in the
former Soviet Union [Zhugin and Krupnikov, 1987] using manganin gauge techniques,
and the present effort was a cooperative study using VISAR diagnostics [Grady and Zhu-
gin, 1994]. There are interesting technical reasons that this critical transition anomaly has
eluded discovery in this country over the several decades of study concentrated on quartz
and quartz rock materials.

In Section 10 new wave profile data along with previously reported data are examined for
properties of elastic wave precursor decay in the shock compression of ceramics. This fea-
ture is believed to be a sensitive indicator of rate-dependent deformation in the shock-
compression failure of materials. With the possible exception of boron carbide, the ceram-
ics investigated indicate no, or very little, elastic precursor attenuation. These observa-
tions, in contrast to some of the results discussed in earlier Sections, are not consistent
with a rate-sensitive fracture process in the shock front.

Taken collectively, the new results provided here offer an interesting perspective on the
dynamic behavior of ceramic materials. Brittie versus ductile deformation in the shock
front is still an open issue. Although magnitudes and relative levels of Hugoniot elastic
limits and spall strengths are not inconsistent with brittie failure, the lack of rate sensitivi-
ty suggested by the precursor evolution data in Section 10 speaks strongly for plasticity
processes in the shock front.

The spall failure of ceramic materials is clearly a brittie fracture process as is the uncon-
fined compressive failure initiated under Hopkinson bar loading conditions. The failure
wave may be the missing key in linking failure under shock compression to dynamic fail-
ure in the tensile and unconfined compressive environment. The characteristic time for the
failure wave event is intermediate between the shock front failure process and the com-
pression bar experiment. The identification of failure waves in ceramic materials other
than glass, and determination of the related properties of failure waves are the critical next
steps in uncovering the nature of shock-induced deformation of brittle solids.

Shock-induced solid state phase transformation is an integral component of the dynamic

deformation of high-strength ceramics within the pressure range of interest in the present
study. At least three, and possibly as many as six, of the nine ceramic materials referred to
in the present report undergo shock-induced phase transitions. Work in progress indicates
that the inclusion of phase transformation in the computational analysis of projectile-
structure interactions can significantiy influence predictions.

The outline above summarizes the key issues investigated in the present ongoing study.
The work adds substantially to the body of data available on high-strength brittie materi-
als. More importantiy, the effort has identified critical research directions necessary to a
fuller understanding of the dynamic compressibility of ceramics.

11
2 Failure in Ceramics in the Compressive Shock Front
A broad class of brittie solids (the attention here focuses principally on high-strength
monolithic ceramics) subjected to large amplitude shock waves can support substantial
shear stress (of order 5-15 GPa), and the accompanying elastic strain energy, within the
compressive shock-wave front, without failing due to the very limited slip systems in
these materials. When failure occurs under sufficientiy intense shock loading, the effect is
usually observed as a wave splitting in the compressive shock front, and details of the fail-
ure process can be inferred from features in the measured shock-wave profiles. Because of
the high confining stress state associated with the failure event in the shock compression
environment, it is no longer clear whether the microstructural processes of deformation
are brittie or ductile. Some, although by no means sufficient, evidence supports a brittie
deformation mechanism in the hard materials of interest within the range of present dis-
cussion. Other evidence, however, argues for microplasticity processes within the shock
front.

If brittie mechanisms control failure in the shock front for a least some of the ceramic ma-
terials of interest, then details of the activation and growth of brittle fracture and flow in
the shock compression process are poorly understood. Mechanisms for the shear failure
process including the process-zone characteristics of individual shear cracks and fracture
densities can only be surmised. In this Section we present new wave-profile data represen-
tative of the shock compression behavior of many high-strength ceramics. These data are
used to explore some of the microstructural deformation issues of interest, and to provide
estimates of shear fracture properties within the deformation shock wave.

2.1 Compression Waves in Ceramics

The compression failure in a ceramic body, subjected to a complex stress state, is achieved
when some measure of the stress difference such as the octahedral shear stress,

1 / 2 2 2
t , = XA/(CJ - O ) + (a -G ) -I- ( a -o ) , (2.1)
oci 3 ^ ^ X y' ^ y 2' ^ 1 x' ' ^ '

achieves a critical condition which may depend, for example, on the mean stress,
p - (a^ + a + a^) / 3 , the rate of loading, e, and the temperature of the body. One par-
ticularly unique loading path is achieved under a constraint of uniaxial strain in which the
two transverse stress components are equal, say ^ = ^^, and are proportional to the axi-
al stress according to,

^ = i^^.' (2.2)

where v is Poisson's ratio.

The latter stress loading path is of particular interest because it is readily achieved in the
controlled planar impact of two plates and the technology for performing such experi-

12
 Failure in Ceramics in the Compressive Shock Front

TARGET
ASSEIMBLY
Projectile
Foam Ceramic 3 Velocity Pins
Backing

To VISAR

Target Lithium Fluoride

Ceramic Window

PROJECTILE
Al Ring
BODY Al Nose Aluminum
Plate Target Fixture

Figure 2.1 Experimental configuration for shock- and release-wave

experiments on ceramic.

ments and diagnosing the consequences of such tests have achieved a mature state of de-
velopment.

One test technology for probing the response of solids under uniaxial strain compression
loading is illustrated in Figure 2.1 [Grady, 1992a; Grady, 1992b]. A disc of the ceramic
being tested is mounted in the projectile and is supported on the main projectile body by a
disc of low density polyurethane foam. For the target, a similar disc of the ceramic is
mounted in the stationary supporting target fixture. An optical quality disc of single crys-
tal lithium fluoride is intimately bonded with epoxy to the back of this ceramic sample.
All critical surfaces are lapped and polished, and are typically flat to within a few bands of
sodium light The bonded litiiium fluoride surface is first lightiy diffused and vapor-depos-
ited with about 100 nm of aluminum. The ceramic-lithium fluoride epoxy bond thickness
is approximately 10 to 20 fim.

The ceramic-on-ceramic planar impact produces a compressive wave of uniaxial strain

which propagates across the stationary ceramic specimen and through the ceramic-lithium
fluoride interface. An equivalent compressive wave propagates through the projectile ce-
ramic specimen and reflects at the low-impedance foam interface as a release wave which
unloads the compressed ceramic. Dimensions of the ceramic discs are selected such that
lateral release waves from the boundaries of the disc do not interfere with the central mo-
tion until after the primary experimental measurement is completed.

13
The compression and release wave behavior is measured by monitoring the time-resolved
longitudinal motion at the center of the ceramic-lithium fluoride interface with laser veloc-
ity interferometry (VISAR) techniques [Barker and HoUenbach, 1972]. Measurements are
recorded on transient digitizers. Lithium fluoride is used as the laser window material be-
cause, although its mechanical impedance is somewhat lower than the ceramic being test-
ed, it is the only material which has been optically calibrated and which remains
transparent when subjected to the 10 to 50 GPa shock stresses generated in the present ex-
periments [Wise and Chhabildas, 1985].

The interference fringes measured with the VISAR system are converted to a time-re-
solved history of the velocity of the interface using the method of Barker and
HoUenbach (1972), with a time resolution of approximately 1 ns. The ampUtude resolu-
tion is approximately 2% per fringe and typicaUy two to three fringes are achieved in the
interface acceleration through the compressive shock front. From these records the dy-
namic stress and strain characteristics of the ceramics are determined through further com-
putational and analytic techniques.

1.5 -I—r T—I—I—r

w 1.0 -
E

o
o
--J 0 5 -

0.0 J L^_k -rHJj -I L. I • I L

1.0 1.5 2.0 2.5 3.0 3.5

TIME (/zs)
Figure 2.2 Shock-compression and release wave profiles for aluminum
oxide measured with velocity interferometry (VISAR)
diagnostics with the experimental configuration of Figure 1.

14
 Failure in Ceramics in the Compressive Shock Front

Measured wave profiles from a series of compression and release wave experiments per-
formed on aluminum oxide ceramic are shown in Figure 2.2. The lowest impact velocity
achieved a Hugoniot state just in excess of the Hugoniot elastic limit. The remaining ex-
periments ranged up to Hugoniot states of approximately 40 GPa which is about six times
the Hugoniot elastic limit of this material. The impact velocities and experimental dimen-
sions for this series of experiments are provided in Table 2.1.

Table 2.1:
Shock Compression Wave-Profile Experiments on Aluminum Oxide Ceramic

Test Impact Target^ Impactor^ Shock Particle Stress Specific

Number Velocity Thickness Thickness Velocity Velocity Volume

(km/s) (mm) (mm) (km/s) km/s) (GPa) (m^/Mg)

CE-57 1.070 10.006 5.019 8.54 0.535 19.1 0.242
CE-58 1.572 10.008 5.008 8.29 0.786 27.6 0.234
CE-59 1.943 10.007 5.013 8.56 0.972 34.1 0.229
CE-60 2.329 9.998 5.005 8.80 1.165 41.7 0.224
CE-61 0.561 9.987 4.989 9.18 0.281 11.0 0.250

' The material tested was AD-995 ceramic provided by Coors Porcelain Company.
The density is 3890 kg/mr', longitudinal elastic velocity is 10.56 km/s, and shear
velocity is 6.24 km/s.
^ Lithium fluoride windows 25.4 mm in thickness and 50.5 mm in diameter were used
in all tests.
^ Polyurethane foam densities backing impactors were 320 kg/m-^ for Tests CE-57, 58,
59, 61 and 640 kg/m"* for Test CE-60.

The wave profile data as shown in Figure 2.2 can be used to determine the Hugoniot prop-
erties of the aluminum oxide ceramic. Here Hugoniot states refer to the peak stress versus
specific volume states achieved in the shock compression process. The amplitude of the
initial precursor corresponds to onset of shear failure (the Hugoniot elastic limit) and for
the present aluminum oxide ceramic was determined to be 6.2 ±0.4 GPa while the precur-
sor velocity was determined to be 10.74 km/s. Hugoniot states calculated from the wave
profile data are provided in Table 2.1.

Both the Hugoniot elastic limit and the subsequent final stress versus specific volume
Hugoniot states are shown in Figure 2.3. The shock data can be compared with the static
x-ray diffraction data to 12 GPa on sapphire of Sato and Akimoto (1979). The latter static
data have been extrapolated to 30 GPa with a Brrch-Murnaghan equation of state to pro-
vide a measure of the hydrostatic compressibUity of AI2O3. The offset between the static
and shock data illustrates, and is commonly used to assess, the sustained nonhydrostatic
stress state following the shock compression wave alluded to in the opening paragraph of
this section. Shock temperature would account for less than 0.1 GPa of the offset. Impuri-

15
 60 T T
Aluminum
Oxide A Shock-Hugoniot Data
50 (Present)
D
Q. • Static Data
O
(Sato and Akimoto, 1979)
40

{/)
UJ
on 30
I—

20 -
O
O
3
10 -

0.20 0.28
SPECIFIC VOLUME (Mg/m^)

Figure 2.3 Hugoniot states for shock-compressed aluminum oxide

compared with extrapolation of hydrostatic x-ray diffraction
data.
ties and porosity, however, can seriously complicate a detailed evaluation of the deviator
stress state under shock compression by this method.

2.2 Compression Failure Process

It is the compressive failure and deformation properties of ceramics under the shear stress
loading brought about by the shock compression process which contributes to the richness
in the compressive shock-wave characteristics of these materials. Volumetric lattice com-
pressibility for the intermetalUc Ught weight compounds of interest is usually a decreasing
function of pressure in the shock stress range of concern, and would, of itself, lead to the
propagation of a single steady shock wave. The exception occurs in those materials in
which phase changes brought about by lattice instabiUties at critical levels complicate the
volumetric compressibility, a not uncommon occurrence in the present intermetalUc com-
pounds.

As compression within the shock front increases during the shock loading process in ma-
terials with strength, shear stresses wUl ultimately achieve levels in excess of the critical
strength of the material. Stresses wUl be amplified at microscopic defects within the soUd,
and at these points inelastic deformation accompanied by stress relaxation will lead to per-

16
 Failure in Ceramics in the Compressive Shock Front

vasive shear faUure in the material. It is these deformation-induced convexities in the

stress-versus-strain curve for ceramics which provide the complexities observed in the
compression wave-profile structure.

Features of inelastic shearing slip emanatingfi^omdefects within the shock front can be
identified as ductile or brittie. Ductile slip is associated with shearing motions in which
material displacements normal to the slip planes never exceed atomic dimensions and
such that material cohesion is retained. Brittie slip (or shear fracture), in contrast, involves
the normal separation of material points over dimensions in excess of atomic dimensions;
consequentiy, cohesion of the material is lost and internal stress-free surfaces are intro-
duced.

Metals are generally associated with ductile failure, while ceramics are usually considered
to be brittle. This intuitive classification becomes less certain, however, when a confining
pressure component of the stress field is considered. In principle, sufficient confining pres-
sure can be applied to cause any material to undergo a brittle-to-ductile transition. Conse-
quently, the possibility of ductile shear faUure within the shock front in ceramics cannot
be ruled out.

Spall experiments, in which shock wave tests were designed to first dynamically compress
the specimen and then test the tensUe strength of the specimen, have been used to examine
the residual cohesion of a material after shock compression. It has been argued that the
loss of spall strength observed in numerous ceramics, when first shock compressed above
the Hugoniot elastic limit, supports a brittie failure process in the compressive failure of
ceramics. However, even these results are not unambiguous since reverse shear yield is
usually achieved on decompression before a tensile axial stress is achieved. The signifi-
cantiy lower confining pressure associated with the reverse yielding would much more
readily accommodate brittle failure and loss of cohesion than would the initial failure on
shock compression.

The reasonable agreement of shock compression spall and Hugoniot elastic limit data
with predictions based on a Griffith brittie failure criterion to be presented in a subsequent
section might also be construed as further evidence for shear fracture in the shock com-
pression process. Even here the evidence is weak and other explanations are possible.

Nevertheless, shear fracture dvu^ing the shock compression process does appear to be a vi-
able mechanism for inelastic deformation, and speculation on physical details of the frac-
ture mechanisms in the shock front is worth pursuing. The shock compression profile in
ceramics generally looks like the example for aluminum oxide shown in Figure 2.4(a).
There are notable exceptions, however. Boron carbide, for example, is observed to exhibit
marked stress softening in post-faUure behavior under shock compression [Grady, 1992b;
Grady, 1994].

The compressive wave profile is characterized by three distinct regions. The first is a sud-
den elastic rise to the failure limit or Hugoniot elastic limit. (The experimental definition
of the Hugoniot elastic Umit is discussed further in Section 10.) For high-strength ceram-
ics elastic stresses of 5-20 GPa are achieved at failure with corresponding elastic volume

17
and shear strains of about 2%-5%. Second, the faUure ramp identifies a region of rapidly
changing compressibiUty, although only smaU additional strains are accommodated by the
material during passage of this wave segment It is likely that only precursors to the in-
elastic failure process occur within this time such as incipient fracture incubation and nu-
cleation processes. FinaUy, formation of a steady rapidly-rising deformation shock is
demanded by the intrinsic nonlinearity and upward concavity of the volumetric elastic
compressibility of the solid. The preponderance of inelastic shear deformation due to
shear fracture growth is presumed to be accommodated within the deformation shock
wave. The width of this wave segment is governed by kinetic and time-limited deforma-
tion processes.

ALUMINUM OXIDE
(Test CE-58)
Deformation /
\
Hugoniot
State

Elastic
Shock Ny^^
Failure
Ramp\ y
^

J
Position
(a)
Characteristic
Fracture
Spacing (K)

Fracture Shear Fracture Comminuted

Incubation Growth within Material at
and Nucleation Shock Front Hugoniot State
(b)
Figure 2.4 Shock wave compression profile for aluminum oxide (a) and schematic
of brittie shear fracture process within the shock-wave front (b).

18
 Failure in Ceramics in the Compressive Shock Front

A reasonable physical model of the shear fracture process within the shock front is shown
in Figure 2.4. Figure 2.4(a) illustrates the stress (or particle velocity) wave profile propa-
gating toward the left through the solid. Material to the left of the elastic front is undis-
tiu-bed. Material to the right is fuUy comminuted due to passage of the shock. Shear
fracture incubation and nucleation occurs within the faUure ramp — mature shear crack-
ing and inelastic fractiu"e deformation follows within the deformation shock wave.

In the faUure ramp and the early deformation shock wave, confining pressures are of order
3 - 1 0 GPa and can rise to values of 30 - 60 GPa or more as the Hugoniot state is ap-
proached. Clearly then, the material must be capable of locally (microscopically) support-
ing stress gradients of this same order if material displacements consistent with brittle
deformation and loss of cohesion are to occur in the zone of fracture. Such behavior could
readily be accommodated in microscopic shear fracture if a film of ultra-fine debris were
generated within the fracture process zone and persisted to support the walls of the shear
fracture during mature growth and subsequent arrest as iUustrated in Figure 2.5.

Size of the fracture zone debris and width of the fracture can be bounded from the size de-
pendence of the strength of brittle objects based on energy balance arguments,

6 = ^^, (2.3,

where 5 is the characteristic debris size, E is the elastic modulus, y is the cohesive sur-
face energy and P^ is the effective stress determining the elastic strain energy within the
supporting particle. Based on £ = 400 GPa and y = 5 J/m", which are appropriate proper-
ties for aluminum oxide, and P^ equal to the local confining pressure of about 5-10 GPa, a
characteristic particle size and fracture zone dimension of 5 ~ 40-160 nm is calculated.
Values for P^ appreciably higher than the confining pressure would be expected due to
stress concentration near the environment of the supporting particles and, consequently,
estimates for 5 probably represent an upper bound. A small degree of local dilatancy is
expected to accommodate the shear-fracture free volume. The parameter 5 will decrease
through the deformation shock as confining pressures approach the Hugoniot state. The
corresponding crushing of fracture zone debris would lead to local energy dissipation and
elevated temperatures. Under certain material conditions localized melting and transition
to near fluid shear zones within the fracture region late in the deformation shock may be
expected [Grady, 1980].

The characteristic spacing of fractures in the shock process zone X (see Figure 2.4(b))
may be predictable from an energy balance approach to fragmentation which has been
successful in other applications [Grady, 1982; Grady, 1988]. The theory leads to,

based on the shearing strain rate e in the shock wave and the fracture energy y. The strain
rate can be calculated from e = 3Aa I2r\ where r\ is the solid viscosity within the shock

19
 Failure Ramp Deformation Sliock

Shear-Fracture Fracture-Zone
Process Zone Debris
^"•"^

Confining
Pressure (P^,)

/
Process-Zone
Width (6)

Figure 2.5 Indicates the incubation and growth of shear fracture in

the ramp and deformation shock front.

front and Ao^ is the stress jump through the deformation shock wave. Although proper-
ties in the shock which determined this viscosity are not understood, measurements of
shock widths in brittle materials can be used to estimate deformation strain rates.

For example, in the aluminum oxide reported in the present work (Figure 2.2), a shock
state of 19 GPa leads to a strain rate in the shock front of about 1.7x10 /s and a fracture
spacing from Equation 2.4 of approximately 130 |im. In contrast, the highest shock stress
of 42 GPa yields a shock front strain rate in excess of l.OxlO^/s and afracturespacing of
less than 8 |j,m. Material parameters of p = 3890 kg/m^ and y = 5 J/m^ were used in
Equation 2.4. Such predictions of fracture spacing exceed the fracture process zone di-
mension 5 by about three orders of magnitude and at the lower strain rates are weU above
the characteristic grain size of this material.

20
 Failure Waves

3 Failure Waves
The emergence and evolution of structure in a compressive shock wave such as illustrated
for aluminum oxide in Figure 2.2 is critically dependent on the amplitude of the input
stress load. If the input stress level is less than the Hugoniot elastic limit, it is expected
that an elastic wave of finite amplitude will propagate in the brittie soUd. This expected re-
sponse for waves of uniaxial compression has been brought into question by recent exper-
imental shock wave studies on K-2 glass by Kanel' et al. (1992). That work has provided
evidence for the propagation of a delayed front of fracture foUowing the initial elastic
compression wave. This new, and not yet well understood, phenomenon has been identi-
fied as a failure wave, and is presumed to be a shear fracture process which is driven by
the large shear strain energy residing in the body behind the large-amplitude elastic uniax-
ial-strain compression wave.

Further evidence supporting the observations of Kanel' and coworkers has been provided
by the studies of Brar et a/. (1991) on a similar soda-lime glass, and those of Raiser et al.
(1994) on an alumina-silicate glass. In the former work, transverse stress gages provided
definitive evidence for substantial increase in the transverse stress component a upon
passage of the faUure wave, thus significantly reducing the shear stress
T = (CT^- <7 ) / 2 . Brar et al. (1991) also tested the tensUe strength of glass behind the
initial elastic shock wave both before and behind the failure wave through the appropriate
design of the spall experiments. These tests indicated a tensile strength in excess of 3 GPa
for glass in front of the failure wave and nearly zero strength behind, suggesting a transi-
tion to fully comminuted material following passage of the failure wave. FaUure wave ef-
fects were noted by Brar et a/. (1991) for soda-lime glass for compressive stress levels in
excess of about 4 GPa but below the Hugoniot elastic limit of approximately 6.5 GPa.
Raisor et el. (1994) investigated surface roughness effects at the impact as a source for ini-
tiation of the faUure wave but found no influence. Spall experiments on alumina-silicate
glass before and behind the failure wave were also in agreement with the results of Brar et
al. (1991).

EarUer researchers have suggested that the failure wave may be a propagating fracture
front trailing the initial elastic shock wave at a velocity substantially less than the shock
velocity [Kanel' et al. 1992]. A faUure wave velocity closer to the Rayleigh wave speed
has been proposed [Raisor and Clifton, 1993] although evidence for this result is not yet
convincing.

A viable alternative is that the delay time between the shock wave and failure wave is
governed by the kinetics of fracture nucleation and growth. Incubation of the shear frac-
ture process would initiate immediately after passage of the elastic shock front. The fail-
ure wave occurring somewhat later would then manifest the catastrophic fracture growth
phase accompanied with observable changes in the longitudinal and transverse stress
state.

Thus, the dweU time between shock front and failure wave could be loosely likened to the
ramp region of the higher ampUtude wave profile Ulustrated in Figure 2.4. It is a regime of

21
the profile in which time-dependent preparatory conditions are being established prior to
the pervasive dynamic shear failiu-e process.

The reported changes in the state of material upon passage of a failure wave are difficult to
reconcile without rather unusual assumptions regarding material response. Several of the
critical observations are:

(1) Velocity interferometry measurements indicate small or negligible change

in particle velocity through the failure wave.
(2) Transverse and normal stress measirrements indicate that the stress difference
(^x'^^y - 2"^ reduces substantially during passage of the failure wave but
most of the change is due to an increase in o ; there is littie change in a^.
(3) The faUure wave is reported to propagate at a velocity significantiy reduced
from the longitudinal elastic velocity in glass. Values of order 2 km/s have
been inferred from experimental data. The evidence here is weak and any velocity
up to the longitudinal sound velocity — including time and ampitude dependence
— should be considered possible.
Consider the possibility that the material response leads to a failure wave with a character-
istic velocity VfaUure ^ ^"^ *^t passage of the failure wave corresponds to a complete loss
of the stress deviator. If the volumetric compressibUity behind the failure wave continues
to correspond to fuUy dense glass then the situation shown in Figure 3.1(a) wUl apply.
State 1 precedes the failure wave and state 2 is achieved foUowing the passage of the
wave. The velocity of the failure wave is determined from the slope of the chord connect-
ing state 1 and state 2 according to,

V,, = l^. (3.1)

failure /Jp^g

The difficulty with this assumed behavior is that a significant increase in particle velocity
(AM = ^^^failure^ would be expected which is not consistent with the first observation.

A second possibility is that material behavior through the faUure wave represents the cata-
strophic relaxation of the shear stress due to a time-dependent fracture nucleation and
growth process. Consider the possibility in Figure 3.1(b) in which complete relaxation of
the stress deviator at constant strain is achieved through the failure wave. Again, in order
to achieve the volumetric compression curve of fuUy dense glass identified as state 2 in
Figure 3.1(b), one of the critical observations is violated. Namely, a substantial drop in a^
through the failure wave — approximately twice the increase in o .

There is another possible material behavior, Ulustrated in Figure 3.1(c), which could be
consistent with the three experimental observations. This explanation involves a change in
the volumetric compressibility through the faUure wave. The alternative pressure (mean
stress) versus volume curve is initiated due to shear fracture induced dilatancy in the fail-
ure wave. As illustrated in Figure 3.1(c), the process could lead to a fixed failure wave ve-
locity and occurs with small changes in particle velocity and longitudinal stress through
the failure wave. The dilatant void volume required to accommodate the behavior can be

22
 Failure Waves

STRAIN

STRAIN

STRAIN

1 Transition states for failure waves in brittie solids. In each

example the shock state at point one is below the Hugoniot
elastic limit O^ and 0 ^ represent uniaxial strain and mean
stress loading paths, respectively.

23
estimated. A pressure change is related to volumetric lattice strain "u^^^ through the bulk
modulus K according to

dp = -K—^, (3.2)

However, if dilatant void volume dv^j, is introduced, the total specific volume change is
dv = dVi^j + dVj-i and

^P = -^^^^-d^dii^- (3.3)

For the purpose of simplicity assume a constant strain process through the failure wave,
dv = 0, and,

^ ' . f . (3.4.

If it is further assumed that do^ = 0 and that o^ = c after passage of the failure wave,
then dp = 4T/3 , where x is the shear stress in the elastic state preceding the failure wave,
and

- ^ =|p. (3.5)

For glass shocked to about 5 GPa on the Hugoniot, x = 1.7 GPa, K == 56 GPa, predicting a
dilatant strain of about 4%. The complete transfer of elastic shear strain energy to dilatant
strain energy yields a comparable result.

The concept of failure waves clearly contributes an intriguing added dimension to the pro-
cesses of dynamic failure of brittle solids, and extends a challenge to the emerging consti-
tutive models intended to describe such failure. Further experimental effort is needed to
more carefully constrain the observable features of failure waves. Dilatant inelastic strain
within the compressive failure process should be examined further. It is critical to identify
other materials in which the effect occurs.

24
 Dynamic Spall Process

4 Dynamic Spall Process

Shock compression techniques also offer the unique opportunity of testing the dynamic
tensile properties of matter without the complications of sample structure or surface prep-
aration considerations. In the high-strength ceramics, dynamic tensile (spall) strengths are
typically about one to two orders of magnitude lower than the corresponding dynamic
compressive strength. Spall strengths have previously been related to fracture surface en-
ergy with some success [Grady, 1985; Grady, 1988], and such comparisons are examined
further in the present work with spall data for selected ceramics.

4.1 Spall Strength of Ceramics

When an isolated and stationary plate of test material is impacted by a thinner plate of a
selected impact material, shock waves are created in both samples at the impact plane.
The transmitted shock waves reflect as waves of decompression at the respective free sur-
faces, meet within the thicker plate, and carry the interior plane at the collision point of
the waves rapidly into tension. If the tensile stress and duration is sufficient, interior frac-

0.25 \
Spall
Precompression • Plane
Shock state
0.20
VISAR

J 0.15 LiF
Window
Decompression PMMA ^11^
>- Wave Sample

0 0.10
_l
LJ
>
0.05 Fullback
Signal

0.00
0.0 0.5 1.0 1.5 2.0 2.5
TIME (/zs)
Figure 4.1 Velocity interferometry profile for a dynamic tensile fracture
experiment in aluminum nitride ceramic illustrating elastic
shock precompression and subsequent spall signal.

25
tures will nucleate, grow and coalesce at this plane, causing material separation and relax-
ation of the tensile stress to zero.

Such a spall experiment can also be performed if a low impedance VISAR window mate-
rial is bonded to the back of the stationary test plate — the amplitude of the reflected de-
compression wave is determined by the impedance difference between the sample and
window plates. The resulting interface velocity profile from such a spall experiment, in
which a plate of aluminum nitride ceramic backed by a lithium fluoride transparent win-
dow material was subjected to planar impact by a thinner plate of polymethyl methacry-
late (PMMA), is shown in Figure 4.1. Laser velocity interferometry (VISAR) was used to
monitor the interface velocity history.

The initial jump in velocity amplitude quantifies the magnitude of the precompression
shock wave. The release wave feature identified as the pullback signal images, at the re-
cording interface, the tension magnitude and duration occurring at the spall plane. Spall
strength values are calculated from the tensile magnitude of this signature through consid-
eration of material impedance differences [Grady and Kipp, 1993]. To first order the spall
stress can be calculated from,

where Z^ and Z^ are the elastic impedances of the sample and window, respectively, and
A M ^ is the magnitude of the pullback signal as seen in Figure 4.1. Further discussion of
experimental spall relations is provided in Section 8 of this report which deals with wave
profile measurements on tungsten carbide.

Spall data for selected ceramics are provided in Table 4.1. The shock precompression in
the experiments providing this spall data was appreciably less in magnitude than the corre-
sponding Hugoniot elastic limit for the material. For later comparisons Hugoniot elastic
limit data are also provided along with other selected properties.

4.2 Calculations of Spall Strength

If the interior of a crystalline body is subjected to a tensile stress, and if the body is free of
micro structural or atomic defects, then the spall strength of that body is determined by the
forces necessary to overcome the molecular cohesive energy of the material. One estimate
for the magnitude of this spall strength is provided by the approximation of Orowan (see
Lawn and Wilshaw (1975)),

a,^ = - . (4.2)

This expression is recognized to over-estimate the theoretical tensile strength and more
reasonable values based on models of the intermolecular forces have been obtained. Such
studies have been performed for ceramics [Izotov and Lazarev, 1985; Kozhusko et ai,
1987] and theoretical tensile strengths a , from their work are provided in Table 4.1.

26
 Dynamic Spall Process

Table 4.1:
Elasticity and Dynamic Strength Properties of Ceramics

P E V be >meas
(kg/m^) (GPa) (nm) (GPa) (GPa) (GPa) (J/m2) (J/m2)

AIN 3250 321 .24 .19 44.9 7-9 0.5-0.6 1.2/6.2 28-48
AI2O3 3970 401 .24 .19 59.5 11-12 0.5-0.8 2.1/7.7 15-61
B4C 2510 462 .17 .16 64.5 17-19 0.4-0.5 1.4/7.5 23-40
SiC 3120 434 .16 .19 55.9 15-16 0.3-0.4 1.4/8.4 7-17
SiOj 2630 90 .08 .16 24.0 7-9 .08-. 11 1.0/1.5 5-14
TiB2 4510 523 .05 .24 42.6 14-17 0.4-0.5 0.8/12.7 20-34
WC 14930 627 .21 .25 52.2 4-6 2.5-3.0 1.0/15.9 490-845
Zr02 6030 218 .31 .20 69.4^ 13-16 1.5-1.9 4.4/4.4 640-1300

^ Izotov and Lazarev, 1985 (see text).

^ Equation 4.12/ Equation 4.4 (see text).
^ Based on a,^ = E/n (see text).

Measured spall strengths of engineering ceramics (as provided in Table 4.1) are observed
to be about two orders of magnitude lower than the theoretical strengths, indicating that
the inherent or induced flaw structure of the material is crucial in determining the dynamic
tensile strength of ceramics.

High quality ceramics are usually exceedingly homogeneous. It is only on the microstruc-
tural scale that material heterogeneities emerge which are responsible for the stress con-
centrations that nucleate internal fracture and spall. Influence of the flaw structure on the
spall strength may also depend on the rate of tensile loading during the spall process.

The strain rate and flaw structure sensitivities of spall in brittle materials which have
emerged from previous theoretical studies can be explored within the context of the plot in
Figure 4.2. For an unflawed material the spall strength will reside on the theoretical
strength curve (upper curve in Figure 4.2) as discussed previously. In general the spall
strengths of engineering ceramics fall within the bounding curves identified.

The fracture-producing flaw structure of a ceramic plays a critical role in the processes of
spall failure. Such a flaw structure may be inherent due to defects, incompatibilities, or in-
ternal stress states in the initial microstructure. In addition, flaw structure may be induced
by large-amplitude compressive shocks which precede dynamic tensions and can cause
microstructural compressive fracture damage.

A possible extension of the latter is the failure wave following the compressive shock
which would be expected to dramatically alter the flaw structure. In either case the materi-

27
 Theoretical
Spall Strength
^o
\ 1
/
LU

0)

(f)
\
V

^
\
\
Allowed
Spall V
Domain

\ ^ -
\ ^
\
\

^
/
/
x Dynamic
Spall Strength

V
/ /
Flaw-Dominated
(Quasi-Static)
Spall Strength

STRAIN RATE
Figure 4.2 Regions of spall behavior inferred from energy-based
theories of tensile failure of brittle solids.
ized by a critical flaw size a^. The characteristic length a^ is regarded as a property of the
material which leads to catastrophic spall fracture when a Griffith criterion,

(4.3)

is achieved, and when strain rates in tension are sufficiently low. Equation 4.3 is based on
the solution for internal penny-shaped cracks of radius a^. The parameter y is the surface
energy which must be overcome by the elastic strain energy to initiate fracture growth,
and is also a property of the material.

In the limit of ideal brittie fracture the surface energy is that necessary to separate molecu-
lar planes over the cohesion dimension h^ and, from the Orowan analysis [Lawn and
Wilshaw, 1975], is estimated to be.

, 2

T = T,y, = (4.4)

More generally, extension of a fracture involves additional dissipation through plasticity

or comminution processes and the fracture surface energy can be written as.

28
 Dynamic Spall Process

1 = <^lth^ (4.5)

where a is greater than unity.

Using Equation 4.3-4.5 the relation for spall strength can be scaled to the theoretical ten-
sile strength of the unflawed crystal.

Tt^c

For a = 1 (theoretical surface energy y^^) the spall strength is provided by the lower
horizontal line in Figure 4.2. For dissipative fracture ( a > 1), the spall strength will lie
within the interior of the region.

When tensile strain rates achieve levels in which the characteristic crack of dimension a^
no longer responds in a quasi-static manner to the time-varying stress field, spall strengths
are predicted to become dependent on the tensile loading strain rate and independent of
the length scale a^. Strain rate dependent spall has been investigated through the time-de-
pendent stress intensity factor of a single crack [Kipp et al., 1980], through statistical
crack distributions [Grady and Kipp, 1989], and, more recently, through thermodynamic
energy conditions for spall fracture [Grady, 1988]. These studies all suggest the same
functional dependence for the spall strength, and the latter provides the strain rate depen-
dent criteria.

( 2 3 V/3
""sp
= [6p C ytj . (4.7)

Again y = ay,^ is the surface energy dissipation in crack extension, and c is the elastic
wave velocity.

Equation 4.7 predicts a spall strength which is independent of the length scale characteriz-
ing the flaw structure at sufficientiy high strain rates. This strain-rate regime is bounded
from below by the transition strain rate e, to the flaw-dominated strength and from above
by the critical strain rate e^ shown in Figure 4.2. The spall strength resides on this line
when y = y,^ and above it otherwise. The dynamic spall strength converges to the theo-
retical strength when the higher strain rate is approached. The dynamic (high-strain-rate)
spall strength can be related to the theoretical strength,

e^l/3
^sp = [ « - J ""th- (4.8)

The critical strain rate e^ can be interpreted as the rate at which the velocity of separation
of atomic planes is such that the kinetic energy of relative separation is comparable to the
cohesive energy. This threshold strain rate.

29
is functionally the same as the crystal Debye frequency and numerically similar. The as-
sumption of reasonable properties for ceramics (see Table 4.1) yields an e^ of the order
of 10^^/s-lO^-^/s — well above the strain-rate range experienced in application. The excess
surface energy parameter a can be expected to depend on strain rate since the spall
strength cannot exceed the theoretical strength in Equation 4.8.

A lower transition to flaw-dominated spall, characterized by the dominant flaw length

scale a^, occurs at the strain rate.

(4.10)

Equation 4.10 would also be applicable for y = ay^^^ by replacing y,^ with y. Assuming
the reasonable values of y,^ = 5 J/m^, p = 4000 kg/m^, and a^ = 10 |im leads to an
e, - 10 /a. Thus characteristic flaw dimensions in excess of about 10 |im are expected to
be necessary to accommodate the dynamic spall behavior predicted by Equation 4.7 since
strain rates of this order are typically achieved in the impact spall experiment. Dynamic
and static spall can be related through,

^.p(dyn) = ( f ) ' ' ' a , , ( s t a t ) . (4.11)

The theoretical surface energy of a crystalline solid is subject to some uncertainty. Based
on the Orowan analysis, one estimate leads to [Lawn and Wilshaw, 1975],

Eb^
y.h = — • (4-12)
71

Alternatively, Equation 4.4 relates surface energy drrectiy to the theoretical strength of the
material. Values from both relations are provided in Table 4.1 for the ceramics of concern,
providing a measure of both the magnitude and the uncertainty in calculating the theoreti-
cal surface energy.

The spall data for the same materials are used, through the relation (Equation 4.7),

3
'P (4.13)
, 2 3.
6p C E

to estimate the surface energy consistent with measured dynamic tensile failure. The strain
rate consistent with the impact tests has some uncertainty. Here we have used the ampli-

30
 Dynamic Spall Process

tude and the duration (width at half maximum) of the spall pullback signal to calculate the
strain rate. Values between lO'^/s to 10^/s were determined depending on the material. The
wave velocity c = JET^ was used in Equation 4.13. The theory leading to
Equation 4.13 does not specify this elastic property in detail.

Experimental values of y^^^^ are provided in Table 4.1 and can be compared with esti-
mated theoretical surface energies. For most of the materials the measured value is less
than one order of magnitude higher than the theoretical value, consistent with their ex-
pected brittle nature. Tungsten carbide and zirconium dioxide are about two orders of
magnitude higher; consistent with the additional deformation complexities of these two
materials. The tungsten carbide was a liquid-phase sintered ceramic (see Section 8) and
the zirconium dioxide was a transformation toughened material.

31
5 A Failure Criterion
A criterion for the onset of fracture within the shock process has been sought by various
authors to describe the behavior of high-strength brittie solids. The familiar Tresca and
von Mises criterion, under constraint of uniaxial strain loading, both reduce to.

where the unconfined tensile failure stress T^ is a property of the material. While this cri-
terion has been reasonably successful in describing the response of shock-loaded metals, it
has not been adequate for ceramics. Other workers have attempted to develop more gener-
al pressure and rate dependent criteria to better describe dynamic failure in ceramics
[Steinberg, 1991; Johnson and Holmquist, 1993; Curran et ai, 1993]. Recentiy
Rosenberg (1993) has suggested the applicability of the Griffith failure criterion for ce-
ramics and has demonstrated the ability of this criterion to capture several of the unique
shock effects in these brittle materials.

The criterion developed by Griffith (1924) is based on the stress concentrations at the tips
of a population of microcracks assumed to pervade the material subjected to a stress load.
The theory identifies a critical stress state at which favorably oriented cracks achieve a
value characteristic of the material at which crack growth initiates. The analysis carried
out by Griffith was strictiy two dimensional and resulted in the criterion,

io^-<5y)^ = 87^(a^ + o p , if a^ + 3ay>0, (5.2)

o = -T^, otherwise. (5.3)

Compressive stresses are positive and Equation 5.2 and 5.3 apply when o^>C5 . The ma-
terial constant T^ is again the uniaxial tensile strength.

A rigorous three-dimensional extension of the Griffith criterion has not apparentiy been
carried out. Rosenberg (1993) has proposed a direct application of the Griffith form
(Equation 5.2) to the three dimensional uniaxial geometry {O^^C5 = o^),

( a ^ - a p ^ = 87^ (o^ + o p , if o^ + 30^ > 0, (5.4)

Murrel (1963) has suggested a logical modification to three dimensions based on geomet-
ric symmetry considerations. When reduced to the uniaxial environment this criterion be-
comes,

(a^ - crp ^ = 127^ (o^ + 20y), if a^ + 5o^, > 0, (5.5)

Although differences in absolute predictions between the two extensions of Griffith's cri-
terion are noted, the important trends are similar. The unconfined compressive strength Y^

32
 A Failure Criterion

1 1 1 1 1 1 1

4
1
/ ,Lwc
• von Mises •

<D • Criterion Griffith

Criterion
Rosenberg's
II
.1 - ^"^^--..^^ Modification
TZr02 -
''"'•~~-~..^ / J.
~-^^^^ -

iTiB2 7--....^^^ ^ \
" f t B4C^>.
SiC / ^ \ .

.01
0.0
ISiOo
'f '^

0.1 0.2
Murrel's
Modification
N>-
0.3 0.4
POISSON'S RATIO

Figure 5.1 Comparison of the spall/HEL ratio for selected ceramics witii
predictions based on von Mises and Griffith failure criteria.

is significantiy higher than the tensile strength — in agreement with observed behavior of
ceramics. For Rosenberg's extension y^ = 87^ while for Murrel's extension 7^, = 127^.

In the shock wave environment the spall strength is equal to the unconfined tensile
strength, a = T^, while the Hugoniot elastic limit o^^^, or compressive strength under
uniaxial strain loading conditions, can be related to the spall strength. Rosenberg's exten-
sion yields.

^hel = 8 - ~~2^sp^ (5.6)

(1-2V)

whereas Murrel's modification results in.

" * „ - 12 il^. . 2 sp-

(5.7)
(l-2v)

In Figure 5.1 a plot of the ratio of the spall strength to the Hugoniot elastic limit for select-
ed ceramics is provided from data in Table 4.1. Also shown are the predicted tensile fail-
ure levels based on von Mises and Griffith failure criteria. Relating the tensile yield based

33
on a von Mises criteria to the ductile spall strength is not strictiy correct although plastic
strains at yield appear to initiate cavity nucleation and growth [Curran et al, 1987] and
observed ductile spall strengths are usually comparable to or somewhat higher than the
yield stress.

Although agreement in detail is not observed, the Griffith criteria for brittie failure is con-
sistent in trend with low value of this ratio for most ceramics. Tungsten carbide has shown
other indications of nonbrittie behavior, while the zirconium dioxide tested was a transfor-
mation toughened material, possibly accounting for the higher spall strengths and higher
values in Figure 5.1. The low value for Si02, the one geological material included, is con-
sistent with comparisons of the Griffith criterion with strength of other rock materials,
[e.g., Janach (1977)].

Like the Griffith criterion, the dynamic relation for spall in brittle materials (Equation 4.7)
also follows from thermodynamic energy principles. Consequently, it is reasonable to ex-
plore the Griffith failure criterion within the dynamic regime of behavior through inclu-
sion of the early spall relation. Working with Murrell's modification (Equation 5.5) a
prediction for the Hugoniot elastic limit based on fundamental material properties is

<^Hel = 12-J^ ^l^^pcyt] . (5.8)

• (l-2v)''
Using reasonable properties for aluminum oxide (v = 0.23, p = 4000 kg/m^, c = 10 m/s,
Y = 5 J/m^) and assuming an effective strain rate range of lO'Vs-lO^/s, yields a Hugoniot
elastic limit of about a^^^ = 14-30 GPa which can be compared with values of 6-12 GPa
for reasonably competent alumina ceramics.

Equation 5.8 is not intended as an accurate predictor of the shock strength properties of
brittle materials. It should be regarded as a first-order expression for the dynamic com-
pressive failure of an ideal flawed brittle material in the sense of Griffith. Numerous mi-
crostructural issues known to influence dynamic strength (i.e., grain structure, porosity)
are not accounted for. Nonetheless, the reasonable agreement with measured strengths of
ceramics would lend support for the brittle failure mechanics leading to Equation 5.8 and
provides additional argument for a fi-acture-dominated process controlling failure in the
compressive shock loading environment.

Equation 5.8 also predicts a strong cube root strain rate dependence of the Hugoniot elas-
tic limit. As will be discussed in Section 10, examination of elastic precursor decay in ce-
ramics does not support this aspect of Equation 5.8, and weakens the argument for a brittie
mechanism in the shock wave front.

34
 Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide

6 Shock-Wave Strength Properties of Boron Carbide and

Silicon Carbide
Time-resolved velocity interferometry measurements have been made on boron carbide
and silicon carbide ceramics to assess the dynamic equation-of-state and strength proper-
ties of these materials. Hugoniot precursor characteristics, and post-yield shock and re-
lease wave properties, indicated markedly different dynamic strength and flow behavior
for the two carbides.

Thus, in this study, shock and release wave profiles in boron carbide and silicon carbide
have been measured from peak stress states that are just in excess of the Hugoniot elastic
limit to peak stresses approaching 60 GPa. Silicon carbide reveals an uncommonly high
Hugoniot elastic limit (~15-16 GPa) compared to many materials. Post-yield strength of
silicon carbide, determined by comparison of Hugoniot uniaxial strain and calculated hy-
drodynamic response, reveals neutral or increasing strength with subsequent deformation
beyond the initial dynamic yield. Boron carbide exhibits a somewhat higher Hugoniot
elastic limit (~ 18-20 GPa). In contrast, however, subsequent deformation indicates a dra-
matic loss in strength supporting capability. Hugoniot and hydrodynamic response for bo-
ron carbide converge at stresses approaching about twice the Hugoniot elastic limit,
suggesting littie or no shear stress component at higher Hugoniot states. The contrasting
dynamic strength characteristics of silicon carbide and boron carbide are further amplified
in the release properties of these materials from the Hugoniot state. The release paths for
silicon carbide indicate reverse yielding and continued strength characteristics of elastic-
plastic material behavior. The unloading stress-volume paths for boron carbide closely
parallel the calculated hydrodynamic behavior suggesting near fluid-like response with
sustained loss of strength. Further features in the measured wave profiles indicate hetero-
geneous deformation in boron carbide in contrast to homogeneous deformation in silicon
carbide. Microscopic mechanisms which may be responsible for the strikingly different
shock and deformation properties of these two ceramics are considered.

6.1 Materials and the Shock-Wave Experiment

The silicon carbide tested in the present study was supplied by Eagle Picher Industries.
The density of this material is 3177 kg/m^. The longitudinal and shear elastic velocities
are 12.02 and 7.67 km/s, respectively. The material has a porosity of approximately 1%
and a nominal grain size of 7 |im. Knoop hardness for this ceramic is 22.3 [Grady, 1994].
Shock wave data for this material have previously been reported [Kipp and Grady, 1989;
Grady and Kipp, 1993]. The boron carbide investigated in the current work was provided
by Dow Chemical Company. The density is 2506 kg/m^. Longitudinal and shear elastic
velocities are 14.03 and 9.65 km/s, respectively. The nominal grain size is 3 |im, porosity
of the order of 1%, and Knoop hardness is 25.6. Properties and microstructure differ
slightiy from another boron carbide provided by Eagle Picher Industries for which shock-
wave data have previously been reported [Kipp and Grady, 1989].

Planar shock and release wave experiments were performed on the monolithic ceramic
samples with a single stage powder gun capable of 2.5 km/s maximum projectile velocity.

35
 2.0 I I 1 r- -I 1 r-

Silicon
Carbide j ^ W » ^ * ^ ^ ^

1.5 -

E
1.0 -
u
o
_J
LtJ
> 0.5 -

0.0
0.5 1.0 1.5 2.0
(a) TIME (/xs)

en
\
E

o
o
_l
LJ
>

(b) TIME (/xs)

ure 6.1 Shock and release wave profiles for silicon carbide and boron carbide
ceramics measured with velocity interferometry diagnostics.

36
 Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide

Plates of the same ceramic, or a high density metal, were backed by low density polyure-
thane foam, mounted on the projectile, and caused to impact stationary target plates of the
test ceramic. Target samples were backed by lithium fluoride windows approximately
25 mm in thickness and 50 mm in diameter. The transmitted particle velocity profiles pro-
duced by the impact-generated shock waves were measured with laser velocity interfer-
ometry (VISAR) techniques [Barker and HoUenbach, 1972]. Details of the experimental
method have been reported earlier [Grady, 1992a; Grady, 1992b]. Measured shock and re-
lease wave profiles for silicon carbide and boron carbide are provided in Figure 6.1. Ex-
perimental dimensions are noted in Table 6.1. In all tests the peak stress (ranging between
about 25 and 50 GPa) exceeded the Hugoniot elastic limit of the ceramic. Striking differ-
ences in wave profile characteristics relating to the elastic precursor wave, the Hugoniot
state, and the release structure are noted between the two ceramics. These differences are
discussed further in the subsequent subsections.

Table 6.1:
Experimental Conditions for Impact Tests.

Foam^ Projectile Projectile Target Target Impact

Test
Density Material Thickness Material Thickness Velocity
Number (kg/m3) (mm) (mm) (km/s)
CE-4 320 SiC 3.987 SiC 8.939 1.542
CE-5 640 SiC 3.995 SiC 8.940 2.100
CE-31 640 Tantalum 1.516 SiC 8.956 2.118
CE-17 320 B4C 4.831 B4C 10.322 1.633
CE-18 640 B4C 4.815 B4C 10.346 2.076
CE-26 640 Tantalum 1.514 B4C 9.680 2.059

^ Polyurethane foam discs backing the projectile impact plate were approximately 6 mm
in thickness.

6.2 Elastic Precursor Characteristics

Details of the elastic precursor waves for silicon carbide and boron carbide are shown in
Figure 6.2. The Hugoniot elastic limit is defined here as the break in slope following the
steep rise of the initial wave arrival — a profile velocity of about 0.55 km/s for silicon car-
bide and about 0.7-0.8 km/s for boron carbide. Stress values for the Hugoniot elastic limit
calculated by impedance matching methods are about 15-16 GPa for silicon carbide and
18-20 GPa for boron carbide. Post-yield characteristics of the precursor waves are dramat-
ically different for the two materials, however. For silicon carbide the positive slope of the
precursor wave reveals a subsequent hardening with further deformation to stresses in ex-
cess of 20 GPa. Whether pressure hardening or deformation hardening plays the more
dominant role cannot be uniquely determined from the shock wave data. Nevertheless,

37
 1.2
Silicon
Carbide
1.0

M
0.8 I-
E
>» 0.6 -
^Post-Yield
O v^^ Hardening
Hugoniot
o 0.4 -
Elastic
limit
0.2 -

0.0
0.7 0.8 0.9 1.0 1.1
(a) TIME (MS)

1.2 1 1 r-

Boron
Carbide
1.0 -
Hugoniot
m Elastic
0.8 -
E
>_ 0.6 -
CJ
O Tost-Yield
0.4 Softening

0.2

/
0.0
0.7 0.8 0.9 1.0 1.1
(b) TIME i/uLs)

Figure 6.2 Details of the elastic precursor wave profiles measured with VISAR
diagnostics for silicon carbide and boron carbide.

38
 Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide

significant resistance to dislocation plasticity or compressive shear fracture continues to

persist in silicon carbide as dynamic shear deformation in the shock process proceeds.

Precursor waves for boron carbide, in contrast, indicate post-yield stress softening. Stress
relaxation suggests a rate-sensitive deformation process, probably accompanied by pre-
cursor attenuation with propagation distance. The latter behavior has been confirmed by
shock profile measurements on thinner samples (see Section 10).

Finally, it is worth noting in Figure 6.2 the nearly a factor of two difference in time sepa-
ration of the elastic wave and first indication of the following deformation shock wave be-
tween silicon carbide and boron carbide. This feature also reveals distinct differences in
post-yield strength characteristics of the two ceramics.

6.3 Hugoniot Properties

Hugoniot states achieved in the shock process are assessed by estimates of the elastic and
deformation shock velocity from the profile data, and particle velocities from either the
symmetry of the impact, the known Hugoniot curve of the impactor material, or the mea-
sured amplitudes from the wave profiles. The velocity of the elastic precursor was not
measured in the present study. Instead the longitudinal elastic velocity was augmented by
an estimate of the non-linearity of the material at the Hugoniot elastic limit
U^^i = Ci + SUf^^i, assuming a representative value of 5 = 1, to provide the elastic shock
velocity t/^^^. This estimated elastic shock velocity is about 4 to 5 percent larger than the
elastic wave speed C^. The deformation shock velocity is then referenced to the elastic
shock wave velocity.

Shock velocity versus particle velocity data for silicon carbide and boron carbide are
shown in Figure 6.3. Both Hugoniot elastic limit and final Hugoniot states are identified.
The dashed curve represents the bulk sound velocity, C^ = C^ - 4C^/3, determined from
ultrasonic data of Manghnani (1994) to 2 GPa and extrapolated to higher pressure using
the reported values for the zero pressure bulk moduli K^ and the pressure derivative K^ .
Hugoniot states for silicon carbide are in good agreement with the extrapolated bulk ve-
locity and consistent with a material which retains a shear strength at the Hugoniot pres-
sure comparable to the strength at the Hugoniot elastic limit. In contrast, shock velocities
for boron carbide are markedly lower than bulk velocities suggesting significant reduction
in the shear strength at the Hugoniot state. Open circles for boron carbide represent the
new data (Figure 6.1 and Table 6.1). Closed circles correspond to an earlier material
which exhibited a slightiy reduced Hugoniot elastic limit and somewhat higher deforma-
tion shock velocities [Kipp and Grady, 1989].

Hugoniot states are plotted in stress versus specific volume space in Figure 6.4. In addi-
tion to the data from Figure 6.3, Hugoniot states for one experiment each for silicon car-
bide and boron carbide below the Hugoniot elastic limit reported previously [Kipp and
Grady, 1992] are shown. Also a recentiy measured Hugoniot point to nearly 58 GPa, using
a two-stage light gas gun on a 3-mm thickness boron carbide sample, is included. Also
shown in Figure 6.4 are estimates of the bulk pressure versus volume response of both ce-
ramics, again based on the ultrasonic data of Manghnani (1994). A serious uncertainty in

39
 16 I I I I I I I I

Silicon
Carbide
(0
14
E O p„ = 3177kg/m3

o
o
10 -

U Deformation
\ Shock
O 8|- Bulk
Sound
Velocity
I I I

0.0 0.4 0.8 1.2 1.6

PARTICLE VELOCITY (km/s)
(a)
16 -1 1 1 1-
-•—r
Boron
Carbide

• p„ = 2517kg/m-
Elastic O Po = 2506kg/m'
Shock
12 -

O
O
—I 10
Ui
> \
Bulk
o Sound
O 8 /
Velocity
Deformation
Shock
• • '
' ' '
0.0 0.4 0.8 1.2 1.6

(b) PARTICLE VELOCITY (km/s)

Figure 6.3 Hugoniot shock velocity versus particle velocity data for silicon carbide
and boron carbide.

40
 Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide

this construction arises from an unclear knowledge of the theoretical density of the ceram-
ic materials. Ideally the theoretical density appropriate to this development should repre-
sent the zero porosity material accounting for any sintering impurities or differences in
chemistry from the common formula unit. The reference densities used in Figure 6.4 sim-
ply account for the nominal 1 % porosities of these ceramics and may be uncertain by as
much as a percent either way.

Comparisons of the Hugoniot data in Figure 6.4 with the estimated bulk response general-
ly support observations relating to shock velocity data in Figure 6.3. For silicon carbide
the measured Hugoniot states are offset substantially from the pressure-volume curve in-
dicating retention of a substantial shear stress component. Boron carbide, on the other
hand, upon exceeding the Hugoniot elastic limit, exhibits a dramatic increase in com-
pressibility and achieves final Hugoniot states near the bulk pressure-volume curve. Post-
yield behavior for boron carbide would suggest nearly complete loss of shear strength.
The fact that the 40 and 58 GPa Hugoniot points actually lie to the left of the estimated
pressure-volume curve is of some concern. It may simply reflect the uncertainty in the ini-
tial theoretical density of this ceramic. It should be remembered, however, that the unique
deformation processes initiated during shock compression could potentially lead to en-
hanced volumetric lattice compression (with corresponding change in lattice structure)
which exceeds the calculated stable lattice compression based on the lower pressure ultra-
sonic data. Considering the unusual, very open structure of the boron-carbide lattice [Em-
in, 1987], an anomalous volume compression under shock loading should not be ruled
out.

Finally, complete shock and release wave profiles from several of the tests have been
matched with one-dimensional wave code calculations by Kipp [Kipp and Grady, 1989;
Grady and Kipp, 1993] using adjustable parameter computational models. The three
dashed curves for silicon carbide and two for boron carbide in Figure 6.4 represent the
stress-volume load and release curves corresponding to a satisfactory fit to these wave
profiles. The nature of the curves for silicon carbide corresponds remarkably well to a
high-strength elastic-plastic material. Simulations of the boron carbide wave profiles, on
the other hand, suggest loss of strength above the Hugoniot elastic limit and near-fluid-
like behavior on stress release.

6.4 Discussion

Markedly different dynamic compression and release behavior are noted in the shock
compression characteristics of silicon-carbide and boron-carbide ceramics. The common
methods of contrasting shock compression states with measured or estimated hydrostatic
pressure-volume behavior suggests near-metal-like shock properties for silicon carbide in
its strength-retention characteristics under a cycle of shock-wave compression and re-
lease. A similar comparison for boron carbide, however, indicates catastrophic loss of
strength above the Hugoniot elastic limit with near-fluid behavior during subsequent de-
formation in the shock load and release cycle. Such comparisons can be misleading, how-
ever, and other interpretations of the data are possible. The possibility of a phase-change-
like volume collapse in boron carbide under shock compression should not be ruled out.

41
 60
Silicon
Carbide
o 50 -
Q.
O
40 - 3177 kg/m3 _

to
bJ Bulk Computational.
30 - Pressure Shock/Release
a: Volume Paths
m Curve
o 20
z
o
o 10 h

0.25 0.27 0.29 0.31 0.33

(a) SPECIFOfiJ^yT/^^/Mg)
60 — ' — I •>•••• 1 ' I ' 1 '

o Boron
Carbide
^50
• \

40 h • 2517 kg/m^ -
(/) O 2506 kg/m^
I/) • \
Bulk / ^
Pressure Computational -
I— Shock/Release
t/) Volume
Curve /Paths
Q 20 -
o
o
\
\ \
Z5 10 -

0 ,_l L_ I

0.32 0.34 0.36 0.38 0.40 0.42

(b) SPECIFIC VOLUME (m^/Mg)
Figure 6.4 Data points are elastic and final shock Hugoniot states. The solid curve is
the bulk pressure versus volume behavior from extrapolation of ultrasonic
data. Dashed curves correspond to wavecode solutions to experimental
VISAR profiles.

42
 Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide

Another intriguing feature in the velocity interferometry data of Figure 6.1 should be not-
ed. f*rofiles for silicon carbide are smooth and regular whereas corresponding profiles for
boron carbide show an erratic and irregular component to the measured motion. The
VISAR laser beam is focused to a spot of about 25-50 fxm in diameter in these experi-
ments. Consequently the measured motion represents an average over this spot size. Other
workers [Atroshenko et al., 1990; Meshcheryakov et al., 1991] have discussed the effect
on VISAR data of differential interface motion at various spatial scales. Differential mo-
tion on a scale less than the spot size will lead to reduction in VISAR contrast because in-
terference maxima and minima at different points within the spot will be achieved at
different times. Differential motion on a scale larger than spot size will not appreciably af-
fect contrast. Random elastic wavelets from nearby points removed from the laser spot
and undergoing differential motion can, however, lead to irregular motions at the record-
ing point. VISAR data on both the subscale (less than spot size) and mesoscale (greater
than spot size) for silicon carbide indicate homogeneous motion under shock loading.
VISAR data for boron carbide, however, suggests homogeneous motion on the subscale,
but heterogeneous motion on the mesoscale. The latter result may suggest a heteroge-
neous deformation process under shock loading within the 50-500 nm spatial scale. The
concentration of deformation energy into discrete deformation zones along with localized
material softening or melting has been proposed as a mechanism for reduced-strength flu-
id-like behavior in brittie solids under shock wave load and release [Grady, 1980].

43
7 Shock Phase Transformation and Release Properties of
Aluminum Nitride
Aluminum nitride is characterized by a non-recoverable volume phase transformation
from the wurtzite (hexagonal) to the rocksalt (cubic) structure that commences under
shock compression at about 22 GPa. The phase transformation has been reported by Voll-
stadt, et al. (1990) in static high pressure experiments, and by Kondo, et al. (1982) in
shock wave experiments. In the latter study, the discrete data, obtained with pin and elec-
tromagnetic gage techniques, clearly indicated the presence of the phase transformation.
The current study expands on that uniaxial data with time-resolved measurement of parti-
cle velocity histories of the shock and release states. Four experiments are reported, all
with peak impact pressures of approximately 40 GPa, well above the phase transformation
pressure. The evolution of the shock wave with propagation distance is addressed with
variations of the target layer thickness. Details of the material response are obtained by
numerical analysis techniques developed to examine the response of several other ceram-
ics to shock loading [Kipp and Grady, 1990]. The data obtained in the present study pro-
vide a basis for development of accurate computational models that can represent the
response of this ceramic to high strain-rate loading in shock wave propagation codes.
Such models can then be used to examine the consequences of the large energy-absorbing
aspect of the volumetric phase transformation of this ceramic during shock loading events.
Useful data for aluminum nitride have also been provided in the longitudinal and lateral
shock wave measurements of Brar, et al. (1992), and shock compression studies of phase
transformations by Nakamura and Mashimo (1994). Elastic properties and their pressure
and temperature dependence have been investigated by Gerlich, et al. (1986), while Van
Camp, et al. (1991) have explained equation of state issues in aluminum nitride as well as
other binary nitrides. Shock and equation-of-state properties of aluminum nitride have re-
centiy been reviewed by Dandekar et al., (1994).

7.1 Materials and Experiment

The aluminum nitride used in this study was produced by DOW Chemical Corporation.
The nominal reference density of the aluminum nitride was 3262 kg/m^. Ultrasonic mea-
surements on samples of this material indicate a longitudinal velocity of 10730 m/s and a
shear velocity of 6320 m/s. Associated elastic properties derived from these velocities are
a Poisson's ratio of v = 0.234, and bulk sound speed of 7866 m/s. The hot-pressed ceramic
had a nominal porosity of 1% and a grain size of approximately 2 |im.

Pure tantalum and tungsten impactors were used in order to obtain initial shock amph-
tudes well above the phase transition point of the aluminum nitride at the impact velocities
attainable with the impact facility employed. The lithium fluoride used as the laser win-
dow material in the present impact experiments was optically clear [100] oriented single
crystals [Wise and Chhabildas, 1985].

Uniaxial-strain compressive shock and release waves were produced in the aluminum ni-
tride with a single-stage powder-gun facility as discussed in Section 2. The impactors
were either tantalum backed by polyurethane foam or tungsten backed by polymethyl

44
 Shock Phase Transformation and Release Properties of Aluminum Nitride

methacrylate (PMMA) and were accelerated with the powder gun to velocities of about
2200 m/s, providing shock amplitudes between 37 and 41 GPa in aluminum nitride. The
thicknesses of the tungsten and tantalum impactor plates were about 1.5 mm. The plate
impactor thicknesses in these experiments were designed so that the complete shock and
release profile could be measured at the recording interface.

The suite of target ceramic thicknesses led to variations in available time for both the de-
velopment of the shock wave structure and the subsequent interaction of the release wave
with the propagated shock wave. Table 7.1 lists the key parameters associated with these
experiments.

Table 7.1:
Uniaxial Strain Impact Experiments on Aluminum Nitride

Test Density Thickness Impactor Impactor Impactor Hugoniot

No. (kg/m3) (mm) Materials Velocity Thickness Stress
(m/s) (mm) (GPa)

CE44 3265 2.507 PMMA / Tungsten 2281 1.502 41

CE43 3262 2.510 Foam / Tantalum 2207 1.528 37
CE52 3258 4.183 Foam / Tantalum 2215 1.526 37
CE37 3265 9.567 Foam / Tantalum 2239 1.508 38

The measured particle velocity profiles of all four experiments are shown in Figure 7.1.
The arrival times of the initial, or "elastic", waves correspond to the ultrasonic longitudi-
nal transit time through the aluminum nitride targets. The two highest amplitude wave
profiles (both with arrival time at 0.2 |is) are for the tungsten (CE 44) and tantalum (CE
43) projectiles impacting 2.5 mm of aluminum nitride. The impactor thicknesses, impac-
tor velocities, and aluminum nitride target thicknesses are of comparable magnitude for
both these experiments (Table 7.1). The higher impedance of the tungsten, and its slightiy
larger impact velocity results in a larger amplitude particle velocity than for the corre-
sponding tantalum case. This in turn leads to a much earlier arrival of the upper shock
wave characterizing the shock-induced phase transformation. The initial break in the
shock wave occurs at about 400 m/s, and represents onset of permanent deformation, or
"yield", as typically observed in many ceramics tested in this one-dimensional strain con-
figuration [Kipp and Grady, 1990; Grady, 1992]. The second break which appears, at
about 900 m/s, is believed to correspond to onset of the wurtzite (hexagonal) to the rock-
salt (cubic) phase transformation, with a non-recoverable volumetric phase transforma-
tion. The release wave structure substantiates the non-recoverable aspect of this transition,
as will be demonstrated by evaluation of the one-dimensional numerical calculations of
the experimental wave profiles.

In three experiments, the thickness of the tantalum flyer plate was maintained at approxi-
mately 1.5 mm, and the impact velocity was fixed at about 2200 m/s (Table 7.1). Al-
though for the 2.5 mm aluminum nitride target a wave with some time at the peak between

45
 2000 T—I—I—I—I—I I I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—r

2 1000
>

^^ 500

Pu
0 J b _i_i I I I t_Ji I I I I I 1 I I I I I I I I I I I I I L

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Time (/xs)
Figure 7.1 Particle velocity records for tungsten impactor (1.5 mm, 2200 m/s)
impacting 2.5 mm aluminum nitride target, and tantalum impactor
(1.5 mm, 2200 m/s) impacting 2.5 mm, 4.2 mm, and 9.6 mm
aluminum nitride targets.

shock and release is recorded, increasing the thickness of the aluminum nitride target ef-
fectively extends the propagation distance of both the release wave, which emanates from
the trailing low impedance interface of the flyer, and the initial shock wave. Given suffi-
cient propagation distance, the release wave will overtake the lower velocity phase trans-
formation wave, and the interaction will attenuate the amplitude of the initial wave. The
third profile in Figure 7.1 (4.2 mm aluminum nitride) shows the upper part of the shock
wave just beginning to be eroded by the release wave at the recording interface. The
9.6 mm aluminum nitride target is of large enough thickness that almost complete attenua-
tion of the upper shock wave has occurred by the time the rear surface of the ceramic tar-
get has been encountered.

There are two other noticeable trends in the particle velocity records for the tantalum im-
pactor series in which the thickness of the aluminum nitride target increases from 2.5 mm
to 9.6 mm: (1) the initial amplitude of the break in the first shock wave decays from about
400 m/s to 320 m/s, and (2) the amplitude of the second break, identified with the phase
transformation, drops in amplitude from 900 m/s to 840 m/s.

7.2 Compression and Release Properties

The complex structure of the particle velocity history, particularly above the phase trans-
formation point in both the shock and release waves, is a consequence of impedance dif-

46
 Shock Phase Transformation and Release Properties of Aluminum Nitride

ferences at the impact and recording interfaces of the target. The tantalum and tungsten
impactors have significantly larger impedances than the aluminum nitride targets, and the
impedance of the lithium fluoride window is less than one-half that of aluminum nitride.
As a consequence, when the release wave from the trailing interface of the impactor ar-
rives at the impact interface, some unloading will occur. In the same way, at the aluminum
nitride / lithium fluoride window interface, the reflected partial release wave will interact
with the higher amplitude, slower shock of the phase transformation. The numerical tech-
niques used here account for aU these wave interactions, and clarify the original wave
structure in the aluminum nitride targets.

Analysis of the data was made with a one-dimensional Lagrangian finite-difference wave
propagation code, WONDY [Kipp and Lawrence, 1982]. The aluminum nitride material
response was characterized with parameterized sound velocity and strain linear segments,
with separate loading and unloading curves defined so that both shock and release proper-
ties of the ceramic could be represented [Kipp and Grady, 1990]. Linear variations in
sound speed with strain result in a quadratic dependence of the local material modulus on
strain. Particular care was required to monitor the current amplitude and rate of loading so
that the cortect unloading and reloading curves were used for the local material response.
The local material modulus was calculated from the sound speed based on the local strain
and the appropriate sound speed segment, and determined the increment in axial stress at
each time step in the calculation. Ultrasonic data were used to determine the initial load-
ing moduli. The parameters of the sound speed curves for the inelastic response were de-
termined iteratively, tailored to match the particle velocity history for each experiment. By
taking account of the interactions of the waves at the lithium fluoride / aluminum nitride
and flyer / aluminum nitride interfaces, this approach provided the capability to examine
the ceramic response interior to the samples, and in particular, to determine the effect of
the volume phase transformation on the transmitted wave structure.

The tantalum and tungsten impactors were modeled with an anisotropic strain hardening
model in WONDY [Kipp and Lawrence, 1982]. Shock and release data for each of these
materials in this regime of pressure verified the parameters used to characterize these ma-
terials. Data for the tungsten were obtained from Chhabildas, et al. (1988) and for the tan-
talum from Chhabildas and Asay (1992). The release wave that originates at the trailing
interface of the impactor is transmitted to the target through the impact interface. It is of
particular importance that the initial structure of this transmitted release into the alumi-
num nitride sample be as accurate as possible. Otherwise, modeling errors in the impac-
tors will corrupt the analysis of the particle velocity data by requiring compensating errors
in the model constructed for the aluminum nitride.

Numerical analysis of the experiment in which a 2.5 mm target was impacted with a
1.5 mm tantalum projectile (Experiment CE 43) resulted in the fit to the particle velocity
data shown in Figure 7.2(a). Nearly all of the features of the data are captured by this anal-
ysis technique; iterations on the sound speed - strain function focused primarily on the
first microsecond of the record in order to reproduce the shock and release response of the
aluminum nitride in as much detail as possible. The fit to the particle velocity data in
which a 2.5 mm aluminum nitride target was impacted by 1.5 mm tungsten projectile (Ex-

47
 zo -|—I—I I I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—r-

WONDY
Data

Experiment CE 43:
1.5mm Ta - > 2.5mm AIN
• • I J
0.0 I I i_ ' '
0.0 0.5 Lo L5 ao Z5
(a) Time (yiis)

ao I I I I I—I I I I I I r I I I I I I I I I I I

WONDY
Data
m
B 1-5 h

O LO I-
p—H

0)
>
'o
05
(-1
(d
PL. Experiment CE 44:
1.5mm W - > 2.5mm AIN
Q Q I • *| • • I ' ' ' ' I • t ' » I ' ' « ' I • • •

• 0.0 05 LO L5 ZO Z5
(b) Time (/xs)
Figure 7.2 (a) Numerical fit to interface particle velocity data for tantalum
impact on 2.5 mm aluminum nitride (CE 43). (b) Numerical fit to
interface particle velocity data for tungsten impact on 2.5 mm
aluminum nitride (CE 44).

48
 Shock Phase Transformation and Release Properties of Aluminum Nitride

periment CE 44) is shown in Figure 7.2(b). The sound speed - strain functions used to fit
these two experimental profiles differ only slighdy at large strains, in order to accommo-
date the increased velocity of the upper shock at the higher impact pressure. The unload-
ing curves are identical for these two cases.

The appearance of a transmitted stress wave in an aluminum nitride target is recorded in

Figure 7.3(a). The three-wave structure is much more well-defined in this representation
than in the particle velocity history at the target / window interface, because no interface
wave interactions have occurred yet. The structure in the particle velocity data in both
Figure 7.2(a) and Figure 7.2(b), beginning at 0.35 |is and ending at about 0.5 |is in the
particle velocity interval 900 m/s to 1250 m/s, is all due to wave interactions at the target /
window interface, and is not present in the transmitted wave prior to encountering the in-
terface. The three primary waves are identified in Figure 7.3(a) as the initial elastic wave,
the deformation (or "plastic) wave, and the phase transformation wave.

The stress - strain response that corresponds to this transmitted wave profile is shown in
Figure 7.3(b). The first (elastic) wave has an amplitude of approximately 10 GPa,
Figure 7.3(a) and Figure 7.3(b), at which point a break in the curve indicates the onset of
yield. This amplitude agrees well with 9-10 GPa reported by Nakamura and Mashimo
(1994), and 9.4 GPa reported by Brar, et al. (1992). The next break in the wave occurs at
22 GPa, which compares well with 21+1 GPa shock transition pressure reported by Kon-
do, et al. (1982), and 19 GPa reported by Nakamura and Mashimo (1994). The total axial
strain at the beginning of the phase transformation is about 7.5%. The maximum ampli-
tude of the shock wave in this case corresponds to about 28% strain, with the phase trans-
formation accounting for about 20% strain between the onset and the maximum load. This
volumetric compression agrees well with the value of 22% obtained by Kondo, et al.
(1982) from shock wave experiments. These results also compare favorably with static
measurements, in which a transition pressure of 16.5 GPa and volume strain of 21% were
reported by VoUstadt, et al. (1990).

Note that as the aluminum nitride unloads from peak compression of about 40 GPa down
to about 10 GPa, the unloading progresses nearly elastically, recovering about 6% strain,
and does not follow the initial compression curve that included the phase transformation
in Figure 7.3(a). This indicates that no reversal of the phase transformation on unloading
to approximately 10 GPa has occurred in the elapsed time of this experimental configura-
tion, and that the deformation (volume strain) accompanying the phase transformation is
permanent. Were there a reversal of the phase transformation during unloading, a much
more complex representation of the unloading curve would be required.

The functional relationship of sound speed to strain determined for the aluminum nitride
that engenders the accurate fit to the particle velocity data in Figure 7.2(a) is shown in
Figure 7.4(a). The discrete nature of the sound speed fitting segments and the distinct
loading and unloading curves are quite evident in this figure. The reference zero-strain ve-
locity is the ultrasonic velocity, followed by a slight increase (non-linear elastic) to the
yield point at about 2% strain. A sudden drop in sound speed to 8300 m/s occurs to repre-
sent the "plastic" deformation wave. At the onset of the phase transformation, there is a
another large decrease in sound speed to accommodate the volumetric compression of the

49
 40 T 1 I I I I I I I I I I I I I I

Phase Release
Transformation Wave
Wave
cd 30 -

10
CO
Q)
20 Plastic
Wave

Cd

10 -
Elastic
Wave ^ .
WONDY
J i_i ' • • J I I I L.

0.0 0.5 1.0 1.5 ao

(a) Time (fis)

40

Pheise
32 Transformation
cd

OH
24 -
CO
m
Q)
u
t/3
i-H
Cd
.1-1
X
<

WONDY
6 12 IB 24 30

(b) Strain (%)

ure 7.3 (a) Transmitted wave profile in aluminum nitride from tantalum
impact, (b) Stress-strain response of aluminum nitride.

50
 Shock Phase Transformation and Release Properties of Aluminum Nitride

phase transformation as shown in Figure 7.4(a). The sound speed then gradually recovers
with increasing strain as the phase transformation deformation limit is approached. The
sound speed function between "onset" and "completion" determines the structure in the
particle velocity histories at the recording interface above the "plastic" wave (0.35 -
0.5 |is, Figure 7.2(a) and Figure 7.2(b)). The separate unloading sound speed - strain func-
tion (Figure 7.4(a)) accommodates the large change in wave velocities between shock and
release states in the aluminum nitride. The parameters for this unloading curve were also
derived iteratively to match the release waves in Figure 7.2(a) and Figure 7.2(b). In the
calculation, when material begins to unload from compressive states, the local modulus
evaluation abruptly changes from the loading curve to the unloading curve.

The shock and particle velocities associated with the transmitted wave profiles for both
tantalum (2207 m/s, CE 43) and tungsten (2281 m/s, CE 44) 1.5 mm impactors onto alu-
minum nitride are plotted in Figure 7.4(b). The shock velocity is calculated from the arriv-
al time of an increment of the stress wave at a chosen monitoring position in the
aluminum nitride, and the distance from the impact interface to the monitored point. The
displaced position of this monitoring point is included in the shock velocity calculation.
Data points from Kondo, et al. (1982) are compared with this data extracted from the
analysis of the present experiments. We observe that there is a good agreement with the
two data points at a particle velocity of 8.6 km/s, defining the shock velocity of the plastic
deformation wave. Three other data points from Kondo, et al. (1982), in the particle ve-
locity range of these experiments, are also in excellent agreement with the shock veloci-
ties determined for the phase transformation wave. The data point near 175 m/s is in the
elastic regime, albeit somewhat lower than that obtained in the material of the current
study. (The reported longitudinal velocity in [Kondo, et al., 1982] is 10000 - 10400 m/s
compared to the 10730 m/s for the material used here.) From the calculated change in
shock velocity with particle velocity at the phase transformation wave, an effective coeffi-
cient, S, in the linear shock velocity - particle velocity relationship is determined to be
about 4.8; this is much larger than the value of about 2.0 determined to be the observed
upper limit for shock compression of the preponderance of single phase materials [Jean-
loz, 1989]. Kondo, et al. (1982) report shock velocity versus particle velocity data to
about 3000 m/s, and a value of S of at least 3.4 is suggested by their fit to that data. These
large values of S indicate that there are significant differences in the nature of compress-
ibilities between single and multiple phase materials.

7.3 Discussion

The present experiments and analysis of aluminum nitride provide substantial evidence
for a non-recoverable volumetric phase transformation under shock compression at a pres-
sure of about 22 GPa. The fine time-resolution of the data permit accurate assessment of
the initial permanent deformation, the onset of the phase transformation, the loading path
through the transformation to final shocked states of 40 GPa, and the unloading behavior
from these shocked states. Numerical analysis provides insight into the development of
the three wave structure that forms from the initial shock at the impact interface, and is an
effective analytical utility for establishing dynamic stress versus deformation behavior.
The data are generally consistent with previously reported data for aluminum nitride.

51
 14 -1—r
Loading Unloading
Curve- Curve

Phase Transformation
^_— Onset
0)
Q) 8 - Completion—.«^
PH
CO

d 6 -
o
CO 4 -

WONDY .
6 12 18 24 30
(a) Strain (%)

14 1 — • — • ' — ' — • — I — ' — < -

Elastic Wave

Plastic Wave

Phase
Transformation
Wave

"O!
P
WONDY. CE 43
—- WONDY. CE 44
O Kondo. et al. [2]
0.0 0.5 1.0 1.5 ao
(b) Particle Velocity (km/s)

Figure 7.4 (a) Characterization of aluminum nitride with sound speed function
of strain, (b) Shock velocity versus particle velocity response of
aluminum nitride.

52
 Shock Phase Transformation and Release Properties of Aluminum Nitride

These data and analyses of deformation provide a solid basis for the construction and
evaluation of material models for this material, particularly in the accounting of the large
energy dissipation that accompanies the phase transformation.

53
8 Shock-Compression and Spall Measurements on
Tungsten Carbide
Velocity interferometry wave profile measurements have been made on several tungsten
carbide ceramics. Hugoniot states for these materials ranging from approximately 5 to 60
GPa were investigated. A Hugoniot elastic limit of approximately 4 GPa, followed by
strong hardening behavior, was determined. Several tests examined the spall failure of
tungsten carbide and provided measurements of spall strengths in the range of 2.7-3.6 GPa
which is markedly larger than for other ceramics. The velocity profiles provided specific
shock and high-strain-rate material property data along with response measurements for
computational model validation.

8.1 Background

Tungsten carbide is a high density ceramic with reasonably attractive shear and tensile
strength properties in certain applications. The current study was undertaken to investigate
the dynamic equation of state and strength properties of this material for purposes of com-
putational modelling under high velocity impact applications. The effort was initiated and
completed within a relatively short time frame and consequently does not represent a thor-
ough investigation of the dynamic properties of tungsten carbide. Material from two sup-
pliers was tested. Both shock compression and spall data were obtained from the series of
seven shock profile measurements conducted.

A search for earlier shock wave data on tungsten carbide was also conducted in this study.
Some previous work was uncovered. Hugoniot data on WC-5wt%Co was generated by
McQueen et al. (1968; 1970) and has been tabulated in the LANL Shock Hugoniot Com-
pendium [Marsh, 1980]. Also, several compression wave profiles on tungsten carbide us-
ing the free-surface capacitor technique were measured by Taylor, and by Hopson, and
published in the LANL Shock Wave Profile Compendium [Morris, 1980]. Gust (1980) in-
dicates shock Hugoniot experiments were performed at Lawrence Livermore National
Laboratories, however only a Hugoniot elastic limit is provided in this reference. Stein-
berg (1991) has assimilated the above data and provided partial material property parame-
ters in the framework of the Steinberg-Guinan model.

8.2 Materials and Experiment

Two tungsten carbides prepared by pressureless liquid-phase sintering were tested in this
effort. The first was a fully dense tungsten carbide provided by Kennametal Inc. The other
was extracted from 14.5 mm AP (BS-41) rounds. The Kennametal tungsten carbide con-
tains 5.7% Co, 1.9% Ta and less than 0.3% Nb and Ti. The reported Rockwell-A hardness
is 93 and the compressive strength is 5.8 GPa. The AP material contains 3-47c Ni, 0.4-
0.8% Fe and 0.05-0.2% Co. The Rockwell-A hardness is 86-92. Static compressive
strength for this material is 44 GPa and split cylinder test results provided a tensile

54
 Shock-Compression and Spall Measurements on Tungsten Carbide

strength of 240-270 MPa [Holmquist, 1994]. Density and ultrasonic properties for the two
materials are provided in Table 8.1.
Table 8.1:
Elastic Properties for Tungsten Carbide
Density c, Co Bulk Poisson's
Material (kg/m3) (km/s) (km/s) (km/s) Modulus Ratio
(GPa)
KM^ 14930 6.895 4.165 4.941 364.5 .213
AP' 14910 6.918 4.149 4.991 371.4 .219
^ KM refers to the K68 material obtained from Kennametal Inc. whereas AP identifies
the tungsten carbide extracted from 14.5 mm (BS-41) armor piercing rounds.

Uniaxial strain compressive shock and release waves were produced in the tungsten car-
bide with a single stage powder gun facility as described in Section 2. A disc of impactor
material (either tungsten carbide or aluminum in the present tests) was mounted in the
projectile and was backed by a disc of polymethyl methacrylate (PMMA).

Table 8.2:
Experimental Conditions for Tungsten Carbide Impact Tests.
Projectile^ Projectile Target'' Target Impact
Test
Material thickness Material Thickness Velocity
Number
(mm) (mm) (km/s)

WC-1 WC(KM) 3.370 WC(KM) 6.566 1.141

WC-2 WC(KM) 3.363 WC(KM) 6.542 1.566
WC-3 Al(6061-T6) 12.87 WC (AP) 2.985 1.039
WC-4 Al (6061-T6) 12.51 WC (AP) 2.994 0.687
WC-5 Al(6061-T6) 1.047 WC (AP) 2.986 0.361
WC-6 Al (6061-T6) 1.038 WC (AP) 2.980 0.446
WC-7 Al(6061-T6) 1.030 WC(KM) 3.357 0.405

^ All impactors were backed by PMMA.

^ Target diameters for test WC-1, WC-2 and WC-7 were slightly greater than
63 mm. For tests WC-3 through WC-6 diameters were just under 11 mm.

55
For the target, a disc of the tungsten carbide ceramic was mounted in the stationary sup-
porting target fixture and an optical quality disc of single crystal lithium fluoride was inti-
mately bonded with epoxy to the back of this ceramic sample. VISAR velocity profiles to
be discussed subsequently were obtained by the methods described in Section 2. The im-
pact parameters and component dimensions for the present series of impact experiments
on tungsten carbide are provided in Table 8.2

8.3 Dynamic Strength

Compression wave profiles for Tests WC-1, WC-2 and WC-3 are shown in Figure 8.1.
Test WC-3, which was performed on an approximately 3 mm sample of the AP tungsten
carbide has been scaled to the 6.55 mm thickness of the Test WC-1 and WC-2 KM sam-
ples assuming centered wave behavior. Thicknesses of both the elastic and deformation
shock for the WC-3 profile is probably unduly spread by this scaling. Comparison of the
three profiles indicates consistent shock properties and comparable dynamic strengths for
the two materials. Arrival times of longitudinal and bulk elastic waves are indicated to
compare with arrivals of the deformation shocks in Figure 8.1. Tests WC-4 through WC-7
also provide compression profile information, although an unusual velocity excursion,
which may be an experimental artifact and is discussed later, compromises this data to
some extent.

The 6.5 mm thickness Kennametal tungsten carbide samples reveal a fairly distinct transi-
tion from the elastic precursor rise to the transition ramp region at approximately 70 m/s.
Based on an acoustic impedance calculation this leads to a Hugoniot elastic limit of
1.4
WC-1
1.2 h

S 1.0 h WC-2
J 0.8

^ 0.6 -
D
3 0.4
WC-3
0.2 -

0.0
0.8 1.0 1.2 1.4 1.6 1.8
TIME (/xs)
Figure 8.1 Compression shock profiles for tungsten carbide.

56
 Shock-Compression and Spall Measurements on Tungsten Carbide

4.1 GPa. The thinner AP sample tested, WC-3, did not provide as distinct a transition lev-
el, however, a yield at about 70 m/s is not inconsistent with the behavior of this material,
and indicates similar yield characteristics for both materials.

Material behavior within the transition ramp regime was analyzed in two ways. First,
computational simulations of the precursor and transition ramp portions of the wave pro-
file were performed using the one-dimensional Lagrangian wave code WONDY-IV [Kipp
and Lawrence, 1982]. With this technique empirical parameters for the longitudinal mod-
ulus as a function of amplitude are adjusted until a best fit to the wave profile in the region
of interest is achieved. The stress-strain relation derived from this fit is then accepted as
the dynamic response of the tungsten carbide. The experimental profiles for Tests WC-1
and WC-2 are shown with the one-dimensional wave code solution in Figure 8.2.
0.4

WONDY Solution
Experimental Profiles
0.3
(n
E WC-2

0.2
O
O

0.1 \ Linear Slope

in Ramp

Q Q l-« nVr~.lr~i>^l I I I I I I I I I I I I I I I I I 1 1-
'0.90 0.95 1.00 1.05 1.10 1.15
TIME (/xs)

Figure 8.2 Compression shock profiles and WONDY computational

solution emphasizing the ramp region of the elastic precusor.

An analytic solution was also performed assuming a linear rising velocity versus time be-
havior within the region of the ramp (Figure 8.2) and writing the wave velocity as,

(8.1)
'• 1 + Au/u'

where u = a l/c^. The precursor velocity is c^, the thickness of the target plate is /, and
a is the constant acceleration determined from the slope of the ramp wave in the mea-

57
sured profile. Also, AM = u-u^, where u^ is the particle velocity at the Hugoniot elastic
limit. Both a^ and Aw have been corrected for the impedance difference between tung-
sten carbide and lithium fluoride assuming an elastic impedance for the WC and LiF,
(AM = (^WC + ^ L / F ) ^ " " /2Zw^P). Stress and strain are calculated from the Riemann
integrals, do = p^c^ (M) du and de = du/c^ (M) .

Results of the stress-versus-strain behavior within the ramp region determined from both
the computational simulation and the analytic solution for the axial CJ^ - e behavior are
provided in Figure 8.3. Comparison is made with the linear mean stress curve calculated
from the ultrasonic bulk modulus and with the difference 3 ( a - a ) / 4 which would
provide the shear stress behavior if the volumetric compressibility during the shock com-
pression process tracked the ultrasonic compressibility.

If there is onset of an inelastic deformation mechanism at the 4 GPa break in the measured
wave profile in tungsten carbide, as common understanding of the process tends to sup-
port, then the ramp structure in the profile cannot be described with simple von Mises
plasticity as indicated in Figure 8.3. Several hardening mechanisms may potentially ex-
plain the observed behavior. The elastic loading path could contact a pressure hardening
yield surface at the Hugoniot elastic limit (HEL). Subsequent loading would lead to in-
creased stress rise due to the increasing shear stress with confining pressure on the pres-
sure hardening yield surface. Altematively, the HEL may correspond to the intersection of
the elastic loading path with a deformation hardening yield surface. Due to post-yield de-
formation in the shock wave the initial yield surface would evolve through a succession of
12.0 I I 1 I

Data — Analytic Solution

Data — WONDY Solution
Elastic Perfect Plastic .•(

.y
8.0 .<?'
.•i"
o.
o 4"
Q-
O - Bulk
Compression
(/) / (Ultrasonic)
LJ
OH
I—
(/) 4.0

^ -(o -a )
4^ J m'

0.0
0.000 0.005 0.010 0.015 0.020
STRAIN
Figure 8.3 Stress versus strain behavior in ramp region of the wave
profile.

58
 Shock-Compression and Spall Measurements on Tungsten Carbide

states to a final yield surface consistent with later states in the ramp wave. Evolution of the
yield surface would be determined by a deformation hardening law and could account for
the ramp behavior characteristic of the early inelastic deformation.

There is a further, less commonly considered, mechanism for hardening which could ac-
count for the observed wave profile structure. Rather than an increase in the stress devia-
tor with inelastic deformation, an increase in the mean stress state above that predicted by
compressibility of the fully dense solid, could account for the observed behavior. Such an
elevated mean stress versus volume curve is possible if dilatant void volume is generated
during the shear deformation process.

Of course, all three of the hardening mechanisms proposed could be operating collective-
ly, and rate dependence of these effects may also be playing a role. Of the three hardening
mechanisms, pressure hardening is the most difficult to reconcile with the second order
character of the elastic-to-inelastic deformation transition. (The transition to a ramp wave
represents a discontinuous jump in the curvature of the stress-versus-strain curve and not
the slope.) In principal deformation hardening or dilatancy hardening laws could be con-
structed which could simulate the observed transition to ramp wave structure.

Although the present wave profile experiments on tungsten carbide serve to illuminate
some of the possible mechanisms of inelastic deformation operating in the shock com-
pression of this material, they are not sufficient to discriminate among the initial yield and
hardening mechanisms. Further dynamic testing is needed. In particular tests measuring
both longitude and transverse stress would be especially revealing.

8.4 Hugoniot Properties

Hugoniot states for the present tungsten carbide were determined assuming a precursor
velocity equal to the longitudinal ultrasonic elastic velocity of 6.95 km/s. Hugoniot states
for Tests WC-1, WC-2 and WC-3 are shown in a shock velocity versus particle velocity
plot in Figure 8.4 and compared with the data of McQueen et al. (1968, 1970). The two
higher amplitude states are in reasonable agreement with the date of McQueen. The rela-
tively broad shock wave for Test WC-3 lends some ambiguity to the experimental shock
velocity at this amplitude. Nevertheless this data point cannot be reconciled with the mea-
surements of McQueen, although the trend of the lower three or four data points of Mc-
Queen tend to fall above the linear expression. Precursor states and ramp region for these
data are also indicated in Figure 8.5. Corresponding Hugoniot pressure versus specific
volume states for this tungsten carbide are provided in Figure 8.6.

The tungsten carbide tested by McQueen et al. (1968, 1970) was also approximately 5%
cobalt by weight, however the reported density of 15,013 kg/m^ is about 1/2% higher than
the present Kennametal material. This density difference accounts for about half of the
discrepancy between the McQueen et al. Hugoniot and the present Hugoniot data in
Figure 8.6. Since the present Hugoniot states were calculated based on an ultrasonic ve-
locity for the elastic precursor (a lower limit), it appears that the present Hugoniot is
slightiy stiffer than that reported by McQueen and coworkers.

59
 8.0 1 1 1 ' 1 ' 1 '

(0
1 •° McQueen et al.
Present Data ^<P^
7.0
E r~ci
^ ^ ^ ^ ^

O6.0 1 ^.1^6 Steinberg ^

o
py ^ ^ - ^ ^
J

o -
X
in 1 °
4.0
Li^s 1 1 , 1 .
0.0 0.5 1.0 1.5 2.0
PARTICLE VELOCITY ( k m / s )

Figure 8.4 Shock velocity versus particle velocity for tungsten carbide.

7.5
—I ' r
Hugoniot
McQueen e\ al.
Elastic
(0 7.0 - ^ / Limit Present Data
E
Ramp
6.5 Wave
O
O

^ 6.0

O Steinberg
o
^ 5.5

5.0
0.0 0.2 0.4 0.6 0.8 1.0
PARTICLE VELOCITY ( k m / s )
Figure 8.5 Shock velocity versus particle velocity data for tungsten
carbide emphasising lower pressure data, including precursor
elastic limit and ramp wave region.

60
 Shock-Compression and Spall Measurements on Tungsten Carbide

80

70 - O McQueen et al.
• Present Data
o
a.
o 60 -

50 -
UJ
DiL
h- 40 -
in Steinberg
^—
O 30
z
o Ramp
o 20 - Wave
Hugoniot"
10 - . , Elastic -
^ / ^ Limit

55 70
SPECIFIC VOLUME (cm^/kg)

Figure 8.6 Hugoniot stress versus specific volume for tungsten carbide.

8.5 Spall Strength

Release waves originating from the rear of the aluminum impactor and from the lithium
fluoride window interface interact within the tungsten carbide samples carrying the mate-
rial rapidly into tension. The dynamic tensile strength (spall strength) of tungsten carbide
was exceeded in the present tests and the time history of the spall process was imaged in
the measured spall pullback signal within the velocity profile recorded at the window in-
terface. Interface velocity profiles for test WC-5, WC-6, and WC-7, in which spall oc-
curred, are shown in Figure 8.7.

The spall strength of tungsten carbide can be calculated from the spall pullback signal
identified in Figure 8.7. The present data provide an opportunity to compare two methods
for determining the spall strength when an interferometer window is used. The appropri-
ate velocity levels are identified for Test WC-6 in Figure 8.7 for the spall analysis. First, if
the full amplitude of the pullback is used, then the appropriate relation for the spall
strength is

% = 2^^s''Z^) ('^max-^mJ Z u (8.2)

w max

61
 0.14

0.12 I- - — WC-5
- - - WC-6
^-.. 0.10 WC-7
E
w 0.08 h
t
o 0.06 -

UJ
> 0.04 -

0.02 -

0.00
0.0
TIME ((IS)

Figure 8.7 Wave profiles emphasizing spall pullback signals in

tungsten carbide.

In Equation 8.2 Z^ and Z^, are the shock impedance of the sample material and window
material, respectively. Equation 8.2 assumes approximately linear behavior of stress-strain
response of the sample and window material and ignores the complications of corrections
for wave dispersion and elastic-plastic behavior which have been discussed in the litera-
ture [Grady and Kipp, 1993]. The down side of Equation 8.2 is that it can be a difference
of two large quantities each of which may have uncertainties. This is not a serious problem
with the present data, however.

An alternative expression for the spall stress is provided by the relation,

a =-(Z+Z)(u-u ,) (8.3)
sp 2 * w' ^ o r,

Equation 8.3 does not suffer from the differencing problem of Equation 8.2. On the other
hand it assumes that stress at the spall plane has relaxed to zero before the reflected wave
has returned to the recording interface. This depends on experiment design and the time
history of the spall process, but often this relaxation is not complete and the spall strength
calculated from Equation 8.3 is a lower bound. With the present data spall strengths have
been calculated with both Equation 8.2 and Equation 8.3. Results are provided in
Table 8.3 and plotted in Figure 8.8.

62
 Shock-Compression and Spall Measurements on Tungsten Carbide

Table 8.3:
Spall Properties of Tungsten Carbide

Test ^max min

Material^
Number (nVs) (nVs) (m/s) (GPa) (GPa)
WC-5 AP 81 1 56 3.56 3.21
WC-6 AP 108 21 82 3.60 3.56
WC-7 KM 96 26 71 2.75 2.62
^ KM refers to the K68 material obtained from Kennametal Inc. whereas AP identifies
the tungsten carbide extracted from 14.5 mm armor piercing rounds.
^ Analysis method one (Equation 8.2).
^ Analysis method two (Equation 8.3).

8.6 Discussion

It is important to recognize that the two tungsten carbide ceramics tested in the present
study are, in fact, metal ceramic mixtures and that this feature may profoundly influence
the dynamic properties uncovered in this study. Metals added in the ceramic processing
congregate at ceramic grain boundaries and junctions, filling most of otherwise porous re-
gions in the ceramic. Volume fraction of metal in the Kennametal (KM) material is 11%-
12% while the AP material is slightiy less at 87c-10%.

This metal content may be responsible for the reduced strength of tungsten carbide
(Hugoniot elastic limit of approximately 4.0 GPa) in comparison with other ceramic mate-
rials. On the other hand the strong hardening behavior of this material may also reflect the
metal component. The higher compressibility of metal relative to the ceramic would lead
to increasing ceramic-ceramic contact with increasing pressure, and consequentiy increas-
ing manifestation of ceramic strength with pressure.

The remarkably large spall strength of tungsten carbide is probably a consequence of en-
hanced toughening brought about by the metal component. Differences in spall strength
between the Kennametal and AP materials need to be studied further but are most likely a
consequence of differences in bonding characteristics of the nickel-iron and the cobalt-
tantalum in tungsten carbide. Anomalous excursions (glitches) in the compression veloci-
ty profiles for Tests WC-4, WC-5 and WC-6 were noted (see compression profiles for
tests WC-5 and WC-6 in Figure 8.7). Considerable effort was expended in attempting to
explain these features but we were unsuccessful. They appeared clearly in the data 1 and
data 2 traces (90 degree out of phase velocity fringe records) and not in the beam intensity
so they appear to record a true velocity excursion. We found nothing in the target geome-
try that could lead to a wave reflection. Nonetheless, it is our opinion that this feature is an
artifact of the experiment.

63
 AP
Material
KM
Material •

2 -
Hugoniot
Elastic
Limit
1 - Analysis method one

Analysis method two

2 4 6 8 10
HUGONIOT STRESS (GPa)

ure 8.8 Spall strengths determined from the present tests on

tungsten carbide.

64
 Shock Compression Experiments on Quartz Rock Investigating a Critical Stress

9 Shock Compression Experiments on Quartz Rock

Investigating a Critical Stress Transition in the Si02
Structure
Two shock compression experiments on the same quartz rock tested by Zhugin and Krup-
nikov (1987) have been performed using high resolution velocity interferometry diagnos-
tics [Grady and Zhugin, 1994]. Peak stress levels of approximately 26-28 GPa were
achieved in the present tests. A Hugoniot elastic limit of 6.0 GPa and a compression
anomaly at 23.4 GPa on the shock Hugoniot were observed in the experimental results.
The latter stress value is in good agreement with a transition stress of 23 GPa previously
reported by Zhugin and Krupnikov (1987). A critical transition stress level identified in
the work of Zhugin and Krupnikov (1987) and in the present study suggests a need for re-
examining both earlier interpretations of shock wave results for quartz, and theories of the
shock transition process. A closer relation between the shock transition process and the
static crystalline-to-amorphous transition at about 20 GPa reported by Hemley et al.
(1988) should be considered. Theories of the transformation based on free-energy/kinetic
concepts (Sikka, 1992) warrant serious consideration in light of the existence of a critical
shock transition level, and the remarkably close dynamic and static transition stress val-
ues. Earlier interpretation of a constant shock velocity segment of the Hugoniot through
the mixed-phase region over a broad stress range may be erroneous in light of the more
complex shock structure observed. Reinterpretation would suggest a concave Hugoniot
segment with higher density dynamic compression states at comparable stress levels. The
present tests suggest transition toward the high-density 6-coordination structure of Si02
rather than the relatively minor crystallographic change associated with the coesite phase.
Similarities among the present wave profile measurements on Si02 and comparable mea-
surements on Zr02 are noted.

9.1 Background

At the All-Union Workshop on Physico-Chemical Properties of Matter convened at Lake

Baikal, Russia during the period of August 17-24, 1990, Zhugin and Krupnikov reported
the first definitive observation of a wave profile feature indicative of a critical transforma-
tion stress in the a-quartz form of Si02 [Zhugin and Krupnikov, 1987]. In that work, sam-
ples of quartz rock were subjected to plane shock waves between about 21 and 29 GPa.
Transmission compressive stress wave profiles were measured with high-resolution pi-
ezoresistive manganin stress transducers. A compression anomaly was noted in the data at
approximately 23 GPa — presumably related to the onset of lattice instability and phase
transformation in the Si02 structure. Since a critical transformation stress had not been
noted in earlier shock-wave studies on Si02, common thinking was that the 4- to 6-coordi-
nation transformation was diffused over a broad range of pressure. This behavior was
thought to account for the anomalous constant shock velocity and linear Hugoniot seg-
ment observed over the correspondingly broad particle velocity range.

At the Lake Baikal Meeting it was concluded that independent measurements of the wave
profile characteristics with a different time-resolved diagnostic technique would lend sup-
port for this previously unreported feature. Time-resolved velocity interferometry mea-

65
surements were made at the Sandia National Laboratories Shock-Compression Facilities
and the authors of the earlier work kindly provided samples of the same quartz rock tested
by them.

9.2 Experimental Method

ControUed-impact planar shock-wave experiments were performed on the propellent

launch facility described in Section 2. Projectile and target configuration for the shock ex-
periments on quartzite are illustrated in Figure 9.1. A tungsten impact plate launched at
near the maximum velocity of the propellent gun (-2.3 km/s) was judged sufficient to ex-
ceed the 23 GPa transition stress reported by Zhugin and Krupnikov (1987). The low den-
sity PMMA (polymethyl methacrylate) backing was intended to provide for a partial
release wave following the initial compressive shock wave. Velocity history of the com-
pressive shock wave was measured at an interface between the quartz rock sample and a
lithium fluoride window with the diffused-surface velocity interferometry (VISAR) meth-
od of Barker and HoUenbach (1972).

Two shock-compression experiments of the configuration just described were successfully

completed. Dimensions and impact parameters for the two tests are provided in Table 9.1

Tungsten
Impact
Target
PMMA Assembly
Backing

3 Velocity Pins

Lithium-Fluoride
Window

Aluminum
Ring Target
Projectile Aluminum Fixture
Body Nose Plate

Figure 9.1 Experimental configuration for shock compression and release

measurements in quartz rock.

66
 Shock Compression Experiments on Quartz Rock Investigating a Critical Stress

Table 9.1:
Experimental Parameters

Sample Tungsten Quartz PMMA Impact

Test
Density Thickness Thickness Thickness Velocity
Number (kg/m3) (mm) (mm) (mm) (km/s)
SVl 2645 2.990 4.996 6.382 2.19
SV2 2645 2.989 4.936 6.396 2.30

Tungsten disc diameter — 35.0 mm.

Quartz disc diameter — 35.0 mm.
Lithium fluoride dimensions — 19.1 mm thickness by 25.2 mm diameter.

9.3 Experimental Results and Analysis

The measured experimental profiles for Tests SVl and SV2 are provided in Figure 9.2. An
expanded view of the wave front is shown in Figure 9.3. The arrival times of the two wave
profiles are arbitrary and were selected to easily visualize the separate profiles. The transit
time of the elastic shock wave through the quartz sample was not measured.

Lithium fluoride is a reasonably good impedance match for quartz (LiF has a slightiy
higher shock impedance) and the observed profiles are believed to be representative of
the in-material motion history.

An elastic shock wave was clearly observed in both tests with a profile amplitude very
close to 0.38 km/s. Based on earlier studies (Grady et al, 1974) on a similar density
quartz rock, an elastic shock velocity of 6.2 km/s was assumed in order to perform Hugo-
niot analysis on the present compressive shock-waves.

A profile amplitude of approximately 1.40 km/s was determined for the second plateau
amplitude for both SVl and SV2 immediately before the ramp in velocity up to the veloc-
ity peak. This plateau level is believed to correspond to the transition level identified by
Zhugin and Krupnikov (1987) with manganin gage techniques.

67
 Table 9.2:
Hugoniot Data

Profile Hugoniot Shock Hugoniot Hugoniot

Velocity Particle Velocity Stress Specific
Velocity Volume
(km/s) (km/s) (km/s) (GPa) (cm^/kg)

Elastic Hugoniot State

0.38 0.363 6.20 5.96 355.9
Transition Hugoniot State
1.40 1.56 5.50 23.4 273.6

Hugoniot analysis based on impedance match methods was performed using the elastic
and transition profile amplitudes determined above. Hugoniot results are provided in
Table 9.2. A transition stress amplitude of about 23.4 GPa is in good agreement with the
23 GPa value reported by Zhugin and Krupnikov (1987).

2.0 n I I T 1 1 r -1 1 r

Transition Plateau
(1.40 km/s)
1.5 -
(/)

E
SVl
1.0 - \ \
SV2
o
o Elastic Plateau
_J
LxJ
(0.38 km/s)
> 0.5 - QUARTZ WAVE
• • • • • • • • • • t(t mi m PROFILES

0.0 J I I I I I L. J .
0.0 1.0 2.0 3.0 4.0
TIME (/xs)

Figure 9.2 Experimental VISAR profiles for Tests SVl and SV2 on quartz
rock.

68
 Shock Compression Experiments on Quartz Rock Investigating a Critical Stress

There are some concerns, however, about the impedance match Hugoniot analysis in light
of the dramatic change in compression above the critical state.

A slight inflection in the compression profile was also observed at a compressive profile
velocity amplitude of about 1.15 km/s. This inflection was more noticeable in Test SVl
than in Test SV2 (see the Figure 9.3 expanded scale). This difference was believed to be
due to a somewhat larger glue bond thickness (~2|im) between sample and lithium fluo-
ride for Test SV2. Consequentiy, Test SVl is considered to be the higher resolution wave
profile. The source of this inflection is not understood and it was ignored in the Hugoniot
analysis.

Above the 1.40 km/s plateau level, particle velocity rises gradually to a peak value. The
rate of ramping and peak value for Test SV2 are somewhat higher — consistent with the
slightiy faster impact velocity for this test (Table 9.1). Overtake by the trailing release
wave prohibits achievement of the maximum particle velocity compatible with the impact
conditions at the present sample thickness.

Quartz Hugoniot states from the present tests (Table 9.2) are compared with the corre-
sponding stress versus particle velocity Hugoniots for tungsten in Figure 9.4. It is clear
that particle velocities at the impact interface closer to 1.85-1.95 km/s were achieved and
that significant peak attenuation occurred after transmission through the 5-mm sample
thickness. Several assumptions were pursued to estimate the impact compression state.
2.0 1 1 1 1 1 1 1 1 1

^1-5

^
o
E
1.0
SVl
[ SV2 "
o
_J QUARTZ WAVE
UJ ; PROFILES
> 0.5 ^ ^ y (expanded scale)

17 , 1 . 1 . 1
0.0
1.0 1.2 1.4 1.6 1.8 2.0
TIME (/zs)

Figure 9.3 Experimental VISAR profiles for Tests SVl and SV2 on Quartz
rock. Expanded scale view of compressive wave.

69
First, it was assumed that the final state was achieved through a compressive shock that
has been overtaken by the release wave. (Test SVl was considered in this analysis.) The
fastest shock velocity possible which is consistent with overtake by the release wave at
5 mm is 3.1 km/s. This leads to a maximum stress jump of 2.6 GPa and a minimum specif-
ic volume change of approximately 40 cm^/kg.

Altematively, the assumption of a linear ramp to the intersection particle velocity state and
simple centered wave propagation led to a lower final stress state and a specific volume
change closer to 55 en?/kg. Stress profiles measured by Zhugin and Krupnikov (1987) in-
dicated an exponential-like convergence to the peak stress and it seems sensible that the
present volume change estimates of 40-55 cm^/kg represent a lower bound.

9.4 Discussion

Although the present tests provide strong support for the occurrence and stress level of the
compression anomaly reported by Zhugin and Krupnikov (1987), they do not demonstrate
the definitive three-wave structure, indicative of elastic yield and first-order phase frans-
formation, observed in certain compounds. Clearly, higher amplitude experiments of the
present type on the quartz rock would be needed to reveal such wave structure. Unfortu-
nately the required impact velocities could not be achieved on the available impact facili-

30
Transition Shock
"^^25 h State Ramp
=• I I
CL I
I
I
I
O I
I
I
I
I I
I I
20 h I
\
I
I
I \

(/)
\
I Tungsten
\
I
\ \ Hugoniots
(i: 15 - t I
\ I
I— Quartz \ \
\ I
Hugoniot I I
Q 10
o
o \ \
I I
I I
Elastic Vi = 2.30 km/s \
State Vi = 2.19 km/s ^"
2.5 3.0
PARTICLE VELOCITY (km/s)

Figure 9.4 Quartz Hugoniot states and assumed interactions with the
tungsten Hugoniot for the present shock-wave experiments.

70
 Shock Compression Experiments on Quartz Rock Investigating a Critical Stress

2.0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1J
/—\ ZIRCONIUM DIOXIDE i
/ \ WAVE PROFILES

in
[ Transition
. Stress

>- 1.0
o
o
_J
LJ
^ 0.5

, . . . 1 . . . . 1 . . . .
0.0
I I I !

0.0 0.5 1.0 1.5 2.0

TIME (/zs)

Figure 9.5 Experimental VISAR profiles for Zr02 indicating yield and
transition characteristics.

ty. Also, other impact configm-ations, such as reshock or symmetric impact, may be more
appropriate to investigate the present phenomenon.

For comparison, however, we show similar VISAR wave profiles measured on another
oxide, Zr02, in Figure 9.5, which reveal a similar ramp structure above the transition pla-
teau at lower impact velocity and a definitive 3-wave structure at a higher impact ampli-
tude. The existence of similar wave structure in quartz would require further testing.

The occurrence of a critical transformation stress has interesting implications concerning

the molecular and microstructure transformation mechanisms in the a-quartz structure.
The existence of metastable mixed-phase Hugoniot states over a broad stress range in
quartz was inferred from the interpretation of earlier shock data. A theory of microstruc-
tural deformation and thermal heterogeneity was proposed (Grady, et a/., 1975; Grady,
1980) which was consistent with this metastable Hugoniot along with much of the dynam-
ic and post-shock recovery data.

Because of the new wave profile data which indicates a critical transition level for lattice
instability in the quartz structure, both the earlier shock data interpretation and phase
transformation theory should be revisited. In light of this observation, recent static com-
pression work of Hemley et al. (1988) should be noted in which a crystal-to-amorphous

71
transition in quartz at approximately 20 GPa was reported. This transition has been ex-
plained in terms of free-energy/kinetics arguments (Sikka, 1992) in which the a-quartz
crystalline state is driven by elevated pressure to a high free energy state with respect to
both the high density crystalline and amorphous state, but transformation is prohibited by
transition kinetics. When the mechanical stability limits are achieved (spinodal decompo-
sition) the material follows the easiest kinetic path to an amorphous high density state.
Molecular dynamics simulations of the a-quartz transition have reasonably reproduced
the expected transition pressure and volume strain (Tse and Klug, 1992; Chaplot and Sik-
ka, 1992). Clearly this proposed transformation mechanism also warrants consideration
for the shock transition process in quartz, considering the remarkably close critical stress
levels reported in the static work of Hemley et al. (1988), and in the dynamic studies of
Zhugin and Krupnikov (1987) and the present experiments.

Earlier interpretation of shock wave results which led to a broad Hugoniot range (approxi-
mately 20-35 GPa) in which the shock velocity remained constant with increasing particle
velocity also needs to be reexamined. Wave structure above the 20 GPa level is apparently
subtle and requires high-fidelity wave profile diagnostics to resolve. Earlier shock-wave
diagnostics would have easily missed this additional structure. The consequence would
have been to report a constant shock velocity (the transition shock velocity) along with the
peak particle velocity. Shock velocities would again begin to increase with particle veloci-
ty only after the 23 GPa transition shock was over driven. This experimental interpretation
of the constant shock velocity region in quartz seems inherentiy more reasonable than ear-
lier explanations based on equation-of-state or kinematic arguments.

The implication of this interpretation is that the dynamic compression curve of a-quartz
would lie below the previously reported Hugoniot curve, achieving higher densities at cor-
responding pressures. As noted in the earlier analysis, transition volumes in excess of 40-
55 cm /kg are suggested for the impact amplitudes achieved. This significantly exceeds
the volume strain predicted from reported Hugoniot curves for quartz. The transformation
volumes at ambient conditions for the a-quartz-to-coesite structure and the a-quartz-to-
stishovite structure are approximately 33 and 143 cm^/kg, respectively. The present tests
would suggest that the shock transition at 23 GPa is proceeding to the high-density 6-coor-
dination structure of Si02 rather than toward the relatively minor crystallographic change
coesite structure.

72
 Precursor Attenuation in Ceramics

10 Precursor Attenuation in Ceramics

Because of the transient times involved in the compressive shock failure and flow process,
the possibility of rate-dependent strength properties of ceramic must be considered in the
computational evaluation of projectile-target interactions. A manifestation of rate-depen-
dent strength is provided in controlled shock wave experiment through the evolution and
attenuation with propagation distance of the elastic-to-inelastic transition precursor wave
observed in wave-profile measurements. In the present study shock wave data on ceramics
have been surveyed for evidence of rate-dependent precursor decay. The ceramics which
have been investigated are aluminum oxide, aluminum nitride, boron carbide, silicon ni-
tride and titanium diboride. Only boron carbide shows evidence of precursor attenuation
within the range of the data.

10.1 Background

The stress versus deformation response of solids subjected to intense impulsive loads is
expected to depend on the rate of loading. It is common, therefore, to incorporate the rate-
sensitive behavior of materials into the constitutive models developed to model the dy-
namic response.

Specimens of the solid material subjected to controlled plane-wave impact experiments

with high-resolution wave profile diagnostics are believed to provide a sensitive test of the
extent of rate sensitivity exhibited by the material. In particular, the evolution of the shape
of elastic precursor waves as they propagate through increasing distances in a target mate-
rial is expected to provide strong evidence for the rate dependent yield and flow behavior
of solids.

The concept of elastic precursor attenuation and the relation of this effect to the Maxwell-
solid behavior of materials has been explored by Duvall (1978). Experimental observa-
tions of elastic precursor decay in quartz rock were reported by Ahrens and Duvall (1966),
and numerous further examples have been documented in the intervening years. Perhaps
the most in-depth investigation of the underlying physics of precursor decay has been
conducted on lithium fluoride [e.g., Asay et al., 1972]. Rate sensitivity is an integral in-
gredient of the computational model for ceramics developed by Steinberg (1991). Recent-
ly Partom (1994) has noted certain difficulties in reconciling rate-sensitive viscoelasticity
models with experimental wave profiles on ceramics

Studies on the rate sensitivity of ceramics, and on precursor attenuation in particular, are
sparse and results to date have been contradictory. The predominant data appears to have
been generated on aluminum oxide ceramic. One study has reported elastic precursor at-
tenuation in aluminum oxide over sample thicknesses of a few millimeters to 25 millime-
ters [Rosenberg et al., 1988]. In contrast, measurements reported by Cagnoux and Longy
(1987) for aluminum oxide did not indicate precursor decay.

In the present investigation the precursor waves of experimentally measured wave profiles
on selected ceramics using velocity interferometry methods were examined for precursor
decay trends. In all cases relatively pure and nearly full density ceramics were tested. In

73
general very little evidence for precursor attenuation was found for the ceramics investi-
gated. Boron carbide may be the one exception, indicating some precursor attenuation
over the thickness range examined.
10.2 Precursor Wave Properties for Ceramics

To investigate the evolution of elastic precursor waves with propagation distance we ex-
amine shock wave data on selected ceramics for which the thickness of the test samples
have been varied over the range of about 2.5 mm to 15 mm. On all experiments discs of
the test ceramic were backed by lithiumfluoridewindow material and full shock compres-
sion and release waves were measured at a central point at the interface between sample
and window material. Details of the experimental method are described in Grady (1992a,
1992b) and discussed in Section 2 of the present report. In the present study only the pre-
cursor wave of the shock profiles have been examined. Ceramics for which samples of dif-
ferent thickness were investigated include aluminum oxide, aluminum nitride, silicon
nitride, titanium diboride and boron carbide. Relevant experimental parameters for the
data examined are provided in Table 10.1.
Table 10.1:
Properties for Shock-Wave Experiments on Ceramics
Impact Sample*
Test Material Density Q Velocity Thickness
Number (kg/m3) (km/s) (km/s)
(km/s) (mm)
AO-03 AI2O3 3890 10.56 6.24 1.55 2.48
AO-01 AI2O3 3890 10.56 6.24 1.57 4.76
CE-58 AI2O3 3890 10.56 6.24 1.57 10.01
CE-43 AIN 3250 10.73 6.32 2.21 2.51
CE-52 AIN 3250 10.73 6.32 2.22 4.18
CE-37 AIN 3250 10.73 6.32 2.24 9.57
CE-101 B4C 2506 14.07 8.87 3.98 2.99
CE-26 B4C 2506 14.07 8.87 2.06 9.68
CE-104 Si3N4 3130 10.30 5.80 1.05 5.01
SN-02 Si3N4 3130 10.30 5.80 1.06 10.02
SN-01 Si3N4 3130 10.30 5.80 1.05 15.01
CE-10 TiB2 4509 10.79 7.43 1.50 5.01
CE-9 TiB2 4509 10.79 7.43 1.50 10.10

* All tests were ceramic-on-ceramic impacts.

74
 Precursor Attenuation in Ceramics

We first display the measured precursor profiles. (It is noted that there can be an illusory
indication of precursor attenuation because of the different wave velocities associated
with the precursor wave and following shock wave, and consequently the increasing dif-
ference in wave separation with propagation distance.) Precursor waves are then plotted in
a manner which scales with propagation distance and examined for self similarity. This
method is believed to be the best technique for revealing precursor decay and rate-depen-
dent behavior in the yield process under dynamic uniaxial strain loading. It is concluded
that with the exception of boron carbide there is no evidence for elastic precursor decay or
rate-dependent yield in the ceramics investigated.

aluminum oxide
Precursor wave profiles for aluminum oxide measured over propagation thicknesses of ap-
proximately 2.5, 5.0 and 10.0 mm are shown in Figure 10.1. In (a) measured profiles are
shown with the precursor arrival aligned to time zero. In (b) the same profiles are scaled to
10 mm sample thickness by multiplying every time datum by the ratio of 10 mm to sam-
ple thickness. The aluminum oxide on which this data was obtained was the Coors Porce-
lain Company AD-995 (CAP-3) product. The density and elastic properties are provided
in Table 10.1.

The precursor profiles for this aluminum oxide are typical of most of the monolithic ce-
ramics tested previously [Grady, 1992a; Grady, 1992b]. There is an initial sharp rise to a
velocity level of a few hundred meters per second which is usually not resolved within the
resolution of the VISAR system (-2-5 ns). This initial rise is followed by a more gradual
ramping rise which merges into a steeper second wave. The second wave is readily re-
solved with the VISAR diagnostics provided the final shock amplitude is not excessively
high. (The second wave rise time steepens with shock amplitude and ultimately also ex-
ceeds the resolution limits of the VISAR system.)

The initial sharp rise and more gradual ramping region is identified as the precursor wave
or elastic precursor, whereas the second steeper wave is identified as the deformation
shock, although a clear transition from one to the other is not always observed. Figure 2.4
and the associated discussion should be considered here.

From examination of the measured wave profiles in (a) of Figure 10.1 it is difficult to con-
clude with certainty whether or not the amplitude of the precursor wave is decreasing with
propagation distance. The scaled plot of the data shown in (b), however, clearly shows self
similarity in the amplitude of the initial rise and the shape of the ramping region. This ob-
servation suggest no precursor attenuation or rate independent yield and initial post-yield
behavior in the present ceramic.

A closer examination of the profiles in Figure 10.1(b) shows the ramp region of the
10 mm sample about 5% lower than the 2.5 mm and 5.0 mm samples. Our experience
over a broader range of tests is that this is within the sample to sample scatter associated
with these wave profiles measurements. It should be noted, however, that the 10 mm sam-
ple was prepared from an earlier supply of the AD-995 (CAP-3) aluminum oxide than that
of the 2.5 and 5.0 mm samples.

75
 1.2 ! ' ' ' 1 ' 1 '
ALUMINUM OXIDE

i ^ ''"
• I t /
'• 1 '
'• 1 1 / I
'• 1 1
w 0.8 L i / ' / H
/
E r i / ' /
•
•
/ '
1 /
L \ 1 1 / \
1 1
1 '

O
O 0.4
_i
LJJ solid — 2.4B mm
> •T sdash — 4.76 mm J

.1 ,
mdash — 10.0 mm 1

1 , 1 , 1
0.0
-0.1 0.0 0.1 0.2 0.3 0.4
(a) TIME (lis)
1.2 ' ' 1 ' 1 ' 1 •

solid — 2.48 mm
sdash — 4.76 mm J
mdash—10.0 mm ^ - ^ri-M-H
•' ' ' ' / ^ " ^
' / /
' ' /
-
0.8 — 1''
' // H
CO
''' /
E ''/
r
y
^y^ 1

H
^^^
O 0.4 - ^^^*V
^^^^•^ J
h
LU / ^ ALUMINUM OXIDE
> h J (Self-similar scaling
} to 10 mm sample thickness)
t,
0.0 . 0.0
I 0.1
1 . 1

0.2
, 1

0.3
,
i

0.4
-0.1
(b) TIME (lis) ^

Figure 10.1 Velocity profiles for aluminum oxide, (a) Measured wave profiles,
(b) Self-similar representation of wave profiles.

76
 Precursor Attenuation in Ceramics

A further interesting observation in the aluminum oxide profiles is the lack of self similar-
ity of the deformation shock waves. That is in the self-similar plot the deformation shock
for the 10 mm test arrives earlier in time than the 5 mm test which in turn proceeds the
2.5 mm test. This implies that shock velocities calculated from the 10 mm test will be
higher than shock velocities from the 5 mm or 2.5 mm test. The maximum difference in
the present tests is about 3%. This effect is not understood, but one possible explanation is
an increased precursor amplitude (and hence precursor attenuation) for very early propa-
gation distances (< 2.5 mm) which would lead to decreased shock velocities within this
region.

aluminum nitride
Similar profiles of the precursor wave for aluminum nitride are provided in Figure 10.2(a)
and (b). These are the same wave profiles discussed in Section 7 and by Kipp and Grady
(1994) where emphasis was on the -20 GPa shock-induced phase transition. This transi-
tion initiates at the top of the second wave shown in Figure 10.2. The aluminum nitride
ceramic tested was provided by DOW Chemical Company. The material was near theoret-
ical density although sample-to-sample variations in density and sound speed of about 2%
suggested a marked lack of homogeneity.

Like aluminum oxide, the precursor wave is characterized by a sharp rise to an initial
yield level followed by a slowly rising ramp. The slope of the ramp is more gradual than
that of aluminum oxide and the transition to the second shock is more definitive. Although
there is some scatter in the sample-to-sample profile structure, the self similarity in the
scaled profiles in Figure 10.2(b) clearly suggests negligible precursor decay and negligi-
ble rate dependence in the initial yield and post-yield behavior of aluminum nitride.

The nearly self similar behavior of the second wave in the scaled plot is curious. If it is
presumed that this is a steady shock segment over a region of positive stress-strain curva-
ture then self-similar wave profile structure is not expected. Considering scatter in the data
this may be fortuitous, however, and further testing would be needed to confirm the obser-
vation.

silicon nitride
Precursor profiles for silicon nitride measured at sample thicknesses of 5, 10 and 15 mm
are shown in Figure 10.3 (a) and (b). This material is the near theoretical density SN-220
ceramic provided by Kyocera Industrial Ceramics Corporation. Maximum shock ampli-
tude in these tests was somewhat lower than those achieved in the aluminum oxide and
aluminum nitride experiments. Consequently, the second shock wave observed in
Figure 10.3 is just beginning to develop and has the characteristic large rise time of rela-
tively low amplitude shock waves.

The self similarity of the scaled wave profiles in Figure 10.3(b) is quite remarkable and
again provides absolutely no evidence for precursor decay with propagation distance. As
in the case of the aluminum nitride, the self similar behavior of the second shock for the
silicon nitride, even though just beginning to develop, indicates that it is still evolving
over the propagation distance and has not achieved a steady-wave rise time. This assumes,

77
 1.2 ' 1 ' 1 ' 1 >-' 1 '
ALUMINUM NITRDE '-'' ^---—
I / ^y^^ -

/ f «\ —
w' 0.8 - /
\ \ 1
E 1
1 1 1
1 1 1
r
1 1

8 0.4 ^ -
-J 1 ' ' "
LU f solid — 2.51 mm
> sdash — 4.18 mm
mdash — 9.57 mm

0.0 , ,.J 1 t i l l

-0.1 0.0 0.1 0.2 0.3 0.4

(a) TIME (|xs)
1.2
solid — 251 mm
sdash — 4.18 nfvn
mdash — 9.57 mm

0.8 -
E
>

O 0.4 -

LU ALUMINUM NITRIDE
> (Self—similar scallna
to 10 mm sample thickness)

0.0
-0.1 0.0 0.1 0.2 0.3 0.4

(b)
TIME (lis)

Figure 10.2 Velocity profiles for aluminum nitride, (a) Measured wave profiles,
(b) Self-similar representation of wave profiles.

78
 Precursor Attenuation in Ceramics

yj.o 1 1 . 1 1 1 1 1 1 , 1

. SILICON NITRIDE .

0.6 y-^^^^^^^y^'^
E
0.4 - -

O r
O
0.2 - ' -
LU
solid — 10.02 mm
sdash— 15.01 mm
• mdash — 5.01 mm •

n n o...^ 1 1 1 1 1 1 1 1 r

-0.2 0.0 0.2 0.4 0.6 0.8 1.0

(a) TIME (^s)
0.8
solid— 10.02 mm
sdash— 15.01 mm
mdash — 5.01 mm
D.6 -
CO

,E
3.4 -
H
O
o
_l D.2 SILICON NITRIDE
LU (Self-similar scalir
> to 10 mm sample thickness)

0.0
-0.1 0.1 0.3 0.5 0.7

(b) TIME (jis)'

Figure 10.3 Velocity profiles for silicon nitride, (a) Measured wave profiles,
(b) Self-similar representation of wave profiles.

79
of course, that this segment of the wave is evolving toward a steady-wave configuration.
More complex constitutive features are quite possible.

titanium diboride
Only two shock-compression experiments on titanium diboride illustrating wave evolu-
tion features as a function of propagation distance have been performed. They are none-
theless interesting and are shown in Figure 10.4 (a) and (b). The material is a near full
density ceramic prepared by CERCOM Inc.

Titanium diboride ceramic is unique in exhibiting a more complex two wave structure as-
sociated with the dynamic yield process although the possibility of a polymorphic phase
transformation within this range contributing to the unusual structure has not yet been
ruled out [Grady and Wise, 1993].

The scaled profiles illustrated in Figure 10.4(b) indicate some deviation from self similari-
ty although the sample-to-sample scatter in profile structure is uncertain. Both the elastic
wave (up to initial yield) and the second shock wave have comparable rise times in the
5 mm and 10 mm samples, indicating quasi-steady behavior. The third wave appears to
still be evolving with propagation distance. Neither the first or second yield, as indicated
by the cusps in Figure 10.4, show signs of amplitude decay with distance.

boron carbide
Precursor waves observed in the previous ceramics have features in common. A rapidly
rising wavefi^ontrounds over to a more gentie, but still rising, ramp region which merges
in turn into the following deformation shock wave. Precursor waves for boron carbide are
unique among the ceramics tested in exhibiting markedly different profile structure. Rep-
resentative precursor profiles are shown in Figure 10.5. A discontinuous wave front (with-
in the several ns resolution of the VISAR diagnostics) is followed by a somewhat ragged
but nonetheless falling ramp region before arrival of the deformation shock wave which
transports material to the final shock state.

Precursor waves in boron carbide have been compared in detail with corresponding waves
in silicon carbide in Section 6 and also in Grady (1994). Briefly it was suggested that the
velocity (and stress) relaxation within the precursor ramp is indicative of a material exhib-
iting rate-dependent loss of strength in the shock-induced yield process. Further, irregular-
ities in the measured profile may be associated with a heterogeneous yield process on a
spacial scale which caused superimposed chaotic motion at the interferometer spot where
velocity is measured.

Within the scope of the present examination of precursor wave attenuation, only one ex-
periment (CE-101) on a reduced thickness sample of boron carbide has been performed.
These tentative results, however, do indicate a measurable precursor decay with propaga-
tion distance. (A 20 GPa Hugoniot elastic limit at a 3 mm sample thickness attenuates to
about 16.4 GPa at approximately 10 mm.) Thus boron carbide is also unique among the
ceramics examined here in being the only material which may exhibit measurable elastic
precursor attenuation within the range of the present data and does not exhibit self-similar
scaling of the precursor wave.

80
 Precursor Attenuation in Ceramics

1.2 1 ' 1 •
TITANIUM DIBORIDE

/ z

. ,
/ /
lo 0.8 - 1 1
1 /
E
,
/
• /
/ , / "
O f
0.4 - I 1
1

1 1
/ 1
LU / /' solid — 5.01 mm
> / y sdash—10.10 mm
'

1 1 1 1 1
0.0
-0.1 0.0 0.1 0.2 0.3
(a) TIME (i^s)
1.2
solid — 5.01 mm
sdash — 10.10 mm

'w 0.8
E

_l
LU TITANIUM DIBORIDE
> (Self-similar scaling
to 10 mm sample thickness)

0.0
-0.1 0.0 0.1 0.2 0.3
(b) TIME (fis)

Figure 10.4 Velocity profiles for titanium diboride. (a) Measured wave profiles,
(b) Self-similar representation of wave profiles.

81
10.3 Hugoniot Elastic Limit Data
We have attempted to provide a representation of the Hugoniot elastic limit (HEL) data
implied by the present VISAR wave-profile measurements on ceramic specimens of vari-
ous sample thicknesses. The results, which are tabulated in Table 10.2 and plotted in
Figure 10.6, require further discussion.
With the high amplitude and temporal resolution provided by the velocity interferometer
profile measurements, the selection of a precursor amplitude with which to correlated with
a specific HEL amplitude is no longer obvious. Examination of the data in the previous
plots for aluminum oxide, or silicon nitride, for example, reveals a gradual rounding and
ramping which makes it difficult to systematically associate a specific amplitude with the
HEL amplitude. Recent computational simulations of profiles measured on tungsten car-
bide, which reveals similar precursor structure, demonstrated that an initial 4 GPa break-
over stress can achieve 8 to 9 GPa through the precursor ramp region before merging into
the deformation shock wave (Section 8).

In this study we have used the following tactic to systematically establish an HEL precur-
sor amplitude for each profile measurement. A characteristic time x^ associated with the

1.6 -I r -1 > 1 1 r -I r
BORON CARBIDE

H 0.8 h --/vi--
O
O 2.99 mm
LU
9.68
> 0.4

0.0 J LJ I L. _l I L.

0.0 0.5 1.0 1.5

TIME (n's) "

Figure 10.5 Precursor velocity profiles for boron carbide at two sample
thicknesses indicating precursor attenuation.

82
 Precursor Attenuation in Ceramics

separation of the arrival of the precursor wave and the deformation shock wave at the re-
cording interface is established from the profile. Although some arbitrariness is intro-
duced by the finite rise of the respective waves (primarily the deformation shock),
reasonable consistency is relatively easy to achieve. The velocity amplitude, u , at the
temporal midpoint (x^/2) is then accepted as the HEL amplitude. This method is expected
to provide comparable HEL amplitudes with lower resolution wave-profile measurement
techniques (piezoresistive stress gages, magneto-inductive velocity gages) where details
of the precursor structure are frequently not resolved. The HEL stress, reported for the
present data in Table 10.2, was then calculated from the u^ data assuming longitudinal
acoustical impedance properties for the ceramic and hydrodynamic acoustical properties
for the lithium fluoride, respectively. To display the dependence of the HEL stress on sam-
ple thickness for the present data we have chosen to establish an effective strain rate corre-
sponding to the dynamic strength measurement. Although loading to the HEL stress is not
a constant strain rate process in a shock wave experiment, the time x^ separating elastic
and deformation shocks provides a time scale over which the precursor amplitude is es-
tablished. Consequently, the ratio of the HEL strain, Sf^^j = ^^hei^Po^L' ^° ^ ^ appropriate
characteristic time x^, provides an effective strain rate, t^ff = E^^^/X^, which is also re-
ported in Table 10.2.

Table 10.2:
Hugoniot Elastic Limit Data for Ceramics

Test Material Strain Rate HEL*

Number (ns) (m/s) (10-Vs) (GPa)

AO-03 AI2O3 68 405 3.8 11.1

AO-01 AI2O3 122 395 2.0 10.8
CE-58 AI2O3 255 388 0.9 10.6
CE-43 AIN 55 402 4.7 9.75
CE-52 AIN 90 398 2.9 9.65
CE-37 AIN 210 395 1.2 9.58
CE-101 B4C 80 820 5.1 20.0
CE-26 B4C 400 670 0.8 16.4
CE-104 Si3N4 175 400 1.6 9.16
SN-02 Si3N4 345 408 0.8 9.34
SN-01 Si3N4 490 402 0.6 9.21
CE-10 TiB2 38 156 2.4 4.86
CE-9 TiB2 72 169 1.4 5.26

* Hugoniot elastic limit— see text.

83
 STRAIN RATE (10^/s)

Figure 10.6 Hugoniot elastic limit versus strain rate date inferred from
VISAR wave profile data for selected ceramics.

The corresponding HEL stress and strain rate data are plotted in Figure 10.6. Consistent
with the earlier examination of the data in the previous section, with the exception of bo-
ron carbide, there is littie evidence to indicate HEL precursor decay over the approximate-
ly half a decade of strain rate inferred from the profile measurements.

10.4 Discussion

The present survey of compression profiles measured on selected ceramics indicates, for
the most part, very little evidence to support the existence of elastic precursor wave atten-
uation in ceramics. Precursor attenuation is expected to be a natural consequence of the
rate sensitive dynamic strength characteristics of solids.

The present investigation, however, has only examined the issues of precursor attenuation
within the limited sample thickness range of a few millimeters to a few tens of millime-
ters. Corresponding effective strain rates are about 5 x lO'^/s to 5 x lO'^/s. Rate sensitivities
outside of this range, or rate sensitivities within this range of as much as 5% per decade,
cannot be commented on.

84
 Precursor Attenuation in Ceramics

This notable lack of precursor attenuation and rate sensitivity in ceramics is surprising.
Compressive failure in the shock process is believed to be a fractare-controlled mecha-
nism and the nucleation and growth of fracture is known to be a time-dependent process.
Such time dependence would be expected to lead to rate-dependent strength and precursor
decay in ceramics. Fracture failure in brittie materials under other modes of deformation
(spall [Grady, 1980], compression bar [Lankford, 1981]) have shown strength sensitivities
on strain rate approaching a one-third power behavior. The model of compressive fracture
leading to Equation 30 in Section 5 predicts a one-third power dependence of the Hugoni-
ot elastic limit on strain rate. The present insensitivity to strain rate raises the question of
whether failure within the front of a compressive shock wave is indeed a fracture con-
trolled process.

The present results are consistent with the conclusions of Cagnoux and Longy (1987).
From their observations of negligible precursor decay in alumina, they concluded that
grain plasticity was a more likely mechanism for yield in the compressive shock front
than was brittle fracture.

85
11 Conclusions
Substantial experimental research, both here and in other institutions, on the dynamic re-
sponse of ceramics has been performed over the past several years. Results of these stud-
ies are providing a sound experimental base for the development and validation of
computational models focused on the dynamic behavior of brittie materials.

Over the time frame of those experimental research efforts, however, there does not ap-
peared to have been a commensurate improvement in the communities understanding of
the mechanisms underlying the transient failure and deformation of brittie solids. Very re-
centiy benefits from the several years of research effort are beginning to be realized. Con-
cepts such as failure waves, high-confinement dilatancy, brittie-ductile (deformation
stability) transitions and granular flow, although not new, are emerging as very real fea-
tures to address in the constitutive modeling of ceramics. These issues can be expected to
drive the direction of experimental research and modeling over the next several years.

86
 References

12 References
Section 1
Grady, D. E., Dynamic Failure in Brittle Solids, in Proceedings Europe-U.S. Workshop on
Fracture and Damage in Quasibrittie Structures: Experiment, Modeling and Computer
Analysis, Prague, Czech Republic, September 21-23 (1994a).

Grady, D. E.. Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide, in
Proceedings of EuroDYMAT 94 - International Conference of Mechanical and Physical
Behavior of Materials under Dynamic Loading, Oxford England, September 26-30,
(1994b).

Grady, D. E. and J. L. Wise, Dynamic Properties of Ceramic Materials, Sandia National

Laboratories Report, SAND93-0610, September (1993).

Grady, D. E. and Yu. N. Zhugin, Critical Transition Stress in the Shock Compression of
Si02 (abstract only). Bull. Am. Phys. Soc, 39, 410-411 (1994)

Kipp, M. E. and D. E. Grady, Shock Phase Transformation and Release Properties of Alu-
minum Nitride, in Proceedings of EuroDYMAT 94 - International Conference of Mechan-
ical and Physical Behavior of Materials under Dynamic Loading, Oxford England,
September 26-30 (1994).

Zhugin, Yu. N. and K. K. Krupnikov,, Structural Peculiarities of Plastic Wave Front in

Quartz in the Mixed Phase Region, in, Non-Linear Acoustics Problems, in Transactions of
the XI International (lUPAP-IUTAM) Symposium on Non-Linear Acoustics, Part II, No-
vosibirsk, 196-200 (1987).

Section 2
Barker, L. M. and R. E. Hollenbach, Laser Inteiferometer for Measuring High Velocities
of any Reflecting Surface, J. Appl. Phys., 43,4669-4680 (1992).
Grady, D. E, Shock Deformation of Brittle Solids, J. Geophys. Res., 85, 913-924 (1980).

Grady, D. E., Local Inertial Effects in Dynamic Fragmentation, J. Appl. Phys., 53, 322-
325(1982).

Grady, D. E., The Spall Strength of Condensed Matter, J. Mech. Phys. Solids, 36, 353-383
(1988).

Grady, D. E., Shock-Wave Properties of High-strength Ceramics, in Shock Compression

of Condensed Matter - 1991, S. C. Schmidt, R. D. Dick, J. W. Forbes and D. G. Tasker,
eds. Elsevier Science Publishing, 455-458 (1992a).

Grady, D. E., Shock-Compression Properties of Ceramics, in Recent Trends in High-Pres-

sure Research, A. K. Singh, ed., Oxford and IBH Publishing, 641-650 (1992b).

87
Grady, D. E., Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide, in
Proceedings of EuroDYMAT 94 - International Conference of Mechanical and Physical
Behavior of Materials under Dynamic Loading, Oxford England, September 26-30 (1994).

Sato, T. and S. Akimoto, Hydrostatic Compression of Four Corundum-Type Compounds,

J. Appl. Phys. 50, 5285-5291 (1979).

Wise, J. L. and L. C. Chhabildas, Laser Interferometer Measurements of Refractive Index

in Shock-Compressed Materials, in Shock Waves in Condensed Matter - 1985, Y M.
Gupta, ed., Plenum Press, 441-445 (1985).

Section 3
Brar, N. S., S. J. Bless and Z. Rosenberg, Impact-Induced Failure Waves in Glass Bars
and Plates, Appl. Phys. Lett., 59, 3396-3398, (1991).

Kanel, G. I., S. V. Rasorenov and V. E. Fortov, The Failure Waves and Spallations in
Homogeneous Brittle Materials, in Shock Compression of Condensed Matter - 1991,
S. C. Schmidt, R. D. Dick, J. W Forbes and D. G. Tasker, eds., Elsevier Science Pub-
lishing, 451-454, 1992.

Raiser, G. V, J. L. Wise, R. J. Clifton, D. E. Grady and D. E. Cox. Plate Impact Response

of Ceramics and Glasses, J. Appl. Phys., 75, 3862-3872 (1994).

Raiser, G. V. and R. J. Clifton. Failure Waves in Uniaxial Compression of an aluminosili-

cate Glass, in High Pressure Science and Technology — 1993, S. C. Schmidt, J. W Shan-
er, G. A. Samara, M. Ross, eds., AIP Press, 1039-1042 (1994).

Section 4
Grady, D. E., The Mechanics of Fracture under High-Rate Stress Loading, in Mechanics
of Geomaterials, Z. Bazant, ed., John Wiley and Sons, 129-156 (1985).

Grady, D. E., The Spall Strength of Condensed Matter, J. Mech. Phys. Solids, 36, 353-383
(1988).

Grady, D. E. and M. E. Kipp, Fragmentation of Solids under Dynamic Loading, in Struc-

tural Failure, T. Wurzbicki, ed., John Wiley and Sons, 1-39 (1989).

Grady, D. E. and M. E. Kipp, Dynamic Fracture and Fragmentation, in High-Pressure

Shock Compression of Solids, J. R. Asay and M. Shahinpoor, eds.. Springer-Verlag, 265-
319(1993).

Izotov, A. D. and V. B. Lazarev, Theoretical Strength of Ceramic Materials, Inorg. Mater.,

21,612-617(1985).

Kipp, M. E., D. E. Grady and E. P. Chen, Strain-Rate Dependent Fracture Initiation, Int.
J. Fracture, 16,471-478(1980).

88
 References

Kozhusko, A. A., 1.1. Rykova, A. D. Izotov and V. B. Lazarev, Strength and Fracture of
Ceramic Materials, in High-Velocity Deformation. Inorg. Mater., 23, 1820-1824 (1987).

Lawn, B. R. and T. R. Wilshaw, Fracture of Brittle Solids, Cambridge University Press,

(1975).

Section 5
Curran, D. R., L. Seaman and D. A. Shockey, Dynamic Failure of Solids, Physics Reports,
147,253-388(1987).

Curran, D. R., L. Seaman, T. Cooper and D. A. Shockey, Micromechanical Model for

Comminution and Granular Flow of Brittle Materials under High Strain Rate Application
to Penetration of Ceramic Targets, Int. J. Impact Eng., 13, 53-84 (1993).

Griffith., A. A., Theory of Rupture, Proceedings First International Congress Applied Me-
chanics, Delft, 55-63 (1924).

Janach, W, Failure of Granite under Compression, Int. J. Rock Mech. Min. Sci., 14, 209-
215(1977).

Johnson, G. R. and T. J. Holmquist, Improved Computational Constitutive Model for Brit-

tle Materials, in High Pressure Science and Technology — 1993, S. C. Schmidt, J. W.
Shaner, G. A. Samara, M. Ross, eds., AIP Press, 981-984 (1994).

Murrel, S. A. F , A Criterion for Brittle Fracture of Rocks and Concrete under Triaxial
Stress and the Effect of Pore Pressure on the Criterion, in Rock Mechanics, C. Fairhurst,
ed., Oxford, Pergamon, 563-577 (1963).

Rosenberg, Z., On the Relation Between the Hugoniot Elastic Limit and the Yield Strength
of Brittle Materials, J. Appl. Phys., 74, 752-753 (1993).

Steinberg, D. J., Computer Studies of the Dynamic Strength of Ceramics, in 3rd Interna-
tional Conference on Mechanical and Physical Behavior of Materials under Dynamic
Loading, Ies Editions des Physique, 837-844 (1991).

Section 6
Atroshenko S. A., Vasil'kov V. B., Meshcherykov Yu. I., Savenkov G. G. and Chernysh-
enko A. I., High-Frequency Vibration of Grains Excited by Pulsed Loading, Sov. Phys.
Tech. Phys., 35, 336-341 (1990).

Barker, L. M. and R. E. Hollenbach, Laser Interferometer for Measuring High Velocities

of any Reflecting Swface, J. Appl. Phys., 43,4669-4680 (1972).

Grady, D. E., Shock Deformation of Brittle Solids, J. Geophys. Res., 85, 913-924 (1980).

89
Grady, D. E., Shock-Wave Properties of High-strength Ceramics, in Shock Compression
of Condensed Matter - 1991, S. C. Schmidt, R. D. Dick, J. W. Forbes and D. G. Tasker,
eds. Elsevier Science Publishing, 455-458 (1992a).

Grady, D. E., Shock-Compression Properties of Ceramics, in Recent Trends in High-Pres-

sure Research, A. K. Singh, ed., Oxford and IBH Publishing, 641-650 (1992b).

Grady, D. E., and M. E. Kipp, Shock Compression Properties of Silicon Carbide, Sandia
National Laboratories Report, SAND92-1832, July (1993).

Grady, D. E., Impact Strength and Indentation Hardness of High-Strength Ceramics, in

High Pressure Science and Technology — 1993, S. C. Schmidt, J. W. Shaner, G. A. Sama-
ra, M. Ross, eds., AIP Press, 741-744 (1994).

Emin D., Icosahedral Boron-Rich Solids, Physics Today, 1,1-8 (1987).

Kipp, M. E. and D. E. Grady, Shock Compression and Release in High-Strength Ceramics,

Sandia National Laboratories Report, SAND89-1461, July (1989).

Kipp, M. E. and D. E. Grady, Elastic Wave Dispersion in High-Strength Ceramics, in

Shock Compression of Condensed Matter - 1991, S. C. Schmidt, R. D. Dick, J. W. Forbes
and D. G. Tasker, eds., Elsevier Science Publishing, 459-462 (1992).

Manghnani, M. H, Elastic Properties and Compressibility of Advanced Ceramic Compos-

ites, J. Appl. Phys., to be submitted (1994).

Meshcheryakov, Yu. I., S. A. Atroshenko, V. B. Vasil'kov and A. I. Chernyshenko, Crite-

ria of Transition from Translational to Rotational Motion of Media under Shock Loading,
in Shock Compression of Condensed Matter - 1991, S. C. Schmidt, R. D. Dick, J. W.
Forbes and D. G. Tasker, eds., Elsevier Science Publishing, 407-410 (1992).

Section 7
Brar, N. S., S. J. Bless, and Z. Rosenberg, Response of Shock-Loaded AIN Ceramics De-
termined with In-Material Manganin Gauges, in Shock-Wave and High-Strain-Rate Phe-
nomena in Materials, M. A. Meyers, et al. eds.. Marcel Dekker, Inc., 1023-1030 (1992).

Chhabildas, L. C , J. R. Asay, and L. M. Barker, Shear Strength of Tungsten Under Shock-

and Quasi-Isentropic Loading to 250 GPa, Sandia National Laboratories Report,
SAND88-0306, April (1988).

Chhabildas, L. C. and J. R. Asay, Dynamic Yield Strength and Spall Strength Measure-
ments Under Quasi-Isentropic Loading, Shock-Wave and High-Strain-Rate Phenomena in
Materials, M. A. Meyers, L. E. Murr and K. P. Staudhammer, eds.. Marcel Dekker, Inc.,
947-955 (1992).

Dandekar, D. P., A. Abbate and J. Frankel, Equation of aluminum Nitride and its Shock
Response, J. Appl. Phys., 76, 4077-4085 (1994).

90
 References

Gerlich, D., S. L. Dole, and G. A. Slack, Elastic Properties of Aluminum Nitride, J. Phys.
Chem. Solids, 47,437-441 (1986).

Grady, D. E., Shock-Compression Properties of Ceramics, in Recent Trends in High-Pres-

sure Research, A. K. Singh, ed., Oxford and IBH Publishing, 641-650 (1992b).

Jeanloz, R., Shock Wave Equation of State and Finite Strain Theory, Journal of Geophysi-
cal Research, 94, 5873-5886 (1989).

Kipp, M. E. and R. J. Lawrence, WONDY V - A One-Dimensional Finite-Difference Wave

Propagation Code, Sandia National Laboratories Report, SAND81-0930, June (1982).

Kipp, M. E. and D. E. Grady, Shock Compression and Release in High-Strength Ceram-

ics, in Shock Compression of Condensed Matter -1989, S. C. Schmidt, J. N. Johnson and
L. W Davison, eds., North-HoUand, Amsterdam, 377-380 (1990).

Kondo, K., A. Sawaoka, K. Sato and M. Ando, Shock Compression and Phase Transfor-
mation of AIN and BP, AIP Conference Proceedings No. 78, Shock Waves in Condensed
Matter - 1981 (Menlo Park), W. J. Nellis, L. Seaman, and R. A. Graham, eds., AIP Press,
303-305(1982).

Nakamura, A. and T. Mashimo, Shock-Induced Phase Transition of AIN, in High Pressure

Science and Technology — 1993, S. C. Schmidt, J. W. Shaner, G. A. Samara, M. Ross,
eds., AIP Press, 981-984 (1994).

Van Camp, P. E., V. E. Van Doren and J. T. Devreese J. T, Theoretical High Pressure
Studies of the Binary Nitrides, in Recent Trends in High-Pressure Research, A. K. Singh,
ed., Oxford and IBH Publishing, 322-327, (1992).

Vollstadt, H., E. Ito, M. Akaishi, S. Akimoto, and O. Fukunaga, High Pressure Synthesis
ofRocksalt Type of AIN, Proc. Japan Acad., 66 Ser. B, 7-9 (1990).

Wise, J. L. and L. C. Chhabildas, Laser Inteiferometer Measurements of Refractive Index

in Shock-Compressed Materials, in Shock Waves in Condensed Matter - 1985, Y. M.
Gupta, ed.. Plenum Press, 441-445 (1985).

Section 8
Grady, D. E. and M. E. Kipp, Dynamic Fracture and Fragmentation, in High-Pressure
Shock Compression of Solids, J. R. Asay and M. Shahinpoor, eds.. Springer Verlag, 285-
322(1993).

Gust, W. H., Hugoniot Elastic Limits and Compression Parameters for Brittle Materials,
in High Pressure Science and Technology, Vol. 2, B. Vodar and Ph. Marteau, eds., Perga-
mon Press, 1009-1016(1980).

Holmquist, T. J., Unpublished. (1994).

91
Kipp, M. E. and R. J. Lawrence, WONDY V - A One-Dimensional Finite-Difference Wave
Propagation Code, Sandia National Laboratories Report, SAND81-0930, July (1982).

Marsh, S. P., LASL Shock Hugoniot Data, S. P. Marsh, Ed., University of California Press,
(1980).

McQueen, R. G., The Equation of State of Mixtures, Alloys, and Compounds, in Seismic
Coupling, G. Simmons, Ed., Proceedings of a Meeting sponsored by the Advanced Re-
search Projects Agency at Stanford Research Institute, Menlo Park, California, January
15-16, (1968).

McQueen, R. G., S. P Marsh, J. W Taylor, J. N. Fritz, W. J. Carter, The Equation of State

of Solids from Shock Wave Studies, in High-Velocity Impact Phenomena, R. Kinslow, Ed.,
Academic Press, 293-417, 521-568 (1970).

Morris, C. E., LASL Shock Wave Profile Data, C. E. Morris, Ed., University of California
Press, (1980).

Steinberg, D. J., Equation of State and Strength Properties of Selected Materials,

Lawrence Livermore National Laboratories Report, UCRL-MA-106439, (1991).

Section 9
Barker, L. M. and R. E. Hollenbach, Laser Interferometer for Measuring High Velocities
of any Reflecting Surface, J. Appl. Phys., 43,4669-4680 (1992).

Chaplot, S. L. and S. K. Sikka, Molecular Dynamics Simulations of Pressure Induced

Crystal to Amorphous Transitions in Some Corner Linked Polyhedral Compounds, in Re-
cent Trends in High Pressure Physics, A. K. Singh, ed., Oxford and IBH publishing, 259-
261 (1992).

Grady, D. E., W. J. Murri and G. R. Fowles, Quartz to Stishovite: Wave Propagation in the
Mixed Phase Region, J. Geophys. Res., 79, 332-338 (1974).

Grady, D. E., W.J. Murri and P. S. DeCarli, Hugoniot Sound Velocities and Phase Trans-
fofmation in Two Silicates, J. Geophys. Res., 80,4857-4861 (1975).

Grady, D. E., Shock Deformation of Brittle Solids, J. Geophys. Res., 85, 913-924 (1980).

Grady, D. E. and Yu. N. Zhugin, Critical Transition Stress in the Shock Compression of
SiOj (abstract only). Bull. Am. Phys. Soc, 39, 410-411 (1994).

Hemley, R. J., A. P. Jephcoat, H. K. Mao, L. C. Ming, M. H. Manghnani, Pressure-In-

duced Amorphization of Crystalline Silica, Nature, 334, 52-54 (1988).

Sikka, S. K., Pressure Induced Glasses, in Recent Trends in High Pressure Physics, A. K.
Singh, ed., Oxford and IBH publishing, 254-258 (1992).

92
 References

Tse, J. S. and D. D. Klug, (1992), High Pressure Phase Transitions in Silica and Berlinite
by Molecular Dynamics Simulations, in Recent Trends in High Pressure Physics, A. K.
Singh, ed., Oxford and IBH publishing, 274-279 (1992).

Zhugin, Yu. N. and K. K. Krupnikov, Structural Peculiarities of Plastic Wave Front in

Quartz in the Mixed Phase Region, in Transactions of the XI International (lUPAP-
lUTAM) Symposium on Non-Linear Acoustics, Part II, Novosibirsk, 196-200 (1987).

Section 10
Ahrens, T. J. and G. E. DuvaU, Stress Relaxation Behind Elastic Shock Waves in Rocks, J.
Geophys. Res., 71,4349-4360 (1966).

Asay, J. R., G. R. Fowles and Y Gupta, Determination of Material Relaxation Properties

from Measurements on Decaying Elastic Shock Fronts, J. Appl. Phys., 43, 744-746
(1972).

Cagnoux, J. and F. Longy, Is the Dynamic Strength of Alumina Rate Dependent, in Shock
Compression of Condensed Matter -1987, S. C. Schmidt and N. C. Holmes, eds. Elsevier,
293-296 (1987).

Duvall, G. E., Maxwell-Like Relations in Condensed Materials. Decay of Shock Waves,

Iranian Journal Sci. Tech., 7, 57-69 (1978).

Grady, D. E., Criteria for Impulsive Rock Fracture, Geophys. Res. Letters, 7, 255-258
(1980)

Grady, D. E., Shock-Wave Properties of High-Strength Ceramics, in Shock Compression

of Condensed Matter - 1991, S. C. Schmidt, R. D. Dick, J. W. Forbes and D. G. Tasker,
eds. Elsevier Science Publishing, 455-458 (1992a).

Grady, D. E., Shock-Compression Properties of Ceramics, in Recent Trends in High-Pres-

sure Research, A. K. Singh, ed., Oxford and IBH Publishing, 641-650 (1992b).

Grady, D. E. and J. L. Wise, Dynamic Properties of Ceramic Materials, Sandia National

Laboratories Report, SAND93-0610, September (1993).

Grady, D. E., Shock-Wave Strength Properties of Boron Carbide and Silicon Carbide, in
Proceedings of EuroDYMAT 94 - International Conference of Mechanical and Physical
Behavior of Materials under Dynamic Loading, Oxford England, September 26-30
(1994).

Kipp, M. E. and D. E. Grady, Shock Phase Transformation and Release Properties of Alu-
minum Nitride, in Proceedings of EuroDYMAT 94 - International Conference of Mechan-
ical and Physical Behavior of Materials under Dynamic Loading, Oxford England,
September 26-30 (1994).

93
Lankford, J., Mechanisms Responsible for Strain-Rate-Dependent Compressive Strength
in Ceramic Materials, J. Am. Ceram. Soc, 64, C33-C34 (1981).

Partom, Y, Difficulties in Modeling the High Rate Viscoplastic Response ofAD995 Alumi-
na, Institute of Advanced Technology Tech. Note, IAT.TN0035, April (1994).

Rosenberg, Z., N. S. Brar and S. J. Bless, Elastic Precursor Decay in Ceramics as Deter-
mined with Manganin Stress Gauges, in 2rd International Conference on Mechanical and
Physical Behavior of Materials under Dynamic Loading, Ies Editions des Physique, 707-
711 (1988).

Steinberg, D. J., Computer Studies of the Dynamic Strength of Ceramics, in 3rd Interna-
tional Conference on Mechanical and Physical Behavior of Materials under Dynamic
Loading, Ies Editions des Physique, 837-844 (1991).

94
DISTRIBUTION:

INTERNAL EXTERNAL
1400 E. H. Barsis T. F. Adams
1404 J. A. Ang Los Alamos National Laboratory
1421 S. S. Dosanjh MS F663
1422 R. C. Allen Los Alamos, NM 87545
1423 E. F. Brickell
1424 A.L.Hale T. J. Ahrens
1431 J. M. McGlaun Geophysics Division MS/252-21
1431 E. S. Hertel California Institute of Technology
1431 J. S. Peery Pasadena, CA 91125
1431 A.C.Robinson F AUahdadi
1431 T. G. Trucano Phillips Laboratory
1431 M.K.Wong PLAVSSD
1432 P Yarrington Kirtland AFB, NM 87117-6008
1432 P J. Chen M. L. Alme
1432 H.E.Fang 102 Stevens Forrest Professional Center
1432 A. V. Farnsworth 9650 Santiago Road
1432 G. I. Kerley Columbia, MD 21045
1432 M. E. Kipp
1432 S. A. Silling C. E. Anderson
1432 P A. Taylor Southwest Research Institute
1433 R L. Stanton 6220 Culebra Road
1433 J. A. Ang San Antonio, TX 78284
1433 L. C. Chhabildas J. A. Bailey
1433 M. D. Furnish U.S. Army Research Office
1433 D. A. Crawford P O . Box 12211
1433 D. E. Grady (35) Research Triangle Park, NC 27709
1434 D. R. Martinez E. Baker
1500 D. J. McCloskey Army ARDE Center
1511 J. S. Rottler SMCAR-AEE-WW
1512 A. C. Ratzel Picatinny Arsenal, NJ 07806-5000
1561 H.S.Morgan
1562 R. K.Thomas P. Bartkowski
1562 J.W. Swegle Army Research Laboratory
2565 S. T. Montgomery Materials Directorate
7141 Technical Library (5) AMSRL-MA-PD
7151 Technical Publications Watertown, MA 02172-0001
7613 Document Processing for D. W. Baum
DOE/OSTl (10) L-35
8523-2 Central Technical Files Lawrence Livermore National Laboratory
9121 W.Tucker Livermore,CA 94550
9121 J. Hickerson
9723 M. J. Forrestal

95
p. Beaulieu J. A. Collins
Army Research Laboratory U.S. Air Force Armament Laboratory
Materials Directorate AD/MNW
AMSRL-MA-PD Eglin Air Force Base, FL 32542-5434
Watertown, MA 02172-0001 J. W Coltman
G. Bishop Simula Inc.
Army Research Laboratory 10016 South 51st Street
Materials Directorate Phoenix, AZ 85044
AMSRL-MD D. Curran
Watertown, MA 02172-0001 SRI International
S. J. Bless 333 Ravenswood Avenue
Institute for Advanced Technology Menlo Park, CA 94025
4030-2 W Braker Lane D. Dandekar
Austin, TX 78759-5329 Army Research Laboratory
W J. Bruchey Materials Directorate
Army Research Laboratory AMSRL-MA
Weapons Technology Directorate Watertown, MA 02172-0001
AMSRL-WT J. Dehn
Aberdeen Proving Ground, MD Army Research Laboratory
21005-5066 Weapons Technology Directorate
G. Bulmash AMSRL-WT-TA
Army Research Laboratory Aberdeen Proving Ground, MD
Weapons Technology Directorate 21005-5066
AMSRL-WT-TA K. Epstein
Aberdeen Proving Ground, MD DOW Chemical USA
21005-5066 Ordnance Systems, 800 Building
Kyu Cho Midland, MI 48667
Army Research Laboratory J. C. Foster, Jr.
Materials Directorate U.S. Air Force Armament Laboratory
AMSRL-MA AD/MNW
Watertown, MA 02172-0001 EgUn Air Force Base, FL 32542-5434
S. Chou K. Frank
Army Research Laboratory Army Research Laboratory
Materials Directorate Weapons Technology Directorate
AMSRL-MA AMSRL-WT-TD
Watertown, MA 02172-0001 Aberdeen Proving Ground, MD
R. Clifton 21005-5066
Brown University G. Gazonas
Division of Engineering Army Research Laboratory
Providence, RI 02912 Weapons Technology Directorate
AMSRL-WT-PD
Aberdeen Proving Ground, MD
21005-5066

96
W Gillich G.Johnson
Army Research Laboratory AUiant Techsystems, Inc.
Weapons Technology Directorate 7225 Northland Drive
AMSRL-WT-TA Brooklyn Park, MN 55428
Aberdeen Proving Ground, MD J. N. Johnson
21005-5066 Los Alamos National Laboratory
W. Gooch MS B221
Army Research Laboratory Los Alamos, NM 87545
Weapons Technology Directorate R. Katz
AMSRL-WT-TA Army Research Laboratory
Aberdeen Proving Ground, MD Materials Directorate
21005-5066 AMSRL-MD
Y. Gupta Watertown, MA 02172-0001
Washington State University P. Kingman
Department of Physics Army Research Laboratory
Pullman, WA 99163 Weapons Technology Directorate
G. T. Gray AMSRL-WT
Los Alamos National Laboratory Aberdeen Proving Ground, MD
Los Alamos, NM 87545 21005-5066
W. Haskell K. Kimsey
Army Research Laboratory Army Research Laboratory
Materials Directorate Weapons Technology Directorate
AMSRL-MA-PA AMSRL-WT
Watertown, MA 02172-0001 Aberdeen Proving Ground, MD
T. Holmquist 21005-5066
AUiant Techsystems, Inc. R. W. Kocher
7225 Northland Drive Advanced Research Projects Agency
Brooklyn Park, MN 55428 Land Systems Office
Y. Horie 3701 North Fairfax Drive
North Carolina State University Arlington, VA 22203-1714
Dept. of Civil Engineering J. Lankford
Raleigh, NC 27607 Southwest Research Institute
C. Hubbard 6220 Culebra Road
Army Research Laboratory San Antonio, TX 78284
Materials Directorate D. Lassila
AMSRL-MA-CA(APG) L-35
Aberdeen Proving Ground, MD Lawrence Livermore National Laboratory
21005-5066 Livermore,CA 94550
K. Iyer K. T. Leighton
U.S. Army Research Office Lanxide Armor Products, Inc.
P O . Box 12211 1300 Marrows Road
Research Triangle Park, NC 27709 P O. Box 6077
Newark, DE 19714-6077

97
D. Lovelace R. Palicka
Army Missile Command CERCOM, Inc.
AMSMI-RD-ST-WF 1960 Watson Way
Redstone Arsenal, AL 35898-5240 P O. Box 70
D. Mandell Vista, CA 92083
Los Alamos National Laboratory R. Paricio
MS F663 Coors Ceramics Company
Los Alamos, NM 87545 600 Ninth Street
M. Manghnani Golden, CO 80401
Mineral Physics Group A. Prakash
University of Hawaii Army Research Laboratory
2525 Correa Rd. Weapons Technology Directorate
Honolulu, HA 96822 AMSRL-WT-WD
H. W. Meyer Aberdeen Proving Ground, MD
Army Research Laboratory 21005-5066
Weapons Technology Directorate W. W. Predebon
AMSRL-WT-TA College of Engineering
Aberdeen Proving Ground, MD Michigan Technological University
21005-5066 Houghton, MI 49931
M. Meyer G. F. Raiser
Univ. of Calif, at San Diego Washington State University
Dept. of Applied Mech. & Eng. Sciences Department of Physics
La JoUa, CA 92093 Pullman, WA 99163
J. D. Morrow E. Rapacki
FMC Corporation Army Research Laboratory
Ground Systems Division Weapons Technology Directorate
1107 Coleman Avenue Box 367 AMSRL-WT-TA
San Jose, CA 95103 Aberdeen Proving Ground, MD
S. Nemat-Nasser 21005-5066
Univ. of Calif, at San Diego A. M. Rajendran
Dept. of Applied Mech. & Eng. Sciences Army Research Laboratory
La JoUa, CA 92093 Materials Directorate
T Nicholas AMSRL-MA
Air Force Wright Aeronautical Labs. Watertown, MA 02172-0001
Air Force Systems Command G. Randers-Pehrson
Materials Laboratory Army Research Laboratory
Wright-Patterson AFB, OH 45433 Weapons Technology Directorate
D. Orphal AMSRL-WT
California Research and Technology, Inc Aberdeen Proving Ground, MD
5117 Johnson Drive 21005-5066
Pleasanton, CA 94566

98
G. Ravichandran T Wright
Graduate Aeronautical Laboratories Army Research Laboratory
California Institute of Technology Weapons Technology Directorate
MS 105-50 AMSRL-WT
Pasadena, CA 91125 Aberdeen Proving Ground, MD
M. Scheidler 21005-5066
Army Research Laboratory
Weapons Technology Directorate
AMSRL-WT
Aberdeen Proving Ground, MD
21005-5066
S. Segletes
Army Research Laboratory
Weapons Technology Directorate
AMSRL-WT
Aberdeen Proving Ground, MD
21005-5066
M. Slavin
Army Research Laboratory
Materials Directorate
AMSRL-MA
Watertown, MA 02172-0001
D. Steinberg, MS L35
Lawrence Livermore National Laboratory
P O. Box 808
Livermore, CA 94550
J. Thompson
U.S. Army Tank-Automotive Command
AMSTA-RSS
Warren, MI 48397-5000
D. J. Viechniki
Army Research Laboratory
Materials Directorate
AMSRL-MA
Watertown, MA 02172-0001
T. Weerasooriya
Army Research Laboratory
Materials Directorate
AMSRL-MA-PD
Watertown, MA 02172-0001