-
in
Pullm
gton
2005
ndureeinume p
on ineraappolycffecdentcantitialhinged tolevect trobshys
to study the dynamic properties of materials under planar ous
approximations have been developed to relate the pull-
JOURNAL OF APPLIED PHYSICS 99, 023528 2006loading, and a variety
of experimental methods have beendeveloped for this purpose,
including explosive loading,1plate impact,2 particle beams,3 laser
irradiation,4,5 and mag-netic loading.6 One application has been to
study dynamictensile behavior under uniaxial strain loading,
usually re-ferred to as dynamic spallation. Since the 1960s,
numerousplate-impact experiments have been conducted to
investigatespall behavior in the intermediate-to-high strain
rateregime.7,8 Quantification of spallation properties under
high-rate loading has resulted in significant improvement in
un-derstanding the basic properties controlling nucleation
andgrowth of damage produced during dynamic tensile
failure.Summaries of these studies are available in several
reviewarticles.713
There are generally two methods for estimating spallstrength in
plate-impact experiments. The tensile stress suf-ficient to
initiate the onset of spallation can be estimatedfrom recovery and
postshot examination of a planar samplesubjected to various initial
shock stresses followed by tensileloading.7 In this approach, shock
amplitudes are varied toinfer the tensile stress required for spall
failure. Another pro-cedure is to use time-resolved diagnostics to
determine thepullback in free-surface velocity resulting from
planar ten-sile failure produced during the loading process. Since
the
back velocity to the tensile stress achieved
beforespallation.8,1417
Spallation is a cooperative nucleation, growth, and coa-lescence
process of void or crack formation during tensileloading.18 Based
on the work of Shockey, damage nucleationgenerally initiates at a
relatively low stress level, which de-pends on material properties
and the tensile loading history,as compared to the peak tensile
stress produced during theloading process. In this regard, the
spall strength is not amaterial property, since the value for this
parameter dependson loading conditions and sample geometry, in
addition tomaterial microstructure.8 Through an extensive array of
pre-vious experiments, researchers have investigated spall
be-havior under varying stress amplitudes and pulse durations,with
the result that the loading conditions leading to dynamicfailure
are well understood for planar loading;7,1820 severalspecific
examples are given in summary articles.11,2125
Gray and co-worker have demonstrated that shock-induced
deformation produces a variety of microstructuraldefects that
strongly influence deformation behavior.26,27Since spallation is
preceded by shock compression, initialmetallurgical properties, as
well as the shock-induced defor-mation structure, are therefore
important to dynamic failure.It has also been observed that the
initial microstructure isclosely related to the operative mechanism
of damage initia-tion and growth;20,2832 particularly noteworthy is
the workaElectronic mail: [email protected] behavior of aluminum
with varyX. Chen, J. R. Asay,a and S. K. DwivediInstitute for Shock
Physics, Washington State University,D. P. FieldSchool of
Mechanical and Materials Engineering, WashinWashington 99163
Received 25 August 2005; accepted 13 December
A series of plate-impact spall experiments was comicrostructural
conditions of aluminum, including thand commercially pure 1060
polycrystalline alumorientations, over the stress range of 422 GPa.
Thsignature of spall strength, is observed to dependduration, and
loading rate. The pullback velocity geGPa and achieves a maximum as
the impact stress100 single-crystal Al is higher than that for both
psamples, indicating that grain orientation strongly aalso show
that the spall behavior is strongly depenshock pulse duration was
observed to be less signifiindicates that the observed differences
depend on ininitial microstructures and impurities have a
diminisnear 22 GPa. However, initial properties are observstructure
of the pullback velocity profile at all stressa sharp slope during
pullback followed by a distinobserved. The occurrence of this
change in slope israte, and grain size. 2006 American Institute of
P
I. INTRODUCTION
Shock wave methods have been used for several
decades0021-8979/2006/992/023528/13/$23.00 99, 02352
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
tog microstructures
an, Washington 99163
State University, Pullman,
; published online 31 January 2006
cted to study the spall strength of sevengrain sizes of 6061 Al
alloy, both ultrapure, and single-crystal Al with two different
ullback velocity, which is a characteristicnitial
microstructure, impact stress, pulselly increases over the stress
range of 414roaches 22 GPa. The pullback velocity ofrystalline
samples and 111 single-crystalts material response. Experimental
results
on sample thickness, while the effect of. Comparison among three
6061 materialsyield strength. The results also show thateffect on
the pullback velocity at stresseshave a pronounced effect on the
detailed
ls. In particular, an interesting feature, i.e.,ansition to a
slower slope, is consistentlyerved to depend on impact stress,
loadingics. DOI: 10.1063/1.2165409
free-surface velocity profiles do not provide direct
measure-ment of the tensile fracture stress in these experiments,
vari- 2006 American Institute of Physics8-1
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-2 Chen et al. J. Appl. Phys. 99, 023528 2006of Curran et
al.,7 who conducted systematic studies of initialmetallurgical
properties on the spallation properties of sev-eral materials. This
work examined the effects of microstruc-tural features, such as
inclusions, grain boundary, subgrainstructure, and texture, on void
nucleation, growth, and coa-lescence, and resulted in the
development of the nucleationand growth NAG models that provide a
fundamental basisfor describing dynamic failure in ductile
materials. The re-cent work of Schwartz et al.29,30 extended the
fundamentalunderstanding by examining the effects of grain size
andinclusions on the spallation of well-characterized pure
cop-per.
Although considerable work has been conducted to es-tablish
basic failure mechanisms, there are several remainingissues: 1
There is still no widely accepted spall criterion orpredictive
damage accumulation model; 2 there are limiteddata regarding the
relative importance of initial metallurgicalproperties;13 and 3
there is no clear understanding of theimportance of loading
conditions, such as shock amplitudeand pulse duration. Much of this
uncertainty can be related tothe lack of carefully characterized
samples that allow corre-lation of microstructural effects with
observed spallationproperties. Except for a few cases, such as
copper,29,30 mate-rial properties and shock loading parameters have
generallynot been systematically varied.
The goal of the present experiments was to systemati-cally study
dynamic spallation in aluminum. Initially, we fo-cused on how
microstructural effects, including grain size,impurities, and
precipitates influence pullback signals forvarying stress
amplitudes, loading rates, and pulse durations.Where possible, a
single material property or loading effectwas changed while holding
other parameters constant. Thesestudies involved a variety of
aluminum materials, includingcommercially pure aluminum 99.6%,
ultrapure polycrystal-line aluminum 99.9998% Al, single-crystal Al
99.999%pure for 100 and 99.98% for 111, and three heat treat-ments
of 6061 aluminum alloy. Plate-impact techniques wereused to produce
planar tensile loading of aluminum speci-mens and velocity
interferometry33 was used to measure par-ticle velocity changes
resulting from the spall process. Inorder to clarify the fracture
mechanisms, we extensivelycharacterized the initial microstructures
and material proper-ties of all materials studied, including
measurements of grainsize and distribution, texture, precipitate
size and distribu-
tion, composition, dislocation density, and microhardness.
Inthis way, specific aspects of spallation properties can be
cor-
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
torelated to metallurgical properties. Through these compari-sons,
we are able to identify potential correlations betweenmesoscale
material properties and the continuum spallationmeasurements.
For all experiments, the peak impact stresses were variedfrom 4
to 22 GPa and different sample thicknesses were usedto evaluate the
loading rate, which varied from about0.131.5106 s1. Results of
these experiments illustratethat pullback velocities of the various
aluminum materialsgenerally increased up to 60% over the range of
414 GPa.Results for 1060 aluminum and two 6061 Al alloys
showessentially constant or slightly declining pullback velocityfor
stresses from 14 to 22 GPa. Ultrapure polycrystalline Alshowed a
monotonic increase from 4 to 22 GPa without evi-dence of leveling
off as observed in other polycrystallinematerials. Besides pullback
velocity measurements, there ap-pears to be a characteristic
structural change in the slope ofpullback signals that is dependent
on impact stress, loadingrate, and grain size. This is thought to
represent a changefrom a brittlelike response to a more ductile
behavior. Pre-liminary results obtained on single crystals also
suggest astrong orientation-dependent spall behavior, with 100
giv-ing much higher pullback velocity and a different structurefor
the pullback signal, as compared with the 111 single-crystal
data.
We first present the experimental configuration used fordynamic
measurements of pullback velocity and then discussthe initial
metallurgical properties of each material studied.Then, the
experimental results are discussed, including theobserved
dependences on impact stress, loading rate, andgrain size.
II. EXPERIMENTAL TECHNIQUEA. Impact experiments
The approach for using planar loading techniques to pro-duce
spallation involves configuring the impactor and sampleto cause
wave interactions that result in controlled states oftension. The
basic configuration is shown in Fig. 1a. Inthese experiments, a
flat plate of specified thickness is im-pacted against a flat
target. The lateral dimensions are chosenso that edge effects
resulting from radial release waves donot reach the central
measurement point of the measuredtarget response during times of
interest.
FIG. 1. Plate-impact configurationused for spallation studies. a
Con-figuration used for quartz or sapphireimpactors in which the
rear surface ofthe impactor was unsupported over thecentral
section. b Configuration usedfor thin quartz impactors or
aluminumimpactors at velocities of13002400 m/s.In previous
spallation experiments on aluminum, the im-pactor was usually
fabricated from aluminum, resulting in
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-3 Chen et al. J. Appl. Phys. 99, 023528 2006symmetric
impact.8,23,24 This configuration complicates theanalysis at low
stress levels, since the measured wave struc-ture is strongly
influenced by the input wave structure. Toavoid this problem, we
designed the experiments so that asingle-step unloading from the
impactor side of the samplewas introduced into the sample for all
experiments. At rela-tively low stresses, thin Z-cut quartz or
C-cut sapphire plateswere used as impactors to produce impact
stresses of 4 and8.9 GPa, respectively. Because both materials are
elastic overthis stress range the dynamic elastic limit of Z-cut
quartz isabout 6 GPa and for C-cut sapphire it is about 12 GPa,
asingle elastic shock wave is produced in the impactor afterimpact.
This shock wave reflects from the rear surface as anunloading, or
rarefaction wave, usually referred to as a rar-efaction fan since
the velocity of different states on the wavedepends on stress
amplitude.7 For experiments involvingthick elastic impactors at low
pressure, the central area, typi-cally about 25.7 mm diameter of
the Z-cut quartz or C-cutsapphire impactor, was unsupported to
produce complete un-loading. For quartz impactor thickness of about
0.5 mm, aLexan backing was used to support the thin plate and
preventfailure during launching.34 For impact stresses of
nominally13.5 GPa and above, symmetric impact with aluminum
im-pactors was used. At these stress levels, the elastic
precursoris overdriven by the plastic wave, so that a sharp
singleshock loading is also produced upon impact.
The technique for producing tensile loading is illustratedin
Fig. 2a. Upon impact, planer shocks are produced inboth the
impactor and sample, as indicated by the x-t traceslabeled S.
Reflection of the shock wave from the impactorrear surface at time
t2, which is either a free surface or lowerimpedance Lexan backing,
produces a rarefaction fan re-ferred to as R in the figure.
Similarly, reflection of theshock wave from the sample free surface
results in a left-going rarefaction fan propagating back into the
sample. In-teraction of the two rarefaction waves produces a state
oftension that can lead to separation of the material.
The plate-impact experiments were performed using ei-ther a
64-mm-diameter light gas gun for impact velocitiesup to 750 m/s or
a 30-mm-diameter propellant gun for
2impact velocities up to 2400 m/s. For the gas gun experi-ments,
both the impactor and target plates were about 30 mm
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toin diameter. For the propellant gun experiments, the impactorand
target were approximately 25 mm in diameter. The im-pact velocities
for the gas gun and powder gun experimentswere determined to within
0.25% and 0.5% accuracy, respec-tively.
Time-resolved free-surface velocities were measuredwith a
velocity interferometer referred to as VISAR for ve-locity
interferometer system for any reflector.33 This interfer-ometer
allows velocity measurements with an accuracy ofabout 1% and a time
resolution of about 2 ns. For spallationexperiments at impact
stresses greater than about 17 GPa, itwas found that interferometer
measurements of the free-surface velocity required well-polished
mirror surfaces tominimize material ejection that obscures the
interferometersignal.35
Figure 2b shows a typical free-surface velocity signalobtained
in a spallation experiment. The velocity differencebetween the
first peak plateau and the first valley, ufs in thefigure, is
generally referred to as the pullback velocity. Thisis the primary
measurement reported in this paper instead ofspall strength *. The
spall strength can be estimated fromthe pullback velocity based on
an assumed relationship be-tween pullback velocity and spall
strength and is generallydependent on the specific elastic-plastic
unloading pathfollowed.8 Since pullback velocity is proportional to
spallstrength, it is an equivalent measure of spall strength, so
ourresults provide a relative measure of spall strength for
thedifferent materials studied.
As shown in Fig. 2a, the initial unloading wave fromthe flyer
plate reaches the VISAR recording area at time t3.Tensile loading
is produced in the sample and then recordedby the interferometer at
times between t3 and t4. The in situtensile loading rate resulting
from the wave interactions can-not be accurately inferred from
free-surface wave profiles.For this reason, we report the directly
observed peak decel-eration rate of the free-surface velocity
occurring before timet4, which provides an estimate of the tensile
loading rate8 forcomparisons between samples. The structure of the
wavesegment from t4 to the second peak, or pullback structure,as
marked in Fig. 2b, is primarily due to the details of the
FIG. 2. Spallation produced by plateimpact. a x-t diagram for
symmetricimpact at high pressure. b Typicalwave profiles observed
on the free sur-face of a planar sample.nucleation and growth phase
of the failure process. A steeprise from the minimum usually
signifies a brittle failure pro-
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-4 Chen et al. J. Appl. Phys. 99, 023528 2006cess, whereas
a more gradual recovery rate generally indi-cates ductile
failure.
B. Materials
Each of the aluminum materials studied was
carefullycharacterized for composition, grain size and
distribution,microhardness, impurity and precipitate distributions,
and anestimate of the dislocation microstructure. The
individualproperties for the materials studied are summarized in
TableI. The average grain size and orientation of the specimenwere
determined with the electron backscatter diffractionEBSD
technique,36 with representative orientation imagesshown in Fig. 3.
EBSD is a scanning electron microscopySEM-based technique that can
be used to perform rapiddiffraction analysis that yields spatially
specific crystallo-graphic data from which grain sizes, individual
crystal ori-entations, and size distributions in a polycrystalline
material
TABLE I. Summary of material properties.
Pure aluminum
1060 Ultrapure 100
Density g/cm3 2.705 2.700 2.699Compositionpurity, wt %
99.6% 99.9998% 99.999%
Grain size m 182 286 N/AHardnesskg f /mm2
HRA N/A N/A N/AHV 20.50.6 19.70.4 14.30.3 1
Textureb Type 110 100 N/AStrength 2.2 10.8 N/A
aEBSD showed that the crystal orientation between neighboring
grains was vnot fully formed into actual grains.bTexture strength
is a measure of orientation randomness of grains. It is
dimemicrostructure would have a texture strength of 1.Downloaded 16
Feb 2006 to 134.121.73.50. Redistribution subject toare inherently
obtained. The triangle shown in Fig. 3 illus-trates the assigned
colors representing specific crystal direc-tions aligned with the
specimens surface normal orientation.A brief summary of these
properties is given in the followingsections. More details on
specific metallurgical properties forthe materials studied here are
given in a recent paper byHuang and Asay.34
1. Pure aluminum 1060, pure polycrystalline, andsingle-crystal
aluminum
As illustrated in Fig. 3a, the 1060 aluminum 99.6%Al studied had
a nearly randomly oriented and equiaxedgrain structure, with an
average grain-size diameter of about180 m. EBSD measurements
indicate that the 1060samples had a relatively low degree of
texture. EBSD mea-surements Fig. 3b of the ultrapure aluminum
99.9998%Al showed a strong recrystallization texture with 100
6061 alloy
6061-02 6061-20 6061-80
9 2.703 2.704 2.706% Al:95.8%, Mg:1.0, Si:0.6% Fe:0.7%,
Cr:0.040.35%
5a 34 4713.50.5 280.5 39.50.5
0.3 59.41.8 80.70.6 1081.4
100 100 1006.9 2.9 1.72.3
mall, indicating that the observed structure consisted of
subgrains that were
less, and higher texture strength indicates higher texture. The
perfect random
FIG. 3. EBSD results of 6061 Al: a6061-02, b 6061-20, c 6061-80,
d1060 Al, and e ultrapure aluminum.111
2.6999.98
N/AN/A
4.5
N/AN/A
ery s
nsion AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-5 Chen et al. J. Appl. Phys. 99, 023528 2006planes
aligned with the normal direction. EBSD measure-ments at different
locations in the samples indicated that thegrain size displayed
considerable spatial variation. The mea-sured average grain size by
area was about 300 m, al-though several of the larger grains had
dimensions of about300900 m. Preliminary data were also obtained
onsingle crystals of aluminum with 100 99.999% pure and111 99.98%
pure orientations. The higher impurity con-tent in the 111 crystal
is mainly due to a few elements: Si150 ppm, Fe 25 ppm, B 12 ppm, P
3.3 ppm, etc. Thesesamples were diagnosed with the Laue diffraction
techniqueto ensure proper orientation.
2. 6061 aluminum alloyThe initial goal of experiments on the
6061 aluminum
alloy was to evaluate the effects of average grain dimensionsof
2, 20, and 80 m on spallation properties. To achievethese grain
properties, a process of preparing the specimenswas developed
consisting of cold rolling followed by heattreatment.37 Following
heat treatment, all samples werewater-quenched and then aged at 160
C for 18 h. Based onthe preplanned grain sizes, we designate the
sample sets as6061-02, 6061-20, and 6061-80. The initial goal of
attainingthese grain conditions was not fully achieved. EBSD
mea-surements indicated that the final grain sizes for the
differentsample sets ranged from 5 to 50 m, as given in Table I.In
addition, the grains for the 6061-02 material were notfully
recrystallized, but consisted of substantial subgrainboundaries, as
shown by the EBSD measurements in Fig. 3;whereas grains for the
other two configurations were welldefined. Pole figure analysis
indicated that the texturestrength for these materials increased
from a maximum peakvalue of 1.7 times random grains nearly randomly
orien-tated for the larger grained 6061-80 to 6.9 times
randommoderate texture for the fine grained 6061-02.
The hardness value for the 6061-02 material was sub-stantially
less than that for the other two material sets, asnoted in the
table. The 6061-80 material had a hardness simi-lar to that for
stock 6061-T6, although the grain size wasconsiderably smaller than
that for a standard plate stock.23Finally, it was found that the
dislocation structure varied be-tween the 6061 alloys studied.38 A
dense dislocation cellstructure with tangled dislocation walls was
observed for all6061 materials, with the dislocation density of
6061-20 esti-mated to be slightly less than that for 6061-02.
Transmissionelectron microscopy TEM was also used to investigate
thenature of impurities and precipitates in the samples.
Verycoarse, blocky precipitate structures with sizes in the rangeof
35 m were observed in 6061-02. For 6061-80, TEMindicated that the
impurity structures consisted of both fineconstituent and coarse
blocky particles.38
III. EXPERIMENTAL RESULTS
By adjusting the impactor and impact velocity as dis-cussed
earlier, the impact stress was varied over the range of422 GPa. In
most experiments, the flyer thickness was
about half the target thickness to produce spallation near
themiddle of the sample. The impactor and sample thicknesses
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
towere generally maintained constant to compare the resultsobtained
on different materials. Both thin, nominally 1 mmthick, and thick
samples, nominally 5.9 mm thick, were usedto study loading rate
effects. In these experiments, the impactstresses were held
constant for comparison of material ef-fects on loading rate. The
combined set of experiments al-lowed evaluation of initial shock
loading on the spallationproperties of the various materials
studied.
As mentioned, determination of spall strength from thepullback
velocity measurements has considerable uncer-tainty. The measured
pullback velocity also has random er-rors that arise primarily from
the measured VISAR velocityhistory, mainly from fluctuations in
peak velocity and uncer-tainties in reducing fringe records, which
result in a com-bined uncertainty of about 5 m/s. Besides this,
some mate-rial property variations also contribute to the
measureddifferences. Ultrapure polycrystalline aluminum and
single-crystal 100 aluminum samples were found to have
poorreproducibility in pullback signals when compared with1060.
This may be partially due to grain structure differencesfor the
polycrystalline materials, as 1060 has more evenlydistributed and
equiaxed grain sizes, while ultrapure Al pos-sesses a nonuniform
grain structure, as illustrated in Fig. 3.
Table II summarizes all experimental results. The spallstrength
values were estimated from the relation *=0.5ceffufs, where ceff is
the effective wave speed
24and is
approximately 5.77 km/s for aluminum. In the
followingsubsections, these are compared for the different
materials,stresses, and loading rates to evaluate the effects of
peakstress, loading rate, grain size, and impurity content on
spal-lation properties.
A. Effects of compressive peak stress
We first discuss the effects of stress amplitude for
thedifferent materials studied. In this set of experiments
thesample thickness was kept fixed and the stress amplitudevaried
over the range of 422 GPa. Figures 4 and 5 show thewave profiles,
normalized in time to the arrival of the plasticwave obtained on
the thick samples. Figure 4a illustratesthe full set of wave
profiles measured on the 1060 alloy atdifferent impact stress
levels and Fig. 4b shows similarprofiles on ultrapure aluminum.
Figure 4c gives the resultsobtained on aluminum single crystals for
100 and 111orientations. Similarly, Figs. 5a and 5b illustrate the
waveprofiles at different stress levels for the 6061-20 and
6061-80sample sets. As illustrated, the shock wave structure at 4
GPain general consists of an elastic wave with a velocity of
about6.4 km/s, followed by a plastic shock wave with a velocityof
about 5.6 km/s. The plastic wave rise time is consistentwith
previously reported values for aluminum.39 As the am-plitude of the
shock wave is increased, the rise time of theplastic shock becomes
significantly steeper and then be-comes unresolvable to within the
time resolution of the re-cording instrumentation. The elastic wave
amplitudes ob-tained on different materials generally have
different
amplitudes. This has been previously observed inaluminum,34
which we will discuss in a later section.
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
m/s
449449502449501502501503499501502500725565335565585566588921348
023528-6 Chen et al. J. Appl. Phys. 99, 023528 2006Figure 6
presents the free-surface velocity profiles ob-tained on thick
samples of 1060, adjusted to the differencebetween the measured
peak particle velocity U and the im-pact velocity Vp at each impact
stress 4.1, 8.9, 13.5, 17.3,and 22.2 GPa Expt. Nos. 5, 13, 17, 20,
and 21 in Table II,respectively. This adjusts the base line to the
same value forthe pullback signals. It is observed that for similar
impactorand sample thicknesses, the pulse duration decreases
slightlywith increased impact stress, because both the shock
velocityand the unloading wave velocities increase with stress.
Fig-ure 6 also shows that for the 1060 alloy, the pullback
velocityincreases by about 50% when the impact stress increasesfrom
4 to 17.3 GPa. This increase was observed to be repeat-able, as
indicated in Table II. For higher impact stresses, aslight decrease
is observed in pullback velocity, although thedifference is
insignificant over the range of 1422 GPa for1060.
The pullback velocities obtained on thick samples for
allmaterials not including Expt. Nos. 912, which correspondto
higher tensile loading rates are plotted in Fig. 7. Thefigure shows
that the pullback velocities for most aluminummaterials studied
exhibit a stress dependence. The major re-sult is that the pullback
velocities obtained on different poly-
TABLE II. Summary of experiments.
Expt.No.
Flyer Sample
VpMaterial h1 mm Material h2 mm
1 Quartz 3.201 6061-02 5.8852 Quartz 3.200 6061-20 5.8063 Quartz
3.197 6061-20 5.8674 Quartz 3.186 6061-80 5.8835 Quartz 3.192 1060
5.8936 Quartz 3.199 1060 5.8867 Quartz 3.185 Pure 5.6628 Quartz
3.197 100 5.8819 Quartz 0.524 6061-20 0.997
10 Quartz 0.524 6061-80 0.99811 Quartz 0.524 1060 1.00012 Quartz
0.524 Pure 0.99413 Sapphire 5.086 1060 5.88514 6061-20 2.969
6061-20 5.850 115 6061-80 3.078 6061-80 5.910 116 1060 3.000 1060
5.904 117 1060 3.247 1060 5.698 118 Pure 3.023 Pure 5.862 119 Pure
2.999 Pure 5.883 120 1060 3.103 1060 5.786 121 1060 3.245 1060
5.691 222 1060 3.247 1060 5.682 223 1060 1.998 1060 5.855 224
6061-80 1.986 6061-80 5.704 225 6061-20 1.982 6061-20 5.707 226
Pure 1.988 Pure 5.731 227 100 1.999 100 5.711 228 6061-T6 1.998 100
5.685 229 6061-T6 1.996 111 5.691 2
a* is the spall strength.crystalline aluminum samples differ
significantly at 4 GPa,but are essentially the same to within
experimental errors at
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
to22 GPa. Another observation is that the pullback velocitiesof all
materials increase for shock stresses between 4 and 14GPa. For
polycrystalline materials, the 1060 alloy shows thelargest rate of
increase with impact stress over the range of414 GPa, followed by
6061-20, ultrapure aluminum, and6061-80 alloy. As the impact stress
increases from 14 to 22GPa, the pullback velocities for 6061-80
show an apparentdecrease at the highest impact stress, becoming
comparableto that for 1060 at 22 GPa. The data obtained on
6061-20between 14 and 22 GPa show a nearly constant or
slightlyincreasing pullback velocity, to within experimental
uncer-tainty, whereas the pullback velocity for ultrapure
aluminumshows a monotonic increase over this range.
Although preliminary, the results obtained on single-crystal
aluminum are interesting and may shed light on fail-ure mechanisms
important to the polycrystalline materials.The 100 single-crystal
orientation has the highest pullbackvelocity, which monotonically
increases over the stress rangestudied, being about 40%
consistently larger than that forpure polycrystalline aluminum.
Only one experiment wasconducted on the 111 orientation, but the
result obtained ata stress level of 22 GPa is surprisingly similar
to the averageobtained on polycrystalline materials. The
significant differ-
Peak stress
GPa HEL kbar ufsm/s * GPaa
4.14 1.74 105 0.824.14 3.54 125 0.984.16 3.53 125 0.984.14 6.32
164 1.284.15 1.1 103 0.804.16 0.69 107 0.8344.15 0.55 117 0.914.17
0.85 174 1.364.14 5.63 202 1.5784.15 7.48 215 1.684.16 1.16 169
1.324.14 1.62 187 1.458.96 0.80 125 0.98
13.5 5.04 160 1.2511.2 5.59 174 1.3613.5 159 1.2413.8 149
1.1713.4 127 0.9913.7 138 1.0717.3 158 1.2322.1 153 1.1622.1 149
1.1622.1 156 1.2121.8 156 1.2321.9 166 1.2921.2 149 1.1621.9 240
1.8722.1 227 1.7621.9 151 1.17352350323338275338345331ence between
100 and 111 at 22 GPa suggests a strongcrystal orientation
dependence of spall behavior, but it
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-7 Chen et al. J. Appl. Phys. 99, 023528 2006should be
noted that there is a minor difference in impuritiesbetween these
materials, which sometimes yields a signifi-cant difference in
mechanical response of nearly pure alumi-num. These differences are
assumed to be insignificant forthe observed results but should be
addressed in future experi-ments.
B. Initial yield strength
FIG. 4. Free surface velocity histories for pure aluminum: a
1060 Al, bultrapure Al with various impact stresses, and c 100 and
111 singlecrystals. Sample thicknesses are nominally 5.9 mm.There
have been only limited studies relating spallstrength to initial
hardness and compressive yield properties.
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toButcher28 showed that the spall strength of steel alloys
in-creased with mechanical hardness, which relates to
yieldstrength, since hardness can be semiquantitatively connectedto
the dynamic yield strength for a specific material. For the6061
aluminum alloys we studied, the elastic profiles at 4GPa exhibit
distinct Hugoniot elastic limits HELs, whichcorrespond to the
various hardness values, as shown in TableI. Therefore, similar to
Butchers observation, we observe adirect correlation of these
values with the measured pullbackvelocities in the three 6061
aluminum alloys at 4 GPa, asshown in Fig. 8a. Although the pullback
velocities of6061-20 and 6061-80 alloys are significantly different
forthick samples 5.9 mm at 4 GPa, they are closer at higherstresses
and essentially equal at 22 GPa Expt. Nos. 24 and25. The pullback
velocities for these two materials are alsofound to be similar.
Since the compressive yield strengths ofthese alloys are found to
be similar at high shock stresses,34it appears that there is an
approximate correlation of spallstrength with yield strength.
Furthermore, as shown in Fig.8b, for experiments on thin samples of
6061-20 and 6061-80, the HEL and the observed pullback signals are
also verysimilar, further supporting a rough correlation between
spallstrength and yield strength.
Although these correlations may not likely hold for allloading
conditions and between materials, since the failuremechanisms are
different for compressive and tensile failureswe have made an
initial comparison of our results with sev-eral models previously
developed to relate spall strengthwith yield stress.10,40 We
observe some notable differencesbut also a rough correlation with
compressive strength insome cases, since the relationship may be
valid for a specificalloy with different heat treatments, but not
likely for differ-ent alloys. As an example, 1060 and ultrapure
polycrystallineA1 both have considerably lower initial yield
strengths than6061-02, but have approximately equal or slightly
higherpullback velocities at 4 GPa. In addition, single-crystal
100aluminum, which has a similar HEL to polycrystalline pureA1,
shows an even higher spall strength than 6061-80 at 4GPa. This
observation is consistent with other results,7,30which indicate
that spall strength is also affected by severaldifferent
microstructural deformation mechanisms, in addi-tion to dislocation
generation and hardening associated withcompressive yield
strength.
C. Rate effects
In previous studies of strain rate effects on spall strength,the
impactor thickness h1 and sample thickness h2 werevaried in order
to study the effects of pulse duration andloading rate. This was
also the approach followed in ourwork to evaluate the effect of
tensile loading rate at a givenstress. For experiments at 4 GPa,
sample thicknesses h2 werevaried from 1 to 5.9 mm with a
corresponding change inimpactor thicknesses h1 from 0.524 to 3.2
mm. Figure 9shows two sets of pullback velocities for four
materials1060, 6061-20, 6061-80, and ultrapure polycrystalline
alu-minum that illustrate loading rate effects. Variations in
pulse
duration of about a factor of 6 and average free-surface
de-celeration by an order of magnitude were achieved. In Fig. 9
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
-80 A
023528-8 Chen et al. J. Appl. Phys. 99, 023528 2006it is
observed that the initial free-surface decelerations for
allmaterials at a specific stress level are nominally equal,
sug-gesting an approximate correlation with loading rate. Foreach
material we studied, the amplitude of the pullback sig-nal for thin
samples was larger than that for the thicksamples. This is
consistent with the large number of priorinvestigations, which show
that spallation is a time-dependent damage accumulation process.41
In Fig. 10, thepullback velocity is plotted versus free-surface
velocity de-celeration normalized by the bulk speed c0, which has
beenpreviously used to approximate strain rate at the
failureplane.8,24 We do not interpret the normalized deceleration
asstrain rate, but use it to make relative comparisons of
loadingrate effects on the pullback signal.
The first set of comparisons we will discuss representsthe case
of low loading rate where the target thicknesses areabout 5.9 mm.
The second set corresponds to target thick-nesses of about 1 mm,
where the rate is about 106 /s. It isobserved that the change in
pullback velocity over this rangeis essentially the same an
increase of about 60% in spallstrength for all materials. This
suggests that loading rate
FIG. 6. Stress dependency of pullback signals for 1060 for
sample thick-nesses of nominally 5.9 mm. The measured profiles are
adjusted by thedifference between the measured profiles and the
peak velocities reported in
FIG. 5. Free surface velocity histories for a 6061-20 Al alloy
and b 6061mm.Table II so that all pullback signals are referenced
to the same value. Slightshifts in time have also been made to
clearly display the results.
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toeffects are not strongly dependent on microstructure
effects,although this observation is based on a limited set of
dataand should be further explored in future experiments.
Theobserved change is close to the predictions of the Tuler-Butcher
model,41 which suggests that spall strength varies asapproximately
0.33 for 6061-T6 this function corresponds toan 80% change in
pullback velocity in our experiments ver-sus the 60% change we
observe. In contrast to the results for6061-T6, Kanel has reported
a spall strength dependence onstrain rate of 0.21 for Al-6% Mg
alloys. We used the tech-nique proposed by Romanchenko and
Stepanov14 to investi-gate the attenuation of the pullback signal
due to propagationfrom the spall plate to the free surface. These
calculationsindicated that the observed differences between the
thin andthick samples are not strongly affected by this effect.
We also investigated separately the effect of pulse dura-tion
through comparison of experiments with different im-pactor
thicknesses, but the same sample thickness. Results oftwo
experiments for the 1060 alloy are shown in Fig. 11.Strictly
speaking, the effects of pulse duration and tensileloading rate are
not completely separated. However, it is ob-served that the pulse
duration varies by a factor of 2, while
FIG. 7. The effect of impact stress on the pullback velocity for
different
l alloy under various impact stresses. Sample thicknesses are
nominally 5.9aluminum materials. The sample thicknesses used for
this comparison are5.75.9 mm.
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
btain
023528-9 Chen et al. J. Appl. Phys. 99, 023528 2006the
free-surface velocity deceleration shows no significantdifference.
It is noted that the pullback velocities are nearlythe same. The
detailed characteristics of the pullback signalsare also similar,
except for the later time ringing. This com-parison suggests that
pullback velocity is largely independentof compressive pulse
duration for the aluminum materials westudied.
D. Shoulder in the pullback signalApart from the pullback
velocity variations with stress
and loading rate, the structure of the pullback signal from
itsminimum value was observed to systematically vary for
thedifferent materials and loading conditions studied, as
illus-trated in Fig. 2b. Specifically, the impact stress, strain
rate,and grain size were observed to influence the structure of
thepullback signal. As shown in Fig. 6, the profiles for
1060aluminum indicate a sharp change in slope from an initialfaster
rise time to a slower slope at stresses in the range of1322 GPa.
This may represent a transition from brittle to amore ductile
failure process.10 Also, the effect occurs at ap-
FIG. 8. Effects of initial yield strength effect on spall
stress. a Profiles osamples 1.0 mm in thickness at 4 GPa.FIG. 9.
Loading rate effect for 6061-20, 6061-80, 1060, and pure Al: a 5.9
mmthickness.
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toproximately the same relative time after minimum velocityfor all
stress levels. We refer to this structural transition as ashoulder
effect in subsequent discussions.
The shoulder feature has not commonly been mentionedin the
literature, but it has occasionally been observed andpartially
analyzed. In particular, it was observed in tantalum,copper, and
6061-T6 aluminum embedded with aluminaparticles.40,4246 The recent
spall data reported by Schwartz etal.29,30 on single-crystal copper
appear to be of the samenature, although spikes or overshoot in
pullback signal ex-ist at the transition of the two slopes. Our
data on 100single-crystal aluminum shown in Fig. 4c also exhibit
ve-locity overshoots at the corresponding point in the profile.
Inthe present work, we consistently observed the shoulder
phe-nomenon for systematic variations in both material proper-ties
and loading parameters.
As mentioned, the spall profile for a 1060 thick sampleat 4 GPa
is smooth, without a shoulder effect. A beginningtrace of the
feature occurs in the pullback profile at 9 GPa.However, the effect
becomes more pronounced with in-creased impact stress, as
illustrated in Fig. 6. Examination ofthe profiles for ultrapure Al
over the range of 421 GPa
ed on samples 5.9 mm in thickness at 4 GPa and b profiles
obtained onin thickness, at a stress level of 4 GPa, and b samples
of 1.0 mm in
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-10 Chen et al. J. Appl. Phys. 99, 023528 2006shows a
similar effect Fig. 4a. It is worthy to note that thetwo slopes and
even the heights of the shoulders from theminimum in particle
velocity are similar for both 1060 andultrapure polycrystalline Al
under similar loading conditions.For all stress levels, the slopes
of the profiles are parallel toeach other both before and after the
transition.
As shown in Figs. 9a and 9b, the shoulder for 1060aluminum is
enhanced as the unloading rate increases. Theshoulder transition is
not observed at the lower rate for anymaterial we studied, except
for the 100 single crystal at 4GPa, whereas at the same shock
stress, but higher loadingrate, it is observed in both 1060 and
ultrapure Al. This sug-gests a loading rate effect on the shoulder
formation; higherrates appear to result in a more prominent
transition feature.
Finally, we have preliminary evidence that the effect
isgrain-size dependent, although experiments were not specifi-cally
designed to explore this effect. For example, it isprominent in
1060 and ultrapure Al, but absent in 6061-20and barely perceivable
in 6061-80 at 14 and 22 GPa for highrates. As illustrated in Table
I, the grain sizes vary consider-ably for these materials.
Grain-size effects on the structure ofthe pullback signals for 6061
alloy and pure aluminum are
FIG. 10. Loading rate effects for 6061, 1060, and ultrapure Al
samplethicknesses are 1 mm at low strain rates and 5.9 mm at high
strain rateson pullback velocity.FIG. 12. Grain size effect on the
shoulder effect for the pullback signals: a Normprofiles for thin
samples at 4 GPa.
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toillustrated in Fig. 12a and 12b. These comparisons sug-gest that
the shoulder effect becomes more apparent as thegrain size
increases and thus that grain size has an effect onthe pullback
velocity structure.
A final speculation is that the shoulder effect could arisefrom
quasi-elastic-plastic response during recompression, asrecently
observed by Huang and Asay.34 They observed thatat about 20 GPa
strength effects appear to be independent ofmicrostructure. This
could account for the similarity of thereshock structure in the
pullback signals near 20 GPa, whereshoulder effects become
pronounced. However, the single-crystal materials show a pronounced
shoulder effect com-pared with polycrystalline materials. Reshock
experimentson single crystals would be useful to help resolve this
issue.
IV. DISCUSSION
A major result of this work is the observed increase inpullback
velocity with increased shock stress. Previous spal-lation studies
on other materials, including uranium,47 Fe,and Cu,48 illustrated a
strong stress dependence of spallstrength. For aluminum alloys,
other researchers have
FIG. 11. Pulse duration effect for 1060 at 22 GPa. The solid
line is for theimpactor thickness of about 3.2 mm; the dashed line
is for the impactorthickness of 2 mm. The fluctuations in particle
velocity are thought to resultfrom internal material
inhomogeneities.alized profiles for thick samples at 13 GPa time
shifted and b normalized
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-11 Chen et al. J. Appl. Phys. 99, 023528 2006reached
varying conclusions about stress dependence.23,24,49The general
observation has been that spall strength in alu-minum is more or
less independent of the stress amplitude.In particular, Stevens and
Tuler23 concluded that shock wavestrengthening is not significant
for the 6061-T6 aluminumalloy and 1020 steel. Kanel49 suggested
that the spallstrength of AD1 and 1100 aluminum does not depend
onpeak shock pressure. Ek and Asay24 also observed a
minimaldependence for 6061 and Al-6Mg. It is possible that the
dif-ferences observed in the present study with previously
re-ported results for aluminum are partially due to microstruc-ture
effects. Also, it is noted that previous studies did
notsystematically address the issue of stress dependence,
inde-pendent of changes in other properties such as loading ratefor
the range studied here.
Another important observation is that the spall strengthsof
several different polycrystalline aluminum materials aresimilar at
22 GPa, whereas major differences exist at 4 GPa.This suggests both
the importance of initial metallurgical ef-fects and the
possibility that these effects are overridden byshock-induced
microstructure at high stresses. In support ofthis hypothesis, Gray
and Huang26 reported a fine subgrainstructure formed in large-grain
pure aluminum at 13 GPa,which is consistent with the decreased
importance of metal-lurgical effects at high stress levels if the
new microstructureplays a role in the failure process. A possible
explanation isthat at lower shock stresses where the spall strength
is closeto the incipient value, defects such as grain boundaries
coulddominate the measurements, thus giving smaller spallstrengths
for smaller grain materials that have a larger sur-face area of
boundaries. Whereas, at high stresses, spallationthrough the grains
themselves produced by a shock-inducedmicrostructure could become
more important, resulting insimilar values of spall strength at
high stresses. Finally, it isnoted that the 111 single-crystal A1
studied here has aboutthe same pullback velocity as polycrystalline
6061-80 alumi-num at 22 GPa, which is also consistent with an
increasingrole of shock-induced microstructural changes at
higherstresses.
It is also noted that Minich et al. reported changes in
thepullback velocity versus shock stress for pure
polycrystallinecopper of different grain sizes and also for
single-crystalcopper,30 in agreement with the present observations.
Theyfound that the pullback velocity of pure copper with grainsizes
larger than 45 m and for different orientations ofsingle crystal
increased by about 50% over the stress rangeof 430 GPa, followed by
a leveling off at higher shockstresses in most cases. The 100
single-crystal copper wasobserved to have the highest spall
strength at all stress levelsfrom 6 to 45 GPa, also in agreement
with our results. Inaddition, at high stresses, the 111 copper had
the same pull-back velocity as that for other polycrystalline pure
coppersamples in the large-grain size range. However, the results
ofMinich et al. on small-grain pure copper 8 m showed anincrease in
pullback velocity over the stress range of 414GPa, and then
saturation to a value lower than that for the 45and 90 m copper at
higher stresses. These results suggest
that there may be a threshold grain size, beyond which
spallstrength becomes grain-size independent at high stresses.
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
toThis effect could be investigated with the line imagingVISAR
technique demonstrated by Chhabildas et al.,42which showed that
microstructural properties in tantalumnear incipient spall have a
pronounced effect on the spallsignal.
Previous spall measurements on single-crystal aluminumare
limited to a few specific examples,5053 so it is difficult todraw
general conclusions from these preliminary results.Kanel et al.52
reported a value of 2.3 GPa for the spallstrength of 100 aluminum
at a stress level of about 7.6GPa, which corresponds to a pullback
velocity of 300 m/s.This is 50% larger than the present results.
However, theirsamples were significantly thinner than those used
here, so adirect correlation cannot be made. However, their results
aregenerally compatible with ours, considering the effect ofsample
thickness on spall strength. Furthermore, their resultsdid not show
stress-dependent spall strength for 100 singlecrystal, as we
observe. Belak53 obtained values of pullbackvelocity ranging from
140 to 150 m/s on various orienta-tions of single-crystal aluminum
at about 1.5 GPa for samplethicknesses similar to ours. Their
results for 111 at 1.5 GPa,when combined with ours at 22 GPa,
result in a smaller slopeas compared with that for the other A1
materials.
Experiments by Stevens et al.50,51 at 1.5 GPa showedthat
tension-produced voids in A1 single crystals were octa-hedral in
shape with surfaces parallel to 111 planes. Thissuggests that the
voids nucleated mainly from dislocationnucleation and growth, since
plastic deformation occurs pre-dominantly on these planes. It is
plausible that tensile failureof a polycrystalline material would
be strongly influenced bythe strength of the weakest subcrystal
orientations, resultingin a lower strength for the composite
response. The big dif-ference observed between 100 and 111
single-crystal alu-minum at 22 GPa may be due to different
shock-inducedmicrostructures or to differences in the dislocation
genera-tion and mobility for the different slip planes activated.
Forexample, a simple geometrical analysis of uniaxial deforma-tion
using Taylor factors, M, indicates that the 111 crystal,M =3.67,
requires considerably more dislocation motion thanthe 100 crystal,
M =2.45, to achieve unit deformation.Similarly, using the ideas of
critical resolved shear stressand the Schmid factor for single
crystals indicates that the111 crystals would have a yield strength
2/3 higher thanthe 100 crystals. Based on the general agreement
observedin the pullback signals for both 100 and 111 orientationsof
aluminum and copper,30 a similar tensile failure mecha-nism is
suggested for these fcc metals. However, a system-atic study of
spallation in single-crystal aluminum is neededto fully understand
the orientation dependence and implica-tions for deformation
mechanisms. Soft recovery experi-ments would be helpful in
identifying the mechanisms op-erative for different crystalline
directions.
Currently, a theoretical basis for the shock stress depen-dence
of spall strength is lacking. It is well known that thespall
failure mechanism in ductile materials is related to thenucleation
and growth of voids, and that the fracture processis controlled by
local plastic deformation for ductile
10,54
materials. A material with a higher yield strength
shouldtherefore exhibit a stronger resistance to void growth
and
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
023528-12 Chen et al. J. Appl. Phys. 99, 023528 2006hence the
spall strength should be higher, as suggested by themodels of
Johnson,54 Grady,10 Gurson,55 and Rajendran etal.56 Since the flow
strength of aluminum increases withshock stress,34 the possible
decrease or saturation of spallstrength in the stress range of 1322
GPa for the polycrys-talline aluminum materials is not expected. A
possibility isthat strain hardening effects saturate at high
stresses, as sug-gested by the work of Huang and Asay.34 It is also
likely thatother factors, such as nucleation and growth effects
orchanges in damage mechanisms, may play a role due to achanging
microstructure, as observed in the work of Grayand Huang.26 In any
case, it is possible that the commonpullback velocities in the
different aluminum materials at 22GPa result from similar
shock-induced substructures at highstresses.
A final comment concerns the observed shoulder phe-nomenon. The
effect could be indicative of a fundamentalchange in failure
mechanism or an additional fracture modethat is initiated under
certain conditions. The phenomenon isintriguing because of its
apparent dependence on stress, load-ing rate, and possibly grain
size. Johnson et al. first ad-dressed this effect in spall
modeling41 and interpreted theslope change as a secondary spall
resistance due to theformation of a second spall plane. However, a
strong linkbetween multiple spall planes and the shoulder has not
beenestablished. Based on the work of Shockey et al.18 on
crackarrest, we postulate that the abrupt change to ductile
re-sponse could result from the resistance of fractured areas
tofurther void growth. It would be useful to systematically
in-vestigate the effect by studying a single material, such
asultrapure Al with different grain sizes. Soft recovery
experi-ments would help identify possible correlation of
fracturemorphology with the appearance of this effect.
V. CONCLUSIONS
Results of pullback velocities in plate-impact experi-ments are
reported for aluminum initially shocked over thestress range of 422
GPa. Several aluminum materials withdifferent microstructures were
studied, including three alloysof 6061 alloy, commercially pure,
and ultrapure aluminum,and two orientations of single-crystal
aluminum, 100 and111. The principal goal was to investigate the
effects ofmicrostructure e.g., grain size and impurities and
shockloading parameters i.e., impact stress and sample thicknesson
spallation behavior. A major observation is that the pull-back
velocity characterizing spallation is observed to in-crease for all
materials over the impact range of 414 GPa.The pullback signals for
polycrystalline aluminum samplesare found to differ significantly
at 4 GPa, while the pullbacksignals are similar at 22 GPa. This
observation implies thatinitial metallurgical properties are
important at low impactstresses, but relatively less so at high
levels and thus that thefailure mechanism may be induced by the
shock process it-self at high stresses. In addition, preliminary
data obtainedon 100 and 111 orientations of aluminum single
crystalssuggest a significant orientation dependence of the
tensile
failure process in aluminum
Grain-size effects on spall strength are observed at low
Downloaded 16 Feb 2006 to 134.121.73.50. Redistribution subject
tostresses, but are indistinguishable at higher stresses. How-ever,
the structure of the pullback signal appeared to changewith
different grain sizes, high stress amplitudes, and differ-ent
loading rates. The change in structure is thought to sig-nify a
transition from a brittlelike to a more ductilelike fail-ure
process. This conclusion is tentative since the grain sizewas not
systematically varied in the present experiments,with other
properties held constant. Further studies should beconducted on
polycrystalline aluminum with varying grainsizes, but with other
properties held constant to investigatethis effect.
ACKNOWLEDGMENTS
The authors would like to thank Kent Perkins, Kurt Zim-merman,
Dave MacPherson, and Nate Arganbright for assis-tance in conducting
the plate-impact experiments. Theywould also like to acknowledge
Professor Y. M. Gupta forproviding overall technical guidance and
discussions and formany technical discussions with Jim Johnson and
M. Winey.Erin Devlin of the Colorado School of Mines is
acknowl-edged for preparing the 6061 aluminum samples. Pablo
Es-cobedo and Pankaj Tvivedi of the Department of
MechanicalEngineering, Washington State University, are also
thankedfor help with EBSD characterizations. This project was
sup-ported by DOE under Grant No. DE-FG52-97SF21388.
1G. E. Duvall and G. R. Fowles, Shock Waves, in High Pressure
Physicsand Chemistry, edited by R. S. Bradley Academic, New York,
1963, p.229.
2G. R. Fowles, G. E. Duvall, J. Asay, P. Bellamy, F. Feistmann,
D. Grady,T. Michaels, and R. Mitchell, Rev. Sci. Instrum. 41, 984
1970.
3V. E. Fortov, V. P. Efremov, G. Kanel, P. Morozov, B. Demidov,
I.Lomonosov, A. Ni, and O. Yu. Vorobiev, in Shock Compression of
Con-densed Matter 1991, edited by S. C. Schmidt, R. D. Dick, J. W.
Forbes,and D. G. Tasker, Elsevier Science, New York, 1992, p.
833.
4I. Gilath, S. Eliezer, M. P. Dariel, and L. Kornblit, J. Mater.
Sci. Lett. 7,915 1988.
5V. E. Fortov, V. V. Kostin, and S. Eliezer, J. Appl. Phys. 70,
4524 1991.6C. A. Hall et al., Rev. Sci. Instrum. 72, 3587 2001.7D.
R. Curran, L. Seaman, and D. A. Shockey, Phys. Rep. 147, 253
1987.8T. Anoun, L. Seamen, D. R. Curran, G. I. Kanel, S. V.
Razorenov, and A.K. Utkin, Spall Fracture Springer, New York,
2003.
9J. S. Rinehart and J. Pearson, Behavior of Metals Under
Impulsive LoadsThe American Society for Metals, Cleveland, Ohio,
1954.
10D. E. Grady, J. Mech. Phys. Solids 36, 353 1988.11D. E. Grady
and M. E. Kipp, in High-Pressure Shock Compression of
Solids, edited by J. R. Asay and M. Shahinpoor Springer, New
York,1993, pp. 265321.
12L. W. Davison and R. A. Graham, Phys. Rep. 55, 256 1979.13M.
A. Meyers and C. T. Aimone, Prog. Mater. Sci. 29, 1 1983.14V. I.
Romanchenko and G. V. Stepanov, J. Appl. Mech. Tech. Phys. 21,
141 1980.15G. I. Kanel, S. V. Razorenov, and V. E. Fortov, J.
Appl. Mech. Tech. Phys.
25, 707 1984.16S. A. Novikov, I. I. Divnov, and A. G. Ivanov,
Phys. Met. Metallogr., 21,
608 1966.17S. Hanim and J. R. Klepaczko, Int. J. Impact Eng. 22,
649 1999.18D. A. Shockey, L. Seaman, and D. R. Curran, in
Metallurgical Effects at
High Strain Rates, edited by R. W. Rohde, B. M. Butcher, J. R.
Holland,and C. H. Karnes Plenum, New York, 1973, p. 473.
19L. Seaman, D. R. Curran, and D. A. Shockey, J. Appl. Phys. 47,
48141976.
20W. B. Jones and H. I. Dawson, in Metallurgical Effects at High
StrainRates, edited by R. W. Rohde, B. M. Butcher, J. R. Holland,
and C. H.Karnes Plenum, New York, 1973, p. 443.21D. R. Christman,
W. M. Isbell, S. G. Babcock, A. R. McMillan, and S. J.Green, DASA
Report No. 2501-3, AD735966, 1971 unpublished.
AIP license or copyright, see
http://jap.aip.org/jap/copyright.jsp
-
22S. Cochran and D. Banner, J. Appl. Phys. 48, 2729 1977.23A. L.
Stevens and F. R. Tuler, J. Appl. Phys. 42, 5665 1971.24D. R. Ek
and J. R. Asay, Shock Waves in Condensed Matter 1985, edited
by Y. M. Gupta Plenum, New York, 1986, p. 413.25Y. I.
Mescheryakov, A. K. Divakov, and N. I. Zhigacheva, Int. J.
Solids
Struct. 41, 2349 2004.26G. T. Gray III and J. C. Huang, Mater.
Sci. Eng., A 145, 21 1991.27G. T. Gray III, Acta Metall. 36, 1745
1988.28B. M. Butcher, J. Appl. Mech. 34, 209 1967.29A. D. Schwartz,
J. U. Cazamias, P. S. Fiske, and R. W. Minich, Shock
Compression of Condensed Matter 2001, edited by M. D. Furnish,
N. N.Thadhani, and Y. Horie, AIP Conf. Proc. No. 620 AIP, New York,
2002,p. 491.
30R. W. Minich, J. U. Cazamias, M. Kumar, and A. J. Schwartz,
Metall.Mater. Trans. A 35, 2663 2004.
31A. K. Zurek, Shock Compression of Condensed Matter 1989,
edited by S.C. Schmidt, J. N. Johnson, and L. W. Davidson
North-Holland, Amster-dam, 1990, p. 433.
32A. K. Zurek, P. S. Follansbee, and J. Hack, Metall. Trans. A
21A, 4311990.
33L. M. Barker and R. E. Hollenbach, J. Appl. Phys. 43, 4669
1972.34H. Huang and J. R. Asay, J. Appl. Phys. 98, 033524 2005.35J.
R. Asay, L. P. Mix, and F. C. Perry, Appl. Phys. Lett. 29, 284
1976.36V. Randle, in Electron Backscatter Diffraction in Materials
Science, edited
by A. J. Schwartz, M. Kumar, and B. L. Adams Kluwer Academic,
Dor-drecht, 2000.
37E. Devlin private communication.38H. J. Kleebee private
communication.39J. W. Swegle and D. E. Grady, J. Appl. Phys. 58,
692 1985.40J. N. Johnson, R. S. Hixson, D. L. Tonks, and A. K.
Zurek, AIP Conf.
Proc. 370, 523 1996.41F. R. Tuler and B. M. Butcher, Int. J.
Fract. Mech. 4, 431 1968.42P. D. Church, W. G. Proud, T. D.
Andrews, and B. Goldthorpe, Shock
Compression of Condensed Matter 2001, edited by M. D. Furnish,
N. N.
Thadhani, and Y. Horie, AIP Conf. Proc. No. 620 AIP, New York,
2002,p. 487.
43R. S. Hixson, J. N. Johnson, G. T. Gray III, and J. D. Price,
AIP Conf.Proc. 370, 555 1996.
44G. T. Gray III, N. K. Bourne, M. A. Zocher, P. J. Maudlin, and
J. C. F.Millett, Shock Compression of Condensed Matter 1999, edited
by M. D.Furnish, L. C. Chhabildas, and R. S. Hixson, AIP Conf.
Proc. No. 505American Institute of Physics, Melville, NY, 2000, p.
509.
45H. Nahme, Shock Compression of Condensed Matter 1999, edited
by M.D. Furnish, L. C. Chhabildas, and R. S. Hixson, AIP Conf.
Proc. No. 505American Institute of Physics, Melville, NY, 2000, p.
517.
46F. Llorca and G. Roy, Shock Compression of Condensed Matter
2003,edited by M. D. Furnish, Y. M. Gupta, and J. W. Forbes, AIP
Conf. Proc.No. 706 American Institute of Physics, Melville, New
York, 2004, p.589.
47L. C. Chhabildas and J. R. Asay, in Shock-Wave and
High-Strain-RatePhenomena in Materials, edited by M. A. Meyers, L.
E. Murr, and K. P.Staudhammer Marcel Dekker, New York, 1992, p.
947.
48V. A. Ogorodnivov, A. G. Ivanov, and E. S. Tyunkin et al.,
Fiz. GoreniyaVzryva 94 1992.
49G. I. Kanel, S. V. Razorenov, A. Bogatch, A. V. Utkin, V. E.
Fortov, andD. E. Grady, J. Appl. Phys. 79, 8310 1996.
50A. L. Stevens, L. Davison, and W. E. Warren, Proceedings of
the Confer-ence on Dynamic Crack Propagation Lehigh University,
Bethlehem,1972, p. 37.
51A. L. Stevens, L. Davison, and W. E. Warren, J. Appl. Phys.
43, 49221972.
52G. I. Kanel, S. V. Razorenov, K. Baumung, and J. Singer, J.
Appl. Phys.90, 136 2001.
53J. Belak private communication.54J. N. Johnson, J. Appl. Phys.
52, 2812 1981.55A. L. Gurson, J. Eng. Mater. Technol. 99, 2
1977.56A. M. Rajendran, M. A. Dietenberger, and D. J. Grove, J.
Appl. Phys. 65,
1521 1989.
023528-13 Chen et al. J. Appl. Phys. 99, 023528 2006Downloaded
16 Feb 2006 to 134.121.73.50. Redistribution subject to AIP license
or copyright, see http://jap.aip.org/jap/copyright.jsp