Vous êtes sur la page 1sur 9

Plate impact response of ceramics and glasses

G. F. Raiser, J. L. Wise, R. J. Clifton, D. E. Grady, and D. E. Cox


Citation: Journal of Applied Physics 75, 3862 (1994); doi: 10.1063/1.356066
View online: http://dx.doi.org/10.1063/1.356066
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/75/8?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
On the shattering of clusters by surface impact heating
J. Chem. Phys. 105, 8097 (1996); 10.1063/1.472663
Effects of geometry in pressureshear and normal plate impact recovery experiments: Threedimensional
finiteelement simulation and experimental observation
J. Appl. Phys. 80, 3267 (1996); 10.1063/1.363268
Scaling of the fingering pattern of an impacting drop
Phys. Fluids 8, 1344 (1996); 10.1063/1.868941
Transient acoustic near field in air generated by impacted plates
J. Acoust. Soc. Am. 99, 700 (1996); 10.1121/1.414646
Modeling of KippGrady plate impact experiments on ceramics using the RajendranGrove ceramic model
AIP Conf. Proc. 309, 749 (1994); 10.1063/1.46481

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

Plate impact response

of ceramics

and glasses

G. F. Raisera)
Division of Engineering, Brown University, Providence, Rhode Island 02912

J. L. Wise
Organization 1433, Sandia National Laboratories, Albuquerque, New Mexico 87185

R. J. Clifton
Division of Engineering, Brown Vniversity, Providence, Rhode Island 02912

D. E. Grady and D. E. Cox


Organization 1433, Sandia National Laboratories, Albuquerque, New Mexico 87185

(Received 3 March 1993; accepted for publication

11 January 1994)

Soft-recovery plate impact experiments have been conducted to study the evolution of damage
in polycrystalline Al,O, samples. Examination of the recovered samples by means of scanning
electron microscopy and transmission electron microscopy has revealed that microcracking
occurs along grain boundaries; the cracks appear to emanate from grain-boundary triple points.
Velocity-time profiles measured at the rear surface of the momentum trap indicate that the
compressive pulse is not fully elastic even when the maximum amplitude of the pulse is
significantly less than the Hugoniot elastic limit. Attempts to explain this seemingly anomalous
behavior are summarized. Primary attention is given to the role of the intergranular glassy phase
which arises from sintering aids and which is ultimately forced into the interfaces and voids
between the ceramic grains. Experiments are reported on the effects of grain size and glass
content on the resistance of the sample to damage during the initial compressive pulse. To
further understand the role of the glass, plate impact experiments were conducted on glass with
chemical composition comparable to that which is present in the ceramic. These experiments
were designed to gain further insight into the possibility of failure waves in glasses under
compressive loading.

I. INTRODUCTlON
The dynamic recovery plate impact experiment has
been considered an attractive method for evaluating the
brittle behavior of ceramics for two main reasons. First,
since microdamage in ceramic materials takes place at very
rapid rates, these experiments provide a means of initiating
microcracks but removing the loads before the microcracks
coalesce into large-scale cracks and rupture the material.
Second, if the plate thicknesses and geometries are chosen
correctly, it is possible to subject the central region of the
specimen to a well-known stress history and still recover it
for microscopy studies.
Several studies have utilized some of these ideas in
plate impact experiments designed to study damage mechanisms in ceramics and ceramic-related materials. Yaziv
used a double-impact technique which led him to characterize tensile damage as comprising a spa11 zone. Longy
and Cagnoux used spa11 and recovery experiments to
study how certain microstructures of alumina ceramics affect their spa11properties and their Hugoniot elastic limit
(HEL). Louro3 tested different aluminas under various
stress pulse durations and magnitudes to highlight how
porosity and grain size alter their dynamic damage properties. Stress histories in the ceramic were unfortunately
indeterminate, and post-test analyses were limited to gross
effects on recovered fragments. This is because these invesakurrent

address: Washington State University, Pullman, WA.

3882

J. Appl. Phys. 75 (a), 15 April 1994

tigations were conducted at very large stresses and impact


velocities, so reusable target holders could not be constructed to stop projectiles at impact. Consequently, flyer
plates were constantly reaccelerated by the projectile, giving rise to multiple impacts on the target.
Stopping the projectile can be accomplished if impact
velocities are reduced. Although low velocities may bring
stresses below the reported HEL values for these materials
(4-9 GPa), there are reasons to expect that some inelastic
response occurs below this elastic threshold.4 Measured
HEL values are often higher than actual values because full
decay of the precursor has not occurred at the distances
where the HEL is measured, Also, the existence of residual
stresses (as in the case of ceramics) can promote loadinduced microcracking.5-7 In fact, given that in a ceramic
with medium to large grain size certain facets will microcrack spontaneously during processing, the strength of a
ceramics elastic response most likely varies continuously
from low to high stresses due to the wide distribution of
residual stresses present in the material. For these reasons,
microdamage can be expected at stress levels below the
HEL.
Recovery
experiments
have been successfully
conducted> in an extensive study of Vistal, an CT-A1203
from Coors Porcelain Company. Velocity-time and stresstime laser interferometry data, electron microscopy pio
tures, and finite-element modeling were presented. Elastic,
three-dimensional finite-element calculations were also carried out to show that the desired stress-history profile was

0021-8979/94/75(8)/3862/8/$6.00

@ 1994 American Institute of Physics

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

imposed on the specimen.g Results of the Vista1 study


pointed to the critical role played by the intergranular
glassy phase in this material, which is present in all sintered Al,Os ceramics.
A more complete understanding is currently needed
for the compressive and tensile damage processes that occur in aluminas below the HEL. In addition, an explanation is required for the influence of grain size and, especially, the intergranular glassy phase on the damage
resistance of these ceramics. To this end, a series of aluminas of different microstructures has been subjected to comparable stress levels and stress histories. Velocity-time profiles and electron micrographs are compared and
conclusions are made regarding the importance of grain
size and glassy phase in these materials.
To understand further the role of the glass, experiments have been conducted on an aluminosilicate, i.e.,
Cornings Cl723 glass, with a chemical composition that is
similar to the one found at triple points within the Vista1
alumina.810 These experiments are intended to aid in understanding the recently reported phenomenon of failure
glass.1-*3 According
to
Brar
and
waves in
co-workers,1112 when soda lime glass is impacted above its
HEL, a wave with speed 2.2*0.2 mm/ps [less than both
the longitudinal and shear wave speeds) is propagated into
the glass from the impact face. Behind this wave the spa11
strength is reported to drop and the transverse stress is
reported to increase. They interpret this behavior as a decrease in shear strength and a comminution of the glass.
Kane1 and co-workersI used thick flyer plates to impact
K19 glass and interpreted a small step in rear-surface particle velocity as evidence of a recompression from the failed
material, and hence, evidence of a failure wave, Moreover,
they reported that failure waves are observed near or below
the HEL of K19 glass. Some questions arise from these
conclusions. For example, how can extensive cracking occur in a nonporous material under a state of uniaxial compression, where any crack would be opening against high
compressive stresses which would tend to close it? Also, if
this phenomenon is indeed a material property, then why
does it start immediately at the surface, but not inside the
material where the compressive stress behind the leading
wave front is unattenuated with distance of propagation
and is held for an extended time prior to failure? To understand the nature of a failure wave, spa11experiments
have been conducted on Cl723 aluminosilicate glass specimens with different initial surface roughnesses. These tests
were intended to probe not only the spall strength of the
glass after compression to different stress levels, but also to
investigate the possibility that the failure wave phenomenon is a surface effect in the sense that it emanates from
surface irregularities which cause nonplanar waves to develop.
II. PROCEDURE
A. Recovery

experiments

on A1203

These experiments were conducted in the Plate Impact


Facility at Brown University. A diagram of the essential
J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

FIG. 1. Star-flyer recovery experiment configuration.

elements inside the target chamber is shown in Fig. 1. A


projectile trips several velocity pins as it emerges from the
63.5 mm .gun barrel and strikes the steel anvil, which is
bolted down inside the target chamber. Dimensions of all
critical parts are set such that the star flyer (Ti6Al4V)
impacts the specimen and the projectile nosepiece (steel)
hits the anvil simultaneously. This stopping of the projectile prevents additional impacts of the flyer on the specimen. Moreover, since the flyer has a lower impedance than
the specimen, there is contact only for a time equal to the
longitudinal wave round-trip time in the flyer. The star
essentially bounces oft the specimen. The initial compressive pulse in the momentum trap reflects from its. rear
surface (becoming tensile) and returns to the weakly
bonded specimen interface, causing the momentum trap
(Hampden steel hardened to 60-62 Ro) to separate from
the specimen and fly off with the momentum that the flyer
lost. The star-shaped flyer minimizes the effects of lateral
unloading waves in a central octagonal region.g*14A short
tensile pulse arises in the specimen due to an intended gap
at the specimen-momentum trap interface. This gap can be
increased by sputtering a thin (several micrometers) layer
of material onto the four corners of the momentum trap
prior to bonding it to the specimen. Moreover, it can be
eliminated by careful lapping and assembly of the two
plates. The compressive pulse will reflect as tension for the
time required to close the gap (usually 5-60 ns). This
tensile pulse reflects from the impact face as a compressive
pulse and travels into the momentum trap prior to separation. The specimen is left at rest inside the aluminum
holder and taken out for microscopy analysis. The Lagrangian t-X diagram in Fig. 2 shows how these waves
traverse the plates and indicates the sequential order in
which they arrive at the rear surface of the momentum
trap. The laser interferometer system monitors these plane
waves at this surface, at four separate points within the
plates central octagonal region, giving redundant, nearly
identical, velocity-time and derived stress-time data.
Each plate is lapped and polished to measured flatnesses within one wavelength of a monochromatic light
Raiser et al.

3863

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

Beam

Second Pulse

M2
Photodiodes

rl3lT
Lens

To Monitoring
FIG. 2. Lagrangian

f-X

diagram for the recovery experiment.


FIG. 3. Interferometer
locations.

source (a = 587.6 mn), and subsequently cleaned ultrasonically. The ceramic specimens are prepared for vapor deposition in a solution of hydrogen peroxide and sulfuric
acid.8 An aluminum layer < 0.1 ym thick is then vapor
deposited in the pattern of four tabs surrounding a central
cross, each tab being electrically isolated on the nonconducting alumina specimen. These conductive tabs and the
cross provide a means of measuring tilt as well as a reflective surface for pm-impact alignment of the star flyer and
specimen. The projectile is assembled and a Delrin key is
inserted in a piston at the back of the projectile. This key
fits into a slot in the gun barrel and prevents rotation of the
projectile during its acceleration toward the target. The
specimen and momentum trap are assembled with four
small drops of epoxy between them at the corners. For
production of extensive tensile damage a sputtered layer of
material several micrometers thick is deposited on the four
corners of the momentum trap prior to assembly to provide
a prescribed gap between the plates. The assembly is then
bonded inside the aluminum holder at the precise location
to make projectile-anvil
impact coincide with flyerspecimen impact. The disposable brass nose, which protects the anvil from damage during numerous impacts, is
then applied with epoxy. Tilt wires are attached outside the
impact area to each of the vapor deposited tabs and to the
central cross on the specimen. These wires are fed into a
logic box and the output is monitored on an oscilloscope
used for tilt measurement. Tilts of less than 1 x 10e3 rad
are routine. The target assembly is aligned with the star
flyer using a technique developed by Kumar and Clifton,15
which renders respective faces parallel to an accuracy of
2.0~ 10m5 rad. After alignment the projectile is pulled
back to the breech end of the gun barrel and a velocity
system similar to the one described by Fowles et al. l6 is
installed at the muzzle. The accuracy of this system is
better than 1%. A disposable mirror is suspended behind
the momentum trap and the laser system is set up to monitor surface motion caused by the stress waves in the experiment. For these experiments, all interferometer data is
confined to a 4 mm square region in the center of the
3864

Surface

J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

setup for monitoring

motion at four separate

momentum trap (well within the central octagonal


region14). The interferometer system is based on a scheme
by Mello, Prakash, and Clifton to monitor four points
simultaneously during the experiment. Figure 3 shows the
general features of this system. It consists of four separate
Michelson inteferometers where the four photodiodes measure experimental intensity variations (one peak-to-peak
variation in light intensity corresponds to a normal displacement of n/2) which are stored on LeCroy wave-form
digitizers. Details of the digital data processing have been
given by Tong.18 The application of this data reduction
procedure yields high-resolution velocity-time graphs with
approximately 1500 points (0.742 ns per point) for these
recovery experiments. The elastic nature of the momentum
trap ensures that the measured particle velocity is linearly
related to the stress at the back of the specimen, thereby
allowing a display of the interface stress on the secondary
y axis of all plots. After the laser is set up, the gun is
pumped down to approximately 50x 10e3 Torr and the
digitizers are set to record data. After the shot, the LeCroy
data is reduced and the ceramic specimen is prepared for
microscopic analysis. The specimens are cut using a lowspeed diamond saw, then highly polished and finally thermochemically etched. The microscope used is a JEOL
JSM-840F scanning electron microscope (SEM) which
records operating conditions on every picture. All pictures
are taken within the central octagonal region of the specimen where the stress history is well known. Knowledge of
the stress history allows certain ranges of distance through
the thickness to be identified as tension dominated or compression dominated. These are indicated in the figure captions.
B. Spall experiments

on glass

These experiments were conducted on the single-stage


powder gun at Sandia National Laboratories. The test configuration is shown in Fig. 4. The projectile has a 6061-T6
Raiser et al.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

NosepieceIlmpactor

FIG. 4. Spa11experiment configuration.

aluminum impactor plate mounted on its forward face, and


the supporting nosepiece has a counterbore in the center,
making the rear face of the impactor a free surface. Upon
impact, compressive stress states are induced in the impactor and target plates. Wave reflections from free surfaces
interact to form a tensile stress state inside the glass specimen. The location where this tension first occurs can be
controlled by varying plate thicknesses. The Lagrangian
t-X diagram of Fig. 5 maps the stress state histories at
different locations within the impacting plates. In region 2
the stress state is compressive while in region 5 the stress
state first becomes tensile in the glass specimen. The particle velocities of states, 0, 3, and 6 are monitored using
VISAR~ interferometry. If there is no failure wave and the
magnitude of the tensile stress in region 5 is less than the
spa11strength of the glass, then the particle velocity will
drop nearly to zero in region 6. The dashed line (representing a failure wave) marks the boundary between failed
material (on the left-hand side) and unfailed material (on
theright-hand side). In this figure state 5 first occurs to the
left-hand side of the failure wave. If the glass has failed in
this region, it will have no tensile strength and a spa11
plane will be created. Left-hand-going waves that reflect

FIG. 5. Lagrangian t-X diagram for the spa11experiment.


J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

from this surface will be weak, and will return to the glass
rear surface through region 6 causing little or no reduction
in free surface particle velocity. The dotted line maps the
path of a reported recompression wavei1-13 arising from
the release wave interaction with the failure wave. The
occurrence of such a wave is explained on the basis of the
failed material having a lower impedance than the unfailed
material.
The 6061-T6 aluminum flyer plate is lapped and polished on both sides to a roughness of 0.02 pm rms. The
Corning aluminosilicate glass is cut, lapped, and polished
by Precision Glass Products Co. of Oreland, PA. Its flatness is measured as in Sec. II A and is better than four
rings. The 60-40 polish has an average roughness of 0.04
pm rms. A thin ( -500 nm) coating of aluminum is applied to the rear surface of the glass to give it the necessary
reflectivity for laserinterferometry measurements. In shots
requiring a rough glass impact face, the glass is lapped
for 0.5 h using 15 pm B&! powder, giving a surface roughness of 0.52 pm rms. An epoxy bond around the targetplate periphery holds the glass concentricially within a
standard aluminum support ring. The impact surface of
the glass is flush with the front face of this ring. Impact
velocity and tilt are measured using coaxial shorting pins
mounted in this ring. The particle velocity history is measured using a VISAR.~ The fringe constants are 0.4028
mm/,&fringe
for shots GLASS1 and GLASS2 and 0.1988
mm/,&fringe
for shots GLASS3 and GLASS4. The two
quadrature records and the beam intensity variation are
recorded on LeCroy digitizers (previously described).
Data reduction is carried out using VISARSS, a program
developed at Sandia.2*

Ill. RESULTS AND DISCUSSION


A. Recovery

experiments

To check the experimental approach, two shots were


conducted using high-strength metal specimens that remained elastic throughout the loading history. Figure 6
shows the data recorded from one of these shots (92-08))
where the flyer is Ti6A14V (acoustic impedance pc, =27.7
GPa ps/mm) and the specimen and momentum trap were
Hampden steel (pcl=46.24 GPa ps/mm) hardened to 62
Rc . The other elastic shot has the same features and therefore is not presented. This figure and all subsequent figures
display two velocity-time curves. One is the measured experimental profile (solid line), and the other is the profile
predicted for planar, elastic; longitudinal waves based on
the experimental impact velocity (dashed line). Although
four points are monitored during all tests, no significant
differences are observed between them, confirming the
one-dimensional nature of the deformation in the central
region of the plates. For this reason, and in the interest of
graph clarity, only one representative data record is presented for a given shot. As shown in shot 92-08, the intended stress history is imposed to reasonable accuracy.
This observation confirms earlier three-dimensional finite
element analyses by Espinosa et al9
Raiser et a/.

3865

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

SHOT 92-08
0.07

Exp. Data

SHOT 89-05
- - - Elastic Prediction

0.07

1-1 ElasticP

1500 T

1500

0.06

0.06

0.05
loo0

0.04
0.03
0.02

IWO

0.04
0.05

0.03
0.02

500

0.01

8
500

0.01

0.00
-250

260

500

750 1000 1250 iSo%

Tlms[me]

Time [nsoc]

SHOT 92-l 1
m,....,--..,-...,-.-.,.~..,--.*,

SHOT 92-09
0.06~.,....,....,....,....,....,...~
-

0.07

0.05
0.06

f7

0.03
0.04

0.02

- - - Elastic Prediction

.Data

250

---

500

ElasticF

750

lool

Time [nut]

Time [nsoc]

FIG. 6. Experimental particle velocity vs time data for shots 92-08 (elastic shot: steel specimen), 89-05 (Coors Vista1 specimen), 92-09 (Coors AD-995
specimen), and 92-11 (Coors AD-999 specimen).

Results from tests run on three different sintered aluminas from Coors Porcelain Co. of Boulder, CO, are also
presented in Fig. 6. Test parameters and ceramic material
properties are summarized in Table I. All shots were conducted at nearly the same stress level, using the same star
flyer and momentum trap materials (Ti6Al4V and Hampden steel, respectively). Comparisons between certain
shots where specimens differ primarily in only one microstructural feature allow conclusions to be made about that
features effect.
Shots 89-05 and 92-09 were performed using Coors
Vista1 and Coors AD-995, respectively. As shown in Table
I, AD-995 has an average grain size similar to Vista1 (17
pm for AD-995,20 pm for Vistal), but more intergranular

glassy phase as indicated by the smaller preprocessing


wt % AlLO,. As indicated in Fig. 6, there is an improvement in compressive response for AD-995 as less attenuation of the first pulse occurs. The additional glass in AD995 appears to have reduced the intergranular residual
stresses, requiring higher applied stresses to initiate damage. The lower residual stresses are attributed to the flow of
glassy phase during processing to accommodate thermalexpansion anisotropy between grains.23 Shot 92-09 shows a
large pull-back signal after the initial compressive pulse
and severe attenuation of the middle (second) pulse, indicating that this material has experienced considerable damage in tension. A previously reported shot on Vistal, with
a tensile pulse of comparable duration, shows a sustained

TABLE I. Summary of recovery experiments on ceramics.

Shot
No.

Impact
stress
WI%)

89-05
92-09
92-l 1
92-14

1291
1372
1473
1476

Material

Specific
gravity

Average
grain size
(pm)

wt % Al,O,
(preprocessing)

Longitudinal
wave speed
(mm/w)

Porosity
(%b)

Coors Vista1
Coors AD-995
Coors AD-999
Browna HP Al,O,

3.99
3.89
3.96
3.96

20
17
3
2.4

99.9
99.5
99.9
99.99

10.8
10.45
10.2
10.9

0.0
2.3
0.7
0.8

Based on adaptation of process described by Staehler and co-workers (Ref. 22).


3866

J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

Raiser et a/.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

SHOT 92-14
XT
3
2
!i.a

are believedto be responsiblefor the strong compressive

0.06
0.07

1500

z
z

0.05
0.04

1000

0
R

8yg

0.03
0.02

500

P(D

E
tj
z

0.01

0.00
-250

250

500

750

1000

1250

150;

Time [nsec]

FIG. 7. Experimental particle velocity vs time data for shot 92-14 (hotpressed alumina).

tensile stress of approximately 400 MPa, indicating that


Vista1 is superior to AD-995 in this regard. The increase in
glassy phase appears to have made the AD-995 alumina
weaker in tension by decreasing the tensile strength of the
grain boundaries.
Shot 92-11 was conducted on Coors AD-999. This material has nearly the same amount of glassy phase as Vista&
but a much smaller average grain size (3 ,um for AD-999,
20 pm for Vistal). Comparison of this shot with 89-05
shows that the reduction in grain size results in a significant improvement in compressive behavior, as the first
pulse is attenuated less than for any of the commercial
aluminas tested. This behavior is consistent with the expected reduction in residual stresses with decreasing grain
size.$ The second pulse in these two shots reveals that the
tensile strengthening is nearly insignificant, as expected for
aluminas with similar amounts of glassy phase.
These experiments demonstrate that the desired microstructure for an alumina under dynamic loading at room
temperature is one with a small grain size (improving compressive response) and a minimum of preprocessing additives (improving tensile response). To test these conclusions, a high-purity,
small-grain-sized
alumina was
processed in-house. The processing procedure was an adaptation of the procedure introduced by Staehler, Predebon, and Pletka.22 A vacuum hot press was performed
on HPA-0.5AF 99.99%purity Al,O, with 50% 0.48 ,um
particle distribution (obtained from Ceralox Corp. of Tucson, AZ) with the following conditions: 1400 C, under
axial pressure of 34.5 MPa for 2.5 h. The heating and
cooling cycle temperatures were ramped at 5-10 C/mm
and pressure was maintained from 1400 C until the end of
the cooling cycle. The final density was measured as 3.952
g/cm3 (99.2% of theoretical). Shot 92-14 (Fig. 7) is a
successful experiment on this material. The compressive
pulse was attenuated less than for any of the commercial
ceramics tested. The grain size of this material was found
to be 2.4 pm by the line intercept method on a typical SEM
picture. The small grain size and reduced residual stresses
J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

response. The tensile pulse rises nearly to its elastic level of


1526 MPa. This marked improvement in tensile strength
appears to result from the higher tensile strength of the
grain boundaries. More grains are bonded directly to each
other, eliminating many of the facets with weak, glassyphase interfaces.
B. Microscope

examination

Representative SEM micrographs made from cross


sections of the central octagonal region of the recovered
samples are presented in Fig. 8. The impact direction in all
cases is left- to right-hand side. Figures 8 (a) and 8 (b)
show typical
compression-dominated
and tensiondominated regions of the Coors AD-995 specimen used in
shot 92-09. This alumina has a grain size similar to that of
Vistal, but a lower preprocessing wt % of Al203 and consequently more intergranular glass. Compared with Vistal,
this specimen was slightly better in compression, but much
weaker in tension. The micrographs bear this out as the
compression-dominated region [Fig. 8 (a)] shows some
triple-point porosity, but relatively little microcracking,
whereas the tension-dominated region [Fig. 8 (b)] shows a
clear spallation with a large intergranular crack and additional microcracking in the surrounding area. Figures 8 (c)
and S(d) show typical compression-dominated
and
tension-dominated regions for shot 93-01 in which the
specimen was hot pressed from Al,O, powder without the
addition of glass. The lack of visible microcracks in either
the compression-dominated or tension-dominated regions
is consistent with the essentially elastic response recorded
for this specimen.
C. Spall experiments
The spa11experiments are summarized in Table II and
experimental data is presented in Fig. 9. GLASS1 and
GLASS2 were conducted at compressive shock stresses in
the range 7.5-7.9 GPa, while compressive stresses for
GLASS3 and GLASS4 were in the range 3.3-3.5 GPa. The
two higher-amplitude curves in Fig. 9 are from GLASS1
and GLASS2 and they show no unloading at the expected
time ( - 1.9-2.0 ys). This is a clear indication that the
glass has lost its spa11strength in the region where tension
first develops. The two lower-amplitude curves in Fig. 9
are from GLASS3 and GLASS4, and those curves here
show an almost complete drop in particle velocity at the
estimated unloading time. This result shows that in these
shots (3 and 4) the glass has a spa11strength of at least 3.5
GPa.
These shots indicate that when the glass is subjected to
a sufficiently large compressive stress it loses its spa11
strength in the region where the reported failure wave has
already passed. If the glass undergoes a sufficiently small
initial compressive stress then the spa11strength is retained.
These observations are consistent with the published results of previous failure wave experiments by Brar and
co-workers11f2 where this loss in spa11strength is reported
at elevated initial shock stress levels.
Raiser et a/.

3867

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

PI

(4

FIG. 8. SEM micrographs of (a) compression-dominated region in Coors AD-995; (b) tension-dominated region in Coors AD-995; (c) compressiondominated region in hot-pressed alumina; (d) tension-dominated region in hot-pressed alumina.

There was no noticeable change in particle velocity in


any of the records at the expected failure wave-induced
recompression time; however, for GLASS1 and GLASS2
there appears to be some small velocity increase after this
time and before the estimated unloading wave arrival. This
could be evidence of a recompression from a slowermoving failure wave, but since this effect is so small, no
definite conclusions can be drawn. Finally, since no significant differences are observed between shots with speci-

SUMMARY OF GLASS EXPERIMENTS


1.50
?? 1.25
8

. . . . . .......(..~~.
--Smooth - 0.04~ rms
- - - -Rough - 0.52~ rms

Estimated Recompression
EstimatedUnloading

2 1.00
.E.
*is8 0-. 75
z
a 0.50
g
g

TABLE IL Summary of spall experiments on glass.

Shot
No.

Impact
velocity
b-44

Shock
stress
(GW

Impact surface
roughness
(8, ms)

Nonzero
spa11
strength

GLASS 1
GLASS2
GLASS3
GLASS4

0.961
0.965
0.450
0.443

7.88
7.51
3.42
3.33

100
5300
700
5100

no
no
yes
yes

3868

J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

q.25
0.00
0.0

0.5

1.0

1.5

2.0

2.5

3.0

Time(psec)
FIG. 4. Experimental particle velocity vs time data for spall experiments
on glass.
Raiser et

a/.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

mens of different impact surface roughnesses, no evidence


of a surface effect on failure waves has been obtained.
IV. CONCLUSIONS
By adhering to the requirements to achieve a reliable
experimental design for star flyer recovery experiments
(explained in previous studies by Kumar and Clifton,14
and Espinosa et al. 9), and proven in the elastic shots of this
investigation, the identification of dominant microstructural failure mechanisms in A&O, specimens has been accomplished. The experiment has made it possible to isolate
and discuss compression-dominated damage processes and
improvements as well as tensile damage processes and improvements. In compression, grain size reduction lowers
average residual stresses at triple junctions and grain
boundaries and makes the material less susceptible to
triple-point and intergranular flow, sliding, and microcracking. Reducing the amount of glassy phase (corresponding to a higher percentage of pure preprocessing
powder) makes tensile damage less likely by improving
grain-boundary strength. These conclusions were tested on
a high-purity,
small-grain-sized
alumina, processed
through hot pressing. The dynamic compressive and tensile
properties were found to be superior to those of all other
specimens tested.
Future experiments on ceramics will focus on the processing and testing of an alumina with an ultrafine grain
size. Recent results in lowering the costs and improving the
efficiency of preparing high-purity, nanometer-sized powders has brought attention to the possibility of large-scale
production of nanophase materials.2526 Current reports
on other types of nanophase ceramics indicate that these
materials have potential for easy, near-net-shape forming
and grain bonding without the addition of sintering
aids.2*28 The development of a nanophase alumina is expected to exhibit the best possible properties of this ceramic.
The spa11experiments on an aluminosilicate glass support the evidence of some shock-stress-dependent failure
phenomenon. At high enough compressive stresses (e.g.,
7.5-7.9 GPa for these tests) the glass loses its spa11
strength in a region where the postulated failure wave has
already passed. When the stress is reduced to 3.3-3.5 GPa,
there is a retention of spa11strength in the same area. Evidence of a recompression, indicating impedance mismatch
across the failure wave boundary, is uncertain-a
small
increase in free-surface velocity appears to occur, but at a
later time than expected. Finally, there is no evidence to
suggest that a failure wave arises from impact surface irregularities of the glass.
ACKNOWLEDGMENTS
For the ceramic work, the authors would like to thank
Professor S. Suresh and Professor B. Sheldon for helpful
discussions on damage mechanisms and processing details

J. Appl. Phys., Vol. 75, No. 8, 15 April 1994

related to ceramics. In addition, thanks are due to C. Bull,


M. Mello, B. Dean, H. Stanton, T. Kirst, and K. Markert.
Free samples of Al,O, were provided by D. Ranney of
Coors Porcelain Co. and complimentary alumina powder
was supplied by D. Bussell of Ceralox Corp. This ceramic
work and Brown Universitys contribution to the glass research were supported by the NSF-MRG at Brown University entitled Micro-Mechanics
of Failure-Resistant
Materials. For the glass research, P. L. Stanton and J. R.
Asay are gratefully acknowledged for their support of the
experiments at Sandia National Laboratories. Also, thanks
are due to J. H. Gieske, E. D. Apodaca, and J. A. Moya for
materials characterization and plate preparation. Portions
of this work performed at Sandia National Laboratories
were supported by the U. S. Department of Energy under
Contract No. DE-AC04-76DP00789.

D. Yaziv, Ph.D. thesis, University of Dayton, 1985.


F. Longy and J. Cagnoux, J. Am. Ceram. Sot. 72, 971 (1989).
L. H. L. Louro, Ph.D. thesis, New Mexico Institute of Mining and
Technology, 1988.
4H D. Espinosa, G. Raiser, R. J. Clifton, and M. Ortiz, J. Hard Mater.
3;285 (1992).
Y. Fu and A. G. Evans, Acta. Metall. 33, 1515 ( 1985).
6A. G. Evans, Acta Metall. 26, 1845 (1978).
S. Suresh and J. R. Brockenbrough, Acta Metall. 36, 1455 (1988).
sG. Raiser, R. J. Clifton, and M. Ortiz, Mech. Mater. 10, 43 (1990).
H. D. Espinosa, G. Raiser, R. J. Clifton, and M. Or&, J. Appl. Phys.
72, 3451 (1992).
OS..M. Wiederhorn, R. J. Hockey, R. F. Krause, Jr., and K. Jakus, J.
Mater. Sci. 21, 810 (1986).
N. S. Brar, S. J. Bless, and 2. Rosenberg, Appl. Phys. Lett. 59, 3396
(1991).
*N S. Brar, Z. Rosenberg, and S. J. Bless, J. Phys. (Paris) IV 1, 639
(1991).
G . I . Kanel, S. V. Rasorenov, and V. E. Fortov, in Shock Compression
of Condensed Matter-1991, edited by S. C Schmidt, R. D. Dick, J. W.
Forbes, and D. G. Tasker (North-Holland, Amsterdam, 1992), pp.
45 l-454.
14P Kumar and R. J. Clifton, J. Appl. Phys. 48, 4850 (1977).
15P: Kumar and R. J. Clifton, J. Appl. Phys. 48, 1366 (1977).
l6 G. R. Fowles, G. E. Duvall, J. R. Asay, P. Bellamy, F. Feistmann, D.
Grady, T. Michaels, and R. Mitchell, Rev. Sci. Instrum. 41, 984
(1970).
M. Mello, V. Prakash, and R. J. Clifton, in Shock Compression of Condensed Mafter-1991, edited by S. C. Schmidt, R. D. Dick, J. W. Forbes,
and D. G. Tasker (North-Holland, Amsterdam, 1992), pp. 763-766.
s W. Tong, Ph.D. thesis, Brown University, 1991.
191. 0. Owate and R. Freer, J. Am. Ceram. Sot. 75, 1266 (1992).
L. M. Barker and R. E. Hollenbach, J. Appl. Phys. 43, 4669 (1972).
L. M. Barker, Sandia National Laboratories Report SAND88-2788,
1988.
r2J. Staehler, W. Predebon, and B. Pletka, in Proceedings of the 12th
Army Symposium on Solid Mechanics, edited by S. C. Chou, 1991, p.
469.
3 J. S. Reed, Introduction to the Principles of Ceramic Processing (Wiley,
New York, 1988), p. 469.
24J. E. Blendell and R. L. Coble, J. Am. Ceram. Sot. 65, 174 (1982).
*T. Beardsley, Sci. Am. 267, 114 (1992).
r6J. Guo, J. Solid State Chem. 96, 108 (1992).
M. J. Mayo, R. W. Siegel, A. Narayanasamy, and W. D. Nix, J. Mater.
Res. 5, 1073 (1990).
*M J Mayo, R. W. Siegel, Y. X. Liao, and W. D. Nix, J. Mater. Res. 7,
973 (1992).

Raiser et a/.

3869

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
103.37.201.8 On: Mon, 01 Feb 2016 13:59:41

Vous aimerez peut-être aussi