Sei sulla pagina 1di 16

Shock-induced chemical reactions in titaniumsilicon powder mixtures of different morphologies: Time-resolved pressure measurements and materials analysis

N. N. Thadhania)
School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0245

R. A. Graham
The Tome Group, Tome, New Mexico 87031

T. Royal
School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0245

E. Dunbar
Department of Metallurgical and Materials Engineering, New Mexico Tech, Socorro, New Mexico 87801

M. U. Anderson and G. T. Holman


Sandia National Laboratories, Department 1152, Albuquerque, New Mexico 87185-1421

Received 4 November 1996; accepted for publication 23 April 1997 The response of porous titanium Ti and silicon Si powder mixtures with small, medium, and coarse particle morphologies is studied under high-pressure shock loading, employing postshock materials analysis as well as nanosecond, time-resolved pressure measurements. The objective of the work was to provide an experimental basis for development of models describing shock-induced solid-state chemistry. The time-resolved measurements of stress pulses obtained with piezoelectric polymer poly-vinyl-di-ouride pressure gauges provided extraordinary sensitivity for determination of rate-dependent shock processes. Both techniques showed clear evidence for shock-induced chemical reactions in medium-morphology powders, while ne and coarse powders showed no evidence for reaction. It was observed that the medium-morphology mixtures experience simultaneous plastic deformation of both Ti and Si particles. Fine morphology powders show particle agglomeration, while coarse Si powders undergo extensive fracture and entrapment within the plastically deformed Ti; such processes decrease the propensity for initiation of shock-induced reactions. The change of deformation mode between fracture and plastic deformation in Si powders of different morphologies is a particularly critical observation. Such a behavior reveals the overriding inuence of the shock-induced, viscoplastic deformation and fracture response, which controls the mechanochemical nature of shock-induced solid-state chemistry. The present work in conjunction with our prior studies, demonstrates that the initiation of chemical reactions in shock compression of powders is controlled by solid-state mechanochemical processes, and cannot be qualitatively or quantitatively described by thermochemical models. 1997 American Institute of Physics. S0021-8979 97 03215-5

I. INTRODUCTION

Materials synthesis and processing under conditions of high-pressure shock-compression loading of powder mixtures has become a subject of increasing attention. Of particular interest and importance is the use of strongly nonequilibrium processes. As the conditions encountered in the shock process are not achieved in any other environment, the process has the potential of yielding novel compounds, metastable phases, and uniquely modied microstructures.17 It is to be expected from elementary considerations6 that the unusual combination of high pressure and rapid loading rates, which produce large plastic deformation and concomitant high pressure in powders can lead to shock-initiated chemical reactions by mechanisms different from those encountered in conventional processes. Unfortunately, there is little
a

Electronic mail: naresh.thadhani@mse.gatech.edu

scientic knowledge of the processes of shock deformation in porous solids with high levels of porosity. The present work attempts to address these fundamental questions with a systematic study of shock compression of titaniumsilicon TiSi powder mixtures under controlled shock loading. Both, nanosecond time-resolved pressure measurements and materials analysis of samples preserved for postshock, microstructural characterization are used in the investigation. Past work has established that the fundamental mechanisms controlling chemical reactions in powder mixtures and leading to synthesis of compounds are dominated by processes occurring during the few to more than a hundred nanosecond stresspulse rise time and the microsecond duration of the peak pressure state.3,4 The critical processes include particle congurational changes principally caused by plastic deformation less often, fracture or comminution , mixing of the reactant powders within and around the voids by plastic ow and dispersion of fragments, enhanced solid 1997 American Institute of Physics 1113

J. Appl. Phys. 82 (3), 1 August 1997

0021-8979/97/82(3)/1113/16/$10.00

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

state reactivity of powders, as well as the usual temperature increases.6 Under such conditions, chemical reactions have been observed to occur in powder mixtures during the highpressure shock state before unloading to the ambient pressure in time scales of mechanical equilibration.816 Chemical reactions can also occur at times after the sample is in the ambient, postshock state. Such reactions occurring in essentially the shock-modied activated material, due to bulk residual shocktemperature increases in time scales of thermal equilibration, involve mechanisms similar to more conventional combustion-type processes,17 but are expected to be dominated by shock-activated solid-state diffusion and thermal quenching. While these two types of chemical reactions, dened, respectively, as shock-induced and shock-assisted chemical reactions,7 have been distinguished on the basis of the time period over which they occur, it is difcult to infer from samples preserved for postshock analysis recovered samples alone, whether the observed reactions occurred during the high-pressure loading and unloading conditions, or if they occurred after unloading to ambient conditions. Furthermore, the highly exothermic nature of these reactions leads to rapid temperature increases, often melting the reaction products. Thus, recovered samples show reaction products with microstructures typical of melted and solidied materials, which mask the characteristic features of the shock-compressed powders at the time the reaction is initiated. The objective of the work reported in the present paper is to experimentally investigate the mechanics of the deformation processes leading to shock-induced chemical reactions. The work involves direct determination of the rates of deformation and chemical reaction in a systematic study of titanium and silicon powder mixtures in a highly porous about 53% of solid density state, using both nanosecond time-resolved pressure measurements and microstructural analysis of recovered samples. Further, in the present work, the effects of morphological characteristics of powders on shock deformation and reaction is extensively investigated with both procedures. The materials system under investigation is of considerable interest in that the product of the highly exothermic chemical reaction, Ti5Si3 is an alloy of considerable technological importance. The starting materials, titanium and silicon, have been investigated in prior studies1820 and represent, respectively, a metal that would be expected to undergo plastic deformation, and a semiconductor that normally deforms in a brittle fashion. The present work builds directly upon prior work on shock compression of titanium and silicon powder mixtures,20 in which the authors showed the relatively lowpressure initiation threshold for onset of shock-induced reaction based upon evaluation of the microstructure of recovered samples. Further, the authors showed that the threshold was strongly dependent on particle morphology size , and initial packing density of the samples. In the present paper, we will rst provide a brief background on the mechanics of shock compression of powders in both inert and chemically reacting states. Next, the experimental procedures and results will be described and discussed. The paper will then draw
1114 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

broad conclusions regarding the fundamental aspects of the deformation and chemical reaction processes, based on both the present work and our prior investigations.
II. SHOCK COMPRESSION MECHANICS OF INERT AND REACTIVE POWDERS

Shock mechanics of the time-dependent compression of highly porous solids powder compacts or distended solids with porosities of tens of percent cause wave propagation features, which are qualitatively different from the same solid in the fully dense state. The sample responses revealed by shock-pressure measurements are completely dominated by the porous state and have the macroscopic appearance of a radically different class of materials, which indeed they are. It is to be expected that it is these deformation features, which lead to the observed chemical reactions in the shock state.6 The overall features of deformation processes associated with highly porous solids in which voids are not isolated in a solid matrix but are interacting in the deformation process , can be considered from a thermodynamic equilibrium viewpoint to identify rst-order features of the wave mechanics. Figure 1 a shows an idealized shock pressure versus volume curve for a porous solid of 50% theoretical density, and a vertical line corresponding to the dense solid. In the case of the porous solid, the application of increasing shock pressure leads to large compression at low pressure as the particles deform into voids. At a critical pressure dened crush strength , the observed compression approaches the solid density and the sample becomes fully dense. With increasing pressure, the compression follows the solid density behavior offset by a small expansion due to shockcompression heating. The rst-order consequences of such a behavior in terms of the expected shock velocities of waves propagating through a sample, can be directly inferred from equilibrium mechanics of shock propagation, as shown in Fig. 1 b . The slopes of the Rayleigh lines connecting the initial state with a shock-pressure state for both a material with a zerocrush strength and a nite-crush strength, represent equilibrium wave velocities; higher slopes represent higher wave velocities. In both cases, as pressure is increased, the wave velocity is expected to increase. As indicated in Fig. 1 c for zero strength, however, the wave velocities will be less than the nite strength case for the same pressure. The wave velocity versus pressure behavior expected from the zerocrush-strength response can be accurately estimated as shown by the dotted line in Fig. 1 c . Corresponding to the crush strength shown in Fig. 1 b , the lled circles describe a typical wave velocity versus pressure behavior for a material with nite strength. Thus, wave velocities higher than those predicted for zero strength are expected at pressures below the crush strength, and approximately the same values at pressures higher than the crush strength. The essential, rstorder differences between the behavior of highly porous solids and fully dense solids are well illustrated by comparing the wave velocities shown schematically in Fig. 1 d . The extraordinarily low wave velocities at low pressure and the large inuence of pressure on wave velocity, dominate the behavior of the porous solid, in contrast to a dense material.
Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

FIG. 1. Schematics illustrating a idealized shock pressure vs volume curve for porous solid of about 50% theoretical density, contrasted with that of solid in the fully dense state; b shock pressure vs volume curve illustrating the expected equilibrium wave velocities depicted by slopes of Rayleigh lines connecting the initial state with shock-pressure states of the porous material with zero-crush strength and nite-crush strength; c Wave velocity vs pressure behavior expected from zero-crush-strength response dashed line contrasted with that for a porous material with nite strength points , deduced from slopes of Rayleigh lines in b ; d rst-order differences between the behavior of a highly porous solid and a fully dense solid, indicated by extraordinarily slow wave velocities at low pressure and large dependence of wave velocity on shock pressure dominating the behavior of a porous solid.

The rst-order consequences of a shock-induced chemical reaction occurring during the rise time of a propagating stress pulse, can be considered as shown in the pressure volume plot in Fig. 2 a . Consistent with the model based on reaction under conditions of constant pressure as previously proposed by Graham and co-workers21 and evaluated by Bennett and Horie,22 the exothermic energy of reaction will cause an expansion to larger volumes. As indicated in Fig. 2 a , the extent of the volumetric effect for the constant pressure model is expected to vary widely from an essentially zero exothermic contribution from a zincferrite reaction,23 to a modest effect from a nickelaluminide reaction,24 to a large effect in the present titaniumsilicon system. Such a pressurevolume curve in which exothermic energy transformation is detectable in the shock response has been called a ballotechnic curve as opposed to a conventional inert Hugoniot curve.21 From the Rayleigh lines of Fig. 2 a , the wave velocity behavior in a chemically reacting powder mixture can be calculated in the equilibrium condition as shown in Fig. 2 b . It can be seen that at low pressure without reaction, higher wave velocities are anticipated until the crush strength is exceeded. At pressures exceeding the crush strength, the obJ. Appl. Phys., Vol. 82, No. 3, 1 August 1997

served behavior will show higher wave speeds than predicted for the zero-strength model if chemical reaction occurs. It is to be emphasized that if chemically driven high wave velocities are observed experimentally, the rate of energy input to the wave must be short compared to the rise time of the wave. Thus, identication of the chemically driven high wave speeds in highly porous powder mixtures is direct evidence for shock-induced chemical reaction occurring on time scales of 100 ns or less. Much of the prior work on mechanics and physical mechanisms of shock compression of powders involve rstorder descriptions based on thermodynamic equilibrium.2527 Extensive work has gone into development of the isolatedpore collapse model,28,29 including incorporation of rate dependent loading and viscoplastic effects.30 The most promising approach is that of Baer,31 based on a continuum mixture theory, which has methods for explicit treatment of local stresses. The Baer model also has the capability to incorporate data from time-resolved pressure measurements to describe the viscoplastic deformation problem of highly porous solids and the resulting chemical reaction. The scientic problem of chemically reacting powder mixtures has not progressed to the stage of realistic fundamental modeling, which
Thadhani et al. 1115

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

mechanical deformations induced by the presence of the voids and the heterogeneity of the stresses at the interparticle solid contacts. In the model, the two reactant particles are separated by a large void.6 The deformation works to close the void in time, and after some time, the reactants can be expected to become mixed in the sense of a more intimate spatial relationship. With further mechanical action and its intense shear, those reactants that are in sufciently intimate contact, begin to react. As reaction proceeds, the system becomes increasingly heterogeneous, and at some stage, proceeds to the degree of completion possible for the mechanochemical conditions. The unreacted starting materials also play a critical role due to their thermal quenching effects.32 An important consequence of the mixing concept is that variables such as the volumetric ratios of substituents play a crucial role. If volumetric ratios are substantially unbalanced, the reaction cannot proceed because mixing will be limited. Thus, the overall CONMAH concept denes stages corresponding to the conguration change of reactants prior to and at the onset of reaction, leading to the evolution of the products. Microstructural characterization of samples preserved for postshock analysis has been successful in bringing out features of the deformation processes and conguration changes in reactants; however, direct quantication of the effects has been difcult. Semiquantitative evaluation of the effects, such as shock activation and mixing, has been provided by differential thermal analysis,3335 revealing reductions in temperatures at which reactions occur upon heating of recovered samples, and x-ray diffraction XRD linebroadening analysis,3638 showing retained microstrain and crystallite size reductions.
III. EXPERIMENTAL PROCEDURES
FIG. 2. Schematic of a pressurevolume curve illustrating rst-order effects of reaction under conditions of constant pressure Ref. 21 , in which the exothermic energy transformation causes an expansion to larger volumes; b equilibrium wave velocity vs pressure behavior in a chemically highly exothermic reacting powder mixture calculated based on the reaction effects shown in a . It can be seen that at low pressures without reaction, higher wave velocities are anticipated until the crush strength is exceeded. At pressures greater than the crush strength, if chemical reaction occurs, then the observed behavior will show higher wave speeds dashed line than those predicted for the zero-strength model solid points .

can account for the reaction mechanisms. Constant volume reactions have been used to interpret the few available shock velocity data obtained by the Batsanov group.10,11 Although such an analysis is sufcient to demonstrate that a chemical reaction is likely to have occurred, it does not provide a physically descriptive model for the mechanical and chemical features of reaction. A model, called CONMAH conguration, mixing, activation, and heating has been proposed by Graham,6 identifying the processes in a conceptional sense and considering all relevant effects occurring prior to, at the onset, and subsequent to reaction initiation. It presumes that chemical reactions in highly porous solids are fundamentally mechanochemical in character, and are controlled by the intense
1116 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

The approach adopted in the present work was to investigate the chemical reaction response of a low-reactionthreshold titaniumsilicon system by performing a systematic series of combined time-resolved pressure measurements and shockrecovery experiments on the same powder mixtures. The titaniumsilicon powder mixture system has the unusual characteristic of a small density differential between reactants, approximately the same shock impedance, and a high reaction exothermicity. Both Ti and Si are expected to undergo shock-induced polymorphic phase transitions at pressures greater than about 10 GPa in homogeneous deformation modes.39 Table I summarizes physical properties of the starting powders along with those of their stable compounds.4042 Prior shock synthesis studies on TiSi powders by the Vreeland group20 revealed that shockinduced chemical reactions in this system occur at very low pressures few GPa and corresponding shock-induced meanbulk-temperatures below the melt temperature of Si, forming the Ti5Si3 compound. It was also reported that the propensity for reaction initiation in TiSi shows a very large dependence on the particle size, with powders approximately 45 m in size reacting at lower shock-pressure thresholds and also with lower porosities, than the coarser about 100 m powders. Such a particle size dependence has also been observed during combustion synthesis as well as mechanical
Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

TABLE I. Properties of elemental Ti and Si reactants and TiSi compounds Refs. 4042 . Sound velocity km/s 5220 7990 Yield strength MPa 140 93 Melt temp. C 1667 1412 1570 2130 1380

Constituent Ti Si TiSi Ti5Si3 TiSi2

Density (g/cm3) 4.500 2.330 4.320 4.315 4.043

V %

H kJ/g atom

Crystal structure hcp Diamond ortho hcp ortho

Electronegativity 1.5 1.8

22.9 27.8 27.5

15.5 73.4 10.7

alloying of TiSi powders,43,44 which suggests that the reaction characteristics of Ti and Si powders may be strongly process dependent.
A. Starting materials

B. Nanosecond time-resolved pressure measurements

Three different morphologies of both silicon and titanium were studied in what are termed coarse, medium, and ne powder mixtures based on a measure of the size of the powder particles. The titanium powders were: Cerac No. m; Aesar No. 10386, medium, T-1191, ne, 13 45 m; and Cerac No. T-1219, coarse, 105149 m. The silicon powders were: Cerac No. S-2021, ne, 10 m; Cerac No. S1053, medium, 45 m; and Cerac No. S1051, coarse, 105149 m. Particle size distributions and scanning electron microscopy SEM images were obtained for each powder. SEM images showing the morphology of the powder mixtures are shown in Figure 3. The medium morphology Ti was found to contain many elongated particles, which may be as long as 100 m, but because of the elongated shape, they pass through the 325 mesh sieve. While the Ti powders were polycrystalline aggregates in all cases, the Si powders were actually single-crystal particles. The Ti and Si powders were mixed in a stoichiometric ratio to form the Ti5Si3 compound, which corresponds to a weight ratio of 74/26 and a volumetric ratio of 60/40.

Direct time-resolved measurements allowing observations of shock-compression phenomena on the nanosecond time scale under precisely controlled loading are essential to distinguish effects occurring during shock compression from those that occur after unloading to ambient conditions. The pioneering work on time-resolved observations of chemically reacting powder mixtures was accomplished by Batsanov and co-workers and reported in 1986.10 These authors used manganin pressure gauges to obtain records of shock proles, and wave speed, and observed that in equimolar tin and sulfur powder mixtures, the measured pressure points deviated towards the right increased volume of the Hugoniot curve calculated for the unreacted mixture. Additional work was reported in this same system45 and in tin teluride,11 where optical pyrometric measurements were also utilized. The aluminumsulfur system was also investigated in work reported in 1992.12 Other work with high-pressure shock waves demonstrating evidence for shock chemistry using various detection methods is listed in Table II. The rst use of nanosecond time-resolved measurements using PVDF poly-vinyl-di-uoride stress-wave gauges in chemically reacting powders was reported by Graham and

FIG. 3. SEM images showing particle morphologies of different 5 Ti3 Si powder mixtures: a Fine, 13 m Ti Cerac No. T-1191 and 10 m Si Cerac No. S-2021 ; b Medium, 1044 m Ti Aesar No. 10386 and Si Cerac No. S-1053 ; and c Coarse, 105149 m Ti Cerac No. T-1219 and 45149 m Si Cerac special No. S-1051 . Si particles are generally blocky single crystals with a shiny contrast, while Ti particles are rounded polycrystalline aggregates. J. Appl. Phys., Vol. 82, No. 3, 1 August 1997 Thadhani et al. 1117

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

TABLE II. Measurements reporting evidence of shock chemistry. Year 1986 1989 1990 1991 1991 1992 1992 Authors Ref. Batsanov et al. Boslough46 Boslough47 Gao and Jing48 Batsanov11 Gogulya12 Yoshida and Thadhani15 Bennett et al.14 Iyer et al.13 Batsanov et al.45 Chen et al.49
10

Method used Pressure, manganin gauges Optical pyrometry Optical pyrometry Manganin gauge Pressure, manganin gauges Optical pyrometry Optical measurements of bulk sound speed Reected-shock pressure, manganin gauges Pressure, manganin gauges Optical pyrometry and manganin pressure gauges Impedance match

Material density Sn S 98% Ni Al Al Fe2O3 50% 2Al Fe2O3 50%70% Sn Te S Al ( 90%) 3 sizes, 3 mix ratios Nb Si 50%

Observations Shift in Hugoniot at 15 GPa Temperature changes Temperature increases above 5 GPa pressure 5 GPa wave Shift in Hugoniot at 50 GPa Temperature increases at 79 GPa pressure Increases in bulk sound speed and Hugoniot changes at 20 GPa Excess reected-shock pressure at 15 GPa Excess reected shock pressure Temperature increases within 0.2 s at 20 GPa 3045 GPa discontinuity

1992 1993 1994 1994

Al Ni 55% Al Ni 55% backed by steel or lexan Sn S 92% BaCO3 TiO2

co-workers in 1993.21 Dunbar and co-workers16 reported the rst time-resolved work in titanium silicon, also in 1993. Table III summarizes the various studies on highly porous solids with the PVDF gauges. The literature cited in Tables II and III, shows that there is considerable evidence for the occurrence of a strong exothermic chemical reaction on times scales of signicantly less than the typical wave transit time of 1 s.16,21,5054 Measurements of wave proles in shock-compressed powders reported during the past few years have also revealed that the powder compression process is complex and highly rate dependent.50,51 Stress pulse rise times after propagation distances of a few millimeters are large, typically varying from many tens to hundreds of nanoseconds. With these characteristics, the wave propagation in powder samples cannot be realistically described on the basis of conventional shock conservation relations. In a particularly interesting case, a study of Al Fe2O3 mixtures51 observed that there was evidence that the transmitted waves had distinct features in their proles associated with individual deformation features of each constituent. In the present work, Bauer piezoelectric polymer PVDF stress-wave gauges55 were used to quantify the responses of the powder samples. Typical PVDF gauge package congurations consisted of insulating lms of FEP Teon or Kel-F on both sides of 25 m thick PVDF elements, with Al sput-

tering of 2000 on powder sides of the gauge package to prevent pyroelectric effects from affecting gauge response during the possible reaction of powders. All gauges were of high quality, biaxially stretched PVDF lm, poled using the Bauer process52 to a 9.2 C/cm2 remnant polarization, and having identical gold over Pt electrodes. The PVDF gauge experiments were performed using the Precision Impact Facility at Sandia National Laboratories, which employs a 63 mm diam, 25 m long single-stage, compressed-gas gun. The facility provides for exceptional control of tilt or misalignment of impact surfaces typically 200 rad or less , impact velocity measurements to 0.1%, wave speed measurements to 0.1%, and Gigahertz frequency response instrumentation. The frequency response of the gauge and recording system result in timing measurements that provide an unusually precise capability for wave speed measurements. For higher pressure experiments, explosive loading with 4 in. diam plane-wave generators was used.51 The overall experimental arrangement is shown schematically in Fig. 4. The powder-mixture samples were pressed directly into copper capsules with PVDF gauge packages placed in intimate contact with the powder-capsule planar surfaces to monitor both input-shock loading and the propagated-wave characteristics. The propagation of the shock wave sensed by the input gauge and propagated

TABLE III. Nanosecond time-resolved measurements with PVDF gauges performed on powders. Year 1993 1994 1994 1994 1994 1996 1996 Authors Graham et al. Anderson et al.50 Holman et al.51 Dunbar et al.16 Shefeld et al.52 Anderson et al.53 Holt et al.54
21

Material Density 3Ni Al, 5Ti 3Si, TiO2 TiO2 60% dense 0.216.1 GPa 2Al Fe2O3 mixture 53% dense , 0.6710 GPa 5Ti 3Si 45%53% HMX explosive (1.24 g/cm3) HMX explosive Teon, two morphologies

Observations Powder compression, wave rise times wave dispersion in porous rutile, rise times in HMX, 5Ti 3Si, 3Ni Al, Structured waves, P V stiffening Wave-prole measurements Wave-prole and reaction behavior of HMX Simultaneous PVDF and VISAR experiments Wave rise time, crush strengths

1118

J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

FIG. 4. Schematic of overall experimental arrangement showing typical conguration used for time-resolved pressure measurements, with the powder mixture sample material encased in a copper capsule and PVDF gauge packages placed in intimate contact with powdercapsule planar surfaces to monitor both input-shock loading and propagated-shock wave characteristics. Congurations with different driver and backer materials on both sides of powder and gauges were used, to impedance match with powder sample and gauges.

listed in Table IV . Different congurations were used in order to match sample shock impedances more closely with the powder-mixture impedance, and to delay the reshock of the powder mixture to clearly detect propagated-wave prole features. The sample assembly was placed on the impact surface of the gun in which a smooth-bored projectile faced with a 10 mm thick copper yer plate was accelerated to a preselected impact velocity. It should be noted that all experiments were designed such that a planar shock wave propagates through the powder thickness without attenuation from the loading or peripheral surfaces. To improve sensitivity, the piezoelectric current was recorded with two complementary amplier sensitivities connected to a current viewing resistor at the PVDF gauge. The combined recordings of both ampliers provide a high-resolution current-versus-time prole, which is then numerically integrated to obtain the stress-versus-time prole using the PVDFSTRESS computer code56 and stress magnitudes using data analysis procedures described elsewhere57 .

gauge at their respective locations also provides precise data on the transit time through the powder sample. OHFC copper yer plates were used in all cases. In some experiments, copper driver plates were used to transmit the impact stress into the samples, and in others TPX or Kel-F polymers were bonded to the impact surface of copper driver plates. The actual sample assemblies consisted of several congurations

C. Sample preservation (shock recovery) experiments and materials characterization

For sample-preservation experiments, the various TiSi powder mixtures were explosively loaded using the Sandia Poppa-Bear xture samples of 45.7 mm diam 6.35 mm thickness with baratol explosive PBB and MommaBear xture samples of 31.8 mm diam 6.35 mm thick , also with baratol explosive MBB . These standardized

TABLE IV. Experimental parameters and results of time-resolved experiments on 5TI 3SI powder mixtures. Impact velocityc km/s 0.280 0.434 0.650 0.871 0.972 TNT PBX-9404 1.043 0.460 0.642 0.382 1.036 0.618 Input stresse GPa 0.21 0.51 1.49 2.08 2.5 2.94 11.5i 2.8i 0.674 1.303 0.625 2.81 1.33 Wave velocityf km/s 0.766 0.878 1.314 1.646 1.842 2.090 2.786i 1.848i 1.080 1.279 0.966 1.803 1.233 0.750 0.825 1.299 1.597 1.827 2.064 N/A N/A 0.969 1.311 0.921 1.745 1.191 Output stressg GPa 0.42 Kel-F 0.69 TPX 1.17 TPX 3.95 Kel-F 11.0 Cu 5.4 Kel-F N/A N/A 0.63 TPX 0.86 TPX 0.64 TPX 2.88 TPX 1.20 TPX

Experiment numbera 2462 a 2475 a 2476 a 2463 a 2407 a BE-231 b BE-232 b 2501 a 2499 a 2497 a 2504 a 2500 a 2505 a
a b

Morphologyb Medium Medium Medium Medium Medium Medium Medium Coarse Coarse Coarse Fine Fine Fine

Congurationd 2 3 4 2 1 5 5 3 3 3 4 4 4

Relative volumeh 1.539 1.243 1.042 1.141 1.169 1.227 0.442i 1.083i 1.322 1.108 1.312 1.047 1.025 1.527 1.157 1.024 1.089 1.157 1.211 N/A N/A 1.223 1.146 1.221 0.968 0.938

a impact loading; and b explosive loading. Medium morphology 1.930 0.012 g/cm3 47% porosity , coarse morphology 1.925 0.003 g/cm3 47% porosity , and ne morphology 1.633 0.003 g/cm3 55% porosity . c Symmetric impact, Cu on Cu or explosive loading. d Conguration, nominal thickness in parentheses: 1 Cu 6.5 mm FEP 50 m PVDF 25 m FEP 62.5 m 5Ti 3Si 4 mm 2 Cu 6.5 mm Kel-F 125 m PVDF 25 m Kel-F 125 m 5Ti FEP 62.5 m PVDF 25 m FEP 50 m Cu 9.5 mm ; 3 Cu 6.5 mm TPX 9.5 mm PVDF 25 m 3Si 4 mm Kel-F 125 m PVDF 25 m -Kel-F 125 m Kel-F 0.7 mm Cu 8.8 mm ; FEP 12.7 m 5Ti 3Si 4 mm FEP 12.7 m PVDF 25 m TPX 9.5 mm Cu 6.5 mm ; 4 Cu 6.5 mm TPX 9.5 mm PVDF 25 m 5 P-040explosive 25.4 m Al 12.7 mm FEP 25 m 5Ti 3Si 4 mm FEP 25 m PVDF 25 m TPX 9.5 m Cu 6.5 mm ; Cu 6.5 mm FeP 12.7 m PVDF 25 m FeP 12.7 m 5Ti 3Si 4 mm FEP 12.7 m PVDF 25 m Kel-F 9.5 mm Cu 6.5 mm . e PVDF measurement. f Wave speed from toe-to-peak half-maximum value . g PVDF measurement polymer backing material . h Calculated from PVDF input stress, unshocked density, and toe-to-peak wave speed half-maximum value . i Values computed based on predictions, not on measurements. J. Appl. Phys., Vol. 82, No. 3, 1 August 1997 Thadhani et al. 1119

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

shock-recovery xtures have been extensively used for generating a range of controlled and reproducible shock conditions in numerous single-component and multiplecomponent powders.2 The plane-wave explosive-loading system utilizes standard P-22 plane-wave generators and explosive pads, designed and manufactured at U.S. DOE Laboratories. The copper capsule encasing the powders permits their pressing to preselected densities with accuracies of less than 1%. To assure well-dened loading conditions, all surfaces incident to the shock direction are mechanically lapped to optical atness. In these xtures, the loading history is dominated by an initial low-pressure planar wave, followed by a radial wave, caused by wave trapping occurring as a result of the low impedance powder mixture relative to that of solidcopper containment. The resulting pressures calculated from numerical simulations on 55% dense rutile powders, show that the PBB xtures subject the powders to an average peak pressure of 5 1 GPa with a shock-induced temperature rise of 150 C in the bulk region and 75 C in the outer volume. The MBB xtures subject the powders to an average peak pressure of 7.5 2.5 GPa in the bulk region and a pressure of 27 GPa along the axial region with temperatures of 225 C in the bulk and 250 C in the outer volume. Because of the low starting density of typical powder compacts and resulting radial wave-focusing effects, the shock pressures and temperatures are relatively insensitive to materials properties. Details describing the recovery xtures and numerical simulations are provided in prior works.5861 The shock-compressed samples preserved for postshock analysis are typically removed intact from the xture after experiment, then sectioned, and polished to view the cross section of the hockey-puck shaped compacts. Optical and scanning electron microscopy were performed to identify spacial variations in sample characteristics, including whether partial or complete reaction was present, and to characterize the shock-deformed conguration of unreacted constituents. A typical reacted sample shows a smooth or uniform contrast of individual grains of the product microstructure along with spherical or rounded voids in the case of highly exothermically reacting systems. Unreacted samples show the dissimilar contrast of the constituent particles, as well as distinct interparticle boundaries. X-ray diffraction line-broadening analysis, using the WilliamsonHall method,62 was also performed on the unreacted powder mixtures of different morphology, to determine the amount of microstrain retained and crystallite size reductions in the shock-compressed reactants. These XRD measurements build upon the extensive work of Morosin36,37 who employed the WarrenAverbach technique, which is important for lower symmetry materials such as ceramics.
IV. EXPERIMENTAL RESULTS A. Time-resolved pressure measurements

FIG. 5. Representative traces of gauge output in current vs time and corresponding integrated traces of stress vs time for gauge packages located at input-shock a and c and propagated-shock b and d locations, separating the 4 mm thick TiSi powder mixture layer for sample No. 2476.

The time-resolved PVDF-gauge pressure measurements were performed in two sets. In the rst set, experiments were conducted on medium-morphology TiSi powders at different stresses and a constant packing density of about 53%
1120 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

TMD.16 In the second set, the experiments were extended to coarse at 53% TMD and ne at 45% TMD powders, through the same range of pressures. Representative traces of the gauge output in currentversus-time and the corresponding integrated traces of stressversus-time for gauge packages located at both input-shock and propagated-wave locations, for experiment No. 2476, are shown in Figs. 5 a 5 d . The input PVDF gauge generates a piezoelectric current as the shock wave transits the gauge, with a rise time less than the shock transit time through the 25 m lm thickness. In interpretation of the records, it is important to consider the detail of wave propagation in the PVDF gauge packages. The input shock propagates through the 50 m FEP Teon insulation lm to the Teonpowder interface, where a reection is caused due to the impedance mismatch between the Kel-F and the powder. The reected release wave then arrives back at the input PVDF gauge about 40 ns after the initial input shock, as shown in Figs. 5 a and 5 b . The magnitude of the input stress entering the powder is taken as the stress at the time 40 ns , where the arrival of the reected release is indicated on the current time trace. Thus, as shown in Fig. 5 c , the input stress of 1.49 GPa propagates through the 4 mm thick powder sample, and arrives as a dispersed wave at the propagated PVDF gauge location, and generates a piezoelectric current Fig. 5 b with wave rise time typically increased by an order of magnitude from that recorded by the input gauge. As shown in the resulting stress-versus-time prole in Fig. 5 d , the output stress recorded by the propagated gauge is 1.17 GPa. The experimentally measured and calculated parameters from all time-resolved PVDF gauge experiments on the powder mixtures and the corresponding sample congurations are listed in Table IV. Data in the input stress column correspond to the stress in the powder measured by the inputshock gauge as indicated above. The wave velocity is the wave speed through the powder, obtained by measuring transit time between the two gauges placed in direct contact with
Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

FIG. 6. Measured input stress plotted as a function of calculated relative volume for medium, coarse, and ne TiSi powder mixtures, along with a calculated vertical curve for a solid inert TiSi powder mixture. Points corresponding to experimentally measured input stress and calculated relative volume for the three different morphologies of powder mixtures show signicantly different trends. Points for medium-morphology powders indicate crush-up to full density at 1 GPa pressure, followed by volume expansion shift to right of calculated solid inert mixture curve , indicating evidence for rapid shock-induced chemical reaction, while points corresponding to coarse and ne powders show crush-up at higher pressures and no signicant deviation from the solid curve, indicating no reaction at pressures up to 3 GPa.

FIG. 7. Plot of experimentally measured wave speed as a function of measured input stress for the three morphologies of TiSi powder mixtures showing trends similar to those revealed by pressurevolume compressibility characteristics shown in Fig. 6. The points for medium-morphology powders approach the calculated curve for the inert powder at the crush-up strength, and at input stresses greater than the crush strength, the experimental points deviate from curve showing increased wave speed corresponding to the occurrence of chemical reaction and formation of an intermetallic compound in the shock state. Points for the ne and coarse mixtures remain close to the calculated curve indicating no reaction.

opposite surfaces of powder compact less propagation times for the insulation . The output stress, measured by the propagated-wave gauge placed between the powder and polymer backing material, corresponds to the stress resulting from the interaction between the powder mixture and the polymer backing. The magnitudes of these stresses will depend strongly on the backing material due to the impedance mismatch. The relative volume shown is calculated from the known initial density, measured input stress and wave velocity, and from shock jump conditions for conservation of mass and momentum. In reality, the stress pulses propagating through the 4 mm thick powder mixtures have a structure characteristic of wave-dispersion effects, with shock rise times greater than several hundreds of nanoseconds.50,51 With such wavedispersion effects, the calculation of the relative volume based on jump conditions applied to a steady-state shock wave, may not be appropriate. One can, however, use the calculated relative volume along with the measured input stress to obtain rst-order powder-compressibility effects. This calculation is more meaningful for fully dense solids than for highly porous solids due to the large compressions involved in the latter case. The measured input stress plotted as a function of the calculated relative volume is shown in Fig. 6. A calculated curve for an inert TiSi powder mixture at 53% density , is also shown in the plot, with crush-up to full density assumed to be occurring at zero stress, followed by marginal expansion from that of the soliddensity curve. This zero-strength inert curve can be calculated with condence due to the much smaller compression compared to the porous state. Points corresponding to experimentally measured input
J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

stress and calculated relative volume for the three different morphologies of the powder mixtures show signicantly different trends. It can be seen that points for mediummorphology powders indicate crush-up to full density occurring at pressures less than about 1 GPa, followed by a subsequent shift to the right of the calculated inert powdermixture curve. The observed volume expansion indicates strong evidence for rapid chemical reaction, consistent with the constant pressure model21,22 and the shock-compression mechanics description provided earlier. The constant pressure model was proposed on the basis of consideration of mechanical relaxation times for heterogeneous states of the reactants and products, and contrasts sharply with the constant volume model in which large pressures will result from the exothermic energy products.10,63,64 The plot of measured input stress versus calculated relative volume Fig. 6 also shows that in contrast to the medium-morphology powders, the points corresponding to the coarse and ne powder mixtures, do not deviate from the calculated inert curve, demonstrating that these powder morphologies do not undergo reaction under the conditions used. It is signicant to note that the data for both the coarse- and ne-morphology mixtures, indicate crush-up to full density at stress levels higher than that for the medium-morphology powders. The observed different crush strengths are overt indications of the strong inuence of morphology on the deformation process. While the results of measured input stress plotted as a function of calculated relative volume may be considered sufcient only to reveal rst-order effects, the measured wave speed as a function of the measured input stress perhaps provides a more direct indication of chemical reactions in the shock state. Figure 7 shows the wave speed-versusinput stress data for the powder mixtures, obtained from the
Thadhani et al. 1121

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

TABLE V. Parameters and results of recovery experiments in 5Ti 3Si powders of different morphology Refs. 5760 . Experiment number NMG-9122 NMG-9121 NMG-9120 NMG-9212 NMG-9123 NMG-9211 NMG-9252 NMG-9251 NMG-9254 NMG-9253 SNL-2506 SNL-2572 SNL-2573
a

Packing density g/cm3 % 1.92 1.93 1.95 1.625 2.147 2.33 1.633 1.916 1.626 1.989 1.912 1.901 1.904 52.8% 53.1% 53.6% 44.7% 59.0% 64.1% 44.9% 52.7% 44.7% 54.7% 52% 52% 52.6%

Porosity % 47.2 46.9 46.4 55.3 41.0 35.9 55.1 47.3 55.3 45.3 48.0 48.0 47.4

Congurationa PB-B MB-B MB-B PB-B PB-B MB-B PB-B MB-B PB-B MB-B gun gun gun

Powder morphology Medium Medium Medium Medium Medium Medium Coarse Coarse Fine Fine Medium Coarse Fine

Reaction behavior None Complete Complete Complete None None None None None Localized None None Not recovered 5 GPa.

MB-B Momma Bear Baratol, Peak P

7.5 GPa; PB-B Poppa Bear Baratol, Peak P

PVDF gauge measurements. Trends similar to those revealed by the pressurevolume compressibility characteristics are observed. At shock pressures above the crush-up strength, the points for medium-morphology powders appear to deviate from the calculated curve for the inert powder, showing increased wave speed corresponding to the occurrence of chemical reaction and formation of an intermetallic compound in the shock state. On the other hand, points for the ne and coarse powder mixtures remain close to the calculated curve, indicating no reaction with these morphologies under the conditions of the experiments. Thus, the compression characteristics, and measured wave speed-versusmeasured input stress reveal consistent evidence for shockinduced chemical reactions in medium-morphology powder mixtures and no reaction in coarse and ne mixtures. The results provide the data needed to determine the extent of reaction. The data also provide the critical observations needed for development of realistic deformation descriptions in mixture theory models.
B. Sample preservation experiments

focusing effects, and at impact velocities of about 370 m/s, approach conditions similar to the PBB xture. Details of powder characteristics and experimental congurations used for the recovery experiments are listed in Table V. The overall results of shock-recovery experiments, based on optical and SEM analysis, are illustrated in the reaction map shown in Fig. 8. For medium-morphology powders, MBB experiments showed complete reaction in samples with 46% porosity, and no reaction in samples with 36% porosity. The PBB experiments showed no reaction in samples with 41% and 47% porosity, but complete reaction with 55%

Six shock-recovery experiments were performed on the powder mixtures of medium morphology packed at an initial density ranging between 53% and 64% with the MBB xtures, and 45% and 53% with the PBB xtures. Two experiments were performed on ne- and coarse-morphology powders, each with packing density of 64% with the MBB, and 53% with the PBB xture. Controlled gun-impact experiments at 52% initial density were also performed, one each on ne-, medium-, and coarse-morphology powders at impact velocities of about 370 m/s. It should be noted that unlike the more dispersed wave produced in the Momma and Poppa Bear explosive loading congurations due to the wave-shaping effect of the explosively generated pulse as it is transmitted through a steel driver undergoing the 13 GPa phase change , a at pulse is generated during experiments with controlled plateimpact loading. Thus, with gun recovery experiments, the incident conditions in the powders are similar to those produced in time-resolved experiments. The peak conditions, however, are dominated by radial wave1122 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

FIG. 8. Reaction map showing overall results of shock recovery experiments, based on the effect of initial density and shock pressure on the reaction threshold in TiSi powder mixtures of medium Ti, Si 1044 m , coarse Ti 105149 m and Si 45149 m , and ne Ti 13 m, Si 10 m . Cross-hatched bars correspond to reacted material, and horizontally hatched bars with dark lettering correspond to unreacted material, as evidenced from microstructural optical and XRD analysis of recovered shock compressed samples. Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

FIG. 9. General characteristics of fully reacted TiSi sample with a scanning electron micrograph revealing the grain structure average size 10 m , and b XRD trace showing peaks corresponding to the Ti5Si3 reaction product.

porosity. The reaction trends for medium-morphology powders were used to guide experiments on ne and coarse powders. In general, the microstructure of fully reacted samples showed a uniformcontrast grain structure, interdispersed with spherical voids, typical of a melted and resolidied material. Figure 9 shows the general characteristics of the reacted sample with the optical micrograph in a revealing the grain structure (size 10 m), and the XRD trace in b showing peaks corresponding to the Ti5Si3 product. The results of materials analysis of reaction propensity in mixtures of Ti and Si powders of different morphology, shock compressed at the same pressures, shows that the medium-morphology powders, have a greater propensity for reaction, than ne- and coarse-morphology powders packed at the same density, similar to the trends revealed by timeresolved instrumented experiments. Furthermore, with higher packing densities or lower initial porosities higher pressures are required for shock-induced reaction initiation. With 53% TMD packing density, the medium-morphology powders react in the higher pressure MBB xture and not in the lower pressure PBB xture, in which case, reaction is observed only with the lower packing density of 47% TMD.
C. Characteristics of unreacted shocked sample compacts

Microscopy and x-ray diffraction line-broadening analyses were performed on recovered compacts, which were shock compressed at conditions below the threshold. Such
J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

samples are useful to qualitatively and quantitatively determine shock-induced congurational changes to the particles and characteristics of reactants prior to reaction initiation. All microscopic analysis was performed on cross-sectional surfaces, along identical regions of every compact, namely, the bulk area away from axis and peripheral edges. Optical micrographs of samples of the three different morphologies of powder mixtures, shock compressed using the PBB shock recovery xture are shown in Figs. 10 a 10 c . The grainycontrast ne-, medium-, and coarse-morphology Ti particles, in all three micrographs show extensive plastic deformation and ow. The ne Ti particles are seen to also form large agglomerates ( 200 m diam). On the other hand, the blocky and shiny-contrast ne and medium Si powders show extensive plastic deformation and ow around the Ti particles, while the coarse Si particles show only fracture and cracking. SEM micrographs of unreacted coarse- and medium-morphology powder mixtures, packed at higher initial density and shocked with the MBB xture, provide a better description of the conguration, as shown in Fig. 11. It can be clearly seen that, while Ti powders of both coarse and medium morphology undergo extensive plastic deformation and ow, the qualitative nature of the response of Si particles depends on the particle size. Medium-morphology Si powders plastically deform and ow along with Ti particles Fig. 11 a , in contrast to coarse Si powders, which exhibit extensive fracture and fragmentation, and remain contained within the Ti particles Fig. 11 b , thereby limiting the intimate mixing of reactants. In all cases, independent of morphology and peak pressure, titanium is observed to form the matrix of the compact, indicating that titanium plastically ows around silicon and encapsulates the silicon fragments. To quantify the deformation effects in shock-compressed unreacted compacts, x-ray diffraction peak broadening analysis was performed on the initial powder mixtures of medium and coarse morphology, and the shocked compacts of medium morphology No. 2506 and coarse morphology No. 2572 . The back surfaces of the compacts were lightly polished to remove an approximately 0.5 mm layer, prior to x-ray scans with a 0.015 step size, and a 5 s hold was used to analyze the data. Prior to performing peak-broadening analysis, instrumental broadening was computed using Si standard and, subsequently, subtracted from the experimentally measured integral breadths for the powder-mixture sample. A 95% condence interval of the y intercept of the linear regression was also determined for the Si standard, to calculate the amount of error in each subsequent WilliamsonHall value. WilliamsonHall plots int cos versus sin were generated for each sample to quantify the effects of microstrain and crystallite size. The Williamson Hall analysis performed on unshocked-, coarse-, and medium-morphology Ti powders showed no microstrain, while Si powders of both morphologies showed some strain. Titanium powders are generally made by chemical techniques and no retained strain would be expected. On the other hand, the strain in Si powders may have been induced
Thadhani et al. 1123

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

FIG. 10. Optical micrographs comparing the conguration of deformed reactants in a ne, b medium, and c coarse morphology powders shock compressed with Poppa Bear Baratol PBB recovery xture at similar packing density. While the grainy-contrast Ti particles in all mixtures reveal consistent deformation effects, the blocky- and shiny-contrast single-crystal Si particles show deformation behavior dominated by particle size effects, with ne and medium Si particles showing extensive deformation and coarse Si particles showing extensive fracture and cracking.

due to attrition processes used for reducing and grading powders to different sizes. A WilliamsonHall plot comparing the medium- and coarse-morphology compacts, at the same peak shock pressure impact experiments Nos. 2506 and 2572 is shown in Fig. 12. The slope of the shock-compressed coarse Ti powder is slightly greater than the corresponding slope of medium-morphology Ti, indicating greater microstrain in the coarse Ti compact. However, the data on Si reveal an interesting trend, with considerable microstrain in the mediummorphology powder as indicated by the larger slope and minimal strain indicated by nearly zero slope in the coarsemorphology Si powder compact. The intercept of the coarse Si is, however, increased, indicating that the crystallite size is reduced to almost the same as medium-morphology Si powders indicating particle fracture .

The calculated residual microstrain of Ti in shockcompressed, unreacted samples of both the medium- and coarse-morphology system are 3.22 10 3 and 3.98 10 3 , respectively. In contrast, the microstrain in Si of the medium-morphology sample is 3.14 10 3 , while the coarse-morphology Si exhibits extensive fracture and no strain. The values of microstrain in Ti and Si correspond to dislocation densities of the order of 1011 cm 2 estimated using the Williamson and Smallman approach.64 Such levels of dislocation densities are typical of heavily cold-worked materials.

FIG. 11. SEM micrographs showing unreacted conguration of a medium morphology and b coarse morphology TiSi powder mixtures shock compressed with the MBB xture, revealing clear differences in their deformation and fracture response. 1124 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

FIG. 12. WilliamsonHall plot comparing the medium and coarse powder mixture compacts at the same peak pressure impact experiment Nos. 2506 and 2572 . The Ti peaks in both cases show substantial microstrain, with the trace for coarse Ti indicating slightly larger strain than the medium Ti shocked samples. Data for Si peaks show extensive strain in the medium morphology shocked sample, and no strain but signicantly decreased crystallite size increased intercept in the coarse morphology shocked sample. Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

V. DISCUSSION OF RESULTS

The combination of traditional shock compression work time-resolved measurements and traditional materials science studies microstructural characterization performed in the present investigation, provides an unusually strong materials behavior database from which the fundamental nature of the deformation characteristics of reactants and kinetics of chemical reactions can be described.
A. Shock-induced chemical reaction behavior

It is clear from the results of time-resolved measurements, that sufcient evidence exists to demonstrate that chemical reactions in TiSi powder mixtures occur during the rise times and the microsecond duration wave propagation of shock loading. Nevertheless, the mechanochemical processes leading to the initiation of reaction are extremely complex. The fact that coarse and ne powder morphologies are less sensitive to the shock initiation of reactions than medium-morphology powders, evidenced both from timeresolved measurements and shock-recovery experiments, is an overt indication of the complex local conditions that lead to reaction initiation. Thermochemical models,18,19,65 as well as voidcollapse models,2931 are not capable of evaluating the powder morphology inuence on reaction behavior. This is not just because of the geometrical constraints that need to be built in the models, but primarily because constitutive relations available do not permit realistic calculation of shapedependent deformation or fracture response. Voidcollapse models also cannot describe situations in which voids are not isolated but are strongly interacting during the deformation process.
B. Shock-induced congurational changes in unreacted powder constituents

The shock-induced congurational changes of reactants in unreacted samples are observed to differ signicantly among the different morphologies. The deformation or fracture response of constituents determines the extent of mixing and the nature of congurational changes occurring during shock compression, which greatly inuences the intimacy of contact between reactants, and therefore, the reaction pressure thresholds. In the powder mixtures, the Ti particles generally undergo extensive plastic deformation irrespective of particle size, as evident from microstrain analysis and microstructural observation. On the other hand, Si particles exhibit either plastic deformation or cracking and fracture of particles, depending on the particle size. The coarsemorphology powder mixtures show extensive fragmentation and containment of Si within deformed Ti particles. The SEM micrograph shown in Fig. 11 b illustrates this effect most clearly. With such a behavior, mixing between the two reactants is inhibited, restricting chemical reaction. Coarse particles of covalently bonded materials such as diamond66 have also been shown to exhibit extensive fracture, while powders of small sizes undergo plastic deformation. The ne-morphology powders, on the other hand, show agglomeration of particles forming separate aggregates of Ti
J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

and Si, thereby limiting simultaneous deformation and mixing. In the present powder mixtures, shock compression of only the medium-morphology powders results in simultaneous deformation of both Ti and Si, which yields an intimately mixed conguration more favorable to chemical reaction. Consequently, it can be argued, after the study of the conguration of shocked and unreacted samples, that the medium-morphology powder mixtures could be expected to have the lowest reaction pressure threshold. The observed strong inuence of particle morphology on chemical reactions is overt evidence that equilibrium thermodynamic models are inadequate to describe the observed behaviors. Indeed, the ne-morphology higher surface area low initial density higher shock temperature compacts would be expected to initiate reaction at the lowest shockloading conditions in many descriptions. Such is not the case; these strong morphological effects are also characteristic of AlNi powder mixtures as observed from our previous studies.23,24,34,35,67 Clearly, a scientic description of the chemical process must be based on the local micromechanical conditions. In the thermochemical sense, the coarse morphology in which interparticle contacts are fewer in number than in ne or medium morphologies, higher local stresses will result in higher temperatures in localized hot spots. 68 For example, in recent studies on silicide-forming powder mixtures, Meyers et al.19 proposed a thermodynamic and kinetic analysis of shock-induced reactions in Nb or Mo and Si based systems. Their analysis is derived on a reaction initiation mechanism requiring a melt phase at siliconmetal interparticle regions. They rationaled that if the energy generated due to chemical reaction is greater than that dissipated by thermal conduction, then a steady-state reaction can start from local hot-spot melt areas and propagate into the interior of the particles. Accordingly, they calculated critical molten hot-spot regions and melt fraction of Si , based on a shock energy threshold corresponding to the mean-bulk temperature, which must be above that required to initiate reactions at ambient pressure. However, if such local temperatures alone were responsible for initiation of reaction, coarse particles would be observed to initiate reaction at a lower mean pressure than the medium-morphology powders. Such is not the case. Indeed, existing thermochemical models, which address computation of local interfacial temperatures the calculated local temperature is strongly model dependent are sufcient to describe chemistry in only a very thin region of material at the interface. The observed chemical reaction behaviors indicate rapid reaction over large volumes of materials. In the mechanochemical sense, the interparticle contact points control the local stresses, thereby producing stress magnitudes and stress states, which result in enhanced plastic deformation and more efcient lling of the voids with the reactants, leading to a more intimately mixed reactant conguration. The role of morphology, then, is to control the magnitude of the local stresses, thereby inuencing the deformation. The present observations are consistent with substantial local stresses. The presence of a signicant, morphologically dependent mean pressure is also required to explain
Thadhani et al. 1125

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

the change from ductile to brittle deformation in silicon. The mechanochemical concept of solid-state reactivity is greatly enhanced by defects resulting from plastic deformation.69 Hence, the mechanochemical process will lead to shockactivated conditions. The present work provides both qualitative conceptual features and quantitative data, which can be used to develop realistic models of chemical reactions induced by shockcompression processes. It also demonstrates and outlines fundamental issues involved in shock compression of highly porous solids. These results can now be used to develop models explaining the mechanisms of the shock-compression response of powders and, subsequently, shock-induced chemical reactions. In the following section, we provide concluding remarks, which discuss these issues in light of the results of the present work and our other prior work.
VI. CONCLUDING REMARKS: SUMMARY OF FUNDAMENTAL ISSUES

The present work adds qualitatively new aspects to our knowledge with the use of both nanosecond time-resolved pressure measurements and shock-recovery techniques, but perhaps its principal signicance is to bring a focus to the prior broad and deep studies of shock-induced solid-state chemistry. The various research studies have moved from the early phrenological investigations of 30 years ago1,5,70 to a contemporary ability and realization of the need to clarify a vision of the fundamental issues involved. Fundamental issues to be considered include i differentiation between thermochemical and mechanochemical processes, ii identication of materials and morphological factors controlling initiation of reaction, modeling of deformation of porous solids, and modeling of the chemical reaction process, and nally iii the role of nanosecond, time-resolved measurements in developing scientic models. The choice between thermochemistry and mechanochemistry as the fundamental paradigm used to provide a scientic description of the deformation and chemical reaction point i above , provides a stark contrast. The literature is abundant in both areas, and both paradigms are supported by theory and experiment. Thermochemical approaches are rooted in the foundations of early shock-compression science, which are based on concepts of thermodynamic equilibrium. It is generally agreed that shock-induced chemical changes in materials are observed principally in the porous state. Even though it is clear that the porous solids must experience heterogeneous deformation, conventional Hugoniot approaches have yielded a reasonable rst-order description in many cases.18,19,65 In this case, the additional pressurevolume energy in the system corresponding to compression of the extended state is assumed to appear as temperature, which adds to the pressurevolume state as an expansion. The temperature experienced can then drive chemical reaction, particularly when the contribution of shear deformation in shocked solids is considered. Elevated temperature can also result in melting with resulting increases in diffusion constants. The validity of the thermochemical paradigm for the description of shock-induced solid-state chemistry must rest upon detailed studies and ex1126 J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

periments, which probe the fundamental assumptions. In fact, rst-principle, elementary considerations and many detailed experiments clearly show that shock-induced solidstate chemistry cannot be scientically described on the basis of thermochemistry. Mechanochemical approaches are rooted in foundations of the substantial mechanochemical literature and elementary, rst-principle considerations. It is obvious that in the highly porous state, an assembly of particles can only be stressed through interparticle contacts. Mean pressure on the powders is, thereby, supported locally with stresses substantially greater than the mean pressure. Many investigations show that the principal feature of shock-compressed powders is their residual strain, typical of cold-worked metals even in covalently bonded, otherwise brittle materials.36,37 The local stresses are strongly dependent on the particle morphology, porosity, and mechanical properties of the materials. The interparticle localized contacts create stress states, which include both mean-pressure, normal-, and shear-stress components. Further, the total mean compression from the initial extended state to the fully dense state typically, 100% must be accomplished through viscoplastic shear deformation. The local thermal environment caused from localized deformation results from a balance between the mechanical energy and the thermal conductivity. The viscous input mechanical energy rates are balanced by the heat ow away from the local site. As a sample compresses from the initial to the nal state, the stress conguration and stress state will be altered qualitatively. From a mechanochemical viewpoint, plastic deformation of the particles will result in substantial concentrations of defects, which will serve to greatly enhance the solid-state reactivity.35 Observations are consistent with the maximum effects possible in solids.36,37 Melting itself is controlled by defect concentrations and defect congurations. Hence, the melt phenomena will also be controlled by the shock-formed defect state in the solid. It is interesting that from the mechanochemical paradigm, any tendency to soften or melt one of the components will reduce the interparticle shear stresses resulting in less overall plastic deformation and reduced tendency for reactions to occur in the duration of the shockcompression state. Observations from time-resolved measurements15 and recovery experiments34 support this view. It is, of course, no doubt that interparticle ow of the melt phase due to shock-compression and capillarity, may favor postshock or shock-assisted reactions occurring via dissolution and reprecipitation or defect-enhanced solidstate diffusion processes. Materials and morphological features ii above further complicate the problem. These morphological features control the interparticle stresses, which in turn, depend explicitly on the mechanical properties of each material system. Thus, a multicomponent mixture of particles will result in different local stresses than that of the same conguration of single components. Equilibrium mixture theory based on data from single-component materials samples would, therefore, not be expected to be descriptive of the mixed system. This situation is well demonstrated in shock-compressed samples of aluminum-hematite71 and aluminumtungsten oxide.72 In the
Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

aluminum-hematite sample, shock compressed in a MBB xture, the hematite is observed to deform around the aluminum particles and form the matrix phase. In the tungsten oxide mixture, the aluminum is observed to deform around the oxide. Whether the effect is due to morphological factors in the tungsten oxide case is not known, but the observation indicates that the total plastic deformation cannot be assumed to be a xed value for a given porous material. The shock-deformation process is substantially different from other intense deformation process as shown in comparisons between intensely ball-milled powders and shockmodied powders.73 The condition of 50% dense solids producing plastic deformations of about 100% at rates of 106 107 s 1, yields a strain versus strain-rate space not accessed in any other deformation experiment. Mechanochemical effects observed in mechanical alloying and shock synthesis also reveal striking differences. Data on paramagnetic defects in shock-modied rutile,73 and chemical changes observed in shock-modied MnO2, 74 provide explicit evidence conrming the uniqueness of the defects produced by the shock-compression process. The results of the present work reveal an extraordinarily low-threshold stress to initiate reaction in the titanium silicon system compared to other systems such as aluminum nickel,24,34 zincoxide hematite,23 and aluminum hematite,51 and raise the question of whether the materials factors controlling stress threshold can be identied. In general, the case can be made that the ease of plastic deformation will lead to improved mixing of potential reactants. Thus, low elastic limits and high viscosity ow should act to promote reaction. In the case of titanium or silicon, neither has exceptionally low strength, indeed silicon has a very high Hugoniot elastic limit. The main distinguishing feature of these two materials is that they both experience polymorphic transformations at pressures in the vicinity of 10 GPa. Even though the mean pressure for initiation is less than this value, local stresses are likely to be in the transformation pressure range. Certainly, volumetric changes accompanying transformations will act to promote materials deformations and could lead to better mixing. Nevertheless, it is not clear what materials behaviors have led to the low-threshold stress for chemical reaction in this system. Under rapid viscous ow, density differences between substituents although not very signicant in the present case 4.5 and Si 2.3 g/cm3 is expected to play a critical Ti role in promoting instabilities, which can lead to materials mixing. In that regard, any materials or morphological feature that tends to promote heterogeneous motions will act to accelerate chemical reaction. Indeed, once reaction occurs, the local reaction products themselves with their different densities and temperature will lead to greater heterogeneity and act to accelerate the reaction process. These mechanochemical characteristics are consistent with the processes dened by the CONMAH conceptual model, which provides a framework from which various scientic models can be developed. Certain critical aspects of the CONMAH processes are contained in the Horie VIR model.32 The local stresses and mechanical deformations can apparently be realistically treated in the Baer mixture theory models.31
J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

The present paper uniquely combines traditional shock compression and materials science techniques to develop a more detailed picture of the chemical processes. It is, nevertheless, important to recognize that the present work follows a continuous path from the early experiments on ZnOhematite,23 aluminumhematite,51 and nickelaluminum24,34 investigations, all performed with the use of standardized sample preservation xtures, which strengthens the comparative nature of the work and gives condence in the subtle conditions encountered. Taken together, the various studies represent a broad and thorough investigation of the fundamental aspects of shock-induced solid-state chemistry. These studies demonstrate that in description of shockinduced solid-state chemistry, the question is no longer whether, but, the extent to which. The question is not whether shock compression can initiate complete chemical reaction in the time of a typical shock-compression process, but under what conditions will these reactions occur. Certainly, the reaction space is very limited, not unlike all aspects of solid-state chemistry.
ACKNOWLEDGMENTS

The authors acknowledge the support provided by Sandia National Laboratories DOE Contract No. DEAC0494AL86000 and the Army Research Ofce Grant No. DAAHO4-93-0062 at Georgia Institute of Technology. The shock recovery experiments were performed using the EMRTC explosives ring facilities at New Mexico Tech with the assistance of Marvin Banks.
G. Duvall, Chairman, Shock-Compression Chemistry in Materials Synthesis and Processing, National Materials Advisory Board Report No. 414 National Academy Press, Washington D.C., 1984 . 2 R. A. Graham, B. Morosin, E. L. Venturini, and M. J. Carr, Annu. Rev. Mater. Sci. 16, 315 1986 . 3 R. A. Graham, Solids Under High Pressure Shock Compression: Mechanics, Physics, and Chemistry Springer, New York, 1993 . 4 N. N. Thadhani, Prog. Mater. Sci. 37, 117 1993 . 5 A. N. Dremin and O. N. Breusov, Russ. Chem. Rev. 37, 392 1968 . 6 R. A. Graham, in Proceedings of the 3rd International Symposium on High Dynamic Pressures, edited by R. Charet Assoc. Francaise de Pyrotechnie, Paris, 1989 , pp. 175180. 7 N. N. Thadhani, J. Appl. Phys. 76, 2125 1994 . 8 S. A. Shefeld and A. C. Schwarz, in Proceedings of the Eighth International Pyrotechniques Symposium, edited by A. J. Tulis IIT Research Institute, Chicago, IL , 1982, p. 972. 9 A. N. Kovalenko and G. V. Ivanov, Combust. Explos. Shock Waves USSR 19, 481 1989 . 10 S. S. Batsanov, G. S. Doronin, S. V. Klochkov, and A. I. Teut, Combust. Explos. Shock Waves USSR 22, 134 1986 . 11 S. S. Batsanov, M. F. Gogulya, M. A. Brazhnikov, E. V. Lazareva, G. S. Doronin, S. V. Klochkov, M. B. Banshikova, A. V. Fedorov, and G. V. Simakov, Sov. J. Chem. Phys. 10, 1699 1991 . 12 M. F. Gogulya, I. M. Voskoboinikov, A. Yu Dolgoborov, N. S. Dorokhov, and M. A. Brazhnikov, Sov. J. Chem. Phys. 11, 343 1992 . 13 K. R. Iyer, L. S. Bennett, F. Y. Sorrell, and Y. Horie, in High-Pressure Science and Technology1993, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross, AIP Conf. Proc. No. 309 AIP, New York, 1993 , Pt. 2, pp. 13371340. 14 L. E. Bennett, K. R. Iyer, Y. Horie, in Shock Compression of Condensed Matter1991, edited by S. C. Schmidt, R. D. Dick, J. W. Forbes, and D. G. Tasker Elsevier, Lausanne, 1992 , pp. 605608. 15 M. Yoshida and N. N. Thadhani, in Shock waves in Condensed Matter 1991, edited by S. C. Schmidt, R. D. Dick, J. W. Forbes, and D. G. Tasker Elsevier Science, Lausanne, 1992 , pp. 586592. 16 E. Dunbar, R. A. Graham, G. T. Holman, and M. U. Anderson, and N. N. Thadhani et al. 1127
1

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

Thadhani, in High-Pressure Science and Technology1993, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross, AIP Conf. Proc. 309 AIP, New York, 1994 , Pt. 2, pp. 13031306. 17 Z. A. Munir and U. Anselmi-Tamburini, Mater. Sci. Rep. 3, 277 1989 . 18 B. R. Kruger, A. H. Mutz, and T. Vreeland, Jr., Metall. Trans. 23, 55 1992 . 19 M. A. Meyers, L. H. Yu, and K. S. Vecchio, Acta Metall. Mater. 42, 701 1994 ; ibid. 42, 715 1994 . 20 T. Vreeland, Jr. and A. H. Mutz, California Institute of Technology, Pasadena, California unpublished results . 21 R. A. Graham, M. U. Anderson, Y. Horie, S.-K. You, and G. T. Holman, Shock Waves: Int. J. 3, 79 1993 . 22 L. S. Bennett and Y. Horie, Shock Waves: Int. J. 4, 127 1994 . 23 R. A. Graham, B. Morosin, Y. Horie, E. L. Venturini, M. Boslough, M. J. Carr, and D. L. Williamson, in Shock Waves in Condensed Matter1985, edited by Y. M. Gupta Plenum, New York, 1986 , pp. 693712, 797 802, 803808, 809814, and 815820. 24 Y. Horie, R. A. Graham, and I. K. Simonsen, in Metallurgical Applications of Shock-Wave and High-Strain-Rate Phenomena, edited by L. E. Murr, K. P. Staudhammer, and M. A. Meyers Dekker, New York, 1986 , pp. 10231035. 25 R. B. Schwarz, P. Kasiraj, T. Vreeland, Jr., and T. J. Ahrens, Acta. Metall. 32, 1243 1984 . 26 W. H. Gourdin, Prog. Mater. Sci. 30, 39 1986 . 27 K. Kondo and S. Sawai, J. Am. Ceram. Soc. 73, 1985 1990 . 28 M. M. Carroll and A. C. Holt, J. Appl. Phys. 52, 2812 1981 . 29 V. F. Nesterenko, in High-Rate Deformation of Heterogeneous Materials Novosibirsk: Nauka-Sib. Division, 1992 , pp. 551 in Russian . 30 G. J. Ravichandran, J. Appl. Phys. 74, 2425 1993 . 31 M. Baer, in High-Pressure Science and Technology, 1993, AIP Conf. Proc. 2, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross AIP, New York, 1993 , pp. 12471250. 32 Y. Horie and A. B. Sawaoka, Shock-Compression Chemistry of Materials KTF Scientic, Tokyo, 1993 , Chap. 6. 33 W. F. Hammetter, R. A. Graham, B. Morosin, and Y. Horie, in Shock Waves in Condensed Matter, 1987, edited by S. C. Schmidt and N. C. Holmes Elsevier Science, Lausanne, 1988 , pp. 431436. 34 E. Dunbar, N. N. Thadhani, and R. A. Graham, J. Mater. Sci. 28, 2903 1993 . 35 R. A. Graham and N. N. Thadhani, in Shock Waves in Materials Science, edited by A. B. Sawaoka Springer, New York, 1993 , pp. 3565. 36 B. Morosin and R. A. Graham, Mater. Sci. Eng. 66, 73 1984 . 37 B. Morosin, in High-Pressure Explosive Processing of Ceramics, edited by R. A. Graham and A. B. Sawaoka Trans Tech, Switzerland, 1987 , pp. 283339. 38 T. E. Royal, N. N. Thadhani, and R. A. Graham, in Metallurgical and Materials Applications of Shock-Wave and High-Strain-Rate Phenomena, edited by L. E. Murr, K. P. Staudhammer, and M. A. Meyers Elsevier Science, Lausanne, 1995 , pp. 629636. 39 G. E. Duvall and R. A. Graham, Rev. Mod. Phys. 49, 523 1977 . 40 Properties and Selection: Non-Ferrous Alloys and Pure Metals, Ninth ed., ASM Metals Handbook, ASM International, Metals Park Vol. 2, 1982. 41 O. Kubaschewski, C. B. Alcock, and P. J. Spencer, Materials ThermoChemistry, 6th edition Pergamon, Oxford, 1993 . 42 S. P. Murarka, Silicides for VLSI Applications Academic, New York, 1983 . 43 S. B. Bhaduri, Z. B. Qian, and R. Radha Krishnan, Scr. Metall. Mater. 30, 179 1994 . 44 J. Trambukis and Z. A. Munir, J. Am. Ceram. Soc. 73, 1240 1990 . 45 S. S. Batsanov, M. F. Gogulya, M. A. Brazhinikov, G. V. Simakov, and I. I. Maksimov, Combust. Explos. Shock Waves 36, 361 1994 . 46 M. B. Boslough, Phys. Chem. Lett. 160, 618 1989 . 47 M. B. Boslough, J. Chem. Phys. 92, 1839 1990 .

48

X. Gao and F. Jing, in Proceedings of Condensed Matter Under Shock Pressures, Oct 1415, 1991, Bombay India. 49 X. Chen, X. Jin, and Y. Musang, Chin. J. High Pressure Phys. 3, 67 1994 . 50 M. U. Anderson, R. A. Graham, and G. T. Holman, in High-Pressure Science and Technology1993, AIP Conf. Proc. 309, Part 2, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross AIP, New York, 1994 , pp. 11111114. 51 G. T. Holman, R. A. Graham, and M. U. Anderson, in High-Pressure Science and Technology1993, AIP Conf. Proc. 309, Part 2, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross AIP, New York, 1994 , pp. 11191123. 52 S. P. Shefeld, R. L. Gustavsen, R. R. Alcon, R. A. Graham, and M. U. Anderson, in High-Pressure Science and Technology1993, AIP Conf. Proc. 309, Part 2, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara, and M. Ross AIP, New York, 1994 , pp. 13771380. 53 M. U. Anderson and R. A. Graham, in Shock Compression of Condensed Matter1995, edited by S. C. Schmidt and W. C. Tao, AIP Conf. Proc. 370 AIP, New York, 1996 , pp. 11011104. 54 W. H. Holt, W. Mock, Jr., M. U. Anderson, G. T. Holman, and R. A. Graham, in Shock Compression of Condensed Matter1995, edited by S. C. Schmidt and W. C. Tao, AIP Conf. Proc. 370 AIP, New York, 1996 , pp. 573576. 55 F. Bauer, U.S. Patent No. 4, 684, 337, August 4, 1987. 56 D. E. Wackerbarth, M. U. Anderson, and A. Graham, Sandia Report No. SAND92-0046, Feb, 1992. 57 R. A. Graham, M. U. Anderson, F. Bauer, and R. E. Setchell, in Shock Compression of Condensed Matter1991, edited S. C. Schmidt, R. D. Dick, J. W. Forbes, and D. G. Tasker Elsevier, Lausanne, 1992 , pp. 883886. 58 R. A. Graham, in High-Pressure Explosive Processing of Ceramics, edited by R. A. Graham and A. B. Sawaoka Trans Tech, Switzerland, 1987 , pp. 2964. 59 R. A. Graham and D. M. Webb, in Shock Waves in Condensed Matter 1983, edited by J. M. Asay, R. A. Graham, and G. K. Straub Elsevier Science, Lausanne, 1984 , pp. 211214. 60 R. A. Graham and D. M. Webb, in Shock Waves in Condensed Matter 1985, edited by Y. M. Gupta Plenum, New York, 1986 , pp. 831836. 61 S. L. Thompson, Sandia National Laboratories, Report No. SLA-73-0477, October 1973. 62 G. K. Williamson and W. H. Hall, Acta Metall. 1, 22 1953 . 63 L.-H. Yu and M. A. Meyers, J. Mater. Sci. 26, 601 1991 . 64 G. K. Williamson and R. E. Smallman, Acta Crystallogr. 7, 574 1954 . 65 M. B. Boslough, J. Chem. Phys. 92, 1839 1990 . 66 T. Akashi and A. B. Sawaoka, J. Mater. Sci. 22, 3276 1987 . 67 E. Dunbar, M. S. thesis, New Mexico Institute of Mining and Technology, 1992. 68 R. B. Frey, Eighth International Symposium on Detonation 1985 , pp. 385388. 69 Y. Horie, D. E. P. Hoy, I. Simonsen, and R. A. Graham, in Shock Waves in Condensed Matter1985, edited by Y. M. Gupta Plenum, New York, 1985 , pp. 749754. 70 R. A. Graham, B. Morosin, and B. Dodson, Sandia Report No. SAND 83-1987, October 1983. 71 R. A. Graham, Sandia National Laboratories, 1995 unpublished results . 72 R. A. Graham, Sandia National Laboratories, 1995 unpublished results . 73 B. Morosin, E. L. Venturini, and R. A. Graham, in Shock Waves in Condensed Matter, edited by W. J. Nellis. L. Seaman, and R. A. Graham American Institute of Physics, New York, 1982 , pp. 7781. 74 P. Newcomer, B. Morosin, and R. A. Graham, in X-ray Diffraction Study of Shock Modied Tetragonal Phases of SnO 2 and M nO 2 , edited by J. V. Gilfrich Plenum, New York, 1993 , pp. 595601.

1128

J. Appl. Phys., Vol. 82, No. 3, 1 August 1997

Thadhani et al.

Downloaded30Jun2004to130.207.165.29.RedistributionsubjecttoAIPlicenseorcopyright,seehttp://jap.aip.org/jap/copyright.jsp

Potrebbero piacerti anche