David J. Erskine and William J. Nellis- Shock-induced martensitic transformation of highly oriented...

8/3/2019 David J. Erskine and William J. Nellis- Shock-induced martensitic transformation of highly oriented graphite to diamo

1/5

Shock-induced martensitic transformation of highly orientedgraphite to diamondDavid J. Erskine and William J. NellisLawrence Livermore National Laboratoty, P.O. Box 808, Livermore, California 94550(Received 9 December 199 1; accepted for publi cation 3 1 January 1992)Shock-wave profiles of highly ordered pyrolytic graphite shocked normal to the basal plane ofthe graphite crystal structure have been measured. For graphite with sufficient orientationalorder a martensitic transformation to a diamond-like phase is observed with a transition onsetpressure 19.6hO.7 GPa, the stability limit of the graphite structure under shock compression.The minimum overpressure required for the transformation is not more than 6 GPa and thetwo-wave structure of the transition is overdriven to a single wave above 40 GPa.

INTRODUCTIONBecause of their great technological importance, theinvestigation of the carbon phases and how they transformfrom one to another under pressure is an important field of

materials research. A significant part of this research isaccomplished using shock waves, since in these experi-ments uniform and accurately determined pressures can beapplied over large sample volumes.It is well known that diamond is formed by the shockcompression of graphite. This process, which occurs in mi-croseconds, happens naturally in the impact of meteors,12within products of explosives,34 and by explosive compres-sion of powders.5*6 However, an important issue is whetherthe shock-induced phase transition of graphite to diamondis martensitic or diffusive. The relation between the crystalstructures of graphite and diamond indicates. that thephase transition should be fast and martensitic if shockpressure is applied perpendicular to the graphitic basalplane.78 Since the lattice planes of graphite are looselycoupled, small amounts of shear stress naturally existing inthe shock state are expected to induce displacive shearmotion between the planes so as to produce the diamondstacking sequence.Recently, several groups have measured the stability ofthe graphite lattice under static compression. Utsumi andYagi have found that single crystalline graphite begins totransform reversibly to a new phase at 18 .GPa, indicatedby a sudden increase in the optical transmittance. Takanoand Wakatsuki have observed two discontinuous changesin the sample volume with applied load, at 17-18 GPa and22-23 GPa. Measurements of the pressure dependent Ra-man spectra by Stishov et al. l1 indicate that graphite trans-forms to a new form above 20 GPa. X-ray diffraction mea-surements by Hemley et al. l2 have found that above 14GPa graphite transforms to a form with a strength com-parable to that of diamond, but with a structure that isneither cubic nor hexagonal diamond. These studies allsupport the observation that the graphite lattice is unstableabove -20 GPa and that the transformation to the newphase is reversible.Previous shock compression studies of graphite pre-sented an inconsistent picture. A consequence of a marten-

sitic transformation is a well-defined transition onset pres-sure. Shock studies13-15 using pressed porous graphitesamples indicate a phase transition near 20 GPa, al thoughinterpretation of the results is complicated by the hetero-geneous temperature distribution due to porosity. l6 In con-trast, other studies using pyrolytic graphite (a quasi-single-crystalline form) have observed phase transitions near 34GPa,17 45 GPa,13 or have failed to see it below 50 GPa.18Since pyrolytic samples are more ideal crystalline graphitethan porous samples, this wide variety of transformationpressures observed previously seemed inconsistent.Our preliminary measurements on pyrolytic graphiteof two different crystalline grades having different orienta-tional order indicate a strong sensitivity of the transforma-tion on graphite microstructure. Thus we believe the ele-vated transition pressures observed with pyrolytic samplesof earlier studies13,17Y18ere caused by insufficient order intheir microstructure, while their high density preventedsignificant shock heating which could provide a thermallyactivated transformation.20 The latter effect likely domi-nated the behavior of pressed porous samples,13-15allowinga transition onset pressure near the crystalline graphitevalue.Our studies are the first real-time shock-wave-profilemeasurements using highly ordered full-density pyrolyticgraphite. We report here the shock properties of pyrolyticgraphite over a range of pressures, exploring the pressure-volume relation of the compressed diamond-like state, con-firming the martensitic nature of the transformation, andrefining the value of the transition onset pressure ( P,4).EXPERlMENTAL TECHNIQUE

The locus of states achieved through shock is called aHugoniot. Because the graphite-diamond transition in-volves a relatively large volume change, the graphite Hugo-niot is expected to have a pronounced kink at the transitiononset. This will cause the shock-wave transiting the speci-men to bifurcate into two waves,l producing a two-stepstructure in the shock pressure and mass velocity profile.This is illustrated in Fig. 1.For pressures between PA and PO the shock occurs intwo steps. The first step represents graphite shocked to astate incipient to the transition at a pressure PA along the4882 J. Appt. Phys. 71 (lo), 15 May 1992 0021-8979/92/104882-05$04.00 @ 1992 American Institute of Physics 4882

Downloaded 28 Feb 2005 to 128.115.175.8. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp


2/5

Time

FIG. 1. Illustration of how a system undergoing a phase transition undershock produces a two-step wave-profile. The pr essure (P)-volume ( V)pIot shows the untransformed and transformed graphite Hugoniots i andii, respectively. The shock OAB produces a stepped wave profile in pres-sure vs time, and a corresponding one in mass velocity vs time.

untransformed Hugoniot i. The second step is a compres-sion into the phase-transformed state -represented byHugoniot ii. The shock speed of the steps ( U,,,U,) de-pends on the slope AP/A V of the jumps in pressure-volumespace. As the net shock pressure increases the speed of thesecond wave approaches the first until there is a single step,when the slope OD exceeds OA (the overdriven condi-tion).The speed of the phase transition is manifested in therisetime of the second step. If a set of simple displacementsexists between two crystal structures the transformation ismartensitic and will produce a fast risetime. Thus by mea-suring the temporal wave-profile we can determine themartensitic nature of the transition and through simpleshock relations we can determine the location of the Hugo-mot.Shock-waves were generated by impact of a Cu diskaccelerated by a two-stage light-gas gun to velocities ofseveral km/s striking a graphite specimen normal to thebasal plane. A schematic of the target is shown in Fig. 2.The specimen was backed by a LiF window through whicha laser beam was reflected off the graphite/LiF interface.Since LiF has a shock impedance similar to graphite, -theperturbation of the first shock by the interface i s relativelysmall. The laser is the-illuminating beam of our VISARinterferometer,22 which measures mass velocities (UP) via

grappte flyerVISA

shorting pin / -Al target body & 1

FIG. 2. Target schematic. The VISAR laser beam is Doppler shifted bythe movement of the LiF/graphite interface. An interferometer producesa fringe shift proportional to the interface velocity.

32.5

52

ia 2.i?0 1.5ak-jj 1

0.5 ~

-0.1 0 0.1 Tin!?e2(~s) 0.3 0.4 0.5

FIG. 3. Wave-profiles of~pyrolytic (ZYB) graphite. The curves have beenstaggered horizontally on the graph for clarity. The two-wave structure isdirect evidence for a phase transition.

the Doppler shift of the reflected light with a response timeof -2 ns. Eight electrical shorting pins flush with the frontsurface -of the sample provide a time zero for the start ofthe shock. The optical/electrical propagation delay of theVISAR is known to f 1 ns so that observation of shockarrival at the sample rear allows shock speed determinationto 0.5%. The advantage of our experimental method overother shock diagnostic techniques is that both shock transittimes and wave profile amplitudes can be accurately mea-sured. Both these pieces of information are necessary tocompute the pressure-volume states achieved duringshock-induced phase transformation.The majority of our samples were highly orientedmonochromator grade (ZYB) graphite from Union Car-bide measuring 3.5-4 mm thick and 15 mm square. Theirdensities were 2.254-2.259 g/cm3, very near the handbookvaluez3 for crystalline graphite of 2.265 g/cm3. Severalshots were also fired with a lower grade graphite (ZYH)from Union Carbide. This grade is distinguished from theformer by beta, the angular width of the (002) x-ray dif-fraction peak, which measures the spread of misalignmentof the crystallite c axes. For the high- and low-grade sam-ples beta is 0.8 and 3.5, respectively.

WAVE-PROFILERESULTSThe measured wave-profiles of several shots are shownin Fig. 3. The clean two-step structure demonstrates a mar-tensitic transformation by the rapid ( < 10 ns) risetime ofthe second steps. As the impactor speed increases the delaybetween the steps decreases until a single shock is pro-duced, analogous to the progression OAB, OAC, OAD inFig. 1. The amplitude of the tirst step is steady in time,constant from shot to shot, and the second step has a shortrisetime. Thus this progression of the wave-profile shapesagrees well with the description of a system undergoing a

martensitic phase transition for pressures above PA..4883 J. Appl. Phys., Vol. 71, No. 10, 15 May 1992 D. J. Erskine and W. J. Nellis 4883



3/5

Mass elocityUp) impactorFIG. 4. Construction in pressure (P)-mass velocity (UP) space for deter-mining compressed states A a nd B from measur ed wave profiles. LiF andCu are Hugoniots of window and impactor. Shock speeds (US,, U,) offirst and second waves determine slopes of OA and OB. The asteriskedstates represent the first and second waves reflecting off the LiF window.

HUGONIOTDATA ANALYSISThe wave-profile data was analyzed using standardrelationsz4 conserving energy, mass, and momentum acrossthe shock front to find the description of the states ofgraphite during its compression and transformation. Thestates corresponding to incipient transformation (A), posttransformation (B), and reshock off the LiF wi ndow (B*)were first found by the construction in pressure-mass ve-locity space illustrated in Fig. 4 and described below.We wish to determine states A and B from the mea-sured shock speeds ( Usl,Usz) and wave-profile amplitudes( UpA+UpBe) of the first and second steps, the impactorvelocity, and the known LiF and Cu Hugoniots. The as-

terisked states A* and B* represent the first and secondshock waves reflecting from the LiF window. Since theymust lie on the LiF Hugoniot they are determined by thecorresponding wave-profile amplitudes UpA* and UpBe.State A must lie on a line from the origin having a slopepoUs,, and its position along that line is determined by theslope of AA*. Since AA* represents a weak compression,we introduce a negligible error in assuming for all shotsthat the slope of AA* is given by the reflection of line ABfound in the lowest pressure shots, or - 10 GPa s/km.

The second wave propagates as state B, which must lieon the Cu impactor Hugoniot and on a line from Ahaving a slope PA J& = ( lJ3z UpA)/vA, where V,= ( 1 - Up;4/Usl)/po. Finally, from Up, P, and U, we cal-culate the specific volume of the states. The relevant mea-surements and calculation results are tabulated in Tables Iand II.HUGONIOTRESULTS

The shots are plotted in the pressure-volume domain inFig. 5. Except for the two highest velocity shots whichoverdrive the transition, a two-step wave-profile exists andthere are the following stages of compression: ( 1) initialshock to transition onset at pressure PA. This is held untilpassage of the second wave whereupon there is (2) a rapidtransition to the denser phase, and (3) reflection of thesecond wave off the LiF window. We note that the value ofPA between the four lowest velocity shots is very consistent(19.6hO.7 GPa) compared to the variety of values ob-tained with the lower grade graphite samples discussedbelow.Three shots using the lower grade (ZYH) graphitewere performed and the results, shown in Fig. 6 are quitedifferent than with the ZYB grade graphite. In contrast tothe sharp transitions seen with the higher grade, the ZYHmaterial shows frustrated transitions, manifested by theelevated and highly variable value of PA (2545 GPa), bythe finite transition risetime (tens of ns), and by the inter-mediate states between phases i and ii through which thetransformation proceeds.

DISCUSSIONNote that the first and second waves interact differ-

ently with the LiF window. Because the Hugoniots of LiFand untransformed graphite are very similar, there is verylittle reflection of the first wave at that interface. However,when the second wave reaches the window there is a sig-nificant reflection and shock-up in pressure. This is becausecompressed graphite immediately ahead of the secondwave is incipient to the phase transition, and thereforemuch more compressible than the window. Figure 5 showsthat the phase transformed graphite has a diamond-likecompressibility. This is indicated by the steep slope of the

TABLE I. Measured shot parameters for ZYB graphite. UPR and UPIp are wave-profile amplitudes behind the first and second steps.Impactor Sample Two wave speeds (km/s) Wave-profile amplitude (km/s)Shot velocity thickness Initial densityname (km/s) (-) (gdcm3 1 v,, US2 UpA UPP

pi 4.350 3.448 2.250gE 3.900 3.977 2.255gD 3.471 3.862 2.254gG 3.120 3.937 2.258gF 2.603 3.453 2.255g9 2.602 2.463 2.258ga 2.605 2.480 2.254BOverdriven. Only one shock exists.bWave-profile amplitude was too noisy for a reliable value.

ovrdrvn 8.077 ovrdrvn bovrdrvn 7.660 ovrdrvn 2.7047.077 6.922 1.223 2.4606.966 6.164 1.173 2.2026.873 4.859 1.110 1.8306.961 4.727 1.181 1.8506.931 4,759 1.164 1.812

4884 J. Appl. Phys., Vol. 71, No. 10, 15 May 1992 D. J. Erskine and W. J. Nellis 4884Downloaded 28 Feb 2005 to 128.115.175.8. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp


4/5

TABLE II. Compressed states calculated from the wave-profiles described by volume, pressure, and mass velocity.Reshock of 2ndIncipient transition Post-transitionShot wave off windowname u, (density - ) PA (GW UpA (km/s) 0.7 pB UPB UE pff

d ovrdrvna 0.2683 58.20 3.200 bgE ovrdrvn 0.2767 49.8 2.880 0.2744 63.30gD 0.3608 21.1 1.320 0.2782 41.0 2.603 0.2768 55.10gG 0.3623 19.9 1.267 0.2817 34.7 2.357 0.2798 47.30!zF 0.3660 18.6 1.200 0.2868 26.5 1.992 0.2843 36.90is9 0.3614 20.1 1.280 0.2876 26.8 1.984 0.2859 37.50ga 0.3629 19.7 1.261 0.2874 26.7 1.989 0.2842 36.40Qerdriven past the transition.bWave-proftle amplitude was too noisy for a reliable value.

Hugoni ot (ii) of second shocked states, as well as the steepslope of the portion of the compression histories corre-sponding to the reflection of the second wave off the LiFwindow.We note that the endpoin ts of the shock trajectories inFig. 5 are very reproducibly and consistently falling alonga Hugoni ot curve (ii) which is distinct from, but parallelto, the diamond Hugonio t.* The density of this state is- 5% less than that of diamond for an equivalent pressure.The density di fference may be ascribed to thermal pressureand to disorder of hexagonal diamond. Poor crystallizationwas found in hexagonal di amond recovered from staticcompression experiments of Bundy. In flash x-ray ditTrac-tion measurements during shock compression of pyrolyticgraphite, Johnson & Mitchell found that above 20 GPa nowell-defined diffraction peaks could be observed, whereaswith the same apparatus shocking graphitic BN, well-de-fined peaks of the phased transformed BN structure wereseen2

0.35Volume (ems)FIG. 5. Pressure-volume histories of high grade (ZYB) pyrolytic graph-ite. In the two highest velocity shots the transition is overdriven by asingle shock which then reflects off the LiF wi ndow. In the other shots theshock splits into two waves and the pressure is held at the incipient valueP.4until passage of the second wave. Shot g8 was omitted for clarity sinceit is almost identical to g9. The reelected portion of the data for shot giwas not discernable due to noise. For comparison the measured Hugoniotof diamond (see Ref. 25), and of untransformed graphite (see Refs. 13,18, and 26) (i) is shown. Curve ii representing the denser phase is a guidefor the eyes.4885 J. Appl. Phys., Vol. 91, No. 10, 15 May 1992 D. J. Erskine and W. J. Nellis 4885

Another possibility is that Hugoniot ii represents anew form of carbon which is distinct from d iamond butdiamond-like in density and compressibility. A distorteddiamond form (n-diamond) has recently been recovered inshocked graphite by Hirai and Kondo. The static com-pression studies-l2 of graphite support the conclusion thatat least one new phase of carbon intermediate in densitybetween graphite and di amond exists. Shu et ai. I2 indicatethat this form has a diamond-like compressibility. That theshock compressed phase is electrically insulating has beendetermined by Mitchell et al29 They observed in real timethe electrical resistance of pyrolytic graphite shocked to 40GPa to increase by three orders of magnitude.The shock temperature in the transformed state is dif-ficult to determine wi thout knowing the cold-compressioncurve of the diamond-like state. In our previous reportwe estimated the temperature of this state using prelimi-nary wave-profile data and the cold compression curve ofdiamond. However , the data in this report imply that thecold-compression curve of the transformed state is signifi-cantly different from that of diamond, so that a reliabletemperature estimate is not possible.However , it is possible to calculate the temperature of

FIG. 6. Pressure-volume histories for the lower grade (ZYH) graphite.As opposed to ZYB grade graphite (Fig. 5), the PA of this material ishighly variable and exceeds 20 GP, and the transition risetimes are sig-nificantly longer. Curves i, ii are the same as in Fig. 5.



5/5

graphite at the transition onset. This was calculated to be(500-550 K) by integrating dT = (V, - V)dP/2C,+ [P/2C, - Ty/V]dV along the graphite Hugoniot,where CJ T) is from Ref. 30. The Gruneisen gamma wasassumed to range between its graphite and diamondvalueq3ty = 0.35 and y = 1.15, respectively. This temper-ature is low compared to the melting point3* of compressedgraphite ( - 4000 K) , and thus supports a martensitic tran-sition rather than a diffusive recons tructive transformation

to diamond, which occurs3 at high shock pressures at3000-4000 K.

CONCLUSIONSOur results support a martensitic transformation forhighly ordered pyrolytic graphite to a diamond-like statewith a 19.63tO.7 GPa transition onset pressure. This pres-sure represents the stability limit of the graphite lattice toshock loading, and is similar to the -20 GPa transitiononset pressures observed in static compressionexperiments.8-3 Martensitic transformations are only ob-

served with graphite of sufFicient orientational order. TheHugoniot of the transformed state lies parallel to the dia-mond Hugoniot (implying diamond-like compressibility),but with -5% less density. This suggests that the state isa disordered or distorted diamond-like form. The mini-mum overpressure needed to drive the transition has notbeen determined, but is as small as 6 GPa. The transition isoverdriven above 40 GPa. These experiments illustrate thediagnostic capability that is now available to investigateshock-induced phase transitions in the laboratory.

ACKNOWLEDGMENTSWe acknowledge Ku Moua for fine construction of thetargets. This work was performed under the auspices of theU.S. Department of Energy by the Lawrence LivermoreNational Laboratory under Contract No. W-7405-ENG-

48.

R: E. Hanneman, H. M. Strong, and F. P. Bundy, Sci ence 155, 995(1967).R. S. Clarke, D. E. Appleman, and D. R. Ross, Nature 291, 396(1981).N. R. Greiner, D. S. Phillips, J. D. Johnson, and F. Volk, Nature 333,440 (1988).4A. I. Lyamkin, E. A. Petrov, A. P. Ershov, G.-V. Sakovich, A. M.Staver, and V. M. Titov, Sov. Phys. Dokl. 33, 705 (1988).P. S. Decarli and J. C. Jamieson, Science 133, 1821 (1961).G. R. Cowan, B. W. Dunnington, and A. H. Holtzman, U.S. Patent No.3401019, 1968.S. Fahy, S. G. Lottie, and M. L. Cohen, Phys. Rev. B 34, 1191 (1986);35,7623 (1987).F. P. Bundy and J. S. Kasper, J. Chem. Phys. 46, 3437 (1967).W. Utsumi and T. Yagi, Science 252, 1542 (1991).OK. J. Takano and M. Wakatsuki, Jpn. J. Appl. Phys. 30, L860 (1991).A. F. Goncharov, I. N. Makarenko, and S. M. Stishov, Sov. Phys. JETP69, 380 (1989).J F. Shu, H. K. Mao, J. 2. Hu, Y. Wu, and R. J. Hemley, Bull. Am.Phys. Sot. II 36, 479 (1991).l3 R G. McQueen and S. P. Marsh, Behavior of Dense Media Under HighDynamic Pressures (Gordon and Breech, New York, 1968).14A V. Ananin, A. N. Dremin, G. I. Kanel , and S. V. Pershin, Zh.Phkl. Mekh. Tekh. Fiz., No. 3, 112 (1978).M. F. Gogulya, Fiz. Goreniya. Vzryva 25, 95 (1989).16R. L. Willia mson, J. Appl. Phys. 68, 1287 (1990).W. H. Gust, Phys. Rev. B 22, 4744 (1980).N. L. Coleburn, J. Chem. Phys. 40, 71 (1964).D. J. Brskine and W. 3. Nellis, Nature 349, 317 (1991).2oS. V. Pyaternev, S. V. Pershin, and A. N. Dremin, Sov. Combustion,Explosion, and Shock Waves 22, 756 (1986).2 Ya. B. Zeldovich and Yu. P. Raizer, Ph ysfcs of Shock Waves and High-Temperature Hydrodynamic Phenomena (Academic, New York, 1967),pp. 750-756.*rWl F. Hemsing, Rev. Sci. Instrum. 50, 73 (1979).23W. N. Reynolds, Physical Properties of Graphite (Elsevier, Amsterdam,1968), pp. 3-5.z4R. G. McQueen, S. P. Marsh, and .I. N. Fritz, J. Geophys. Res. 72,4999(1967).M. N. Pavlov&ii, Sov. Phys. Solid State 13, 741 (1971).2bD. G. Doran, J. Appl. Phys. 34, 844 (1964).Q. Johnson and A. C. Mitchell, Phys. Rev. Lett. 29, 1369 (1972).28H. Hiiai and K. Kondo, Proc. Jpn. Acad. 67, 22 (1991); Science 253,

772 (1991).2gA C. Mitchell, J. W. Shaner, and R. N. Keeler, Physica B 139 & 140,386 (1986).3oW. C. Morgan, Carbon 10, 75 ( 1972).M. van Thiel and F. H. Ree, Int. 3. Thermophys. 10, 227 (1989).32A. V. Kurdyumov, N. F. Ostrovskaya, and A. N. Pilyankevich, Sov.Powder Metallurgy Metal Ceram. 27, 32 (1988).

4886 J. Appl. Phys., Vol. 71, No. 10, 15 May 1992 D. J. Erskine and W. J. Nellis 4886Downloaded 28 Feb 2005 to 128.115.175.8. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

David J. Erskine and William J. Nellis- Shock-induced martensitic transformation of highly oriented...

Documents

Transcript of David J. Erskine and William J. Nellis- Shock-induced martensitic transformation of highly oriented...