SCIENTIFIC CORRESPONDENCE
hardness of stishovite has been reported Discovery of hardest known oxide to be 17-20.8 GPa along different direc tions7, but the sample used in this case SIR - Microhardness measurements on and is metastable under normal condi had been synthesized at about 9.5-10 GPa synthesized samples of stishovite, a high tions7; its bulk modulus, 298 GPa (ref. 8), and 1,200-1,400 °C. The transformation pressure phase of silica, show that it is the is significantly greater than that of alumi was not complete at this pressure, as later hardest oxide yet discovered. Among na, 252 GPa, which is itself a hard oxide. shown 11• The small amount of lower-pres polycrystalline materials, its hardness (33 Here, we report microhardness measure sure phases present drastically modified gigapascals, GPa) rivals those of the hard ments on synthesized stishovite. the indentation results. est materials. Despite many searchesl-4, We performed synthesis experiments Polycrystalline stishovite is now the no material with a measured hardness using a 1,200-ton uniaxial MA-8 multi hardest oxide known; it is harder than comparable to diamond or cubic boron anvil apparatus (see ref. 9 for method), alumina and boron oxides12• Other super nitride has been identified until now. taking Raman and X-ray diffraction (see hard polycrystalline materials include dia Hardness (H) of ionic and covalent ref. 10 for image plate system) measure mond and cubic boron nitride compacts; materials increases with bulk modulus1- 5_ ments. Microprobe analysis showed only their hardness is much lower than that Diamond has the highest known bulk silicon in the investigated region. We did of single crystals, with hardness values modulus, B = 444 GPa, and is also the the hardness tests, using a Knoop micro of 50 and 32 GPa, respectively4• Thus, hardest material known, with single-crys hardness tester (Shimadzu type M) with stishovite is among the hardest polycrys tal H = 90 GPa. It is followed by cubic talline materials known. boron nitride ( cBN), with corresponding Bulk moduli and Knoop hardness for Stishovite is thus potentially an impor values of B = 369 GPa and single-crystal polycrystalline hard materials tant technological material. It may be the H = 48 GPa (refs 2, 4). ----······--·· ----- first member of a new family of superhard Theoretical calculations have been Material B(GPa) H(GPa) materials: the metastable high-pressure used in this search for hard materials, B4C 200 30 phases of high-valence cation oxides with principally by looking for high-bulk 850 200 30 high bulk moduli induced by an increase modulus compounds. A semi-empirical SiC 248 29 of the coordination number. theory1 was applied to group IV elements Al 20 3 252 21; 19* J.M. Leger and 111-V and II-VI compounds: B = Si02 298 33* J. Haines Sintered cBN 369 32 N/4 (1,971-220/)d-3 5 (where N is the Laboratoire de Physico-Chimie des Sintered diamond 444 50 average coordination number, I is an Materiaux, empirical ionicity parameter and d the Values from ref. 4, except for 86 0 (ref. 12) CNRS, 1 Place Aristide Briand, bond length). However, no superhard and asterisks (this work). 92190 Meudon, France material has been synthesized as a result M. Schmidt* of these calculations and the stability of loads of 490, 980 and 1,960 mN, on Bayerisches Geoinstitut, the predicted compounds has not been quenched samples after polishing. The D-95440 Bayreuth, verified. hardness values obtained are indepen Germany Under ambient conditions, the bulk dent of load and those of a test sample of J.P. Petitet modulus of ionic compounds is given by a alumina are in excellent agreement with Laboratoire d'lngenierie des Materiaux et general relationship: B rxZaZJV, where values cited in the literature (see table). des Hautes Pressions, Za and Zc are the formal anion and We treated a sample of a-quartz at CNRS, Avenue J. B. Clement, cation charges respectively, and V is the 14 GPa and 1,000 °C; X-ray diffraction 93430 Villetaneuse, specific volume per ion pair6• High bulk indicated that the sample was >99% France moduli require high charges and small stishovite. Micro-Raman spectroscopy A. S. Pereira volumes; thus tetravalent cation dioxides indicated the presence of coesite and pos Esco/a de Engenharia & lnstituto could be hard. Under pressure, packing sibly quartz (line around 490 cm-1) at vari de Fisica, efficiency increases at phase transitions ous places on the sample surface; the lines UFRGS, 91501-970 Porto Alegre, at which the cation coordination number of stishovite were observed in all cases. RS, Brazil (Ne) rises. As structures become more The hardness measurements ranged J. A. H. da Jornada compact, compression becomes increas between 19 and 33 GPa. This scatter was lnstituto de Fisica, UFRGS, ingly difficult and the bulk modulus rises. attributed to the presence of small 91501-970 Porto Alegre, High-pressure phases may thus be hard amounts of quartz or coesite, both of RS, Brazil materials. The transformation from which have very low hardness, on the sur 1. Cohen. M. L. Science 261, 30 7- 308 (1993). graphite to diamond is the archetypical face of the sample, as revealed by the 2. Riedel, R. Adv. Mater. 6, 549-560 (1994). example. Raman spectroscopy. 3. Leger, J. M., Haines, J. & Blanzat , B. J. Mater. Sci. Lett. 13. 1688- 1690 (1994). Although silicon is the smallest To test this hypothesis, we synthesized 4. Sung, C. M. & Sung, M. Mater. Chem. Phys. 43. 1- 18 tetravalent cation, the common forms of another sample at higher pressure: 20 (1996 ). 5. Goble. R. J. & Scott, S. 0. Can. Mineral. 23, 273- 285 silicon dioxide are not hard because of GPa and 1,100 °C, using amorphous silica (1985). their open structures in the cristobalite as the starting material. The transforma 6. Anderson, 0 . L. & Nafe, J. E. J. Geophys. Res. 70, or quartz phases (Nc=4). However, the tion to polycrystalline stishovite was com 3951- 3957 (1965). 7. Stishov, S. M. & Popova, S. V. Geochemistry 10, 7 high pressure phase of silica, stishovite plete and we could detect no foreign 923- 926 (1961). (Ne= 6), could potentially be a hard phase by Raman spectroscopy or X-ray 8. Hemley, R. J. , Prewitt, C. T. & Kingma, K. J. Rev. Mineral. 29, 41-81 (1994). material. It is much denser than quartz, diffraction. The seven indentation mea 9. Rubie, 0 . c., Karato, S., Yan, H. & O'Neill, H. St. C. surements indicated a very high hardness Phys. Chem. Minerals 20, 315-322 (1993). ranging from 30.9 to 34.7 GPa, with an 10. Haines, J., Leger, J. M. & Schulte, O. Science 271, Scientific Correspondence 629-631 (1996). average value of 33 GPa, irrespective of 11. Sclar. C. B., Young, A. P., Carrison, L. C. & Scientific Correspondence is intended to the orientation of the indentor and the Schwartz, C. M. J. Geophys. Res. 67, 4049- 4054 provide a forum in which readers may load. We noticed no change in the Raman (1962). raise points of a scientific character. 12. Srikanth, V., Roy, R., Graham. E. K. & Voigt, 0 . E. J. Am. Ceram. Soc. 74, 3145-3147 (1991). Priority will be given to letters of fewer spectrum before and after indentation. than 500 words. The second experiment definitely * Present address: CNRS, URAlO. Universite Blaise Pascal, shows that stishovite is very hard. The 63038 Clermont-Ferrand, France. NATU RE · VO L 38 3 · 3 OCTOBER 1996 401