American Mineralogist, Volume 71, pages 977-984, 1986 High-pressurecrystal chemistry of beryl (BerAlrSi.Otr) and euclase(BeAlSiO4OH) Rosnnr M. HaznN, ANonpw Y. Au, Llnnv W. FrNcnn GeophysicalLaboratory, CarnegieInstitution of Washington, 2801 Upton Street,N.W., Washington, D.C. 20008 Ansrnlcr Compressibilities and high-pressurecrystal structures of beryl and euclasehave beon determined by X-ray methods at severalpressures. Beryl (hexagonal,space group P6/mcc) has nearly isotropic compressibility; linear compressibilitiesperpendicular and parallel to the c axis ateB':1.72 + 0.04 x l0-a kbar-'and B,:2.10 + 0.09 x 10-akbar-'. The correspondingbulk modulus is 1.70 + 0.05 Mbar if the pressurederivative of the bulk modulus K'is assumedto be 4. Euclase(monoclinic, spacegroup P2r/a)has anisotropic compression,with maximum compressibrlity of 2.44 + 0.05 x l0-a kbar-l parallel to the unique monoclinic b axis, and minimum compressibility of 1.50 + 0.03 x 10-a kbar-' approximately parallel to [01]. The intermediate axis of compressionhas a magnitude of 1.96 + 0.05 x lO-akbar-'in the a-c plane.The bulk modulus of euclaseis 1.59 t 0.03 Mbar if K' is assumedto be 4. The bulk moduli ofBe, Al, and Si cation coordination polyhedrain beryl are all consistent with 1.7 Mbar, which is the crystal bulk modulus. Beryl compressionoccurs primarily by shorteningof cation-anion bond distances.In euclase,the 2.3-Mbar polyhedralbulk moduli are significantly greater than the observed 1.6-Mbar crystal modulus. Compression in euclase,particularly along the b crystallographicaxis, results from a combination of poly- hedral compressionand changesin interpolyhedral angles. INrnooucrroN calculations for the oxides of Be, Al, and Si by modified Among the more fundamental goals of mineralogy and electron-gasmethods. petrologyisthepredictionofphysicalpropertiesandphase The crystal structure of beryl [BerAlrSiuO,r:hexagonal equilibria from a knowledgeof structureand bonding. The P6/ mcc, Z : 2l has been reported by Gibbs et al. ( I 968) study of minerals must proceed on a broad front of ex- and Morosin (1972). It is characterizedby six-member periment and theory if microscopic structural character- rings of Si tetrahedra, crossJinked by Be tetrahedra and istics are to be related to macroscopicmineral behavior. Al octahedra (Fig. la). The tetrahedral rings lie in the One promising approach is the characteization and ra- (0001)planeandaresuperimposedtoformchannelsalong tionalization of properties-structural, thermochemical, the c axis. Though often classifiedas a ring-silicate, beryl elastic,and vibrational-for a group of closely related canalsobedescribed(Zoltai,1960)asathree-dimensional minerals. The system BeO-AlrOr-SiOr-HrO provides an framework of Be and Si tetrahedra (Fig. lb). ideal suite of large, well-formed, stoichiometric, and or- The structure of euclase [BeAlSiO.(OH): monoclinic dered crystalline phasesin a variety of structuretypes. All P2r/a, Z: 4l was described by Mrose and Appleman atoms are of low atomic number and are thus amenable (1962). Euclasehas a-axis chains of three-member rings to the proceduresof computational quantum chemistry. formed by interconnectedBe and Si tetrahedra (Fig. 2a), Only three types of cation coordination polyhedra-Be crossJinked by Al octahedra (Fig. 2b). Although techni- and Si tetrahedra and Al octahedra-provide the struc- cally an orthosilicate with isolated Si tetrahedra, euclase tural building blocks for most of the phases,thus facili- can be interpreted equally well in terms of chains of Be tating tests of the "polyhedral approach" for modeling and Si tetrahedra. mineral properties (Hazen, 1985). Recent studies, com- The principal objectives of this high-pressureresearch pleted as part of this integratedeffort, include vibrational on beryl and euclaseare (l) to determine pressure-volume spectroscopy (Hoering and Hofmeister, 1985), elastic equation-of-stateparameters and compression anisotro- moduli (Au and Hazen, 1985; Yeganeh-Haeriand Weid- pies by measuringthe pressurevariation of unit-cell di- ner, 1986), thermochemistry (Barton, 1986), and com- mensions, (2) to calculate polyhedral bulk moduli from parative crystal chemistry (Hazen and Au, 1986).Also in high-pressurestructure data in order to test the effectsof progress are modeling of Be polyhedral clusters by ab structure on these moduli, (3) to identifu geometrical initio molecular orbital proceduresand lattice dynamic changesin the beryl and euclasestructures that result in 0003-004x/86/0708-o977$02.00 977 978 HAZEN ET AL.: BERYL-EUCLASE HIGH-PRESSURE CRYSTAL CHEMISTRY Table l. Unit-cell parametersof beryl and euclaseat several Table 2. Beryl refinement conditions and refined atomrc pressures parameters BERYL 1 bar 19 kbar 36 kbar 57 kbar N@ber of obs (I>2o) 299 100 701 94 P(kbar) range c/a a(A) c( A) v(E3) R(z)t 3.1 3,5 3.8 4.r weishred R(Z)+ 2.6 2-8 3.0 Extinicion, .* (*ro5) e.;(z)* r.4(7) 5(1) 1.4(8) 0,001 45-55 9.214(1)* 9.194(1) 676.0(1) 0.99?8 1.001 6 0.001 25 9.208(3) (3) 9.188(3) 674 .7 0.9978 1 .0000 ParaDeter 15 25 9.183(3) 9.159(3) 669.1(3) a,99',t4 0.9918 1B 25 9.179(2) 9.157(3) 668.2(3) 0.99?6 0.9904 0.5 0.5 0.5 0.5 (5) 30 25 9.153(10) 9.13r 663.5(11)0.9965 0.9835 0 0 0 0 33 25 v 9.155(3) 9.118(2) 661.6(3) 0.9960 0.9806 z 0.25 025 0.25 o.25 36 25 9.153(3) 9.119(1) 661.6(3) 0.9960 0.9806 B 0.s1 (7) 0.3(2) 0.4(3) 0.8(4) 40 25 9.r38(ro) r.rr3fi ) 659.2(5) 0.9973 o.9771 43 25 9.135(6) 9.103(2) 657,9(7) 0.9964 o.9752 7/3 7/3 7/3 713 \7 9.r34(4) 657.0(3) 9.096(2) 0.9958 0.9238 2/3 2/3 2/3 2/3 57 14 9.127(3) 652.4(15)0.9931 v 9.064(r5) 0.9670 o.25 o.25 o.25 0.25 B o.sr(3) 0.9(1) 1.1(1) 1.0(1) EUCLASE 0.3876(1) 0-3882(3) 0,3883(3) 0.3893(4) v 0.1159(1) 0.1ls5(3) 0.1161(3) 0.1166(4) P(kbar) a(A) b(a) c( A) v(E3) 0000 B 0.44(3) 0.68(8) 0.87(8) 0.95(10) 0.3103(3) 0.3122(6) 0.3111(6) 0.3107(8) 0.001 4.7800(3) 14.322(1) 4,6335(2) 1oo.31o(5) 312.30(3) 1.0000 0.2369(3) O.2376(7) 0 .2377(6) 0. 2366(8) 21 \.759(1) t\.255(6) 4.612(r) 1o0.24(1t J07.89(10) 0.9859 v 0000 22 4.761(1) 14.254(4) 4.610(1) r00./7(1) J07.85(15) 0.9859 0.82(5) 0.58(13) 1.06(14) 0.30(17) 42 4.746(1) 14.189(|) 4.599(1) roo.r8(2) 304.82(26) 0.9760 B rl9 4.739(1 ) 14.1 58 (3 ) 4.589(1) r00.15(1) 303.11(1r)0.97a7 0.4985(2) 0.4987(4) 0.4984(5) O.4992(6) 62 4.730(1) r4.136(9) 4.580(1) 100.16(2) 3A1.\A(22) 0.9651 v o.74s6(2) 0.1459(4) 0.1450(4) 0.7452(s) z 0.14s3(1) 0.1449(5) 0.L447(5\ 0.1470(6) * Parenthesized figures represenl esd's. B 0.68(3) 0.78(10) 0.98(10) 0.72(72) + Measured unit-cel1 parafreLers nay depend systemabically on the range of 2e used in lheir determinalion (Swansonet a1., 1985). * Parenthesized figures represent esd's Only 25. data were used in equation-of-state deierminations, + ! = rtlrol-lFcll/:Fo weishted R = [tw(lI L-lF.l)2/:"F^zl!/2 compression, and (4) to relate structural changes in beryl pressure at high to the elastic moduli determined by Yoon Lehmann and I-arsen (1974) with an option for manual inter- and Newnham (1973). vention. Refinementconditions and refinedstructural parameters for beryl and euclaseappear in Tables 2 and 3, respectively. ExpnnrvrnxrAr METHoDS Refined anisotropic temperatureparameters and the magnitudes Specimendescription and orientation of thermal vibration ellipsoids for both minerals room appear in Tables 4 and 5.1 Crystals of synthetic beryl (variety emerald) were provided by at conditions Richard (Vacuum Ventures, M. Mandle Inc.). This material con- Data collection at high pressure forms to the ideal beryl composition, with the exception of 0.37 platelike were mount- wt9oCr, correspondingto 1.3olooccupancy ofCr at the AI octa- Flat, crystals approximately 40 ;rm thick pressure for X-ray with hedral site. The synthetic material was dehydrated at 800'C for ed in a diamond-anvil cell ditrraction, 4 h, yielding a sample with less than 0.3 moleculesof HrO per an alcohol mixture of 4:1 methanol: ethanol as the hydrostatic formula unit. pressuremedium and 5-10-pm chips of ruby as the internal pressure Crystalsofnatural euclasefrom Minas Gerais,Brazil (National calibrant. Pressure-celldesign, loading, operation, and (1 Museum of Natural History, Smithsonian Institution, specimen calibration wereas described by Hazen and Finger 982). Special gasket parts no. #121350),were provided by John White. Chemical analysis care was taken to avoid X-ray shielding by the of gasket pm of this colorless,gem-quality material by Barton (1986) showed the diamond cell. Large holes of400 were used, and pm no impurities. Crystal fragments of dimensions approximately the 100 crystals were well centered in the diamond cell 100 x 100 x 40 pm were used for X-ray diffraction studies at throughout the experiments.