Compressibility and Crystal Structure of Andalusite at High Pressure
Total Page:16
File Type:pdf, Size:1020Kb
American Mineralogist, Volume 69, pages 513-519,l9A Compressibility and crystal structure of andalusiteat high pressure RussBrr L. Relpn Department of Geological Sciences, University of Washington Seattle, Washington 98195 Llnnv W. FIncEn, RoBEnr M. HezeN Geophysical Laboratory, Carnegie Institution of Washington Washington, D. C. 20008 nxo SusRAre GnosB Department of Geological Sciences, University of Washington Seattle, Washington 98195 Abstract The unit-cell dimensions and crystal structure of andalusite Al2SiO5have been refined from X-ray data on single crystals mounted in a diamond anvil cell at pressuresof 12,25, and 37 kbar. Structure refinementswith anisotropic temperaturefactors yielded weighted R factors of 3.4, 4.9, and5.2%o respectively. The bulk modulusof andalusiteis 1.35t0.10 mbar and the axial compressionratios of orthorhombic unit-cell axes a:b:c are approxi- mately2.1:1.5:1.0. The relativelygreater compressibility of the A(IFOD bond resultsin a beingthe most compressibleaxis. Those bondsthat compress)3obetween I bar and 37 kbar at room temperature, are the bonds that also expand significantly between 25 and 1000"C at room pressure. Polyhedral bulk moduli for the Al(l) octahedron, the Al(2) trigonal bipyramid and the Si tetrahedronare 1.3t0.2, 1.6!0.5, and 4.1+1.5 mbar' respectively. Thus, the aluminum polyhedra are significantly more compressiblethan the silicon tetrahedron. The omega step-scanningtechnique of X-ray intensity data collection results in a significant improvement in accuracy and is recommended for structure determination with the diamond-anvil high-pressurecell. Introduction spectroscopicdata, the phonon spectra ofandalusite and High pressurestructure determination contain valuable their temperaturedependence have been determined and data on the equations of state, interatomic forces and interpreted on the basis of a rigid-ion model by Iishi et al. chemical bonding in minerals. The Al2SiO5polymorphs, (1979). andalusite,sillimanite and kyanite, provide an interesting In this paper are presented data on the high pressure system where aluminum occurs in three types of coordi- structural response of andalusite to 37 kbar, as well as nations; in addition to the octahedral coordination found correlations between high-pressure changes and high- in all three minerals, aluminum also occurs in tetrahedal temperature response, elasticity, and phonon spectra. It coordination in sillimanite and in five-fold trigonal bipyra- has been possible to observe directly the structural ele- midal coordination in andalusite. Knowledge of the re- mentsof andalusitethat are responsiblefor the compress- sponse to pressure and temperature of these different ibility and, hence, the elastic constants. The present data types of Al-O bonds in the presence of relatively rigid on andalusite at high-pressure,combined with high-tem- SiOa tetrahedra is necessaryfor an understandingof the perature data of Winter and Ghose (1979),may be used to stability relations and phase transformation within the test the inverse relationship betwedn structural responses Al2SiO5system. The temperature efects on the crystal due to temperature versus pressure (Hazen and Finger, structures of andalusite, sillimanite and kyanite have 1982).Andalusite is a light-atom structure with a small been determinedby Winter and Ghose (1979).The elastic unit-cell and orthorhombic symmetry and thus also con- constants of andalusite and sillimanite have been deter- stitutes a test casefor the improvement in accuracy of X- mined by Vaughan and Weidner (1978), who used the ray intensity data collected at high pressureby the omega Brillouin scatteringtechnique. From infrared and Raman step-scanningtechnique. The resulting improvement in 0003-004)v84l0506-05l 3$02 .00 5I 3 514 RALPH ET AL.: ANDALUSITE AT HIGH PRESSURE the data resulted in a successful refinement of the aniso- A. Auiomotic bockgroundposltlons tropic temperature factors in a complicated silicate struc- ture at high pressure. Experimental A fragment of a gem-quality andalusite crystal from Minas Gerais, Brazil, (from the samesample used by Winter and Ghose (1979)for their high-temperatureexperiments) was usedfor high- pressurestudy. A square crystal plate approximately 100 x 100 x 50 g,m was selectedand mounted along with small ruby chips Iqsls in a diamond-anvilhigh-pressure cell of the Merrill-Bassett type "3893 (Merrill and Bassett, 1974)modified (1977). by Hazen and Finger B. Monuolbockground posltlons The pressuremedium was a 4: I methanol:ethanol mixture. The ruby fragmentswere used for the pressurecalibration by the R1 ruby flourescencetechnique. The andalusitecrystal was mount- ed with its (001) plane approximately parallel to the diamond faces and was securedwith a small dot of petroleumjelly. The first data set at 37 kbar was collected at the University of Washington on the Syntex Pf four-circle automatic diffractom- eter with the d-20 techniqueand MoKa radiation monochroma- tized by reflecting from a graphite "single" crystal. All other Is61s.341 .32i7 data s€ts were collected at the Geophysical Laboratory on an Isslg automatedfour-circle diffractometer and Nb-filtered MoKa rad! Fig. l. Comparison of manual and automated background ation. All accessiblereflections for which sin 0/)r < 0.70 were selection for evaluation of integrated X-ray intensities: (a) collected. The data were corrected for specimenand diamond- continuous omega scan, automatic background selection; (b) cell absorption as well as Lorentz and polarization factors. omega step-scan, manual background selection. Integrated Results of unit-cell refinements at the pressures studied are intensities procedures, shown in Table 1. for weak reflections are similar with both but strong reflections, which may have white radiation Data at 12 kbar w€re collected first with continuous omega shoulders, axe often overestimated with automatic background scansand background measurementswere determined from the selection. intensities at the extremes of the scan. (Omega scans axe necessaryin diamond-cell studies because beryllium, which is used to support the diamonds, produces powder rings that modified manually prior to integration. Reflections were again interfere with 0-20 scans.)Data from symmetrically equivalent corrected for absorption, and symmetrically equivalent reflec- reflections were averagedprior to least-squaresrefinement with tions were averaged. Refinement based on 312 observed data the program nrrNe4 (Finger and Prince, 1975).Neutral scattering from the step-scanprocedure yielded a significantly improved (3.4% factors of Cromer and Mann (1968)and coefficientsfor anoma- anisotropic model, with a weighted R factor of 3.4Vo lous scatteringfrom Cromer and Libermarl (1970)were used for unweighted). all atoms. Refinement of a model with anisotropic atoms (46 Ofi-line selection of background positions, as facilitated by parameters, including scale factor, extinction parameter, 14 om€ga step-scandata collection, is an improvement over auto- p positional and 30 thermal parameters), based on 380 observed mated backgroundcollection for at least two reasons. When averagedreflections (I > 2 o) yielded a weightedR factor of 4.3Vo filtered radiation is used with a moving-crystal fixed counter (7.2Votnweighted). technique, shouldersof intensity are observed in the vicinity of Intensity data for andalusiteat 12 kbar were recollected with the Bragg peak (Fig. l). As the crystal is rotated away from the stepwiseomega scans (4 secondcounts at0.025"intervals). Step- scan data were processedand displayed visually on a graphics CRT for each reflection; backgroundpositions were selectedby Table 2. Refinement conditions and refined extinction the technique of Lehman and Larsen (1974), but could be parametersfor andalusiteat high pressure lZ kbar 12 kbar 25 kbar 37 kbar Continuous Sterscan Steo-Scen Table l. Unit-cell parametersof andalusite l{umberof observatlons I> 2o) 380 312 2A7 351 t{eightedR(X) + 4.3 I barr 12 kbar 25 kbar 37 kbar 3.4 4.9 S.z 8(t) +i t.z 3.4 6-2 5.4 7.79W(71 7.7599(',r8)7.7382(10) 7.7ue(6) r* (xror) "(l) l.o(3) 2.8(4) 1.4(4) 0.8(3) b(A) 7.9031(r0) 7.873s(17) 7.8571(241 7.8375(r6) s.ss66(s) 5.s440(17) s.5338(40) 5.5262(r3) + "([) r"ightea R = [r{Fo - rrl2lx,roz]llz v(i3) 342.4s(6) 338.72(rs) 336.46(27) 333.71(l I ) i+ R-rllFol - IFcll/ rl Fo Parentheslzedfigures represehtesd's. r Fm Uinter and chose (1979) RALPH ET AL.: ANDALUSITE AT HIGH PRESSURE 515 Bragg position, the white radiation from the tube is ditrracted ture factors are adjustedin such a way that the discrepanciesfor into the counter first. Eventually the limits of the counter strong reflections are minimized at the expenseofweaker ones. aperture are reached and the only radiation observed is general As a result, the eflect on unweightedR is much greater than for scatteredradiation. Usually the scanlimits are chosento be large weighted R. in order to reduce sensitivity to small shifts in crystal orientation; A secondadvantage to manual selection of background posi- thus backgrounds are measured outside the shoulder regions. tions is the opportunity to view the shapeof each peak in detail. For weak reflections there is little difference in the intensities Anomalies in peak shapedue to diamond or ruby diffraction or measured with continuous versus step-scan procedures. The other interference, are quickly identified and those aberrent effect is significantfor a strong reflection for which the measured reflectionsmay be rejectedfrom the data set prior to refinement.