High-Pressure Crystal Chemistry of Cubanite, Cufers

American Mineralogist, Volume 77, pages 937-944, 1992

High-pressure crystal chemistry of cubanite,CuFerS,

CarnnnrNr McClvrmoNr* JrNvrrN ZruNG, RonBnr M. HlznNo Llnnv W. FrNcBn GeophysicalLaboratory and Centerfor High-PressureResearch, Carnegie Institution of Washington, 5251Broad Branch Road NW, Washington,DC 20015-1305,U.S.A.

ABSTRACT We have studied cubanite, CuFerS, (orthorhombic, space gtoup Pcmn, a: 6.46, b : 11.10, c : 6.22 A), using single-crystalX-ray diffraction in a diamond-anvil cell at room temperature from 0 to 3.68 GPa. Refinementswere performed aI 0, 1.76, and 3.59 GPa, and cell parameterswere measuredat 20 pressuresup to 3.68 GPa. The linear compressibilities of the a, b, and c crystalaxes are 0.00513(5),0.00479(5),and 0.00575(4)GPa ', respectively.Compressibility data were fitted to a Birch-Murnaghan equation of statewith parametersKo : 55.3 + 1.7 GPa with Ko' constrained to be 4. High-pressurerefinement data indicate that the principal responseof the crystal structure to compression is a reduction in Cu-S bond lengths, whereasFe-S bond lengths remain essentiallyunchanged. Results are compared to bulk moduli and polyhedral bulk moduli of other sulfides.

INrnonucrroN lography provides an excellent means for examining the effect of pressure on individual atomic configurations. The crystal structure of cubanite, CuFerSr, is ortho- Cubanite is a worthy candidate becauseof the interest in rhombic with ordered tetrahedral cation sites and S at- the effect ofpressure on intervalence transitions (see,for oms in approximately hexagonal closest packing. The example,Burns, I 98 l) and becausevery few sulfideshave structure was first determined by Buerger (1945, 1947) beenexamined with this technique.We undertook a crys- and described as slices of wurtzite structure parallel to tallographic study of cubanite at high pressurewith the (010) with widthb/2,joined such that every other slab is following goals: (l) to determine the relative axial com- inverted, resulting in edge-sharedpairs oftetrahedra. The pressibilities and bulk modulus of cubanite, (2) to ob- valenceof Cu has been establishedas * I on the basis of serve changesin cation-anion configurations with pres- neutron diffraction data (Wintenberger et al., 1974), sure through variations in bond lengths,bond angles,and whereasFe atoms appearto occupy adjacent edge-shared tetrahedral geometries,and (3) to examine the effect of tetrahedra as Fe2*and Fe3*,with rapid electron exchange pressrueon the electron structure of Fe, basedon changes betweenthem. Crystal structure refinementsindicate cen- in the atomic arrangements. trosymmetric spacegroup Pcmn, which implies that the two Fe atoms are in identical oxidation statesacross the ExprnrlrnNTAr, METHoDS centerof symmetry on a time-averagedbasis (Fleet, I 970; Roorn conditions experiment Szymanski, 1974). Mtissbauerspectra show only one site for Fe from 4.2Klo room temperature(Imbert and Win- A single crystal of cubanite was selectedfrom sample tenberger,1967; Greenwood and Whitfield, 1968),which R17907 (Morro Velho, Nova Lima, Brazil) provided by is consistent with Fe atoms in the same valence state on the National Museum of Natural History, Smithsonian the time scaleof the Mdssbauereffect (1.5 x l0-' s). Institution, Washington, DC. A rectangular prismatic From neutron diffraction data, Wintenbergeret al. (1974) cleavagefragment, approximately 30 x 60 x 100 pm, found a singleFe moment of 3.2 p,",consistent with rapid was used for the X-ray diffraction experiments. Data on electron exchange.Electrical resistivity data of Sleight and a crystal mounted in air were obtained with a Rigaku Gillson (1973) have shown that cubanite is a semicon- AFC-5 goniometer with a rotating anode generator pro- ductor, implying that rapid electron exchangeis confined viding MoKa radiation with a graphite monochromator, -9 -15 to pairs of edge-sharedtetrahedra. ourro (sin0)/)\: 0.1/A,with < h < 9, = k < < Goodenough (1980) has shown that the edge-shared 15, and 0 < / 8. The 212,212, and 060 reflections configuration of Fe tetrahedra should stabilize the Fe2*- were monitored as intensity and orientation standards Fe3*configuration. An interesting problem, therefore, is every 150 reflections; their intensities varied by up to the efect ofpressure on electrontransfer, sincethe crystal l.5olofrom the mean. A total of 2138 symmetry-allowed structure will inevitably change. High-pressure crystal- reflections were measured,of which 1504 were observed at I > 2or. Unit-cell parameters were determined from the positions of24 centeredreflections in the range 3l < * Presentaddress: Bayerisches Geoinstitut, Postfach l0 12 5 I, 20 < 39. Absorption correctionswere made with the Ag- W-8580 Bayreuth, Germany. nost program supplied as part ofthe diffractometer con- 0003404x/ 92109l 04937$02.00 937 938 MCCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY

TABLE1. Refinementconditions and refined parameters of cubanite at several pressures

0 GPa in air 0 GPa in cell 1.7GPa 3.6 GPa

Number of observations(, > 2o) 479 180 166 IAR R(%l 22 c.u 4.6 4.6 Weighted R (%) 29 ao 8.3 7.6 Extinction,r- (x 105) 4.0(21 5.4(5) 5.2(s) 0.6(1) Atom Parameter Cu x 0.5836(1) 0.5827(12) 0.5834(13) 0.5831(1 2) v V4 th V4 t/4 z 0.1227(11 0.1209(7) 0.11 68(7) 0.11 14(7) B 1.49(1) 1 90(10) 2.06(e) 1.67(9) Fe x 0.41479(8) 0.4124(9) 0.4118(9) 0.4066(9) v 0.41293(4) 0 4124(3) o.4127(2\ 0.4132(2) z 0.63664(7) 0.6377(s) 0.633s(4) 0.629e(5) B 0.92(1) 1.26(8) 1.50(7) 1.32(7) S1 x 0.58s7(4) 0 5e6(6) 0.5e6(6) 0.611(5) v Y4 V4 V4 z 0 7577(2) 0.7595(12) 0.7556(10) 0.7s03(11 ) B 0.96(3) 1.55(28) 1.43(26) 1.67(34) S2 x 0.4116(2) 0.4080(30) 0.4124(34) 0.4245(29) v 0.41545(8) 0.4158(8) 0.4151(7) 0.4146(8) z 0.26703(11) 0.2657(7) 0.2621(7) 0.2564(7) B 0.se(2) 1.42(15) 1.66(1 s) 1.19(1 6) trol software. Equivalent reflections were averagedin point refinements, furthermore, relied on reflections within a glovp mmm to give 681 unique reflections,of which 479 relatively narow 20 runge, from l8 to 33o, in order to with 1 > 2or were used for structure solution and refine- reduce systematic errors associatedwith measuring lat- ment. tice parameters at different 20 ranges (Swanson et al., l 985). High-pressure experiments In an effort to improve pressurecalibration, the pres- The single crystal (30 x 60 x 100)rrm studied at room sure cell was opened, and the cubanite crystal was re- pressurewas mounted in a diamond-anvil cell designed mounted with a cleavagofragment of calcium fluoride for single-crystal X-ray diffraction studies (Hazen and 100 x 100 x l0 prmthick as an additional pressurestan- Finger, 1982).An Inconel 750X gasketwith a hole 0.35 dard, along with ruby chips. The 2d values of fluorite 220, mm in diameter was centeredover one 1.0-mm diamond 202, and022 reflectionsprovide a sensitiveinternal pres- anvil, and the crystal was centered on the face with a sure standard(Hazen and Finger, l98l). Unit-cell param- small dot of the alcohol-insoluble fraction of Vaseline eterswere measuredat an additional l2 pressuresto 3.68 petroleum jelly. We used a mixture of 4:l methanol to GPa, and pressureswere measuredby both ruby and cal- ethanol as the hydrostatic pressuremedium, and several cium fluoride methods. At severalpressures only calcium l0 pm fragmentsof ruby were sprinkled near the crystals fluoride pressureswere obtained, owing to the failure of for internal pressurecalibration. Special care was taken the laser in the ruby calibration system. to ensure that a clearance of 0.1 mm was maintained When comparing unit-cell parametersat several pres- betweenthe crystalsand gasketwall. This precaution pre- sures,it is important to employ the sameset of reflections vents significant X-ray shadowing of the crystal by the for each measurement.Cell refinements at all presswes gasketwall. were determined with the same set of eight reflections: All high-pressureX-ray data were obtained on a Huber 440, 042, 033, 403, 442,442,460, and 270 (the eight automated four-circle difractometer with graphite Friedel pairs are also included by the eight-reflectioncen- monochromatized MoKa radiation. Unit-cell parameters tering routine). for the crystal were measuredat six pressuresto 3.6 GPa Intensity data were obtained for the cubanite at 0.0, with the procedure of King and Finger (1979), whereby 1.76, and 3.59 GPa, prior to the inclusion of the calcium several reflections are measured in eight equivalent ori- fluoride internal pressure standard. The rectangular entations. This procedure minimizes errors associated cleavagefragment of cubanite was mounted with the c with crystal centering and diffractometer alignment. All axis approximately perpendicularto the flat diamond fac-

TABLE2. Anisotropicthermal parameters ( x 103) for cubaniteat room conditions

R A P22 v23 UU e-51(15) 3.2s(5) 8.2s(14) 0 0 0 Fe s.53(11) 1.74(3) 6.27(10) -0.06(6) 0.37(10) 0.10(5) al 5.86(44) 1.79(9) 6.63(26) 0 0 0 S2 6.17(25) 1.s8(6) 6.18(15) -0.06(19) -0.43(17) -0.51(8) McCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY 939

Tnele 3, Cubaniteanisotropic thermal ellipsoids at room con- TeeLe5. Selectedinteratomic distances and anglesfor cubanite ditions at severalpressures

Rms 0 GPa 0 GPa Angle with respect to '1.7 amplitude Parameter in air in cell GPa 3.6 GPa Axis (A) Cu tetrahedron 1 0.127(1) 90 90 0 Cu-Sl 2.271(1) 2.250(1) 2.22s(9) 2.209(10) 0.142(1) 0 90 90 Cu-S1 2.295(3) 2.249(35) 2.23(4) 2.12(3) 3 0 142(1) 90 0 90 Cu-S2 [2] 2.327(1) 2.338(13) 2.306{13) 2.242(12) I 0.103(1) 65(10) 34(11) 11 1(6) Mean Cu-S 2.305(1) 2.294(1) 2.268(5) 2.203(51 2 0.107(1 ) 137(1 0) 57(12) 66(7) S1-Cu-S1 111.1(1) 110.4(7) 110.5(8) 109.0(7) 3 0.113(1) 57(6) 8s(6) 33(6) S1-Cu-S2[2] 107.6(1) 107.3(6) 106.9(7) 104.7(6) S1 1 0.106(3) 90 0 90 S1-cu-S2 [2] 112.9(1) 113.7(6) 113.9(7) 115.4(5) 0.11 1(4) 0 90 90 S2-Cu-S2 104.3(1) 103.8(7) 1O4.2(8) 106.6(7) 3 0 114(21 90 90 0 Fe tetrahedron S2 1 0.101 (2) 74(s) 51(5) 43(5) Fe-S1 2.249(21 2.28s(19) 2.276(21) 2.324(19) 2 0.11 4(3) 1s1(15) 62(13) 95(17) Fe-S2 2.259(11 2.216(19) 2.225(21) 2.238(18) a 0.11 9(2) 113(19) 128(11) 47(7) Fe-S2 2.290(11 2.311(11) 2.298(12) 2.272(11\ Fe-S2 2.300(1) 2.314(61 2.291(51 2.280(s) Mean Fe-S 2.275(11 2.282(71 2.273(8) 2.279(7) S1-Fe-S2 110.0(1) 1O7.7(4) 107.6(4) 106.0(4) '112.2(9) All reflections re- S1-Fe-52 110.3(1) 112.1(9) 114.8(8) es. accessible [approximately 300/oof S1-Fe-S2 110.4(1) 110.5(s) 109.8(6) 107.1(6) flections to (sin 0)n\ < 0.71 were collected with @ step S2-Fe-S2 104.8(1) 1O4.7(3) 105.7(3) 106.0(3) scansof 0.025" for 4 s per step. The fixed-@mode of data S2-Fe-S2 110.5(1) 110.3(4) 110.9(5) 113.1(4) S2-Fe-S2 110.6(1) 111.2(4) 110.4(4) 109.3(4) collection (Finger and King, 1978) was used to optimize Sl tetrahedron reflection accessibility and minimize attenuation by the S1-Cu 2.271(1) 2.250(10) 2.22s(s) 2.209(10) diamond and Be components of the pressurecell. Data S1-Cu 2.295(3) 2.249(351 2.23(4) 2 12(3) 51-Fe[2] 2249(2) 2.285(19) 2.276(211 2.324(19}, were corrected for Lorentz and polarization effects,crys- Cu-S1-Cu 111.8(1) 114.7(13) 114.7(14) 118.2(121 tal absorption, and X-ray absorption by the diamond cell Cu-S1-Fe[2] 109.4(1) 108.2(8) 108.2(8) 105.7(8) (Hazen Cu-S1-Fe[2] 109.5(1) 110.6(6) 110.q7) 112.6(5) and Finger, 1982). Fe-S1-Fe 107.1(1) 104.1(12) 1O4.O(14) 100.2(11) 52 tetlahedron Rnsulrs S2-Cu 2.327(1) 2.338(13) 2.306(13) 2.242(12) S2-Fe 2.259(1) 2.216(19) 2.225(211 2.238(18) Room-pressurerefi nernent S2-Fe 2.290(1) 2.311(11) 2.298(121 2.272(111 Refinement conditions, refined isotropic extinction co- S2-Fe 2.300(1) 2.314(6) 2.291(5) 2.280(5) Cu-S2-Fe 107.3(1) 107.4(5) 107.3(5) 106.2(5) efficient (Zachaiasen, 1967), refined positional parame- Cu-S2-Fe 111.8(1) 111.5(5) 112.4\51 114.3(5) ters, and equivalent isotropic temperaturefactors are list- Cu-S2-Fe 121.6(1) 120.5(8) 121.8(1 0) 124.9(9) parameters Fe-S2-Fe 124.1(1) 1252(6) 124.6(6) 123.6(6) ed in Table l. Refined anisotropic thermal Fe-S2-Fe 11 1 .5(1) 111.7(71 110.8(8) 107.3(6) are in Table 2, corresponding magnitudes and orienta- Fe-S2-Fe 75.2(11 75.4(3) 74.3(3) 74.0(3) tions of vibration ellipsoids appear in Table 3, and tab- Fe-Fe 2.800(1) 2.827(7) 2.773(7) 2.738(7) ulated calculatedand observedstructure factors are given Note-' Bracketed numbers represent multiplicity of bond distances or in Tables 4A-4D.1 ano|es. Single-crystal refinements of cubanite have been re- ported by Azaroffand Buerger(1955), Fleet (1970),and Szymanski (1974). The refined parameters agree most Linear compressibilities closely with the data of Szymanski (1974), who used a The variations in the lattice parametersof cubanite as sphericalcrystal to eliminate systematicerrors in the data a function ofpressure are listed in Table 6 and illustrated due to absorption, extinction, and Renniger effects.The in Figure l. The cell parameterscan be fitted to straight positional parametersagree to within 0.0004 in all cases. lines within the resolution of the data set: The vibration ellipsoids in the present data set are close - to spherical (Table 2), in agreementwith previous data. a: 6.458(r) 0.0328(3)P (1) Interatomic distances and angles are given in Table 5; b : rr.104(r)- 0.0s27(6)P (2) thesealso agreewith the data of Szymanski(1974). In particular, our data show the same systematicpattern of and Fe-Sbond lengths,which supports the presenceofan Fe- c : 6.220(r)- 0.03s4(2)P (3) Fe repulsion force, in contrast to the Fe-Fe attraction force suggestedby Fleet (1970). The linear compressibilitiesare as follows: 0": 5.13(5)x 10-3/GPa (4) pb: 4.79(5)x l0-3lGPa (5) I A copy of Table 4 may be ordered as Document AM-92-506 from the BusinessOffice, Mineralogical Society of America, 1130 and Seventeenth Street NW, Suite 330, Washington, DC 20036, U.S.A. Pleaseremit $5.00 in advance for the microfiche. p,: 5.75(4)x l0-3/GPa. (6) 940 McCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY

TABLE6, Cubaniteunit-cell parameters vs. pressure 1.01

P (GPa) Ruby CaF, a (A) b (A) c (A) Y(A1 o.ss 0 0 6.460(4) 11.058(2) 6.22s(8) 444.7(11) Lf; 0 0 6.456(7) 11.108(5) 6.224(2) 446.3(5) 0 0 6.451(8) 11 .101(3) 6.222(31 445.6(6) 0 0 6.459(5) 11.104(2) 6.220(2) 446.1(3) 8 o.sz 0.04 6.455(6) 11.102(2) 6.226(2) 446.2(4) q.) 0.38 0.25 6.446(1) 11.082(4) 6.207(1) 443.4(2) ' 0.26 6.451(6) 11.096(6) 6.215(3) 444.9(4) alao 1.07 0.87 6.427('t) 11.051(s) 6.185(1) 439.3(2) o b/bo 0.87 6.432(21 11.064(2) 6.193(1) 440.7(11 E o.e5 ^ 1.17 6.425(7) 11.043(2) 6.184(2) 438.8(5) c/co '1.2 1.36 6.420(1) 11.032(4) 6.177(1) 437.5(2) * 1.76 6.413(6) 11 .019(2) 6.167(2) 435.8(4) v/vo 1.76 6.3s8(8) 10.996(2) 6.150(2) 432.715) 0.93 1.78 1.51 6.401(1) 11.015(4) 6.154(1) 433.9(2) 01234 1.99 1.82 6.393(1) 11.016(22) 6.147(1) 432.9(9) 2.32 6.391(5) 10.988(2) 6.145(2) 431.5(3) Pressure(GPa) 2.34 6.381(2) 10.98s(2) 6.138(1) 430.2(2) Fig. 2. Relative compression of cubanite at high pressure. 2.88 6.364(1) 10.948(4) 6.113(1) 425.9(2) 3.08 6.359(7) 10.937(2) 6.109(2) 424.9(5) The solid lines through the data for a/ ao,b/ bo,and c/ co arelinear 3.22 3.28 6.359(1) 10.942(4) 6.108(1) 425.0(2) least-squaresfits to the data, whereasthe solid line through the 3.35 6.356(4) 10.929(4) 6.107(2) 424.2(3) V/ Vo data is a Birch-Murnaghan equation of statewith K : 55.3 422.9(3) 3.46 6.353(4) 10.915(5) 6.098(2) GPa and Ko' : 4. 3.59 6.344(4) 10.s1e(2) 6.098(1) 4224(3) 3.65 3.68 6.331(1) 10.892(6) 6.079(2) 419.2(3)

P : 3f(t + 2fv'It - 3f/2(4 - Ko',)l (7) The relative compressibilitiesare illustrated in Figure 2. The compressibilities of the a and b axes are similar, where the Eulerian frnite strain is given by 2/3- whereasthe c axis is approximately 150/omore compress- f : Yzl(V/V6) ll. (8) ible. This differencein relative compressibilitiesis easily understood in terms of the crystal structure. The approx- A least-squaresfit of the cubanite data to a third-order : imately hexagonal closest packing arrangement of S at- equation of stategives the valuesK 56.8 + 3.8 GPa, :2.2 oms in layersperpendicular to the c axis resultsin a larger Ko' + 1.0,and Vo:447.3 + 0.6 A3,whereas a : compressibility of the c axis. second-orderfit of the dala(Ko' 4) givesa bulk modulus of K: 55.3 + 1.7GPa. Zowas considered a fit parameter Bulk modulus because"0 GPa" values measuredbefore and after high- The V/Vo compressiondata were fitted to a third-order pressureexperiments are actually made at some pressure Birch-Murnaghan equation of state Taylor expansion of between 0.0 and 0. I GPa. The reason is that pressure energy in terms of Eulerian finite strain (Birch, 1952, fluid must be retained in the cell; otherwise the surface I 978): tension of expandingair bubbles in the pressurechamber can push crystals out of orientation. Our fitted value of Zo is in excellent agreement with previous data (Fleet, B 1970;Szymanski, 1974). The compressiondata are illustrated in Figure 2, where the theoretical curve corresponding to a Birch-Murnaghan equation of state with parameters K : 55.3 GPa and Ko' : 4 is included for comparison. The uncertainty of the bulk modulus is best illustrated l0 20 30 plot pressure Pressure (GPa) in a of normalized (F) vs. finite strain (/). The normalized pressurecan be written

63 450 C D F : P/l3f(r + 2f)",1 (9) 4q k and so Equation 4 becomes o E 430 F: Koll - 3/2f(4 - Ko',)l (10) 4m (Birch, 1978; seealso Jeanloz, 1981).The compression 410 1.0 20 30 40 1.0 2.0 30 4.0 data are illustrated in a plot of .Fvs. /in Figure 3. The Pressure (GPa) Pressure (GPa) horizontal line correspondsto Ko' : 4, and the degreeto Fig. 1. Compression of cubanite at high pressurefor (A) a which K is dependent on the value of Ko' can be clearly axis; (B) D axis; (C) c axis; and (D) unit-cell volume. The solid seen.It is evident that the simplifoing assumption Ko' : lines representleast-squares fits to the data. 4 (horizontal line) is consistentwith the data, and in view MCCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY 94\

80 P=0 GPo P=3'6GPo '70

^60 c1 (,s0 (100) fr 40

20 0 0.01 o.o2 0.03 f Fig. 3. Compressiondata of cubanite plotted on an -F (normalized pressure)vs. /(finite strain) plot. The line represents the best fit with a constrainedK"' : 4 and a fitted value of tr/": 447.3 43. ofthe uncartainties we feel that the data do not support a third-order fit (Ko' unconstrained). Fig.4. Polyhedralrepresentation ofthe (100),(0 l0), and(00 l) High-pressure structures layersofcubanite at 0 and 3.6GPa. The Fe tetrahedraare shad- grayand the Cu tetrahedraare shadedin light gray. Selectedinteratomic distancesand anglesare given in ed in dark The orientationof eachsection is shownby an arrowindicating Table 5, unit-cell parametersappear in Table 6, and poly- eitherthe c or a axis. hedral volumes and distortion indices for the four atoms (all of which are tetrahedrally coordinated) are in Table 7. When describing the changesof a crystal structure re- nates agree within about 2 esd, but the in-cell thermal sulting from high pressure,it is important to use a ref- parametersaverage almost 500/ogreater than those mea- erenceroom-pressure refinement based on intensity data sured in air. obtained using the crystal in the diamond cell. Significant The principal responseof the cubanite crystal structure differencesare typically observed,particularly in thermal to pressureis a dramatic shorteningof Cu-S bonds. From parameters,between refined parameters for room-pres- Table 5 the mean Cu-S distancecontracts between 0 and sure crystals in air and in the pressurecell (Tables l, 5). 3.6 GPa from 2.305 to 2.203 L (4.4o/A.This changecor- Systematic errors arise in the diamond cell data from respondsto a 13.4o/oreduction in the volume of the Cu absorption complexities and the limited accessto recip- tetrahedron (Table 7). In contrast,the mean Fe-Sdistance rocal space. In the case of cubanite, all atomic coordi- remains virtually unchangedat 2.279 A, whereasthe Fe- Fe distanceis shortenedfrom 2.800 to 2.738 A' (Z.Zoto). TABLE7. Polyhedralvolumes and distortion indices" of cubanite The effect on the crystal structure is a change in bond vs. pressure anglesto accommodatethe shrinking Cu tetrahedra.Fig- 0 GPa 0 GPa ure 4 illustrates a polyhedral view of the crystal structure in air in cell 1,7 GPa 3.6 GPa at 0 and 3.6 GPa. The (100) layer illustrates the smaller Cu tetrahedron Cu tetrahedron at 3.6 GPa accompaniedby a changein volume (43) 6.258(7) 6.16(8) 5.94(8) s.42(71 Cu-S-Fe bond angles,resulting in a relative tilting of the Quad.elong. 1.003(1) 1.004(18) 1.004(11) 1.008(10) tetrahedra.The degreeoftilting is best seenin the (010) Anglevar. 12.0 15.9 16.6 25.2 relative Fe tetrahedron layer, where the smaller size of the Cu tetrahedra vorume(43) 6.027(5) 6.08(6) 6.01(6) 6.04(s) to Fe is evident from the increasein the size ofthe gaps Ouad.elong. 1.0010) 1.002(8) 1.001(8) 1.004(7) betweentetrahedra. The (001) layer illustrates the closing Anglevar. 5.4 7.8 5.8 14.4 anglesto accommodatethe smaller Cu tetra- Sl tetrahedron of Fe-S-Fe vorume (43) 5.963(2) s.96(5) 5.84(5) 5.75(4) hedra at 3.6 GPa. Ouad.elong. 1.001(1) 1.003(19) 1.003(18) 1.009(16) Anglevar. 2.4 12.3 12.8 41.9 DrscussroN 52 tetrahedron volume (43) 5.408(3) 5.43(3) 5.27(3',, 5.07(3) Comparisonof cubanite bulk rnoduluswith other sulfides Ouad.elong. 1.095(1) 1.094(8) 1.100(29) 1.107(8) of cubanite appears anomalous be- Angle var. 309.9 308.3 325.1 346.5 The compression large differencein the relative compressibil- ' causeof the Robinsonet al. (1971). ities of the cation tetrahedra, and it is useful to compare 942 McCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY

TABLE8. Bulk moduliof some sulfidecompounds 0.04

Compound Ko(GPa) Ko' Reference 0.03 ZnS (wurtzite) 77(81 Simmonsand Wang(1971) i FeS, 148 b.5 Clendenenand Drickamer(1 966) 147.9(5) 4 Jephcoatet al. (1983) =- oo? 11 8(4) 4 (1984) Will et al. o CuS(IID FeS- 82(7) 4 King and Prewitt(1982) I CUS(IV) CoS. 11 5(4) 4 Will et al. (1984) - . 0.01 FeS2 CuS. s4(1) 4 Tak6uchiet al. (1985) + CoS2 * Sz' a Fes(troilire) Volume data from reference were fitted to second-order Birch-Mur- A FeS(MnP) naghanequation of state. 0.00 o cubuite(Fe-S) a cubmite(Cu-S) x HF(1981)corpiladon -0.01 the derived bulk modulus with data from other 0102030 sulfide , 12 compounds. Table 8 lists values obtained from the lit- (d)' I A' I erature and demonstratesthat the bulk modulus of cu- SZi \esut/ banite is low compared to similar sulfides. The closest Fig. 5. Polyhedral bulk modulus data for some sulfides.Val- structural analogue to cubanite is ZnS in the wurtzite ues from the previous compilation of Hazen and Finger (1981) structure, which has a bulk modulus that is 400/ogreater are indicated by crosses,and the line indicates the generalbulk than that of cubanite. The highest axial compressibility modulusrelationship Krd'/S'zzg": 750 GPa A'. ofcubanite is along the c axis, however, which is perpen- dicular to the wurtzite layers; therefore, a direct comparison of the structure compressibilitiesis probably not ap- K"d3/S2z.z": 750 GPa A3 (l l) propriate. The compressibility of cubanite is determined primarily by the compressibility of Cu'* tetrahedra, and where rK" is the efective polyhedral bulk modulus, d is one might thereforeexpect a similarity in the bulk moduli the cation-anion separation, 52 is an empirical ionicity of copper sulfides.The bulk modulus of CuS determined term, and z,and z"are the cation and anion formal charg- from single-crystalcompression experiments of Tak6uchi es,respectively. For sulfides,52 was determined to be 0.4. et al. (1985) agreesvery well with the bulk modulus of We have updated the compilation of Hazen and Finger cubanite, although it should be noted that the structure (1979) with recent data from experimentson sulfides;the of CuS consistsof triangular CuS, groups as well as CuSo data are listed in Table 9 and plotted in Figure 5. Note tetrahedra. The negligible compressibilities of the Fe-S that the slight expansion in the mean Fe-S distance with bonds observed in cubanite are not duplicated in either pressure(Table 5) results in a negative polyhedral bulk ofthe iron sulfidesFeS or FeSr.In pyrite, FeSr,the unit- modulus for the Fe tetrahedron,a value inconsistentwith cell translation length is twice the Fe-S, distance, so the the empirical Equation I l. From Table 9 it is evident bulk modulus directly reflects the reduction in cation- that the data for FeSr,CoSr, FeS (troilite), and FeS (MnP anion bond distance. High-pressuresingle-crystal refine- type) are in good agreementwith the relationship between ments of King and Prewitt (1982) have shown that the bulk modulus and volume; however, the data for CuS Fe-S distance in FeS also shortenswith pressure. and CuFerS,are not. From Figure 5 one seesthat the Cu In order to quantifu the comparison of the compress- tetrahedra are more compressiblethan predicted by the ibilities of the different sulfides, it is useful to consider relation, whereasthe Fe tetrahedra of CuFerS. are much the individual cation polyhedra. Hazen and Finger (1979) lesscompressible. As noted by Hazen and Finger (1979), showed that a relationship between bulk modulus and deviations from relationship l l tend to occur with sig- volume exists for a wide variety of cation coordination nificant covalenceof chemical bonds, which may provide polyhedra, including sulfide compounds. They found the an explanation for the deviation ofthe Cu tetrahedra. In generalrelationship the caseof the Fe tetrahedra, there is likely a large elec-

TABLE9, Polyhedralbulk modulifor some sulfidecompounds

Compound K" (cl)" (polyhedron) Structure type (d) (A) Kp (GPa) za s"zz. Reference. CUS(lll) covellite 2.191 (1 ) 64(2) 420(5) 1 CUS(lV) covellite 2.314(2) 32(1) 2 2 249(3) 1 FeS, pyrite 2.26(21 99(86) 2 2 716(1 82) CoS, pyrite 2.316(5) 108(28) 2 2 842(61) 2 FeS troilite 2.492(21 6e(9) 2 2 663(23) e Fe5 MnP 2.465(12) 77(46\ 719(112) 2 CuFerS3(Fe-S) cubanite 2.275(1) -965(2239) z.c 2 -5678(3842) this work CuFerS"(Cu-S) cubanite 2.305(1) 28(1) 1 2 423(6) this work " Referencesare (1) Tak6uchiet al. (1985);(2) Wiil et at. (1984);(3) Kingand prewitt (1982). McCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY 943 tronic contribution to the bulk modulus from the electron pressibility by the electron transfer interactions of Fe2t- exchangeoccurring between Fe2*and Fe3t. Fe3*, Alternatively, slight rotations and increaseddistortions 5. We infer that the nature of electron transfer between of all cation tetrahedra (Fig. 4, Table 7), traits not ob- Fe2*and Fe3*is essentiallyunchanged between 0 and 3.6 servedin other sulfide structuresstudied at high pressure, GPa on the basis of the constancy of Fe-S distances. may contribute to the lower than predicted compressibility of the Fe tetrahedra. AcxNowr-stcMENTS This work was performed while one of the authors (C.A.M.) was a at the GeophysicalLaboratory. The visit was funded Effect pressure Fe2*-Fe3*electron exchange visiting investigator of on by the Natural Sciencesand EngineeringResearch Council of Canada, The nature of the electron configuration of Fe is sen- through both the University ResearchFellow Program and the Operating sitive to cation-anion bond distances,since these deter- Grant Program High-pressure crystallographic studies are supported at Laboratory by the Camegie Institution of Washington' mine the relative strength of electronic interactions. In- the Geophysical by NSF grant EAR-8916709, and by the NSF Center for High-Pressure tervalence charge transfer interactions have been observed Research. to intensif! in oxides and silicates at high pressure because of the contraction of metal-anion distances (see RnrunnNcns crrED Burns, 1981, and referencestherein). The Fe-S distance in cubanite remains constant to 3.6 GPa, however, im- Azaroff L.V., and Buerger, M.J. (1955) Refinement of the structure of American Mineralogist, 40, 213-225. plying process largely cubanite, CuFe,S.. that the electron transfer should be Birch, F. (1952) Elasticity and constitution ofthe Earth's interior. Joumal unchangedby the application ofpressure. of GeophysicalResearch, 57, 227-286 It is interesting to speculateas to why Fe-S distances -(1978) Finite strain isotherm and velocities for single-crystaland remain essentially constant in cubanite with increasing polycrystalline NaCl at high pressuresand 300 K. Joumal of Geo- -1268. pressure.In FeS the Fe-S distance contracts ftom 2.492 physical Research,83, 1257 Buerger,M.J. (1945) The crystal structure of cubanite, CuFe:S:,and the to 2.452 A between0 and 3.3 GPa (King and Prewitt, coordination offerromagnetic iron. Journal ofthe American Chemical 1982),and in FeS, the Fe-S distanceshrinks from 2.26 Society,67, 2056. to 2.23 A between0 and 3.8 GPa, so it seemslikely that -(1947) The crystal structure of cubanite. American Mineralogist, the behavior of CuFerS,is related to the electron transfer 32,4r5-425 (1981) Electronic spectraofminerals at high pressure:How process.We Fe2+-Fe3+ Burns, R.G. can speculatethat the configura- the mantle excites electrons. In W Schreyer,Ed., High-pressurere- tion in cubanite is unusually stable, and that, further- searchesin geoscience,p. 223-246. E. Schweizerbart'scheYerlagsbuch- more, the energyis lowest near2.27 A, the observedFe-S handlung, Stuttgart. distance in cubanite. Compression of the crystal struc- Clendenen,R.L., and Drickamer, H.G. (1966) Latti.ceparameters of nine ofChemical Phys- ture, which is compatible with reduction in the Fe-S dis- oxidesand sulfidesas a function ofpressure.Joumal ics. 44. 4223-4228. tance, results in a higher relative energy of the electron Finger, L.W., and King, H.E. (1978) A revised method of operation of configuration, and therefore the Fe polyhedra resist com- the single-crystaldiamond anvil cell and the refinementofthe structure pression. This behavior may be compared to vivianite, of NaCl at 32 kbar. American Mineralogist, 63,337-342. which exhibits localized Fe2+-Fe3+electrons acrosspairs Fleet, M.E. (1970) Refinement of the crystal structure of cubanite and polymorphism of CuFerS,. Zeitscrift fiir Kristallographie, 132, 276- ofedge-sharedoctahedra. Initially, vivianite is easily ox- 287. idized, but it reachesa maximum level of Fe3* concen- Goodenough,J.B. (1980) Structural chemistry ofiron sulfides' Materials tration because of the stability of the p.:+-fsr+ pnir ResearchBulletin. 13. 1305-1314. (McCammon and Burns, 1980). Greenwood.N N.. Whitfield, H.J. (1968) Mdssbauereffect studies on cubanite (CuFe,S,)and related iron sulfides.Joumal of the Chemical So- ciety (London) A, 1697-1699. CoNcr,usroNs Hazen, R M., and Finger, L.W. (1979)Bulk modulus-volume relationship polyhedra. of Geophysical Research, 84, 6723- high-pressure for cation-anion Journal We conclude the following based on our 6728. crystallographicexperiments between 0 and 3.7 GPa: -(1981) Calcium fluoride as an internal pressurestandard in high- l. The c axis of cubanite is approximately l5olomore pressure/high-temperaturecrystallography. Joumal of Applied Crys- compressible than the a or b axes, consistent with the tallography,14, 234-236. -(1982) Wiley, New York. approximate hexagonal closest packing of S atoms per- Comparative crystal chemistry. Imbert, P., and Wintenberger,M. (1967) Etudes des propri6tes magn6- pendicular to c. tiques et des spectresd'absorption par effet Mdssbauerde la cubanite 2. The bulk modulus of cubanite is 55.3 + 1.7 GPa et de la sternbergite.Bulletin de la Soci6t6Frangais de Min6ralogie et with Ko' constrained to be 4. de Cristallographie,90, 299-303. phases 3. The principal structural responseto pressureis the Jeanloz,R. (198l) Finite-strain equation ofstate for high-pressure GeophysicalResearch Irtters, 8, 1219-1222. 4.40lobetween 0 and 3.6 shortening of Cu-S bonds by Jephcoat, A.P., Mao, H.K., and Bell, P.M. (1983) Pyrite: Hydroslatic GPa. The Fe-S distance remains essentiallyconstant. compressionto 40 GPa. Eos,64,847. 4. The compressibility of both Cu and Fe tetrahedra King, H.E, and Finger, L.W. (1979) Diffracted beam crystal centering do not agreewith the empirical relationship betweenbulk and its application to high-pressurecrystallography. Joumal ofApplied 4-378. modulus and volume of Hazen and Finger (1979); we Crystallography,12, 37 King, H.E., and Prewitt, C.T. (1982) High-pressureand high-temperature suggestthis disagreementmay be due to covalency of the polymorphism ofiron sulfide(FeS). Acta Crystallographica,B38, 1877- Cu-S and Fe-S bonds, as well as contributions to com- l 887 944 MCCAMMON ET AL.: CUBANITE CRYSTAL CHEMISTRY

McCammon, C.A., and Bums, R.G. (1980) The oxidation mechanismof covellite CuS under high pressureup to 33 kbar. Zeitschrift fiir Kris- vivianite as studied by M6ssbauerspectroscopy. American Mineralo- tallographie, 173, I 19-128. gist,65, 361-366. Will, G., Laute{ung, J., Schmitz, H., and Hinze, E. (1984) The bulk Robinson, K., Gibbs, G.V., and Ribbe, P.H. (1971) Quadratic elongation: moduli of 3d-transition element pyrites measured with synchrotron A quantitative measureof distortion in coordination polyhedra. Sci- radiation in a new belt type apparatus. Materials Resefich Society ence.172. 567-570. Symposium Proceedings,22, 49-52. Simmons, G., and Wang, H. (1971) Singlecrystal elastic constants.Mas- Wintenberger,M., Lambert-Andron, B., and Roudaut, E. (1974) D6ter- sachusettsInstitute of Technology Press,Cambridge, Massachusetts. mination de la structure magn6tique de la cubanite par diffraction neu- Sleight, A.W., and Gillson, J.L. (1973) Electrical resistivity of cubanite: tronique sur un monocristal. PhysicaStatus Solidi, A.26,147-154. CuFerS,.Journal ofSolid State Chemistry, 8, 29-30. Zachariasen,W.H (1967)A generaltheory of X-ray diffraction in crystals. Swanson,D.K., Weidner, D.J., Prewitt, C.T., and Kandelin, J.J. (1985) Acta Crystallographica,23, 558-564. Single-crystalcompression of 7-MgrSiOo.Eos, 66, 370. Szymanski,J.T. (l 974) A refinementofthe structureofcubanite, CuFerS, Zeitschrift fiiLrKristallogaphie, | 40, 21 8-239. MeNuscnrp'rRrcErvED DeceL{sen 13, l99l Tak6uchi, Y., Kudoh, Y., and Sato, G. (1985) The crystal structure of Mmuscnrsr AcCEPTEDMw 2l, 1992