Carwdian Mineralogist Vol. 16, pp. 131-137 (1978)
KESTERITE,Cu2(Zn,Fe)SnS+, AND STANNITE,Cuz(FeZn)$nsa, STRUCTURALLYSIMILAR BUT DISTINCT MINERALS*
S. R. HALLt, J. T. SZYMA(SKI eIn J. M. STEWART Mineral SciencesLoboratory, CANMET, Departtnent of Energy, Mines & Resources, 555 Booth Street, Ottawa, Ontario KIA ocl
AssrRAcr (12,/2,0)et S en 8g (0.7560(2),0.7566Q),0.8722(2)). Dans la structure de la stannite, les atomes se si- The crystal structure of kesterite from Oruro, tuent diff6remment:Cu en 4d (0,/2,/e), (FeZn) en Bolivia has been determined by single-crystal X-ray 2a (0,0,0),Sn en 2D(Yz,r/2,1/z) et S en 8t (0.7551(1), diffraction metlods and refined to an R value of 0.7551(1),0.8702(l)). Les diff6rencesstructurales 0.0,14 (atl data). Th_e mineral is tetragonal, a qui r6sultent de l'emplacementdes atomesde cuiwe 5,427(l), c 10.871(5)4, Z=2, space group /4 and expliquent pourquoi ces deux min6raux ont des composition Cu1.u(Zne.7sFeo.1rCd6.e1)Snq.esSa.oo. The mailles et des propri6t6s optiquesdistinctes. crystal .structure of the coexisting mineral sfannils Clraduit par la R6daction) has been refined to an.R value of 0.025 (all data). Stannite is tetragonal, a 5.449(2), c 10.757G)4,, Z-2, space grotp I42m and composition Cu1.sg- INrr.opucrroN (Fee.s1Zne.1sCdq.o2)Sns.qnSa.6s.The kesterite structure is characterized by a cell that is pseudocubic The minerals stannite and kesterite are recog- Qa = c). The Cu atoms are in the separate positions nized as separatespecies because of their differ- 2a (0,0,0) a\d2c (0,Vz,Ye).The (tu,ps;, Sn and S ent Fe to Zn compositional ratios, and their atoms are in positions2d (/2,0,/e),2b (Vz,Vz,O)and distinct optical and physical properties. These and 89 (.7560Q), .7566(2), .8722(2)), respectively. minerals occur at a large number of localities, In the stannite structure the Cu, (Fe,Zn), Sn and S often as twolphase intergrowths. The composi- atoms are in positions 4d (O,Vz,Yq), (0,0,0), 2a 2b tional and crystal data for a number of tlese (Yz,/z,Vz) and 8i (.7551(1),.7551(l), .3702(1)), minerals from different localities are being com- respectively. The structural differences due to the piled (in positioning of the Cu atoms account for the distino by Kissin & Owens prep.). The results nnil gells and optical properties of these two min- of their study tend to discount the possibility that erals. kesterite and stannite are members of pseudo- binary solid solution series, as was originally suspectedfrom the wide compositional range of SoMMAIRE Fe and 7n, because their cell dimensions are largely independent of. Fe/ Zn variation, La structure cristalline de la kesterite a €t6 d€- In order to confirm that kesterite was indeed termin6e sur un cristal d'Oruro (Bolivie) de compo- structurally distinct from a coexisting stannite, sition Cur.ee(Zne.6Fe6.2eCdo.or)Sn6.eeSa.n6;elle a 6t6 crystals were selected for X-ray structure anal- jusqu'i (en affin6e un r6sidu de 0.044 utilisant tou- ysis from the specimens analyzed by Kissin & tes les rdflexions). T6tragonale, avec a 5.427(l), c Owens (in prep.). The structure of stannite was 10.871(5)A, Z-2, elle possede la sym6trie t4. On (1934), a aussi affin6, jusqu'au rdsidu R=0.025 (sur don- determined by Brockway who assumed n6es complbtes), la structure d'une stannite co-exis- a stoichiometric composition CuzFeSnSa,but the tante de composition Cu1.ee(Feo.g1Zn6.1sCdo.o2)Sn1.ss-structure is redetermined here in order to pro- Sa.s6,t6tragonale avec a 5.449(2), c 10.757(3)A, vide a suitable comparison with that of kesterite. Z-2, de groupe spatial 142m. llne maille quadru- ple de la kesterite est pseudo-cubique (2a=c). Les atomes Cu occupent deux positions distinctes 2a EXPERIMENTAL (0,0,0) et 2c (O,r/z,Ye)i les autres atomes se placent comme suit: (Zn, Fe) en 2d (Y2,0,/s), Sn en 2D Kesterite An irregular crystal fragment of kesterite of approximate dimensions 0.25X0.10x0.06mm *Minerals ResearchProgram, ProcessingContribu- tion No. 57. was extracted from specimen 2R(2) Kissin & tPresent address: Crystallography Centre, Univer- Owens,in prep.), from Oruro, Bolivia. Gandolfi sity of Western Australia, Nedlands 6009, Western powder photographs of the fragrnent show that Australia. it is a single phase, and that it contains none of t3L 132 THE CANADIAN MINERALOGIST
TABLE1. CRYSTALDATA
K.este.rit.e Stannite Source: Oruro,Bolivia 0ruro,Bolivia gr. 2R(2),(Kjssin & Owens,in prep.) gr. 2R(3),(Kissin & 0wens,in prep.)
Composi : (F"0.812n0. 5n --(friii6liobeti on cur.gB(2n0. 73F"0. 2gcd0. o1)Sno. 99s4. 00 c'1. 99 18cd0.02) 1. 00s4. 00 analysis ) CelI dimensions: a = 5.427(I), e = 10.871(5)i a = s,44s(2), c = 10.757G)"A tematic absences: h+k+L= zyL+I h+k+t = 2n+1 Spacegroup: 14 (#82), z = 2 rlzn (r1,2r),z = 2
Absorption : u(Moro)= 151.3cm-1 u(Morro)= 147.5cm-1 Data: (I/U Or LwarospnereJ (1/16of Ewaldsphere) 1288measured 3 times 692measured 3 times 1119with I>o(I) 684with I>o(I)
the stannite with which it soexists. Precession monitor tlle crystal alignment and instrument photographs show the crystal to be single, un- stability. lwinnsd, and to have a diffraction pattern with The three octants of data were merged after systematically absent reflections when h+k+l= the application of generalized Gaussian absorp- 2ni1,, The absencesand the general intensity tion corrections (Gabe & O'Byrne 1970) into an equivalences of the pattern suggest the Laue asymm€tric data set containing the mean net in- group 4/ mmr?. In turn, this limits the space tensity Io"t atrd the r.m.s. deviation o(I). Negative group to i42m because of the as$umed sphaler- net intensitieswere set to zero. Of the 1288 inde- ite-like arrangement of sulfur and metal atoms, pendent reflections, 1119 had mean intensities which usually occurs in this type of structure. greater than
(each L/L6 of tle Ewald sphere) of stannite in- tensities. The seg ents collected were hkl, khl and khl with h>k, and they were merged into PrcposedModels wlth ReflnedAverage Temperature Factors an asymmetric data set of 692 reflections. Of and AgreenentValues tlese, 684 had mean net intensities above o(I), and an overall agreement factor )AI/X of (esterlte Stannlte 0.o21. fi l'lodel 1 llodel 2 i-42n 'oos i tion posltion Atom B(A'?) Aton B(i'?) Arom B(i'?)
Srnucrunr Sor-urroN AND RBFnIEIVTENT 2q (0,0,0)Zn,Fe Cu I.44 24 (0,0,0) Fe,zn 0.93 (!,rr,o) sn 0.55 Sn 0.65 b (\,\,0) sn 0.78
Kesterite 2c (0,!,k) cu Cu L.29 4d (0,k,!) cu 1.55 u (L,o,ra) Cu 0.99 Zn,Fe 1.U 8i (x,x'z) s 0.93 The consistently stoichiometric proportions of Cu, (Zn,Fe), Sn and S in kesterite for the range 89 0.85 s 0.85 of specimensand locations studied by Kissin & Owens (in prep.) suggestthat these atoms fully R (al1 data) 0.045 0,044 .P (a] l data) 0.025 occupy specific sites in the structure. As dis- 0.027 0.025 rD (a11 data) 0.017 cussed in the experimental section, the initial R 0.036 0.035 space-groupchoise of. I42m for kesterite neces- 0.026 0.025 sitated the placement of the Cu atoms in the 4d (0,y2,y4) position, and Sn and (Zn Fe) at either 2a (0,O,0) or 2b (Yz,Yz,0).Least-squares refine- composite (Zn,Fe,Cd) atom is not easy' and in- ment convergedrapidly with this model but pro- tercianging these two atomic types in a given vided thermal parameters for the Cu and (Zn, position affects the R value very slightly. Of the Fe) which were insonsistent with those of stan- iwo models given in Table 2, model 1 is equi- nite and other com,parable sphalerite-like struc- valent to spacleqroujp i42m. but with positions tures (Hall 1975; Szymaiski 1976, 1978; 2c and 2d independent, whereasmodel 2 is only Kudoh & Tak6uchi 1976). possible in space group /4. The final R values The unacceptability of the thermal parameters lor both structures are similar, but favor model was enhanced by the fact that the structural si- 2. Significantly, however, the thermal parame- milarity of. the AZm model with that of stannite ters for the secondmodel are consistentbetween was totally unexplainable: kesterite and stannite the two Cu positions, and are in the same pro- are found coexisting adjacently with no indica- portion as lhose observed in other sphalerite- tion of miscibility; there appears to be a com- iike structures (Hall 1975). For a final compa- positional discontinuity in the Fe/Zn ratio be- rison, the /4 intensity data were averaged be- tween them, and this is reflected in a discon- tween the hkl and kftl reflections, and model 1 tinuity in the sell parameters (Kissin & Owens, was refined in space group 142m. The R values in prep.). These factors are strongly indicative were higher than in space group 14 (R=O.446' of a structural break between stannite and kes- R-=0.Ot8), and the therrnal parameters for Va, terite, and the i42m assignment cannot explain Fe), Sn, Cu and S were L.46, 0.66, L.24 ar.d such a break. This reasoning suggests_thatthe 0.86#, respectively. These values are in poor correct space group may in fact be 14, as this agreement with those in similar structures' would permit a reordering of the metal atoms. Refinement with the full-matrix least-squares The data were subsequentlyrecollected assuming program CRYI-SQ (Stewart et al, 1972) em' the lower symmetry Laue group 4/ m and sev- ployed weights derived during the data-merge eral small but significant differences between process.Thi scattering-factor curves used were hkl and &ft/ intensities were obseryed.Structural ihose of Cromer & Mann (1968). The anomalous models were refined for the permutations of dispersion coefficients of Cromer & Liberman metal types in tle possible positions of /4. The (tf70) were used in the structure factor and two with best agtee.mentare shown in Table 2. least-squarescalculations to define the enantio- It should be pointed out that the composite (Zn, morphic configuration of the structure, with right-handed Fe,Cd) f curve, generated from the atomic scat- respect to the arbitrary choice of tering curves in the proportion indicated by the axJs selected at the time of data collection' microprobe analysis, differs from the cdpper I Analysis of the structure factor agreementin the curve by a maximum of. 0.32e and has a mean later stagesof refinement indicated the need for difference of O.11e over tle whole range used. an extinition parameter g (Larson 1970) in the Ilencen differentiating the Cu atoms from the least squares.The parameter g refined to 1.36X 134 THE CANADIAN MINERALOGIST
1ABLE3. ATOMICPARAMETERS The anisotropic tenperature factors are expressedin the fom: T. expt-2r2(U,.a*2h2+ Zur.a*brhk + ...)1, andthe valuesquoted are x 10{.
KesterJte xy t ur, 0zz ugg otz ut: uz3 posrtr0n 2a Cu 00 0 184(1)184(1)192(2) o 0 0 2b Sn \k 0 85(t) 85(1) 77(r) o o 0 2c Cu 0, t 171(8)171(8) 148(11) 0 0 0 2d Zn,Fe k0 b 146(8)146(8) 153(11) 0 0 0 -4(2) 8ss . /tou(4.1 . /loolzJ .8722(r) ro8(2) 106(2) 107(2) 4(2) 5(2)
Stanlj te
44 Atom x z utt uzz urs utz ut: uzg poslt]0n v 2a Fe,Zn 0 0 0 105(2)105(2) 139(2) 0 o o 2bSn\ r, 0 95(2) 95(2) 107(2\ 0 o 0 MCu0 t \ re0(3) 1e0(3)209(3) 0 0 o 8n s .7s51(1) .7551(1) .8702(1)118(2) 118(2)rr7(2) -4(2J 3(2) 3(2)
10i. The final atomic paxameters are listed in bles 4a and 4b, which are available at nominal Table 3. charge from the Depository of Unpublished Data, CISTI, National Research Council of Canada, Ottawa, Canada KlA 0S2. Stannite
The atomic paramete$ of stannite were re' Dnscnrprrou oF THE Srnucrunss fined by full-matrix least-squares, starting with tle model determined by Brockway (1934). Steps These analysesshow that kesterite and stan' in the refinement processwere identical to those nite are structurally distinct minerals. The prin- employed for kesterite and resulted in final R cipal structural difference is that in stannite, the valuesfor all data of R = 0.025 and R-=0.016. (Fe,Zn) atoms share the z=0 and' z=r/z melal The refined atomic parameters are listed in Ta- layers with Sn, whereas in kesterite these layers ble 3. Final g was 2,72x1'0'. The observed are occupied by Cu and Sn atoms (Fig. 1). This (10XfJ and calculated (10Xf'J structure fac- metal interchange means tlat the metal layers tors for stannite and kesterite are given in Ta- at z=Ve atd, z=/q, which in stannite are oc-
@ Cu /.tr\ \UJ Sn @ ZnFe N FqZn
(, s (o) SrANNfi-EG4zm) (b)KESTERITE(I4) Frc. 1. Diaexammaticrepresentation of the kesterite and stannite structures, emphasizingthe difference in metal ordering. The radius of the spheres is arbitrary. KESTERITE AND STANNITE 135
one sheetof the (111) plane of the pseudocubic ORDERINCIN SHEETOF THE substructure. This is shown in Table 4. PLANESIN CHALCOPYRITE,STANNITE AND KESTERITE. This metal re-ordering is a profound struc- tural chaqgq and would certainly account for a chalcopyrlte stannite kesterite the siguificant miscibility gap observedby Kissin 0 Cu Fe Cu Fe Fe 5n Fe Sn Cu Sn Cu Sn & Owens (in prep.), and the consistently differ- Fe Cu Fe Cu cu Cu Cu Cu zn Cu zn cu ent unit cells for the two minerals. The energy \/z Cu Fe Cu Fe Fe Sn Fe Sn Cu Sn Cu Sn considerations for this structural change are Fe Cu Fe Cu Cu Cu cu Cu cu zn cu zn not clear, though it appears that the similarity 1 Cu Fe Cu Fe Fe Sn Fe 5n Cu sn Cu sn of the effective radii ofZn and Fe accounts for the almost identical cell volumes of the two cupied only by Cu sites, are shared both by Cu species. and (Zn,Fe) atoms in kesterite. An alternative The bond leng:ths and angles are shown in way of illustrating the metal ordering in these Figures 2 and 3, and indicate that whereas the structures, and in chalcopyrite, is obtained by close-packing of tle S atoms is the dominant considering the arrangement of metal atoms in structural mechanism accounting for the spha-
i( ro9.9
0)/' i toa.g(
Cu1
1'
Frc. 2. The atomic coordrnation of each atom site in kesterite showing in- teratomic distances in Angstroms and angles in degrees' The estimated standard deviations are given in parentheses. The atoms are shown as thernral ellipsoids, plotted at Ibe 99Vo'probability limit (Johnson 1965). 136 THE CANADIAN MINERALOGIST
,' t-l,
Sn
Frc. 3. The atomic coordination of each atom site in stannite showing in- teratomic distances in Angstroms and angles in degrees. The estimated standard deviations are grven in parentheses. The atoms are shown as tlermal ellipsoids, plotted at the 99Vo probability limit (Johnson 1965).
lerite-like subcell, the larger radius of the Sn 2334, Most significant of the metal-sulfur dis- atom causes considerable movement from the tances is (Fe,Zn)-S, 2.348Q)A in stannjte, much "idealo' tetrahedral S sites in both structures. longer than that for CuFeSiz,2.3O2(l)A (Hall & The "squashing" effect of the Sn atom results Stewart L973), ard outside the Fe-S range of in angles Sn-${u = 108.0(1') and Sn-S-(Fe, 2.26-23A4 for chalcopyrite-ty,peminerals Qlall Zn) = 108.1(1)' in stannite, and angles Sn-$- 1975). This sontrastswith the kesterite (Zn,Fe)- Cu(l) - LO8.2Q)", Sn-SCu(2) - 108.2(2)" S distance ot 2336Q)4, which is shorter than and Sn-$-(Zn,Fe) = 108.1(2)' in kesterite.As both the ZnS value ot 2.342(L)A and the expected, other coordination angles about the (Fe,Zn)-S distance in stannite. The reason for S atoms are expandedto Cu-$-Cu = 112.2(1)" this apparent reversal of radii is not clear, but and Cu-S--(Fe,Zn) - 110.2(1)" in stannite, and it is no doubt related to the interchange of Cu(1)-${u(2) - LLO.9(2)', Cu-Sn-(Zn,Fe) = Cu and (Zn,Fe) positions between stannite and Ll0.6Q)" and Cu(2)-S-(Zn,Fe) - 110.8(2)' kesterite. in kesterite. The thermal parameters for stannite and kes- For these structures the Sn-S interatomic dis- terite are consistentvrith those observedin maw- tancesof 2AO8Q) and 2.411(1)A are the same sonite (Szymanski 1976) and other chalcopyrite- within the standard deviation and agree well type structures (Hall 1975). T\e mean isotropic rrith the value of 2.409(I)A in mawsonite. .B value of O.78A' for the Sn atom in stannite CueFelSnSe (Szymairski L976). The kesterite is marginally higher than the values of 0.65 and Cu-S distancesof 2.330(2) and 2.332Q)A agree 0.684' in kesterite and ma'wsonite, as are the closely, and are significantly larger than the equivalent sulfur I values; this probably reflects 2.320Q\A in stannite. These values are margin- overall differences in the diffraction data rather ally larger than those observed in the chalcopy- than increasedthermal activity in stannite. Most rite structures where values ranged from 2.3G- inportant, however, the relative magnitudes of KESTERITE AND STANNITE 137
the (Fe,Zn) and the Cu thermal parameters in Heu, S. R. (1975): Crystal structuresof the chal- stannite are very close to those observed in the copyrite series.Carz. Mineral. L3, t68-172. chalcopyrite-like structures and in cubanite _-, KrssrN,S. A. & Srrwenr, J. M. (1975): (Szymanski1,974). Similarly, the Cu B valuesin Stanniteand kesterite:distinct minerals or com- kesterite are both larger than the (Zn,Fe) B potrents of a solid solution? Acta Cryst. A}L, value, though the expectedaverage for the latter 567 (abstr.) cannot be well established from comparable & Srewanr, J. M. ( 1973) : The crystalstruc- structures in the literature. The importance of ture refinement of chalcopyrite,CuFeSg. Acta the relative rnagnitudes of the thermal parame- C'ryst.829, 579-585. ters in identifying metal sites in this type of JonNsoN,C. K. (1965): ORTEP: A FORTRAN structure has been demonstrated for chalco- thermal ellipsoid plot program for crystal struc- pyrite-like minerals (Hall 1975), and was one ture illustrations..Rep. ORNL-3794,2nd Rev. of the principal rea$onswhy model 2 for kes- ORTEP-II addition L97L, Oak Ridge Nat. Lab., terite (Table 2) was more acceptablethan model Oak Ridge, Tennessee;modified for use in the 1. X-RAY STEWART System by J. F. Gu6don, S. Hall, P. Richard and S. Whitlow, L974. Kuoon, Y. & Terfucru, Y. (t976): The super- ACKNOWLEDGMENTS structure of stannoidite,Z, Krist. L44, L45-L60. We are grateful to Mr. D. R. Owens of this LansoN,A. C. (1970): The inclusion of secondary laboratory for the rnicroprobe analyses,to Drs. extinction in least-squaresrefinement of crystal L. J. Cabri and S. A. Kissin for supplying the structures.fu CrystallographicComputing (F. R. crystal specimens, and to Mr. D. Lister for Ahmed, ed.), Munksgaard,Copenhagen. preparation of some of the diagrams. SrEwanr, J. M., Knucrn, G. J., AMoN, H. L, DIcKrNsoN,C. H. & Har-r-, S. R. (1972): The REFBRENcES X-RAY system of crystallographic programs. Univ. Maryland Comp. Sci. Ctr, Tech, Rep. TR- Bnocrwev, L. O. (1934)l The crystal structure of 192. stannite, CuzFeSnSa.Z. Krist. 89, 434-441. Szvrvreisrr, J. T, (1974): A refinement of the BusINc, W. R. (1970): Irast-squares refinement structureof cubanite,CuFe2S3. Z, Krist, L{A, ZIE- of lattice and orientation parameters for use in 239. automatic diffractometry. /n Crystallographic (1976): The crystal structure of mawsonite, Computing (F. R. Ahmed, ed.), Munksgaard, CusFegSnSe.Can, Mineral. L4, 529-535. Copenhagen. (1978): The crystal structure of iernfite, (1970): CRoMER,D. T. & Lnenvrex, D. Relativistic CuzCdSnSa,a cadmium analogueof stannite. Car?. calculation of anomalous scattering factors for Mineral. L6, 139-L46, X rays. J. Chem. Phys.53, 1891-1898. & MINN, J. B. (1968): X-ray scattering fac- tors computed from numerical Hartree-Fock wave functions. Acta Cryst. 424, 32L-325. Gesr, E. J. & O'BvnNe, T. (1970): An absorption correction program for the PDP-8. Amer. Cryst. Received February 1977; revised manwcript ac- Assoc. Summer Meet. Abstr. A4. cepted lanuary 1978,