American Mineralogist, Volume 78, pages 433-436, 1993

Povondraite. a redefinition of the ferridravite

Jonr, D. Gnrcn, T. Scorr Encrr Mineral SciencesSection, Canadian Museum of Nature, Ottawa, Ontario K1P 6P4, Canada FnaNx C. HlwrHoRNE Department of Geological Sciences,University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada

Assrnlcr Povondraite, previously the tourmaline ferridravite, is redefined. It is rhombohedral, R3m, with a: 16.186(2)and c:7.444\l) A. fne new chemicalformula, derived by analysis, is (NaoroKo ru),, ou (Fel.jrFeS j,Mgo rr)",o, (FeljrMg, .uAlo.r)r,nr- B.Si5e6O2,88(OH)r,,and the ideal end-memberformula is NaFel*Fel+(BO.).(SiuO,')- (o,oH)4.

INrnooucrroN nm). Thesevalues are greaterthan those given by Walen- (1979). During the investigation of a seriesof tourmaline crys- ta and Dunn model tal structureswith varying contents of Fe and Mg (Grice Povondraitewas chemicallyanalyzed using a Jeol and Ercit, 1990, 1993) the "ferridravite" structurewas 733 electronmicroprobe. Wavelength-dispersionanaly- 15 kV, refined. It was discovered that the formula proposed by sesused an operatingvoltage of a beam current Walenta and Dunn (1979) for the speciesferridravite was of 25 nA measuredwith a Faradaycup, and a beam di- pm. were incorrect. They adopted the standard schemefor assign- ameter of l0 The following standards used: gehlenite(Al), (Si, ment of cationsto the X, Y, and Z sites,i.e., with Fert synthetic almandine Mg), synthetic (Fe),microcline (K), amphibole (Na), VP,O, and Al3* assignedto the Z site and Mg2* and Fe2+to the fayalite sodic (V), (Ti). were Y site, yielding the empirical formula (NaoroKoro)- and titanite The elementsF, Mn, and Ca sought but not detected. Data reduction was performed (Mg,rrFel-i,)(Fe31nAlo,,)SiuBr (O,OH)30 (OH,F). As with the structure refinement of feruvite (Grice and Robinson, with a conventional ZAF routine in the Tracor Northern programs. 1989),it becameevident that the assumedsite assign- Task seriesof The analysesof the singlecrys- given in ments of Mg'?*and Fe3* were incorrect. The crystal struc- tals used in the crystal structure refinementsare Walenta ture refinement of "ferridravite" yielded a new empirical Table l, along with the chemical data of and ( well with present formula that bears no resemblanceto "the ferric analog Dunn I 979), which agreevery the anal- yses. of dravite" (Walenta and Dunn, 1979); therefore renam- Also in Table I are analysesof K-dominant tour- (10 pm) portion ing of the specieswas required.The new name, povon- maline that forms a thin outer rim on a draite, is in recognition ofthe extensivework on the chem- of the Royal Ontario Museum sampleno. M35899. Un- properly istry of the tourmaline group by Dr. Pavel Povondra, fortunately, there is not sufficient material to Karlovy University. The data and new name for this re- characterizethe K-dominant tourmaline as a new species, presence germane defined specieswere approved by the Commission on but its is very from a crystal-chemical New Mineralsand Mineral Names,I.M.A. Type material vrewpornt. povondraite its potassicrim is for povondraite is preservedat the National Museum of Chemical zoning of and in Natural History, SmithsonianInstitution (no. 144478)and shown in Figure l. There is a sharp discontinuity the povondraite at the Royal Ontario Museum (no. M35899). ratio K:Na between and the K-rich tour- maline at the rim; however, chemical trends established Prrysrc.q.L AND CHEMICALDATA within povondraite continue into the rim tourmaline. For + whereas The two samples used in this study (Royal Ontario the entire system,Fe,., and K/(Na K) increase, Museum no. M35899 and SmithsonianInstitution no. Al, Ti, and V decreasewith crystallization. 144478)are both type material of "ferridravite" from the SanFrancisco mine, nearVilla Tunari, Bolivia. The min- CRvsrll sTRUCTUREREFTNEMENT eral is described(Walenta and Dunn, 1979)as occurring Initially the crystal structure of povondraite was re- in metamorphosedsedimentary rocks. The schist is com- fined using a crystal fragment supplied by the Royal On- posed of , potassium feldspar, alkali amphibole, tario Museum (sampleno. M35899). The quality of in- , and tourmaline (schorl and povondraite). tensity data from this crystal was such that it was evident Povondraiteis stronglypleochroic with O yellow-brown there was a problem with the empirical formula given by and E dark red-brown. It is uniaxial negative with co: Walenta and Dunn (1979), but a better crystal structure 1.820(5)and e : 1.751(3),measured with Na light (590 refinement was desirable.Five grains were obtained from 0003-004x/93/0304-o43 3$02.00 433 434 GRICE ET AL.: POVONDRAITE: FERRIDRAVITE REDEFINED

TABLE1. Electronmicroprobe analyses of povondraite CORE RIM Structure Structure K-rich tourm. Walentaand Fe 144478 35899 35899 Dunn -€- sio, 30 7 31 1 29 6 31.58 F Tio, 0.0 01 0.0 AI Al,o3 1.4 4 2 0.8 2.30 -a- V,O" trace 0.1 0.0 o TI Fe.O.. 45.0 421 45 6 38.37 \ -t- FeO' 1.7 07 0.9 7.25 \ MgO 65 6.9 63 5.57 \ V K,O 10 0.9 22 0.98 z \ -l- Na,O 21 2.2 1 1 2j8 o \ B,O"'. 9.0 91 8.7 gg8 F \ H,Ot 24 2.5 23 3.5 \ TorAL 99 I 99.0 97 5 100.71 rD_ rh Number of cations on the basis ot O : 31 t Si 5.96 5JZ 5.91 6.02 F Ti 000 0.01 0.00 z AI o32 no( o.20 u.Jz IIJ 0.01 0.02 0.00 O Fe3* o.c/ 6.04 6.87 R A1 z Fe2* o27 0.11 0.15 1.16 Mg 1.89 1.95 1.87 1.58 o K 0.26 0.21 0.56 0.24 O NA 0.80 0.80 0.42 0.81 I B 2 ? 3 H 3.12 3.12 n,tz J.C r \,p Fe3+lFe2+ calculated by stoichiometry z-A-- ** B,O" from B : 3 atoms in formula. t H,O from bond valencesum H :3.12 exceptfor Walentaand Dunn r-r_ L.O.t. a o.2 o.3 0.4 0.5 0.6 the Smithsonian Institution (sample no. 144478)and the K/(Na+K) Fig. l. Chemicalzoning in povondraite(core) and K-rich best of these crystals, as determined by the quality of tourmaline(rim), Royal Ontario Museum sample no. M35899. reflections on precession films, was used to measure a Thereis a sharpdiscontinuity in the ratio K:Na (dashedline) second data set. Intensity data were obtained on a fully betweenpovondraite and the K-rich tourmaline.Scales are rel- automated Nicolet R3m four-circle diffractometer using atrve. the method of Grice and Ercit (1986).The data relevant to the structure refinement are given in Table 2. No X-ray powder diffraction data are listed here, as there rs no the Z site, and refinement gave occupancyfactors corre- changefrom that presentedby Walenta and Dunn (1979). spondingto 3.84(12)Fe atomsand 1.86(12)Mg atoms. The structure refinement of povondraite was essential These refined values agreewithin 3o of the electron mi- for the definition of the species.From the chemical data croprobe analysis of this crystal. given in Table I, the number of Fe and Mg atoms are We have found that the residual index R for a tour- known to be 6.84 and 1.89,respectively. The distribution maline crystal structure refinement usually falls between of these cations in the Y and Z sites determines the spe- 2 and 3o/o.The R index in this experiment is controlled cies. The atomic coordinates of feruvite (Grice and Rob- primarily by the poorer quality of povondraite crystals, inson, 1989)were usedto beginthe structurerefinement which tend to have alarge mosaic spread.This gives rise of povondraite. Refinement of positional and anisotropic to larger standardelTors in observedstructure factors and thermal parametersand the occupancyfactors of Fe and thus to larger uncertainties in structure parameters.AI- Mg in the Y andZ sitesgave a residualindex R : 5.06 though the standarddeviations of Fe and Mg occupancies and R* : 5.07o/o.For the Y site,refinement gave 2.52(7) in the Y and Z sites is rather large, there is no doubt Fe atoms,and 0.48(7)Mg atoms. All Al was assignedto about the partitioning preferenceof these elements be-

TABLE2. Structurerefinement data for povondraite

ldealformula: NaFeS*Fe3* 83Si6O,s(O,OH)3 a: 16186(2) A Space group: HJM c:7.444() A Z: 3 v : 1688.9(3)A3 Radiation/Mono: Mo/graphite no. of F. 1214 p: 47.9cm . no of F" > 5o(F) 717 Min. transmission: 0.463 FinalR: 5.06% Max. transmission: 0.552 FinalF": 5.07o/o F -- >(|F"t lF.ly2lF.l & : P w(lF"l- lF"l)'l>wlF"l"l*, w : o 2(F.) GRICE ET AL.: POVONDRAITE: FERRIDRAVITE REDEFINED 435

Tlele 3. Positionaland anisotropicthermal parameters ( x 103 A'?) for povondraite

Y Z U,. U"" U"" U,, Na(x) 0 0 0.228(21 17(3) 17(3) 34(6) 0 0 8(1) 23(3) 't2(1) Fe(Y) 0.1224(2\ 0.0612(1) 0.6429(5) 4(1) 4(1) -1(1) -1(1) 2.2(61 6.5(8) Fe(z) 0.2987(1) 0.2626(1) 0.6124(s) 4(1) 4(1) s.1(8) 10(7) 5(6) 2.2(71 5.8(7) B 0.1097(7) 0.219(1) 0.456(2) 12(6) 4(7) 7(6) 1(6) 0(3) 2(3) s(5) -1(1) - 5l 0.1892(2) 0.1876(2) 0 5(1) 4(1) s(l) 1(1) 3(1) 6(1) o1 0 0 0.772(3) 7(s) 8(5) 7(s) 0 0 4(3) 7(41 02 0.0610(4) 0.1220(8) 0.4s0(2) 10(4) 4(4) 13(s) 2(41 1(2) 2(2) 10(3) o3 0.2s78(8) 0.1289(4) 0.512(1) 12(5) 7(3) e(s) -1(21 -3(4) 6(3) s(3) o4 0.0924(4) 0.1848(8) 0.069(2) 8(3) 16(6) 12(4) -2(41 -5(e) 8(3) 11(3) o5 0.1827(8) 0.0913(3) 0.086(1) 10(5) 4(3) 18(5) 0(21 0(4) 5(2) 10(3) o6 0.1922(6) 0.1829(5) 0.784(1) 12(4) 7(3) 8(3) -2(3) 0(3) 6(3) 8(3) 07 0.2811(5) 0.2814(s) 0.075(1) 8(4) 7(4) 12(3) 0(3) 0(3) 2(3) 10(3) o8 0.2069(5) 0.2678(6) 0.442(1) 5(3) 10(4) 17(41 6(3) 4(3) s(3) 10(3) H3 0.24(1) 0.120cl) 0 41(3) 20

tween the two sites. Only one H atom could be located an OH molecule. As far as we can determine, there is with certainty from a AF-synthesis. Refining the structure little if any occupancyof the H site often associatedwith of the other enantiomorphgave an R index of 5.80/0,ver- Ol. Final atomic and thermal vibration parameters are ifying the correct choice of enantiomorph. given in Table 3. Selectedinteratomic distancesand an- Bond-valencesums for eachatomic site are X : 1.03, gles are given in Table 4. The observed and calculated Y : 2.7t,2 : 2.87,8: 2.94,Si: 4.10,O1 : 1.65,02 structurefactors are depositedas Table 5.' : 1.98,03: 1.17,o,4:2.09,05:2.04, 06 : 2.01, 07 : 2.02, and 08 : 2.02 vu. This confirms that 03 is DrscussroN The most important point in this paper is the illustra- tion of the need for crystal structure analysis of minerals Trau 4. Selectedinteratomic distances (A) andangles (') for whosespecies status is dependenton the groupingofel- povondraite ements within the empirical formula according to their Na (X site) polyhedron ordering within the crystal structure. The data for the Na-O2 2.60(1)x 3 description of "ferridravite" presented by Walenta and 04 2.85(1)x 3 Dunn (1979)were correct,but the formula was in error 05 2.77('ll x 3 Average 2.74 becauseof assumptionsmade on assignmentof cations Fel (Y site) octahedron to various sites. The data presentedhere are but a small Fe1-O1 1.97(1) O1-O2 84 5(4) x 2 portion of presentedin Grice and Ercit (1993),which 02 2.055(8)x2 o1-Oo 99 8(5) x 2 those 03 2.13(1) 02-06 89 3(4) x 2 clarifies the relationships of Fe and Mg substitution in OO 2.007(8)x2 O2-O2 92.3(5) the Y and Z sites. The proportion of Mg in the Y site Average 2.037 O2-O3 97 0(3) x 2 o3-o6 787(3) x 2 decreaseslinearly with the fraction Fe/(Fe * Mg), where o6-o6 88.8(5) Fe and Mg are numbers of atoms. This simple relation- Average 90.0 ship shouldhelp petrologistsand mineralogistsdetermine Fe2 (Z site) octahedron Fe2-O3 2.061(6) 03-06 80.3(4) the correct formula and speciesname as well as Jrelpre- 06 2 009(8) O3-O7 95.2(41 veal the many erors that presentlyexist in the literature. 07 1.991(8) O3-O8 87.7(41 With the present crystal-chemical model it may be pos- 07' 2.022(7) O3-O8 97.2(41 08 1.99(1) O6-O7', 93.3(4) sible to better interpret Mtissbauerspectra. Unfortunately 08' 1.97(1) 06-08 90.7(3) there is insufficient povondraite for such studies at pres- Average 2.007 06-08' e3.e(4) 07-o7', 91 7(1) ent. Also there are many possibilities for studying the o7-o8' 96 1(3) additional complexity of this scheme when significant o7-o8' 79 2(3) amounts of Ti, V, Mn, or Cu are included in the crystal o7-o8 79 5(3) o7'-o8 96 4(3) stnrcture. Average 90.1 In the tourmaline group there are presently ten species B triangle povondraite B-O2 1.39(2) O2-O8 120.2(8)x 2 (Table 6), and is unique among these as the OB 1.37(1)x2 O8-O8' 12O(2) specieswith the largestunit cell. The large cell volume of Average 1-38 Average 120.1 povondraite is pnmarily due to the large averagebond Si tetrahedron si-o4 1.628(s) o4-o5 104.2(6) length of the Z polyhedron(2.007 A) and to a lesserex- o5 1 640(5) 04-o6 111.4(5) o6 1.614(8) 04-o7 109.9(6) 07 1.603(6) o5-o6 109.6(5) I A copyof Table5 maybe ordered as Document AM-93-518 Average 1.621 o5-o7 110.s(5) from theBusiness Ofrce, Mineralogical Society of America,I I 30 06-07 1110(4) Washington,DC 20036, Average 1094 SeventeenthStreet NW, Suite330, U.S.A. Pleaseremit $5.00in advancefor the microfiche. 436 GRICE ET AL.: POVONDRAITE: FERRIDRAVITE REDEFINED

Trele 6. Tourmalinegroup: compositions,cell dimensions,bond lengths

Species 26 a(A) c(A) v (4.) x-o (A) Y-O(A) z-O(A) Ref."

Povondraite NA Fe3* Fe3* 16.186 7.444 1689 2.738 2.037 2 007 1 Buergerite Na Fe3* AI 15.874 7.196 1570 2.706 2.004 1 919 2 Schorl Na Fe2* AI 15 990 / tYc 1593 2.690 2.045 1.924 Feruvite Ca Fe,* AI 16.012 7 245 1607 2 654 2.055 1 944 Uvite Mg AI 15.973 7.231 1594 2.651 2.050 1.928 Dravite Na Mg AI 15.941 7.201 1585 2 681 2.019 1.928 2 Chromdravite Na Mg Cr 16.11 1634 4 Olenite NA AI AI 1s.802 7 086 1532 2.677 1.936 1.898 Elbaite Na AI,Li AI 15.838 7.103 1543 2.670 2002 1.905 Liddicoatite Li AI 15 875 7j26 1555 2.645 2 038 1.909 o * References:1 : presentstudy; 2: Griceand Ercit,1993; 3: Griceand Robinson,1989; 4: Rumantseva,1983; 5: Gorskayaet al, 1982; 6 : Nuberand Schmetzer,1 981 .

tent the X polyhedron (2.738 A;, bottr of which exceed and Emie Nickel for their suggestionsand comments during the IMA the averagebond lengths for all other tourmaline species. processof redefinition for this mineral In addition, the paper has been (1989) improved by the constructive comments ofthe referees,Donald Burt and Foit has shown that there is a very high linear Hughes. Bob per- : Nick Foit, and the Associate Editor, John M. Gault correlation (r 0.98) between cell volume and the formed a number of electron microprobe analysesthroughout the tour- weighted mean octahedral bond length. The Z-site octa- maline project, which was sponsoredin part by a grant to the authors hedron ofpovondraite is considerablylarger than that of from the ResearchCommittee, Canadian Museum of Nature other tourmaline becauseof substitution of Fe3+for Al3* . Chromdravite is the only other speciesof the tourmaline RnrrnnNcBs cITED group principal that does not have Al3* as the Z-site cat- Foit, F F, Jr (1989)Crystal chemistry ofalkali-deficient schorl and tour- ion. With Cr3+ in this site the Z-O distance should be maline structuralrelationships American Mineralogist, 1 4, 422-4f l. quite large. Although the crystal structure of chromdra- Gorskaya,M.G., Frank-Kamenetskaya,O.V., Rozhdestvenskaya,I V , and vite has not been refined, that of Cr-rich dravite has (Nu- Frank-Kamenetskii,V.A. (1982)Refinement of the crystalstructure of ber and Schmetzer,1979). Nuber and Schmetzer(1979) Al-rich elbaite,and someaspects ofthe crystalchemistry of SovietPhysics Crystallography, 27, 63-66 have assignedAloouCrr.rFeo24 to the Z site and deter- Grice, J D., and Ercit, T S ( 1986)The crystalstructure of moydite Ca- mined the averageZ-O distanceas 1.956 A, which, as nadianMineraloBist, 24, 675-61 8 predicted,is larger than for the Al3* species.In povon- -(1990) Crystal chemistryof Fe-rich tourmalines GeologicalAs- draite the ninefold X site is exceptionally large because sociationofCanada Annual Meeting,Program with Abstracts,I 5, A5 l. - ( of Fe and Mg in tourmaline:The conect formula. presence (to 1993)Ordering of the unique of K the extent of approxi- Neues Jahrbuch fiir Mineralogie Monatshefte,in press. mately one quarter of the site occupancy).For all of the Grice, J D, and Robinson,G W. (1989)Feruvite, a new memberof the other tourmalines listed in Table 6, the large cation K is tourmaline group, and its crystal structure.Canadian Mineralogist, 27, absent. t99-203. (1979)Die desCrr* im Tur- The unique geochemistryresponsible for the formation Nuber, B , and Schmetzer,K Gitterposition malin: Strukturverfeinerungeines Cr-reichen Mg-Al-Turmalins. Neues of povondraite (an Al-poor and K- and Fe-rich environ- Jahrbuchfiir MineralogieAbhandlungen, 137, 184-197. ment) explainsits rarity. The possibilityof a K end-mem- -(1981) Structurverfeinerungvon Liddicoatit.Neues Jahrbuch fiir ber tourmaline should encourageinvestigators to look MineralogieMonatshefte, 21 5 -2 | 9. further for sufficient material to characterizethe species. Rumantseva,E.V (1983)Chromdravite, a new mineral Zapiski Vseso- yuznogoMineralogicheskogo Obshchestva , ll2, 222-226(in Russian) Walenta,K., and Dunn, P.J. (1979)Ferridravite, a new mineral of the ACKNowLEDGMENTS tourmalinegroup from Bolivia. AmericanMineralogist, 64,945-948. This project would not have been possiblewithout the specimensfur- nished by the Smithsoman Institution, Washington, and the Royal On- MeNuscnrpr REcETvEDJeNunnv l'7, 1992 tario Museum, Toronto The authors thank Pete Dunn, Kurt Walenta, MnNuscrrn ACCEpTEDNoveussn 13, 1992