American Mineralogist, Volume 58, pages 869-872, 1973

MeteoriticAmphiboles Eownno OrsEN

Field Museum of Natural History, Chicago, Illinois 60605, and Department of Geological Sciences, Uniuersity of lllinois at Chicago Circle, Chicago, Illinois 60680

J. SrspHENHursNBn U.S. Geolo,gicalSuruey, Washington,D.C. 20242

J. A. V. Douctas'aNn A. G. PreNr

Geological Suruey of Canada, Ottawa, Ontario, Canada

Abstract

The crystal chemistry of richterite found in two new occurrences,as well as that of the original richterite found occurring in a meteorite, is presented.Microprobe chemical analysis indicates that two of the richterites, from the iron Wichita County and Canyon Diablo, are not fully fluorinated; however, their a cell parameters suggestthey are fully fluorinated. The cell parametersand the chemical analysesare reconciled by postulating that oxygen (O=) substitutes for hydroxyl (OH-) on the O(3)-site. The third richterite, from an enstatite (), shows a and p parameters lower than those for fully fluorinated richterite. This is attributed to Mg distributed on the M(4)-site, as well as full fluorination, suggestingthat a partial solid solution between normal fluorichterite (NaCaNa- MguSLOoF,) and Ca-free fluorichterite (NaMgNaMg;Si,O,,F,) may exist in nature. There is no evidence for structural water in any of these meteoritic amphiboles.

Introduction matrix and richterite occurs as clusters of interpene- The first noted occurrence of an amphibole of trating needles (<3.5 mm long) within some of extratenestrial origin was in the coarse thesepods. Enstatite (Fs = O) is also present. Other , Wichita County (Olsen, 1967). Sub-. metal pods lack richterite but contain enstatite, sequently, tv{o new occurrences have been found: troilite, niningerite, albite, cristobalite, and old- the coarse octahedrite, Canyon Diablo, and the Type hamite. I , Abee (Douglas and Plant, Analytical Data 1968). This report will discussthe crystal chemistry of these amphiboles. In the original electron microprobe analysisof the The Canyon Diablo occurrence is similar to that Wichita County richterite (Table 1) the fluorine described by Olsen for the occurrence in Wichita value was determined against the only two available County. Richterite grains up to a millimeter long are standards at the time, an analyzedhornblende and a enclosed within a centimeter size graphite inclusion synthetic barium fluoride. Both of these were fairly in iron metal; associated minerals are forsterite unsatisfactory in terms of their bulk compositions (Fa 1), krinovite, roedderite, chromite, ureyite, high relative to the richterite, although they did give albite, troilite, and sphalerite. internally consistent fluorine count rates. It was not The occurrence in the Abee meteorite is very possible at the time to make accurate mass absorp- different. Abee consists of a of angular tion corrections because of the absenceof tables of fragments set in a fine-grained matrix with metal reliable coefficients for fluorine radiation. Because concentrated along the margins of the fragments. of the microscopic quantity of richterite, no method Some podlike segregations of metal occur in the of water analysis was applicable, and an amount of water was assumedto be present to satisfy the con- 'Deceased, June18, 19?1. dition (F + OH) = 2 pet amphibole formula unit'

869 870 E. OLSEN, AND I. S. HUEBNER

TesI-s 1. Electron Microprobe Analyses of Three be drawn about underpopulations of tetrahedral Meteoritic Amphiboles in Wt Percent sites,or under-and-overpopulationsof l-sites (Table

Canyon Diablo* Abeex Wichita County* 2). ln addition, Table 2 is presented in the standard sio2 57.2 57.9 57.8 style of Deer, Howie, and Zussman (1962). This Tio2 2.2 0.01 1.5 format cascadeserrors into the l-site. Ar203 0.38 0.15 o.4 Structural Formulae and Lattice Parameters ct2o 3.5 0.00 o.7 3 The richterite analyses were first recalculated on Fe0 0.44 0.17 0.5 negative charges, assuming O : 22 MnO 0.05 0.00 0.00 the basis of 46 Mco 20.9 26.r 22.4 and OH * F: 2 (Table 2). Iron, chromium, and Cao 3.7 5.5 5.8 titanium were assumed to be present as Fez*, CrS*, and Ti4*, and distributed among the cation sites Naro 8.3 7.3 7.5 (Table 2). Next, fluorine was added to the O(3) K2o r.2 0.36 0.6 position, and the formula then could be adjusted in F 3.6 4.6 2.3 any of four ways. 1) Assume the fluorine analysis is Sum t01.5 102 .1 99.5 correct, OH is present, and that all 24 anionic sites O=F -t.9 -0.9 are filled. Enough water is then added to fluorine to Total 100.0 100.2 98.6 fill the O(3) site (F + OH :2).2) Assumethat the fluorine analysis is correct, that no hydrogen is present, and that there are still 46 positive and negative charges per formula unit. This distribution This procedure resulted in OH = 1.0 mole per results in oxygen vacancieswithin the structure (not amphibole formula unit and H2O = 1.1 wt percent necessarilyrestricted to O(3)), a highly unlikely in the analysis. situation. 3) Assume the fluorine analysis is in- In an attempt to find a more suitable microprobe correct and that no hydrogen is present; fill the standard for fluorine, we used two Na-fluorichterites O(3) site with 2F (balancing the charge of one of synthesizedby Huebner and Papike (1970). Un- the 23 oxygens in the original formula unit). 4) As- fortunately, by this time all of the richterite from sume that the fluorine analysis is correct, that there Wichita County had been expended due to research is no structural hydrogen, and that there are no requests from other investigators. The original anionic vacancies. In this case, oxygen substitutes richterite grains were so small that they were lost in on the O(3) site, and there are more than 46 the process of repolishing the original probe mount. negative charges per formula unit. The cationic The richterite from Canyon Diablo was analyzed formula is recalculated to preserve electroneutrality. by microprobe and found to be similar to the Wichita Note that there are two ways to calculate the formula County specimen except for slightly higher K and of oxygen-substitutedamphibole. In the first method, much higher Cr (balanced by lower Ca and Mg). charge balance is achievedby increasing the assigned The fluorine analysis, against the synthetic standard, valence of various cations; the only possible cation was repeated in two different laboratories by two here is Fe'*, which is present in such a small quantity different persons with identical results. A single that charge balance cannot be achievedby converting fluorine analysis was made for the Abee sample. all iron to Fe3*. The other possibility is to increase Although the precision of these microprobe the number of cations relative to 24 anions. How- analysesis within five relative percent in the original ever, the number of cations in any site as given by count rate variations, the accuracy is not generally the idealized structural formula, A(M4)2(M(I- that good. For some elements (Fe, Mn, Ti, Cr, Ca, 3))rXaOr:(O, OH, F)2, should not be exceededby K) the accuracy may be as good as two percent, as- more than the analytical uncertainty. suming we know the compositions of our standards Huebner and Papike (1,970) present cell param- exactly. For other elements (Si, Al, Mg, Na, F) the eters for the substitution of F for OH in richterite accuracy is certainly within ten relative percent. In and suggestthat the Wichita County richterite is fully calculating structural formulae (see below) three fluorinated. The cell parameter a (or a'sin B) is decimal places were carried for consistency; how- sensitiveto this substitution (as well as to the substi- ever, the accuracy is such that no conclusions can tution of K for Na). Fixed indexine of Olsen's METEORITIC AMPHIBOLES 871

(1967) Wichita County richterite X-ray powder TrsrF, 2. Analyses Recalculated to Formula Units

difiraction pattern yields the parameters given in Canyon Diab Table 3 (Huebner and Papike, 1970).In addition, AB single crystalsof the Wichita County richterite were st 7.884 7.956 7.972 7.9L4 1.958 8.000 A1 0.062 o.044 0.001 0.001 0.042 measureddirectly by Dr. M. Cameron (personal Ti 0.054 0.024 0.024 communication)with the cell dimensionsalso given Sun 8.000 8.000 7.937 7.939 8.000 8.000 in Table 3. When these are comparedto the cell si 0.130 parameters(especially a and a.sin B) of synthetic A1 0.018 0.023 0.666 Na-fluorichterite (Huebner T1 0.L74 0,230 0.155 0.159 and Papike, l97O), it Cr o.382 0.386 0.076 0.078 would appear that the Wichita County richterite is 4.293 4.333 5.000 5.000 4.596 4.567 Fe 0.051 0.032 0.0s8 indeedfully fluorinated. Mn 0.006 Similarly cell parametersobtained for the Canyon 0.094 o.092 Diablo richterite (from fixed indexing of the powder SW 5.000 5.000 5.000 5.000 5.000 5.000 pattern) indicateit too is fully fluorinated.The cell 0.316 0.3r7 o.L29 parameters Fe 0.018 0.019 0.019 0.059 of the Abee specimenindicate full fluori- lfn 0.006 nation as well as other peculiaritiesto be discussed 0.452 0.551 0.805 0.805 0.764 0.874 1.548 1.424 0.860 0.859 L236 0. 938 below. Sun 2.000 2.000 2.000 2.000 2.000 2.000 The Abee richterite showsfull stoichiometricfluo- rine (F = 2 moTes/formulaunit). Na 0.670 0.814 r.074 1.075 0.766 1.107 However, the K 0.2rL 0.2r3 0.063 0.063 0.105 0.108 Diablo Canyon and Wichita specimensshow F < 2 SUD 0.881 7.021 r.r37 1. 138 0.871 L2r5 and must thus have significant OH- or O= content F 1.569 1.583 1. 988 1. 988 1.001 1.023 in O(3). OH 0.431 0.0r2 0.999 The microprobe chemical analysesfor fluorine o o.4I7 0.0r2 o.977 may be reconciledwith the richterite cell parameters Sun 2.000 2.000 2.000 2.000 2.000 2.000 by a combinationof factors. Most important, the o 22.O00 22.000 22.000 22.000 22.O00 22.O00 substitutionof eitherF or O=for OH- on the O(3) A colums: Calculated to 46 posltlve = 46 negative charges, site would be expectedto decreasea and.a.sin B, B colws: Calculated to O = 24 - F - OH. 24>0>22. largely due to the removal of the proton2 associated with O(3). We would expectlittle changein a or andB arethe smallestvalues reported for any natural a.sin B as F- = O=, becausehydrogen is not in- richterite. This is supported by the data of Gibbs, volved.Such substitutionswould not changeb or c Miller, and Shell (1962) on synthetic Ca-free, Na- appreciably. Mg fluoriclrteritewhere a : 9.677 A and B = The valuesa anda.sin B of an amphibolewill be 102"57'.It alsosuggests at least apaftial solid solu- changedto a lesserextent as minor cation substitu- tion (in the probabletemperature range of 700"C) tions and reorderings are made amo{rgthe M sites betweennormal fl uorichterite( NaCaNaMgSieOepF2) and betweenthe M(4) and A sites.These changes and Ca-freefluorichterite (NaMgNaMgsSisOzeFz). will also afrectb and,c. Discussion The Abee richterite presents some interesting Evidencefor the presenceor absenceof structural and unique features.Repeated analyses consistently rvater in extraterrestrialmaterials has interested show more than 5 Mg per formula unit (Tables 1 and 2). In computingthe structuralformula, it was Trnre 3. Unit Cell Data for Richterites necessaryto insert some of this surplus Mg into M(4) as well as a small Canyon amount of the slightly Wlchita County Dlablo Abee SFthetlc larger Fe ion. If this is actuallythe case,then this Rlchterite Rlchterlte Richterlte F-Rlchterite would strongly affect the a parameter (by 4 9.832(6)l* e.824(2)A** 9.812(9)L 9.767(15)l 9.838(r)l*** about b !7,9s4QlE 17.955(r)L tt.szalt'sR. u.966(14)A 1i.963(3)n 5%). This fact, coupled with full stoichiometric c s.269(4)X 5.26e(1)X. 5,269(1);. 5.253(8)i s.262(Lr,. fluorine content, accountsfor the fact that both a I 104'2r(3), r04"11(3)' r04"r0(3), 103"55(7)'i r04'15(7), a's1trB 9.525(4).{. s.sz4',, 9.520(iX 9,4sol 9.526(r)l

'The I/ 90r.2(7)8 " 901.0(2)13 899.1(9)n! 894.8(r1.6)X,3900.3(3)43 hydrogen ion position is such that it tends to push apart the amphibole structure along the a axis; see Huebner *f!ffi Huebne? and Pdpike (1970), lefinaent of Olsea's (196il data. *afrrlependentllJ neasured on singXe crystaL by Dt, M. C@eton, and Papike (197O); Papike er al (1969). *4*Fran Huebner andPapike (19?A). 872 E. OLSEN, AND ]. S. HUEBNER numerous investigators, most recently those working of us (EO) was supported by a grant from the National on the lunar samples. Extraterrestrial amphiboles Aeronautics and SpaceAdministration, and this is gratefully acknowledged.Drs. Malcolm Ross and David R. Wones, are an excellent potential water barometer or indica- U.S. Geological Survey, provided helpful reviews of the tor of the presence or absence of significant H2O manuscript. in the environment in which the amphibole formed. In the case of the three meteoritic richterites, we References find no compelling evidence for the presence of Cnenrrs, R. (1972) The pltysical properties and phase structural water in the amphibole, and thus in the equilibria of the hydrous Mg-Fe richterites.Ph.D. thesis, environment in which the amphibole last equili- MassachusettsInstitute of Technology, 157 pp. Depn, W. A., R. A. Howrr, eNo I. ZussueN (1962) Roct brated. We would like to caution future investigators Forming Minerols. Longmans, London. against assuming that incomplete fluorination indi- Douclrs, J. A. V., eNo A. G. Pr-eNr (1968) Amphibole: cates hydroxylation of amphibole. first occurrence in an enstatite chondrite (abstr.) Mete- We expect that additional richterite occurrences oritics. 4. 166. (1971) will be found as more enstatite are ex- FoRBEs,W. C. Synthesisand stability relationsof richterite,Na'CaMgoSi^O.(OH). Amer. Mineral. 56, 997- amined in detail. It is also clear from an ongoing 1004. study of silicates in iron meteorites that many of Grsns,G. V., I. L. MTLLER,exo H. R. SsEr-r (1962) Syn- Wasson's Group I irons will contain richterites thetic fluormagnesio-richterile.Amer. Mineral. 47, 7 5-82. (Wasson, 1970). Occurrences of richterite in these HuEnNrn, I. S., rNn J. J. Pepxr (1970) Synthesisand (Na,K) primitive meteorites (compared to the Earth's crust crystal chemistry of sodium-potassiumrichterite NaCaMg"SLO,,(OH,F),:A model for amphiboles.Amer. and mantle) will have implications on fractionation M i ncrul. 55, l9'73-1 992. in the Earth and the initial storage of alkalies (espe- PlerxE, J. J., M. Ross, eNo J. R. Cunr (1969) Crystal cially K) and volatiles (especially water) in the chemical characterization of clinoamphiboles based on early mantle. Early mantle conditions could be five new structure refinements.Mineral, Soc. Amer. Spec, Pap. 2, studied by examining the high pressure breakdown 11,7-136. OI-sEN,Eowlno (1967) Amphibole: first occurrencein a products of OH-bearing richterite under water de- meteorite.Scie nce, 156, 6l-62. ficient conditions (as has been done for common WessoN, JouN T. (1970) The chemical classification of amphiboles by Wyllie, 1977a, b) as well as under iron meteorites.IV. Irons with Ge concentrationsgreater the excess water conditions examined alreadv bv than 190 ppm and other meteoritesassociated with Group I. Icarus,12, 407-421. Forbes (19'71) and Charles(1972'). Wvrr-tE, P. I. (1971a) Experimental limits for melting in Acknowledgments the Earth's crust and upper mantle. Amer. Geophys. Union, Geophys.Monograph Ser. V, 14,279-301. We would like to acknowledgeDrs. Kurt Fredericksson (1971b) Role of water in magma generationand (SmithsonianInstitution) for performingan independent initiation of diapiric uprise in the mantle. Iour. Geophys. fluorine analysis, Maryellen Cameron (SUNY/Stony Res.76,1328-1338. Brook) for her cell measurementsof the Wichita County richterite,and LouisFuchs (Argonne National Laboratory) Manuscript receiued,Nooember 15, 1972; accepted for his X-ray patternof the CanyonDiablo richterite.One for publication, March 14, 1973.