American Mineralogist, Volume 63, pages 874-877, tgZS Crystalstructures of thehumite minerals: V. Magnesianmanganhumite Cnnl A. FnaNcrsr,qNo Peul H. Rrssr Department of Geological Scjences Virginia Polytechnic Institute and State Uniuersity B lac ksburg, Vi rginia 2406 I Abstract The crystalstructure of typemanganhumite from the Brattlorsmine, Nordmark, Sweden [(Mno,u,MgoroFeoo,Cao.o,),(SiO.)s(OH),,Pbnm,a:4.815(l),b:10.580(2),c:21 .448(5)Al hasbeen refined byconventional least-squaresmethodstoR :0.051 (R- = 0.081),using 1562 structurefactors with F"u" ) 2o(Fo6").Mean Si-O distancesfor the isolatedtetrahedra are (Si(f)-O) : 1.628Aand (Si(2)-O): 1.632A.The apicalSi-O distancesof 1.600and 1.6044 are the shortestyet observedin humite-groupminerals. Site refinement yielded Mgl(Mg+ Mn) ratiosof 0.30in the M(l)Ou octahedron(mean M-O distance:2.1'70A),0.0 in the M(2)O6 and M(2)O,OH octahedra((M-O, : 2.222 and 2.206A), and 0.75 in rhe M(3)O4(OH),octahedron ((M-Ot : 2.117A).The correlationcoefficient of Mgl(Mg+Mn) occupancyand meanoctahedral bond lengthis r : 0.986,and as expected,octahedral sizes correspondalmost exactly to theweighted radii of thelarger Mn andsmaller Mg cations(r : 0.989). lntroduction been observed in all subsequentrefinements of hu- The humite minerals are a homologous seriesof mite group minerals, including several titanian hu- magnesium orthosilicates based on hexagonal mites(Robinson et a|.,1973;Kocman and Rucklidge, (1971) closest-packedarrays of anions and structurally re- 1973).Ribbe and Gibbs concludedthat dis- lated to forsterite. Microprobe analyses of fifty-five tortion and ligancy rather than sizeof the octahedral govern humite samples(Jones et al., 1969)led to the estab- sites the ordering of Fe in Mg humites, Fe preferring (F,OH) lishment of the generalformula: those octahedra with no ligands, i.e.,M(l)Ou and M(2)O6. nlM al . M F,T|,(OH,F)2-2,O2" The recent discovery of Mn humites containing "SiO where M is Mg,Fe,Mn,Ca,Zn in decreasingorder of considerableMg provides an opportunity for further abundance,0 < x < l,andn : I fornorbergite,2for investigationof cation ordering in humites.This pa- chondrodite,3 for humite, and 4 for clinohumite (see per reports the resultsof a site occupancyrefinement Table l). Manganeseend-members of three of the of magnesianmanganhumite. four homologues are well-known, and unpublished Experimental procedures microprobe analysesby C. Richardson(private com- munication, 1976)indicate extensive, if not complete, Crystalsof manganhumitefrom the type specimen from were gen- solid solution betweenMg and Mn end-members. Brattfors Mine, Nordmark, Sweden, Ordering of the small concentrationsof Fe nor- erously donated by ProfessorP. B. Moore, Univer- mally presentin Mg humiteswas initially detectedby sity of Chicago, and upon completion of this study Ribbe and Gibbs (1969,l97l) in a humitewith com- will be deposited in the Mineralogical Museum of Harvard University. Moore (1978) reports their position MguuFeo osirO,rF(OH). Mg/Fe orderinghas composition to be (Mno..rMgo.roFeo.orCao.or)r(SiOa)g (OH),' ' Presentaddress: Mineralogical Museum, Harvard University, Manganhumite is orthorhombic and was refinedin 24 OxfordStreet, Cambridge, Massachusetts 02138. the nonstandardspace group Pbnm to conform with 0003-004x/7 I /09l 0-0874$0200 FRANCIS AND RIBBE: MAGNESIAN MANGANHUMITE 875 Table l. Nomenclatureof humite group mineralsr[M,SiOo]. assigned,and after two cycles of refining the scale M(OH,F), with referencesto moderncrystal structure refinements factor and several more cycles of refining both the Honologue Mg end-menber Mn end-nenber scale factor and the positional parameters,the con- q= 1 Norbergite MAN* ventional R factor dropped to 0.22. At this stage, (clbbs and Ribbe, 1959) : \= 2 Chondrodlre Alleghadylte diffractions of the type h + k 2n i I showed good (clbbs er at., 1970) (Rentzeperis, 1970) agreement between .Fou.and F"ut", while diffractions != 3 Hulte ManganhmiEe (Ribbe and Cibbs, 1971) (This 6rudy) of the type h+k : 2n showed poor agreement.This g= 4 Clinohmite Sonolite systematicdiscrepancy was attributed to Mn/Mg or- (Robinson eE a1 , 1973) (Karo, 1n preparaEion) dering, and when Mg was assignedto the four non- *Manganese Analog of Norbergice (},IAN), synrhesized by CAF: !=4.869(2'), b=10.796(2), c=9.179(2) A. equivalent octahedral sites on the basis of average M-O bond distances,the R factor improved suf- previousstudies (Taylor and West,1928; Ribbe and ficiently to releasethe temperaturefactors. Using the Gibbs, l97l) and the recommendationsof Jones temperature factors, atomic coordinates, and scale (1969).Unit cell dimensions,4 : 4.815(l),b : factor generated in this manner, site refinement fol- 10.580(2),c = 21.448(5)A,were determined by least- lowed. squaresrefinement (program of Appleman and Being the chief substituent,Mg was selectedas the Evans,1973) of 22 powderX-ray lines (16-60"U) independentvariable and, subject to the restriction recordedwith monochromatizedCuKa radiation(X that the total magnesium content must equal 2.1 : l.54l8A) on a PhilipsNorelco powder diffrac- atoms, was first distributed randomly over all four tometerusing BaF, (a : 6.198A)as an internalstan- octahedral sites.The trace levelsof Fe and Ca were dard. ignored throughout the site refinement.The concen- A cubicfragment 0.2 mm on an edgewas mounted trations of Mg in M(2), and M(2). dropped to with b nearlyparallel to thephi axisof a PickerFecs- slightly negative values, so these sites were sub- I four-circlediffractometer. Intensity data were col- sequentlyconsidered to be fully occupiedby Mn. The lectedin two octants(20 < 70") usingNb-filtered site occupancies then converged to the values MoKa, radiation(I : 0.709264)and a20 scanrate 0.304(6)Mg in M(l) and 0.746(6)Mg in M(3), irre- of lo/minute.Background measurements were made spectiveofthe sitechosen as the dependentvariable. for 20 secondson eitherside of dispersion-corrected Data for which Fou" I 2o(Fo6.)were considered scanranges (1.2'm basewidth). unobserved.The 238 data rejectedfrom the refine- Symmetricallyrelated standard diffractions mon- ment by this criterion are indicated in the structure itoredafter every 50 datashowed a maximumvaria- factor table (Table 2)'zby asterisks.In the final cycles bility of t5 percent.No interpolationof thedata was made,but an "ignorancefactor" of 0.025was incor- 2To receivea copyof Table2, order DocumentAM-78-086 from poratedin the calculationof the weightingfunction the BusinessOffice, Mineralogical Society of America,Suite 1000 (Fingerand Prince,1975, p.5). The datawere cor- Lower Level, 1909K Street,N.W., Washington,D.C. 20006. rectedfor background,Lorentz,polarization and ab- Pleaseremit $1.00 in advancefor the microfiche. sorptioneffects (p*o*o: 63.1cm-') and thenaver- Table3. Positionalparameters, isotropic temperature factors and agedto yield a setof 1800unique structure factors r.m.s.equivalents, p, for manganhumite usingthe programs DnrnLIs andDnresonr from the qt(A) Atom xla ytb z/c B(A2) World List of CrystallographicComputer Programs M(1) o.oo1s(1)* 0.37s7(1)0.r7s8(r) 0.90(2) o.107(1) (3rd ed. and supplements). M(2)6 .s141(2) .1s83(r) .2s .63(2) .089(l) M(2)s .008s(r) .0s74(r) .10e1(r) .s3(2) .l0s(i) M(3) .4s01(2) .86s7(l) .0270(L) .84(3) .103(2) Refinement si (r) . 07s0(4) .e6s9(r) .2s .63(2) 089(l) A full-matrixleast-squares refinement was carried si (2) . s7s3(3) .2844(r) .1043(1) s7(2) .o8s(r) out usingthe programRnINn4 (Finger and Prince, o(2.3) .71S0(6) .2178(2) .16s9(l) .84(4) .103(2) 0(1.3) .2168(6) .038r(2) .1896(1) 86(4) .104(2) 1975),and atomic scatteringfactors for neutral o(2.4) .71s7(6) .2130(2) .0447(r) . 7s(4) . 098(3) atomstaken from Volume4 of the InternationalTa- o(2.1) .24s2(7) .28s0(3) .1029(1) .89(4) t06(2) bles for Crystallography(1974, p. 99, 149).The re- o(r 2) .278s(8) .32e3(3) .2s .80(s) .101(3) o(r.1) .742r(s) .s680(4) .2s .s7(6) .1rr(3) finementwas initiatedusing the positionalparame- o(2.2) .7778(6) .92s2(2) .1031(l) .72(4) .0s6(5) tersof humite(Ribbe and Gibbs, l97l) andassuming oH .263r(6) .0306(2) .0333(l) .87(4) .los(2) that all octahedralsites were completely occupied by rNunbers in parentheses are estinated standard deviations and Mn. Reasonableisotropic temperature factors were refer to to the last decinal Place. 876 FRANCIS AND RIBBE: MAGNESIAN MANGANHUMITE of refinement,corrections for anomalousdispersion tancesand angles are reported in Table4. Theatomic and extinctionwere includedwithout significantly designationsfollow thoseused by Ribbeand Gibbs alteringthe results.The final unweightedR factoris (re7l). 0.051(R* : 0.081)for the observeddata; R : 0.058 andR* : 0.082for all 1800data. Atomic coordinates Resultsand conclusions and isotropictemperature factors and their r.m.s. Manganhumiteis confirmedto be isotypicwith equivalentsare listed in Table 3. Interatomicdis- humite,and becauseof its closesimilarity to humite, the readeris referredto Ribbeand Gibbs (1971)for Table 4. Si-O, M-O and O-O distances(in angstroms) and O-Si- structurediagrams and detailed discussion of theste- O and O-M-O angles (in degrees)in manganhumite reochemistry.We note,however, that the apicalSi-O bonddistances of 1.600and 1.6044in manganhumite Isio, I rerrahedra st(1)-o(1,1)a 1.604* sr(2)-o(2,1)A 1.600* (Table 4) are the shortestyet observedin humite- o(1,2) r.645 o(2,2) r.648 group minerals. o(r,3) [2] 1.533 o(2,3) r.640 Mean 1.628 o(2,4) r.540 The cationsin manganhumiteare ordered(Table ltean 1,632 Angles at Angles at 5), the magnesiumbeing concentrated in the M(l)06 0...0 distances* S1(1)** 0...odlstances* sl(2)** and M(3)O.(OH),octahedra, each of whichare "in- (r,3)-(1,2) [21 2.5598 r02.7 <2,3)-(2,2) 2,s74r rO3.O (1,3)-(r,3) 2,59L 105.0 (2,4)-(2,2) 2.570x r02.8 terior" sitesin the serratedchain of edge-sharing (1,2)-(1,1) 2.735 LL4.7 (2,3)-(2,4) 2.599t 104.8 octahedra(Ribbe and Gibbs, 1971,Fig.