Canadion Mineralogist Yol.24, pp. 605-614(1986)
THE CRYSTAL STRUCTURE OF BOBFERGUSONITE
T. SCOTT ERCITI, FRANK C. HAWTHORNE ENOPETRIERNV Depafimentof GeologicalSciences, University of Manitoba, Winnipeg,Manitoba R3T 2N2
ABSTRAcT dant, Al et Fd+ sont ordonndsentre les chaines,ce qui donnedeux chaines distinctes en composition. Le deuxidme Bobfergusonite Na2Mn5Fe3+Al@O/6 is monoclipic, feuillet,identique d sonanalogue dans la wyllieite,contient F2r/n, witb, a 12.776(2),b 12.488(2),c 11.035(2)A, B deuxtypes de chainede polyddresd cationsX (Na,Mn) 97.21(l)", V 1746J(4) 13, Z:4. The crystal structure,a parallblesi X. Unechaine contient une alternance de poly- more highly ordered derivative of the wyllieite structure, tdresd Na et Mn d facespartagdes; I'autre contientexclu- was refined on the basis of a transformed wyllieite struc- sivementdes polyidres i Na d ar€tespartag4s. Ces deux ture, resulting in an R index of 3.8 Vo for 2889 observed sortesde chaine,qui ne sont pasentreJi6es dans le feuil- (3o) reflections measuredon a single crystal of red-brown let, lient les feuilletsde polybdresi cationsM aux tdtra€- holotype material. The structure of bobfergusohiteis topo- dresPO4. logicaly identical to tlat of wyllieite and alluaudite, but differs from both these structure types in terms of its M- (Traduit par Ia Rddaction) cation ordering. It is a layo structure, and has alternations of two types of layer along L One type of layer consists Mots TABLE1. MISCELLANE0USINFoRIIArI0N FoR BoBFERGUS0NITE SfnUCfUnf ANALYSIS a 12.776(2)A Spacegroup P2\/n it was evident that the b r2.488(2) Crysial size 0.18x0.29x0.29m From the precessionstudy, . 1r nlqf2l ll, Radiatjon 44.5 cm-r, MoKo bobfergusonite structure is a more highly ordered B 97.21(1\ : Total Fo, obs. Fo 4204,2889 v r746.7(4)[3 Final R, wR 3.8,3.4g derivative of the wyllieite structure. The bobfer- from the wyllieite struc- 6"116e11qn15.(Z=4):. gusonite structure differs Naz.cllnq.aFe2', orFe3'. zgAl r.oCar gMg.oZn.oPo0z g.a0H.z* ture in having a doubled a period and the spacegroup R = t(lFol-lFcl)/tlFol P2/n, whereasthe wyllieite structure obeysP21/c = B=l wn Irw(liol-lFcl)'z/trlFol'?l'' @icit et at. L986).Comparison of the symmetry of a P2/c cell doubled along Xto the symmetryof a * for charge balance 0H:0 calculated P21/n cell of equivalentdimensions givestwo pos- sibilities for the origin of the bobfergusonitecell with reference to a wyllieite subcell: (0,0,0)ra and tensediffractions, and were adjusted automatical- (12,0,0,)w,where w refers to wyllieite. ly. Backgrounds were measured for half the scan Structure analysis was done with the program X time at the beginningand end of eachscan. Three in the STIELXTL package.Scattering curves for neu- standard reflections were monitored every 45 reflec- tral atoms from Cromer & Mann (1968) and tions for changesin beam intensity or crystal orien- anomalous-dispersioncoefficients from Cromer & tation. All such changes were insignificant. Data Liberman (1970)were used. A direct-methodssolu- were collected over I asylnmetric unit of reciprocal tion of the structure uniquely gave the origin as space,initially to sina/)r= 0.7035,but later only to (0,0,0)w, so that refinementwas initiated with this sin0/r: 0.5946,to reducethe mrmber of data. In all' setting. Co-ordinates for the wyllieite $tructure 4204reflections were collected, of which 2889were (Moore & Molin-Case 1974),transformed to com- consideredobserved (I>3o). ply with its P2r/c setting (Ercit et al. 1986),_were Subsequentto the collection of the main data-set, used as starting values for the refinement. For a additional intensity-data were collected on I I strong (0,0,0)w origin, the suppressionof certain symmetry- diffraction-maxima evenly dispersedover a range of ilements by the doubled'F21/cto P21/n transfor- 6to 53"?1.These data were collectedevery 10" while mation splitseachM-, P- and O-site,X(la) andXQ) rotating each reflection 360' about its diffraction of the parental wyllieite structure into two non- vector (ry'),using the sameset of scan parametersas equivalent sites;X(lb) remains unsplit but degener- the main collection of data. Thesery'-scan data were atis from a specialto a generalposition [seeMoore used as a calibration data-set by an empirical & Molin-Case (1974)for site nomenclatureof the absorption-correction routine in the SHELXTL wyllieite structurel. This resultsin 6Msites, 5 Xsites, packageof programs, a modification of the proce- 6 P sites arLd?llO sitesin bobfergusonite.The site dure of North e/ al. (1968).Absorption correction nomenclature adopted for bobfergusonite and its was done by assuminga pseudo-ellipsoidalshape for relation to the system used for wyllieite- and the crystal, and by refining the lengths and orienta- alluaudite-groupminerals (Moore & Ito 1979,Moore tions of the ellipsoid semi-axe$while holding p'R > L91l) are given in Table 3. constant.Absorption correctionreduced the merg- For early stagesof the refinement, a wyllieite-like iqg R of the ry'-scandata-set from 3.2 to l.6t/o. Data ordering schemewas assumed,and by using the non- reduction was done with a SIIELXTL program, and geneticmethod of Moore & Ito (1979)for the calcu- included background and L.P corrections, and scal- lation of the formula of wyllieite-groupminerals, ini- ing on the standard reflection-data. tial site-populations were assigned (Table 4). Becauseof the variable composition of bobfer- Subsequently,the total occupancyof eachM andX gusonite,the crystalused in the data collectionwas site was refined using mean scattering-curvesbased embeddedin an epoxy mount and was analyzedwith on the wyllieite-like model, while constraining the the electronmicrbprobe, using the sameanalytical temperaturefactors of the membersof eachsplit site techniquesas in Ercit et ul. (1986). The composition to be equal. is given in Table 2; the resulting formula given in Several cycles of refinement of the scale factor, Table I was used for the structure refinement. all positional and isotropic thermal parameterqand th; X- ard M-site occupanciesgave R indices of 9.0' weightedR 8.090.A differenceFourier map at this TABLE2. ELECTON-IIICROPROBEDATA ON BOBFERGUSONITT CRYSTAL stage showed two large loci of residual electron- sites,suggesting Naro Mgo cao l,4n! Feo zno Fe203 A1203 Pros H20 Total densityon oppositesides of the Xl strong anisotropic vibration for atoms of the site. 6.8 0.4 1.1 31.6 0.3 0,4 6.7 7.5 45.1 0.3 100.2 Modeling the atoms of the site for anisotropic vibra- tion reducedthe R indicesto R:6.6, wR:5.7t/0. Fe2":Fe3* by ilossbaler [email protected], H20 by the@gravlmetry. No other strong indications of anisotropic vibration THE CRYSTAL STRUCTURE OF BOBFERGUSONITE 607 weredetected at this stage;refinement of the model In the next stageof refinement of the (0,0,0)w with the (0,0,0)w origin restedhere. model, all atoms were modeled as vibrating Refinementof the model with its origin at (th,0,0)w anisotropically. After several cycles of refinement, began with the transformation of wyllieite co- Ihe M- and X-site occupancies were critically ordinates,which resultedin the samenumber of M examined for the first time. The occupanciesof the X, P and O sitesas the first model; however, the following pairs of sitesrepresenting split sitesof the secondmodel hastheX(la) siteunsplit but degener- wyllieite structure were statisticallyidentical: Ml and ated to a generalposition, and X(lb) split yet remain- M2, M3 andM4, X2 atd X3, X4 and X5, indicat- ing in specialpositions. The refinementprocedure ing that no siglificant cation ordering existsbetween for the secondmodel was the sameas for the firsr. the membersof eachpair. However, the occupan- Refinement of the scale factor, all positional cies of M5 and M6 are not equivalent; the scatter- parameters,isotropic temperature-factorsfor all sites ing from M5 is much stronger than that from M6 exceptthe split X(lb) sites(strongly anisotropic) and (significantwell abovethe 99.890confidence-level). of all X- and M-site occupanciesgave R:8.1, On the basis of a wyllieite substructure, the main wR:6.8V0, clearly inferior to the first model with occupantsof thesesites are Fe and Al; therefore, the 8 : 6.6, wR : 5,7clo. We concludedthat the correq M5 site must be preferentially occupied by Fe, and origin for bobfergusoniteis at (0,0,0)w. the M6 site, by Al. The model of Moore & Ito (1979)for the crystal chemistry of the wyllieite-group minerals indicates TABLE3. SITE-CORRETATIONTASLE FOR THE ALLUAUDITE that the Ml and M2 positions of bobfergusonite STRUCTI'REAND DERIVATIVES shouldbe fully occupiedby Mn; however,the num- ber of electrons at these sites is 990 lower than Al I uaudlte l,lyl1 I ei te Bobfergusonite expected. In all refinements of wyllieite-group minerals to date (Moore & Molin-Case 1974, M(r) M(l) Ml , t42 M(2) M(2a) M3, M4 Zheshenget ol. 1983),the site in the wyllieite struc- M(2b) M5, M6 plus x(l) x(rb) ture correspondingto Ml M2 of bobfergusonite X(la) x2,x3 has always been assumedto be fully occupied by x(2) x(2) x4, i(5 P(1) P(l) Dl D2 either Mn or Fd*. On the basis of data currently P(2) P(2a) P3, P4 available, it is not possibleto determine whether the P(2b) P5, P6 0l 0(1a) 0t, 02 low occupanciesare unique to bobfergusonite,or are 0(rb) typical of wyllieite-group minerals and their deriva- 02 0(2a) 05, 06 0(2b) 07, 08 tives. Furthermore,without more data it cannot be 03 0(3a) 09,010 determined whether the low occupanciesare due to o(3b) 0ll, 012 04 0(4a) 0l3, ol4 vasanciesor to the presenceof elementslighter than 0(4b) 015,016 Mn or Fe at thesesites. Refinementsof two wyllieite 05 0(5a) 017,018 o(sb) 0r9, 020 samplesand a secondcrystal of bobfergusonitehave 0(6a) 021. 022 resolve this, and other 0(6b) 023,024 been initiated in order to ambiguities in the crystal chemistry of the minerals. For the present time it is assumedthat the Ml and IA sfies have small amounts of vacancies. TABLE4. BOBFERGUSONITE:INITIAI ANDFTNAL lit- ANOX-SITE The populafions of the M3 to M6, Xl and,Y2 sites POPULATIONS calculated on a wyllieite-substructure model refined to full occupancy, apparently indicating that the Populatlon Moore &Ito (1979)model adequatelypredicted the bond-valencecal- Initia l chemistryof thesesites. Howevern culations for M3, M4 and M5 are quite nonideal. Ml , r.t2 1.00 iln !r3, M4 0.02 7n + 0.04 trlg-10.02 Fe2** 0.09 Fes++ 0,83 Mn Specifically, M3 and M4 are found to have l7rlo 145,1460.69 Al + 0.31 Fe' was XI 0.55 liln+ 0.19 Ca* 0.26 ila more valenceassociated with them, and M5 x2,x3 1.00Na found to have 1l9o lessvalence associated with it x4,x5 0.40 Na than anticipatedfrom the refined site-populations. Flnal To minimize both problems, all divalent cations Ml 0.92(l) l4n exceptMn2* weretransferred from M3 and M4 to M2 0.91(l)Mn il3, 144 0.82Mn + 0,06(l) Fe3-* 0,12(1)Al M5, it exchangefor trivalent Al and Fe. lilS 0.48(l) Fe" . 0.34(l) 41 " 0.04 Fe2"+ 9.99 I'ls+ 0.05 zn had refined occupancieslower 146 0.80(l) Al + 0.20[t) Fe3' The X2 to X5 sites xl 0.81 Mn+ 0.19 Ca than l; the correspondingsites for the wyllieite struc- x2 0.85(l) Na x3 o.81(l ) Na ture are typically only partly occupied, so that this x4 0.59(1) Na behavior is not unexpected. 0.58(l ) t{a Taking the above points into consideration,the 608 THE CANADIAN MINERALOGIST following occupancy model was refined: TABLE5. B0BFERGUS0NITETP0SITIoNAL AND THERI'IAL PARAIiETERS l. Ca: All Ca was assignedto the Xl site at the z U(eq) microprobe-determined concentration. I'fl o.r3re4(7) 0.23856(7) -0.00190(8) 138(3) ) Na: All Na was assignedto the X2 to X5 sites t42 0.62989(7) 0.23753(7) -0.00'107(8) 137(3) of these sites, and the total M3 0.29562(6) 0.14643(7) 0.72421(7) 113(2) as the sole occupant M4 0.19574(7\ 0.14588(7) 0.72617(B) 125(2) occupancy of each site was refined. M5 0.46125(8) 0.16153(8) 0.2798'(e) 108(3) 0.96067(10) 0,16220(lo) 0.?7950(11) 98(4) 3. Zu,ME: All Zn and Mg wereassigned to the M5 It6 xl 0.24944(lt ) o.o0l03(9) -0.00188(ll) 280(2) site at their microprobe-determined concentra- x2 00 0 309(r4) bond-valence requirements. x3 1/2 0 0 307(r5) tions, to optimize x4 0.3734(5) 0.0169(4) 0.4972(4) 444(re) 4. Al, Fe: The Al:Fe ratio of M6 was refined. x5 0.8738(4) 0.0156[4) 0.5004(4) 377(17) Enough Al + Fe was added to fill the M5 site (in Pl 0.38r6(l ) 0.2183(r) 0.0033(r) 102(4) aU pz 0.8826(l ) 0.2t52(1) 0.0067(1) 'r05(3)105(4) addition to the Zn+Mg already assiened); P3 0.2013(',l ) 0.r'r41 0 ) 0.2578(l) 'roe(3) remaining Al+Fe was split equally among M5 P4 0.7041( r ) 0.r'r 34(1 ) 0.2598(l) 'r18(3) 0.0597[1) 0.0e57(1) 0.7384(l) and M4. By constraininC Al(R) to equal Al(T)- P6 0.5588(r ) 0.10000 ) 0.7400(r) 122(3) N(Ms) and Fe(R) to equal Fe(T)-Fe(M5), and 0l and 02 Ar(M3):AL(MA):Al(R)-Al(M5) 03 Fe(M3)=ts(l,at(1: Fe(R)-Fe(M5), whereAl(R) 04 05 and Fe(R) are Al and Fe at M3 to M5 and Al(T) 06 07 and Fe(T) are the total Al and Fe contents from 08 analysis, the M3- to 09 results of the microprobe 0I0 M5-site Al and Fe contents were determined. 011 012 5. Mn: Mn was assignedto the Xl site with an 0t3 014 irullllillmlililllfillllfilll = occupancyofMn l-Ca, andtotheM3 andM4 015 Mn= l-(Zn+Me+Fe+Al) for each 0'16 siteswith 017 site. All remaining Mn was assumedto reside at 018 019 the Ml and M2 sitesas the sole oocupant of these 020 siteswere 021 sites,and the total occupanciesofthese 022 refined. 023 024 Owing to limitations of the computer programs lfifillllmllllfifililfilll usedin the structure refinement, random error in the All urs are x10". microprobe data was ignored in the refinement. For this final model, the constraint of equal temperature-factorsfor eachmemba of the split sites was relaxed. The model had 372 least-squares Final positional parameters and equivalent parametersand was refined to R indices of 3.8, isotropic iemperature-factorsare ervenin ]ablg-s' The wR=3.4V0, A difference-Fouriermap made at this The final site-populations are given in Table 4. last stage had no residual electron-density maxima standard deviations reported in Table 4 are higher greaterthan 0.45 e /43.T:he largestof thesedid not than those estimated from the refinement by a fac- of correlatewilh any possibleproton positions;conse- tor of 1-3, to reflect the higher imprecision quenfly, all such dtfferenceswere consideredas back- microprobe analysis, on which the site-occupancy ground. refinements were based. Bond lengths are given in The only microprobe-unconstrainedcompositional Table 6, and selectedintrapolyhedral angilesand O-o par€uneterswere the Mn and Na occupancies.The separationsare given in Table 7. Observedand cal- refined Mn and Na contentsof 4.28(1) afi2.0lQ) culated structure-factors and anisotropic atoms per formula untt (Z:4) comparevery well temperature-factors have been submitted to the with microprobe-determinedvalues of 4.21 and2.07 Depository of UnpublishedData, CISTI, National atoms, an indication of the correctnessof the model Reiearch Council of Canada, Ottawa, Canada KIA and the relative freedom from error of the intensity 0s2. data. The M5 and M6 sites were found to partition Al DESCRIPTION OF THE STNUCTT.TNN and Fe very differently. M5 hu a refined Fe:Al ratio the of 1.51, and M6 has an Fe:Al ratio of 0.25. The The bobfergusonite structure is identical to Al:Fe ordering is also reflected in the bond lengths alluaudite structure (Moore l97l) and the wyllieite topol- of the M5 and M6^polyhedra;M5 has a meanbond- structure (Moore & Molin-case 1974)in cross length of 2.025 A, whereasM6, with its greater ogy. The bobfergusonite structure can be described proportiolt of (smaller)Al, has a meanbond-lengfh al-a hyer structure, with alternations of strongly to I! of 1.956A. bonded and weakly bonded layers normal TIIE CRYSTAL STRUCTURE OF BOBFERCUSONITE 609 TABLE6. BOBFERGUSON1TEIBOND LEHEMS (T) types of chain: one type consistsof alternatingXl (Mn) - and X2-X3 (Na) polyhedra in a face-sharing tft 02 2.222(4) t{? - 0l 2.241(3) M3- 0l 2.127(41 - 03 2.194(3) -04 2.2r5(3) - 06 2.100(4) relationship;the other type consistsonly of X4 and - 01o 2.252(41 -09 2.308(4) - oil ?.097(4) X5 (Na) polyhedra, yia - 012 2.238(4) - 0I1 2,236(4) - 018 2.216(4) linked edge-sharing. - 0r4 2.352(4) - 0I3 ?.305(4) - - 0re 2.097(3) The co-ordination polyhedra of the X3 to M6 and 016 2.z',t3(4) - 015 2.231(3\ - 022 1,994(4) < Mr-o > ZZ{5 < l'12-0 > 7:286 < t43-o > z:IOd Xl cations are reasonably octahedral and only - il4 02 2.140(4) r.r5- 03 2.002(4'l |'|6- 04 l.e4l(4) slightly distorted,T\e Ml and luf2polyhedraare very - 05 2.117(4) - 08 1.92t(4) - 07 r.861(4) distorted octahedra, which Moore (1971) - 012 2.083(4) - 010 2.000(4) - 09 r.934(4) has - 017 2.181(4) - 017 2.063(4) - - 018 1.992(4) describedas bifurcated tetragonal pyramids. The co- 020 2.0e6(4) - 020 2.231(4\ - 0r9 2.135(4) - s2r r .998(4) - 024 1.932(4j - 023 1.874(4) ordination polyhedra of X2 andX3 are very distorted < M4-o > z:mT < lit5-0 > z.TzF < l'16-0 > T:t56- cube, the irregularity arising from two anomalously xr - 05 2.186(4) X2-OS x2 2.781(4) x3 - 06 x2 2.835(4) large, symmetrically equivalent edgesin poly- - 06 2,185(4) -07 x2 2.390(4) - 08 x2 2.383(4) each - 013 2.430(4) - 0'14x2 2.393(4) - 013x2 2.385(4) hedron (O5-Ol4 for X2 arLd06-013 forX3, Table - 014 2.364(41 - 016xz 2.707(4) - 015x2 2.711(3) - 0I5 2.150(4) < x2-0 > < x3-o > T:{jq 7). The co-ordination polyhedra of X4 andX5 are - 016 2.t58(4) < xt-o > ZZ{6 diminishedgable disphenoids. x4 - 0l 2.692(6) x5 - 02 2.651(6) - 03 2.851(6) - 04 2.817(6) CanroN OnoenlNc - 012 2,757(61 - 011 2.740(6) - 02r 2.560(6) - 022 2.564(6) - 022 2.467(7\ - o2't 2.465(7\ - 0t4a 2.701(6) - 023a2.738(6) The cation-ordering schemeof the bobfergusonite - o2qt z.soz(ti - 023b2,480(6) structure is very similar to that of the wyllieite struc- < X4-0 > Z:6ifd < x5-o > 2:6-15' ture. Complete disorder of cations existsbetween the Pr - 02 t.528(4) P2- 01 1.528(4) P3-09 'r.535(4)1.546(4) - 04 r.544(4) - 03 1.538(4) _ 016 pseudosymmetricalpairs of sitesMl-M2, M3-M4, - 06 1.547(4) - 05 1.544(4) - 017 1.545(4) X2-X3 and X4-X5; however, - 08 r.532(4) - 07 t.533(4) - 021 1.50e(4) significant Al versus < Pl-o > T:536 < P2-0 > T:5t6' < P3-0 > Fe3+(+F*+ +ll'4:g+Zn) ordering occurs between 'r.548(4) P4 - 010 r.543(4) P5- 0ll r.543(4) P6- 012 the M5 andM6 sites,which is not permitted in the - 015 1.544(4) - 014 r.536(4) - 013 r.s33(4) - 0t8 1.544(4) - 020 1.536(4) - 0r9 r.549(4) wyllieite structure, M6 showsa strong preferencefor - 022 1.5lr (4) - 023 r.525(4) - 024 1.533(4) N; M5 prefers the larger Fe3* and lower-charged < P4-0 > < P5-o > T:535 < P6-0 > T:F4T cations such as F4* , Zn and Mg. The validity of using a wyllieite substructure as a starting point for determining the ordering scheme which accountsfor the perfect {010} cleavage.The of cationsin bobfergusonitewas proven by the low repeat pedod along I is four layers thick. R-indicesand by the good agreementbetween the The strongly bondedlayer has a mean intralayer Mn2* and Na+ contentsfrom the structure refine- Pauling bond-strengthof 0.6 v.u. It consistsof M- ment and microprobe-determinedvalues. flowever, cation polyhedra (distorted octahedra)and phos- mean polyhedral bond-lengths and bond-valence phate tetrahedra. The octahedraof the layer link vla sums can also be usedto assessthe validity of the edge-sharingin cir configurations to form very stag- model. In Table 8, a comparisonis made between geredchains along [01] (Fig. l). Thesechains are the observed mean polyhedral bond-lengths and cross-linkedby isolated phosphatetetrahedra; the valuescalculated from Shannon(1976a), using the P-O bonds of the tetrahedraare shown as spokes final site-populationsof Table 4. In general,agree- in Figure l. In terms of M-cation orderingthere are ment betweenthe observedand calculated mean 1wo types of chain: one type essentiallyconsists of bond-lengthsis reasonable.Differences increase with only Mn (Ml andM4 sites)and Fe3* (M5 site), the increasingbond-length; this is to be expected,as the other of Mn (M2 and M3 sites) and Al (M6 site). weaker the averagebond-valence to a cation, the One-third of all phosphatetetrahedra are completely more easily the mean bond-length is perturbed from bonded within the layer; the remaining two-thirds ideality by such factors as polyhedron distortion. haveone P-O bond of eachtetrahedron involved in With only 6090 occupancy of the X4 and X5 sites cross-linkingthe layers of the structure. by Na, there is a significant increasein the mean The weakly bonded layer has a mean intralayer bondJength over that expected;this effect has been Pauling bond-strengthof 0.2 v.u. It consistsof X- well documentedin Li compounds(Shannon 1976b). cation polyhedra only (Fig. 2). The Xl site is mainly Bond valences@rown 1981)may also be usedto occupiedby Mn and has an octahedralco-ordination assessthe validity of the results. Becauseof the large polyhedron (stippledin Fig. 2). The X2 to X5 sites number of sites, only bond-valencesums to the sites are occupiedby Na. The X2 and X3 cations are 8-co- are presentedhere (Table 9). The following section ordinate; the X4 and X5 cations are 7-co-ordinate summarizesthe results: (Na-O bonds are shown as spokesin Fig. 2). The l. Ml, IuI2: The bond-valencesums are lL-140/o co-ordination polyhedra of the X cations form lower than expectedfor a model with 9l-92t/o straight chainsparallel to the X axis; thesechains occupancyof Mn at thesesites, which may indi- are not cross-linkedwithin the layer. There are two cate the presenceof minor Na. 610 THE CANADIAN'MINERALOGIST IABLE7. BOBFERGUSONITE:SELECTED ANGLES (O), O.OSEPARATIONS (E) l'll Polyhedror I'12Polyhedron l'll3 Polyhedron 0t-0l l 78.4(l),2.669(5) -018 82,2(2\, 2.856(51 ":313?l:3[f]: l:3llt3] ",il -019 83.00). 2.7e9(5) 'l?:itl ,?i,?iii,i,iiiiii -022 '|r3.6(2),3.511(7)97.4(2),3,0e9(6) :3ll l:3:3l3tll 06-0ll -:ii; -018 79,8(1 2.76e(5) ,.:til ), i!.:iliii.iiii;i -019 83.2(1),2.788(6) ;i:?lili:liliii -022 10r.3(2),3.168(6) 0ll-018 94.4(l),3.165(6) ,il.liji: -022 82.2(2) 2.693(5) ,,,1u,li':iil, ,':lli , i.iiiiii 7e.4(r),2.759(5) i.i?:lli 018-019 ::]ssisisl;!li. ol9-022 103.9(2),3.225(6) siilili.ffflii fffli: i]iii8i <0-M3-0> Pl Tetrahedron P2 Tetfahedron P3 Tetrahedron *:31i33:3[!]: i:i]8til o':3i 133:3[iJ: l:i8tli] ":!ii il8:?[?i: i:ii!l!i *:33li:::iii: ?:ci:i o.-31lll:3iij: !:i8s[i] ',:3i] 139:?[i]: i:il?l8l ,,:lglffi:3iti:,:xifisi,,:31 - l3i:6[ti:,:xt8[s] liU[3]:ilEl3] <0-pl-0' T095 .0-P2-0>TO9- "'-3i1<0-P3-0>109.5 io-ot 2.510 <0-0> 2 '506 <0-0> 2 '504 P4 Tetrahedron P5 Tetrahedron P6 Tetrahedron 0r0-0ls |10.9(2),2.542(5 0r1-or4 110.6(2),2.531(6) 0r2-013llo.5(2), 2.531(6) -018 r08.3(2),2.502(s -o2o 107.'t12\,2.485(6) -0r9'r08.4(2),2.sll(6) -022 r08.3(2),2.475(5 -023 r08.1(2), 2.483(5) -024 r08.5(2),2.sol (6) 0rs-018 107.4(2),2.488(5 0'r4-020108.8(2), 2.497(5) 013-019108.s(2), 2.502(5) -022 111.2(2),2.520(5 -023 il0.2(2),2.510(5) -024 1r'l.l(2), 2.s29(5) 109.7(2\, 2.520(6) olB-022il0.7(2), 2,514(5 020-023'l09,5111 .4(2), 2.528(6) 019-024 .o-P4-o> T09:5 <0-P5-0> M3-M5: As mentionedearlier, the starting popu- within expectations. There is no strong indica- lations for these sites (Table 4) gave quite tion of any preferredsites for O-for-OH substi- nonideal results, necessitatinga transfer of Zn, tution. Mg and Fd" from M3 andM4to M5, in return The observedcation-ordering scheme for bobfer- for Al and Fe3*. This reduced the observed gusonite indicates that its structural formula is: ver$rs expectedbond-valence discrepancies from +l7Vo to +1190 for M3 and M4, and from 1x54x441[x32n2]xl4 -llVo to -7Vo for M5. tM L4tW2 4lW 4tt4441 M5 41464(POJza J. M6, Xl-X3: The bond-valencesums to thesesites where, ideally, )(2-X5:Na, Xl, Ml-M4:Mn, are well within expectations. M5:F€+ , andM6: N, Full occupancyof all sites 4. X4, X5: The sums to these sites are low by by the cations indicated would give an exceischarge approximately 2090with respectto a model with of +4 per 24(POr); however, the X4 and X5 sites 58-5990occupancy of thesesites, a reflection of are only half-occupied by Na, thus neutralizing the the bond-length perturbations with low site- excesspositive charge. The simplified ideal formula occupancydiscussed earlier. is flNarMn5Fe3+Al@O)6(Z = 4). ). P1-P6: All sumsto the P sitesare within expec- Ordering €rmongthe X-cation sitesof the bobfer- tations. gusonite structure is identical to that in the wyllieite 6. 01424: All bond-valencesums to the O sitesare structure, although a potential for higher degreesof It c l FIo. l. The strongly bonded layer of the bobfergusonite structure. M-cation poly- hedralink ula edge-sharingto form staggeredch:iins parallel to [l0l]. Key to the polyhedra: Mn, denselystippled; Fe3+, liebtly stippled; Al, unshaded.P-O bonds of PO4 tetrahedra at the samelevel as, to slightly above, these octahedral chains are shown as spokes(bold rule). 6t2 THE CANADIAN MINERALOGIST a+l I c l FIc.2. The weakly bonded layer of the bobfergusonitestructure. The Xl polyhe- dron is an octahedron, and is representedin densestippling' to signify occupancy by Mn. The.f2-X5 sitesare occupiedby Na; Na-O bonds of their co-ordination polyhedra are representedas spokes.Spokes without arrows denotebonding within the layer; spokes with arrows denote bonds that serve to link adjacent weakly and strongly bonded layers, BONDVALENCE SUMS (v.u.) TABLE 8. B0BFERGUSONITE:l,lEfrt'l M- AND X-SITE BONDLEN0THS ([) TABLE9. B0BFERCUS0NITE: Site Exp. Bond obs. Calc. Bond obs. Calc. Calc. ExP. '1.99 > < > 2.24 t41 1.61 1.8 0r il1.0 2.245 2.20 xl-o 2.246 '| 02 2.00 I'12-0> 2.256 2.20 < x2-0 > 2.568 t42 .54 I .8 I'13 2.42 2.2 03 2.06 M3-0 > 2.106 2.16 < x3-0 > 2.578 2.05 > < >* 2.646 2.50 l,l4 2.44 2.2 04 t'14-0 2.103 2.16 x4-0 05 2.05 l't5-0 > 2.025 1.99 < x5-0 >* 2 -50 M5 2.65 2.4 146 2.90 3 06 2.05 I'16-0> 1.956 1.93 07 2,04 08 2.01 '| (1976a)' .95 2 09 1.95 Bond lengths calculated from the lonlc radll of shannon uslng 010 1.96 the slte populatJons of Table 4. \2 0.94 0.85 l3 0.89 0.81 011 1.97 012 L96 * only 601 occupJed, x4 0.46 0.58 x5 0.41 0-59 013 1.86 014 1.86 015 1.98 P1 016 2.02 P2 5.06 017 I.95 order exists. The alluaudite structure is lessordered P3 5.09 018 1.96 P4 5.06 019 1.91 than these; specifically, intrachain ordering of X 5 .07 020 1,90 P6 4.99 021 2.06 cations is not shown (Moore & Molin-Case 1974). 022 2.05 y',J3+- - 023 2.04 Figure 3 illustrates idealizsd Fe3+ Mz+ 024 1.95 cation ordering in the M-cation octahedral chains of the alluaudite, wyllieite and bobfergusonite struc- Bond valences calculated fron Brc*n (1981). Expected bond-valences tures. The M-cation chains of the alluaudite struc- fron Table 4 slte Populatlons. ture (Fig. 3a) are host to Fe3* and a wide range of divalent cations,but Al is presentonly in subordinate quantities, and plays a crystallogaphicaly indistinct dis- role from Fe3+. Consequently,Al is not a variable cation order rangesfrom completeif"-cation (top 3a) to complete in the ordering schemeof the alluaudite structure. order over both sites of Fig. - one site and The alluaudite structurehas only two crystallographi- Fe3* IuPj order, i.e., with all IUP*at (bottom 3a). cally distinct M-cztion sites (Table 3);Fe3+ - luF all Fe3* at the other of Fig. T}IE CRYSTAL STRUCTURE OF BOBFERGUSONITE 6t3 a b c FIc.3. Al - Fe3+- M+ cation ordering in the alluaudite (a), wyllieite (D), and bobfergusonite(c) structure-types. Polyhedron shading is as in Figure l, except that the useof densestippling is extendedto representpolyhedra of any.rll+ cation. 614 THE CANADIAN MINEMLOGIST Wyllieite-group minerals are typified by dramati spectra, and R.A. Ramik of the Royal Ontario cally higher Al-contents than alluaudite-group Museum for the TG analysis. Financial support for minerals (Moore & Ito 1979);consequently, Al takes this work was provided by Natural Sciencesand a crystallographicaly distinct role from that of Fe3* Engineering$.esearch Council of Canadaoperating in the wyllieite structure. The wyllieite structure has grants to PC and FCH, and by the Canada- three crystallogaphicaily distinct M-cation sites.The Manitoba Interim Agreement on Mineral Develop- most disordereddamples of wyllieite should have Al ment, 1982-1984. in one M site lMQb)], arrd all luP* disorderedover the other two siteslM(2a) alrdM(l); top of Fig. 3bl. REFEP.ENCES The most highly ordered samples of wyllieite have Al, Fe3* and Fd+ in separateM sites (bottom of BRowN, I.D. (1981): The bond-valence method: an Fig. 3b). As with the alluauditestructure, M'cation empirical approach to chemical structure and bond- Bonding in Crystals I! (M. ordering produces only one crystallographically dis- ing; /r Structure and OXeeffe & A. Nawotsky, eds.). Academic Press' type of chain for each structure. tinct New York. Nonideal cation-ordering in wyllieite results in minor alluaudite-structure character. Specifically' Cnovrn. D.T. & LBSRMAN,D. (1970): Relativistic cal- alluaudite-like site mixing in wyllieite-croup minerals culation of anomalous scattering factors for X-rays. results in the presenceof minor Al at the Fe3* site I. Chem.Phys.53' 1891-1898. IMQ)], and of minor Fe3+ atd lu?* at the Al site lM(?h)1, as with type ferrowyllieite (Moore & Molin- & MaxN, J. (1968): X-ray scattering factors Case'1974).None of the wyllieite samplesdescribed computedfrom numericalHartree-Fock wave func- in Moore & Ito (1979)show idealM cation order of tiors. Acta Cryst. A24,321'324. the wyllieite structure; all have a slight to significant' EncIr,T.S., AuornsoN,A.J., 6ERNf,P. & Hewnro-nNs, but not dominant, alluaudite-structure character. F.c. (liS6): Bobfergusonite:a newprimary phos- The bobfergusonite structure (Fig. 3c) has Al tak- ohate'mineralfrom CrossLake, Manitoba. Can' ing a crystallographically distinct role from that of Mineral. 24, 599'604. Fe3+, but unlike the wyllieite structure, additional ordering results in nonequivalence of adjacent M- Moonr, P.B. (1971): Crystal chemistry of the al- cation chains. The bobfergusonite structure has six luauditestructure type: contributionto the paragen- gSant crystallographicallydistinct sites;however, only four esisof pegmatitephosphate crystals.Amer. 5. of these sites are chemically distinct in the refined Mineral. 56, 1955-197 refined structure has two crystallo- structure. The & Iro, J. (1979):Alluaudites, wyllieites, arroja- Ideally graphically distinct types of M-qtion chains. dites: crystal chemistry and nomenclatute. Mineral, one such chain consists of Al at one M site, with Mag. 43,227-235. lW+ disordered over its two remaining M sites; the other lype of chain has a similar pattern of order- & Mot tN-Cese, J . (1974):Contribution to peg- ing, but has Fe3* in place of Al. However, the matite phosphate giant crystal paragenesis.II. The refined structure has a minor wyllieite-structure or crystal chemistry of wyllieite, Na2Fe2+2Al[POa]3'a alluaudite-structurecharacter. Wyllieite-like site mix- primary phase. Amer. Mineral. 59' 280-290' ing should result in exchange of Al and Fe3* PHrt-rps, D.C. & MeruBws, F.S. between M5 and M6: alluaudrte-like site mixing Nontr, A.C.T., (1968):A semi-empiricalmethod of absorption cor- between should result in Al and Fe3* exchange rection. Acta CrYst. XU,35l-359. M5-M6 andM3-M4. Although the data suggestthat alluaudite-like site mixing is prevalent, more bob- SsarNoN, R.D. (19?6a)l Revised effective ionic radii fergusonite refinements are needed in order to be and systematic studies of interatomic distances in conclusive.The implication from Moore &Ito (L979) halides and chalcogenides. Acta Cryst' 432, is that postcrystallizationoxidation of wyllieite-group 75t-767. mirrerals and of bobfergusonite may be partly of interatomicdis- responsible for nonideal cation ordering in the two (1976b):Systematic studies in oxides.,tz The Physicsand Chemistryof of structure; additional structure-refinements tances types Mineralsand Rocks(R.G.J' Strens,ed.). J. Wiley of wyllieite and bobfergusonite will be done to test & Sons.London. this model. ZnrsrsNc, Me, Nrcrnlrc, Sru & Znrzsor.rc,psN; (tSAl): - qin- AcrNowrnncplanNrs Crystal structure of a new phosphatic mineral gheiite. Scientia Sinica BXi, 876884. The authors would like to thank C.A. McCam- mon, formerly of the Department of Physics, Received October 23, 1985,revised manwcript accepted University of Manitoba for collectingthe Mdssbauer September 17, 1986.