Scapolite Crystal Chemistry: Aluminum-Silicon Distributions

American Mineralogist, Volume 61, pages 864-877, 1976

Scapolitecrystal chemistry:aluminum-silicon distributions, carbonate group disorder, and thermal expansion

LoutssLrvtnu a,No J. J. Peprrr Department of Earth and Space Sciences State Uniuersily of New York, Stony Brook, New York 11794

Abstract

Crystal structure parameters have been determined for a compositionally intermediate scapolite (Na, nrCa,.rrK' ro) (Si8.0sAl3e5)Or4Cl' u"(COr)0.(SOr)o on in space group P4r/nbefore and after a heating cycle. In addition, unit-cell parameters have been determined with room- temperaturedata collected before and after heating, and with data collectedat 400",600o, 700', 800',900o and 1000"C. Both a and volume increasewith temperaturewhile c remains constant.Thermal expansionof a resultsfrom rotation of four-memberedrings of tetrahedra in the (001) plane. Tetrahedral bond distances suggestthat this sample has a highly ordered Si-Al distribution with Tl and 73 occupied by Sio+ and T2 occupied by Al3+. The irreversible change in tetrahedral bond distancesafter the sample was heated to 1000'C suggestsa slight amount of disordering has taken place. This change also affectedthe unit-cell parameters.The carbonate group was refined as a rigid body with realistic bond distancesand anglesusing the least-squarescomputer program RrtNE4. The refinement confirms the model of Papike and Stephenson(1966) and suggeststhat the group tilts only slightly out of the (001) plane. Normalized primitive reflections which violate 14/m symmetry decreasein intensity with temperatureup to 1000'C but do not dissappear.

Introduction and that of a Ca-richscapolite by Papikeand Ste- phenson(1966). Lin and Burley(1973a,1973c,1975) Scapolites,a group of rock-formingsilicates that reportedthree additional structural refinements. The exhibitmany structuralcomplexities, can, as a first spacegroup of scapoliteis a functionof composition. approximation,be consideredas solid solutions The theoreticalend-members, marialite and meion- between marialite, NaoAlrSi"O2oCl,and meionite, ite, display diffraction symmetry consistentwith CaoAl.SiuO2nCO..These formulae are written in the spacegroup 14/m, bfi intermediatecompositions form mostconsistent with the scapolitestructure, but havereflections that violatethis symmetryand reduce can also be written as marialite,3NaAlSirOr.NaCl the spacegroup to a primitive type. Papikeand Ste- and meionite,3CaAlrSirOr.CaCO, to showan anal- phenson(1966) reported diffuse reflections violating ogy with plagioclasefeldspar chemistry. Three thebody-centered symmetry but did not treatthem in coupledsubstitutions are evidentin this solid solu- the structuralrefinement. Lin and Burley (1973a, tion series:Na'+ c Ca2+'Si4+ = Al3+;Cl'- c CO3-. 1973b,1973c,1975)state that the true spacegroup is All threesubstitutions are activein the more sodic P42/n,and their threerefinements are reported in this part of the series[Ca/(Ca + Na) < 0.75]while in the spacegroup, eventhough they did not observereflec- lesssodic part lCa/(Ca * Na) > 0.751,the sub- tions violating 14/m symmetryin their meionite-rich stitutionis NaSi z=lCaAl, the sameas in the plagio- scapolite.Ulbrich (1973)also statesthat the space clasefeldspars (Evans et a|.,1969).AtCa/(Ca * Na) group for intermediatecompositions is P4"/n. Pha- > 0.75 the anion site is filled with CO3- (papike, key and Ghose(1972) and Buseckand Iijima (1974), 1964;Evans et al.,1969). however,conclude from electrondiffraction studies The first reasonablycomplete structure models of that thespace group for intermediatecompositions is scapolitewere reported by Pauling(1930) and Schie- P4 or P4/m. bold and Seumel(1932). The structureof a Na-rich Infrared spectra (Papike, 1964; Schwarcz and scapolitewas first refined by Papikeand Zoltai (1965) Speelman,1965) suggest that the carbonategroup is,

864 SCAPOLITE CRYSTAL CHEM ISTRY 865 in fact, a trigonal group similar to those in other controlled Picker four-circle diffractometer with substances.The roleof a trigonalcarbonate group in MoK" radiation(0.7107A) monochromatized by a a tetragonalmineral was first addressedby Papike graphitecrystal. A total of 1778independent reflec- and Stephenson(1966), who suggestedthat the car- tionswere measured in the before-heatingexperiment bonate group in each unit cell is displacedoff the and 1657in the after-heatingexperiment. Intensities four-fold axis(14/m) and disorderedover four differ- werecollected using an o-N scan(2olmin) with five- ent positionsin the (001)plane. Because scapolites secondbackground counts on eachside of the peak, are volatile-containingsilicates, there has recently and then convertedto structurefactors by applying beenrenewed petrologic interest in them as possible Lorentzand polarizationcorrections, but no absorp- storagesites for volatilesin the lower crust or upper tion corrections(pr : l2.2cm-'). A standardreflec- mantle.Lovering and White (1964)described sulfur- tion, (040), which was repeatedevery 20 reflections rich scapoliteas a primary constituentin granulite duringdata collection, showed less than one standard inclusionsoccurring with eclogiteand ultrabasic deviationvariation in its structurefactor. rocksin the breccia-and basalt-filledpipes ofeastern Initial positionalparameters and temperaturefac- Australia.Newton and Goldsmith (1975) report that tors for all atomsexcept the CO3- group weretaken meionite,CanAl.Si.OrrCOs, has an unusuallyhigh from Lin and Burley(1973a). Original carbonate po- thermal stability, exceeding1500"C at 20 kbar. sitions were estimatedfrom Fourier and difference Meionitethus may play a role asa primary mineralin Fouriermaps. deep-seatedmagmatic processes, and may be a major Both isotropicand anisotropicrefinements were storagesite for COzin the lower crust. executedusing scatteringfactors from Doyle and In this study we haveinvestigated: (l) the Al-Si Turner (1968).The structurewas initially refined in distribution in a scapolitederived from a slowly the space group P4r/n using the least-squarespro- cooledmetamorphic rock; (2) the preciselocation of gram,RrtNn2, written by L. W. Finger.Final refine- the carbonategroup within the scapolitestructure; mentsused RnINn4, (Finger and Prince,1975) a pro- and (3) the thermalexpansion of this mineral. gram which permittedrefinement of the CO!- group asa rigidbody. The carbonategroup was constrained distanceof 1.34. The Experimental to be planarwith a fixedC-O refinedparameters for the group includedthe posi- Data werecollected on a singlecrystal from sample tion of the center,the rotation and tilt of the triangle ON-6 (Naz.r7Ca,.rrKo.ro) (SiE.06AL.eu)O24C10.6e(COs)o.s? of oxygens,and thermal parameters corresponding to (SO,)o.on(Shaw, 1960) (0.25 mm X 0.15mm X 0.10 the motionassociated with uniaxialtranslation of the mm). Thissample is from a scapolite-pyroxenegran- entiregroup. ulite outcrop in Monmouth Township, Ontario Errors on isotropictemperature factors were calcu- (Lin and Burley, 1973b). latedusing the programBcoNv, and errorson bond The crystalused was found to beof suitablequality distanceswere calculatedusing the programBoNux. by examiningWeissenberg photographs, which dis- Both programswere written by B. A. Wechslerand played sharp spotswith no split reflections.Next it usediagonal matrix elements only. was mountedparallel to its a axis on a quartz fiber In the anisotropicrefinement of the before-heating with specially prepared high-temperaturecement data, all 1660reflections of the P4z/n data set were (Brown et al., 1973).The crystaland fiber werethen acceptedand resultedin a final unweightedR of 0.072 placedin a silica-glasscapillary (0.5 mm in diameter) and a final weightedR of 0.052.In the anisotropic which was evacuatedand sealed. refinementof the after-heating data, again all reflec- Becauseof the scapolitespace-group controversy, tions wereaccepted resulting in a final unweightedR the first data set was collected assumingno ex- of 0.080and a final weightedR of 0.054.The weight- tinctions. All h + k : odd for / : 0 reflectionswere ing schemewas W : l/or2 whereap is the standard found to be indistinguishablefrom background,and deviation of the structurefactor basedon counting wereassumed to be of zero intensity.This suggestsstatistics(Prewitt and Sleight,1968). P4r/n as the most likely choicefor the scapolitespace Cell-parameterdeterminations at eachtemperature group. The primitive reflectionswhich violate I4/m weremade using 2d valuesof 13 to 17 independent symmetry(h + k + / : odd) werefound to be present reflections.2d values were collected at eachsuccessive and were treated in the refinement.Integrated in- temperatureas temperaturewas increased,and then tensitieswere measuredusing a PDP-15 computer- at room temperatureafter the heatingcycle. The cell L. LEVIEN AND J. J. PAPIKE

Tlnt-s 2. Final positional parametersand equivalentisotropic TesI-r 2, continued temperaturefactors (A) for scapolitebefore and after heating. Theseparameters are (/r, t/e,%) greaterthan thosereported for Before After the 14/m spacegroup by Papikeand Zoltai (1965) Heating Heating

Before After x 0.36(7) 0.36(7) y Atom Heating Heating 0.28(7) 0.28(7) z 0.24(7) 0.24(7) B 4.3(4) Na,Ca x 0.6136(l) 0.6127(l) 4.3(4) y 0.5361(l ) 0.5358(r) ^^ z 0.7641(2) 0.7634(2) vr x 0.17(7) 0.17(7) B 1.ee(2) 2.14(3) y 0.28(7) 0.28(7) z 0.24(7) 0.24(7 ) cl x 0.7500 0.7500 B 4.3(4) 4.3(4) y 0.7500 0.7500 z 0.7500 0.7500 B 3.87(e) 5.02(e) parameterswere refined using the program CELRT, x 0.5892(l) 0.ssel ( I C. T. Prewitt'smodified version of Poonx. y 0.6592(l) 0.659s (I z 0.2469(1) 0.2472(2 Observedand calculatedstructure factors are listed B 0.57(r) 0.65(r) in Table l,' positionalparameters and equivalent x 0.9097(l) 0.e0e7 (I isotropictemperature factors in Table 2, interatomic y 0.l6se(l) 0.1662(r distanceswithin tetrahedrain Table 3. interatomic z 0.04e2(1) 0.0487 (r B 0.6s(2) 0.6e(2) anglesof linking tetrahedrain Table4, interatomic distancesbetween the Na,Ca siteand its coordinating IJ x 0.6626(l) 0.662e(r) oxygensand Cl in Table 5, betas for each of the y 0.0854(l) 0.08s7(1) z 0.961s(l) 0.e613(r) positionalparameters in Table6, andcell parameters B 0.60(2) 0.68(2) for the crystalat room temperatureboth beforeand x 0.7084(2) 0.7080(2) afterheating and at 400o,600o, 700o, 800o, 900o, and y 0.6016(2) 0.6011(2)1000"Cin Table7. z 0.2523(3) 0.2525(4) B r.r7(4) 1.27 (5) Scapolitestructure 02 x 0.9430(2) 0.9424(2 y 0.1268(2) 0.1 277 (2 The scapolitesare framework aluminosilicates with z 0.2626(3) 0.2617(4 a unique arrangementof 4-memberedtetrahedral B r.r4(4) r.2e(5) ringswhich form two distincttypes of channels,the 03 x 0.6000(2) 0.6005(2) first of which containseither Na1+ or Ca'+. These y 0.1e06(2) o.rer5(2) oval-shapedchannels run parallelto the c axisand z 0.0447(3) 0.043e (4 ) B r.32(s) r.43(6) arebest shown in FiguresI and2, projectionsdown the c axis of marialiteand meioniterespectively. The 04 x 0.7949(2) 0.7es3(2) y 0.0e64(2) 0.0s74(2) secondchannel type containsthe volatileconstituent, z 0.9681(3) o.e685 (4 ) CO3 shownin Figure2 or Cll- shownin FiguresI B r.28(5) 1.44(6) and 3. FigureJ, which is a projectiondown the a, axis 05 x 0.5202(2) o.5re8(2 ) perpendicularto sectionAA'in Figurel, showsthe y 0.617e(2) 0.6r84(2) five-membered z 0.0774(3) 0.0767(4) tetrahedralrings where thb aluminum B 1.42(6) r.56(7) avoidancerule is brokenas meionitecontent is increased.Meionite contains half aluminumand half x 0.6216(2 0.6216(2) y 0.9783(2 0.e788(2) silicon tetrahedra,so half of the five-memberedtet- z 0.0775(3 0.0766 (4 ) rahedralrings must contain3 Al3+and 2 Sia+tetra- B r.20(5) r.43(7) hedra.One oxygenmust link two aluminums,thus x 0.?6(2) o.26(2) breakingthe aluminum avoidancerule. The solid- v 0.23(2) 0.23(2) z 0.24(2) 0.24(2) solution limit (for ordered scapolitestructures) be- B 4.3(4) 4.3(4) I n7 To obtain observed and calculated structure factors from the x 0.27(7) 0.27(7 ) final cycle of each anisotropic refinement, document y 0.12(7) 0.12(7) order AM-76- z 0.2s(7) 0.25(7) 021 from the BusinessOffice, Mineralogical Society of America, B 4.3(4) 4.3(4) Suite 1000 lower level, 1909 K Street, N. W., Washington, D. C., 20006. Pleaseremit $1.00 for the microfiche. 867 SCAPOLITE C RYSTAL CH EM I STRY

anglesof linkingtetrahedra TesLr 3. Interatomicdistances within tetrahedra TrsLs 4. Interatomic

After After Before Before Heating Heating Heati ng Heati ng ' 0l -T1-01 110.4(r) ilr.o(l) r-o distances E) 0l -Tl -05 rro.6(r ) 110.4(2) 0r -Tl -06 loB.o(1) r07.e(2) T1-01 r.5e8(2) .60l(3) ' ' 05-T1-01 ilr.r(l) 110.7(2) Tl -0.l r.605(3) .607(3) -01' roB.8(l r08.e(2) -05 .617(3) 06-T1 ) Tt r.6ro(2) 05-Tl-06 r07.e(l) r07.e(1) Tt-06 1.625(2) .632(3) mean I09.5 109.5 mean I.6t0 Al ( 02-r2-03 t07.B(r) r07.8(l) r?-02 '1.736(3)1.732(2) 1.725(3 02-r?-04 il2.6(1) il2.5(l) T2-03 '1.723(31.724(3 02-T2-05 r04.3(l) 104.3(l) r2-04 1.732(3) 03-T2-04 I15.0(r) il4.6(r) 1.742(3) r.734(3 03-T2-05 ilr.e(r) |r 2.r(l ) T2-05 105.2(1) mean I .736 1.727 04-T2-05 104.8(l) mean 109.4 109.4 r3-02 1.608(2) t.607(3) rr4.2(l l .6ll (3) 02-T3-03 1r4.2(l) ) T3-03 1.605(3) r08.2(l r08.3(l) 1.603(3) 1.607(3) 02-T3-04 ) T3-04 '1.638(2) 02-T3-06 r04.7(l) r04.8(l) T3-06 1.637(3) 03-T3-04 il3.1(l'r05.8(l)) r2.s(r) mean 1.614 1.615 03-T3-06 r06.o(l) |r0.5(l il0.8(l) o-o distances (fr) 04-T3-06 ) mean 109.4 109.4 Tl Tetrahedron' ' 0t-0r 2.63r(3) 2.644(3) Tl-0] -Tl 'Ir5e.5(2) r59.0(2 0t-05 2.637(3) 2.643(4) r2-02-13 3e.0(2) r39.5(2 T2-03-T3 145.e(2) 146.4(2 01-06 2.608(3) ?.614(4) '| (2 '-05 (3) 2.653(4 T2-04-T3 r47.e(2) 48.5 0l 2.6sr ) -05-T2 r37.r(2) r37.1(2 '-06 ?.626(3) 2.636(4) T't '138.4(2 0l Tt-06-T3 r38.s(2) 05-06 2.617(3) 2.627(4) mean 2.628 ?.636 T2 Tetrahedron 02-03 2.803(3) 2.786(4) 1000'C, where it undergoespermanent shortening' 2.868(4 02-04 2.883(3) ) change in volume with temper- 2.743(3) 2.731(4) Figure 4c shows the 02-05 again is 03-04 2.s25(3) 2.eot(3) ature, which is linear until l000oc, and (4) 03-0s 2.881(3) 2.867 anomalously large at 1000"C and after heating' We 2.i53(3) 2.746(3) 04-05 permanent changes in cell parameters 2.831 2.816 believe these mean Al3+ T3 Tetrahedron are causedby a small amount of disorderingof (4) 02-03 2.698(2) 2.701 and Sin+in the three tetrahedral sites. 2.602(3) 2.605(4) 02-04 3 lists the tetrahedral bond distances for 02-06 2.570(3) 2.570(4) Table 03-04 2.676(3) 2.675(3) both the before-heating and after-heating data' The 03-06 2.587(3) 2.5e4(3) bond distancesin the before-heating data are consist- 04-06 2.663(3) 2.670(4) mean 2.633 2.636 Te,sI-B5. Interatomic distancesof the Na,Ca coordination PolYhedron yond which the aluminumavoidance rule must be Distance (A) : very closeto the violated occurswhen Al/Si Vz, Before After ratio which occursin samPleON-6. Atom Heating Heating Aluminum-silicondistributions Na,Ca- 02 2.34r(3) 2.346 (3 ) Figure 4 shows changesin cell parameterswith Na,Ca- 03 2.516(3) ?.527(3) temperature.Figure 4a showsthe expansionof the a Na,Ca- 04 2.484(3) 2.4s8(3) anomalous cell dimension with temperature'The Na,Ca- 05 2.808 (3 ) 2.80e(3) permanentchange lengtheningof a aI 1000'C is a Na,Ca- 05' 2.74o(3) ?.739(3) dimensionafter the heat- and is reflectedin the a cell Na,Ca- 06 2.87e(3) 2.870( 3) 4b showsthe variation in the c cell ing cycle.Figure Na,Ca- 06' 3.000(3) 2.e8e(3) asa functionof temperature.Cell constant dimension Na,Ca- Cl 3.064(1) 3.076(r) c is constant with increasing temperature until L LEVIEN AND J J PAPIKE

TlsI-B6. Betasfor positionalparameters

Before After Before After Heating Heatinq Atom Beta Heating Heating 'n Na,Ca 0.00328(7) 0.00321(8) 04 lt 22 0.00283(7 ) 0.00303(7 ) 22 33 0.0t05(2) o.ol2t(3) 33 't312 0.00167 (5) 0.00r64(6) 12 -0.0002(t) -0.0003(2) 13 z5 -0.ooo0(t) 0.0oor(r ) 23 'n 'tl cl 0.0066 (2 ) 0.00e5(2) 05 0.0024(2 0.0027(2 22 0.0066(2 ) 0.00e5(2) 2? 0.001 I (2 0.0020(2 33 0.0170 (e ) 0.or70(e) 33 0.0079(4 0 .0085 't312 0.0 0.0 12 0.0002(1 0 .000]t; 0.0 0.0 13 -o.oo28 (2 -0.00t9i; 23 0.0 0.0 23 -0.0009(2 -0.0014l7 'll TI 0.00077(4 ) 0.00087(5) 06 1l o.o0t5(2) 0.0020 22 0.0007 e(4 ) o.ooo9l(s) 22 0.00r8(2) 0.002 1 33 0.oo3s(1) 0.0040(l) 33 0.007r(4) 0.0084 12 -0 (3 -0.00005(4) -0.000r -0.0004 't3 .0000s) 't312 (r) -0.0001 2(8 ) -0.0002(t) -0.0007(2 ) -0 .0002 23 -0. 00012(8 ) 0.000r(r ) 23 0.00r3(2) 0.0024 'll T? 0.0009s(6) 0.00101 (7 ) 1'l o.oo8(r) 0.008(r ) 22 0.0one(6) 0.0olle(7) 22 0.008(l) 0.008(r) 33 0.0030(r) 0.0034(2) 33 o.or6(3) o.or6(3) ta o.0ooo8(5)-0.00002(5) 't312 0.0000(0) 0.0000(0) l3 0.00024(i) 0.00034(e) 0.0000(0) 0.0000(0) 23 0.000il(8) 0.00021(e) 23 -o.ooot(1 ) -0.000](t ) 'll T3 1'l 0.000et(s) 0.00r0](7) 07 0.008(r ) o.oo8('r) 22 0.00112(6) 0.0or3l(7) 22 o.oo8(l) o.008(t) 33 0.0027(l) 0.0030(2) 33 0.0t6(3) o.or6(3) 12 0.00001(4 ) -0.00005(6) 12 0.0000 ( ( 'l? -0.00013(7) 't3 0) 0.00000 ) -0. 00013(8 ) 0.0ooo(o) 0.0000(o) 23 0.00003(7) -0.oo0lB(e) ZJ -0.000r(1 ) -0.000](1) 't1 0l 1l 0.00]0(l ) 0.00t3('t) oe o.oo8(r 22 ) 0.0023(r) 0.0020(r) 22 0.008(1) JJ (4 0.0071 ) 0. 0083(4 ) 33 0.0t6 (3 ) l2 0.0002(t) 0.0004(t) 't312 o.oooo(o) IJ 0.0002(2) -0. 0008(3 ) 0.0000(0) 0.000](3 ) -0. 0002(3 ) 23 -0.000r(r) 'tl 'll 02 0.0027(1) 0.0030(2) 09 22 0.00]e( l ) 0.002t(l ) 22 JJ 0.0032(3) 0.0038(4 ) 33 12 0.0009(l) 0.0010( t ) 12 t3 0.000s(2 ) 0.0005(3 ) t3 23 -0.0003(2 ) -0.0003(3) 23 03 11 0.0025(2 ) 0.0027(2) 22 0.00r6 (2 ) 0.0018 (2) 33 0.006e(4) o.oo72(5) '1312 0.0007( t ) 0.0006(2 ) 0.0002(2) -0.000](3 ) ?3 -0.0007(2) -0.0007(2 ) ent with essentiallycomplete order, with Zl and T3 composition,because of the intensityof its primitive containingonly Sia+and T2 containingonly Als+ reflections,will havea totally orderedAl-Si distribu- (Levien et al., 1975).Lin andBurley (1975) predicted tion. Our sample'scomposition is just past the what we actuallyfind, namelythat a scapoliteof this criticalAllSi ratio of 72,this ratio beingthe point SCA POLITE C RYSTAL CH EM ISTRY 869

Tnsl-r 7. Unit cellparameters of scapoliteON-6at severaltemperatures

24'C Before 24"CAfter 400'C 600'c 700"c 800"c 900'c ]000'c '12.2102(4) a(A) 12.0686(4) l2.o8t7(3) 12.'t206(4) r2.r48o(3) r2.r606(4) r2.r74s(3) 12.r867(3) (3) 7.5750(4) c(A) 7.s8r2(3) 7.5758(3) 7.57e0(3 ) 7.57e8(3) 7.5792(4') 7.5798(3) 7.s7e7 il2e.3s(s) V(A") r'r04.21(4) il0s.82(4) iln3.42(4) iln 8.59(4 ) l r20.8r (s) 1r23.45(4) n 25.70(4)

+ O1

Shaded Flc. l. The structure of marialite scapoliteprojected down the c axis. 71 tetrahedraare striped; 72, white; and 73, dotted' z = h Z1 tetrahedraare at approximatelyz = t/e,unshaded at z : t/t while shadedchlorines (CL) and Na,Ca'sare at approximately (e.g, and Ol') and unshadedat z : 3/r.The primiiivecell and all symmetryelements are delineated.Symmetrically equivalent atoms Ol l'4' is shown are distinguishedonly for the purposeof identifyingdistinct bond distancesand angles.A projectionperpendicular to in Fig.3. 870 L. LEVIEN AND J, J. PAPIKE

o2 I

+Or

Ftc. 2. The structureof meionitescapolite projected down the c axis showingthe positionsof the disorderedcarbonate group.

after which an ordered scapolite must violate the the tetrahedralnetwork would contain fewer large aluminum avoidance rule. If Z2 is totally filled with Al3+ tetrahedra,creating a net shorteningof c. As Al3+, then any additional Als+ must enter Tl or 73. Al3+ substitutesinto Tl, a pathparallel to a would All Zl and ?T tetrahedra are linked to T2 tetrahedra containmore Al3+ tetrahedra, and c wouldundergo a (Fig. 3); therefore an Al3+-O-Al3+ bond would be net lengthening. made, and the aluminum avoidance rule would be In an independentstudy, N valueswere collected broken. on a crystalof sampleON-8 (Shaw,1960). Several The tetrahedral bond distancesindicate that, after crystalswere then heatedfor one weekat 990oCin a heating, Zl increases and T2 decreasesin size, while sealedplatinum capsule. A good-qualitycrystal was Z3 remainsthe same.These changes, although small, selectedfrom thoseheated, and 2fi valueswere col- may reflect a slight amount of disordering that has lectedon it as well. The changesin cell parameters taken place with Al3+ exchangingfor Sia+in Zl, and after heating are furthe samedirection and of the Sio+replacing Al'* in 72. The way this disordering samemagnitude as ON-6 with a increasingand c would affect the cell dimensions can be seenin Figure decreasing. 3. As Sia+is addedto T2, a path parallel to c through We do not believethat this changein cell dimen- SCAPOLITE CRYSTAL CH EM ISTRY 871

Frc. 3. The structureof marialitescapolite projected down theaz axis, perpendicular to sectionA A' in Fig' I . Tetrahedraare shaded to be consistentwith Fig. l. sions reflectssignificant loss of the volatile com- map puts the disordered CO3- group further from the ponentsfor two reasons.First, the crystalwas heated center of the cavity, which has been enlarged during in a sealed,evacuated silica-glass capillary with an heat treatment, If CO, were lost and O'- left in its inner diameterof 0.5 mm and a lengthof 14 mm. place,it would probably take the sameposition as the Very little volumeof gaswould be neededto crack chlorine anion. When the Clr- electron density is this capillary; yet it did not break during or after subtractedusing a Fourier differencemap, thereis no heating.Fourier maps also support carbonate reten- electron density left that could be construed as being tion. The carbonate-oxygenpeaks are the same 02-, again indicating carbonateretention. height on both the before-heatingand after-heating of carbonate maps.A lossof 10percent or more of the carbonate Refinement should be detectableon the Fourier maps. In the In past refinements of scapolite, carbonate has ei- before-heatingFourier map the four carbonpositions ther not been refined (Papike and Stephenson,1966) superimposewith the chlorineto form one largecen- or has been refined in a manner inconsistent with the tral peak (Fig. 5). The lack of changein this peak constraints of a carbonate group (Lin and Burley, implies no chlorine loss. The after-heatingFourier 1973a, 1973c, 1975). Using the computer program L LEVIEN AND J. J, PAPIKE

RRtNn4,(Finger and Prince, 1975) we were o (A) able to t22 refine the carbonate group as a rigid body, with realistic bond distancesand angles.Figure 5 shows the contours of the Fourier map overlain by the four a refined, disorderedcarbonate-group positions. Over t2 half the volatile-containing cavities in this sample contain chlorine. The others would have only one of the carbonatepositions filled at any one time (Fig. 6). Therefore, each unit cell containing carbonate must t2 actually be triclinic, but the averagesymmetry of the entire crystal is consistentwith the tetragonal space t? group P4r/n This refinement indicates that tilting of the CO3- group out of the (001) plane is minimal, probably lessthan three degrees,which is contrary to earlier published results (Lin and Burley, 1973a, 400 600 800 tooo 1973b, 1973c, 1975). T lo^\ We think that this is a good model for several reasons.First, the refinementconverged with the positions of the carbonateoxygens and the carbon fall- ing within the areas of highest electron density. In Figure 5 oxygens07 and 08 fall within one electron (A densitypeak, and 09 falls into a secondpeak of equal c intensity. The peak containing 09 is partially caused by a Fourier termination error, so it doesnot actually representequal electron density to the peak containing 07 and 08. An essentiallyflat differencemap resultsfrom this model. 400 600 800 tooo This model is also the only model which gave rea- Tec) listic calcium to carbonate-oxygenbond distances;all other models gave at least one distance that was much shorter than would be predictedby ionic radii. The high uncertaintieson the CO!- positional parameters,causing the large errors in thesebond distances(0.84), are due to the low atomic weight of the atoms involved (carbon and oxygen), to the large amount of thermal motion in this group, and to the small amount of CO3- in the crystal. Each of the disorderedCO!- positions contains lessthan l0 percent CO3-. The Na,Ca site has a large and fairly anisotropictemperature factor, probably causedby a slight splitting of the positions of the Nal+ and Ca2+ cations.The positional parametersfor this site are the averageof the position for Nal+ and Ca2+.The Nar+ cation is probably located closer to the 4, axis on which the Cll- position is fixed. Therefore,the Nal+ to Cll- distancelisted in Table 5 is anomalouslylong. By the sameargument the actualposition of the Ca2+ ?oo 400 600 800 tooo cation is located further from the 4raxis, making the T (oc) distancescomputed for Ca'+ to carbonate-oxygen FIc. 4. Variation in unit-cell parameters with increasing tem- bonds anomalouslyshort. Refinementattempts using perature. Arrows point. to anomalous cell dimensions discussed a different site for Na1+ than for Ca2+,instead of a in the text. singleNa,Ca site, were unsuccessful. SCAPO LITE CRYSTAL CH EM I STRY E73 O2I

Frc. 5. Fourier(Foo") map of Clr- and CO3- in sectionat z = V+overlain by the four refinedpositions for the CO!- group.

Ca2+has an unusualcoordination polyhedron in showsa as a function of both meionitecontent (dot- scapolite.In Figure 6 Ca' and Ca"' arecoordinated ted line) and temperature(solid line) to compare by two carbonateoxygens while Ca" and Ca"" are chemicaland thermaleffects. Sample ON-6 wasused coordinatedby only one.Ca2+ is alsocoordinated by for the thermal study, so it is plotted on the figure sevennoncarbonate oxygens (Table 5). Theseseven both at 33.5percent meionite (Papike, 1964) and at oxygensplus either one or two carbonateoxygens room temperature.The two linesare drawn with the give calciuma coordinationnumber of either8 or 9 sameslope, so a changein a reflectsparallel changes dependingon the position of the CO3- group in the in chemistryand temperature.An increasein meion- unit cell. This refinementof the carbonategroup ite content from zero to 67 percentcauses the same confirmsthe disorderedCO3- model of Papikeand amount of expansionof the a cell dimensionas rais- Stephenson(1966). ing the temperature from room temperature to 1000"c. Changescaused by temperatureand chemistry Although increasingthe meionitecomponent and Eugsteret al. (1962)and Papike(1964) describe a increasingthe temperaturehave similar effects on the linear variation of the cell parameterswith composi- c cell dimension,the changein a is causedby two tion.ln Figure4 we showan almostlinear relation- differentmechanisms. Papike and Stephenson(1966) ship of the cellparameters with temperature.Figure 7 reportedthat increasedmeionite conten( causes rota- 874 L. LEVIEN AND J, J, PAPIKE

, orlgln

Ftc. 6. Positionof singlecarbonate group in its Ca'+ environment.Ca'and Ca"' arc coordinatedby two CO3- oxygenswhile Ca" and Ca"" arecoordinated by only one.Errors on calciumto oxygenbond distancesare 0.8A

tion of the tetrahedralrings "in" (Fig.8) causingthe causeit is difficult to displaceor rotate tetrahedra short axisof the cationchannel (measured fiom 06 that link in that direction. to 06" in Fig. l) to decreasein length. Increased temperaturehas the oppositeeffect on thesechannels. Scapolitesubsolidus The short axisincreases in lengthwhile the long axis Scapolitewas originally refinedin the spacegroup between05 and 05" staysrelatively constant. The I4/m by Papike and Zoltai (1965)and Papike and channel width divided by. the channel length is Stephenson(1966) because h + k * / : odd reflec- plotted as a function of temperaturein Figure 9. tions are weak and diffuseeven in long precession Thermalexpansion causes the tetrahedrato rotate in photographsand absentin nearend-member compo- the oppositedirection from increasedmeionite con- sitions(Lin and Burley, 1973b,1973c). More recent tent (Fig. 8), resultingin a more open structure.The refinementsincluding thosein this paper have used changein channelwidth can accountfor most of the the primitive space group P4r/n (Lin and Burley, expansionof a with increasingtemperature. The c cell 1973a,1973c, 1975). To seeif the structurebecomes dimensionremains constant with both compositional body-centeredat a higher temperature,we plotted and thermal variation.As meionitecontent is in- normalizedstructure factors of primitive reflections creased,Al$+ cations enter tetqshedra that affectthe a which violate14/m as a function of temperature.To cell dimensionbut not the c cell dimension(Il ). The avoid seeingan artifactof ddta collectionat different dimensionc remainsconstant with temperaturebe- temperatures,all reflectionswere normalized by di- SCAPOLITE CRYSTAL CH EM ISTRY 875

T (oc)- 2P 600 ro@ 825, 2

t2 2l

t2 t7

I t? t3 o (A) 1209 ./ tooocc <------..--'{. ON 6 s67c/cMe t2 .l--u'

t2'ot --.----cHEMTCAL EXPANSIOI - THERMAL EXPANSIOA

il97

'1" Me - Ftc. 7. Variationin a with increasingtemperature and increasingmeionite composition. The thermal studywas performed on sampleON-6 so it is plottedboth at 33.5percent meionite composition and at room temperature.The sameamount ofexpansion is causedin a by increasingthe meionitccontent from 0 to 67 percentas by increasingthe temperaturefrom room temperatureto l000oC.Unit-cell data for scapolitesother than ON-6 are from Papike(1964). viding them by the averagestandard reflection at eachtemperature. If at any temperaturethese normalized reflectionswere all of zero intensity,the structure would be body-centeredand have14/m symmetry. Thesereflections do not go to zero intensityat 1000"Cbut do decreasesignificantly. Because these data werecollected on an intermediatecomposition, the primitive reflectionsare rather strong.Near end- membercompositions, which have weak or no primitive reflectionsand are consideredto approachthe body-centeredstructure at low temperatures,may display body-centeredsymmetry at temperatures lowerthan 1000'C. Conclusions (l) With increasedtemperature the volume in- creasesin a linear fashion,the a cell dimensionin- creasesin a nearly linear fashion, while the c cell dimensionremains constant. (2) SampleON-5 has essentiallytotally ordered Frc. E. Dotted arrows show rotation of tetrahedra"in" as meionitecontent increases, closing the channel. Solid arrows show Sia+and Al8+distribution with Sia+filling Tl and T3 rotationof tetrahedra"out" astemperature increases, opening the and Al3+ filling T2 as predictedby Lin and Burley channel.This rotationtakes place in the (001)plane. Both effects (197s). resultin the expansionof the a cell dimension. 876 L. LEVIEN AND J, J. PAPIKE

but do not disappearin the intermediatecomposition studied.Thus this sampleapproaches body-centered symmetryat higher temperatures,but has not yet reachedit at 1000"C.

Acknowledgments Dr. Larry W. Fingeris gratefullyacknowledged for his help in I usinghis computerprogram, RntNe4, and for his commentson the F (9 manuscript.K. J. Baldwin and C. T. Prewittare also acknowl- z U edgedfor their helpful discussionsand aid in collectingthe data. J This researchwas supportedby National ScienceFoundation J GrantNo. AO-41137. ztrl z () References

F Bnowu,G E.,S. SupNo rNo C. T. Pn'ewtrr(1973) A newsingle- o = crystalheater for the precessioncamera and four-circlediffrac- tometer.Am. Mineral.58, 698-704. U z BusEcK,PrrEn R. rNo Suuro Iurur (1974)High resolutionelec- z tron microscopyof silicates.Am. Mineral. 59, l-21. I DovI-e,P. A. lNo P. S. TURNER(1968) Relativistic Hartree-Fock x-ray and electron scattering factors. Acta Crystallogr. !t24, 390-397. Eucsrnn. Hlrs P.. Henolo J. Pnosrr,r eNo DANTSLE. Applr- ueN (1962)Unit-cell dimensions of naturaland syntheticscapo- lites.Science 137, 853-854. EvrNs,B. W., D. M. SuewrNo D. R. HeucsroN (1969)Scapo- lite stoichiometry.Contrib. Mineral. Petrol. 24, 293-305. Toc- FINcER,1. yf.^r+iiP E. PRINcE(1915) A systemof Fortran IV Ftc. 9. Variation in channel width (06-06") divided by channel Computer Programsfor Crystal Structure Computation.lJ.S. length (05-05") with increasing temperature. Oxygens 06,06",05, Nat. Bur.Stand. Tech. Note 854(Feb., 1975) 133 pages. and 05" are shown in Fig. l LEVTEN,L., J. J. Pnprrr,K. J. BelowrNrNo C. T. Pnewrrr(1975) High-temperaturecrystal chemistry of scapolite.(Abstr.) Irans. Am. Geophys.Union 56, 462. LtN,S. B. eNoB. J. BunI-sv(1973a) Crystal structure of a sodium (3) All threecell parametersundergo an irrevers- and chlorine-rich scapolite. A cta Crys tal log r. B.29,127 2-127 8. ible changeat 1000"C.Both a and volumeincrease -AND- (1973b)On the weakreflections violating body- while c decreases.This change,coupled with the di- centeredsymmetry in scapolites.Tschermaks Mineral. Petrogr. rection of the changesin the tetrahedralbond dis- Miu 20,28-44. -and- (1973c) tances,suggests amount and The crystalstructure of meionite.Acta a small of Al3+ Sia+ Cry stal I ogr. B.29, 2024-2026. disorderinghas taken place. -AND- (1975)The crystalstuctufe of an intermediate (4) The refinementindicates that the trigonal car- scapolite-wernerite. A cta Cry stal log r. B3l, I 806-1 I I 4. bonategroup can be disorderedin a mannerconsis- Lovrntxc, J. F. rNo A. J. R. Wune (1964)The significanceof tent with Ihe P4r/n spacegroup and that the group is primary scapolitein granulitic inclusionsfrom deep-seated pipes. very (001) "/. Petrol.5,195-218. closeto beingin the plane, tiltingless than NEwroN,RossRt C. ,cNoJuLreN R, Golosutru (1975)Stability threedegrees. of scapolitemeionite (3CaAlzSizOaCaCOs) at high pressuresand (5) Both increasedmeionite content and temper- storageof CO, in the deepcrust. Contrib. Mineral. Petrol.49, ature lengthenthe a cell dimension,but by different 49-62. mechanisms.Increased meionite substitutes larger PAprKE,J. J. (1964)The crystal sffucture and crystalchemistry of Ph.D. thesis,University of Minnesota. Al3+ for smallerSia+ causing lengthen scapolite. a to and at the -AND TtsoR ZoLreI (1965)The crystalstructure of a maria- sametime causingtetrahedra to rotate "in" decreas- lite scapolite.Am. Mineral.50, 641-655. ing the width of the Nal+ and Ca2+containing chan- -AND NevrLLeC. SrrpHsNsor.r(1966) The crystalstructure of nel. Increasedtemperature allows the tetrahedrato mizzonite,a calcium-and carbonate-richscapolite. Am. Min- rotate"out" wideningthe cation containing channels eral. 51, 1014-1027. (1930) and creatinga more Plur-tNc, L. The structureof somesodium and calcium openstructure. alumino-silicates.Proc. Wash.Acad. Sci. 16. 453-459. (6) At high temperature,the primitive reflections PHAKEv,P. P.lNo S.Guosr (1972)Scapolite: observation ofanti- which violale I4/m symmetry decreasein intensity ohasedomain structure. Nature Phvs. Sci. 238. 78-80. SCA POLITE C RYSTAL CH EM I STRY 871

PREwrrr,C. T. lNo A. W. Slprcnr (1968)Structure of Gd,S". Sulw, DnNts M. (1960)The geochemistryof scapolite.Part I. Inorg. Chem.7, 1090-1093. Previouswork and geneqalmineralogy. J. Petrol.1,218-261. ScHrcnoro,E. rNo G, Sruprpr-(1932) Ober die Kristallstruktur UrBntcu, Honsr H. (1973)Crystallographic data and refractive von Skapolith.Z. Kristallogr.Sl, 110-134. indicesof scapolites.Am. Mineral.58, 8l-92. ScHwencz,H. P. lNo E. L. SprBr-lleN(1965) Determination of sulfurand carboncoordination in scapoliteby infra-redabsorp- Manuscript receioed,Augusl I l, 1975; accepted tion spectrophotometry.Am. Mineral. 50, 656-666. for publication,April 27, 1976.