1039 The Canadian Mineralo gist V o l . 3 4 , p p . 1 0 3 9 - 1 0 5 0( 1 9 9 6 7
MARIALITE: RIETVELD STRUCTURE-REFINEMENT AND ASi MAS AND 27AISATELLITE TRANSITIONNMR SPEGTROSCOPV ELENA V. SOKOLOVAT eru YtlRtr K. KABALOV Deparxnzntof Crystallography,Faculty of Geolog, MoscowStateUniversity,Moscow 119899,Russia BARBARA L. SHERRIFF ANDDAVID K. TEERTSTRA Departurcntof GeolagicalSciences,Universityof Manitoba Winnipeg,Manitoba R3T 2N2 DAVID M. JENKINS Departmmtof GeologicalScimcesandEnvirownentalSndies,BinghamtonUniversity,BinghamtorlNew York13902-6000,U.SA. GERALD KTINATH-FANDREI. STEFFEN GOETZ ANDCHRISTIAN JAGER Institutfir Optik and QwmtenelelaroniltFriedrich-Schiller-Universitttt,Mat-Wien Platz I, D-07743Jena"Germany AssrRAcr and three samplesof NaCl-rich scapolite The crystal structureof syntheticend-membermariatite (Ma) NaaAl3SieO2aCl from Pamir (central Asia) were refined using Rietveld methods. Compositional measulementsindicate meionite (Me) CaoAl6Si6OroCO3 contentsof 0 (SYN MAR), 4.6 (PAM-l), 7.5 (PAM-2), axld7.6Vo(PAM-3). The crystal structureswere refined in spacegroup IAm using ionized X-ray scatteringfactors: Rp 4.89 - 5.92Vo,Rwp 6.78 - 7.28Vo,ls 2.53 - 3.4OVq, RF2.49- 3.34Vo . The syntheticend-membermarialitehasunit-cell parametena = 12.0396(2) A, c = 7.54n Q) A , s | .26--2.O9 and V = 1093.3(4)Al. A linear conelation was found betweenthe a and c unit-cell dimensionsand the Si contentof these samplesof marialitic scapolite.Additional electron-densitymaxima were found on the differenceFourier mapsD(xyz), and correlationwith anincreasewith H2Ocontentsuggestspartial occupancyby H2Oalongthe channelsofthe marialiteframework. 27AlsatellitetransitionNMR spectrashowthat Al is in only oneenvironmentin thenaturalsamples,and2esiMAS NMR spectra show that Si alone occupiesthe Z1 site. Calculation of the numbersof A1-0-Si bonds from peak fitting to the 2esi NMR spectraindicatethat up to 807oof the AI atomsin the T2 site areinvolved in oneA1-O-AI bond. Keyword,s:scapolite,marialite,volatiles, synthesis,Rietveld refinement XRD, satellitetransitionNMR, MAS NMR. SoMlaans Nous avons affin6 la structure cristalline de la marialite synth6tique(compositionid6a1e:NaaAl3SbOr4Cl)et de trois 6chantjllonsde scapoliteprochesde ce p61e,provenantdes montagnesPamir, en Asie centrale,par m6thodesde Rieweld. est 0 (SYN MAR), 4.6 (PAM-I)' l,a compositionde ces quatre6chantjllons,en termesde leur teneuren CaaAl6Si6O2aCO3, 7.5 (PAM-2), andT.6Vo@AM-3). Leur structurea 6t6affin6edals le groupespatial/4/m en utilisant desfacteursde dispersion desrayonsX appropri6saux espdcesionisdes:Rp 4.89 - 5.92Vo,Rwp 6.78 -7.287o, RB 2.53 -3.407o,Rp2.49 -3.34V9, s 1.26 - 2.O2, IE p6le marialite synthdtiquepossddeles paramdtresr6ticulaires suivants: a 12.0396(2),c 7.54nQ) A, V 1093.3(4)A3. Une relation lindaire existeentreles dimensionsa et c etle contenude Si. Des maximaen densit6d'6lectrons ont 616document€ssur descartesde diff6renceFowier D(xyz); ceux-cimontrentune corr6lationavecla teneuren H2O,ce qui sembleindiquer une occupationpartielle des canauxdans 1atrame par des mol6culesde H2O. ks spectresde r6sonance magndtiquenucldaire(RMIrf dessatellitesassoci6si la transitiondesatomes7Al montrentquel'aluminium setrouve dansune seuleposition dansles &hantillons naturels.ks spectresRMN obtenuspar spin du 2eSil anglemagiquemontrentque seulle Si occupela position (1). Un calcul de la proportionde liaisonsAl-O-Si par interprdtationdespics d la lumibre desspectres RMN de 2esi indique qu'un maximum de 807odes atomesAl occupantla position Z2 seraientimpliquds dansune liaison Al-o-At. (Traduit par la R6daction) Mots-clds: scapolite,marialite, phasevolatile, synthdse,affinement par m6thodede Rietveld, diffraction X, transition des satellites,r6sonancemagn6tiquenucl6aire,spin i anglemagique.
IE-mnil address.'
[email protected]
1040 Ivrnoluct:loN
the A site of greater than 1.0 atom per formula unit (apfu), tf H2O is included with Cl-, CO32and SOIScapolite-groupminerals have a general formula (Teertstra& Sheniff 1996b).The naturalsamplesused M4T,O24A, and constitute a solid-solution series in this study were analyzedfor H2O, and intensity betweenthe idealizedend-members NaoAlrSinOroCl peaksfound on difference-Fouriermapsfrom powder (marialite, Ma) and CaaAl6Si6O2oCO, (meionite,Me). XRD data are used to investigatethe position of the Scapolites have ttree main forms of isomorphous volatilespecies. substitution:Sia+for Al3+in the Z site. Na+for Ca2*tn In this study, Rietveld structural refinementsof a the M site, Cl- for CO]- or SO?-in the A site. There syntheticsampleof end-membermarialite and of three canalsobe minor or traceamountsof K, Sr,Ba andFe, samplesof marialitewith meionitecontentsof lessthan but only trace quantitiesof Mg, Mn,Ti, P, Br and F 8Voenableus to examinetrendsin the cell parameters have been measured.Two changesin compositional for the marialitic portion of the scapolitesolid-solution and cell-parametertrends, at cation contents of series. Difference-Fourier maps are exarnined for (Me1r) and Na,.4Car.6Al4.7Si7.3 Nar..Cao.uAlr.6Sis.o additionaldensitythat would shedlight on the position (Me6) divide the seriesinto tlnee portions(Teertstra& of volatiles in scapolites.We studied the degree of Sherriff 7996a,Zo7otarcv 1993).The Me < 15 portion Si-Al order in the tetrahedralsites with MAS and is the focusof this study. satellitetransitionNMR spectroscopy. The role of volatile species,includingHrO, within the scapolitestructureis not understood.Therearefew Rsvrw oF TIIESTRUCTITRAL ANarvsss recent measurementsof volatile contents"becauseH and C can not be measuredusing the electronViewed along the c axis, the tetrahedralsites in microprobetechnique.Also, it is exhemely challeng- scapoliteform two types of 4-memberedrings. One ing to find the position of these light atoms by ring consistsof Zl tetrahedrathat have their apices refinements of the structure using X-ray-diffraction pointing in the samedirection along the c axis. In the QRD) data.Although XRD datado not indicatewhere other ring, the apicesof the tetrahedrapoint alternatfie H atoms are situated,infrared (IR) spectroscopic tively in opposite directions along the c axis. In the dataindicatethe presenceof abundantbicarbonateand spacegroupIAm, with a 4-fold rotationaxis anda center bisulfate(Swayze& Clarke1990),andRamanspectra of inversion, the latter tetrahedraare symmetrically suggestthe presenceof HCI @onnay et al. 1978). equivalent,so arelabeledT2 (Fig.1), but in the space Formula calculationshave indicatedan anion sum for grolp P42ln,thesebecomeT2 arrd13. Viewed alons
cl , c%,so4 N o ,C q ,K
o cl,co3,so4 a No,Co,K
Ftc. 1. Crystal structme of scapolite viewed (a) along the c axis and (b) along the a axis. The T2 autdT3 sites are equivalent in space goup l4lm.
STRUCTTJRAL ASPECTS OF MARIALITE
the a axis, these rings join to form 5-membered rings and large cavities, which each enclose one A anion surroundedby four alkefu(W cations.In the present study, the composition of the samples of marialite will be quoted by their meionite content and alsoby the Si fVoMe= 100X divalentcations/41, contents(apfu). Teertstra& Sheniff (L996a)showedthat variations in the a and Vcell parameterscorrelatewith Si:Al ratio ratherthan with substitutionsin the M or A site. There is little overall variation of the c cell edge with compositionacrossthe series.Cell volumes(V = azc) seemto be constantoverthe rangeG-lSVoMe,because with decreasingSi content,a decreasein a is matched by a slight increasein c. There has been one complete refinement of the structureof a samplewith a meionite contentof less than l57o or Si greaterthan 8.4 apfu (Si: 8.7I apfu; Belokonevaet al.1993); partiat datawere reportedby Comodi et al. (1990)for a samplewith Si = 8.47apfu. Single-crystalX-ray refinementsof the structureshow that intermediatemembersof the series obey space group P42ln, and that the end-membersobey I4lrn (Belokonevaet al. L991,1993,Comodi et al. 1990, Papike& Zoltai L965,Lin & Burley I973a,b, 1975, Levien & Papike 1976,Papike& Stephenson1966, Aitken et al. 1984,Ulbrich 1973a,b).In this study, structuralrefinementsderivedfrom powderXRD data, by the absenceof the weak reflectionsviolating bodycenteredsymmetry,confirm that the spacegroupl4lrn is correct for the long-rangesymmetry of marialite. Tbe T2 and 13 sitesin the spacegroupP4y'n become symmetrically equivalent n l4/rn, leading previous investigatorsto suggestthat, in contrastto mid-series scapolites,both end membersof the scapolite series have a high degreeof Al-Si disorder.The changesof symmetry occur near Me15 and Meur (Teerfstra & Sherriff 1996a). It should be emphasizedthat in compositionswith Me < 40Vo,Al-Si disorder is restricted to the T2 site, as the Z1 site is occupied predominantlyby Si. )(RD observationsare consistent with a long-rangeaveragestructuralmodel.On a unitcell scale, transmissionelectron microscopy (TEM) observationssuggest lower symmetry and Cl- and CO]- order(Hassan& Buseck1988). Magic angle spinning nuclear magneticresonance (MAS NMR) spectroscopy hasbeenusedto investigate the degreeof order of Si andAl in the tetrahedralsites, as it provides a short-rangeview of the structure (Sheriff et al. 1987),but in that investigation,owing to the lack of Na- and Cl-rich samples,models of Al-Si order had to be extrapolatedto the marialite endmember.Also, there was a problem with quadrupolar interactionscausingbroad unresolvablepeaks in the 27Al spectra.In this sody, this problem is solved by using a combinedanalysisof the centraltransitionand spiming sidebandsof the t l/2 = X 3/2 satellrte transitions, including their envelopes,to investigate
t04l
the Al sites. Least-squaresfitting of the peaksin the 2eSispectragives the proportions of Si in different environmentsand allows the degreeof Si andAl order amongthe tetrahedralsitesto be studied. ExpsRn/ENTAL PnocEounrs Materials The naturalsamplesof scapolitearefrom the Kukurt hydrothermalscapolitedepositof fhe Muzcol'sk alpine metamorphiccomplex in the EasternPamir mountain belt in Russia. This Paleozoic gneiss complex has experiencedamphibolite-grademetamorphismat the core of an anticline,and arnphibolite-and greenschistgrade metamorphismon the flanks. Scapotte is widespreadthroughoutthe complex, associatedwith zones of Na-metasomatism.The Kukurt deposit consists of hydrothermal veins of scapolite with cavities containing transparentcrystals of scapolite associatedwith rutile" iftnenite. titanite and albite. Scapolite also is found in secondarycavities within gfanitic pegmatitesalong the Turakuloma mountain ridge (Zolotarev 1993, and referencestherein). The scapolite crystals in the Kukutt deposit are usually violet, although colorless and yellow crystals also occur. These crystals are usually prismatic, with the dominantforms being {010} and {110}, but {120}, {llll, {2211and {001} may alsobe present. Three transparent violet-colored crystals of inclusion-free,gem-quality, marialitic scapolitewere selectedfor study (PAM-l, PAM-2, PAM-3). The euhedralprisms,between0.5 and3 cm in length,show a variation in the intensity of the violet color, PAM-2 being the darkest, and PAM-3, the lightest. The crystalswere analyzedand checkedfor homogeneity and purity by powder XRD, electron-microprobe analysisand optical microscopy. Synthesisof marialite Synthetic marialite was produced specifically for NMR spectroscopicstudies in a piston-cylinder apparatusfrom a mixture of reagentgrade Na2CO3, NaCl, Al2O3,SiO2,and Fe2O3.A preliminarylrllvp investigation of the marialite synthesized,with the composition NaaAl3SieOraCl(without Fe), indicated that extremelylong Z1relaxationtimes wereneededto acquirethe 2esispectrum.Accordingly, a mixture was preparedcontaining0.1 wt.VoFe2O3in order to induce a small amountof paramagneticFe into the marialite structureto attempt to reducethe Z1 relaxation time (e.g.,Sheniff & Hartrnan1985).If we assumethat all of the Fe substitutesfor Al asFe3+andthat it remained as Fe3+,the composition of the resultant marialite unfortunately, would be Naa.ssAlr.eeFee.elSie.q0O2acl; this Fe content was not confirmed by EMP analysis (seebelow).
1042
THE CANADIAN MINERALOGIST
Saqle Codo
StsdlS odslel
T Cc)
MAR l-l! MAR l-7 MAR l-12 MAR 1-13 MAR2-I
dde-N8(]@fi Ori&-NaClmir MAR l-ll MAR l-7 feneaiogorile - Nfl nii Fe-bosiuSolide - NaClBii MAR2-I&Z-Z MAR 2-1 &2-2
l0D(10) 1024(10) tm3(r0) 103300) 1Bq5)
r72rt 16.(4) t73(3) 17.q3) 1&6(4)
180(5)
r8.q4)
1032(5) 1014(18)
18.1(4) 18.3{3)
MAR2-2 MAR2-3 MAR 2-4
P(lb6)
t(h)
45 v) 123 q t94 16r
M4 Qt4 Ab Ma.aE, Ab(?) M4 taal Me. IQtzl M4 halie, Qe
Rutherford mine, Ameta Courthouse, Virginia, anorthite from Sitkinak Island, Alask4 and tugtupite from the type localify in soutlem Greenland(R.O.M. #M32790).Eachsamplewas checkedfor homogeneity by analysis. Compositions were determined on the same samples from which the X-ray data were measured.The elementsMn, Mgo Ti, P, Br and F
M!,hane,QE were sought, but not detected. Mr. halnp Mq ia&e, tQtzl
I Al*rniatiw Abr alt'itc, Ma @irli!e, Qtd qurrta BtE l@b Mloab fis ee Fhe itr tae ?,fuft. U@taiEi€€ h las digit @ g!@ tn!@ds. FM
i8
Absolute quantities of HrO were determined at 900'C by Karl Fischer titration using a Mitsubishi moisture meter for the three natural samples of marialite. Prior to the determination,the sampleswere dried at 110'C to remove any water adsorbedon surfaces. The stoichiometric formulae for the four samples were calculated using the method of Teertstra & Sherriff (1996u b). This calculation consistsof four basic steps: (1) The formula of scapoliteis initially calculatedby normalizing to Si + N = 12 apfu. Monovalentand divalent cations(including total Fe as Fe) are assignedto the M site. If XM is greaterthatr 4 apfu, this may indicate-thatsome of the Fez*tt M should be assignedas Fer+in Z. The forrnula can be renormalizedto Si + Al + Fe3+= 12,generatinga lower AuI. (2) An excesspositive charge(EPQ is calculated by subtractingthe negative charge generatedby the framework (7O;) from the positive chargeof the M cations(M+) as follows: EPC = M+- IQ = A-, w[s1s 'IOA= + N Fe3+,andM+= Na + K+2(Ca+ Mg + Sr + Ba + Mn + Fe2). (3) Cl and S contentsare usually determinedby EMP analysis.By assuminga divalentS species,the remaininganion-chargemay be calculated by subtracting Cl, F, and 25 from the EPC. Tllre residual charge may be assigned to a divalently chargedcarbonspecies,grving a calculatedCO|- pfu. (4) If H2O is determinedand there is a remaining excesspositive chargeothis may be assignedas OH- to balancecharges.ff there is a net negativecharge,this may be assigned as H+ (i.e., bicarbonate).Any remainingH, when the chargesbalance,is assiguedas molecularH2O.
The starting mixture was roasted in air at about 1100"C for 30 secondsto drive off CO. from the Na2CO3andto partially fuse the mixture.An additional 15 wt.Vo NaCl was mixed into the decarbonated starting material in order to ensurethat the marialite is saturatedin NaCl. Portions of this mixture were sealed(dry) in Pt capsulesandtreatedat l0l4-1030'C and 18.1-18.8kbar for 99-123 hours in a Vz-inch piston-cylinder apparatususing solid NaCl pressure media (Johannes1978).No specialattemptwas made to control the fugacity of oxygenin the piston-cylinder pressureassemblage. High but incomplete yields of light blue-grey marialite were obtainedfrom this frst treatment.there being minor quartz and albite presentalong with the excess NaCl. To maximize the yield of marialite, the material obtained from the first ffeatment was thoroughly ground and treated a secondtime at the samepressure-temperature conditionsfor an additional 161-194hours.This secondtreatmenteliminatedall of the albite, and left only a trace (< l%o)of quartz, as judged from theXRD powderpattern.The excessNaCl was rinsed from the sample with distilled water. Examination of the marialite under the petrographic microscoperevealedblocky, equantgrains 25-30 p"m X-ray p owder diffraction on a side,with low first-orderbirefringenceanda mean index of refraction of 1.536. The conditions for the X-ray powderpat0ernswere collectedon an ADP-2 synthesisof individual runs and also the run products diffractometerusing CuKa radiation (Ni-filter), a step aregiven in Table 1. width of M0 of 0.02", a 2Q rangeof 10 to 150' and count times of 5 s per step. Chernicalanalyses Rietveld structurerefinement The scapolitecrystalswere mountedin epoxyresin, polishedand then analyzedusing a CAMECA SX-50 Structure refinements were carried out with the electron microprobe operating at 15 kV and 20 nA, Wyriet, version 3.3 program written by Schneider with a beamdiameter10 pm and count times of 20 s. (1989). The model of Belokonevaet al. (1993) for The datareductionusedthe PAP procedure@ouchou Mer1,with a spacegroupof I4/m. was usedas a base & Pichoir 1985). We used, as principal reference for the refinement of the three samplesfrom Pamir. standards,gem-qualitymeionitefrom Brazil, U.S.N.M. Later, for the sample of synthetic marialite #R6600-l (Dunn e/ al. 1,978),albite from the (SYN-MAR), the model for the Mea.6(PAM-I) was
rM3
STRUCTTJRALASPECTSOF MARIALITE
used. Two mixing parameterswere refined for the pseudo-Voigtprofi.lefunctions, selectedusing six firll peak-widthsat half maximum height @WHM), and the background was graphically modeled. Refined non-structural parameters included 20 zero point, sampledisplacement,profile parametenu, v, and w, peak-asymmetrycorrections for 20 < 40o, and preferred orientation. Scattering facton for ionized specieswere selectedfor the refinement of atomic coordinates,isotropic and anisotropic displacement factors.The occupanciesof Si wererefinedat 7l = 8(fr) and T2 = 16(r, as were thoseof Na, and Ca at the 8(ft) site. Tablesof structurefactorsare availablefrom the Depositoryof UnpublishedData, CISTI, National ResearchCouncil, Ottawa,CanadaKlA 0S2. For comparative puqposes,the structure of the synthetic marialite was refined in the space group P4t/n as well as I4/m. T\e reflections with h + k + I * 2n, violating the body-centeredlattice, would be expected if the structure obeys the space group P4"ln. After refinement, difference-Fouriermaps D(xyz1 for the three natural samples indicated additional intensity, which might representadditional positions for volatile speciesin the structure. Nuclear magneticresonancespectroscory
obtain these spectrawith different relaxation delays between pulses. The second set of sampleswas preparedwith the addition of Fe to try to increase the relaxation rate. Even with the doped samplethe relaxation rate was slow, and the best signal-to-noise ratio was obtainedwith 148transientsand a 15-minute delay betweenpulses. RESULTS AND DISCUSSION
Chemicalcompo$itionsandstoichiometricformulae of the three natural and one synthetic samples of marialite are given in Table 2. T\e three samples PAM-I, PAMJ and PAM-3 have compositions Mea.6,Me7.5and M9.6, and the Si/Al values are2.85, 2.72, and 2.71, respectively.It shouldbe noted that PAM-I and PAM-2 contain more K than PAM-3. Therefore, the low-Me PAM-I is mainly due to the high valuesfor K and the low levels of Ca rather than the high Na contentClable 2). SamplesPAM-I and PAM-2 have similar H2O contents,0.05 and 0.06 wt.Vo.Although the stoichiometric calculations indicatethat this is in the form of HrO for PAM-I and OH- for PAM-2 to achievea balanceof charges,&ere may also be minor COi. PAM-3 has a much higher content with 0.16 'xt.Vo H2O distributed by chargebalance calculations between 0.08 OH pfu and 0.M H2Opto, or 0.08H2Oand0.04CO3-.
2esi and 27Al MAS NMR spectrafor the samples PAM-I. PAM-2 and PAM-3 were obtained on a Bruker AMX400 instrumentat frequenciesof 79.5 and 104.2MHz, respectively.Rotation rates of 8-I4 Wz were obtainedusing a Bruker high-speedprobe. 2eSi spectra were recorded for PAM-I, PAM-2 and PAM-3 with recycledelaysof 5 s, asthis gavethe best signal-to-noiseratio for a given time-period. Spectra (m%) recordedfor PAM-I with recycledelaysof 1, 5, 30 and StO, Nro3 180s showeddifferencesin relative peak-intensitiesof F%o. less than !2Vo using these differing delays. Relative Na.O intensitiesof the 2eSipeakswere found by simulating r;o the spectra on a computer using a least-squares sto iterative process,which varied the isotropic chemical BaO cl shift, the Gaussianand Lorentzian broadeningpara- so3 meters, and the intensity of each line. Short pulses HrO correspondingto tip anglesof lessthann/15 wereused S@ o4l,F for 27Al in conjunctionwith a 1 s recycle delay. The Total baseline roll causedby the finite pulse-length and St (rptu) dead-time was corrected using the cubic spline fit N in the spectrometersoftware (Kunath et al. 1992). Fc3' Na The spectra were referenced against the 27AI(VI) resonanceof Y3A15O,at +0.7ppm. Simulationsof the satellitetransitionswere performedon a 486 processor cl (Kunathet al. L992)using the theory of Skibstedet al. sOH (1991). HrO zeSispectraof the two samplesof syntheticmarialite svAr M9 were obtainedat the Prairie RegionalNMR Centreon an AMX 500 with a Doty high-speedprobe at a TtE Frp6d@ frequencyof 99.35MHz. Many attemptsweremadeto deb49d
of lto,
64.16 18.01 0.m
62l.3 $62 0.f7
6r:71 1924 0.6
6Lv 1931 ol}?
14.60 o.02 0.05 0.00 0.m 4.lr 0.m d. r0.95 4g2 100.(p
t2,g t37 1.15 0.11 0.m 1,16 o.lz 0.6 101.00 -.094 1m.06
t2.K 1.48 192 0.04 0.02 3,94 0.15 0.06 1m.90 -o.t9 10.01
tLw 0.94 LM 0.14 0.(B 3.!lt 0.16 0.16 10)96 -{g) lq).06
9.4, L98 0.m 3.98 0.(! 0.01 0.m 0.98 0.m rd. rd. 3.@ 0.z)
E.EE 3.12 0.m 3.49 oa 0.18 0.01 1.0r 0.01 0.m 0.@ 285 4S
8.78 32 0.0 3.4t oz, 0a 0,01 095 0.o2 0.06 0.m LN 75
436 324 0.01 353 0.17 0,9 0.m 0.96 0.u 0.6 o04 L1I 75
6 d$lb€dh&em
uLd
oI! Fe'4 a!iFe3'
v@ €r€hr€d
1044
TT{E CANADLA.NMINERALOGIST TA3LB 3. UNII-o\2) u.t\rt r(2)_06) r(2)q3) 0.454{r) TQ>{x.4\
o.@4(2t 0.88?.(8) 0 1.4(3)
Clroj0J0J roj0J0J zO30J0J Br 33(2)
rJ84{t 1.608(t L6Al4) t.612
rOF(l) roFc(l)' r(lF q4) x 2 aw
av@99
2 2
rrrxQ) rvts(2) l@,4
ar/s(t
2910(t LWA u893(6) 2m
1045
STRUCTIJRALASPECTSOF MARIALITE Y
Y
]D
f
1.0t
1.0 t-
a
*3:i!
b
0.08 A v o.2o
0.6
0
o.42Q
a0.10
I
o 6) o o.ro Jo.+z
O0.36
0
@
0 0.0t0.0
0.0 0.0
0.4
Y
t
c
@0.20 00.30
L4)
00.20
@0.:o
n
o a+, LY./
/Nl
o
o
0.0F 0.0
Frc. 2. Differenceelectron-densitymapsD(002) (0 < z < 0.5) for samplesof marialite PAM-I (a), PAM-2 O), PAM-3 (c). The map valuesrangefrom 2.0 ta 2.587(PAM-I), I.7 to 2.553(PAM-2), 1.7 to 2.474(PAM-3). Contours aredrawnat valuesof electrondensityof 2.00 and2.30for PAM-I, 1.70,2.00, and 2.30 for PAM-2 and PAM-3, with intervals of 0.3. The numberson the mapsindicate the e value for eachmaximum.
range from 2.0 to 2.587 (PAM-l), 1.7 to 2.553 @AM-2), 1.7 ro 2.474 @AM-3). Figures2a, b and c summarizethe positions of electon-density maxima for PAM-I, PAM-2 and PAM-3, respectively. To removebackgroundnoise and show the projectionsof the intensemarima only, electron-densitycontoursare drawn at 2.00 and 2.30 for PAM-I, L.70,2.00, and 2.30for PAM-2 andPAM-3. with intervalsof 0.3. The numbers on the maps indicate the e value for each maximum.Comparisonof thesethreefigures indicates that there are additional maxima (002) in the large
channelsof fhe sfructureaboveand below the A site, which is most pronouncedin the H2O-rich sample PAM-3. Othermaxima which do not haveany specific spatialpositionbut alsoaresituatedin the intersticesof the structure,occur for all thrce natural samples.All of the maxima are interpretedas being due to partial occupancyby volatile species,possibly neutrally chargedHrO molecules.Swayze& Clark (1990)found several different OH-stretching and OH-bending vibrationsin their IR spectraof carbonate-andsulfaterich scapolites. From spectra taken from oriented
@ 19
o
o
6\, v
TM6
THE CANADIAN MINERAI,OGIST
ppm
-85
-90
-95
-100
-105
Frc. 3. 2eSi MAS NIVIR specrra of (a) SyN MA& (d) PAM-3.
sections,they determinedthat the speciescausingthe peakswerein specificcrystallographicorientationsand not randomly distributed adsorbed water. They concludedthat in their samples,OH was related to tlte carbonateor sulfategroup. However,they did not haveany samplesasCl-rich asthe marialite specimens from Pamir or discussthe possibility of neutral water moleculesin thechannelaboveor belowtheA site.The
-110
O) PAM-I,
-11F" (c) PAM-2,
and
additional maxima in our difference-Fouriermaps are the first direct information on additional positionsfor volatile speciesin the scapolitestructure. The 2esi MAS NMR specrraof PAM-I, PAM-2 and PAM-3 each consist of six peaks, at -92.3, -95.8, -98.7, -102.0, -105.8, -110.7 t 0.1 ppm (Figs. 3 b, c, d). The relativeintensitiesof the six peaks vary for each spectrum.The relative amountsof the
t047
STRUCTURALASPECTSOF MARIALITE
l].
or, *n NMRspecfium ;; .ix peaksfittedbytheleast-squares or rot-i-rl* between andthedifference process; spectrum, fittedpeaksandenvelope, theexperimental fittedandexperimental slrectra areshown.
atoms, in the specific environmentsrelating to each sile, were calculatedby fitting the spectrato the sums of peaks,which had a combinationof Gaussianand lorentzian line shapes(Frg. 4). The areasunder the fit0edpeakswere convertedto relative intensitiesand related to the number ofSi apfu contributing to each peak (Table 6). The elror on the measurementof relativeintensitywasfound to bet2Vo fromthe specha of PAM-I obtained at differing values of relaxation delays.The 2esiMAS NMR spectrumof the synthetic marialitehada very poor signal-to-noiseratio (Fig. 3a); however,it can be seenthat peakssimilar to those of the natural samplesare present.The signal-to-noise ratio ofthis spectrumwas consideredto be too low to allow a msnningful integration of the peaksor for it to be worth undertakingthe laboriousfitting procedure on the 27Al center band and sidebandsof satellite transitions. Previouswork has shownthat wherethe Si:Al ratio becomes2:1, there are only two peaks in the 2eSi spectrum,at -92.3 and -105.8 ppm, due to the two Si environmentsof I3(3Al lSi) and n(lAf 3Si) (Sheniff et al. 1987). Allowing for a shift of about 4 ppm to high field for each substitution of a neighboringAl for Si, the six peaksin the 2esispectra of the marialitesamplesar-92.3, -95.8, -98.7,-102.0, -105.8, -110.7 ppm were allocatedto ?2(1si3A1), T2(2Si2Ar), r2(3Si1A1), T2(4Si), 71(3Si1Al) and 71(4Si), respectively.
from 3.12to 3.24apfuftomPAM-l As Al increases to PAM-3, there is an increasein the intensity of the ?2(1Si3Al) and 71(3SilAl) peak, at -92.3 and -105.8 ppm, relativeto the peaksdue to f2(2si2Ai), Z2(3Si1Al), T2(4Si), and Zl(4Si) sites,as expected (Table 6). Thereare4.00 Si apfu n the 11 sitesfor all threesamples,within experimentalerror,conoborating that the singlepeakin each27Alspectrumis due to Al n theTZ site, andthat therels nq significantamountOf Al in the 71 site.
TABLE 6. @--ITYE PeL poafdoa aod all@dono
-92.3w -95.8ppo -rE7 plu -ro2-op!@
I'TTF.ISTTBS FOR,gsi MAS NMR SPECTRA
hak 8lloc€i@
PAM-I
17%(rs) 14%(r2) 13%(1,1) rrEo0n)
I2{ISi3AD T2(25,12lJ,) I2(3Si,1Al) r2({sD
16%(r,4) 14%(1.3) Be $2) 12%(r,o)
r5%(r3) u% 0a r47o(12) 12%{r.r)
55&(4,9)
55%(4.8)
55%GA)
I(3Sl,1Al) r (4Si)
MQA 15%(r.4)
3M(L7) 15%(\s)
WoW ts%(rs)
TdlfoIl
$%@,4
45%(4.0)
45%(4,0)
TffiI
1m%(8.9)
l@, (8.8)
l@% (&E)
Total fq f 2
-16€pln -u0.7 !!o
numbd ol Si mm p€r lomula mit itr p@!ths, at rcf6 b emhedol siB @neiniag eitlq si 6 Al.
1048
TIIE CANADIAN MINERALOGIST
-100ppml
-200
FIc. 5. 27Alsatellitetransitionspectraof PAM-I showingthe experimental(solid line) and fitted (dashedline) central transition (CT) and satellite transition (ST) line-shape, andthe experimentaland simulatedsidebandenvelope.
The 27Alspectrafor the three natural sampleshave a similar single center band and satellite transition sideband peaks. The isotropic chemical shift of 58.7 ppm, quadrupolarcoupling constantCo of 1.98MHz and asymmetryparametern of 0.8 to 1.0 werecalculatedby fitting the shapeofthe centerband and sidebandandthe sidebandenvelope(Fig. 5) using 350 Hz Gaussianline broadening.A1 appearsto be in a single 72 environment in the three samples,in agreementwith the I-O bond distancescalculated from the Rietveld refinementsand with the 2esi MAS NMR spectra. However, there is a distribution in the quadrupolar coupling constant of 85 kHz, indicating someslight variationin environmentamong Al sites.
It would be expectedfrom the Al avoidancerule (Lowenstein1954)that whereA1 is sufftciently dilute in the system, there will be no AI-O-A1 bonds. Therefore,only (4Si) sites should be presentin these compositions of scapolite. However, if the total number of Al atoms is calculatedfrom the relative intensities of the peaks fitted to each 2eSispectrum using the formula: Al=1a(ml4)Iq.^ Bngelhardt & Michel (1987) and referencesthereinl, the results for PAM-I, PAM-2 and PAM-3 are 2.6, 2.6 and2.7,respectively,insteadof the expectedvalues of 3.1, 3.2, and3.2. This translatesinto 3.4. 3.2 and
tu9
STRUCTURALASPECTSOF MARIALITE
3.3 Al-O-Si bondsper Al atom insteadof the 4 that would be expected.Either there is less Al in the unit cell than has been calculated from the electronmicroprobeanalyses,or up to 807oof the Al atomsare involved in one AI-O-A1 bond. in violation of Lowenstein'srule. Tossell (i993) calculated that the difference in energybetweenpairedand alternatingSi andAl atoms in a four-memberedring of Si2Al2Ot2Hs2is only 63 kJ/mol ratherthan the previously calculatedvalues of >400 kJ/mol (Hasset al. 1981.,Sauer& Engelhardt 1982,Nawotsky et al. 1985,Derouaneet al. 1990, Pelmenschikovet al. 1992). This value is further reducedby about22 kJ/mol by the addition of a single Na cationto the atom ofbri.lgrng oxygen,asis present in the structureof marialitic scapolite.This 40 kJ/mol result is consistent with the calorimetric data of Nawotskyet al. (L982,1985)andthe studyof latticeenergy minimization(Bell et al. 1,992).With these small differences in energy, either of the two configurationsis possible.Therefore,we suggestthat the single2741peakin the marialite samplesoriginates from the Z2(3SilA1) environment.The lack of a small 27Alpeakexpectedfrom the fl(4Si) may be dueto the lack ofresolutionofthe (4S0and(lAl3Si) peaksunder the broad centerand sidebandpeaks,but also may be the cause of the distribution of the quadrupolar coupling constant. A1 this point, thereis no evidenceas to whetherthe A1-O-AI bondsarewithin or betweenthe 4-membered T2 rings. The distancefrom Na to -O(4),betweenthe 4-memberedrings, is 2.9 to 3.0 A, whereaswithin the 4-memberedrings, the distancefrom Na to O(2) is 2.4 A, and to O(3), it is 2.6 A. Therefore,the stabilizing influence would be greater within the fow-memberedrings, wherethe Na-O bond is the shortest.We interpret thesedatato indicate that there are Al-O-Al bonds within the four-membercd T2 rings. CoNcLustoNs 1. Rietveld refinementfrom X-ray powderdataof the marialite samplesfrom the Pamir Mountains confirm tetragonalsymmetry describedby space group l4lm. There is a linear trend in the cell parametersfrom the synthetic marialite end-memberthrough the natural samples(Mea.6,Me7.5andMez.o). 2. Our structuralrefinementof Mea.6@AM-l) is the closest so far to the natural marialite end-member. The syntheticMa end-memberhasunifcell pammeters a 12.0396.(2) A, c 7.54/7(2) A, with V equal to 1093.3(4) A3. 3. Additional electron-density maxima on the differenceFourier mapsindicatethat there are volatile species,probably neutral HrO molecules, in the intersticesaboveand below the A site in the structure of marialite.
4. nAl satellitetransitionand2esiMAS NMR spectra show that Al is locatedonly in the Z2 site. There is a strong indication that A1-O-AI bonds are present betweenmostof the Al atomsin thesecompositionsof scapolite,in contraventionof Lowenstein'srule, but in agreementwith the calculationsof Tossell (1993). AcrNowleocEl'{El.lrs Sampleswere kindly provided by S. Sergeevand F. Rafikova. Researchexpenseswere supportedby a RussianFund for Basic Researchgrant 95-{5-15699 to E.V.S. and Y.K.K., a Natural Sciences and Engineering ResearchCouncil University Research FellowshipandResearchGrantto B.L.S., andNational ScienceFoundationGrant EAR-9316W9 to D.M.J. Specialthanksaregiven to David A. Vanko for advice on the synthesisof marialite. We thank Jian-jie Liang, an anonymousreferee and AssociateEditor Peter B. LeavensandRobertF. Martin for helpful commentson this manuscript. REInnsNCEs The H.T.,JR.& Kotnrranr, J.A.(1984): AnrrN,8.G.,EvANs, crystalsffuctureof a syntheticmeionite.NeuesJahrb. Mineral.Abh.149,309-342. R.A. & Ceuow, C.R.A.(1992): BELL,R.G.,JAcKsoN, studv. l,owenstein's rule in zeoliteA: a computational kolites 12,870-871. E.L., SoKoLovA, N.V. & DoRoKHovA,G.I. BELoKoNEVA, (1991):Crystalsffuctureof naturalN4Ca-scapolite- an intermediatememberof the marialite-meioniteseries.Sov. Phys.Crystallogr.36, 828-830. & Unusov, V.S. (1993): Scapolitescrystalline structuresof marialite (Mell) and meionite (Me88) - space group as a function of composition. Crystallogr. Rep. 38(1), 25-28 (translated from Kristallogr. 38, 52-77). BSRAR,J.-F. & Lrnmt, P. (1991):E.S.D's and estimated probableerror obtainedin Rietveldrefinementsv/ith local conelations.J. Appl. Cryxallogr.24, l-5. CoMoDr,P., Mum,u, M. &Zxezzt, P.F.(1990):Scapolites: variationof structurewith pressureandpossiblerole in the storageof fluids. Eur. J. Mirrcral. 2, 195-202. DERoUANE, E.G., Frupnr, J.G. & Baurtoos, R. (1990): Quantum mechanicalcalculationson molecular sieves. II. Model cluster investigation of silicoaluminophosphates. J. Phys.Ch.en.94,1687-1'692. DoNNAy,G., SHAw,C.F., m, Buu-sR,I.S. & O'Nen, J.R. (1978):The presenceof HCI in scapolites.Can.Mineral, 16.341-345. DUNN,P.J., NneN, J.E. & NoRBERc,J. (1978): On the J. Gemmol.16,4-10. compositionof gem scapolites.
1050
T}IE CANADIAN MINERALOGIST
ENGHfiARDT, G. & Mrcru, D. (1987):HiSh ResolutionSoli.d StateNMR of Silicatesanl T*olites. JohnWiley andSons, New York" N.Y.
Psr-MEr,rscHrKov, A.G., PAItKsrms, E.A., Eotsnnnasnvnq M.O. & Znnounov, G.M. (1992):On the lowenstein's rule and mechanismof zeolite dealumination.J, Phys, Chem.96,7051-7055.
HAss, E.C., MEay, P.c. & Pmm, P.J. (1981): A nonempirical moleculm orbital study on Lowenstein's rule andzeolitecomposition.J. Molecular Struct.76, 389-399,
Poucuou, J.L. & Hcrror& F. (1985) '?AP- (phi-rho-Z) procedure for impmved quantitative microanalysis.1n MicrobeamAnalysis (J.T. Armstrong,ed.). SanFrancisco hess, SanFrancisco,California(104-106). Hassal, I. & Busncr, P.R. (1988):HRTEM characterization of scapolitesolid solutions.Am. Mineral.73, 119-134. Saunn,J. & ENcnunror, G. (1982):Relativestability of A1-O-A1 linkagesin zeolites.A non empirical molecular Htrr, R.G. & FLAcK,H.D. (1987):The use of the Durbinorbital study.Z. Natur. 37L, 277-279. Watson d statistic in Rietveld analysis. J. AppI. Crystallogr. 20, 356-36'1. Scnnnnn, J. (1989): Profile refinement on IBM-PC's. JoHeNNss,W. (1978): Pressurecomparingexperimentswith I.U.Cr. Int. Worl
