American Mi,weralogist Vol. 57, pp. 732-750(1972)

THE CRYSTAL STRUCTURES OF PYROPHYLI,ITE, lTC, AND OF ITS DEHYDROXYLATE RoNer,oWennr,n AND G. W. Bnrxor,ur, Materinls R'esearch Laboratorg, artd,the Department of Geosci'ences, The Pennsgluania StateUniuersitE, [Jni'uersitAPark, Pa'

Agsrnect

INrnoouctroN

material may have some advantages despite the limitations imposed by the powder diffraction technique.

ExpnmunNtal crushed powder passing a 200-mesh screen was used. Diffractometer powder pattems were recorded with Ni-filtered cuKa radiation at l/4 (2d),/min and 732 STRUCTURE OF PYROPHYLLITE 733

with 1"(2d)/inch chart paper, and were calibrated rvith respect to silicon powder (o - 5.43062A at 21"C). Some reeordings were made with FeKa radiation. AIso Debye-Scherrer powder photographs were recorded wiih CuKa and with CoKa radiations. For intensity (see Brindley and .Wardle, data, side-packed samples were used 1970). Overlapping reflections were resolved graphically and intensities obtained by measurement of the resolved areas. Basal intensities were recorded from samples oriented from a slurry. Dehydroxylations were carried out to con- stant weight in platinum thimbles in air at temperatures ranging from 550'C (500 hr) to 950"C (5 hr). The diffractometer traces show no differencesin samples fully dehydroxylated over this temperature range.

Srnucrunn ANer,vsrs Unit CeLIs The unit cell parametersof pyrophyllite. lTc and its dehydroxylate phase previous reported (Brindley and Wardle, 1970) have been refined by the least squares refinement computer program due to Evans,Appleman, and Handwerker (1963),with the followingresults:

Pyrophyllite, 1 Tc Pyrophyllite, 1 Tc, dehydroxylate a 5.1614+ 0.00L A 5.191,+ 0.001,;. b 8.9576+ 0.002' 9.1221+ 0.0014 c 9.351r + 0.0018 9.4990+ 0.001u d. 91.030 + 0.02 91.170+ 0.02 p 100.37" + 0.02 100.21" + 0.02 89.750 + 0.02 88.620 + 0.02 v 425.21 + 0.1, i.' 442.5a + 0.1, ;,' These data differ only slightly from those previously given but the departureof 7 from 90o is an interestingresult. In particular it pro- vides indexing for a reflection from the dehydroxylate for which no indiceswere found previouslywith y = 90o (seeBrindley and Wardle, 1970,p. 1266). Idealiaed,Structu.re of Pgrophgllite,1Tc Previouswork has shownthat pyrophyllite has a 2:1 layer struc- ture, with a dioctahedral Al, O (OH) sheet betweentwo Si, O tetra- hedral sheets.The structural arrangementcan be developedwith the Al ions at the zero level with respectto c in the unit cell and in con- formity with a C-centeredcell. In the ideal model, close pagked planes of O, OH anions are placed above and below the level of Al ions. The possiblearrangements within a one-layer cell (seeFig. 1) are obtained as follows: Three possibleoctahedral cation sites are occupiedby 2 Al and one vacancy,the latter occupyingone of the 7M RONALD WANDLE AND G. W. BRINDLEY O( -\ (

6)'i \ C,o r( O(r\ r:) o O( (iro Frc. 1. Possible positions in. arrangement A of upper and lower hydroxyls shown by full a.nd dashed Iarge circles respectively, and possible positions of octahedral cation vacancies shown by small circles. three sites labelled I, II, IIL In the anion planes of composition O2(OH),the (OH)- ionscan occupyone of the positionslabelled 1,2, 3, and one of the positionsl',2',3'. Thesetwo planescan be respec- tively above and below the Al plane, as shown by full circles and dashedcircles respectively in Figure 1, (arrangementA), or vice versa (arrangementB). The oxygen ions in these planes form the apices of the SiOr tetrahedra.In the ideal model,the positionsof the silicon atoms and of the associatedbasal oxygen atoms are then fixed. The total number of arrangementswithin the single layer cell is 3 x 3 x 3 x2=54. The hydroxyl combinationswill be labelledas follows: STRUCTUREOF PYROPHYLLITE 735

1 = (1,7'),2: (1,2'),3: (1,3'),4 = (2,I,),8 = (2,2,),G,=(2, 3'),7 = (3, 1'), 8 = (3, 2') and,g: (8, B,),and eachcombination has the A or the B anion plane arrangement.Preceding each arrangement 1, 2, . . ' 9, a symbol I, II, III, denotesthe vacant cation site. The ideal layer structureconforming most, closely with the observed diffractionintensities was determinedby calculatingthe intensitiesfor the first twenty reflectionsof the 54 possible ideal structures using a computerprogram written by Smith (1968,1967). The atomic scat- tering factors built into the computer program were from self-con- sistentfield calculationsfor Alst and Siaiand from a Suzuki curve for O'-; a temperaturefactor B = 1.0 was usedfor all atoms.The results showedthat only 12 different intensity distributions resulted from the 54 arrangements.Table I showsthe equivalentAl, OH arrangements, each of which has the A and B arrangementof O, OH anion planes. of these twelve arrangements,one gives an intensity distribution con- siderably nearer than the others to the observeddata, namely the equivalent structures IzL, II7A", and 1116A. The detailed intensity data are given elsewhere(Wardle, 7972).Projections of theseequiv- alent structures on (001) show that they differ only in the position of the origin along the b axis. ArrangementI2A has a centerof sym- metry at the origin.

Refi,nedStrwchne ol Prophgllite,lTc The idealizedstructure has beenrefined by a trial and error process on the basisof (i) the known crystal-chemicalcharacteristics of.2:l type layer silicates,and (ii) the fit betweenobserved and calculated X-ray diffracted intensities, and observedand calculated diffractom- eter patterns. As shownby Radoslovichand Norrish (1962),a morenearly correct

Table 1

Egulvalent alumlnum, hydroxyl arrangements

IrL = l'5 = II|]- = II,9=II'II5=f.II'9

It3 = !,8 = II,4 = IIrS= IfL3= IfL4

Tt6 = !,7 = II,2 = Tl'6= TII;2= IIIt'l fr2*IIr7=III,6

Ir4 = Ifr3 = fII,8

It9 = lll5 = fILI 736 RONALDWARDLE AND G, W. BRINDLEY layer model is obtainedby rotating the SiOr groupsin eachhexagonal ring about c'* alternately in opposite directions so that the basal hexagonalring becomesditrigonal. The magnitude of this rotation can be estimatedfrom the observedb parameter'an assumedvalue for the Si-O distance,1.62;Ft, and a regular tetrahedralgeometry' For a given pair of SiO+tetrahedra, one in the upper and one in the lower tetrahedral sheet,the rotations may be in the same or in opposite directions,and canbe denotedby (*, +), (+, -), (-, +), (-, -)' The basal oxygen positionswere modified by rotations of i10" for eachof thesefour schemesand the intensitieswere calculatedfor the first thirty reflectionsof the I1A, "' , I9A structures'The results showedthat the I2A and equivalent structures gave the best fit with the observedintensities with rotations (*, -), i.e.,in oppositedirec- tions for the two tetrahedral sheet's;the senseof the rotations is in- dicatedin tr'igure2. For dioctahedrallayer structures,it is well established(see survey by Bailey, 1966)that the Al, O(OH) dioctahedralsheets depart from the ideal arrangementdepicted in Figure I principally in the following ways: the sharedoctahedral edges are shortenedcausing a counter- rotation of upper and lower O, OH triads, and the sheetthickness is modifiedso that the sheetdimensions parallel io (001) conformwith the o and b parameters.It appearsalso that the aluminumions do not depart from the hexagonalarrangement depicted in Figure 1' If each triad OzOH of octahedral anions lies parallel to (001) with equal O-O and O-OH distances,then their coordinatescan be related to two parameters,namely the Al-O, OH bond length and the octahedral sheetthickness. In the fi.rst stage of refinement,the octahedral sheet thickness was held constanbat the value2.14 A givenby Raynerand Brown (1966), and the Al-O, OH bond length was varied to give projectionson (001) ranging from 1.55-1.65A thereby coveringthe likely value of the bond length. Each projecteddistance gives possiblepositions for the O, OH anions,and the ideal model already selectedenables the individual O and (OH) ions to be identified.The oxygenions are the apical oxygensof the tetrahedral sheets and must be bridged by Si-O-Si bonds. When the tetrahedra are assumedto be regular with Si-O = 1.62 L, this bridging of the apical oxygensimposes twists and tilts on the tetrahedra and also severelimitations on the projected Al-O, OH lengthswhich must lie in the range 1.63-1.64A. The twists of the tetrahedra derived from this geometricalapproach are of the same magnitude,:tlQo, and in the same senseas those already ob- tained. STRUCTA NE OF PYROPHYLLITE 737

\-i ---\ri

o ,\

O 5oro1+033d0353 o.. Jt +0.299 +ottg L, (/o,oH OAI oo /A -orr3 U Q) o,oH o Si -0289

Frc. 2. Layer structure of pyrophyllite projected on (001).

Ideally the thickness of the octahedral sheet should be determinable from an 001synthesis but with 007 as the highest observablebasal reflectionthe synthesisdid not providesufficient accuracy. Accordingly diffracted intensities were calculated for structures with octahedral sheetthicknesses ranging from 2.14to 2.02L and a value 2.08'A was selectedwhich agreeswell with the data collectedby Bailey (1966). An isotropictemperature factor B was used for all atoms; values of B ,= 0.5, 1.0,and 1.5 were usedand B = 1.0 gavea slightly better fit to the experimentaldata than the other two. The calculatedintensities RONALD WARDLE AND G. W. BRINDLEY

3 AdNOONN;Fm6OAo6

Fl $ s3q$iqBHHpsS qa F ,++/+d- p i+oNAido Nh3::tt: A g$$$9il$$f, ssspsd e$$$r. =, -,',eEi =: -t BFfi $nHi z { 11 1 {1 1 -1 j j ; ; ; ; ..i ; i ; -.i..i -i ..i..i i €{ fi'8grf i,n E n'H s's 5 R'E p',H'HrErH'E'$'ff SErf, 5 $ a

+ N N 6 HH A4 p N N

N zv dda + . b oodJ+ + NN o F n S az >o $$s$ $E € s aBHsRgs*E s++ q a2 €io;oi d o; d oi N o Ndddd ++'+ .---_^- N 6do6dd+ da N i N H d d dN 6mo{ N Nqod)N 6N nnN d aa .-H r H -:a

!L t $oooajoooNNNNNNoNiHdillitirln::::i:lii:iti:lgl*is99:9: ; ,2, HF OZ E'NR'ifi fl H $rNE,$ fiH'drndt'R,a',d frN s'E'f iTR'$H SH B'S'$

14 @ + N6 ss g 3 n @ -ess 5^- ts h s ts t H 6o$@3 qqq R.€.qEqqEqqlq E3Fq\\ €q !q fit H S ANN O N N o5 + $JJ O OO60 ''--i- F * - - - = S d I e€ g 6 - * p 9 B g - j R ft $ * t g F H d nAflRei? gHi{ qq X 5R?FRfi6pSqSfi EHHsp **R N N N d + -' + i i + d -i di ni 6i o; o; .; o; ai ai d d oi d oi o; a; oi oi ri "i "i "i d g g g L. : EE'6 E i'i H$ E,i i'i'i i',3'HfiE R'f i HE'i H i Hfl H STRUCTUREOF PYROPHYLLITE ?39

for the optimum arrangementare given in the form of a comparison of observedand calculated intensities (Table 2) , and,also, following Smith (1968), as a comparisonof the observedand computer-simu- lated diffraction patterns; the latter was drawn using a Calcomp Plotter (Figure3). The finally selectedatomic coordinateswhich are consistentwith a center of symmetry at the origin, the interatomic distances,and the bond anglesare given in Table B. Figure 2 showsa projectionof the layer structure on (001), with the basal oxygen networks indicated by dashed lines. The rotations of the SiOa groups away from the hexagonalarrangement are readily seen,and are in oppositesenses, as stated earlier. The angle of rotation measuredfrom the..projection is ilOo. The tilt of the tetrahedra,indicated by the positionsof the silicon atoms with respectto the apical oxygens,is closeto 4o. Figure 4 showsa projectionof adjacentbasal oxygennetworks viewed along c* and the a-parametersof the oxygensare indicated. An aly si,sof Pgrophyllite, 1T c, D ehydr orgl,ate The similarity of the powder diffraction patterns of the initial

PYROPHYTLITEITc

F\c. 3. Experimental (B) and computer-simulated (A) diffractometer patterns of pyrophyllite, lTc. 740 RONALD WARDLE AND G. W. BRINDLEY

fable 3

for Pyrophylllter AtoElc CooralLnates, IDteratoElc dl.stances and augles Itc. -Yr -zi ! y, Atool.c Coorallnates xr Yr zi -x, ll2 ! x, ll2 z'

Center of syoretry at (0, 0, 0). Space group Cf'

0b slgniftes o)rygens I'n basal netnorks; 0 are apical o)

al 0.500 0.167 0.000 0E 0.221 0.186 0.113 sl(1) 0.748 0.000 0.289 ob(1) o.os5 0.387 0.353 s1(2) 0.759 0.331 0.289 ob(2) 0.724 0.167 0.353 0(1) 0.671 0.004 0.113 ob(3) 0.550 0.448 0.335 o(2) o.72L 0.319 0.113

Iateratonlc distances r ff, and angles.

3104 tetrahedla: neao Si-0' 1.62

Eean O-O' 2.64; O-S1-O' 109.5o

A106 octahedra: nean Al-0(0H)' 1.94 g4o 0(0n) parallel to laYers' nean 0-0, 2.85; O-A1-O' 0(0H) tnclined to layers' mean O-0, 2.82; O-A1-0' 93o ehbred octahedral edges' nean 0-0, 2.49; 0-A1-0, 8Oo

Interlayer dLstances: (Dlstances > 3.4R are onltte'l) 4' Ob(l). . .0b(6) correspon

ob(1) - 0b(4), ).Jt ob(2) - ob(5), 3.12 0b(2) - 0b(4), 3.07 ob(3)- ob(4), 3.12 0b(2) - 0b(5), 3.37 ob(3)- ob(6), 3.32

mineral and of its dehydroxylateindicates that dehydroxylationpro- ducesonly small structural changes.Two approachesto the structure analysis have been considered.(i) A direct approachexamines the effect of the dehydroxylationon the 001 Fourier synthesis.(ii) An intuitive approachconsiders that adjacentpairs of (oH)- anions (see Figure 2) react to form O'- and HzO and that the residual O'- ions remainin the layer structureeither distributed over the original (on;- sites,or in midway positionsat the a-level of the Al ions. STRUCTURE OF PYROPHYLLITE 741

t

Frc.4. Projection of adjacent basal oxygen networks viewed along c*. Upper network solid circles, lower network dashed circles. Oxygens labelled l, 2, ... 6 refer to Ob(l), ... , Ob(6) given in Table 3; z-coordinates(X1000) are shown.

The 001 synthesesof the initial and dehydroxylateforms of the mineral are shownin Figure 5. For the latter synthesis,the signs of F'(00r) weretaken first to be the sameas thosefor the initial mineral. The resulting electron density curve showeda diminution of the elec- trons associatedwith the O, OH ions and a correspondingincrease of electronsassociated with the Al ions aL z = 0 which points to removal of OH ions and location of the residual oxygen at the a = 0 level. This result is consistentwith the intuitive approachmentioned earlier. Accordingly,the signsof F(002) wererecalculated with pairs of (OII)- 742 RONALD WARDLE AND G. W. BRINDLEY

\ oto \ o,oH c o \\ at oC AI ! o I u I I I \ L I \ I I I I

0 0.1 o2 05 d(oor)

Frc. 5. One-dimensional Fourier syntheses of pyrophyllite, full line, and its dehydroxylate, dashed line' ions replacedby a residual O'- ign lqcatedin the plane a = 0' Only the sign of F(004), a weak reflection,required changing. The resulting Fourier synthesisis shownby the brokenline in Figure 5. The resolved peak areashave beennormalized with respectto 100 electronsin the combinedSiaO6 peak. For the initial pyrophyllite, 55 electronsare found in the Oe(OH)ePeak and25 electronsin the lz(Ll| peak as compared.with 60 and 20 respectively from the atoms involved. The total electrons,howevet, are 180 as expected.For the dehydroxylate, 36 electronsare found in the O+peak and 32 electronsin the % (A1+Oz) peak, as comparedwith the expectedvalues 40 and 30 respectively; the total electronsare 168 as comparedwith 170 expected. The model which emergesfrom the replacementof 2(OH)- ions by O'- in a midway position involves a five-fold coordinationof Al by oxygen.Supporting evidence for this model has been obtained from measurementsof the AlK, fluorescencewavelength which has been shown to vary almost linearly with the averageAl-O distancein a number of crystals for which the averageAl-O distancesare ac- curately known (Wardle and Brindley, 1971). These measurements lead to an averageAl-O distanceof about 1.80A in pyrophyllite de- hydroxylate as comparedwith 1.70-1.74A- tot AlrY and 1.9G-1.94for AIvr. Evidently a meanA1-O distanceof 1.80A could correspondwith Alv or with a l:1 mixture of AlrY and Alvr. A mixture of four-fold and six-fold coordinatedAl ions cannotbe ruled out )r priori, but sincethe STRUCTUREOF PYROPHYLLITE 745

environments of all OH-OH pairs are identical in the pyrophyllite structure, it is not unreasonableto suppose- that the residual o*ygun, also will have identical environments. A detailed model has been developedwith the,residual oxygensin the Al plane at z = 0, with five Al-O distancesof approximaielyt.g0 A, and o-o distancesin the range 2.85*8.10A. Theseconditions in- volved displacementsof the Al ions towards the va.cantoctahedrar sites and related movementsof the o ions; arso adjustmentsof the sio+ tetrahedra were necessitatedby the movement of the apicar oxygensand the assumedregular form and size of the tetrahedra. various atomic coordinateswere derived to conform with these re- quirements; those finally chosenas giving the best agreementbetween observedand calculatedintensities (Table 4) and betweenobserved and computer-simulateddiffraction patterns (Figure 6) are risted in Table 5. The resultingstructure has a centerof symmetry and the co- ordinatesin Table b are basedon axeshaving th! origin at the center of symmetry.The labellingof the basaloxygens Ob (l), . . . , Ob(6) in Table 5 is consistentwith that used in Tanre B and Figure 4. tr'igure7 shows the Al-o sheet of the dehydroxyrate structure projected on !9otl. The basal oxygen networks are similar to those depictedin Figure 2 and therefore are omitted. from Figure z in order to show more clearly the structure of the Al-o sheet. The relations between adjacentbasal oxygenplanes also are similar to those shownin Fig- ure 4.

Drscussror Struchne Reliabititg The reliabilitv factor.E = -F(carc) r lr'(obs) l/rP(obs) is not applicableto the present analysis becauseof the many unresorved reflections. A sirnilar factor B', defined as folrows, thereiore has been used:

R, : D l.I(obs)'/, _ r(calc),/,1/Dl(obs),/" For unresolvedgroups of reflections,.I(obs) and rikewise-I(calc) are the sumsof the intensitiesof the unresolvedreflections. For the pyrophyllite lTc structure, .8, basedon the first thirty_one calculated reflections for the ideal structure is 24.0 percent. For the same reflections,with the tetrahedral groups rotated =10 degreesbut with no changein the octahedrargroups, R' = 7z.g percentlFor the final model, with modified octahedrar and tetrahedral sheets,g, for all the reflectionslisted in Tabre 2 becomes10.6 percent. For the de- 744 RONALD WARDLE AND G. W. BRINDLEY

6dd d + N d F 6 N scs s $ q $ E 4q q as ;); j j .i i ; H -

6 d d t-i-; ^c? ;: t*eet ol]-?;-i ;;-;; 6

H 3::::iiii:ii:liliilBlf ili 'E E,f i,tsBts'5 n 5 *,fi,tfi H'H:fl'ErW',S"HrflE'$$'E

Qdodddd+dF

@@ 9SR3A qq FF9 qq A

* ;-;--83;F;-e-46-il;-i N N o a . :?-f;' 6^i :;;-: i tqiiiqgg::::sl::g35I55B}ii3iiS:i 's'ff E H$ $f; H'tsrff$'f; 'fl ,3 i'$rH'p i'f; HHtsE:HrtsE'fl *'Sfi HE'i'H

F OH6 O O/NNN

ss s 6 eE dg8)q $€3$8fr [] dr in i,i E o .,io; d J d d o; a;oiN -''i-l -' d N;-iT;-i;i^i - .?^i'6 / + d Ci *'

****1*33::::33:::i$lig:9Ellll A OE g'5'?Eri'f,'$ E f i'E!*,R5 fi,fl'H'N',E fi'EH!fr d g'f ifi ft',$

+ _u 9Vl od1 N o+NS€i.* I 9: 35Fgsd3o I q q3 H.fi o s,*s$snr5€fls*alE{€. oi aoi * NNN li*Yi1;;;; .;;a;"iddidoi "i o rE A+ .-_.Fo 6Fe i Fd F _. gsgndii:-. E*. i69o@ *:51lIl5ss5:$s3:t55$n$5:*:::t E g g iEEri'EE i,i'fi H'E B E Hri'fl'fl'H RE!f, f;'fl'H'fi A!fl STRUCTURE OF PYNOPHYLLITE 746

PYROPHYLLITEITC DEHYDROXYLATE

Frc' 6. Experimental (B) and computer-simulated (A) diffractometer paf,rerns of pyrophyllite, lTc, dehydroxylate.

hydroxylate structure and using all the reflectionslisted in Table 4, R' :71.2 percent. On the basis of these results, and the very close fit betweenthe observedand the calculateddiffractometer patterns, Figures B and 6, it is concludedthat the derivedstructures approach the true structures. Further refinementon the basis of X-ray powder data is not war- ranted.

The Structuresof PgrophEllitelTc and its Dehydroryl,ate rn the case of pyrophyllite, the octahedral and tetrahedrar sheets are similar to those of other dioctahedral layer silicates. of particular interest are the interlayer relations illustrated in Figure 4. Becausethe SiOagroups are tilted with respect to (001), one oxygen of each basal 746 RONALD WARDLE AND G. W. BRINDLEY marked tendency for oxygens of one layer to fit between- two or three oxygens of the adjacent layer along lines parallel to [110]. This re- latiJnship may well be a consequence of attractions between Si** ions

table 5

AtoEtc Coordlnate3, IsteratoElc dtstabces aail angles for Pyrophylllte' llc, dehydroxylate. -zi Atoulc cooEdinatea. x' lt 2i -xt 'y, !12 t x. U2 ! y' z'

Certer of sy@try at (0' 0' 0). Space grouP cl.

A1 0.552 0.149 0.000 o(2) 0.728 0.292 0.115 s1(1) 0.225 0.486 0.286 ob(1) 0.037 0.378 0.355 sl(2) O.749 0.312 0.286 ob(2) o.7r7 O.L52 0.355 Or 0.250 0.250 0.000 ob(3) 0.522 0.425 0.320 0(1) 0.123 0.492 0.115

Interatoolc distances' 8, and agles.

5104 tetrahedra3 Ee s1-0, 1.62 reaa 0-0, 2.65 o-s1-0, 109.50

Al05 trLgonal bipyraEid: A1-0r' 1.80 @& A1-0, 1.82 0r- (1) or-0(4), 3.1r 0r-A1-o(1) (4)r 1190 . 0r- (2) 0r-0(3), 2.s7 or-A1-o(2) (3) r 9oo looo 0(1)-0(2) o(3)-o(4), 2.79* o(1),(3)-A1-o(2),(4), 0(1)-0(3) o(2)-o(4), 2.34 o(1),(2)-A1-o(3),(4), 8oo o(1)-o(4), 3.18 o(1)-A1-O(4),1230 o(2)-o(3), 3,65 o(2)-Ar-o(3),1800

*shared eclge

Interlayer dlstances: (dtstuces > 3.4R are onttte't)

ob(l)-ob(4), 3.28 ob(2)-ob(5),3.28 ob(3)-ob(4),3.23 0b(2)-0b(4),3.22 0b(2)-0b(6),3.23

0(3) and O(4) are related resPectively to O(I) and 0(2) by a ceuter of sye€tr)r' Ob(5), Ob(4) an

z O +olrs o O olo oo O -o:ts Frc. 7. The Al-O sheet of the dehydroxylatestructure projected on (001). increaseindicated an untwistingof the tetrahedralsheets following the reorganizationof the octahedtalsheets. The presentanalysis indicates that the expansionsarise largely from the rearrangementof the Al ions and the insertion of oxygenions in the Al planes.Seen in projection on (001),the tetrahedraltwists are not much changed. The five-fold coordination of Al is an interesting result. Previous discussionsinvolving pyrophyllite dehydroxylate(Heller et aI., lg62; Brett et al., 1970) have mentionedthe possibility of a five-fold co- ordination but definite evidencehas beenlacking. Heller et at. (1962) stated that infra-red evidenceprecludes four-fold coordinationbut "does not excludefive-fold eoordination".The form of the five-fold groupnow found is approximatelya trigonal bipyramid and is similar 748 BONALDWARDLE AND G. W. BRINDLEY to that found in andalusite (Taylor, 1929; Burnham and,Buerger, 1961) for which the latter authors found an averageAlv-o distance of 1.836A and a'bent'pyramidal axis with O-AI-O = 162"' In the

and Buerger. T he D ehydrorylation R eaction There has been considerableargument (seereview by BteLt,et aI., 1970) wheiher dehydroxylation of hydrous layer silicates proceeds by homogeneousor inhomogeneousprocesses' The reaction in pyro- phyllite appearsto be homogeneouswith pairs of (oH)- ions reacting Iocally to form 02- and.Hzo. An inhomogeneousreaction is visualized as a migration of protons to favorablereaction sites with a counter- migration of cations,Al3' ions, to maintain electricalneutrality. on this hypothesis,oxygen; ions are not lost from that part of the struc- ture which forms the dehydroxylate, but only from reaction zones where they combinewith protons to form water and liberate Al3* ions for counter-migration. Thesemechanisms have beendiscussed particularly by Nicol (1964) in relation to a comparableproblem involving the dehydfoxylationof muscovite,which had been studied earlier by Eberhart (1963)"The latter also obtained001 syntheses of muscoviteand its dehydroxylate' and concludedthat a homogeneousreaction replaced 2(OH)- by O'z-in the planeof the Al ions.Nicol (1964)criticized the conclusionof Eber- hart and concludedthat the i,nhomogeneousmpchanism "explains the observeddata at least as well as the homogeneousone". It is difficult to reconcilethe inhomogeneousmechanism with a coordinationof Al other than six-fold. The AlK" wavelengthmeasurements clearly in- dicate a ilecreasein coordination;four-fold coordinationis precluded by the infrared data. The availabledata, includingthe presentstruc- ture analysis,all indicate a five-fold coordinationand on this basisit is consideredthat the pyrophyllite dehydroxylation reaction (and probably also that of muscovite)is homogeneous. The increasein c and d(001) on dehydroxylationis related partly to the increasedtilt of the sio+ tetrahedra,which increasesthe effec- tive thicknessof eachtetrahedral sheet by 0.037A, and partly to the increasedthickness of the Al, o sheetby 0.071A. The total increase, 0.037 + 0.071 + 0.037 : 0.145A, in the layer thicknessis about identical with the observedincrease of 0.150A in d(001). In other STRACTU RE OF PYROPHY LLITE 749

words,the separationof adjacent basal oxygensshows little change; the averageseparations of near basal oxygensin Table B, (8.28A) , and in Table b, (9.2b A) are nearly identical. The net result of the parameterchanges is a 3.g percentincrease in unit cell volume with dehydroxylation. In trioctahedral layer silicates, dehydroxylationgenerally occurs at highertemperatures than in dioctahedralsilicates, and recrystalliza- tion generally occurs at about the same temperature.The present results for pyrophyllite, and the probability that a similar process occursin muscovite,suggest a possiblestructural reasonfor the dif- ferent thermal behaviors,namely that in the trioctahedral minerals there is less freedomof movementfor cation and oxygenreorganiza- tion when reaction occurs. Consequentlyhigher temperaturesand major structural changesare involved with the trioctahedralminerals. Nomenclature Previously (Brindley and Wardle, 1gZ0)the dehydroxylatedphase was deseribedas an anhydride,but following the recommendationof BretL et al. Q97$ the term "dehydroxylate"is now preferredas being a more accurateterm. Anhydride is a term more aptly used when structuralwater is removed. The one-layer triclinic structure of pyrophyllite is now labelled "pyrophyllite, lTc". Similarly the two-layer monoclinicstructure can be labelled "pyrophyllite,2M". By placing lTc and 2M alter rather than beforethe nameof the mineral,the indexingof the mineral under ttPl' .(Mt, rather than "T" or is assured.

Acrwowlnncunnrs

of computer facilities was made possible by a grant from The pennsylvania State University.

R^urnnpNcns Berrnv, s- w. (1966) The status of clay mineral structures. clavs claa Mi,neral. 14. t-23. Bnnrr, N. H., K. J. D. MecKrr.rzrn,exn J. H. Snenp (f920) Thermal decomposi- tion of hydrous layer silicates and their related hydroxides. euart. Reu.24, 185-207. BnrNnr,nv, G. W. (1971) Reorganization of dibctahedral hydrous layer silicates by dehydroxylalion. M,i,neral. Soc.Amer. Spec. Pap. t, ZO-ZB. 750 RONALD WARDLE AND G. W. BRINDLEY

ANDRoNALD Wenor,p (1970) Monoclinic and triclinic forms of pyrophyllite and pyrophyllite anhydride. Amer. Mirteral. 55, 1259-7n2' Bunxrrervr, C. W., aNo M. J. Buincnn (1961) Refinement of the crystal stnrcture

March. 19,63,p.4243. Hnr,r,nn, L., V. C. FAnunn, R. C. MacKnNzru, B. D. Mrrcunr,r,, ern I[' F' W' Teyr,on (1962) The dehydroxylation and rehydroxylation of trimorpltic dioctahedral clay minerals. CIag Mineral. BuII. 5, W72, Nrcor,, A. W. (1964) Topotactic transformation of muscovite under mild hydro- thermal conditions. Clags Clng Mi,neral. 12, 1l-L9. Rerosr,ovtcn, E. W., eNo K. Nonnrss (1962) Cell dimensions and symmetry of layer lattice silicates. I. Some structural considerations. Amer. Mi,neral. 47, 599-616. Ravxon, J. I{., eNr G. Bnowx (1966) Stmcture of pyrophyllite. Clags Clag Mi.neral. 13, 73-84. surrrr, D. K. (1963) A Fortran program for calculating X-ray powder diffrac- tion patterns. Lawrence Radiation Laboratory, Livermore, california. UCRL-71e6(April f9$). - (1967) A revised program for calculating X-ray powder difrraction pat- terns. Lawrence Radiation Laboratory, Livermore, california. ucRL-50264 (June 1967). - (1963) Computer simulation of X-ray difiractometer traces. Norelco Rep. Apr-June 1968. (Reprinted by the Lawrence Radiation Laboratory. Liver- more, California, as UCRL-70674). Swrr.roer,n,L. D., axo I. R. Eucsps (1968) Hydrothermal assosiation of pyrophyl- lite, kaolinite, diaspore, dickite and quartz in the Coromandel Area, New Zealard. NZJ. Geol. Geopltys. 11' 1163-1185. Tarr,on, W. H. (1929) The stnrcture of andalusite, ALSiO". Z. Krtstallngr' 77, 205-218. Wennr,n, Rower,o (1572) A Struntural Study of Pyrophyllite ITc aln'd i'ts Dehy' dronglate. Ph-D. Thesis, The Pennsylvania State University. eNn G. W. Bnrrvnr,ov(1971) Dependence of the wavelength of AlKc radiation from alumino-silieates on the Al-o distance. Amer. M'i,neral. 56, 2123-2728.

Manuscnpt receiued, Nouember 29, 1971; acceptedJor ptblicotinm, January 31, Im2.