American Mineralogist, Volume 80, pages 1I 16-1131, 1995

Complex polytypism: Relationshipsbetween serpentine structural characteristicsand deformation Jrr,r.r.lNF. B.lNrrnr,o,Sruncrs W. B.llr,rvrt Wrrrrarr W. B,lRKEn Department of Geology and Geophysics,University of Wisconsin-Madison, 1215 West Dayton Street, Madison, Wisconsin 53706, U.S.A. Ronrnr C. Sernn II Bureau of Topographic and Geologic Survey, Department of Environmental Resources, Harrisburg, Pennsylvania 17 I 05-8453, U.S.A.

AssrRAcr from Woods Chrome Mine, l,ancaster County, Pennsylvania, consists of planar serpentine,randomly interstratified serpentine-chlorite,a seriesofphases basedon regular interstratification ofserpentine and chlorite, minor chlorite, polygonal serpentine, and antigorite. The serpentinemineralogy is complex, including structureswith long-range order in (l) the octahedralcation sequenceand (2) the sequenceofdisplacements between adjacent 1:l layers. In addition to previously described (but generally less common) 2I, 2Ht,2H2,6R,, and 6Rrlizardite polytypes,the assemblagecontains planar serpentines with long-rangeorder in octahedral cation sequencesbut with random b/3 (and possibly a/3) displacementsbetween adjacent layers. By comparison with calculated electron dif- fraction intensities,we identifiedthree- (I,I,[), four- (I,I,[,II and I,I,I,II), five- (I,II,I,II,II), six- (II,I,I,I,II,II), seven-(I,II,I,II,II,II,II and I,II,I,II,I,II,II), and nine-layeroctahedral se- quences.In addition, the assemblageincludes rare three- and four-layer serpentineswith regular stacking; in some casesthe stacking is nonstandard, insofar as zero and +b/3 displacementsoccur together. By comparison between electron diffraction intensities and calculated patterns, we identified a threeJayer regular stacking sequencethat involves 0,0,-b/3 displacementsbetween adjacent layers (a : 98, 0 : 90",.y : 90'), a fourJayer monoclinic sequencewith 0,-bl3,0,+b/3 (a : 90o,A :90", ? : 90"; I,I,II,II), and three four-layertriclinic sequenceswith 0,-bl3,-b/3,-b/3 (a : 90o,0 :9C', ^y: 90"),0,0,0,-bl3 (a:96",0:90, ? : 90'),and -bl3,-b/3,-b/3,+b/3(": 96",B:90',7: 9g'; displacementsbetween adjacent layers. Planar layer silicatesexhibit macroscopicpreferred orientation. The strong lineation in the foliation defined by the silicate sheetsparallels either a or b, suggestingserpentine crystals were rotated, recrystallized, or both during sheardeformation. We suggestthat for crystalswith b parallel to the lineation, deformation induced regular layer displacements.For crystals with a parallel to the lineation, periodic displacementsof OH planesmay have promoted development of regular octahedralcation sequences. Two-layer (I,II) and three-layer (I,I,II) serpentinesand randomly interstratified serpen- tine-chlorite contain frequent but nonperiodic planar defectsperpendicular to a* and the pseudo-a* axesthat are interpreted to displacepolytypic sequences.These defects predate chloritization and, in some cases,appear to serve as sites for chlorite nucleation. Layer silicates are crosscut by late-stagepolygonized two-layer serpentine with disordered or regular stacking. In some caseswith regular stacking, enantiomorphic 6R, segmentswith a common c-axis direction parallel to their boundary are separatedby sectors with 2H, stacking. Polygonized serpentinesare nucleated on steps at the surfacesoflayer silicates and may have developed at a late stageby recrystallization of curved serpentine.

Inrnooucrrox origin. They are common products of hydrothermal al- Structural and compositional details of layer silicate teration of periodotite, lherzolite, harzburgite, etc., in assemblagesmay provide insights into the petrogenetic which primary mafic silicates are converted to assem- history of the rocks that contain them. Serpentinemin- blagesthat include chlorite, brucite, lizardite, antigorite, erals characterize greenschistfacies rocks of ultramafic , and polygonal serpentine.They form in low- temperaturereactions near the Earth's surface.Serpentin- t DeceasedNovember 30, 1994. ization of ultramafic rocks associatedwith spreading zones 0003-004x/95/l l l 2-l l 16$02.00 1116 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION tt17 profoundly changestheir geophysical(magnetic suscep- related to II by a 180"rotation. Structural variations arise tibilities and seismic velocities), isotopic, and chemical when layer sequencesinvolve combinations of type I and characteristics. Serpentine are constituents of II octahedra.The secondsource ofpolytypic variation is chrysotile asbestos,and thus they are also of interest be- the positioning of adjacent planar layers in one of three causeofconcern over their effectson human health. ways to ensure H bonding. The sixfold rings can exactly Although proceduresfor distinguishing standard poly- superimposealong c (zero shift) or be shifted by a/3 along is types of l: I layer silicates, , and chlorites are well the three pseudohexagonala axes (the senseof shift established(see Bailey, 1988a),the significanceofthe de- determined by whether cations in the underlying layers velopment of one layer silicate polytype in place of an- are type I or II) or +b/3 along the three b axes. polytype other is poorly understood. In general,factors considered Various assumptions were made in all deri- to influence polytype development include (l) the chem- vations to limit the number of possible structures-Bai- istry, temperature, and pressureduring gowth; (2) crys- ley's (1969) derivation ofthe 12 standard l:1 layer sili- tal-chemical details, including structural distortions of in- cate poly"typeswas achievedby excluding structureswith dividual layers; and (3) the growth mechanism shifts along both a and b and those with both zero and (particularly ifassociated with spiral growth on screwdis- +b/3 or +a/3 mixed in the same crystal. In the vast locations;Baronnet, I 975). majority of studies, serpentine polytypes identified by Experimentalists have attempted to define P",o-I sta- standard powder XRD techniquesare reported to be one bility fields for polytypes (e.9.,Yoder and Eugster, of the l2 standardpolytypes (Bailey' 1969),I Zbeing the 1955), and observational data have been used to argue most common. However, structures containing more (e.g', that specificpolytypes develop in reaction seriesthat cor- complex sequencesdo form under some conditions relate with increasing temperature (e.9., type I chlorites the 6R, polytype reportedby Steadmanand Nuttall, 1962, are replacedby the IIDb polytype during metamorphism; and the 6I' polytype inferred for Unst serpentineby Hall seeBrown and Bailey, 1962;Karpova, 19691,Curtis et al., er.al., lg76). Results presentedhere illustrate that, under 1985;for other discussionson chlorite seeHayes, 1970; some circumstances,these polytypes may be important Velde, 1965;Walker and Thompson,1990; Walker, 1993; constituents of assemblagesof l: I layer silicates' de Caritat et al., 1993). Other experimental studies on Polytypes based on +b/3 displacementsand charac- micas suggestthat the degreeof supersaturationmay in- terized by long-period regular stacking must involve more fluence stacking (Amouric and Baronnet, 1983). How- complex sequencesof displacements,including sequences (simply re- ever, results such as those of Bigi and Brigatti (1994), that intermix zero and tbl3 displacements -). which reveal numerous complex polytypes in a ferred to below as 0, +, or Possiblesequences for all t/3 single assemblage,suggest that factors other than tem- three- and four-layer structures (excluding reg.ulat (1995)' perature, pressure,and saturation state must also influ- displacements)are listed by Bailey and Banfield ence polytypism. who describe the procedure to identify three- or four- O'Hanley (1991) describedvariations in serpentine- layer polytypes (regular stacking involving zero ot b/3 : (lizardite, antigorite, chrysotile, and polygonal shifts, regardlessof slant of octahedra, 0 90"). Their (or serpentine)distribution with proximity to fault zonesand method involves recognition of the high-intensity the importance of faults and fractures in controlling the missing) 02/ reflections, regardlessof whether the [100]' (1995) accessof fluids neededin olivine + enstatite + serpen- [1 l0], or IT0] zone is obtained.Bailev and Banfield with tine reactions. However, a direct connection between also calculateddiffraction data for l: I layer silicates (up complex polytypism and deformation was not proposed. long-period order in the octahedral cation sequence Some corroborating evidence for such a link might be to trlr"n layers) and provided criteria on the basis of20/ inferred from microbeam X-ray diffraction (XRD) cam- intensity data to identify each sequenceuniquely. Here the era results of Wicks and Whittaker (1977), who reported we use results from this theoretical study to decipher polytypes' the common l7 polytype in pseudomorphic and nonpseu- detailed structuresof previously undescribed ser- domorphic serpentinesbut multilayer polytypes in slick- Lizardite, antigorite, and chrysotile are varieties of to ensided veins. Although it has been shown that mechan- pentine distinguished by the mechanismsthey employ ical grinding during sample preparation for XRD relieve strain associatedwith lateral misfit between the (e.g., experiments can change the chlorite polytype (Shirozu, tetrahedral and larger octahedral sheets Pauling, 1963), relationships between deformation and polytyp- 1930). Although polygonal serpentine(described as Pov- ism have not been explored. len-type by Krstanovi6 and Pavlovie, 1o64)is sometimes Planar polytypes of l:l layer silicates were proposed considered a fourth l:l layer structure (Mellini, 1986)' by Newnham (1961), Zvyagin (1962, 1967), Steadman most results suggestit is simply a special lizardite ar- (1964),Zvyagsnet al. (1966),Bailey (1969),Dornberger- rangement(high surfacearea may modi& its stability rel- de- Schiffand Durovi6 (1975a, 1975b),and Durovi6 et al. ative to lizardite). Polygonal serpentine has been (1981). Polytypism in l:l layer silicatesarises because scribed from numerous localities. It is generallybelieved (and cases' stacking of these layers varies in two fundamental ways. to have formed by modification of in some First, cations occupy one of two setsof positions (I or II), overgrowth on) preexisting chrysotile (e.g., Cresseyand which results in one of two octahedral slants. Set I is Zussman,1976: Mellini, 1986;Mitchell and Putnis,1988). lllS BANFIELD ET AL,: SERPENTINE POLYTYPISMAND DEFORMATION

that intruded into schistsand gneissesin the Ordoyician. The serpentinite is now exposed in the Woods Chrome Mine, Lower Britain Township, LancasterCounty, penn- sylvania (Lapham, 1958). Specimenswere collected circa 1955 (by R.C.S.) from mine dumps (39043'52,N, 76006'2A'W) and correspond to the material termed chromian antigoriteby Glasser al. (1959). The Woods Chrome Mine produced at least 88000 metric tons of chromite from boudinJike pods in the latter part of the nineteenth century. Most of the chromite deposits were located within 100 m of the presently pre- served base of the of the Baltimore Mafic Complex (Hanan and Sinha, 1989). Zircons associated with monazite-(La) and baddeleyite in clinochlore veins in this complex yielded slightly discordant age estimates of 510 Ma (A. A. Drake, personal communication to R.C.S.).Shaw and Wasserburg(1984) reported an ageof 490 + 20 Ma for igneous crystallization. Serpentinized Fig. l. XRD precessionphotograph supporting optical mi_ dunites and gabbroswere describedby McKague (1964). croscopeand electronmicroscopic information indicating that In addition, the complex includes a continental margin layer silicatesadopt one of two orientations.All crystalshave c* island arc or back arc (the JamesRun formation; Hanan normalto the foliationdirection. Approximatelyone-half of the and Sinha, 1989), the Port Deposit Tonalite, and asso- crystalshave b* ratherthan a* parallelto thefoliation direction. ciated melange (including the Bald Friar metabasaltand the Conowingo Creek metabasalt).Together, these units There have been several attempts to explain why vir- represent an ophiolitic island arc complex that was ob- tually all polygonal serpentinescontain either 15 or 30 ducted onto Laurentia during the closure of the Iapetus segments(Cressey and Zussman, 1976; Cressey,1979; during the Taconic orogeny at approximately 455 Ma. It Yada and Liu, 1987; Mitchell and putnis, 1988; Wicks is probable that the chromian serpentinitesformed dur- and O'Hanley, 1988).Chisholm (1992) proposeda model ing the deformation and metamorphisrn that accompa- basedon geometric and strain considerations(particular- nied the Taconic obduction and thrusting. ly the addition of integral numbers of octahedraand tet- Most of the chromite ore in the Woods Chrome Mine rahedra to successivelayers at sector boundaries).More contained 48 wto/oCrrOr. An analysisof the serpentinized recently, Baronnet et al. (l 994) presentedan elegantmod- dunite from a nearby quarry yielded SiO, 37.3010,AlrO, el for the growth of 15- and 30-sector polygonal serpen- 0.15o/o,total Fe as FeO 6.30/o,MgO39.3o/o, CaO 0.230/0, tine. Baronnet et al. (1994) noted that the chrysotile-to- MnO 0.080/0,KrO 0.060/o,NarO 0.010/0,NiO 0.30lo lizardite transition requires unit cells in successivelayers (McKague, 1964). The ore boudins were reported typi- to move into registry along the b axis to establishH bond- cally to be surrounded by a I m thick sheath oftranslu- ing. Baronnet et al. (1994) proposed that the five addi- cent green Ni-bearing antigorite. Chromian chlorite and tional unit cells per layer (neededto complete successive- chromian serpentine,the latter typically having a cross- ly larger diameter layers) reorganize to form five edge fiber or slickensided habit, occurred in vein or breccia dislocations that dissociatesymmetrically, resulting in l5 fillings in brittle fracturesin the chromite. Other minerals arrays of partial edge dislocations that form the sector include common brucite, clinochlore, and dolomite; trace boundaries. The model predicts that l5-sector polygonal maucherite, millerite, pentlandite, and zaratite; and rare serpentine should be characterized,byfive sets ofthree- uvarovite. The serpentinite consists of 810/oserpentine- sector groups with specific polytypic characteristics. group minerals, 5oloolivine (Forr), 60loopaque minerals, In this paper we describethe planar and polygonal ser- lolochromite, lo/otalc, 2o/ocarbonate minerals, l0lobruc- pentine mineralogy of altered and highly deformed meta- ite, the remainder being chlorite and other minerals. Tex- morphosed peridotite and proposea link betweenunusu- tural relations reported by Lapham (1958) suggestser- al polytypism and deformation. The importance of the pentinization precededchloritization. structural characteristicsof these serpentines,which rep- resent precursors for associatedregular serpentine-chlo- Characterization of the layer silicate assemblage rite interstratifications, is explored in a companion paper Powder and precessionXRD patterns were obtained (Banfieldand Bailey, 1996). from serpentine minerals. Three-millimeter diameter ExpnnrurNrt, METHoDs sampleswere removed from a thin section, mounted on copper support washers,and thinned to electron Samples trans- parency by Ar ion milling. Sampleswere C coated before Samples are from the State Line serpentinite, which examination in a 200 kV Philips CM20 UT high-reso- formed by metamorphism of an alpine-type peridotite lution transmission electron microscope (HRTEM) with BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION r 119

cation sequence:(a) two layer (I'If; Fig.2. The [010] SAED patterns ofCr-rich serpentineexhibiting order in the octahedral (note effectsin d- (b) three layer (e.g.,I,I,II); (c) four layer (I,I,II,D; (d-0 duplicate f.o- a thicker area the strongerdynamical "-" streakingof lkL k + 3n reflections in all 0; (g) five layer; (h) six layer; (i) seven layer (nine tayer pattern not shown). Continuous between successivel:1 layers. libif SeEl putterns indiiates disorder in b/3 (and possibly a/3) displacements

by a point resolutionof about 0. l9 nm. Ten ion-milled sam- lated for samplesbetween 3 and 20 nm in thickness poly- ples were examined. Compositional analysis in the Bailey and Banfield (1995).This analysisassumes +a/3. pat- HRTEM was perfcrmed using a NORAN Ge detector types do not contain regular shifts of SAED and NORAN Voyager software. Tilt capabilities of the terns were recorded from the thinnest possible areas revealthat, CM20 UT (a: 115", P: +20") werejust sufficientto (probably usually 7-10 nm thick). Calculations were not great- allow both [010]-typeand [100]-typeselected-area elec- in most cases,distinct intensity variations tron diffraction (SAED) patterns to be obtained from the ly obscured by dynamical diffraction effects in samples same crystal under the most favorable conditions (e'9., up to 10 nm thick. [010]as well as tl l0l). with image Images were interpreted by comparison RBswrs AND TNTERPRETATToN simulations calculatedusing EMS (Stadelmann,l99l). descriPtion The 00/ periodicities in SAED patterns oriented to elim- General sample signiflcant inate dynamical diffraction effectsinvolving 20/ rows were Specimensare mauve in color becauseof the samplesdis- used to distinguish layer silicates. In casesof standard Cr iontent of all minerals. Strongly foliated plane. Specimens serpentine and chlorite polytypes, identification was es- play a distinct lineation in the foliation lineation tablished by comparison with calculated and observed break into elongate,blocky aggregateswith the patternsreported by Bailey(1988b). Octahedral sequenc- parallel to the length of the fragments. Petrographic op- down Z reveals that op- es were analyzedfrom [010] SAED patterns by interpret- tical examination of fragments have extinc- ing variations in intensity of 20l reflections, which are tically continuous millimeter-sized regions perpendicular sensitive to octahedral cation order and essentiallyunaf- tion directions approximately parallel and + fected by variations in stacking order (Bailey and Ban- to the lineation direction (generally l0')' Furthermore' in addi- field, 1995).Identification of stackingsequences in non- XRD photographs from subsamplesshow that, of crystals standard three- and four-layer polytypes involved tion to a common c* direction, the majority (Fig' l)' matching experimental SAED patterns with those calcu- have either a* or b* parallel to the lineation Il20 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION

basedon displacementsalong the b (asin group D) rather than the a axis (as in group B).

Transmission electron rnicroscopy HRTEM imagesand SAED patterns indicate the sam- ple consistsof a complex assemblageof serpentine-group minerals as well as chlorite and large unit-cell structures that are basedon regular interstratifications of serpentine and chlorite. Within the resolution of the energy-disper- sive X-ray analytical technique, all minerals have the same Mg-rich, Al- and Cr-bearing, Fe-poor compositions. The chlorite (Ibb) and regular serpentine-chloriteinterstrati- fications, their microstructures,reaction mechanisms,and origins are discussedin a companion paper (Banfield and Bailey,1996). The serpentine-groupminerals in this sample can be subdivided into three categories.The most abundant are polytypes of planar serpentine. Of lesserabundance are antigorite and polygonal serpentine.

Planar serpentine The most common serpentine in the assemblagehas continuously streakedUkl, k + 3n reflections and a two- layer repeat in 20/ (Fig. 2a and 2d). In [ 100] images,dis- placementsofal3 along the a and pseudo-aaxes (such as found in group B) and b/3 along the pseudo-b axes (as found in group D) both have projected displacementsof zero or b/6. Thus, HRTEM image details cannot be used to distinguish disordereda/3 from b/3 stacking.As noted above, we assumethe displacementsare along b. There are two distinct categoriesofplanar serpentines, Fig. 3. Dark-field imagefrom an areaof serpentineshowing each of which contains members that exhibit long-period a threeJayer20/ periodicity.The regularperiodicity is inter- poll'types not previously. groups pretedto indicatea high degreeoctahedral cation order. described Both have B : 90". The first group (l) has a periodicity of >0.7 nm in 20lreflections (Fig. 2), indicating octahedralorder (Fig. Reflections are arc shaped,suggesting some (-15.) mis- 3) and )kl, k + 3n reflections that are continuously orientation. The strong preferred orientation is also ap- streaked(see Fig. 4a). The secondgroup (2) is character- parent when samplesare examined in the electron micro- ized by periodicities of > 1.4 nm in 20l and discretespots scope. rather than streaks in the 0kl, k + 3n rows, indicating Thin section examination revealed that samples are regular sequencesofb/3 displacementsbetween adjacent comprised almost entirely of layer silicate minerals pep- l:l layers (Fig. a). Serpentineswith regular stackinghave pered by small opaque crystals. The opaques,shown by either a + 90" (triclinic symmetry) or a : 90o(monoclinic analytical electron microscopy (AEM) to be chromite, oc- or triclinic symmetry). In all patterns from serpentine cur in trails that parallel the lineation direction. (00/ periodicity : 0.7 nm) wirh a 201 peiod > 2.8 nm, Powder XRD patterns could be indexed as derived from the }kl, k + 3n reflections are continuously streaked. either a six-layer serpentineor from a mixture of dozyite Polytypes with disordered stacking. Some serpentines (l:l serpentine-chlorite;Bailey et a1., 1995; Banfield et display a 0.7 nm periodicity in 201 (group C: I,I,I . . . al.,1994), serpentine,and chlorite. XRD precessionpho- octahedral tilts). The [010] SAED patterns normally ex- tographs sr.ggest that the layer silicate assemblageis dom- hibit multiples of the 0.7 nm periodicity in 20/ rows. A inated by serpentine with a two-layer periodicity in 20/ two-layer periodicity is most common (Fig. 2a) and can (regular alternation in adjacent layers of set I and set II be distinguished from patterns from regularly interstra- octahedral cations). These could be either group B or group tified minerals (2.1-6.3 nm periodicities in 00/ and 201 D serpentines,as classified by Bailey (1969). Group B see Banfield and Bailey, 1996) only ifthe patternsare serpentinesare very rare, however, and none were iden- from the very thinnest areasor if crystalsare subsequent- tified in this study. Here we describe serpentineswith ly tilted so that 20l reflectionsare not diffracting strongly. semirandom stacking and a two-layer repeat in 201 re- Pattems with a three-layer (2.1 nm) periodicity in 20l flections as two layer (I,II) and assume the stacking is are relatively common (Fig. 2b and 2e). CalculatedSAED BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION lLzl

Fig. 4. The [100] SAED patterns of standard polytypes of Cr-rich serpentine:(a) disordered stacking;(b) 2H,; (c) 2T; (d) 2Hr; (e) 6R,; (f) 6R,. All polytypes except2T (group C) show a two-layer period in 20/ ($oup D).

patterns show that 20/ periodicities and intensities are independent of the sequenceof b/3 displacements be- tweenadjacent l:l layers.Calculations demonstrate that a 2.1 nm periodicity can be explained only by a sequence in which the octahedral slants alternate in a I,I,II-type sequence(Bailey and Banfield, 1995). Although calculat- ed 201 and 201intensities are quite diferent (Bailey and Banfield, 1995),SAED patterns(Fie. 2b) have approxi- mate mirror symmetry. The match between the experi- mental and calculated pattern is much improved if 180' twinning (superimposing201 and 201) is invoked. SAED patterns for serpentineswith a four-layer repeat in 201 exhibit two distinct patterns of 201 intensities. Comparison with calculated patterns reported in Bailey and Banfield(1995) indicates that thosewith intensesat- ellites on the 0.7 nm subcell reflection (Fig. 2c and 2f)

-)

Fig. 5. The [100] SAED pattern from serpentinecharacter- ized by three-layer periodicity in lkl, k + 3n, and a : 98'. Arrows indicate the strongest02/ reflections. rr22 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION

Fig. 6. The [100]SAED patterns from serpentinecharacter- Fig.7. The [00] SAEDpatterns from serpentinecharacter- izedby fourJayerperiodicities in)kl, k + 3n, anda : 90.; (a) izedby fourJayerperiodicities in}kl, k + 3n, ando: 96; (t) -,0, 0, + and(b) -, -, -,0. Arrowsin b indicatethe strongest 02/ 0,0,0,- and(b) +,+,+,-. Arrowsindicate the strongest 02l re- reflections. flections.

are derived from serpentinein which the octahedral cat- matchedwith the resultsof Baileyand Banfield(1995) to ions alternate in a I,I,II,II pattern, whereas those with identify uniquely the following sequences:I,II,I,II,II; uniform 201 = 202 * 203 intensities are from serpen- I,I,I,II,II,II; I,II,I,II,I,II,II; and I,II,I,II,II,II,II (the nine- tineswith regularI,I,I,II alternationofoctahedral cations. layer structureswere not analyzed). Serpentineswith five-layer (Fig. 2g), six-layer (Fig. 2h), Images formed using 20l reflections in close to two- seven-layer(Fig. 2i), and nineJayer (not shown) repeats beam conditions (e.g., using two strongly diffracting 20l in the 20/ reflectionsmust have 3.5, 4.2, 4.9, and 6.3 nm reflections) indicate a very high degreeof octahedral or- periodicities in their octahedral cation sequences.Pat- der (especiallyfor three- and four-layer serpentines)over terns show distinct intensity variations that can be largeareas (Fig. 3). In someregions, the 2.1 nm period- BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION tt23

(a) illustratingserpentine crosscut Fig. 8. (a) Imagedown illustratingdefects perpendic- Fig. 9. Imagedown [010] [001] planar perpendicular Sightingat a low angle ular to the reala* and pseudo-a*axes. (b) SAEDpattern for a. by defects to a*. in 1.4nm fringesun- Carefulexamination ofdiffraction spotsindicates they are star alongthe fringesindicates a displacement (b) SAED patternillus- shapedwith streaksperpendicular to defectorientations. der somethickness and tilt conditions. tratingthat the serpentinehas a two-layerperiod (see arrows) in 201.The weakI .4 nm reflectionin the 00/ row is dueto dynam- icity is modulated every three unit cells, producing a 6.3 ical diffraction. nm periodicity. In rare areas,images indicating imperfect periodicities of up to 50 nm were recorded. Polytypes with regular stacking. Stacking order is in- Four types ofSAED patterns, distinguished by the po- dicated by discrete lkl, k + 3r reflections. For crystals sition of the strongest02/ reflection or an absent reflec- with a 1.4 nm 02/ periodicity, polytypes establishedby tion, were obtained from the fourJayer serpentine,indi- comparisonwith resultsof Bailey(1988b) and Baileyand cating four distinct four-layer, regularly stackedsequences. Banfield (1995; for the caseof 6Rr) were 2H, (group D, A unit cell with a : 90ois appropriate for two groups of Fig. 4b), 2T (group C, Fig. 4c), 2H, (group D, Fig. 4d), crystals with a 2.8 nm repeat in 021, and a triclinic cell 6R, (group D, Fig. 4e), and 6R, (group D, Fig. 4f). In with a : 96owas selectedfor crystalswith two other types some cases,these crosscut well-ordered (e.9., 2I1, cross- of SAED patterns. Comparison between intensities cal- cutting 6Rr) and randomly interstratified serpentine- culated for model structures(a : 90") by Bailey and Ban- chlorite crystals. field (1995) and those in Figure 6a clearly indicates reg- A triclinic unit cell with a : 98" was chosen for ser- ular 0,-,0,+ stacking,whereas intensities in Figure 6b pentines that exhibit a threeJayer periodicity along 0ft1 indicate a regular -,-,-,0 sequence.Likewise, intensi- k + 3n rows (i.e., a 2.1 nm periodicity in 02/). The po- ties in Figure 7a indicate regular 0,0,0,- stacking, and sition of the strongest02l reflection is the same in each thosein Figure 7b indicate +,+,+,- stacking.Although of the three-layer patterns recorded, indicating a single poll'types with fourJayer regular stacking commonly have threeJayer triclinic polytype (Fig. 5). On the basis of re- I,II octahedralcation sequences,patterns with I,I,II,II were sultsreported by Baileyand Banfield(1995), the stacking also recognized. sequenceis determineduniquely to be 0,0,-. The 2.I nm Planar microstructuresin serpentine.Planar serpentine periodicity in 20/ indicates a I,I,II octahedral cation se- frequently contains nonperiodic defectsparallel to (100), quence. It is probable that the -bl3 displacement cor- (l l0), and (1T0).Although thesefaults are commonly de- respondswith the changefrom type I to type II octahedra. veloped in approximately equal numbers parallel to each tl24 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION

Fig. 10. High-resolution [010] image of serpentinesurround- ing beam-damagedregions associatedwith defectsillustrated in Fig. 14. Arrows indicate that the twoJayer periodicity is offset. Fig. I 1. High-resolution [010] image of discontinuousdefects in two-layer serpentine.Details in contrastand fringe orientation suggestdefects separate polytypically distinct sequences. ofthese planes(Fig. 8), they occur in only one orientation in some areas.Faults may terminate without major struc- tural disruption of surrounding serpentine,may be cor- Antigorite related in position across one packet of serpentine,and The sample contains small (usually <0.2 pm in length havea maximum width of -3 nm (Fig. 9). In someareas, and width) antigorite crystals that are both concordant fringes are apparently offset across faults and in others (Fig. l2) and discordant with respectto surrounding pla- they are not (Fig. 9). Rotation offringes by a few degrees nar serpentine.Small crystalsthat crosscutthe planar ser- acrossthe faulted region occurs in some areas.Faults are pentine and serpentine-chlorite probably postdate the especiallyprone to beam damage,so that it is difficult to chloritization event. Images and [010] SAED patterns record images that reveal their detailed structure (Fig. from antigorite crystals indicate a - 5 nm modulation l0). Comparison betweenordering patterns on either side periodicity approximately parallel to a*. Groups of sat- of the zone suggeststhat, at least in some cases,faults ellites around the main serpentinereflections form lines separateregrons in which the polytypic sequenceis out that deviate from a* by < l'. On the basis of the work of of phase (Fig. l0) or is quite different (Fig. I l). Distinct Spinnler et al. (1984) and Otten (1993), the negligible patterns of serpentine-chlorite are sometimes observed deviation of the direction of modulation periodicity from on either side ofthe fault, suggestingthat chlorite nucle- a* in SAED patterns is interpreted to indicate relatively ation postdated development of the planar faults. defect-freecrystals formed during late-stagealteration. BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION tt25

Fig. 12. (a) HRTEM image down [010] illustrating antigorite intergrown with lizardite. The SAED pattern in b indicates a -5 nm antigorite modulation. Rows ofsatellites deviate < 1'from a*, indicating rhat antigorite is relatively perfect (e.g.,Otten, 1993).

Polygonal serpentine pseudo-a*-c*)are separatedby sectorswith a : 90. (some areas,2H, or 2Or; other areas,2H, or 2Or-pseudo-b*-c*; In some regions, planar serpentine polytypes and in- Figs. 15 and 16).It is unlikely that $oup B and D poly- terstratified serpentine-chlorite are crosscut by regions types are mixed in polygonal units. On the basis of the partially filled by polygonal serpenrine(Fig. 13). Secrors abundance of group D polytypes, we suggestthe most generally emanate from a single point that is commonly common arrangementinvolves repetition of the sequence located on a surfacestep on planar serpentine.Within an 6R, 6Rr-enantiomorph, 2H, (Fig. 16). This pattern is individual sector,b is parallel to the layersand a parallels consistentwith predictionsofBaronnet et al. (1994),un- the tube axis, as in normal chrysotile. Occasionally, like the alternative consistingof group B structures(2Mr- chrysotile coexists with polygonal serpentine (Fig. l4a). pseudo-a*-c*enantiomorphs plus 2Or-pseudo-b*-c*).In In some areas,sectors serve to link adjacent layer silicate the former case,the 02l intensity pattems indicate that packets in different orientations (Figs. 13 and l4). Vari- b* of adjacent 6R, polytype sectorssuperimpose, so that ations in fringe periodicities in thicker areas indicate a common c-axis direction parallels the sectorboundary. mixtures of polytypes. For caseswhere the sectorbetween the enantiomorphic In many cases,the two-layer serpentine (I,II) is char- pairs has a one-layer period in 021,our results are con- aclerized by random b/3 (and possibly a/3) displace- sistentwith the predictionsofBaronnet et al. (1994)only ments between adjacent layers. However, polygonal seg- if the sectorsare all group B serpentines(enantiomorphic ments with regular stacking are also present. Where regular 2Mr-pseudo-a*-c* pairs plus 2Or). The {062} surfaces stacking is observed, polygonal arrangementsconsist of (for the 2H, or 2Or polytype, and related planesfor other repeating patterns of three sectors (five sets of a three- polytypes) form grain boundaries between adjacent sec- sector sequence for complete polygonal units). Most tors and separatesectors from planar serpentine-chlorite commonly, pairs of enantiomorphic sectors(6R, or 2Mr- crystals. lt26 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION

Ftg. 13. Image down [00] illustrating polygonal serpentine in a zone that crosscuts planar serpentine. Polygonal crystals appear to have nucleated at steps on the surface of the planar serpentme.

DrscussroN AND coNcLUsIoNs Serpentinepolytypism Comparison between calculated (Bailey and Banfield, 1995) and observed SAED patterns indicates that pro- ceduresfor decipheringpolytypic stackingand octahedral Fig. 14 (above and right). The [00] imagesof polygonal ser- sequencesin l:1 layer silicateson the basis ofrecogrrition pentine. (a) Polygonal segmentsbridge planar serpentine.Beam of strong or missing reflectionsare feasibleif patterns are damage is concentratedat segmentinterfaces. A indicates anti- gorite, serpentine.(b) Polygonal obtained from thin areas where the effectsof dynamical and C indicates chrysotileJike serpentine developed at grain boundaries between planar ser- diffraction are minimized. Using this approach, we de- pentine-chlorite intergrowths. Arrows in the top right corner in- mineralogy of shearedsam- termined that the serpentine dicate that (inclined) planar faults on (100) may be antiphase ples of the Cr-rich State Line serpentinite is distinctive boundaries between chlorite units and separate serpentine and in two important ways. First, the polytypic assemblageis chlorite. diverse. Table 1 lists 2l different planar serpentinestruc- tures identified in our study. Second,the assemblageis nation of octahedral tilts with a mistake (twin?) every few characterized by less common structures, including at least unit cells or as intergrowths ofgroups C and D serpentine. l4 polytypes not recognizedpreviously. Although it is possible that group C serpentines are SAED patterns and dark-field imagesindicate that ser- stabilized by lower Al contents, no compositional differ- pentine octahedral cations are highly ordered. However, ences between polytypes were noted by analytical elec- the degreeof stacking order is variable. Table I lists ten tron microscopy. However, subtle variations (e.g., in Al octahedralcation sequences.More complex sequencescan content) may have gone undetected. The long-period oc- be considered as based on the group D pattern of alter- tahedral cation sequences and ordered stacking might have BANFIELD ET AL.: SERPENTINEPOLYTYPISM AND DEFORMATION rt27

Fig. 14. Continued rr28 BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION

Fig. 16. Diagramillustrating the polytypic structure of polyg- onizedserpentine as inferred from SAEDpatterns.

been establishedat the time of growth. Alternatively, the structures might have been modified subsequentto crys- tallization.

Polytypism and deformation The most obvious macroscopic characteristic of the sample is that it displays a strong lineation within the foliation plane, indicating the senseof sheardeformation. The lineation direction is approximately parallel to either a or b (and a* or b*, as B : 90) of the layer silicates. Overall, approximately equal numbers of crystals devel- op in the two orientations. This observation suggeststhat serpentine crystals were rotated during deformation or recrystallizedin a shearenvironment. Thus, polytypic di- versity might relate directly to the deformation. In a series of papers, Baronnet and coworkers (e.g., Baronnetand Amouric, 1986;Baronnet, 1989; Baronnet and Kang, 1989) discussedthe formation of long-period mica poll.types (ordered stacking) by growth controlled by screwdislocations in either a perfect or faulted matrix. Becauserelationships clearly exist between deformation, dislocations, and recrystallization, this mechanism might explain polytypic sequencesif they were created during $owth. The alternative to primary polytypic diversity involves Fig. I 5. SAED patterns ofpolygonal serpentine.(a) Adjacent modification oforiginal stacking sequencesduring defor- segmentsconsist of different polytypes. Note the oneJayer pe- plausible riod in 02l for the sector with a : 90" (see text). (b) Adjacent mation. This is in our samplesbecause the di- enantiomorphic segments(probably 6Rr). Note the alternation rection of shear corresponds to the direction in which of positive and negative c axes and that b* is common for two layers must be displaced or O planes shearedto change adjacent segments. one polytype into another. For example, for crystals that have b parallel to the shear direction, systematic inter- layer displacementsof b/3 (to optimize H bonding) could generatelong-period polytypes characterizedby regular BANFIELD ET AL.: SERPENTINE POLYTYPISM AND DEFORMATION lt29 stacking (the regular I,II alternation of the majority of Tlau 1. Serpentinepolytypes serpentineswould be unaffected).Alternatively, the oc- Disorderedb/3 tahedral slant (i.e., I vs. II) could be reversedby shearof disolacements Orderedstacking the OH planes in crystals with a parallel to the lineation One layer (group C) (shearcould also createa/3 stacking disorder, a possibil- Two layer(l,ll; group D) 2T ity we were not able to rule out). Complex sequencesof Threelayer (l,l,ll) 2H, Is and IIs might be generated occasionally periodic Four layer(l,l,ll,ll) 2H" by Fourlayer (l,l,l,ll) 6F twinning or shear. This implies that macroscopic shear Fivelayer (l,ll,l,ll,l) 6R, can, in some circumstances,be expressedmicroscopically Six layer(ll,ll,ll,l,l,l) 3Tc, asperiodic -bl3 Sevenlayer (l,l l,l,ll,l,ll,l) 4M' displacements,e.g., displacement of ev- Sevenlayer (l,ll,l,ll,l,l,l) 4Tc. ery fourth layer (0,0,0,-). The detailed mechanism is not Nine layer not determined 4Tc understood. It should be noted that an efficient mecha- 4Tq 4Td nism for generation of long-period polytypes is not re- quired becausemost crystals in the sample have disor- A/ofe.-Group information refers to the group assigned to the standard polytypes by Bailey (1969). For polytypes with disordered b/3 displac+ dered stacking,and long-period octahedral sequencesare ments between adjacent layers, intormationin parenthesesgives the se- rare compared with twoJayer structures. quence of octahedralcations, if known. For polytypeswith ordered stack- At this time it is not known how common long-period ing, information in parentheses indicates the stacking sequence and sequenceof octahedralcations. In all cases, B : 90 (however,o may vary serpentinesare. Most previous polytypic characterization for three- and fourlayer triclinic structures with regular stacking). involved powder X-ray diffraction. Thus, it is possible that long-period serpentinestructures are present as mi- nor constituents of many assemblagesbut have been structures and planar microstructures suggestthat if oc- overlooked. This may be resolved by more routine poly- tahedral sequencesresulted from shearofOH planes,the typic characterization involving interpretation of SAED process operated only within triangular-shaped subdo- patterns using proceduresdescribed above. New data are mains. required to test our hypothesis that more unusual and complex polytypes occur in deformed rocks. Polygonal serpentine The development of polygonal serpentine in regions Planar defectsin serpentine that crosscut serpentine-chloriteand at $ain boundaries Faults are common microstructures in the planar ser- betweenadjacent planar serpentinecrystals indicates that pentine (particularly group D serpentine), and they ap- the polygonal sectorsdeveloped at alate stagein the pe- parently predate chloritization. These defects share the ridotite alteration history. On the basis of angular rela- orientation of antigorite-type reversalsreported in lizar- tionships in SAED patterns, it can be deduced that if dite by Livi and Veblen (1987). However, faults do not polygons were complete they would have 15 sectors,in closely resembleantigorite offsetsimaged at high resolu- accordance with previous observations of this mineral tion by Otten (1993) or shown in Figure 12. In some (e.g.,Jiang and Liu, 1984; Mellini, 1986; Mitchell and regions, faults terminate without any evidence in lattice Putnis, 1988; Yada and Liu, 1987; Baronnetet al., 1994). fringe images for structural perturbation at the implied However, unlike most previous occurrences,polygons in octahedral-octahedral and tetrahedral-tetrahedral (e.g., our specimenshave no open core. Direct restructuring of OT-TO or TO-OT) surfaces.Although we have not es- preexistingcurved lizardite, such as that reported by Veb- tablished that faults are regions where the polarity of the len and Buseck (1979), may have occurred. Polygonal serpentinelayers reverses,the much improved match be- units appear to have nucleated at steps on serpentine- tween observed and calculated [010] SAED patterns if chlorite surfaces, sometimes forming bridges between 180'twinning is invoked does support this interpretation. layers that terminate inclined crystals. Two-dimensional [010] lattice fringe images indicate Baronnet et al. (1994) predicted sectorpolytypes based that faults representboundaries that separateserpentine on the distribution of 15 partial dislocations (b/3) be- regions in which the octahedral sequenceis out ofphase tween the sectors,with resulting b/3 shifts between suc- (e.g.,I,[ from II,!. Displacement of OH planes parallel cessivesectors around the fiber. The combination of 6R, to a within one region could createbounding faults nor- and 2H, sectors is predicted by Baronnet et al. (1994), mal to the three a* directions, resulting in triangular- but the 6R, and 2H, combination is not. However, if the shapedslabs ofserpentine in which the octahedral slants second case involves group B polytypes, the observed are out of phasewith surrounding areas(e.g., see Fig. l0). combination is consistentwith the Baronnet et al. (1994) Such boundaries are also expectedto be associatedwith model. In general, our data support the view that the screw dislocations where the Burgers vector has a mag- stacking sequenceresulted from polygonization of chrys- nitude equivalent to one layer and the octahedral peri- otile. odicity is > I layer. However, in a shear environment it seemsimprobable that abundant screwdislocations would AcrNowr,BocMENTs develop normal to the shear direction (with the fault-cut Thanks are extended to Scott Sitzman for assistancewith sample prep- planes in three orientations). Together, the polytypic aration and to Laurel Goodwin and Steven Ralser for helpful discussion. I 130 BANFIELD ET AL,: SERPENTINE POLYTYPISM AND DEFORMATION

This research was supported by NSF grants EAR-91 17386 and EAR- Hall, S.H., Guggenheim,S., Moore, P., and Bailey, S.W. (1976)The struc- 9317082 to J.F.B. and a grant from the Department ofGeology and Geo- ture of Unst-type 6Jayer serpentines.Canadian Mineralogist, 14, 314- physics, University of Wisconsin-Madison to S.W.B. We thank Alain 32t. Baronnet and Huifang Xu for their careful reviews of this manuscript. Hanan, B.B., and Sinha, A.K. (1989) Petrology and tectonic affinity of the Baltimore Mafic Complex, Maryland. In S.K. Mittwede and E.F. RrrpnrNcts crrED Stoddard, Eds., Ultramafic rocks of the Appalachian Piedmont. Geo- logical Society ofAmerica SpecialPaper, 231, l-18. Amouric, M., and Baronnet, A. (1983) Effect of early nucleation condi- Hayes,J.B. (1970) Polftypism ofchlorite in sedimentaryrocks. Clays and tions on synthetic polytypism as seen by high resolution Clay Minerals,18. 285-306. transmission electron microscopy. Physicsand Chemistry of Minerals, Jiang, S., and Liu, W. (1984) Discovery and its significanc.eof Povlen- 9,146-159. type hydrochrysotile.Acta GeologicaSinica, 58, 136-142. Bailey, S.W. (1969) Polytypism oftrioctahedral l:1 layer silicates.Clays Karpova, G.V. (1969) post-sedimentaryranks in terrigenous and Clay Minerals,17,355-371. rocks. Sedimentology,13, 5-20. -(1988a) Polytypism of l:1 layer silicates.In Mineralogical Society Krstanovi6, I., and Pavlovi6, S. (1964) X-ray study of chrysotile. Amer- of America Reviews in Mineralogy, 19, 725. ican Mineralogist, 49, l7 69-177 1. -(1988b) X-ray diffraction identification of the polltypes of mica, Iapham, D.M. (1958) Preliminary report on the chromite occurrenceat serpentine,and chlorite. Clays and Clay Minerals, 36,193-213. the Wood Mine, Pennsylvania. Pennsylvania Geological Survey Prog- Bailey, S.W., and Banfield, J.F. (1995) Derivation and identincation of ressReport,153,1lp. nonstandard serpentinepolytypes. American Mineralogist, 80, I 104- Livi, K.J.T., and Veblen, D.R. (1987) "Eastonite" from Easton,Pennsyl- 1115. vania: A mixture ofphlogopite and a new form ofserpentine.American Bailey, S.W., Banfield, J.F., Barker, W.W., and Katchan, G. (1995) Do- Mineralogist, 72, 113-125. zyrte,a l:1 regular interstratification of serpentineand chlorite. Amer- McKague, H.L. (1964) The geology,mineralogy, petrology, and geochem- ican Mineralogist, 80, 65-7 7. istry of the State Line Serpentinite and associatedchromite deposits, Banfield, J.F., and Bailey, S.W. (1996) Formation of regularly interstra- 166 p. Ph.D. thesis, PennsylvaniaState University, University Park, tified serpentine-chlorite minerals by tetrahedral inversion in long-pe- Pennsylvania riod serpentinepolytypes. American Mineralogist, 81, in press. Mellini, M. (1986) Chrysotile and polygonal serpentinefrom the Balan- Banfield, J.F., Bailey, S.W., and Barker, W.W. (1994) Polysomatism,po- gero serpentinite.Mineralogical Magezils, 50, 301-305. lytypism, microstructures and reaction mechanisms in regularly and Mitchetl, R.H., and Putnis, A. (1988) Polygonalserpentine in segregation- randomly interstratified serpentine and chlorite. Contributions to Min- textured kimberlite. Canadian Mineralogist, 26, 99 l-997 . eralogyand Petrology,I 17, 137-150. Newnham, R.E. (1961) A refinement of the structure and some Baronnet, A. (1975) Growth spirals and complex polytypism in micas: L remarks on polytypism in minerals. Mineralogical Magazine,32, 683- Pol1.t1pestructure generation.Acta Crystallographica,^A3 1, 345-3 55. 704. -(1989) gowth Polytypism and crystal of inorganic crysrals.In H. O'Hanley, D.S. (1991) Fault-related phenomenaassociated with hydra- growth Arend and J. Hulliger, Eds., Crystal in scienceand technology, tion and serpentine recrystallization during serpentinization. Canadian p. 197-204. Plenum, New York. Mineralogist, 29, 2l-35. po- Baronnet, A., and Amouric, M. (1986) Growth spirals and complex Otten, M.T. (1993) High-resolution transmission electron microscopy of lytypism in micas: II. Occurrence frequencies in synthetic species.Bul- polysomatism and stacking defects in antigorite. American Mineralo- letin Min6ral, 109,489-508. gist,78, 75-E4. polytypes. Baronnet, A., and Kang, Z.C. (1989) About the origin of mica Pauling, L. (1930) The structure ofchlorites. Proceedingsofthe National -493. PhaseTransitions, 16/17 , 477 AcademyofScience, 16, 578-582. (199a) polygonal Baronnet, A., Mellini, M., and Devouard, B. Sectorsin Shaw, H.F., and Wasserburg,G.J. (1984) Isotopic constraintson the or- serpentine:A model basedon dislocations.Physics and Chemistry of igin of Appalachian mafic complexes. American Journal of Science, Minerals. 21.330-343. 284.319-349. Bigi, S., and Brigatti, M.F. (1994) Crystal chemistry and microstructures Shirozu, H. (1963) Structural changesofsome chlorites by grinding. Min- plutonic of biotite. American Mineralogist, 79,63-72. eralogicalJournal (Japan),4, I-ll. polytypism: Brown, B. E., and Bailey, S.W. (1962) Chlorite I. Regularand Spinnler, G.E., Self, P.G., Iijima, S., and Buseck, P.R. (1984) Stacking semi-random one-layer structures. American Mineralogist, 47, 819- disorder in chinochlore chlorite. American Mineralogist, 69, 252-263. 850. Stadelmann,P. (1991) Simulation of HREM imagesand 2D CBED pat- polygonal Chisholm, J.E. (1992) The number of sectorsin serpentine. tems using EMS softwarepackage. Software manual I2M-EPFL, Lau- Canadian Mineralogist, 30, 355-365. sanne,Switzerland. Cressey,B.A. (1979) Electron rnicroscopestudies of serpentinetextures. Steadman,R. (1964) The structure oftrioctahedral kaolin-type silicates. Canadian Mineralogist, 17, 74l-7 56. Acta Crystallographica,17, 924-927. Cressey,B.A., and Zussman, (1976) Electron microscopic studies of J. steadman, R., and Nuttall, P.M. (1962) The crystal structure of amesite. | 4, -3 13. serpentinites.Canadian Mineralogist, 307 Acta Crystallographica,15, 5 l0-51 l. Curtis, C.D., Hughes, C.R., Whiteman, J.A., and Whittle, C.K. (1985) Veblen, D.R., and Buseck,P.R. (1979) Serpentineminerals: Intergowths Compositional variation within some sedimentary chlorites and some and new combination structures.Science, 206, 1398-1400. comments on their origin. Mineralogical Magazine, 49,375-386. Velde, B. (1965) Experimental determination of muscovite polymorph de Caritat, P., Hutcheon, I., and Walshe, J.L. (1993) Chlorite geother- stabilities. American Mineralogist, 50, 436-449. mometry: A review. Clays and Clay Minerals, 41,219-239. -(1993) polltype geothermometry. Clays and Clay Min- Dornberger-Schifi K., and Durovi6, S. (1975a) OD-interpretation ofka- Chlorite 41. 260-267. olinite-type structures: I. Symmetry of packets and their erafs. stacking possibilities.Clays and Clay Minerals, 23,219-229. Walker, J.R., and Thompson, G.R. (1990) Structuralvariations in chlorite - (1975b) OD-interpretation of kaolinite-type structues: II. The reg- and in a diagenetic sequencefrom the Imperial Valley, California. ular polytypes (MDO-pollqpes) and their derivation. Clays and Clay Claysand Clay Minerals,38, 3 15-321. Minerals, 23,231-246. Wicks, F.J., and O'Hanley, D.S. (1988) Serpentineminerals: Structures Durovi6, S., Miklos, D., and Dornberger-Schiff,K. (1981) Polytypism of and petrology. In Mineralogical Society of America Reviews in Min- kaolinite-type minerals:An aid to visualize the stackingoflayers. Crys- eralogy,19, 9l-159. tal ResearchTechniques, 16, 557-565. Wicks, F.J., and Whittaker, E.J.w. (1977) Serpentinetextures and ser- Glass, J.J., Vlisidis, A.C., and Pearre, N.C. (1959) Chromian antigorite pentinization. Canadian Mineralogist, I 5, 459-48 8. from the Wood's Mine, Lancaster County, Pennsylvania. American Yada, K., and Liu, W. (1987) Polygonal microstructure of Povlen chrys- Mineralogist, 44, 651-656. otile observed by high resolution electron microscopy. Proceeding of BANFIELD ET AL.: SERPENTINEPOLYTYPISM AND DEFORMATION I l3l

the Sixth Meeting ofthe European Clay Group, Seville, Spain, 596- Zvyag1n,B.B., Mishchenko, K.S., and Shitov, V.A. (1966) Ordered and 697. disordered polymorphic varieties of serpentine minerals and their di- Yoder, H.S., and Eugster,H.P. (1955) Synthetic and natural . agnosis.Soviet Physics-Crystallography,10, 539-546 [translatedfrom Geochimica et CosmochimicaActa, 8,225-280. Kristallografiya, l0(5), 635-643, 19651. bryagin, B.B. (1962) Polymorphism of double-layer miner.alsof the ka- olinite rype. Soviet Physics-Crystallogaphy, 7, 38-5 1 [translated from Kristaflografiya, 7(l), 51-65, 19621. - (1967) Electron difraction analysis of clay mineral structures, 364 Mer.ruscnrrr REcETvEDAucusr 4, 1994 p. Plenum, New York. M.lNuscmpr AcIEPTEDJur,v 13, 1995