Transmission electron microscopy of chloritoid ...tural equilibrium. The intermediate episode of...

American Mineralogist, Volume 74, pages 549-564, 1989

Transmission electron microscopy of chloritoid: Intergrowth with sheet silicates andreactions in metapelites

Jrr,r.r.lN F. BaNrrnr,oDepartment of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, Maryland 21218, U.S.A.

P,q.uL K-q.RAnrNosDepartment of Geology, Williams College, Williamstown, Massachusetts 01267, U.S.A.

D.lvro R. Vrsr-rNDepartment of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, Maryland 21218, U.S.A.

AssrRAcr

Chloritoid specimens with contrasting metamorphic histories have been studied withhigh-resolution and analytical transmission electron microscopy. The observations showthat chloritoid, an orthosilicate, can intergrow coherently with several sheet silicates andthat a wide variation in microstructures can result from different metamorphic histories.

Chloritoid crystals in garnet- and staurolite-grade, aluminous metapelites from the GreenMountains massif near Jamaica, Vermont, experienced a complex polymetamorphic his-tory including both prograde and strong retrograde events. They contain narrow lamellaeof chlorite (generally 8.5 to 20 nm wide), stilpnomelane (generally 3.6 1o 14.4 nm wide),as well as larger chlorite, paragonite, and berthierine lamellae (80 to 400 nm wide). Theintergrown minerals exhibit stacking disorder and mixed layering. In many cases, almostperfect intergrowth on (001) is observed, with structural continuity across many of thegrain boundaries. In such intergrowths, b of chloritoid (C2/c) is parallel to a of chlorite(C2/m) and b of stilpnomelane. All of the minerals can intergrow readily on (001) becauseof the dimensional similarity between the unpolymerized tetrahedral layer of chloritoidand the tetrahedral sheets of the other minerals. Although the intergrowth textures do notprovide conclusive evidence, it is suggested that these coexisting layer structures formedby retrograde replacement of the chloritoid, rather than by parallel growth or progradereaction ofchlorite to form chloritoid. The nucleation ofarrays ofvery narrow chloritelamellae throughout the chloritoid may have resulted from supersaturation with respectto Mg during retrograde metamorphism.

In contrast to the higher-grade Green Mountains samples, chloritoids from the TaconicRange, Vermont, grew at the peak of a single prograde metamorphic event and are tex-turally simple, with no intergrown sheet silicates. Although much of the chloritoid nu-cleated and grew in a muscovite, paragonite, and chlorite matrix, it did not replace thesheet silicates by direct topotactic reaction. The reaction that defines the chloritoid isogradin the Taconic Range occurs by a dissolution and reprecipitation mechanism.

INrnooucrroN

Although seldom abundant, chloritoid is an importantindex mineral in aluminous, Fe-rich metamorphic rocks,its first appearance in some cases being mapped as anisograd. The structure of chloritoid consists of two dis-tinct types of octahedral sheets that are bonded togetherby layers of isolated silicate tetrahedra. Thus, it is anorthosilicate (island silicate), but the octahedral sheets aresimilar to those of sheet silicates. As a result. chloritoidexhibits physical properties commonly associated withthe latter group, such as perfect cleavage, and structuralproperties such as ordered polytypes (Halferdahl, 1961)and stacking disorder (Jefferson and Thomas, 1978). Thecrystal chemistry of chloritoid has been reviewed in detailby Ribbe (1982).

0003404x/89/05064549$02.00

Although their structures are related, the intergrowthof chloritoid with sheet silicates in crystallographicallycontrolled orientations has been reported only recently(Karabinos, I 9 8 I ) and has been examined only with stan-dard petrographic and analytical techniques. In the pres-ent work, we have used high-resolution transmissionelectron microscopy (rnrpu), X-ray analytical electronmicroscopy (arvr), and selected-area electron diffraction(saro) to study chloritoid from two different regions inVermont having contrasting metamorphic histories. Theseobservations clarify the crystal-chemical principles in-volved in the fine-scale intergrowth of chloritoid withother minerals. In addition, they shed some light on themechanisms of reactions involving chloritoid, includingthe isograd-defining reaction that results in the first ap-

549

550 BANFIELD ET AL.: CHLORITOID INTERGROWTHS

*CHLORITE - CHLORITOIDCHLORITE + CHLORITOID-cABNET

450-4750C

GARNET -CHLORITE + CHLORITOID*CHLORITOID-CHLORITE (?)

CHLORITE

SpncrlrnN DEscRrprroN AND ExpERTMENTAL

METHODS

Samples of chloritoid-bearing rocks used in this studyare from two areas with contrasting geologic histories.The first set of samples (specimen numbers 625,333, and561 of Karabinos, l98l) is representative of Hoosac For-mation schists from the east side of the Green Mountainsmassif near Jamaica, Vermont, which have been subject-ed to multiple episodes of deformation and regionalmetamorphism (Fig. l). During an early episode of de-formation, these rocks reached the garnet grade of meta-morphism. This prograde event was followed by wide-spread and intense retrogression, which led to partialresorption ofgarnet and produced some retrograde chlo-rite (Karabinos, 198 l, 1984). A later deformation wasaccompanied by a second stage ofgarnet growth, whichresulted in textural unconformities described by Rosen-feld (1968) and Karabinos (1984). These textures wereillustrated by Karabinos (1984, Fig. 9). Some of the Ja-maica rocks reached staurolite grade during this secondprograde event. Medium-pressure facies-series condi-tions have been estimated for these rocks (Karabinos andLaird, 1988). Temperature estimates based on the min-eral assemblages and results ofSpear and Cheney (1989),are included in Figure l.

The Green Mountains schists we have investigatedcontain chloritoid and chlorite, coarse-grained musco-vite, fine-grained paragonite, quartz, centimeter-sizedgarnet, epidote, apatite, rutile, and tourmaline; stauroliteis present in some higher-grade samples (Karabinos,l98l). The chloritoid crystals are subhedral and generallybetween 0.2 and 0.3 mm in longest dimension. Typically,both chloritoid and chlorite grains are elongate and par-allel to the dominant foliation, although some randomlyoriented grains of retrograde chlorite also are present.

lst progradeevent

Ret rog rade

event

2nd prograde

e v e n tCHLORITOID - GARNET

+ CHLORITE + STAUROLITE

475-5000C

Fig. l. Simplified compilation of the mineral reactions andpeak metamorphic temperatures determined for the Hoosac For-mation schists, Jamaica, Vermont (Karabinos, 1981). Asterisksindicate reactions discussed in the text.

pearance of chloritoid during prograde metamorphism ofaluminous, Fe-rich pelitic rocks.

Observations using Hnreivr have repeatedly demon-strated that mineral intergrowths recognized in the petro-graphic microscope can occur on a very fine scale, downto the unit-cell level. In addition, such observations,combined with other geologic information, commonly canbe used to infer the processes that result in such inter-growths. In sheet silicates, intergrowths can arise fromexsolution (Veblen, 1983a) or primary growh processes(Iijima andZhu, 1982). However, the fine-scale interlay-ering of sheet silicates most commonly has been attrib-uted to replacement reactions that may result fromretrograde metamorphism, hydrothermal alteration,weathering, diagenesis, or other processes (e.g., Veblenand Ferry, 1983; Banfield and Eggleton, 1988; Ahn andPeacor, 1986).

Fig. 2. Bright-field transmission electron micrograph of narrow chlorite lamellae (indicated by arrows) in chloritoid from Ja-maica, Vermont.

BANFIELD ET AL.: CHLORITOID INTERGROWTHS

Fig. 3. Large chlorite crystals with variable orientations in a channel in chloritoid.

5 5 1

Karabinos (1981) has shown that during each simulta-neous prograde metamorphic and deformational event,these rocks remained approximately in chemical and tex-tural equilibrium. The intermediate episode of retrogrademetamorphic mineral growth occurred under static meta-morphic conditions.

Samples 2l 12 and 2 I I 3 were collected along VermontRoute 4 west of Rutland, very close to the chloritoid iso-grad in the Taconic Range, as mapped by Zen (1960,1967). These low-grade metapelites are thus interpretedto represent the first growth of chloritoid resulting froma single, prograde regional metamorphic event. As de-scribed previously by Zen ( I 960), these fine-grained phyl-lites contain muscovite, paragonite, quartz, chlorite,chloritoid, rutile, and hematite. In thin section, chloritoidcrystals up to approximately 20 pm long can be observed,but rnu study shows that most chloritoid crystals in theserocks are smaller than I prm.

Thin foils were prepared for electron microscopy byion milling 2.3-mm discs removed from petrographic thinsections. Specimens were lightly coated with C and ex-amined in a Philips 420T transmission electron micro-scope operated al 120 keV. Images were obtained fromthin areas at a tarrge of underfocus values around -80

nm. Interpretation of HRrEM images from the sheet sili-

cates follows that of previous papers (e.g., Spinnler et al.,1984; Veblen, 1983a, 1983b; Eesleton and Banfield, 1985;Jefferson and Thomas, 1978; Iijima and Buseck, 1978;Amouric et al., 1981). Because image simulations forchloritoid have not been published previously, we per-formed calculations of images as a function of crystalthickness and microscope defocus in order to interpretour chloritoid images correctly. The results from thesesimulations are outlined in Appendix 1. The identity ofminerals was confirmed using seno and lnr"r. The X-rayanalyses were obtained with an EDAx energy-dispersivespectrometer and Princeton Gamma-Tech System IVanalyzer. aeu procedures follow those described by Liviand Veblen (1987).

Crrr,onrrorn MTCRoSTRUCTURES

Green Mountains samples

The subhedral chloritoid crystals from the Jamaica areacontain abundant, narrow lamellae of chlorite and stilp-nomelane. These sheet silicate lamellae occupy about 30loof the crystals, are typically -8 to 20 nm wide, and arein crystallographic continuity with the chloritoid (Fig. 2).Occasional larger lamellae of chlorite and a 1: I layer sil-icate mineral compositionally consistent with berthierine

,:*

ch lor i te


Fig. 4. nnrnu image of intergrown 2M, and 1?"c chloritoid polytypes. The apparently strain-free termination of a packet of 1Tcinthe 2M, chloritoid is illustrated, and part of a chlorite lamella is shown at the right.

II

i

(up to 400 nm wide), as well as arrays of semirandomlyoriented chlorite, paragonite, and muscovite crystals, haveapparently developed in channels within chloritoid orformed at low-angle grain boundaries (Fig. 3). Electron-difraction patterns from chlorite, stilpnomelane, andberthierine indicate that the lamellae possess varying de-grees of stacking disorder.

Electron micrographs and saro patterns from chlori-toid reveal the presence of both the 2M, and 1Zc poly-types. Two-layer (monoclinic) chloritoid is generally themost abundant polytype in the specimens studied. Planardefects produced by variations in layer-stacking sequenceare similar to those reported by Jefferson and Thomas(1978). Terminating stacking defects (Fig. 4), where pack-ets of the 1Ic polytype pass directly into the twoJayerpolytype, occur without observable strain contrast, re-sembling terminations associated with the 2M, and 2M,polytypes reported by Jefferson and Thomas (1978).

Chlorite lamellae. The narrow lamellae of sheet sili-cates in chloritoid occur in precise, crystallographicallycontrolled orientations. Electron-diffraction patterns ofintergrown chlorite and chloritoid (Fig. 5) show that a ofchloritoid is parallel to b of chlorite, and b of chloritoidis parallel to a of chlorite. The c* axes of chlorite andchloritoid, and hence their (001) planes, also are exactlyparallel. Thus, the chloritoid/chlorite intergrowths can bedescribed as topotactic.

Electron micrographs of narrow chlorite lamellae ter-minating in chloritoid suggest that intergrowth of these

minerals occurs without appreciable strain of either struc-ture; assuming that replacement occurred, it did so byconserving volume (Fig. 6). The interfaces illustrated inFigure 6 are generally sharp, and there are ledges in the(001) boundaries. This texture is similar to those ob-served in exsolution and replacement reactions involvingchain silicates (Vander Sande and Kohlstedt. 1974:Champness and Lorimer, 197 6; Veblen and Buseck,l98l). Although the direction of the reaction is not rig-orously indicated, these observations suggest that re-placement took place in part by the nucleation and prop-agation of ledges on the lamellar interfaces.

The possibility that some structural elements are in-herited across non-(001) interfaces is suggested by thetightness of the interfaces and the continuity of somefringes across these boundaries in Figure 6. Indeed, bothstructures possess similar octahedral sheets, so that sucha relationship also might be expected purely on structuralgrounds. Interfaces oriented at a low angle to (001) alsoshow apparent structural continuity. Low-angle bound-aries between chlorite and chloritoid, such as the bound-ary shown in Figure 7, similarly are tight, suggesting againthat there may be some bonding between the two min-erals across their interfaces, even when they are not per-fectly oriented with respect to each other. These and otherimages of non-(001) interfaces typically show the narrowdark fringes of chlorite turning into either the white orblack fringes of chloritoid. Under our imaging conditions,the narrow dark cYte fringes correspond to the bruc-

BANFIELD ET AL.: CHLORITOID INTERGROWTHS 553

Fig. 5. Electron-diffraction patterns from an intergrowth ofchlorite (1.4-nm basal spacing) and chloritoid (0.9-nm basal spacing).(a) ftOl chloritoid and Okl chlorite; (b) 0icl chloritoid and ftOl chlorite.

Fig. 6. A chlorite lamella terminating in chloritoid. There are no major gaps between the two structures, and there is apparentcontinuity of some lattice fringes across the boundary.


I IriEl r

Fig. 7. Transmission electron micrograph of a low-angle grain boundary between chloritoid and chlorite; the interface is at

approximately a right angle to the (001) planes. The 0.9-nm chloritoid fringes represent one-half of the monoclinic cell, and the

dark, nonperiodic fringes are stacking faults. The continuity of some lattice fringes suggests that there may be bonding across theinterface.

Fig. 8. The widest stilpnomelane lamella observed in chloritoid. Considerable stacking disorder is apparent in the stilpnomelane

and can best be seen by sighting along the crooked cross-fringes approximately normal to the layers.

':E;r:tiiDT

nil

1I


2.44

2.12| .oc

0.010.000.000.00

10.008.009.78

555

Tlsle 1, Average structural formulae from AEM analyses of in-tergrown chloritoid and chlorite, Jamaica, Vermont

Chloritoid

SiAIMgFe'MnNaKCa

oH*

1.994.010.371 . 6 10.030.000.000.00

10.002.008.01

- All Fe calculated as Fe,*... O and OH by stoichiometry.

iteJike sheets (Spinnler et al., 1984), and the centers ofboth the white and black fringes in chloritoid are at thepositions of the two different octahedral sheets (App. l).The nnrnvr data thus suggest that either of the two oc-tahedral sheets ofchlorite can be continuous with eithersheet ofchloritoid.

Average eeu analyses of the chloritoid and chlorite aregiven in Table l. We attempted to obtain analyses of thechloritoid with a fine spot (-10-nm beam diameter) todetermine whether interpretable zoning patterns, such assolute depletion adjacent to chlorite lamellae, could bedetected. All of the variation in chloritoid compositionwith distance from the lamellae was within the uncer-tainty associated with the analytical technique.

Stilpnomelane lamellae. In addition to chlorite, narrowlamellae of stilpnomelane were identified within thechloritoid by sneo patterns and their periodicities in nn-rnvr images. Most of these lamellae are three or four unitcells wide, but they range up to a maximum of 20 unitcells. Approximately l0o/o of the narrow lamellae ob-served in the Green Mountains chloritoids are composedat least in part of this mineral.

Streaking of stilpnomelane reflections parallel to c* inelectron-diffraction patterns from some lamellae is great-er than that expected from the shape effect for thin crys-tals. This streaking suggests that the stilpnomelane pos-sesses considerable stacking disorder. The variableorientation of cross-fringes in the HRTEM image of thewidest observed stilpnomelane lamella (Fig. 8) confirmsthe irregular stacking sequence. Like the chlorite, stilp-nomelane lamellae can terminate in chloritoid. Figure 9shows such a feature, with some strain contrast associatedwith the termination. Continuity of lattice fringes be-tween chloritoid and stilpnomelane may be indicated inthis image.

Electron-diffraction patterns from more-ordered la-mellae (Fig. l0) show that, like the chlorite, stilpnomel-ane tends to grow in an orientation crystallographicallyrelated to that ofthe host chloritoid. The D-axis electron-diffraction patterns from the intergrown minerals indicatethat c* axes of stilpnomelane and chloritoid are parallel,

Fig. 9. unrerrr image showing strain contrast associated withthe termination of a stilpnomelane lamella in chloritoid.

whereas a* of stilpnomelane is inclined to c* of chlori-toid, representing a 30' angle between the a axes of theseminerals (Fig. lOa). Figure lOb shows that [311]* ofchloritoid is almost exactly parallel to I l1]* of stilpno-melane. In real space, this indicates that the (31l)"0 and(ll1)",o planes are parallel, and the [110]- and [10].,pdirections are nearly parallel.

Berthierine lamellae. A sheet silicate with a 0.7-nmbasal spacing is also present within the chloritoid (Fig.I la); 1.4-nm 001 spots are not present in the seeo pat-terns, showing clearly that the mineral is not a chlorite.The Fe-rich, highly aluminous composition and diffrac-tion patterns are consistent with berthierine, as definedby Bailey (1980). The berthierine forms lamellae of vary-ing thickness that appear to penetrate into the chloritoidstructure. Some slightly wider layer spacings are also ob-served (Fig. I lb; see arrows). These measure about 0.9nm and possibly represent individual layers of mica (e.g.,paragonite) or talc intergrown with the berthierine. Likestilpnomelane and chlorite, the berthierine lamellae areoriented crystallographically with respect to the hostchloritoid. The c* directions ofchloritoid and berthierineare parallel, and although the structural form of berthier-


Fig. 10. SAED patterns from chloritoid-stilpnomelane inter-growths. (a) Pattern obtained with the electron beam parallel toD of both chloritoid and stilpnomelane. The c* axes of the twominerals are parallel, and a of stilpnomelane is inclined 30' tothat ofchloritoid. (b) Pattern obtained with electron beam par-allel to [110] of stilpnomelane. The [10] directions of the twominerals are nearly paralle..

Fig. 11. (a) s,,rro pattem from berthierine (0.7-nm basalspacing) intergrown with chloritoid. (b) nnrervr image of steppedberthierine lamellae penetrating into chloritoid. Arrows indicatelattice spacings that are wider than normal in berthierine (ap-proximately I nm, instead of 0.7 nm).

ine is uncertain, a chloritoid is probably parallel to bberthierine (and vice versa).

Other intergrown minerals. Paragonite and goethite alsohave been observed to form narrow, oriented lamellaeparallel to (001) in chloritoid. Like the chlorite, stilpno-melane, and berthierine lamellae, these other mineralswere identified with a combination of sero, Hnteu, andaenr methods. Although they do occur alone, paragoniteand goethite more commonly occur in hybrid lamellaewith one or more other minerals, as discussed below.

Hybrid lamellae. Both stilpnomelane and chlorite co-exist within some individual narrow lamellae in chlori-toid. This is shown in Figure 12, where one layer passesdirectly from chlorite to stilpnomelane, apparently withsome structural continuity and with only minor disrup-tion or deflection ofsurrounding layers. This micrographclearly illustrates the ease with which chloritoid, chlorite,and stilpnomelane can intergrow. Packets of a few micalayers with 0.95-nm layer widths also are locally inter-grown with chlorite in some lamellae. Similar mica layersalso form larger lamellae; their composition and basalspacing are consistent with paragonite (Figs. l3a, l3b).Goethite also forms most commonly as hybrid lamellaewith berthierine (Fig. l4).


Fig. 1l -Continued.

557

Taconic Range samples

AEM and electron diffraction were used to identifuchloritoid in the Taconic samples. Although a smallamount of chloritoid can be seen in these specimens witha petrographic microscope (see above), rnu observationsindicate that most of the chloritoid in these rocks is smallerthan the resolution of a light microscope (crystals typi-cally are 20 to 50 nm in longest dimension). These mi-nute crystals of chloritoid commonly are enclosed by oradjacent to chlorite, paragonite, or muscovite.

sAED patterns indicate that the chloritoid is predomi-nantly a two-layer polytype with some stacking disorder.It is not rigorously crystallographically oriented with re-spect to the surrounding sheet silicates. Furthermore, rnrraobservations (Fig. 15) show that these small crystals aredevoid of inclusions and lack sheet-silicate lamellae suchas those in the Green Mountains chloritoid. These tex-tures suggest that the growth of chloritoid in the Taconicphyllites did not occur by topotactic replacement of chlo-rite or another sheet silicate by chloritoid; such a mech-anism would produce oriented intergrowths and wouldprobably leave areas of sheet silicate in the chloritoid.Instead, it appears that chloritoid formed by unorientednucleation and growth, without direct inheritance ofany

structural elements from the minerals of the lower-gradeassemblage.

DrscussroN

Origin of chloritoid-sheet silicate intergrowths

As noted above, previous studies of the Jamaica regionindicate that the Green Mountains samples experienceda complex polymetamorphic history (Karabinos, 1981,1984).It is thus likely that the assemblage of chloritoid,chlorite, paragonite, stilpnomelane, and berthierine re-sulted from a combination of prograde and retrogradeevents. Although it can be difficult to determine the originof sheet-silicate intergrowths using textural evidence alone,it is clear in these samples that retrograde chlorite par-tially replaced garnet, and it is also likely that relativelylarge, semirandomly oriented chlorite laths found in frac-tures in chloritoid (e.g., Fig. 3) represent the retrogradereplacement of chloritoid. On the other hand, the narrowlamellae of chlorite in chloritoid conceivably could resultfrom several processes: (1) prograde, topotactic replace-ment of chlorite by chloritoid, which left relict lamellaeof the pre-existing chlorite (first prograde event: Fig. l);(2) simultaneous parallel nucleation and growth of bothchloritoid and chlorite; (3) topotactic growth of chlorite


Fig. 12. A hybrid lamella in chloritoid containing both stilp-nomelane and chlorite. The area indicated by the arrow showsa unit cell of stilpnomelane passing directly into a unit cell ofchlorite.

as a result of the staurolite-forming reaction (Fig. l) dur-ing the second prograde event; or (4) topotactic growthof chlorite lamellae in chloritoid during the strong retro-grade metamorphic event.

Several reactions involving replacement of chlorite, asin possibility l, have been suggested previously as mech-anisms for chloritoid formation in metapelites (e.g.,Thompson and Norton, 1968; Miyashiro, 1973, p. 205).However, the rigorous orientation relationship betweenchloritoid and sheet-silicate lamellae implies that such areaction would have to be topotactic, and we have seenthat chloritoid growth in the Taconic Range samples isnot a topotactic reaction. We believe that application ofresults from the low-grade, regionally metamorphosed anddeformed, Al-rich metapelites of the Taconic Range tothe higher grade, regionally metamorphosed and de-formed, Al-rich metapelites of the Green Mountains isvalid, although we cannot demonstrate that all aspects oftheir metamorphic history were comparable (e.g., detailsof fluid flow).

Large chlorite crystals in the matrix of the GreenMountains samples have Fe-rich compositions and tex-

ures consistent with the interpretation that they formedduring the first prograde greenschist-facies metamorphicevent. These crystals show no indication of partial re-placement by chloritoid. If chloritoid were topotacticallyreplacing chlorite, it is not clear why some crystals wouldhave been almost entirely replaced, while other, adjacentcrystals were left completely unreacted. Consequently, wesuggest that the dissolution-reprecipitation mechanismobserved at the chloritoid isograd was also responsiblefor the crystallization of chloritoid in the Green Moun-tains samples (hence, relict, topotactically oriented chlo-rite lamellae should not be observed within chloritoid).We also note that the chloritoid contains narrow lamellaeof several minerals other than chlorite, which would nothave been produced by simple replacement of homoge-neous chlorite by chloritoid. These textures are more con-sistent with the formation of narrow nuclei of chloriteand other minerals within chloritoid than they are withdirect replacement of chlorite by chloritoid.

The second possibility, that the intergrowths result fromsimultaneous, oriented growth of chloritoid and sheet sil-icates, seems unlikely for many of the same reasons. Fur-thermore, the reaction sequence outlined in Figure I in-dicates that chloritoid and chlorite are only producedsimultaneously as a consequence of destruction of garnet(and these reaction products are intimately associated withthe garnet).

The third possibility, that chlorite formed because ofbreakdown ofchloritoid during the second prograde event,is rejected on the basis that the chlorite lamellae are foundwithin chloritoid in samples where the necessary stauro-lite-forming reaction did not occur.

The fourth possibility, that the sheet-silicate lamellaeformed by replacement of chloritoid during retrogrademetamorphism, is consistent with the petrological historyof these rocks and with the observed textures. The GreenMountains samples experienced intense retrograde meta-morphism. Indeed, the chlorite-chloritoid intergrowths arebest developed in samples in which the effects of the in-termediate retrogression, described by Karabinos ( 1 984),are greatest and those of the second prograde metamor-phism are least developed. Furthermore, replacement re-actions in layer minerals commonly result in very naffowlamellae of the reaction products that occasionally ter-minate in the host, similar to the texture observed here(Veblen and Ferry, 1983; Eggleton and Banfield, 1985;Banfield and Eggleton, I 988; Maresch et al., 198 5; OlivesBaflos and Amouric, 1984).

Although we cannot unequivocally discern between thetwo most probable origins for the chlorite lamellae (thefirst prograde versus the retrograde event: see asterisks inFig. l), we argue that the evidence strongly favors theinterpretation that the narrow sheet-silicate lamellae inthe Green Mountains chloritoids resulted from topotacticreplacement during retrograde metamorphism. Further-more, we suggest that a lamellar intergrowth microstruc-ture is the predictable result of retrograde metamorphismthat proceeds without simultaneous strong deformation.

w+

BANFIELD ET AL.: CHLORITOID INTERGROWTHS 559

Nucleation of widely spaced lamellae is characteristic ofa reaction that is controlled by diffusion (Fisher, 1978),a factor that is unlikely to limit reactions that occur dur-ing dynamic, prograde metamorphism.

Chemistry of the chloritoid-sheet-silicate reactions

In occurrences where sheet-silicate intergrowths haveresulted from replacement, a reaction can be written basedon the chemistry of the reactant and product mineralsand the structural reaction mechanism as deduced fromHRTEM observations (as in replacement reactions involv-ing biotite and chlorite: Veblen and Ferry, 1983; Eggletonand Banfield, 1985; Maresch et al., 1985). In efect, theHRTEM observations allow the volume change associatedwith the reaction to be estimated, which gives the ratioof the reaction coefficients for the reactant and product

---)

Fig. 13. (a) A hybrid lamella in chloritoid (Cd) containingchlorite (Clt), paragonite (P), and beam-damaged stilpnomelane(S). (b) s.q.eo pattern from this hybrid lamella illustrating the0.95-nm basal spacing associated with paragonite, 1.4-nm repeatofchlorite, and 0.9-nm spacing from the surrounding chloritoid.


minerals. It has been observed in these studies that thechemistry of the daughter mineral and that of the parentcan be closely related, with the bulk of the material forthe product mineral being inherited from the parent andonly some components necessarily being added or sub-tracted, presumably via a fluid phase.

In the Green Mountains chloritoids, HRrEM images in-dicate that chlorite formed narrow lamellae in chloritoid,rather than producing a finely interstratified, mixed-layerchlorite-chloritoid material. Assuming that the texturaland geologic interpretation in the last section is correct,

TlaLE 2, Chemical balance for equal-volume reaction of chlori-toid to chlorite"

Fig. 14. A hybrid lamella of berthierine and goethite in chloritoid.

- Assumes that no changes in composition of the chloritoid or chloritehave occurred since replacement; see text..- Negative numbers indicate that the component is consumed bv thereaction, i.e., is removed from the fluid.

the lack of substantial strain at lamellar tips in rEM im-ages indicates that chlorite replaced approximately anequal volume of chloritoid. In the present case, an equal-volume transformation implies that almost exactly al unitcells ofone-layer chlorite replaced each unit cell ofmono-clinic (two-layer) chloritoid. (The chlorite coefficient is 7,within the error of the ratio of unit-cell volumes from ourelectron-diffraction patterns. In addition, the chlorite unit-cell volume calculated with the regression equations ofHey, 1954, is 0.693 nm3, so thal. %. V"n : 0.924 nm3; thiscompares with the choritoid unit-cell volume from Hans-com, 1975, of 0.925 nm3.)

The coefrcients for an equal-volume reaction of chlori-toid to chlorite are given in Table 2 (the chlorite coefr-cient of,l results from the fact that Z : 4 and 2 formonoclinic chloritoid and one-layer chlorite respectively,for the formula units used in Table l). Table 2 suggeststhat some exchange with metamorphic fluids is necessaryfor all major chemical components, in order for the re-action to proceed. Substantial amounts of H and Mg mustenter the reaction site and be removed from the meta-morphic fluid, whereas Si, Al, and Fe would be given upto the fluid.

The reaction in Table 2 assumes that the chloritoidbeing replaced had the typical composition given in TableI since the onset of reaction. However, the narrow la-

1 chloritoid % chlorite

SiAIMgFeMnoH

1 9 94.010.371 . 6 10.03

12.O02.00

1.622.371.411 . 1 00.01

12.O0E 2 e

0.371.64

-1.040.510.020.00

-3.33


Fig. 15. A small, randomly oriented chloritoid crystal formed by simple prograde reaction in a metapelite from the TaconicRange. The chloritoid does not contain lamellae ofa precursor mineral, although the inset SAED pattem and other images indicatethat it possesses stacking disorder. Arrows indicate the extent ofthe chloritoid crystal.

561

mellae of chloritoid and chlorite have experienced a pro-tracted regional metamorphic history, so that it may beexpected that they will have re-equilibrated following re-placement. The Fe/Mg in the chloritoid and chlorite ap-pear to preserve compositions indicating equilibrationbetween garnet and staurolite (based on the calibrationof Ashworth and Evirgen, 1984), appropriate for eitherthe early stage ofthe retrograde event, or reflecting com-positional re-equilibration during the second progradeevent. The minerals may also have changed compositionduring the reaction. Consequently Table 2 probably doesnot accurately represent the precise chemical exchangesoccurring at the time of replacement, but reflects the ef-fect ofthe initial structural replacement process, plus anysubsequent exchange reactions, involving for example,FeMg-, and FeAl ,.

It is possible that much of the chemical exchange re-quired to form the chlorite occurred between the lamellaeand the remaining chloritoid, rather than with the meta-morphic fluid. Only about 3olo by volume of the chloritoid(average Mg/[Fe + Mg] = 0.187) is occupied by chlorite.The initial Mg/[Fe + Mg] of the chloritoid need be onlyabout 0.201 for all the Mg required to form the chloriteto have been derived from chloritoid, and all the Fe re-leased to have been absorbed by chloritoid by continuousreaction. Indeed, the formation of the chlorite lamellaemight be a response to the decreasing stability of Mg-richchloritoid with decreasing temperature. Although it seemslikely that the reaction proceeded by exchange ofSi, Al,Mg, Fe, and H with both fluid and host chloritoid. it is

not possible to determine the exact sources and sinks forthe reaction. In addition, the presence ofother sheet-sil-icate lamellae, such as stilpnomelane and berthierine,suggests that an accurate description ofthe reactions wouldhave to include these solid phases, as well as the chloriteand chloritoid. These other minerals may well have actedas sinks for Si, Al, and Fe produced by the chlontoid-chlorite reaction.

As noted in the last section, it is likely that stilpno-melane lamellae, like those of chlorite, formed duringretrograde metamorphism. In any case, the fact that stilp-nomelane commonly forms hybrid lamellae with chloritesuggests that the origin of the two minerals is closely re-lated. Stilpnomelane is relatively poor in Al, and its de-velopment in intimate intergrowths with chlorite andchloritoid is rather surprising. Chloritoid and stilpno-melane typically are self-exclusive in low-grade metape-lites (e.g., Zen, 1960). Although a small three-phase fieldmay exist far to the Fe-rich side of an AFM diagram atlow grade, the interpretation of stilpnomelane * chlori-toid + chlorite as an equilibrium assemblage would im-ply equilibration within the chloritoid crystal (not withthe whole-rock bulk composition), and a chlorite com-position far more Fe-rich than that observed for theselamellae. It is possible that stilpnomelane lamellae grewmetastably, or during a separate metamorphic event byreaction of chloritoid or chlorite with Al-poor fluids. Al-though it is not possible to pinpoint the process that ledto growth of stilpnomelane, this occurrence shows thatthe assemblage chloritoid * chlorite * stilpnomelane is


attainable and demonstrates that stilpnomelane can crys-tallize in rather aluminous environments.

It is possible that the other types of lamellae in theGreen Mountains chloritoids, such as paragonite, ber-thierine, and goethite, also formed during the same retro-grade event in which chlorite formed. However, theseother minerals could well have formed at some other stagein the history of these intergrowths. Since chlorite is byfar the most abundant mineral intergrown in the chlori-toid, we have not considered the detailed chemistry ofthe reactions that produced these other lamellae.

Crystallographic controls on the intergrowths

The ability of chlorite, chloritoid, stilpnomelane, pa-ragonite, and berthierine to form fine-scale, crystallo-graphically controlled intergrowths is attributed to (l) thesimilarity of lattice periodicities normal to c* in all thesestructures and (2) the fact that they all possess two-di-mensionally infinite sheets of octahedrally coordinatedcations (and, of course, tetrahedral units that can be ar-ticulated to these sheets). On structural grounds, it is like-ly that the lamellar interfaces consist of octahedral sheetsthat are shared by two different structures.

Measurement of numerous sAED patterns shows thatinterlayering occurs predominantly with a of chlorite par-allel to b of chloritoid, and vice versa. However, twoother orientations in which the a and b axes of chloritoidare inclined to those of chlorite by +60' were also ob-served (because of pseudohexagonal symmetry in thea-b plane). Since the fits between the dimensions of chlo-ritoid and chlorite are similar in all three orientations, itis not surprising that the alternate arrangements were ob-served occasionally.

The linear misfit between a of chloritoid and b of chlo-rite is approximately 3ol0, and that between b of chloritoidand a of chlorite is about 2o/o. To apply this comparisonto chloritoid and chlorite at elevated temperatures andpressures, it is necessary to assume that these mineralshave similar thermal expansivities and compressibilities,or that the amount of change in unit-cell parameters issmall compared to the misfits. In the present case, it ap-pears that the relative changes in a and D dimensions are,in fact, negligible compared to the misfits, based on thetemperature and pressure conditions for the Jamaica rocks(Karabinos, l98l) and on thermal-expansion data forchlorite and chloritoid and the compressibilities typicalfor sheet silicates (Symmes, 1986; Ivaldi et al., 1988; Ha-zen and Finger, 1978). The detailed mechanism by whichthe misfit is accommodated is not completely under-stood.

Electron-diffraction patterns indicate that the latticesof intergrown chloritoid and stilpnomelane fit togetherexactly in the (001) plane, within the resolution of themno technique. The zone illustrated in Figure lOa con-tains a* and c* of both minerals, indicating that the stilp-nomelane and chloritoid D axes coincide (D of stilpno-melane is four times that of chloritoid). As in the case of

the chloritoid-chlorite intergrowths, the chloritoid-stilp-nomelane intergrowths can be described as topotactic.

CoNcr,usroNs

We have used high-resolution and analytical transmis-sion electron microscopy to show that chloritoid crystalsin garnet-grade schists from Jamaica, Vermont, containcomplex mixtures of intergrown sheet silicates. Thesesheet silicates form very narrow lamellae in orientationsthat are precisely controlled by the structure of the hostchloritoid. In contrast, chloritoid that formed by simpleprograde reactions in phyllites from the Taconic Range,Vermont, developed in random orientations relative tothe sheet silicates in the matrix. This chloritoid containsno relict lamellae of precursor minerals. A variety of tex-tural, chemical, and geologic ovidence suggests that theintergrowths from Jamaica result from retrograde re-placement of chloritoid by the sheet silicates.

The apparent ease with which chloritoid, chlorite, stilp-nomelane, paragonite, and berthierine can intergrow on(001) is attributed to similarities in lattice spacings par-allel to their sheets of octahedra. Continuity of latticefringes between chloritoid-chlorite, chloritoid-stilpno-melane, and chlorite-stilpnomelane may indicate that re-actions between these minerals involve partial inheri-tance of octahedral sheets. Intergrowth of all minerals inthe Jamaica samples occurred without the developmentof regular gaps or significant disruption of the structures,indicating the isovolumetric nature of the reactions.

Although the crystallography of the intergrowths is rel-atively straightforward, the detailed chemistry of the re-actions that produced the intergrown sheet silicates is noteasy to assess. It is possible that chloritoid re-equilibratedto more Fe-rich compositions stable at lower tempera-tures by the formation of chlorite lamellae that are rela-tively rich in Mg. However, the detailed chemical budgetsbetween chloritoid, the various sheet silicates, and meta-morphic fluids cannot be calculated with any certainty.It is likely that observed compositions have been alteredby exchange reactions during subsequent prograde andretrograde metamorphism.

AcrNowr.nncMENTS

This paper benefited from discussions with John M. Ferry, Robert FManin, George W. Fisher, E-an Znn, Kenneth J. T. Livi, and Eugene S.Ilton. Helpful reviews were provided by Adrian Brearley and LaurelGoodwin. The research was supported by NSF grant EAR8609277. JoLaird is thanked for editorial handling. Electron microscopy was per-formed at the Johns Hopkins high-resolution rer"r lab, which was estab-lished with partial support from NSF instrumentation gmnt EAR8300365.

RnrnnpNcrs crrnoAhn, J.H., and Peacor, D.R. (1986) Transmission and analytical electron

microscopy of diagenetic chlorite in Gulf Coast argillaceous sediments.Clays and Clay Minerals, 34, 165-179.

Amouric, M., Mercuriot, G., and Barronet, A. (1981) On computed andobserved HRTEM images of perfect mica polytypes. Bulletin de Mi-n6ralogie, 104, 298-313.

Ashworth, J.R., and Evirgen, M.M. (1984) Mineral chemistry of regional

chloritoid assemblages in the chlorite zone, Lycian nappes, south-westTurkey. Mineralogical Magazine, 48, 159-165

Bailey, S.W. (1980) Structures of layer silicates. In G.W. Brindley and G.Brown, Eds., Crystal structures of clay minerals and their X-ray iden-tification, p. l-124 Mineralogical Sociery, London.

Banfield, J.F., and Eggleton, R.A. (1988) A transmission electron micro-scope study ofbiotite weathering. Clays and Clay Minerals, 36, 47-60.

Champness, P.E., and Lorimer, G.W. (1976) Exsolution in silicates. InH.-R Wenk, Ed., Electron microscopy in mineralogy, p 174-204.Springer-Verlag, Berlin.

Cowley, J M, and Moodie, A.F (1957) The scattering of electrons byatoms and crystals, I. A new theoretical approach. Acta Crystallogra-phica, I0,609-619.

Eggleton, R.A., and Banfield, J.F. (1985) The alteration ofgranitic biotiteto chlorite. American Mineralogrst, 7 0, 902-9 10.

Fisher, G.W. (1978) Rate laws in metamorphism Geochimica et Cos-mochimica Acta. 42. 1035-1050.

Halfer{ahl, L.B. (1961) Chloritoid: Its composition, stability, and occur-rence. Joumal of Petrology, 2, 49-135.

Hanscom, R.H. (1975) Refinement of the crystal structure of monoclinicchloritoid. Acta Crystallographica, 83 l, 7 80-7 84.

Hazen, R.M., and Finger, L.W. (1978) The crystal structures and com-pressibilities oflayer minerals at high pressure. II Phlogopite and chlo-rite. American Mineralogist, 63, 293-296.

Hey, M.H. (1954) A new review of the chlorites. Mineralogical Magazine,30,277.

Iijima, S., and Buseck, P.R. (1978) Experirnental study ofdisordered micastructures by high-resolution electron microscopy. Acta Crystallogra-phica, A34,'709-719.

Iijima, S., andZhu, J. (1982) Electron microscopy of a muscovite-biotiteinterface. American Mineralogist, 67, l 19 5-1 20 5.

Ivaldi, G., Catti, M., and Ferraris, G. (1988) Crystal structure at 25 and700 'C of magnesiochloritoid from a high-pressure assemblage (MonteRosa). American Mineralogist, 7 3, 358-364.

Jefferson, D.A., and Thomas, J.M. (1978) High resolution electron mi-croscopic and X-ray studies of non-random disorder in an unusuallayered silicate (chloritoid). Proceedings of the Royal Society of Lon-don, ser ies A, 361, 399-41l .

Karabinos, P. (1981) Deformation and metamorphism of Cambrian andPrecambrian rocks on the east limb of the Green Mountains anticli-norium near Jamaica, Vermont. Ph.D. thesis, Johns Hopkins Univer-sity, Baltimore, Maryland.

-(1984) Deformation and metamorphism on the east side of theGreen Mountain massif in southem Vermont. Geological Sociery ofAmerica Bulletin. 95. 584-593.

Karabinos, P, and Laird, J. (1988) Structure and metamorphism fromJamaica to the Athens dome, Vermont. In W.A. Bothner, Ed., Guide-book for field trips in southwestern Vermont, and north-central Mas-sachusetts, p. 281-292. New England Intercollegiate Geological Con-ference, 80th Annual Meeting, 1988, Keene, New Hampshire.

Livi, K.J.T., and Veblen, D.R. (1987) "Eastonite" from Easton, Pennsyl-vania: A mixture ofphlogopite and a new form ofserpentine. AmericanMineralogist, 7 2, I 13-125.

Maresch, W.V., Massonne, H.-J., and Czank, M. (1985) Ordered anddisordered chlorite/biotite interstratifications as alteration products ofchlorite. Neues Jahrbuch fiir Mineralogie Abhandlungen, 152, 79-100.

Miyashiro, A. (1973) Metamorphism and metamorphic belts, 492 p Wi-ley, New York.

O'Keefe, M.A. (1984) Electron image simulation: A complementary pro-cessing technique. In Electron optical systems for microscopy, micro-analysis, and microlithography, p. 2O9-220. Scanning Microscopy In-ternational, AMF O'Hare, Chicago.

Olives Bafros, J., and Amouric, M. (1984) Biotite chloritization by inter-layer brucitization as seen by HRTEM. American Mineralogist, 69,869-87 l

Ribbe, P.H. (1982) Chloritoid. Mineralogical Society of America Reviewsin Mineralogy (2nd edition), 5, 155-170

Rosenfeld, J.L. (1968) Gamet rotation due to major Paleozoic deforma-tion in southeastern Vermont. lr,E. Znn, W.S. White, J.B. Hadley, andJ.B. Thompson, Jr., Eds., Studies in Appalachian geology-Northernand maritime, p. 185-202. Wiley, New York.

563

Spear, F.S., and Cheney, J.T (1989) A petrogenetic grid for pelitic schistsin the system SiOr-Al,O,-FeO-MgO-KrO-HrO. Contributions to Min-eralogy and Petrology, 101, 149-164.

Spinnler, G.E., Sel{ PG., Iijima, S., and Buseck, P.R. (1984) Stackingdisorder in clinochlore chlorite American Mineralogist, 69, 252-263.

Symmes, G.H. (1986) The thermal expansion of natural muscovite, pa-ragonite, margarite, pyrophyllite, phlogopite, and two chlorites: Thesignificance ofhigh P/?"volume studies on calculated phase equilibria.B.A. thesis, Amherst College, Amherst, Massachusetts.

Thompson, J.B., Jr., and Norton, S.A. (1968) Paleozoic regional meta-morphism in New England and adjacent areas. In E. Zen, W.S. White,J.B. Hadley, and J.B. Thompson, Jr., Eds., Studies in Appalachiangeology-Nonhern and maritime, p. 319-328. Wiley, New York.

Vander Sande, J.B., and Kohlstedt, D.L. (1 974) A high resolution electronmicroscope study of exsolution in enstatite. Philosophical Magazine,29, t04r-1049.

Veblen, D.R. (1983a) Exsolution and crystal chemistry of the sodiummica wonesite. American Mineralogist, 68, 554-565.

-(1983b) Microstructures and mixed layering in intergtown wone-site, chlorite, talc, biotite, and kaolinite. American Mineralogist, 68,566-580.

Veblen, D R, and Buseck, P.R. (1981) Hydrous pyriboles and sheet sili-cates in pyroxenes and ualites: Intergrowth microstructures and reac-tion mechanisms. American Mineralogist, 66, 1107-1134.

Veblen, D.R., and Ferry, J M. (1983) A rrrr.r study of the biotite-chloritereaction and comparison with petrologic observations. American Min-eralogist, 68, I 160-l 168.

Zen,E-an. (1960) Metamorphism of Lower Paleozoic rocks in the vicinityof the Taconic Range in west-central Vermont. American Mineralogist,45 ,129 -175 .

-(1967) Time and space relationships of the Taconic allochthon andautochthon. Geological Society ofAmerica Special Paper 97 107 p.

Meruscnrsr REcErwD Jurv 28, 1988M.qN'uscrrpr AccEprED Ja.rguenv 20, 1989

Appexorx 1. CoupurnR sTMULATToN oFCHLORITOID HRTEM IMAGES

In order to interpret the experimental chloritoid nnruvr imagesrigorously, simulated images were computed using the sHRLr setof programs (O'Keefe, 1984), version 8oF. These programs usethe multislice method of Cowley and Moodie (1957) to calculatethe eflects of dynamical diffraction in the specimen, and theoptics and operating parameters of the microscope are simulat-ed. Refined atomic parameters for 2M, chloritoid from Han-scom (1975) were used for the input structure. Images were cal-culated with the electron beam parallel to both the a and b axesfor a range of microscope defocus values and specimen thick-nesses ranging up to 14.2 nm. Microscope parameters for thesimulations were appropriate for the Philips lzor at operatingconditions used for this study, as follows: objective aperture ra-dius: 4.37 nm r, semi-angle of convergence: 0.65 mrad, chro-matic halfwidth : 15 nm, spherical aberration coefficient : 2.0mm, vibration halfwidth : 0.05 nm.

The simulations show that for our imaging conditions, a-axisimages consist of one-dimensional fringes, whereas imaging par-allel to the D axis produces more complicated, two-dimensionalimages. This is consistent with our observations and with imagespublished in the literature (e.g., Jefferson and Thomas, 1978).For images with one-dimensional fringes, such as many of thoseshown in this paper, there are only a few essentially differentimage types that occur for our experimental conditions, assum-ing that the specimen is in perfect orientation with respect to theelectron beam; examples of these image types are shown in Ap-pendix Figure 1. There is little variation in the basic image char-acter with changes in specimen thickness up to our maximum



-80n m - l lon m

App. Fig. l. Computer-simulated nnrru images of monoclinic chloritoid obtained with the electron beam parallel to the a axis.Crystal thickness is 9.5 nm. (a) Defocus A/: -80 nm. (b) Af : -l l0 nm. (c) Af : -50 nm. In schematic diagram to the left, rowsof small circles indicate (Al3O8)-? and rows of asterisks indicate [FeiAlOr(OH).]-'.

simulated thickness of 14.2 nm, and the calculated images shownin Appendix Figure I are all for a thickness of 9.5 nm.

For underfocus values around -80 nm, there are four darkand four light fringes per monoclinic unit cell, and the dark fringesare centered on every octahedral sheet (App. Fig. la). For morenegative values ofdefocus starting at about - 100 nm, there areonly two dark fringes per unit cell (App. Fig. lb); in this case,the dark fringes are centered on the relatively pure Al octahedralsheets (M2 sites), while the light fringes are centered on the rel-atively Fe-rich sheets (Ml sites). Similarly, for more positivevalues of defocus starting at about -60 nm, there also are onlytwo dark fringes per cell, but in this case they overlie the Fesheets (App. Fig. lc). Since this contrast reversal takes place overa relatively narrow range of defocus, it is not possible to saywhether the dark fringes correspond to the Al or the Fe-richoctahedral sheets, unless the microscopist has excellent controlon his/her focus conditions. However. it is possible to state that

the fringes do correspond to the positions of one of the octahe-dral sheets, and this is sufrcient information to interpret thestructural details of chloritoid-chlorite intergrowths, for example(see text).

For all one-dimensional chloritoid images that are obtainedwith the electron beam oriented perfectly parallel to the layers,the image periodicity is 0.9 nm, rather than the true 1.8-nmstructural periodicity for twoJayer, monoclinic chloritoid. Thus,information on the polytype of the chloritoid can only be ob-tained from two-dimensional, D-axis images or from one-dimen-sional images that have been recorded from crystals that areslightly out of orientation. Images exhibiting the polytypic pe-riodicity, such as that shown in Figure 4, therefore cannot beinterpreted reliably in terms of the detailed layers in the struc-ture, but they are invaluable for elucidating the variations instacking ofthe layers.

Transmission electron microscopy of chloritoid ...tural equilibrium. The intermediate episode of...

Documents

Transcript of Transmission electron microscopy of chloritoid ...tural equilibrium. The intermediate episode of...