American Mineralogist. Volume 76, pages 1973-1984, I99l

Combined freeze-etch replicas and HRTEM images as tools to study fundamentalparticles and the multiphase nature of 2zl layer silicates

Ho.r.q.ror,r,.lu Vall RnrNruno flnssBDepartment of Geological Sciences, McGill University, 3450 University Street, Montreal, Quebec }{3A 247 , Canada

Ew.q.Lo E. Konr.BnInstitut fiir Angewandte Geologie, University of Regensburg, UniversitiitsstraBe, D-8400 Regensburg, Germany

Ansrucr

Comparative studies of illite and glauconite dispersions ( < I .0 pm fraction) using trans-mission electron microscopy on freeze-etch replicas and ultrathin sections provide detailedinformation on surface morphology and layer structure of these 2:l layer silicates. Resultsshow that the arrangement of layer packets observed in ultrathin section is both a clumpingartifact caused by dehydration required for embedding and a primary structure. Adjacentpackets (40- I 50 A; of Z: t layers usually do not split along joint interfaces even in diluteand Na-saturated dispersions. The thickness of these packets imaged on ultrathin sectionsis in agreement with the size of illite "fundamental particles." However, they do notusually occur as free particles in dispersion. The thickness of individual free particlesobserved in freeze-etch replicas varies between 50 and 500, A. Thus most free particles ofillite and glauconite are composed of several packets.

Treatment of samples with octadecylammonium ions reveals three different types ofinterlayer spacings corresponding to (l) a high-charged nonexpandable component (10 Ain XRD and HRTEM), (2) a high-charged expandable component (29-30 A in XRO; Z+A n URtgVD, and (3) a low-charged expandable component as single interlayers betweenadjacent packets and as expandable components in the l: I ordered mixed layers (no XRDpeak; 16- 17 A in URtEVt). It is concluded that the illitic materials contain at least threedistinct phases: a nonexpandable illite phase, an expandable illite phase, and a l: I orderedmixedJayer phase. The IIIS or IIIIS structures identified in XRD are due to interparticlediffraction. The smectiteJike interlayers in ordered mixedJayer clays do not represent aseparate phase.

IurnonucrroN

Comparison of the surface morphology of naturally oc-curring clay particles with their detailed layer structureshas important implications for the fundamental particlehypothesis and the multiphase nature of illitic clays. Thispaper reports the results of a comparative study of thesurface morphology, layer structure, and crystallite sizeofillite and glauconite particles using freeze-etch replicasand ultramicrotome sections.

The surface morphology of clay minerals can be stud-ied in dispersed systems. For the investigation of diluteaqueous dispersions, the advanced cryofixation technique(Robards and Sleytr, 1985; Bachmann, 1987) followed byfreeze etching (Moor and Muehlethaler, 1963) has greatadvantage over conventional preparation methods suchas sample drying and impregnation. These techniques en-able the observation of detailed surface features of claymineral particles dispersed in HrO (Vali and Bachmann,1988). Since they do not dehydrate samples as occursduring conventional shadowing (Baronnet, I 972; Nadeauet al., 1984a, 1984b, 1984c), the true surface morphologyof clay minerals can be studied. Furthermore, the TEM

resolution of platinum-carbon-shadowed replicas (ap-proximately l0 A for layer silicates, Baronnet, 1972) ismuch higher than that achieved with scanning electronmicroscopy (SEM).

High-resolution transmission electron microscopy(HRTEM) allows the direct observation of illite andsmectite layers as well as their stacking sequence. How-ever, comparison with computer-simulated images showsthat determination of smectite layers in illitic material ispossible only under certain TEM focus conditions; thus,prudence is required in the interpretation of HRTEM im-ages (Guthrie and Veblen, I 989, I 990; Veblen et al., 1990:-Ahn and Peacor, 1989).

Application of the alkylammonium ion-exchangemethod oflagaly and Weiss (1969) to XRD (Lagaly,1979,1982; Rtihlicke and Kohler, l98l; Stanjek and Friedrich,1986; Laird etal.,1987; olis et al., 1990; vali, 1983) andto TEM investigations (Ahn and Peacor, 1986a; ke andPeacor, 1986; Vali and Kdster, 1986; Bell, 1986; Kli-mentidis and Mackinnon, 1986; Marcks et al., 1989;Ghabru et al., 1989) has had great impact on the char-acterization ofexpandable and nonexpandable layer sil-

0003-{04xl9 r/ l l r 2-t 973$02.00 1973

t974

icates and made HRTEM investigations of these mineralsmost profitable. Using this method, Vali and Hesse (1990)introduced a new preparation technique for HRTEM,which allows the estimation of interlayer charge in ex-pandable clay minerals on ultrathin sections.

An important aspect of clay minerals is particle thick-ness. Besides being needed for the calculation of simu-lated X-ray diffraction patterns (e.g., Reynolds, 1989;Bethke and Altaner, 1986; Inoue et al., 1989), clay par-ticle thickness is critical for the fundamental particle hy-pothesis ofNadeau et al. (1984a, 1984b, 1984c) and Na-deau (1985); this hypothesis provides a conceptualinterpretation for the difference in origin between randomand ordered smectite/illite mixed layers (S/I). To under-stand the origin ofthese interstratified clays, however, itis necessary to determine whether fundamental clay par-ticles occur commonly in nature (Mackinnon, 1987; Al-taner et al., 1988). More recently, the multiphase natureof illitic materials has been the subject of extensive dis-cussions by Altaner and Vergo (1988), Eberl et al. (1988),and Rosenberg et al. (1990), which again center on theproblem of the existence of fundamental particles for dif-ferent phases. In HRTEM images of illite and glauconite,Vali and Kdster (1986) observed successive stacks of lay-ers with varying swelling capabilities. They could not de-termine, however, whether the different packets were partsof the same crystal or whether they belonged to differentcrystals that, upon desiccation, became stacked by chance.

These problems are here further pursued with a com-bination of newly developed methods for the study of thesurface morphology and ultrafine structure of 2: I layersilicates.

Terminology

The terms for various components of layer silicatestructures used in this paper are applied in the sense ofBrown (1984). In addition, the term "packet" is used todescribe a coherent sequence of 2:l layers in ultrathinsection. The term "free particle" is used for a particleobserved in freeze-etch replica. In colloid chemistry, theterm "free particle" applies to particles that, in disper-sion, behave kinetically independently (Hiemenz, 1986).The cooling rate of the applied cryo-technique is highenough to prevent phase separation and to preserve therandom distribution ofparticles in dispersion because theice crystals formed remain significantly smaller than theobjects to be studied (Vali and Bachmann, 1988). There-fore the particles shown in freeze-etch replicas representfree particles in dispersion in the sense of the colloidchemist's definition. The term is used as an equiv-alent of packets for lath-shaped and platy morphologicalentities, which are visible in freeze-etch replicas on thesurface ofa free particle as coherent surface elements sur-rounded by steps or furrows. A free particle can be com-posed of numerous units. "Particle" is the general termthat applies to all types of particles occurring in bothfreeze-etch replicas and ultrathin sections. "Fundamentalparticle" is a term introduced by Nadeau et al. (1984b,

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

1984c) and Nadeau (1985) for the smallest units of smec-tite and illite in mixedJayer clays. Fundamental particlesof smectite consist of single 2:l layers t0 A thick. Fun-damental particles of illite consist of two or more suchlayers which are 20,30, or up to 90 A thick, all coordi-nated by planes ofK interlayer ions. A "phase" is "a stateof matter that is uniform throughout, not only in chem-ical composition but also physical state" (Gibbs, quotedin Atkins, 1978; Eberl et al., 1988). In the following, weuse the acronym ODAT as an abbreviation for octade-cylammonium treatment or octadecylammonium treated(R. Reynolds, Jr., personal communication).

Ml.rnnrar. AND METHoDS

Samples

Illite from Fiizerradvany, Hungary, which formed byhydrothermal alteration of feldspar (Nemez, 1973); glau-conite from Bichl, Bavaria, which formed from gelJikehydroxide precipitates and amorphous biogenic silica un-der reducing conditions (Kohler and Kdster, 1976); andan epimetamorphic slate from the northern Appalachiansin Quebec, Canada, were selected for this study. The sam-ples were treated with a 0.1 M solution of di-sodium-ethylenediamine-tetracetic acid (EDTA-Nar) three timesfor 24 h at room temperature for carbonate removal andNa saturation; they were then washed with distilled HrOuntil all excess EDTA Na, was removed (K0ster et al.,1973). After dispersion, the fraction of the samples thatwas < 1.0 p was separated with a centrifuge.

Ultrathin sectioning

Fifty milligrams of freeze-dried sample material wereembedded in I mL of a low-viscosity resin (Spurr, 1969)for HRTEM study (Lee and Jackson, 197 S;Yali and Kds-ter, 1986). To preserve the uniform distribution of theclay mineral particles, the samples were carefully stirreduntil the resin became highly viscous. Ultrathin sectionswere then prepared using an ultramicrotome (Reichert-Jung Ultracut E). For the examination of expandability,the samples were treated with octadecylammonium ionsbefore resin impregnation for 2 d at 60 "C, following themethod of Lagaly and Weiss (1969) and Vali and Kdster(1986). The sample from Quebec was treated after sectionpreparation (Vali and Hesse, 1990). The ultrathin sec-tions were studied with a JEOL 100 CX transmissionelectron microscope (TEM) at 100 kV. Focus conditionsapproximated the Scherzer defocus, depending on thethickness ofthe sections, which varied between 500 and700 A.

Cryofixation and freeze etching

For cryofixation and freeze etching, the samples werefrozen following the spray method of Bachmann andSchmitt (1971). Suspensions of 590 illite and glauconitewere prepared using different electrolytes. To obtain com-plete dispersion of the particles, the Na-saturated 5olo sus-pensions were prepared at least I week before investiga-

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES t97 5

tion and were ultrasonicated periodically (two times perday for 3 min). Approximately 300 pL of the suspensionwere sprayed with a commercial retouching air brush intoa propane-propylene mixture (temperature of 77 K or-196 oc) from a distance of 20 cm at a pressure of 1.5kp,/cm'z. The perfectly dispersed particles were frozenwithout cryo-artifacts at a cooling rate of 105 Ws in drop-lets of fluid 10-50 pm in diameter. The liquid propane-propylene mixture, which served as a cryomedium, wasevaporated in a preparation box under slightly reducedpressure at 187 K (-86 "C). The frozen clay suspensionwas then mixed with one or two drops of ethyl benzeneat 187 K and transferred onto a Au specimen holder. Inthe freeze-etch unit (Balzer's model BAF 400T), the sam-ples were fractured in a vacuum of 10-6 torr at a temper-ature of 173 K (-100 "C). The fractured surfaces wereetched by ice sublimation for I min at 173 K and l0 6torr. They were then platinum-carbon shadowed at anangle of 45" and coated with a C layer 100 A thick forstabilization. A more detailed description of the processis given by Moor and Muehlethaler (1963) and Robardsand Sleytr (1985). The replica was floated in distilled H,Oand kept for 24 h in 60lo HF to dissolve adhering silicateparticles. Replicas were imaged with a JEOL 100 CX at1OO KV TEM.

Rnsur,rsXRD patterns of ethylene-glycol-saturated Hungary il-

lite of this study indicated the presence of R3-orderedmixed layers with about 100/o smectiteJike expandablelayers, estimated by comparison with the patterns ofReynolds and Hower (1970) and Srodori (1984). Beforeglycolation, the (001) reflection of Hungary illite gave ad value of 10.3 A, which after glycolation split into a 9.8A and a I 1.0-l 1.6 A reflection (Fig. l, sections a and b).The ll.6- and 9.8-A peaks indicate R3-ordered mixedlayers. The XRD pattern of this illite (Fig. l, section b)is similar to that of the Kalkberg bentonite (Fig. 2 inReynolds and Hower, 1970) and the Devonian bentonite(Fig. 6 in Nadeau et al., 1984b). After glycolation, the10.2-A peak ofthe Bavaria glauconite changed to 10.0 Aand broadened (Fig. l, sections c and d). The XRD pat-tern of illite from Quebec did not change upon glycola-tion (Fig. l, sections e and f). This sample also containedabout 500/o chlorite. Figure 2, sections a-i show X-raydiffraction patterns of the same samples described abovetogether with some reference samples after ODAT.

Comparison of freeze.etch replicas andultrathin sections

Replicas of spray-frozen and freeze-etched clay mineraldispersions allowed the observation of free particles inTEM. It is of interest to compare the surface morphologyand thickness of these particles with the layer structureobserved in ultrathin sections. Measurement of the thick-ness offree particles from freeze-etch replicas poses prob-lems because the angle of observation in TEM is not pre-cisely known. Determination of the thickness was possible,

2gCuKa

Fig. l. X-ray diftaction patterns: a : Hungary illite (Fiizer-radvany), air dried. b : The same specimen after glycolation.The (001) reflection at 10.3 A (before glycolation) splits into a9.8-A peak and a peak at ll.0-11.6 A after glycolation. c :

Bavaria glauconite (Bichl), air dried. d : The same specimenafter glycolation. The 10.2-A peak changes to 10.0 A and broad-ens. e : Quebec illite (northern Appalachians), air dried. f: thesame specimen after glycolation does not show any changes inits XRD pattern. All samples are fractions < 1.0 rrm. Ethyleneglycol solvation accomplished by exposure to the vapor at 60 "Cfor 24h. Abbreviations are c : chlorite; i : illite; q: qruartz:'f: feldspar.

3.34 | 3.54 7.121 1O'0

4 5 4 0 3 5 3 0 2 5 2 0 1 5 1 0 5 2

r976

2gCuKa

Fig. 2. X-ray diffraction patterns of ODAT (octadecylam-monium treatment) samples: a : Hungary illite; b : Bavariaglauconite; s : Quebec illite; d : S/I mixedJayer clays, MancosShale (CMS sample ISTM-I); e : rectorite, Little Rock, Arkan-sas; f : vermiculite, Llano, Texas; g : calcium montmorillonite,Polkville, Mississippi (API standard 21); h : sodium mortmo-rillonite, Upton, Wyoming (API standard 25); i: finely groundmuscovite, Wacker Chemie, Munich. All samples (fraction < 1.0rrm) were treated two times for 24 h (for preparation method,see Riihlicke and Kohler, l98l).

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

10152 02 530

however, for particles that were oriented perpendicularlyto the freeze-etched surface and parallel to the directionof shadowing (arrows in Fig. 3b). In this case, thicknesscould be measured with a precision of at least l0 A.

Figure 3a shows an overview ofan ultrathin section ofa 50/o dispersion of Hungary illite (< I pm fraction). Mostof the particles embedded in epoxy resin were distinctlyaggregated. This aggregation probably resulted from de-hydration required for impregnation. In contrast, thefreeze-etch replica of the same 5olo dispersion (Fig. 3b)showed that the silicate particles were perfectly dispersedin HrO as free particles. The thickness of these particles(50-500 A) was l-2 orders of magnitude smaller com-pared with the aggregates (up to 0.5 pm) shown in ultra-thin section (Fig. 3a). Higher magrrification of the samematerial in ultrathin section revealed that the particlesembedded in epoxy resin consisted of numerous packetsstacked on top ofone another (Fig. 4a). The thickness ofindividual packets within the stacks varied between 40and 150 A and was comparable to that of some free par-ticles in the freeze-etch replicas. Single isolated packetswere rarely encountered in ultrathin sections.

The surface of most free particles in freeze-etch replicas

Gig. ab) showed a layered or terraced stmcture, indicat-ing that these particles were composed of many units.Under the experimental conditions applied here, obser-vation of l0-A steps (: thickness ofa single 2:l layer) onfreeze-etch replicas is difficult (see Fig. 8 in Vali andBachmann, 1988). Therefore the height of these steps (:thickness of the units) must be more than t0 A. If so,these units may represent the packets observed in ultra-thin sections. In other words, a complete disintegrationof all particles into individual packets did not occur indispersion. The shape and diameter of the stacked unitsvaried considerably and included both euhedral laths andplates. Laths and plates usually occurred within the samefree particle (Figs. 4b and 5b). In some cases, the laths intwo adjacent units are interlaced parallel to their longaxes. The contacts between adjacent units were clearlyvisible on crystal surfaces in the freeze-etch replicas (ar-rows in Fig. 5b). The ultrathin section shown in Figure5a may represent a cross section, perpendicular to thebasal plane, through a particle that is similar to the oneimaged in Figure 5b. The steps along the outer edges ofthe particle (arrows in Fig. 5a) may correspond to thecontacts between adjacent units visible in the freeze-etchreplica (arrows in Fig. 5b).

Particles of glauconite showed terracelike margins inultrathin section (solid arrows in Fig. 6a). The 2:l-layerpackets (50-200 A thick) were not always stacked paralleland their thicknesses were not always constant. Some ofthe packets appear to be tilted and wedge out (open ar-rows within particle of Fig. 6a). Freeze etching revealedthat both irregular platy crystals and very well developedpseudohexagonal crystals commonly occur in the sameparticle. Figure 6b shows surface features of the glauco-nite particles that display well-developed crystal edgesstacked in a coherent crystallographic orientation. Aggre-gates of extremely small platy crystallites (< 100 A thick

3.54 4.98t 7.1

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

\ ,

.3b

t977

a r

\

a

.:

Fig. 3. Low-magnification images of 5olo suspensions of Hungary illite, < 1.0 pm fraction. (a) Ultrathin section showing particle

aggregates in epoxy resin. (b) Freeze-etch replica representing dispersed free panicles ofthe same material embedded in ice. Arrows

point to particles oriented perpendicularly to freeze-etched surface and parallel to direction ofshadowing. (Note: shadows are white.)

Fig. 4. High-magnification images of Hungary illite. (a) Ultrathin section showing stacking of individual layer packets. (b) Freeze-etch replica of a free particle. The surface structure indicates that this free particle is composed of numerous units. Part a could bea typical cross section ofa particle similar to the one shown in b.

# . . ' , l ' f;;1 ; .'t

ia, : ; , , , , ,

t978

a

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

o.o5 p,m-

Fig. 5. High magnification of Hungary illite exhibiting intergrowth structure. (a) Ultrathin section showing stacks of packets withwedged intergrowth structure (arrows within particle). Arrows indicate the positions ofcontacts bet\i/een adjacent packets. (b) Freeze-etch replica showing intergrowths oflaths and irregular plates within an individual particle. Arrows denote contacts between adjacentlaths. Part a could be a typical cross section ofa particle similar to the one shown in b.

and < 1000 A in diameter) were also observed, but thesewere irregular in both shape and stacking arrangement.

ODAT of illite and glauconite

A stable expansion of expandable interlayers in illiteand glauconite was achieved by treating the samples withoctadecylammonium ions (n. : l8). Two sets of expand-ed interlayer spacings were observed in HRTEM: onewith an average 24-A gange:22-25 A) hyer spacing (sol-id diamonds in Figs. 7a and 7b) corresponding to a par-affinlike layer configuration in a high-charge expandablecomponent (Lagaly,1979;Yali and Hesse, 1990); the oth-er with spacing of 16-17 A corresponding to a bilayerarrangement of n.: l8 in a low-charge component (openarrows in Figs. 7a and 7b). The proportion of expandablecomponents was estimated by counting the number ofinterlayers. The proportion of expanded interlayers inglauconite determined in HRTEM after ODAT was ap-proximately 20o/o greater than in illite. Whereas the esti-mates of Srodori et al. (1990) and Veblen et al. (1990) are

based on an assumed two-component system of expand-able smectite and nonexpandable illite, our material re-vealed the presence of more than two components. Sincethe ODAT measures the interlayer charge generated inadjacent layers (Lagaly, 1979), the true structure andcomposition of individual 2:l layers in each of the com-ponents could not be determined. However, different re-sponses to ODAT indicated the presence of several dis-tinct phases in the same sample, at least three of whichcould be characterized as follows:

The first phase was a nonexpandable component (40-500/o ofall layers), and therefore its layer charge could notbe measured directly. Average layer charges calculatedfrom the chemical compositions of the Hungary illite,I(o,uNao orCao o, (Al, ,nMgo ,r), e8[Al0 57Si3 o3O,.(OH)J andthe glauconite from Bavaria, Ko, Nao o, (Alo.ou Fef j, Fefr -f,

Mgo rr), e7[Alo 20Si3 80O,0(OH)r], are 0.82 and 0.77, respec-tively, per [O,'(OH)r] (Vali and Kdster, 1986). The layercharges of the first phase in both minerals must be greaterthan these values because the second phase has charges

VALI ET AL.:FREEZE-ETCH REPLICAS AND HRTEM IMAGES t9't9

Fig. 6. High-magnification images of Bavaria glauconite exhibiting inter$owth structure. (a) Ultrathin section showing packets

with stacking disorder (open arrows within particle). Solid arrows point to the terracelike margins ofparticle as well as the interfaces

between adjacent packets. (b) Freeze-etch replica of a crystal with well-developed crystal faces showing pseudohexagonal symmetry

(solid arrows). Part a could be a typical cross section ofa particle similar to the one shown in b.

less than these averages (see below). The interlayer po-sitions of this phase must be largely filled with K becausethe layers did not expand even after ODAT for 6 weeks.Such behavior has been observed only in muscovite (Fig.2, section i; Vali and Kdster, 1986) and biotite (Ghabruet al., 1989). This phase was identified in HRTEM imagesas nonexpanded packets of varying thickness after ODAT(Figs. 7a, 7b,8a, and 8b). The l0-A peak on X-ray dif-fraction patterns (Fig.2, sections a-c) probably receiveda contribution from thicker packets (>60 A) of this phaseand also from the (003) reflection ofthe second phase (seebelow). Packets consisting offour to six nonexpanded 2:llayers were predominant in the Hungary illite (Fig. 7b).These packets may represent a 47-A or 57-A superstruc-ture ofthe type IIIS or IIIIS, which causes the superlatticereflections at 9.8 and I 1.0-11.6 in glycolated XRD pat-terns (Fig. 1, section b; R. C. Reynolds, Jr., personal com-munication, 1991). In the glauconite and illite from Que-bec, thicker packets containing 10 or more 2:l layerspredominated (Figs. 7a and 8).

The second phase (20-300/o of all layers) woS €rn ex-pandable high-charge component. An accurate determi-nation of mean interlayer charge from XRD analysis (fol-lowing the method of Olis et al., 1990) yielded values of0.74 and 0.70 per [Oro(OH)r] for illite and glauconite, re-spectively. The interlayers ofthis phase expanded after 2d of ODAT, giving an XRD peak of 30.2 A for illite and

29.2 Afor glauconite (Fig. 2, sections a and b). In HRTEMimages, this phase appeared as sequences of coherent-appearing layers with a 24-A spacing (Figs. 7a and 7b).

This difference may be due to radiation damage of then-alkylammonium ions under the electron beam (Vali andHesse, 1990). Similar expansion behavior is shown byvermiculite (weathered biotite) and standard "Llano ver-

miculite" (Ghabru et al., I 989; Vali and Hesse, I 990; Vali

and Kdster. 1986). Laird et al. (1987) observed analogousXRD behavior for some illite using alkylammonium ex-change and XRD techniques, and they described the ex-pandable component as "expanded illite layers"'

The third phase (5-100/o of all layers) consisted of l:l

ordered mixed layers, which occurred as thin packets ofa few nonexpandable double layers sandwiched betweenexpandable smectiteJike interlayers ( I 6- I 7 A) or as singlenonexpandable double layers (Figs. 7a and 7b). Orderedmixed layers of this kind were also imaged in Marbleheadillite (Vali and Hesse, 1990).

In addition, 5-l0o/o of all layers in ultrathin sectionswere single 2:l layers or thin packets ofexpanded layers,which occur either isolated or within larger particles (Fig.

7a). Their main characteristic is that they are short andbent. Aggregates ofvery fine particles were also observedin freeze-etch replicas that might represent this material.These layers appear to be smectite-like, but it was notpossible to identify the mineral. Their morphology hinted

l 980 VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

Fig. 7. Ultrathin sections of(a) Bavaria glauconite and (b) Hungary illite, showing stacked nonexpanded and expanded componentsafter treatment with octadecylammonium ions. Open diamonds: packets of 2:l layers with a 10-A spacing (first phase); soliddiamonds : packets of 2: I layers with an average 24-A spacing (second phase); open arrows : interlayer spacing of 16-17 A; solidtriangles : 1:1 ordered-mixed layers (Rl structure, third phase); S : short and bent 2:l layers ofa smectiteJike component (fourthphase?); open triangles : thin packets offour to six 2:l layers (R3 structure); black arrows : interfaces between adjacent packets.

at smectite, which would represent an additional separatephase.

ODAT provides characteristic XRD patterns for well-defined minerals such as rectorite, vermiculite, mont-morillonite, and muscovite (Fig. 2, sections e-i), whichcan be correlated with HRTEM images as well as X-raydata on glycolated specimens. We want to emphasize,however, that for illitic material such as illite, smectite/illite, and glauconite the X-ray data on glycolated mate-rials did not correlate with ODAT results obtained fromXRD or HRTEM. The Bavaria glauconite and the Quebecillite did not reveal interstratification with smectite afterglycolation, whereas the Hungary illite did (Fig. l, sectionsa-f). However, after ODAT all three samples showedexpansion (Fig. 2, sections a-c, Figs. 7a, 7b,8a, and 8b).It is obvious that some l0-A hyers, which were nonex-pandable after glycolation, did expand by ODAT, as seenon HRTEM images (the second phase in Figs. 7a, 7b,8a,and 8b). ODAT apparently exerts a stronger expandingforce than glycol treatment. It is most likely that the peakscorresponding to d-values around 30 A, 15 A, and l0 Ain ODAT X-ray diffraction patterns (Fig.2, sections a-c)represent this type oflayer [i.e., the (00 l), (002), and (003)

reflections ofthis expanded phasel, whereas the strong I 0-A reflection (compared to rectorite, Fig. 2, section e, orvermiculite, Fig. 2, section f) also includes a contributionfrom the nonexpandable illite phase, as mentioned above(Figs. 7a, 7b, 8a,and 8b).

The reproducibility of ODAT results depends largelyon uniform experimental conditions such as washing anddrying (under a l0 a-bar vacuum for at least 12 h) pro-cedures, use of the same particle size-fraction, and du-ration of treatment. The similarity in the appearance ofthe ODAT X-ray pattern of the interstratified Hungaryillite and the noninterstratified Bavaria glauconite (Fig. 2,sections a and b) indicates that the original structure ofthese materials may be modified during ODAT. Thesharpness of the 30-A peak indicates that the expandedlayers indeed occur as coherent sequences, as is visible inHRTEM (Figs. 7a and 7b). In contrast, the difiirse reflec-tion in the Quebec illite at 3142 A Fig. 2, section c)indicates the presence of stacking disorder of expandableand nonexpandable layers in this sample, as can also beseen in HRTEM images (Figs. 8a and 8b). Although someother illite and glauconite samples show similar behavior(Vali, 1983; Vali and Kcister, 1986), a detailed compar-

a

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES 198 r

Fig. 8. Ultrathin section of Quebec illite treated with octadecylammonium ions. (a) Stacked nonexpanded layers (10 A) andexpanded layers (25 A1. 6; Ctose-up ofa showing that both packets as well as single double layers (20 A; ofnonexpandable 2:llayers are present (open arrows).

ative XRD and HRTEM investigation of ODAT samplesof these materials was beyond the scope of this study.

As mentioned above, the HRTEM images of the firstand second phases of illite and glauconite after ODAT(Figs. 7a and 7b) could be correlated with their ODATX-ray patterns (Fig. 2, sections a and b). However, thethird phase as well as the smectiteJike phase could notbe differentiated in diffraction patterns because of(l) pos-sible peak overlap, (2) the small amounts present of thesephases, and (3) the random distribution ofvery thin pack-ets or layers. For comparison, an ordered l: I mixed-layermineral (rectorite) produced a series of XRD peaks thatoverlapped those ofthe expanded phase in illite and glau-conite (compare Fig. 2, sections a and b, with Fig. 2,section e) and could not be detected by XRD even ifpresent in sufficient quantity. Vermiculite has the sameexpanded reflections as those ofthe second phase (at 29-30 A in XRD, Fig. 2, section f, and,24 A in HRtEVt,Vali and Hesse, 1990). The presence of vermiculite canbe excluded, however, because ofthe absence ofcharac-teristic XRD reflections for this mineral in glycolated ma-terial (Fig. l, sections b and c). Absence of a 17-20 Apeak which is characteristic of ODAT smectite (Fig. 2,sections g and h), confirms that the expanded phase is notsmectite. Therefore the second phase is interpreted as ex-pandable illite in the sense ofLaird et al. (1987).

Preliminary HRTEM investigation of the illite fromQuebec revealed that both thin (< 100 A) and thick pack-ets (> 100 A; were stacked together. After ODAT, ap-

proximately 20o/o of the interlayers were expanded in bothtypes of packets. Figures 8a and 8b show nonexpanded(10-A) and expanded (25-A) interlayers alternating in arandom array.

DrscussroN

Combination of cryofixation and freeze-etching tech-niques with ultrathin sections and XRD methods is apotentially powerful tool that may hold a key to the so-lution of controversial issues concerning the natural oc-currence of fundamental particles and the multiphase na-ture of illitic materials.

Free particles vs. fundamental particles

Free particles observed in freeze-etch replicas are thick-er (average thickness of 200 A; ttran the packets of 2:llayers imaged in ultrathin section (average thickness of100 A). The former are composed of several units as ev-idenced by their surface structure (Figs. 4b, 5b, and 6b).These units correspond to the packets seen in ultrathinsection (Figs. 4a, 5a, and 6a). In some cases, the well-developed edges of units that repeat consistent crystal-lographic orientations suggest that the free particles dis-play primary growth features (Fig. 6b). The results of freezeetching combined with those from ultrathin sectioningthus reveal that the stacking arrangement ofthe packets(Fig. 4a) can be a primary feature of these materials inboth organic resins and aqueous dispersion and is not onlya clumping artifact. Under the applied experimental con-

1982

ditions it is not possible to tell whether the stacking ar-rangements and surface features of the particles in dis-persion are the same as those existing in the original insitu mudrock fabric.

Thin packets (30-50 A) observed in ultrathin sections(Figs. 7a and 7b) have the thickness of fundamental illiteparticles in an R3-ordered mixed-layer clay. However,they do not occur as individual free particles in dispersion.This observation confirms Ahn and Buseck's (1990) con-clusion that the crystallites (: stacks of packets in ourterminology) in Rl- and R3-ordered S/I mixed-layer claysare thicker than the fundamental particles of Nadeau andcoworkers. The free particles ofillite and glauconite (50-500 A) are considerably larger than fundamental particlesof illite in S/I mixed layers described by Nadeau et al.(l98aa) and Nadeau (1985). Thus the natural occulrenceof fundamental illite particles in the materials studied isunlikely.

In contrast, preliminary freeze-etching results of theWyoming sodium montmorillonite show that, in thissample, free particles may exist as single dispersed smec-tite 2:l layers (Vali and Bachmann, 1988). This obser-vation supports the definition of fundamental particles forsmectite by Nadeau et al. (1984a, 1984c) and Nadeau(1985). Thus in the case of smectite, the size of free par-ticles can be the same as that of fundamental smectiteparticles.

Nature of bounding faces of packets

It has been suggested that natural S/I mixed-layer claysconsist of thicker particles, which may split into funda-mental particles because of osmotic swelling along theirsmectite interlayers during the sample preparation pro-cedure (Altaner and Bethke, 1988; Altaner et al., 1988;Ahn and Peacor, 1986b; Veblen et al., 1990; Ahn andBuseck, 1990). Altaner and Vergo (1988) and Altaner etal. (1988) viewed such smectite interlayers as a separatephase. If smectite layers are present as individual layers,they can be imaged in ultrathin sections (Shomer andMingelgrin, 1978; Srodofi et al., 1990; Vali and Kdster,1986) and in dispersions after platinum-carbon shadow-ing (Nadeau et al., 1984a, I 984b, I 984c) or freeze-etching(Vali and Bachmann, 1988). In this case, they form aseparate phase. However, single expanded interlayers be-tween illitic packets (solid arrows in Figs. 7a and 7b) inour illite and glauconite do not represent a separate smec-tite phase. As mentioned, freeze etching reveals that mostfree particles do not split into units in HrO, even in diluteand Na-saturated suspensions. Moreover, the interfacesof adjacent packets within some particles do not respondin a consistent way to ODAT (solid arrows in Figs. 7aand b). This observation tends to support the model ofEberl and Srodori (1988) that the basal surfaces of"shortstacks" are nonswelling "stacking faults." The problemof characterization of interlayers between the packets isclosely related to the question ofwhat the basal surfacesof these packets are. Determination oftheir chemical com-position is not yet possible. However, the top and bottom

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

layers ofindividual packets ofillite and glauconite ofthisstudy do not appear to be smectite.

Muttiphase nature of illitic materials

Illitic materials commonly contain a smectite compo-nent and these smectite/illite mixed layers are consideredby some as members ofa continuous series of single-phasesolid solutions or by others as mixtures of two or morethermodynamically distinct phases. This difference inopinion is highlighted by a recent animated discussionconcerning the multiphase nature of hydrothermal sericitefrom the Silverton Caldera, Colorado, by Altaner and Ver-go (1988), Altaner et al. (1988), Eberl et al. (1987, 1988),and Rosenberg et al. (1990). The answer to this long-debated problem (Lippmann, 1977, 1982; Garrels, 1984;Velde, 1985) has important repercussions for (l) the so-lution chemistry of H,O/solid-silicate systems, (2) themechanism of the smectite-to-illite transformation duringburial diagenesis, and (3) the evolution of mica polytypesduring advanced-level diagenesis and low-grade meta-morphism.

As far as solution chemistry is concerned, a two-phasesystem restricts the composition of an aqueous solutioncoexisting in equilibrium with the solid phase more thana single-phase solid solution does. This consequence ofthe phase rule has been used in the design of equilibrationexperiments by Rosenberg et al. (1990). Their results sug-gest that, in the illitization process of smectite under hy-drothermal conditions, at least four discrete mica-likephases appear.

Morphological studies by Inoue et al. (1987, 1988) ofthe smectite-to-illite conversion and the transformationof the lM to 2M polytype of illite from the Shinzan hy-drothermal alteration area of Japan, on the other hand,revealed the occurrence ofthree distinct morphologic typesof mica-like minerals that appear successively in the il-litization process: flakelike (LMd), lathlike (1M) andplatelike (2M) crystallites. There is considerable overlapin the chemical composition of the different morphologictypes, which the authors attribute to a dissolution-re-crystallization mechanism corresponding to an Ostwaldmaturation process. Eberl et al. (1987,1988) also preferOstwald ripening of a single-phase solid-solution S/I inthe course of which larger particles with lower surfaceenergies grow at the expense of smaller particles. Thefundamental-particle concept of Nadeau et al. (1984a,1984b) and Nadeau (1985) is well in line with such amechanism.

Our results support the suggestion ofAja et al. (1991)and Rosenberg et al. (1990) that illitic materials consistof several thermodynamically distinct phases. In the illiteand glauconite studied, the different phases exist withinthe same free particles. If the thermodynamic definitionof a phase given in the introduction is followed, then thethree types of layer stacks identified in this study by theirdifferent expansion response to ODAT can be interpretedas three distinct phases in illitic materials. However, noevidence was found for the presence of smectite as a sep-

arate phase in ordered mixed layers, contrary to Altaneret al. (1988).

Illitic materials that are composed of more than onephase occur in a variety of different environments. Theyform authigenically (Bavarian glauconite), hydrothermal-ly (Hungarian illite), and during burial diagenesis andmetamorphism (Quebec illite). In this study, we haveshown that the freeze-etch and ultrathin section tech-niques together with the n-alkylammonium method en-able the particle size and the multiphase nature of illiticmaterials to be investigated. Further work may unravelimportant details of the phase transformations of clayminerals during progressive burial diagenesis and meta-morphism.

AcrNowr-nocMENTs

The cryofixation and freeze-etching work was performed in L. Bach-mann's EM Laboratory at the Institut fiir Technische Chemie, TechnicalUniversity of Munich, Germany. Helpful discussions with L. Bachmannare gatefully acknowledged. We thank R. Reynolds, Jr., G. Guthrie, J.Srodori, and an anonymous reviewer for their constructive suggestions intheir reviews of this paper. This research was supported by grants fromthe Deutsche Forschungsgemeinschaft (DFG), Germany; the Natural Sci-ences and Engineering Research Council (NSERC) of Canada; the Petro-leum Research Fund administered by the Arnerican Chernical Society(ACS), Amoco Canada Petroleum Company, Ltd., and the Faculry forGraduate Studies and Research of McGill Universitv. Montreal.

RrrnnrNcrs crrEDAhn, J.H., and Buseck, P.R. (1990) Layer-stacking sequences and struc-

tural disorder in mixed-layer illite/smectite: Image simulation andHRTEM imagrng. American Mineralogist, 7 5, 267 -27 5.

Ahn, J.H., and Peacor, D.R. (1986a) Transmission electron microscopedata for rectorite: Implications for the origin and structure of"funda-mental particles." Clays and Clay Minerals, 34, 180-186.

-(1986b) Transmission electron microscope data for rectorite: Im-plications for the origin and structure of"fundamental particles." Claysand Clay Minerals, 34, 180-186.

-(1989) Illite/smectite from Gulf Coast shales: A reappraisal oftransmission electron microscope images. Clays and Clay Minerals, 37,542-546.

Aja, S.U., Rosenberg P.E., and Kittrick J.A. (1991) Illite equilibria insolutions: L Phase relationships in the system KrO-AlrOr-SiO,-H,Obetween 25 .C and 250 qC. Geochirnica et Cosrnochimica Acta, 55,r353-1364.

Altaner, S.P., and Bethke, C.M. (1988) Interlayer order in illite/smectite.American Mineralogist, 7 l, 7 66-77 4.

Altaner, S.P., and Vergo, N. (1988) Sericite from the Silverton Caldera,Colorado: Discussion. American Mineralogist, 7 3, | 47 2- | 47 4.

Altaner, S.P., Weiss, C.A., Jr., and Kirkpatrick, R.J. (1988) Evidence fromnsi NMR for the structure of mixedJayer illite/smectite clay minerals.Nature. 331. 699-702.

Atkins, P.W. (1978) Physical chemistry. W.H. Freeman and Company,San Francisco.

Bachmann, L. (1987) Freeze-etching ofdispersions, emulsions and mac-romolecular solutions ofbiological interest. In R.A. Steinbrecht and trLZierold, Eds., Cryotechniques in biological electron microscopy, p. I 92-204, Springer-Verlag, New York.

Bachmann, L., and Schmitr, W.W. (1971) Improved cryofixation appli-cable to freeze-etching. Proceedings of the National Academy. of Sci-ences, U.S.A., 68, 21 49-21 52.

Baronnet, A. (1972) Growth mechanisms and polytypism in synthetichydroxyl-bearing phlogopite. American Mineralogist, 57, 127 2- 1293.

Bell, T-E. (1986) Microstructure in mixed-layer illite/smectire and its re-lationship to the reaction of smectite to illite. Clays and Clay Minerals,34, t46-154.

1983

Bethke, C.M., andAltaner, S.P. (1986) Layer-by-layermechanism of smec-tite illitization and application to a new rate law. Clays and Clay Min-erals,34, 136-145.

Brown, G. (1984) Crystal structures of clay minerals and related phyllosili-cates. Philosophical Transactions of the Royal Society, I-ondon, A 3 I I ,221-240.

Eberl, D.D., and Srodori, J. (1988) Ostwald ripening and interparticle-diffraction effects for illite cryslals. American Mineralogist, 73, 1335-1345.

Eberl, D.D., Srodori, J., ke, M., Nadeau, P.H., and Northrop, H.R. (1987)Sericite from the Silverton Caldera, Colorado: Correlation among struc-ture, composition, origin, and particle thickness. American Mineralo-g;st, 72, 914-934.

Eberl, D.D., Srodori, J., Lee, M., and Nadeau, P.H. (1988) Sericite fromthe Silverton Caldera, Colorado: Reply. American Mineralogist, 73,147 5-1477 .

Garrels, R.M. (19E4) Montmorillonite/illite stability diagram. Clays andClay Minerals, 32, 16l-166.

Ghabru, S.K., Mermut, A.R., and Sr. Arnaud, R.J. (1989) Iayer-chargeand cation-exchange characteristics of vermiculite (weathered biotite)isolated from a gray luvisol in northeastern Saskatchewan. Clays andClay Minerals, 37, 164-172.

Guthrie, G.D., Jr., and Veblen, D.R (1989) High-resolution transmissionelectron microscopy of mixedJayer illite/smectite: Computer simula-tions. Clays and Clay Minerals, 37, l-lt.

- (1990) Interpreting one-dimensional high-resolution transmissionelectron micrographs of sheet silicates by computer simulation. Amer-ican Mineralogist, 7 5, 27 6-288.

Hiemenz, P.C. (1986) Principles of colloid and surface chemistry (2ndedition). Dekker. New York.

Inoue, A., Kohyama, N., Kitagawa, R., and Watanabe, T. (1987) Chemicaland morphological evidence for the conversion of smectite to illite.Clays and Clay Minerals, 35, I I l-120.

Inoue, A., Velde, B., Meunier, A., and Touchard, G. (1988) Mechanismofillite formation during smectite-to-illite conversion in a hydrothermalsystem. American Mineralogist, 7 3, 1325-1334.

Inoue, A., Bouchet, A., Velde, 8., and Meunier, A. (1989) Convenienttechnique for estimating smectite layer percentage in randomly inter-stratified illite/smectite minerals. Clays and Clay Minerals, 37,227-234.

Klimentidis, R.8., and Mackinnon, I.D.R. (1986) High-resolution imagingofordered mixed-layer clays. Clays and Clay Minerals, 34,155-164.

Kohler, E.E., and Kdster, H.M. (1976) Zur Mineralogie, Kristallchemieund Geochemie Kretazischer Glaukonite. Clay Minerals, ll,273-302.

K6ster, H.M., Kohler, E.8., Kmhl, J., Krdger, J., and Vogt, IC (1973)Veriinderungen am Montmorillonit durch Einwirkung von 0. I m AeDTE-Liisungen, I n NaCl-Liisung und 0.1 n Salzsiiure. Neues Jahrbuch fiirMineralogie Abhandlungen, I 19, 83-100.

I-agaly, G. (1979) The "Layer charge" of regular interstratified 2:l clayminerals. Clays and Clay Minerals, 27, l-10.

- (l 982) Iayer charge heterogeneity in vermiculites. Clays and ClayMinerals. 30.215-222.

Iagaly, G., and Weiss, A. (1969) Determination of the layer charge inmica-type layer silicates. Proceedings ofthe International Clay Confer-ence, Tokyo, l ,6 l -80.

Laird, D.A., Scort, A.D., and Fenton, T.E. (1987) Interpretation of al-kylammonium characterization of soil clays. Soil Science Society ofAmerica Joumal. 51. 1659-1663.

ke, J.H., and Peacor, D.R. (1986) Expansion of smectite by laurylaminehydrochloride: Ambiguities in transmission electron microscope ob-servations. Clays and Clay Minerals, 34,69-73.

ke, S.Y., and Jackson, M.L. (1975) Micaceous occlusions in kaoliniteobserved by ultramicrotomy and high resolution electron-microscopy.Clays and Clay Minerals, 23, 125-129.

Lippmann, F. (1977) The solubility products of complex minerals, rnixedcrystals and threeJayer clay minerals. Neues Jahrbuch liiLr MineralogieAbhandlungen, | 30, 24t -263.

-(1982) The thermodynarnic status of clay minerals. In H. VanOlphen and F. Veniale, Eds., Intemational Clay Conference 1981, p.475-485. Elsevier, New York.

Mackinnon, I.D.R. (1987) The fundamental nature of illite/smecdre mixed-

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

I 984

layer clay particles: A comment on papers by P.H. Nadeau and co-workers. Clays and Clay Minerals, 35,74-76.

Marcks, C., Wachsmuth, H., and Reichenbach, H. (1989) Preparation ofvermiculites for HRTEM. Clay Minerals, 24,23-32.

Moor. H.. and Muehlethaler. K. (1963) Fine structure in frozen-etchedyeast cells. Journal ofCell Biology, 17, 609-628.

Nadeau, P.H. (1985) The physical dimensions offundamental clay par-ticles. Clay Minerals, 20, 499-5 14.

Nadeau, P.H., Tait, J.M., McHardy, W.J., and Wilson, M.J. (1984a) In-terstratifed XRD characteristics of physical mixtues ofelementary clayparticles. Clay Minerals, 19, 67 -7 6.

Nadeau, P.H., Wilson, M.J., McHardy, W.J., and Tait, J.M. (1984b) In-terparticle diffraction: A new concept for interstratified clays. Clay Min-erals. 19. 757-769.

- (l 984c) Interstratified clays as fundamental particles. Science, 225,923-925.

Nemez, E. (1973) Tonmineralogie. Akademischer Verlag, Budapest.Olis, A.C., Malla, P.B., and Douglas, L.A. (1990) The rapid estimation of

the layer charges of 2:1 expanding clays from a single alkylammoniumexpansion. Clay Minerals, 25, 39-50.

Reynolds, R.C., Jr. (1989) Diftaction by small and disordered crystals.In Mineralogical Society of America Reviews in Mineralogy, 20, 145-l 8 l .

Reynolds, R.C, Jr., and Hower, J. (1970) The nature ofinterlayering inmixed layer illite-montmorillonites. Clays and Clay Minerals, 18, 25-36.

Robards, A.W., and Sleytr, U.B. (1985) Low temperature methods inbiological electron microscopy. In A.M. Glauert, Ed., Practical methodsin electron microscopy, p. 10. Elsevier, Amsterdam.

Rosenberg, P.E., Kittrick J.A., and Aja, S.U. (1990) Mixed-layer illite/smectite: A multiphase model. American Mineralogist, 75, I 182- I 185.

Riihlicke, G., and Kohler, E.E. (1981) A simplified procedure for deter-mining layer charge by the n-alkylammonium method. Clay Minerals,16, 305-307.

Shomer, I., and Mingelgrin, U. (1978) A direct procedure for determiningthe number of plates in tactoids of smectites: The Na./Ca-montmoril-lonite case. Clays and Clay Minerals, 26,135-138.

VALI ET AL.: FREEZE-ETCH REPLICAS AND HRTEM IMAGES

Spun, A.R. (1969) A low-viscosity epoxy resin embedding medium forelectron microscopy. Joumal of Ultrastructure Research, 26, 31-43.

Srodoi, J. (1984) X-ray powder diffraction identification ofillitic mate-rials. Clays and Clay Minerals, 32,337-349.

Srodof, J., Andreoli, C., Elsass, F., and Robert, M. (1990) Direct high-resolution transmission electron microscopic measurement of expand-ability of mixed-layer illite/smectite in bentonite rock. Clays and ClayMinerafs. 38. 373-379.

Stanjek, H., and Friedrich, R. (1986) The determination oflayer chargeby curve-fitting of Lorentz- and polarization-corrected X-ray diagrams.Clay Minerals, 21, 183-190.

Vali, H. (l 983) Vergleichende elektronenoptische und r6ntgenographischeUntersuchungen zur Kristallstruktur und Morphologie von quell{iihigen

und nicht quellenden 2: I Schichtsilikaten. Ph.D. dissertation, Fakultiitfiir Geowissenschaften, Technische Universit.?it Miinchen, Germany.

Vali, H., and Bachmann, L. (1988) Ultrastructure and flow behavior ofcolloidal smectite dispersions. Journal ofColloid and Interface Science,126,278-291.

Vali, H., and Hesse, R. (1990) Alkylammonium ion treatment of clayminerals in ultrathin section: A new method for HRTEM examinationof expandable layers. American Mineralogist, 7 5, | 443 - | 446.

Vali, H., and Kdster, H.M. (1986) Expanding behaviour, structural dis-order, regular and random irregular interstratification of2:1 layer-sil-icates studied by high-resolution images of transmission electron mi-croscopy. Clay Minerals, 21, 827 -859.

Veblen, D R, Guthrie, G.D., Jr., Liyi, K.J T., and Reynolds, R.C., Jr.( I 990) High-resolution transmission electron microscopy and electrondiffraction of mixedJayer illite/smectite: Experimental results. Claysand Clay Minerals, 38, l-13.

Velde, B. (1985) Clay minerals. A physico-chemical explanation oftheiroccurrence. In Developments in Sedimentology, 40, 428 p.

MeNuscprn R-EcErvED JuI-v 17, 1990

MeNuscrrrr ACCEPTED Jure 26, l99l