American Mineralogist, Volume 70, pages 517-528, 1985

Characterization of synthetic tridymites by transmission electron microscopy

Mtcn.lpr, A. C.mpeNrBn

Department of Earth Sciences, Uniuersity of CambridgeDowning St., Cambridge CB2 3EQ, Englanil

lNn MsctrrHILn Wur.rNslr,cn

I nstitut fiir Kristallogr aphie und P etr ogr aphieE.T .H.-Z entrum. Ziirich C H -809 2, Switzerland

Abstract

Eight tridymites, synthesized in a variety of ways, have been examined by transmissionelectron microscopy (TEM). Each sample appeared to be heterogeneous, containing a mixtureof low temperature structural forms. Two mqin structure types were identified: one with apseudohexagonal superlattice, having o: J3ar, 9:2cw and the second with a pseudo-orthorhombic C-face centered cell, having a:2J3at, b :2an and wi^th either a multiple crepeat or (001) stacking disorder (where a, and cs are respectively -5A, -8.2A and refer tothe high temperature hexagonal sublattice). The former structure seemed to predominate ingrains with few (001) stacking faults. In addition, some samples contained regions of consider-able (001) disorder intergrown with more highly ordered material on a scale of a few hundredtngstrtims. Diffraction patterns from these grains could not be indexed easily on a knowntridymite or cristobalite superlattice and have been tentatively ascribed to mixed (tridymite/cristobalite) layering.

Highlow transformations were inducod by beam heating. The ZJ3asx 2a^x nc' typesuperstructures could be transformed reversibly to the hex4onal higlt temperature sublatticeuntil radiation damage started to occur. In contrast, the ./3ag x 2c" superlattice gave way to

the high temperature hexagonal sublattice on heating but invariably reverted to a 213a,x 2a^ x ncH structure on subsequent cooling.

The diversity of superstructures observed in each sample helps to a@ount for differencesbetween the transformation behavior of many synthetic tridymite powders and larger naturalor synthetic single crystals. In this context, TEM provides a useful adjunct to X-ray powderdiffraction for the characterization of tridymites prepared for other types of experiments.

IntroductionIt has long been known that tridymites undergo displa-

cive structural transformations on heating and cooling.These transformations have been studied as functions ofpressure and temperature by a variety of techniques, in-cluding optical microscopy with heating stages (Fenner,1913; Fliirke, 1955; Dollase et al., l97l; Nukui et al., 1978),single crystal or powder X-ray heating cameras (Gibbs,1926; Buerger and Lukesh, 1942;Hill and Roy, 1958; Sato,1963a" b, 1964; Dollase and Buerger, 1966; Tagai and Sa-danaga" 1972;Kihara, 1977,1978:' Nukui et al,, 1978; Hofr-mann et al., 1983), the diamond anvil cell (Nukui et al.,1980), differential thermal analysis (Fliirke, 1955; Fldrkeand Miiller-Vonmoos, 1971; Cohen and Klement, 1980)and calorimetry (Mosesman and Pitzer, 1941; Shahid andGlasser, 1970; Thompson and Wennemer, 1979). One ofthe most striking features of the complex series of changeswhich have been observed is that different tridymites usu-ally do not show the same sequence of structural changes

0003-o04x/85/050H5 I 7$02.00

with the same transformation temperatures (reviewed byFlcirke, 1955, 1967; Sosman, 1965; Nukui et al., 1978).

When tridymite samples used for transformation studieshave been characterized by single crystal X-ray diffraction,the nature of each superstructure observed has been ratherwell defined (Dollase, 1967; Kihara, 1977; Nukui et al.'1978; Nukui et al., 1980; Hoffmann et al., 1983)' even tothe extent of having full structure refinements from inten-sity data collected at elevated temperatures (Dollase' 1967;Kihara, 1977, 1980). Many synthetic tridymites, however,are too fine grained for single crystal X-ray analysis andthe general method ofcharacterization has been by powder

diffraction. Using the distribution and intensity relations ofdiffraction peaks in powder patterns, Hill and Roy (1958)'Fl<irke (1961) and then Sato (1963a) distinguished at leasttwo different forms of synthetic low tridymite and also sug-gested that some of their products consisted of a mixture ofthese forms. with or without cristobalite as an additionalphase. Unfortunately, this method of characterization can

517

5 1 8 CARPENTER AND WENNEMER: SYNTHETIC TRIDYMITES

be rather unsatisfactory for phases with weak superlatticereflections, and, ifthere is a mixture ofphases present, peakassignment may become difficult. In this context. transmis-sion electron microscopy (TEM) may provide a useful ad-junct to traditional X-ray methods, because it can be usedto examine even very small grains and obtain diffractioninformation from them individually, as shown by previousstudies on natural tridymite (Appleman et al., l97l) andcristobalite (Champness et al., lgjl).

Since a thorough appreciation of the nature of phasetransformations in any material is greatly assisted by theobservation of anomalous heat capacity effects, tridymiteshave recently been examined by dynamical calorimetry(Shahid and Glasser, 1970; Thompson and Wennemer,1979). Thompson and Wennemer (1979) proposed an ap-proximate correlation between the positions of anomalousC" effects and the transformations recognized in relativelylarge single crystals. They also suggested that a sampleshowing X-ray powder diffraction peaks appropriate forcristobalite + tridymite was an intimate intergrowth ofthese two phases rather than a mechanical mixture. Sincecristobalite and tridymite may be represented as having apolytypic relationship to each other, there is clearly thepossibility that more-orless ordered mixed layer phasesmight form. The present TEM investigation was under-taken in the light ofthese observations and its purpose wasto define the nature of superstructure types present in anumber of synthetic tridymite samples, consider theirtransformation behavior and have a closer look at the pos-sible mixed layer phases. Differential scanning calorimetry(DSC) results obtained from the same samples and a de-tailed structural interpretation of the transformations arepresented elsewhere (Thompson and Wennemer, 1979:Wennemer and Thompson, in prep.).

Sample descriptionSilica gel or finely ground quartz powder were mixed with a

variety of carbonate and tungstate fluxes. These were groundunder acetone, pressed into pellets and heated in air for differenttimes at different temperatures. Details of the high purity silica geland of the preparation procedures for samples TR-G4, TR-G3and CR-1 were given by Thompson and Wennemer (1979) and aresummarized in Table 1 In contrast with the silica gel/carbonatemixtures, which were heated on a ceramic tablet, the quartz pow-der/tungstate mixtures were mixed and heated in ceramic cruc-ibles. With the exception of T-lsa, the preparation of which hasbeen described by Flcirke and Langer (1972), all the samples weresynthesized by M. Wennemer.

Powder X-ray diffraction traces over the range 20: lg-37"(CuKa radiation) for the eight tridymites and one cristobalite areshown in Figure 1. No attempt has been made to index thesepatterns and determine lattice parameters because the TEM ob_servations (see below) indicate that each sample contained a mix_ture of low temperature superstructure types. The powder diffrac-tion data do show, however, that the synthetic tridymites have aspread of structural states, including Sato's (1963a) M, S and MStypes. In addition, both Li-Trid and K-Trid have some weak re-flections at the positions of cristobalite peaks (e.g., at -4.05A),TR-G3 has strong cristobalite reflections and its powder pattern isquite distinct from all the other samples, both in terms of the peak

positions and intensity distribution, and CR-l gives only sharpreflections of cristobalite.

ObservationsTridymite grains were ground under alcohol and deposited

onto thin carbon films supported by copper grids. They wereexamined in an AEI EM6G electron microscope operating at 100kV. The crushed fragments, typically less than a few pm in size,suffered radiation damage in the electron beam; after -l minuteexposure they developed a speckled texture and the diffractionpatterns became increasingly diffuse (see below). Using a defo-cussed beam, however, it was possible to photograph microstruc-tures and diffraction patterns before degradation became a senousproblem. At the necessary low levels of illumination. and becauseof the need for rapid recording of the diffraction patterns, carefultilting to produce perfect reciprocal lattice orientations was notfeasible, and the orientation of the grain fragments had to beaccepted as they were found. Although a large number of grainswere examined, it was also dillicult to avoid the tendency torecord, preferentially, diffraction patterns with small spacings andgrains with distinct microstructures. No attempt has been made,therefore, to estimate the proportions of grains in each sampleshowing specifi c features.

A further limitation of the TEM observations is that, in general,reciprocal lattice dimensions cannot be determined accuratelyfrom electron diffraction patterns. Thus we have been unable todistinguish between hexagonal, orthorhombic and monoclinic(with B near 90") lattices that have closely related unit cell dimen-sions. We use the terms pseudohexagonal and pseudo-orthorhombic in our descriptions of the superlattice types to sig-nify this uncertainty, and present only approximate lattice param-eters. The terminology used for describing tridymite super-structures is also rather involved and di.ffers among authors (seeNukui et al., 1978; Nukui and Naka"awa, 1980). All our diffrac-tion patterns have been indexed with respect to the high temper-ature, hexagonal cell (an r 5.0A, c" r 8.2A) and the super-structures are described using conventional superlattice notation.We have then tried to correlate these observations with some ofthe structures described elsewhere in the literature.

Perhaps the single most important observation was that eachsample appeared to be inhomogeneous, consisting of a mixture ofdifferent types of low tridymite. A summary description of eachsample is given in Table 1.

Superstructures

Hexagonal sublattice reflections (a, r 5.0A, c, = g.2A)could be identified easily in electron diffraction patterns bytheir relatively high intensities. At least two distinct super-lattices were distinguished. A common superlattice in sam-ples T-lsa, Li-Trid and Na-Trid has reflections along( I l0)* indicating a tripling of the dr ro spacings and reflec-tions along c* that indicate a doubling of door (Fig. 2a,b)(all Miller indices are given in terms of the hexagonal sub-lattice). This superlattice can be described using a primi-tive, pseudohexa&onal unit cell with a * 8.7A and c I16.44, i.e. o : J3ar, c :2cs. This will be referred to asthe J3aH x 2c" superlattice. Many orientations were in-dexed but the most obvious section contains c* and [110]*(Fig. 2b). Diffraction patterns taken with the electron beamparallel to the c-axis were not found so commonly butappeared to be consistent with the proposed unit ceil (Fig.

CARPENTER AND WENNEMER SYNTHETIC TRIDYMITES

: : : -x 2c,, 2 F 3 a . - x 2 a . . x t c , , c r € p e a t c r i s t o b a l i t e

(nult ip Ie

519

Table 1 summary of tridymite synthesis methods, X-ray powder diffraction characteristics and TEM observations. Tentative assign-

ments to the S, M and MS tridymite types of Sato 1te6fa; have been made on the basis of X-ray powder patterns (Fig. l)' only to

illustrate the wide range of structural stat€s represented by these samples. A cross signifies that a feature was observed, a dash that it was

not, and a question mark that it was suspected but not positively identified. See Thompson and Wennemer (1979) for a description of the

synthesis procedure for TR-G4, TR-G3, CR-l and Fliirke and Langer (1972) and T-lsa.

Smplenane

Start ing mter ial l {eat t reatrent Productposder

r i

T - l s a 2 : L , s i l i c a g e l :Na2WO4

TRI I I s i l i ca ge l + 1nolZ K2CO3

TRIV s i l i ca ge l + Irc17 K2CO3

TR-c3 sil ica gel + 1mlz Na2CO3

TR-c4 sil ica gel + Irc12 K2CO3

72 hrs l4oooc ,cleaned, then1 day 140o-c

1 6 1 h r s ,1200-c

1 1 2 h r s ,1200-c

ovelnight

20 hrs 8oo:C,15 hrs 8o0:C,20 hrs gOO"F,20 hrs 10O0-C

ovegnight750-C,20 hrs 8oO"C,reground, ^15 hrs 80o;C,20 hrs 90O-F,20 hrs IOOO-C

87 hrs l toooc ,cleaned

192 hrg1 loooc ,cleaned

22 d.*ys1100-c ,cleaned

20 hrs lsoooc

s tr idynite

S tr idymite

S tr idyDite

TrldyEite doesn o t f i t M , Sor MS. Strongc r i I t o b a l i E er e f l e c t i o n s .

S tr idyoite

x (only onegra in obs . )

X (only oneg r a i n o b s . )

x 3 , 4 , 5 , 6

severed isorder

severed isorder

aeveredisorder

x5, and mnYgrains with

severedisordet

one case of

a u u l t i p l e

c-rePeat

x 4 , 5 , a o n e(oo1)stackingf a u l t s

general lydisordered

L i -Tr id 1 :1 f inequarlzpowder:Li2u0q

Na-Tr id 1 :1 f inequarEzpowder:Na2WOq

K-Tr id 1 :1 f inequartz

Powder:K2WO4

M tr idyoitesomec r i s t o b a l i t e

re f lect ions

HS tr idyEite

M t r i d y n i t e ,sore

c r i s t o b a l i t er e f l e c t i o n s

cr i stobal i te

^

X (only oneg r a i n o b s . )

cR- t s i l i c a g e l

2a). Twinning was not obviously present and the reflectionswere almost invariably sharp. Only one diffraction pattern(from a grain of K-Trid) showed signs of streaking alongc*.

A second superlattice had extra reflections along a* di-rections, representing a doubling of the droo spacings of thesublattice (Fig. 2c). This can be described either with aprimitive, psiudohexagonal unit cell having a N 994(:2ar), c ̂ , n8.24 (:ncJ, or with a C-facec€ntered pseu-

doo.thorhombic cell having a - L7 A 1: 2J 3 a"l, b x 9.9 A(:zai, c-^8.2L (:ncH); the c dimensions were eithersome multiple (n) of 8.2A, where observed values of n were3, 4, 5, 6, or could not be specified because of strong streak-

ing along c*. There was typically some c* streaking even

where multiple c-repeat reflections were present (Fig' 2d)'

This superlattice will be referred to as the 2J3arx2a,x ncH type.

In quite a number of cases, Particularly for grains from

TRIV, TRIII, TR-G4 and TR-G3, the diffraction patterns

taken with the electron beam parallel to the c-axis resem-

bled the 2rfiorx zq"x ncH types but with large distor-

tions from orthorhombic cell parameters, having c t

16.6--17.6A, b t 8.3-9.5A and 7 ! 81-88'. Distortions of

such magnitude are too great to be due entirely to some

artefact of the electron microscope (astigmatism, misalign-

sphere in misoriented crYstals'Cristobalite grains were positively identified in TRIV

and TR-G3 from their diffraction patterns and a character-

istic cross-hatched twinning texture (Fig. 3)' which has also

been described by Champness et al. (1971)' A sample of

almost pure cristobalite (CR-l) was examined in the elec-

tron microscope to be sure of this identification'A number o1 grains, particularly in specimens TRIII and

520 CARPENTER AND WENNEMER: SYNTHETI1 TRIDYMITEI

37 36 35 34 33 12 3t 30 29 2S n 6 25 A 23 22 A m 19o2e

A few grains of tridymite from the Steinbach meteoritewere crushed and examined for comparison with the syn_thetic samples. Again the small crystJ fragments seemed toshow more than one superstructure type. Many grains gavediffraction patterns that could be indexed using-the mono_

CARPENTER AND WENNEMER: SYNTHETIC TRIDYMITES

Fig. 2. Selected-area electron diffraction patterns from grains of synthetic tridymite showing additional reflections consistent with

a ,n ' ) l , doo rx2 \ : r f r o ^x2c "pseudohexagona l supe rs i r uc tu re ) (ah )end d . ^ ^x2 -d^^ , xn ( : 2 . / 3a . x2auxnc *pseudoo r tho r -

521

ur lo ^ r , uoo l ^ . \ -v JoH ^ 4LH Povuvv lv

hombic superstructure) (cd). Indexing is based on the high tem

superlattice. Note tripling reflections along (110)*. T-lsa. Arrows

section of the J3dH x 2cr superlattice. Note tripling reflections a

2J3a, x 2a^ x nc, superlattice. Note doubling reflections along (1

;t tOli'ltroriiontal)- section of the zuEa, x 2a^ x ncr superlattice.

i"r.itr. c* repeat is close to n : 5 but may actually be incommensurate;there is also some streaking. T-lsa'

peared, leaving t!" o" = 5.0A, c'x82L sublattice' On

cooling, the 2J3a, x 2a, x ncH superlattice reflections

reappeared (Fig. 6a--c). The heating and cooling cycle could

be followed a number of times until beam degradation set

in. At this point the superlattice reflections failed to reap-

pear and very diffuse streaking between the sublattice re-

flections developed. In some cases the intensity distribution

in the recooled superstructure was not identical to the ini-

tial distribution, suggesting that the transformation is not

necessarily perfectly reversible; for example, the multiplici-

ty (n) of t!: c repeat might change'

The J3a, x 2cH superlattice could also be heated and

5b,e,f). They were notablv free of microstructure.

Beam heating experiment s

It was possible to induce a high/low transformation inthe tridymites by beam heating. Grains were examined ini-tially using a broadly defocussed beam. By carefully focuss-ing the beam the grains could be heated and then recooledsimply by defocussing again. This technique has been usedextensively for the study of phase transformations in sul-phides (Putnis, 1976). Grains with the 2J3a" x 2a, x nc^superlattice could be reversibly transformed; at some rela-tively high temperature the superlattice reflections disap-

522 CARPENTER AND WENNEMER: SYNTHETIC TRIDYMITES

Fig. 3. Typical cross-hatched twinning texture in cristobalite(CR-l). Similar textures were observed in grains from TRIV andTR-c3. Bright field image.

transformed to the high temperature sublattice. However.on cooling, it invariably transformed to the 2uf3a^ x 2a^x ncH superstructure, sometimes with a non_rational

multiple c-repeat (Fig. 6d-f). This could then be reversiblytransformed in the same way as grains that started with the2J 3a x 2a, x ncrsuperlattice.

Grains whose diffraction patterns could not be indexed

(001 ) stacking disorderTridymite grains in TRIII, TRIV, TR_G3 and TR-G4

had severe (001) stacking disorder, as shown by streakingalong c* in diffraction patterns and abundant stackingfaults in bright field images (Fig. 7a). Grains in K_Tridwere generally disordered, but in the other specimens farfewer stacking faults were observed.

-In some grains there appeared to be well ordered regionsa few hundred ingstr<ims wide intermixed with regions ofvery closely spaced stacking faults (Fig. 7b). These grainsmay be intimate intergrowths of more_or_less ordered

tridymite and cristobalite. An attempt to image (001) latticefringes in a JEM l00C electron microscope proved fruitlessbecause of a high rate of beam damage to the specimens inthis instrument.

- The severely disordered grains showed superlattice re_flections of the 2J3arx 2a, x ncH type and- these couldbe removed reversibly by beam heating. Similarly, thegrains with diffraction patterns that could not be indexedreadily as cristobalite or tridymite superstructures, and thatalso had some stacking disorder, showed the doubling be_havior. It appears that stacking disorder is associated withdoubling type superstructures whereas well crystallized, or-dered tridymites tend to have tt

" ,fio, i 2c" type ofsuperstructure.

Electr on inadiation damag eFigure 8 illustrates the effects of radiation damage on the

tridymite structure. The mottled texturo that developed inindividual grains exposed to the electron beam was accom_panied by changes in their diflraction patterns. With ex_posure to the beam, first the superlattice reflections pro_gressively fade away, leaving some diffuse streaking (hig.8a-d). With further exposure even some of the sublattice

cause the grains to be stranded with the sublattice structure

radiation damage.

DiscussionIt should be emphasized at this point that the TEM

observations have two shortcomings. First, since the sam_ples appeared to contain a mixture of low temperatureforms and each electron diffraction pattern gives infor_mation in only two dimensions, the task of reconstructingthe complete reciprocal lattice geometry of each super_structure was not straightforward. proper indexing of someof the possible mixed layer phases was not achieved.Second, the beam heating technique produces temperaturesthat are unknown and are not well controlled. These dilli_culties do not, however, preclude a discussion of some ofthe more general issues raised by the observations.

Flcirke (1955), Sato (1963a) and Hill and Roy (1958) have

C ARP EN T ER AN D W EN N EM ER : SY N T H ET IC T RI DY M I,I ES

Fig. 4. (a{) Selected-area electron diffraction patterns from grain of TRIII shown in (d). (a) Initial diffraction pattern; a* vertical,

[013j* shown in possible twin related orientations. Typical doubling reflections are circled. (b) Same grain as (a) but with focussed, beam

heated conditions; doubling reflections now absent. (c) Same grain as (b) after cooling. Note the reappearance of doubling reflections

(e.g. circled). (d) Bright field image of grain from which (a)-(c) wire taken. The stacking faults visible in this micrograph were not afected

iftne beam leatirig transformations. The spotty texture is due to radiation damage. (e) Typical diffraction pattern from a disordered

grain of TRIV, with multiple reflections and streaking.

523

suggested that some synthetic tridymite samples containmore than one phase. This supposition was supported byShahid and Glasser (1970), who observed tridymite grainswith different optical properties in some of their run prod-ucts, and is confirmed in the present study. We have foundtwo main types-of well defined superstructures, the J34Hx 2c, and ZJ3at x 2at x ncH types. In the latter group

the value of n can vary from grain to grain in a singesample (Table 1). In addition, a third general group ofstructures had severe (001) stacking disorder.

Our .r/3a" x 2cH superstructure appears to correspondclosely to the MX-l structure described by Hoffmann et al.(1983), who used a C-face centered unit cell with a: as,b : Jlar,, : ,r, P :91.5" and incommensurate superlat-tice reflections at approximately h+n213, k, llnU2(n : 1,2).In all our diffraction patterns, however, the .r/3atx 2cr superstructure seems to be commensurate. Hoff-

mann et al. (1983) suggested that MX-l is essentially thesame as S1 of Sato (1964), and the cell dimensions they giveare almost identical to those given by Fl<irke and Langer

(1972) for T-1sa. They also reported that, on heating, MX-1grains transform to pseudo-orthorhombic (PO in the ter-minology of Nukui and Nakazawa, 1980) structures, equiv-alent to the S1=S2 transformation of Sato (1964). Our^,,hU

" 2c" structure also transforms to a PO structure

QJ3a, x 2a, x nc") but the transformation is irreversibleand occurslia the high temperature aH x cH structure.

Ov 2J3a" x 2an x ncH superstructures correspond tothe pseudo-orthorhombic (PO) superstructures of Nukuiand Nakazawa (1980) and are also consistent with the Mand S types of Sato (1963a,b; 1964). Natural terrestrialsamples have essentially the same superstructure, usuallywith n: 5 or 10 (PO-5, PO-10 in the terminology ofNukui and Nakazawa, 1980) (Buerger and Lukesh, 1942;Tagai and Sadanaga, 1972; Gardner and Appleman, 1974;Konnert and Appleman, 1978; Kawai et al., 1978; Nukui etal.. 1980). Nukui and Nakazawa (1980) have summarizedthe full range of values of n reported previously; these aren : 1, 1.5, 2, 5, 6 and 10, to which may now be addedn : 3 a n d n : 4 ( T a b l e l ) .

524 CARPENTER AND WENNEMER: SYNTHETIc TRIDYMITES

Fig' 5' Selected-area electron diffraction patterns from grains of tridymite from the steinbach rneteorite. (a) aE" - c!" section of cctridymite' This diffraction pattern has been indexed on the cc lattice of Kato and Nukui (1976) and the u.roro, u.. parallel to a!"ely. l: odd reflections are absent due to the c-glide and tr: odd are

for orientation relationships between the monoclinic Cc lattio', and,as (a) but from a grain with the ,Eorr 2c" superlattice. Arrows, afi horizontal. Note doubling reflections along c* (e.g., circled). (c)n (b). Superlattice reflections now absent. (d) Cc tridymite. Based on11fi indicated by second arrow which is also cf. (i.e., cr of the Ccote 1). (e) J3a, x 2cH superlattice. Same orientation as (d) showingafter prolonged exposure to the electron beam.

Monoclinic Cc tridymite (:MC structure of Nukui etal., 1978, Nukui and Nakazawa, l9g0) was not found inany of the synthetic samples. Its apparent absence cannothave been due solely to a problem of identification since it

Given these detailed observations of a diversity of super_structures and microstructures in individual tridymite sam_

ples, it is not surprising that the transformation behavior ofsynthetic powders does not correlate exactly with that ob_served in single crystals (Nukui et al., l97g; Thompson andWennemer, 1979). Even sample T-lsa, which Flcirke andLanger (1972) described as a well ordered low tridymite,shows pressure dependent effects for the displacive trans_formations that do not appear to be consistent with the

and Langer, 1972; Schneider and Fl6rke, lgg}), but grainsize may also be important. Since the high/ow transforma-tions are thought to be displacive in character (Buerger,

1951) strain effects probably contribute to the stability and

kinetic accessibility of the most favored structures. Clearlythe strain distribution across a tiny flake of crystals, lessthan a few pm across and only hundreds of ingstrcimsthick, may be different from that in a much larger crystal.It is possible, therefore, that the superstructures we havefound in the synthetic powders are either the products ofgrinding or are merely the low temperature modificationsmost favored by small crystals. The presence of t5a^x 2c" grains in the Steinbach sample is consistent with

this suggestion if they form from the large Cc crystals on

being crushed into smaller fragments. Our failure to ob-

serve intermediate structures in the beam heating experi-ments may further indicate different behavior for very

small grains but could also be due to the lack of precise

temperature control.As has been discussed by many authors (Flcirke, 1955,

1967; Eitel, 1957; Sosman, 1965; Shahid and Glasser, 1970;

52s

Flcirke and Miiller-Vonmoos, 1971) the transformation be-havior of individual tridymite grains also appears todepend on their degree of (001) stacking disorder' Our ob-servations are that well ordered grains may have the uEatx 2c, superstructure, in -contrast with the disordered

grains, which have 2"/3a"x2atxnc" type super-structures. Again, the strain distribution during structuralcollapse is bound to be affected by the presence of (001)stacking faults though variations in the multiplicity of thec4epeat did not correlate obviously with the presence ofstacking faults. Grains that appeared to be intimate inter-growths of more-orJess ordered regions seemed to havetheir own diffraction characteristics, distinct from eithertridymite or cristobalite. The problematical diffraction pat-terns from such grains could be due to a compromise low-temperature distorted structure. Specimen TR-G3, whichcontained material of this type, was used by Thompsonand Wennemer (1979\ for differential scanning calorimetry.

C ARP EN 7: ER AN D W EN N EM ER : SY N T H ET I C T RI DY M IT ES

Fig. 6. Selected-area electron diffraction patterns obtained during beam heating experiments. (af(c) and (d)'-(f) are representative of

two separate sequences. (a) (001) section of the 2rfior*2a^xnc^superlattice, initial state. TRIV. (b) High temperature sublattice.

Same grain as (a) but with highly focussed electron beam. Doubling reflections now absent. Diffuse intensity may be due to thermal

diffuse scattering. (c) Same grain as (a) and {b), but with electron beam defocussed. Doubling reflections have reappeared but not with the

exact initial intensity distribution. (d) c* - [110]* section of the uEa^ x 2c" superlattice. Defocussed beam, initial state. c* vertical'

[110]* horizontal. T-lsa. (e) High temperature form (focussed beam); superlattice reflections absent. This diffraction pattern was not

obtained from the same grain as (d) but is typical of the low-high-low sequence induced by beam heating. (f) Low temperature form,

after defocussing beam. Same grain as (d). Note that [110]* is now doubled and c* shows a multiple repeat.

526 CARPENTER AND WENNEMER: SYNTHETI1 TRIDYMITES

Fig' 7' Mixed layer (?) phases. (a) Bright field image of a grain from TRIV with abundant (001) stacking faults. Inset is the diffractionpattern (c* - [110]*) from the same grain. (b) Bright field image of a grain from TR-G3 showing ordered regons intersp€rsed withregions ofabundant stacking faults' Inset is the diffraction pattern from this grain (a*-<*). Note the streaking along c* and the doublingreflections along ai.

They showed that, in spite of X-ray and infra_red proper_ties appropriate for a mechanical mixture of tridymite andcristobalite, the excess Co effects were distinct from either.The DSC and TEM observations are thus consistent inimplying the presence of mixed-layer phases that have theirown distinctive transformation behavior.

With regard to the influence of impurity atoms, the pro_portions of Na*, K+, and Li+ that can enter the tridvmite

ficient flux action, or from the incorporation of smallamounts of Na+ and Li+ ions, is not clear. It is notablethat Shahid and Glasser (1920) also found that samplesprepared with a K2WO' flux have a high degree of stack_ing disorder.

A variety of low temperaturg superstructures has nowbeen recognized in tridymites, including the monoclinic Ccform, both as natural and synthetic crystals (Fleming andLynton, 1960; Dollase and Buerger, 1966; Hoffmann, l96j:Dollase, 1967; Dollase et al., l97l; Dollase and Baur. 1976:

Kato and Nukui, 1976; Kihara, 1977; Tagai et al., 1977;Kawai et al., 1978); pseudo-orthorhombic types with a :v2J3ar, b x 2a, and with c disordered or some multiple (n)of 8.2A, n : l, 1.5, 2, 3, 4,5, 6, l0 (Buerger and Lukesh,1942; Sato, 1963a,b, 1964; Hoffman and Laves, 1964; Dol-lase et al., 1971; Kawai et al, 1978: Nukui et al., 19g0;Nukui and Nakazawa, 1980; Schneider and Fldrke, l9g2;this stud-y); and a pseudohexagonal form with a x g.7A,c x 16.44 (Hoffmann et al., 1983; this study). Konnert andAppleman (1978) have refined the pseudo-orthorhombicstrucrure with n: 10 under Fl (triclinic) symmetry. It iseven possible to find two different forms coexisting innature (e.9. Kawai et a1., 1978). The complex series of trans_formations that occurs during heating is rarely repro-ducible in different samples (Kihara, 1977; Nukui et al.,1978) and is frequently irreversible (Buerger and Lukesh,1942;Kawai et al., 1978; Thompson and Wennemet, 1979;Wennemer and Thompson, in preparation; this study).Such an association of properties indicates a very delicateenergetic andlor kinetic balance between possible distor-tions of the tridymite framework, and is typical of alter-native transformation behavior, in which an equilibriumstate is not achieved. Metastable modifications that lowerthe energy and are kinetically favored develop instead, and

CARPENTER AND WENNEMER: SYNTHETIC TRIDYMITES 527

Fig. 8. Effects ofbeam degradation. Prolonged exposure to the elcctro] beam causes the disappearance ofsuperlattice reflections andthe development of diffuse scattered intensity. (a) (001) s€ction of the 2J3a, x 2a, x nc, superlattice. TR-G3. Arrows are along [100]*and J010]* of the hexagonal sublattice. (b) Same grain as (a) after exposure to the electron beam for 1-2 minutes. (c) Initial pattern froma J3arx 2crgrain; c* vertical and [110]* horizontal. Steinbach tridymite. (d) Same grain as (c) after exposure; superlattice refl€ctionsnow absent. (e) Same grain as (c) after further exposure. Some ofthe sublattice reflections have now faded, leaving a 4A lone layer) repeatalong c+.

the delicate balance is such as to be influenced readily byany feature, such as heat treatment, grinding, impurity con-tent, defect concentration and associated flux or fluid com-position, which can cause one particular distortion to bemore accessible than a slightly different one. Under thesecircumstances, the understanding of transformation behav-ior in both synthetic and natural specimens requires verythorough sample characterization, and the methods out-lined here may provide a useful means towards this end.Closely analogous issues may arise in the study of transfor-mations in sub-potassic nephelines, which have a stuffedtridymite type of structure (Henderson and Roux, 1977;Henderson and Thompson, 1980).

AcknowledgmentsM.A.C. acknowledges financial support from the Natural Envi-

ronment Research Council of Great Britain, a grant for Europeanstudy leave from the Royal Society and the generous hospitality ofProf. A. B. Thompson during a stimulating visit to E.T.H., Ziirich.We thank Prof. O. W. Fkirke not only for providing sarnple T-lsabut also for his critical comments on the manuscript and Prof. D.

R. Veblen, Prof. C. T. Prewitt and Dr. M. G. Bown for criticismsand discussions, which resulted in the clarification of a number ofissues. The paper is Cambridge Earth Sciences contributionnumber ES 319.

ReferencesAppleman, D. E., Nissen, H.-U., Stewart, D. 8., Clark, J. R.,

Dowty, Eric, and Huebner, J. S. (1971) Studies of lunar plagio-clases, tridymite, and cristobalite. Proceedings of the SecondLunar Science Conference. 1, l1'l-113.

Buerger, M. J. (1951) Crystallographic aspects of phase transfor-mations. In R. Smoluchowski, J. E. Mayer and W. A. Weyl,Eds., Phase Transformations in Solids, p. 183-211. Wiley, NewYork

Buerger, M. J. and Lukesh, Joseph (1942) The tridymite problem.Science, 95,20-21.

Champness, P. 8., Dunham, A. C., Gibb, F. G. F.' Giles, H. N.,MacKenzie, W. S., Stumpfl, E. F., and Zussman, Jack (1971)Mineralogy and petrology of some Apollo 12 lunar samples.Proceedings of the Second Lunar Ssience Conference, 1, 359-3',76.

Cohen, L. H. and Klement, William, Jr. (1980) Tridymite: effect of

528 CARPENTER AND WENNEMER: SYNTHETIC TRIDYMITES

hydrostatic pressure to 6 kbar on temperatures of two rapidlyreversible transitions. Contributions to Mineralogy and pet_rology, T 1,401-405.

Dollase, W. A. (1967) The crystal structure at 220C of orthor-hombic high tridymite from the Steinbach meteorite. Acta Crys_tallographica, 23, 617 -623.

Dollase, W. A. and Baur, W. H. (1976) The superstructure ofmeteoritic low tridymite solved by computer simulation. Ameri-can Mineralogist, 61, 971-978.

Dollase, W. A. and Buerger, M. J. (1966) Crystal structure of somemeteoritic tridymites. Geological Society of America Abstractswith Program, 1966 Annual Meeting, 5zt-55.

Dollase, W. A., Cliff, R. A., and Wetherill, c. W. (1971) Note ontridymite in rock 12021. Proceedings of the Second Lunar Sci-ence Conference, l. 141-142.

Eitel, Wilhelm (1957) Structural anomalies in tridymite and cris-tobalite. A review of the paper with the same title by O. W.Flcirke published in Berichte der Deutschen Keramischen Ges-ellschaft, 32, 369-381 (1955). American Ceramic Society Bulle-tin,36,142-148.

Fenner, C. N. (1913) The stability relations of the silica minerals.American Journal of Science, 36. 331-384.

Fleming, J. E. and Lynton, H. (1960) A preliminary study of thecrystal structure of low tridymite. physics and Chemistry ofGlasses, 1, 148-154.

Flcirke, O. W. (1955) Strukturanomalien bei Tridymit und Cristo-balit. Berichte der Deutschen Keramischen Gesellschaft, 32,369-381.

Fliirke, O. W. (1961) Die Kristallarten des SiO, und ihr Umwand-lungsverhalten. Berichte der Deutschen Keramischen Gesells-chaft. 38. 89-97.

Fl<irke, O. W. (1967) Die Modifikationen von SiO,. Fortschritteder Mineralogie, 44, l8l-230.

Fliirke, O. W. and Langer, K. (1972) Hydrothermal recrystalliza-tion and transformation of tridymite. Contributions to Mineral-ogy and Petrology, 36,221-230.

Fkirke, O. W. and Miiller-Vonmoos, M. (1971) Displazive Tief-Hoch-Umwandlung von Tridymit. Z€itschrift fiir Kristallo-graphie, l33,l9l-202.

Gardner, S. P. and Appleman, D. E. (1924) X-ray crystallographyand polytypism of naturally-occurring tridymite, SiO2. funeri-can Crystallographic Association Program and Abstracts, 2,271.

Gibbs, R. E. (1926) The polymorphism of silicon dioxide and thestructure of tridyrnite. Proceedings of the Royal Society olLondon, A113,351-368.

Henderson, C. M. B. and Roux, J. (1977) Inversions in sub-potassic nephelines. Contributions to Mineralogy and pet-rofogy,61, 279-298.

Henderson, C. M. B. and Thompson, A. B. (1980) The low-temperature inversion in sub-potassic nepheline. American Min-eralogist, 65, 970-980.

Hill, V. G. and Roy, Rustum (1958) Silica strucrure studies VI. Ontridymites. Transactions of the British Ceramic Society, 57,496_5 r0 .

Hobbs, L. W. (1975) Transmission electron microscopy of ex-tended defects in alkali halide crystals. In M. W. Roberts and J.M. Thomas, Eds., Surface and defect properties of solids,volume 4, p. 152-25O. Chemical Society, London.

Hobbs, L. W. (1976) Radiation effects in the electron microscopyof beam-sensitive inorganic solids. In J. A. Venables, Ed.. Devel_opments in electron microscopy and analysis, p.2g7_292. Lca_demic Press, London, New york, San Francisco.

Hoffmann, W. (1967) Gitterkonstanten und Raumgruppe vonTridymit bei 20'C. Naturwissensch aften, 54. ll4.

Hoffmann, W. and Laves, Fritz (1964) Zur polytypie und poly-tropie von Tridynit. Naturwissenschaften, 51, 335.

Hoffmann, W., Kockmeyer, M., Liins, J., and Vach, Chr. (1983)The transformation of monoclinic low-tridymite MC to a phasewith an incommensurate superstructure. Fortschritt€ der Min-eralogie, 61, Bciheft 1, 96-98.

Kato, Katsuo and Nukui, Akihiko (1976) Die Kristallstruktur desmonoklinen Tief-Tridymits. Acta Crystallographic4 B.32, 2486-2491.

Kawai, Kazuhide, Matsomoto, Takeo, Kihara, Kuniaki, and Sa-kurai, Kin-ichi (1978) The first finding of monoclinic tridymitein terrestrial volcanic rocks. Mineralogical Journal, 9,231-235.

Kihar4 Kuniaki (1977) An orthorhombic superstructure of tridy-mite existing between about 105 and 180"C. Zeitschrift fiir Kri-stallographie, 146, 185-203.

Kihara, Kuniaki (1978) Thermal change in unit-cell dimensions,and a hexagonal structure of tridymite. Zeitschrift fiir Kristallo-graphie, 148,237-253.

Kihara, Kuniaki (1980) On the split-atom model for hexagonaltridymite. Zeitschrift fiir Kristallographie, 152, 95-101.

Konnert, J. H. and Appleman, D. E. (1978) The crystal structureof low tridymite. Acta Crysrallographica, 834, 391-403.

Mosesman, M. A. and Pitzer, K. S. (1941) Thermodynamicproperties of the crystalline forms of silica. Journal of theAmerican Chemical Society, 63, 2348-2356.

Nukui, Akihiko and Nakazawa, Hiromoto (1980) Polymorphismin tridymite. Kobutugaku-Zasshi (Journal of the MineralogicalSociety of Japan), 14, Special Volume no. 2, 364-386.

Nukui, Akihiko, Nakazawa, Hiromoto, and Akao, Masaru (1978)Thermal changes in monoclinic tridymite. American Mineral-ogist,63, 1252-1259.

Nukui, Akihiko, Yamaoka, Shinobu, and Nakazawa. Hiromoto(1980) Pressure-induced phase transitions in tridymite. Ameri-can Mineralogist, 65, 1283-1286.

Putnis, Andrew (1976) Alternative transformation behaviour insulfides: direct observations by transmission electron mi-croscopy. Science, 193, 417418.

Sato, Mitsuo (1963a) X-ray study of tridymire (l). On tridymire Mand tridymite S. Mineralogical Journal,4, l15-130.

Sato, Mitsuo (1963b) X-ray study of low tridymite (2). Structure oflow tridymite, type M. Mineralogical Journal, 4,13l-146.

Sato, Mitsuo (1964) X-ray study of tridymite (3). Unit cell dimen-sions and phase transition of tridymite, type S. MineralogicalJournal, 4,215-225.

Schneider, H., and Fl<irke, O. W. (1982) Microstructure, chemicalcomposition, and structural state of tridymite. Neues JahrbuchIiir Mineralogie Abhandlungen, 145, 280-290.

Shahid, K. A. and Glasser, F. P. (1970) Thermal properties oftridymite: 25'C-300'C. Journal of Thermal Analysis, 2, l8l-190.

Sosman, R. B. (1965) The Phases of Silica. Rutgers UniversityPress, New Jersey.

Tagai, Tokuhei and Sadanaga, Ryoichi (1972) Tridymite, f€aturesof its highJow transitions and structure of its 20Jayer polytype.Acta Crystallographica, A28, S121.

Tagai, Tokuhei, Sadanaga, Ryoichi, Takeuchi, Yoshio, andTakeda, Hiroshi (1977) Twinning of tridymite from the Stein-bach meteorite. Mineralogical Journal, 8, 382-398.

Thompson, A. B. and Wennemer, Mechthild (1979) Heat capaci-ties and inversions in tridymite, cristobalite and tridymite_cristobalite mixed phases. American Mineralogist, 64, t0lg_1026.

Manuscript received, December 12, 1983;accepteilfor publication, January 3, 1985