American Mineralogist, Volume 72, pages 1204-1210, 1987

Incorporation of Al, Mg, and water in -A: Evidencefrom speleothems JonN A. Wrsn Department of Geology, La Trobe University, Bundoora, Victoria 3083, Australia Bnr.q.NL. FrNr,.q.vsoN Department of Geography,University of Melbourne, Parkville, Victoria 3052, Australia

Ansrru.cr Opal-A speleothemsfrom a variety of caves show a wide range of composition, con- taining up to 50/oAlrO, and MgO. The infrared spectraof the samplesshow that the Al is substituting for Si, with OH groups compensatingfor the resultant chargeimbalance. The Mg occursas interstitial ions to which OH groups are bonded to balancethe charge.Thus the hydroxyls in opal are presentnot only as surfacesilanol groups,but also as OH groups within the opal structure to balance the charge of the substituted Al and as OH groups bonded to interstitial elementslike Mg.

present opal speleothemswere obtained Iu'nonucrroN For the study, oparisa naruraily occurring hydrous "amorphous" sil- 11"3,il:#[l::il*'i ;Ti'fi J#ffiH"11'f1"fi: ica that forms in a variety of surface and near- Ou"."rfu"d. Severalspeleothems from a small area (Frondel, 1962; Wilding al., surface environments et "ur,oi.uJi, cave were combined to give a single sample, and 1977). Opal from many of these environments has been rn two caves,samples were collected at two different lo- paid studied intensively, but little attention has been to ca[ties within the cave. occurrencesof opal in cavesas secondarymineral O"O.g:- In addition speleothemswere collected from a granite (speleothems). "speleothem" is its In this study, the term cave in southern California, U.S.A.; these samplescon- general precipitated used in a senseto encompassopals tain interlayers of poorly crystallized sepiolite (a hydrous in cavesfrom both high- and low-temperature solutions, -agnesi.rm silicati clay mineral), which is useful for gel i.e., to cover both hyalites and as defined by coriparison with the opal speleothems. Langer and Fliirke (1974). Opal speleothemshave been frequently recorded (Hill Basaltcaves and Forti, 1986),but little detailed work has been carried Muchof westernVictoria is occupiedby a largevolcanic prov- out on them. However, the analytical data available show ince,containing basalts collectively known as the NewerVol- a wider range of composition than otherwise known for canics(Joyce, 1975). The main periodof volcanicityprobably Holo- opal (Cody, 1980; Webb and Finlayson, 1984), so opal^for commencedin the late Plioceneand continuedinto the physiographicfeatures, such speleothemsappear to have considerable potential cene(McDougall et al., 1966),so lavacav,es, are well preserved'More than 50 lavacaves have throwing new light on the way in which variou, -ino. 1s elementsareincorporated intothe opal *',,",,,i". wla- nf,l*:lt:Li:jTil"JilTjiTfiJi l"J,'Jj'ji,?.::T:; ing et al. (1977) noted that impurities in opal may- be ;;;; ; Jntui' opat speleothems:Skipton Cave and Mt. occluded, chemisorbed,or in solid solution, but to date Hu,,'it,o' cu.r.. little work has been done on the relative contributions of The exactage ofthe lavassurrounding these two cavesis un- these three processesto the minor-element composition known,but in eachcase it is likely to be 2.5 Ma or less(Webb, of opal. 1985;J. A. Webb,E. B. Joyce,and N. C. Stevens,unpub. manu- occunnrNcEsoFopAL spELEorHEM" ::lft'ifff'l'}o."ffil'::1"1':fttX'i."i,:T.llt".,'.";:'fi:TJ; areabun- Opal speleothemshave beenfrequently recorded in lava of the southernmostchamber in Mt. Hamilton Cave crystals;these are composed ofopal and cavesand are alSOOCCaSionally present in limestoneCaveS, dantrosettes ofacicular representpseudomorphs after a fibrouszeolite (webb' granite caves, and sandstoneoverhangs (Fith;t;;;;; ?Poffit"ot^-il Webb, 1985). Small irregular coralloids or botryoidat GreatDividing Rangein southeasteueensland is made coatingsare the most common form taken by the speleo- up of 900m of nearlyhorizontal basalts and minor trachytesof thems, although they also occur as stalactites,stalagmites. laleOligocene to earlyMiocene age (Webb et al., 1967).A small and flowstone (Hill and Forti, 1986). Maximum dimen- cave(Holy Jump Lava Cave) is presentin basalttoward the top sionsrarely exceed5 cm. ofthesequence.Thewallsofthecaveareencrustedbysecondary 0003-o04x/87/rrr2-r204s02.00 1204 WEBB AND FINLAYSON: COMPOSITION OF OPAL-A SPELEOTHEMS r205 silica : glass-clearbotryoidal layers of hyalite (opal-A) and milky stalactitesand flowstone of opal-CT and (Webb, 1979).Only the opal-A crustswere examined during the courseofthe presentstudy.

Granite caves The "Granite Belt" in southeastQueensland is a large area of granite terrane that contains a number of caves and sinking streams(Finlayson, 1982).Most of the caveshave formed where a stream has developed an underground passagealong a joint, and two (South Bald Rock Cave and River Cave) have opaline coralloids coating the undersidesof boulders roofing the caves (Webb and Finlayson, 1984). South Bald Rock Cave also con- tains allophane flowstone on the walls and floor. Extensive areasof granite terrane occur in California, and in one of these, about 100 km north-northeast of San Diego, is Cahuilla Creek Cave (Finlayson, 1985). This cave consists of granite boulders roofing a stream. Coralloids occur on the walls of the cave and flowstone on the floor; both speleothemtypes consist largely of banded calcite, although one of the coralloid samplescontains very thin interlayers of poorly crystallized se- piolite.

MrcnosrnucruRE Freshly broken surfacesof all the opaline speleothems were examined under the scanning-electronmicroscope. A variety of microstructuresare evident, including glassy conchoidal fracture surfaces(Fig. lc), irregular colloidal aggregates(Fig. 1a), and in one instancecolloform or bot- ryoidal surfaces(Fig. lb). Similar microstructures have been documented previously for opal-A (e.g., Segnit et al., 1970, 1973).

ANalyttcl.L METHoDS AND RESULTS Chemical analysesof the speleothemswere carried out using X-ray fluorescenceand an electron microprobe, and showedthat severaldifferent minor elementsare present (Table l). Apart from Al and Mg, which will be discussed further below, these elements are present in very small amounts comparable to, or less than, those recorded in opal from other environments (Frondel, 1962; Wilding et al., 1977;Segnit and Jones,pers. comm.). The X-ray diffraction patterns of all the opal speleo- thems are typical of opal-A (Jonesand Segnit, l97l). Langerand Flcirke(1974, p. 18) subdividedopal-A into two varieties:opal-AN (hyalite),with "no remarkablelow Fig. l. Surfacemicrostructures of opal speleothemsunder angle X-ray scattering," and opal-AG, distinguished by the sEM.(a) Colloidalaggregate (8a16-12); coralloid from South "low angle scattering typical for gelJike structures of BaldRock Cave. (b) Colloform surface (8316-10); zeolite pseu- monocrystalline SiOr." The speleothemsamples all have domorphfrom Mt. HamiltonCave. (c) Conchoidal fracture sur- (8516-3); xRD patterns very similar to the latter type, including the face coralloidfrom Mt. HamiltonCave. hyalite specimenfrom Holy Jump Lava Cave. The infrared (rn) absorption spectraofthe opal speleo- thems (Fig. 2) were obtained using pressed KBr discs of any Christianseneffect (seePrice and Tetlow, 1948) in (preparedin the standard manner, e.g., Russell, 1974) the spectra(Fig. 2) shows that the grain size of the sam- and a JascoA302 spectrometer.Each disc contained 0.2 ples was sufrciently small for high-quality spectra to be mg of sample and 200 mg of KBr, both accurately obtained. Since the specimensdid not contain Cl, anion weighed. The samples were dry ground; grinding in al- exchangewith the KBr of the disc could not occur. To cohol is preferablefor crystalline materials (Tuddenham ensure maximum consistency between the spectra, all and Lyon, 1960),but may lead to water loss in amor- samples were run in the one batch, immediately after phous samples(C. Barraclough, pers. comm.). The lack weighing. t206 WEBB AND FINLAYSON: COMPOSITION OF OPAL-A SPELEOTHEMS

3 16-38

h Bald Rock Cave 8416-12

South Bald Rock Cave

SkiptonCave

uJ o 2 f = U)z (r Mt. HamiltonCave F

)u I si-o-A Ar)-oHJ

Fig. 2. Infrared spectraof speleothemsamples. Cahuilla Creek specimenis poorly crystallized sepiolite; all others are opal-A. Assignmentsofabsorption peaks to bond vibrations applicableto all spectra. WEBB AND FINLAYSON: COMPOSITION OF OPAL-A SPELEOTHEMS r207

TABLE1. Chemicalanalyses of speleothems(in wt%)

Loss Si/(Si on +AD Cave Sampleno and type SiO, TiO, AlrO3 Feroo MnO MgO CaO NarO KrO PrOsignition Total (molar) CahuillaCreek' 8316-38; coralloidinterlayers 62.18 0.07 0.32 0.04 0.o5 24.62 1.80 0.02 0.02 0.00 n.d. 89.12 South Bald Rock 8416-12;coralloid 80.71 0.05 5.96 0.37 0.00 0.17 0.09 0.19 0.26 0.40 10.93 99.13 0.93-. South Bald Rock 8316-8;coralloid 81.44 0.04 5.55 0.40 0.00 0.10 o.12 0.16 0,28 0.67 10.64 99.47 South Bald Rock- 8316-8;coralloid 86.64 0.00 4.41 0.02 0.00 0.03 0.13 0.09 0.50 O.12 n.d. 91.94 0.94 River 8316-7;coralloid 69.70 0.08 9.70 2.54 0.01 0.04 0.07 0.23 0.27 1.66 15.61 99.91 River. 8316-7;coralloid 82.42 0.02 8.41 0.06 0.06 0.04 0.06 0.01 0.05 1.2'l n.d. 92.34 0.93" Skipton 8516-4;coralloid 89.740.11 2.30 0.65 0.00 0.10 o.21 0.2s 0.10 0.21 6.18 99.85 0.971 HolyJump Lava 8616-19;botryoidal crust 93.08 0.01 1.02 0.53 0.00 0.28 0.06 0.16 0.06 0.11 3.78 99.09 0.99 Mt Hamilton 8516-3;coralloid 81.79 0.05 0.64 0.35 0.00 4.71 329 0.13 0.07 0.30 8.02 99.35 0.99 Mt Hamilton- 8316-10; zeolitepseudomorph 92.51 0.09 0.00 0.00 0.05 0.07 0.04 0.01 0.o2 0.00 n.d. 92.79 1.00 .Indicates electron-microprobeanalyses; all other analysesare xRF(n.d, not determined).Detection limit on xRFanalyses is 0.01% or better; on microprobeanalyses, it is 0.04% or better. -'Si/(Si + Al) molarratio adjustedusing additional EDs and NMRdata; see discussionin text. t Minimumvalue because of clav contamination.

The spectraprovide interesting data on the short-range (0.94; Table 1). These results show that the allophane (nearestneighbor) structure in the opal and show some content of the whole speleothemsvaries from about l0lo differencesfrom the opal spectra obtained by previous (South Bald Rock Cave) to approximately 50/o(River workers,e.g., Moenke (1974), Wells et al.(1977), Wilding Cave). et al. (1977), Plyusnina (1979). These differencesare re- In order to check the amount of allophane in the spe- lated to the minor-element composition of the opal, as leothems, two sampleswere analyzedby "Al Nvrn. This discussedbelow. technique determines the relative proportion of tetrahe- drally coordinated and octahedrally coordinated Al in a Al rNr opAl, specimen.Allophane contains both types of Al in varying Previous studies have reported AlrO3 levels of up to proportions(Farmer, 1985,pers. comm.); if the propor- 4.70/oin soil opal (Wilding et al., 1977} Curtiss et al. tion is known, then the amount of allophane in a sample (1985)and Segnitand Jones(pers. comm.) recorded opals can be calculatedfrom the 27AlNvrn results. with I 7 to 22o/oAlrOr, but thesecontained large amounts Analysis of one of the opal samplesfrom Mt. Hamilton of admixed surfacesoil and clay. Cave (85 16-3) showedthat it contained only tetrahedral- In the presentstudy, the highestAlrO, levelswere found ly coordinatedAl and lacked allophanecompletely. How- in samples from the Queenslandgranite caves, ranging ever, one of the opal coralloids from South Bald Rock from 4.4 to 9.7o/oAlrO, (Table l). Part of this AlrO, is Cave (8416-12)has ll0lo of its total Al octahedrallyco- due to inclusions of feldspar and allophane (an amor- ordinated.xvn analysesof allophanedeposits in the same phous hydrous aluminosilicate clay mineral). Very little cave show that about 700/oof their Al is octahedrally co- feldspar is present in thesespecimens, as shown by thin- ordinated. Assuming that the allophane in the opal con- section examination and the very low Na, Ca, and K tains octahedrallycoordinated Al in the sameproportion, levels in the analyses.On the other hand, allophane is this means that of the 60loAlrO. in the opal (Table l), common as very fine grained inclusions concentratedin approximately 10/ois present as allophane (comparable dark, cloudy growth bands separatedby relatively clear with the other South Bald Rock sample).The remaining interlayers. 50/ooccurs as tetrahedrally coordinated Al within the opal, In an effort to obtain opal analysesrelatively free from giving a molar Si/(Si + Al) ratio of 0.93. This result com- allophane contamination, the transparent layers in two pares closely with the ratios obtained above for allo- specimenswere analyzedby electron microprobe (Table phane-freeopal from the same area. l). The analysis of the South Bald Rock Cave sample Confirmation of theseresults comes from the rR spectra (8316-8)appears to represent"pure" opal, but allophane of the speleothems(Fig. 2). At about 550-560 cm-' there is present in the River Cave analysis, as shown by the is a small peak or weak shoulder on most spectra,which relatively high phosphatecontent (allophane from these representsthe vibration of octahedrally coordinated Al cavesis phosphaterich; Webb and Finlayson, 1984).In in the gibbsite sheetsofclays, e.g.,kaolinite or allophane an effort to obtain a better result, a number of clear in- (Stubican and Roy, 196l; Farmer et al., 1979). Thus the terlayers in the latter sample were analyzedby quantita- presenceofthis peak in the opal spectraindicates clearly tive energy-dispersivespectrometry (Eos). The maximum that an occluded clay phaseis present in many samples. SiOr/AlrO3 weight percentageratio obtained was 16.4, The River Cave opal, with the highest allophane content which converts to a molar Si/(Si + Al) ratio of 0.93 (using (-50lo),has the most prominent 560-cm-' peak (Fig. 2). the SiO, content obtained in the microprobe analysis). The South Bald Rock samples contain less allophane This ratio is comparable to that obtained for the allo- (- to7oland have smaller 560-cm ' peaks.The Mt. Ham- phane-freelayers in the South Bald Rock Cave sample ilton coralloids(8516-3), which accordingto the 27AlNun 1208 WEBB AND FINLAYSON: COMPOSITION OF OPAL-A SPELEOTHEMS results lack clay inclusions altogether,have no 560-cm ' speleothemsfrom the Queenslandgranite caseshave low peak. The presenceof a 560-cm ' peak in the Skipton levels of interstitial ions and must contain silanol groups Cave spectrum indicates clearly that this sample has a to balance the charges.There is evidence for this in the small but signiflcant clay content. rR spectra(Fig. 2). The 950-cm-' peak in the opal IR spec- One of the Mt. Hamilton samples(8316-10) does not trum has been ascribed to the Si-O stretching of silanol appear to fit this pattern; the rR spectrum has a marked goups (Moenke, 1974). In the present caseit probably 560-cm 1peak, yet the analysisshows that Al, and hence also contains a contribution from AIOH groups; micas occluded clay, is absent. However, in this casethe sam- and clays have well-defined AIOH peaks in this region, ples were collected from fallen material on the floor of due to the in-plane rocking vibration (libration) of hy- the cave and were as a result partially coated with clay, droxyls linked to two Al ions (Stubicanand Roy, 196l; which proved impossible to remove completely. Thus the Russellet al., 1970).Thus all speleothemspectra with a sample crushedfor n analysiscontained a small amount 560-cm-r peak due to clay inclusions also have a small of clay, and this can be seen on the rR spectrum. The 950-cm Ipeak or shoulder.However, in the high-Al sam- chemical analysiswas obtained using an electron micro- ples, particularly those from South Bald Rock Cave, the ' probe; care was taken to keep the beam well away from 950-cm ' peak is much largerthan the 560-cm shoul- the clearly visible clay coating, so the analysis shows no der. This must be due to the contribution of silanol groups evidenceofclay. within the opal structure, acting as charge balancesfor The rn spectrain Figure 2 also confirm the presenceof the substituted Al. The speleothems with very weak tetrahedrallycoordinated Al substitutingfor Si in the opal 950-cm-' shoulders,e.g., Holy Jump Lava Cave,have few samples.The very strongpeak at 1090-1100 cm-' is due silanol groups;apparently the small amount of substituted to antisymmetric Si-O-Si stretchingvibrations (Moenke, Al presentis balancedby interstitial monovalent and di- 1974).In tectosilicatesthe wave number of this peak de- valent ions. creasesregularly with increasing substitution of tetrahe- dral Al into the framework of the mineral (Milkey, 1960), Mg rN orar, and this effect can be seenin Figure 2. The Queensland Most previously described opals have MgO contents granite cave sampleshave the highest AlrO. contents of of lessthan 2olo(Wilding et al., 1977).However, the opal the speleothems(molar Si/(Si + Al) ratios of 0.93 to 0.94), variety menilite, which occurs as nodules in limestones and their peaksare at about 1090 cm '. The basalt cave and dolomites, typically contains 7 to 8o/oMgO (Damour, speleothemshave Si/(Si + Al) ratios of 0.97 or greater, 1884),and Cody (1980)recorded an opal speleothemfrom and their peaks are at around I 100 cm r. a New Zealand lava cave with 140/oMgO. Unfortunately, Thus this study has shown unequivocably that the Al there are no rn data for any of thesehigh-Mg opals, so it content ofopal is due to occlusionofclay particlesas well is impossible to compare them in detail with the present as substitution of Al for Si. Many authors have proposed specrmens. or assumedthat Al can substitutefor Si in the silica min- The highest MgO content of the opal speleothemsana- erals (e.g.,Dennen, 1966; Frondel, 1962; Flcirkeet al., lyzed in this study is 4.70/o(Mt. Hamilton coralloids, 1982), as both elementscan occur in tetrahedral coordi- 8516-3:Table l). Examination of thin sectionsof these nation and they have similar ionic radii, but direct con- speleothemsusing nos identified calcite inclusions, hence firmatory evidencehas been lacking. explaining the CaO content of the sample,and confirmed Replacementof Si4*by Al3* results in a chargeimbal- that the Mg is uniformly distributed through the opal. ance that must be compensatedby the entry of mono- The Mg is more likely to be interstitial than substitut- valent or divalent interstitial ions (e.g., Li, N4 K, Mg, ing for silica, as Mg has alarger ionic radius and a lower Ca) or by the substitution ofa hydroxyl ion for an oxygen charge than Si or Al. Spacesin the poorly ordered opal to form silanol(SiOH) groups(Dennen, 1966; Wilding et structure can easily accommodateions of this size,as the al., 19771,Flcirke et al., 1982).The speleothemsamples well-ordered structures of a-, a-, and with high Si/(Si + Al) ratios (0.97 or greater) appear to a- all contain gapslarge enough to allow Mg to contain sufficient interstitial ions to match, or almost enter (Eitel, 1964). match, the charge imbalance resulting from the Al sub- If the interstitial Mg is more abundant than the ions stitution (exactcalculations are impossible becauseof the substituting for Si, then there will be a chargeimbalance, presencein some samplesof very small amounts of oc- which must be countered by the introduction of OH cluded feldspar). Similar results have been reported for groups chemically bonded to the interstitial ion rather microcrystalline quartz of volcanic origrn (Fldrke et al., than the Si of the opal. The IR spectrum of the Mt. Ham- 1982).Analyses of the latter samplesshowed a direct cor- ilton coralloids(8516-3, Fig. 2) providesstrong evidence relation between the content of interstitial ions (Ca * that this is indeed the case.The spectrum has a small but Mg + K * Na * Li) and that of substituted ions (Al + definite peak at about 660-cm-', which is not present in Fe), implying that few, if any silanol groups were incor- any of the other opal spectra. The hydrous magnesium porated into the quartz structure to balancethe charges. silicate talc possessesa prominent n peak at approxi- However, this conclusion is definitely not applicable to mately 670 cm-', representingthe in-plane vibrations of some of the opal speleothemsunder study. The high-Al hydroxyls coordinated to three Mg atoms (Russell et al., WEBB AND FINLAYSON: COMPOSITION OF OPAL-A SPELEOTHEMS r209

1970).The rR spectrumof sepiolite,a hydrous magne- Water in opal occursas molecular water, silanol (SiOH) sium silicate clay mineral, has a peak in a comparable groups,and hydroxyls bound to interstitial ions. The mo- position (650 cm-l; Nagataet al., 1974), apparentlydue lecular water is presentas capillary condensationin pores to a similar Mg-OH vibration. Thus the presenceof this w'ithin the opal (Segnitet al., 1965)and as isolatedmol- peak in the Mt. Hamilton spectrum indicates clearly that eculestrapped in small intersticeswithin the silica matrix the Mg in the sample is directly bonded to hydroxyls. (Langer and Florke, 1974). The silanol groups exist both The Cahuilla Creek sample is a poorly crystallized se- as a monolayer ofhydroxyl groupsattached to the surface piolite. It has a molar Mg/Si ratio of 0.59, closeto that Si atoms (i.e., a surfacesilanol layer; Iler, 1955)and as of sepiolite(0.61-0.67; Wollast et al., 1968),but irs bi- isolated SiOH groups within the SiO, framework, com- refringence is much lower than that of sepiolite, and it pensating for the charge imbalance causedby ions sub- lacks well-defined xno peaks. Nevertheless,the rR spec- stituting for Si, as discussedpreviously. The presentstudy trum (Fig. 2) is very similar to that of sepiolite, particu- has also shown that water in opal may be presentas OH larly in the 450-900-cm ' region. The major Si-O peak groups bound to interstitial ions, particularly Mg, when in the sepiolitespectrum is at 1020cm-', with a smaller these ions are more abundant in the opal than ions sub- peak at about 1070 cm '; the shapesand positions of stituting for Si. thesepeaks can vary slightly (compareWollast et a1.,1968; Thus the minor-element content of an opal is of prime Nagataet al., 1974;Yeniyol, 1986),probably due to dif- importance in determining the proportion of OH groups ferencesin composition. The very broad profile of these in the opal, bound to both Si and interstitial ions. For two peaks in the Cahuilla Creek spectrum is similar to example, the high-AlrO. samplesfrom South Bald Rock that figured by Wollast et al. (1968), although the Ca- have abundant silanol groups (as shown by the strong huilla Creek peaks are at slightly higher wave numbers 950-cm-' rn peakson Fig. 2), in this caseformed to charge- (1030and 1090cm '). balance the substituted A1. On the other hand, the Mt. Comparison of the Cahuilla Creek and Mt. Hamilton Hamilton coralloids(8516-3) have very few silanol groups (8516-3)rR spectrain the 1000-1100-cm I region raises (no significant950-cm ' peak),but the 660-cm I MgOH an interesting point. There is a marked shoulder on the peak shows clearly that hydroxyl groups are present in Mt. Hamilton spectrumat about 1030 cm r. By analogy the structure, in this casebonded to the interstitial Mg. with the Cahuilla Creek spectrum, this could be due to The 1630-cm-r rn peak results from the H-O-H bend- Si-O vibrations of a magnesium silicate phasewithin the ing vibrations of molecular water (Langer and Florke opal; the Si-O vibrations ofthe opal itselfare represented 1974),so the strength ofthis peak reflectsto some extent by the very strong I100-cm-' peak. If this assignmentof the content of molecular water in the sample. All the peaks is correct, it indicates that the Mg in the opal is speleothemshave definite 1630-cm-' peaks(Fig. 2) and presentnot as individual interstitial ions, but as hydrous so contain molecular water. The River Cave speleothem magnesium silicate domains of sufficient size and abun- has the largest peak, probably due to the contribution dance to be evident on the rn spectrum. nos scanningof from molecular water in the allophane inclusions. thin sectionsof the Mt. Hamilton coralloids showedthat Langer and Fldrke (1974) used near-infrared spectrato the Mg distribution in the opal is quite uniform, so if the estimatethe relative proportions of molecular and silanol above hypothesisis correct, then the magnesium silicate water in opal samples.They found that, in general, hy- domains must be relatively small and randomly distrib- alites had the highest percentageofsilanol groups (up to uted. 400/oof total water),whereas opal-AG specimenshad only The incorporation of interstitial Mg within the opal-A l0 to l5oloof their water as silanol groups. Langer and structureis not unexpected,as Mg(OH), displaysa strong Fldrke (1974) did not considerthe minor-elementcon- afrnity for silica and can flocculate undersaturatedsilica tent of the opals when attempting to explain this differ- solutions (Williams and Crerar, 1985).Because Mg(OH), ence. Instead they suggestedthat the temperature of the carries a positive surfacecharge under most natural con- depositing solution might have been a significant cause. ditions, it can neutralize the negative surface charge of Numerous OH groups could be more easily incorporated silica particles in solution and allow them to precipitate. into the hyalite structure because it is deposited by The neutralizing cations are readily incorporated into the quenching from high-temperaturesolutions. growing surfaceofthe silica precipitate. However, in the present study, the single hyalite sam- ple (Holy Jump) has very few silanol groups (as shown Wlren rN opAL by the very weak 950-cm ' peak), and, in fact, one of Previous studieshave shown that opal usually contains Langer and Fl6rke's hyalite specimensalso contained a between3olo and I lolowater (Segnitet al., 1965;Wilding low proportion of silanol groups (about l0o/o of total etal., 1977).Allophane-free opal in the speleothemsam- water). Furthermore, several of the opal-AG speleo- ples contains up to 8o/owater (microprobe analyses in thems, particularly those from South Bald Rock Cave, Table l). The loss on ignition for the granite-cavespeci- have a much greater proportion of silanol groups. The mens is greater than this (up to 150/o)because of the oc- presentstudy has shown that minor elementsplay a ma- cluded allophane, which contains about 400/owater and jor role in determining the hydroxyl content within the organic compounds (Webb and Finlayson, 1984). opal structure, and this effect is apparently sufficient to l2l0 WEBB AND FINLAYSON:COMPOSITION OF OPAL-A SPELEOTHEMS override any diflerences due to the temperature of the Milkey, R G. (1960) Infra-red spectra of some tectosilicates.American solutions forming the opal. Mineralogist,45, 990-1007. Moenke,H.H W. (1974)Silica, the three-dimensionalsllicates, borosili- catesand beryllium silicates.In V.C. Farmer, Ed., The infra-red spectra AcxNowr,eocMENTS of minerals. Mineralogical SocietyMonograph, 4, 365-382. Nagata,H, Shimoda,S, and Sudo,T. (1974)On dehydrationofbound We expressour appreciation to Maunu Haukka and David Sewell of waterofsepiolite. Clays and Clay Minerals,22,285-293. the Geology Departmenl, Melbourne University, Max Beyer and Colin Plyusnina,I.l. (1979) Infra-red spectraof opals. Soviet PhysicsDoklady, Barracloughof the Chemistry Department, Melbourne University, and 24,332-333. Dr. David Cookson,B H.P. ResearchLaboratories, Melbourne, for their Price, W C , and Tetlow, K.S. (1948) Infra-red Christiansenfilter effectin assistancein the analysisofthe samples.Dr. Ralph Segnitkindly provided slurries in organic crystals. Joumal of Chemical Physics, 16, ll57- data from a forthcoming paper on the composition ofopal and also made tt62 helpful suggestionsduring the courseof this study. Russell,J D. ( I 974) Instrumentation and techniques.In V C Farmer, Ed , The infra-red spectraof minerals. Mineralogical Society Monograph, RrrBnnNcns crrno 4, ll-26 Russell,J D., Farmer,V.C., and Velde,B (1970)Replacement of OH by Cody, A (1980) Stalagmitic depositsin Auckland lava caves.New Zea- OD in layer silicates,and identification ofthe vibrations ofthese groups land SpeleologicalBulletin, 6, 337-343 in infra-red spectra.Mineralogical Magazine,37, 869-879. Curtiss,B, Adams,J.B., and Ghiorso,M.S. (1985)Origin, development Segnit,E R , Stevens,T.J , and Jones,J.B. (1965)The role of water in and chemistry of silica-alumina rock coatings from the semi-arid re- opal.Journal ofthe GeologicalSociety ofAustralia, 12,2ll-226. gions of the island of Hawaii. Geochimica et Cosmochimica Acta, 49, Segnit,E.R., Anderson,C.A., and Jones,J.B. (1970)A scanningmicro- 49-56 scopestudy ofthe morphologyofopal. Search,1, 349-351. Damour, M.A. (1884)Essais chimiques et analysessur la menilite.Bul- Segnit,E.R., Jones, J.B., and Anderson,C.A. (1973)Opaline silicas from letjn Soci6t6min6ralogie de France,7,219-241. the Murray River region west of Wentworth, N.S.w., Australia Mem- Dennen,w H (1966) Stoichiometric substitution in narural quartz Geo- oirs ofthe National MuseumofVictoria, 34,187-194. chimica et CosmochimicaActa,30, 1235-1241. Stubican,V, and Roy, R. (1961)Isomorphous substitution and infrared Eitel, W (1964) Silicate science.Vol. l, Silicate structures.Academic spectraofthe layer silicates.American Mineralogist, 46,32-51. Press,New York. Tuddenham,W.M., and Lyon, R.J.P.(1960) Infra-red techniques in the Farmer, V C., Fraser,A.R., and Tait, J.M (l 979) Characterizationof the identification and measurementof minerals. Analytical Chemistry, 32, chemical structuresof natural and synthetic aluminosilicate gels and l 680-l 684. sols by infrared spectroscopy.Geochimica et Cosmochimica Acta,43, Webb, A W , Stevens,N.C., and McDougall, L (1967) Isotopic agedeter- 1417-1420. minations on Tertiary volcanic rocks and intrusives of south-eastern Finlayson, B L. (1982) Granite cavesin Girraween National Park, south- Queensland.Proceedings of the Royal Society of Queensland,7 9, 79- eastQueensland. Helictite, 20, 53-59. 92. -(1985) The formation of cavesin granite.In M.M. Sweetingand Webb, J.A. (1979) Morphology and origin of Holy Jump Lava Cave, K. Paterson,Eds., New directions in karst. Geobooks,Norwich. southeastemQueensland. Helictite, 17, 65-74. Finlayson,B.L., and Webb, J A (1985)Amorphous speleothems.Cave -(1985) Acicular opalinespeleothems from Mt. Hamilton lava cave, Science,12, 3-8 westernVictoria. Australian Mineralogist, 49, 290-294 Fliirke, O.W., Kohler-Herbertz,B, Langer,K., and Tdnges,I. (1982) Webb, J A., and Finlayson, B.L. (1984) Allophane and opal speleothems Water in microcrystalline quartz of volcanic origin: .Contribu- from granite caves in south-eastQueensland. Australian Journal of tions to Mineralogy and Petrology, 80, 324-333. Eanh Sciences,31, 341-349. Frondel, C (1962) Dana's systemofmineralogy. Vol. 3, Silica minerals. Wells, N., Childs, C.W., and Downes, C.J (1977) Silica Springs,Tonga- Wiley, New York. riro National Park, New Zealand-Analysis of the spring water and Hill, C.A, and Forti, P. (1986) Cave mineralsof the world. National characterisationof the alumina-silicate deposit. Geochimica et Cos- SpeleologicalSociety, Huntsville, Alabama. mochimicaActa. 41. 1497-1506. Iler, R K. (1955) The colloid chemistry of silica and silicates. Comell, Wilding,L P., Smeck,N.E , and Drees,L.R. (1977)Silica in soils:Quartz, Ithaca. New York. cristobalite, tridymite and opal In J B. Dixon and S B. Weed, Eds., Jones,J.B., and Segnit,E R (1971)The natureofopal. l. Nomenclature Minerals in soil environmefis, p. 471-522 Soil Society of America, and constituentphases Journal ofthe GeologrcalSociety ofAustralia, Madison r 8. 57-68 Williams, LA., and Crerar, DA. (1985) Silica diagenesis,II. General Joyce, E B. (1975) Quatemary volcanism and tectonics in southeastern mechanisms.Joumal of SedimentaryPetrology, 55, 312-321. Australia. In R.P. Suggateand M.M. Cresswell,Eds., Quatemary stud- Wollast,R., Mackenzie,F.T., and Bricker,O P. (1968)Experimental pre- ies,p. 169-176.Royal SocietyofNew Zealand,Wellington. cipitation and genesisofsepiolite at earth-surfaceconditions. American Langer, K., and Flirrke, O.W. (1974) Near infrared absorption spectra Mineralogist, 53, 1645-1662 (4,000-9,000cm') of opalsand the role of "water" in theseSiOr. nHrO Yeniyol, M (1986) VeinJike sepiolite occurrenceas a replacement of minerals. Fortschritte der Mineralogie, 52, 17-51. magnesitein Konya, Turkey. Clays and Clay Minerals, 34,353-356. McDougall,I., Allsopp,H.L, and Chamalaun,F.H. (1966)Isotopic dat- ing of the Newer Volcanics of Victoria, Australia, and geomagnetic MeNuscnrsr RECETVEDJexunnv 2, 1986 polarity epochs.Journal of GeophysicalResearch, 7 1, 6 107-6 11 8. M-rrNuscnrrrAccEprED JttNs 26, 1987