Mineralogical Magazine, February 2003, Vol. 67(1), pp. 1–13

LaserRaman identification of silica phases comprisingmicrotexturalcomponents of sinters

1 2, K. A. RODGERS AND W. A. HAMPTON * 1 Research Associate, Australian Museum, College Street, Sydney, NSW,Australia 2 Department of Geology, University of Auckland, Private Bag 92019, Auckland, NewZealand

ABSTRACT LaserRaman spectroscopy is particularly well suited for routine characterization of the crystalline and thebetter ordered paracrystalline silica phases occurring in silica sinters: opal-C, quartz and moganite. Thenon-crystalline and poorly ordered paracrystalline silica phases (opal-CT and opal-A) are identified withdifficulty, partly as a resultof the high level of fluorescence exhibited by the younger sinters that aredominated by these phases, and partly as a resultof the ill-defined nature of the broad phonon scatteringbands produced by them. Nonetheless, given the limited number of phases present and the highlevel of phonon scattering from the better ordered phases, judicious use of a Ramanmicroprobe enablesthe character of most siliceous microtextural sinter components to be determined readily. Microprobeexamination of a seriesof New Zealandsinters, ranging from Late Quaternary to Pliocene inage shows that a silicaphase inhomogeneity that exists in some outcrops reflects an underlying phaseheterogeneity obtaining in the microstructure of the sinter; it is a consequenceof the phase transformationprocess. In the older sinters, quartz, moganite and opaline cristobalite may be present in asinglefabric element but in quite different proportions to their abundance in adjacent elements. Identificationof opaline lepispheres partially pseudomorphed by quartz+moganite and the association ofquartz+moganite in microcrystals exhibiting common quartz forms, cautions against the use of morphologyas a meansof identifying silica phases at the microstructural level.

KEY WORDS: silica,sinter, laser Raman spectroscopy, opal, New Zealand.

Introduction thatthrive in the hot-spring environments, and hencethe resulting sinters furnish invaluable SURFACE depositsof silica sinter are a common informationabout mineral-microbe associations expressionof geothermal systems. They are a inhydrothermalsystems (e.g. Walter et al., 1996; signatureofthehydrologicalconditions Farmer,2000). Sinters are a remarkable,persis- prevailingat the time the sinter  occulatedas tent,terrestrial, sedimentary deposit. opalinesilica from near-neutral alkali chloride hot WhenŽ rstdeposi ted,silic ainsinter is springsderived from deep reservoirs hotter than characteristicallynon-crystalline opal-A. With 175ºC (Fournierand Rowe, 1966). Consequently, timethe sinter crystallizes to paracrystalli ne sintersprovide an important exploration guide to opal-CT±opal-C and, subsequently, recrystallizes locatingand interpreting geothermal systems tomicrocrystallinequartz, with moganite forming whosesurface activity has changed, declined or asa transitoryphase (Herdianita, et al., 2000a; evenceased. Further, the texture of the freshly Rodgersand Cressey, 2001). This phase matura- depositedopaline silica is determined, in part, by tionis accompanied by microtextural changes. templetsafforded by the microbial communities Thecomplete transforma tioncan take from 30,000to 50,000 years and is parallelled by changesin thedensity, porosity and water content *E-mail: [email protected] ofthe successive silica phases (Herdianita et al., DOI: 10.1180/0026461036710079 2000a).Thecorrect identiŽ cation of the different

# 2003The Mineralogical Society K. A.RODGERSAND W.A. HAMPTON silicaphases is essential to elucidating sinter Herdianita et al. (2000a)andRodgers and history. Cressey(2001)report edsomeprelim inary X-raypowder diffraction is well suited to Ramanscans of opal-A, opal-CT and opal-C, characterizingeach of the different crystalline thesilica phases common among late Pleistocene silicaminerals (Smith, 1997, 1998). DifŽ culties andRecent sinters, they made no systematic exist,however, in using convention alX-ray study.The present report describes the results of a powdertechniques to identify opal-CT, with its Ramansurvey of all sinter specimens used by limitedlattice order, in the presence of opal-C Herdianita et al. alongwith some additional New whichpossesses an enhanced but still restricted Zealandexamples, particularly from quartz-rich crystalstructure (Smith, 1998; Herdianita et al., Tertiarydeposits. These spectra are presented in 2000b; Campbell et al., 2001).Further, the theirraw, unvarnished form in order to demon- progressionin phase transfo rmationis not stratethe strengths and weaknesses of the laser uniformwithin an outcrop, a fabricor textural Ramanmicroprobe as aroutinedeterminative tool type,or evena singlehand specimen. In orderto inmicrotextural sinter studies. The results prove fullyascertain the stage in maturationthat a sinter sufŽcient to demonstrate that a silicaphase hasattained and, importantly, to gain an under- inhomogeneityofte nobservedin a sinter standingofthe process esinvolve dinthat outcropalso pertains at the microscopic level. transformation,it is desirable to determine the phasecomposition of individual morphological componentsof different sinters. This is not Samples possiblewith conventional X-ray powder techni- Herdianita et al. (2000a)reportedon 29 samples ques.The standard 20 61061mm aluminium ofsilica sinter from ten different active and diffractionsample holders require 200 –300 mg extinctgeothermal systems, speciŽ cally collected ofpowder .Thisis some 5 –10 orders of toprovide a rangeof ages and occurrences of magnitudegreater than the mass of many of the typicalthin-bedded moderate- to low-temperature sinterfabric elements .Clearly,each X-ray sinters.Less common textural types and silica powderscan serves to identify only the dominant residuewere excluded. Mineral phases other than phasespresent in several million microtextural silicawere rare. Twenty one were from active and components.Phases present at low levels tend to extinctQuaternary geothermal Ž eldsof theTaupo beoverlooked as has proved to be the case with VolcanicZone (TVZ). Fivecame from the long- moganite(Smith, 1998; Rodgers and Cressey, livedbut still active Ngawha Ž eldof northern 2001).Heaney and Post (1992) found it necessary New Zealand,and three from Tertiary occur- toundertake Rietveld reŽ nement of their data rencesin North America (Hausbr ouckand afterhaving counted for 2 –10 s/0.03º2y from 2 to Lincoln,Montana). All had been deposited from 80º2y,i.e.for 1.4 –7.2h persample, in order to near-neutralalkali chloride waters. None had conŽrm the presence of moganitein microcrystal- sufferedsubsequent heating or steamalteration as linequartz. Such requirements are not conducive faras couldbe ascertained.All were examined in forroutine analyses of silica-rich samples and an thepresent study. Specimen details are given in alternativemethod to X-raypowder techniques is Herdianita et al. (2000a),althoughthe ages of requiredif the mineralogy of individual sinter someof theirsamples have recently been revised. microtexturesis to be determined and the inter- Inaddition, the present study includes samples relationshipsbetween the different silica phases fromthe quartz -richmid- to late-Pl iocene comprehendedat a microscopiclevel. Whenuaroasi nterat Plumduf fandMount Theroutine identiŽ cation of theprincipal silica Mitchell,Northland, two of three southernmost specieswith Raman microspectroscopy has been sintermasses of the fossil Puhipuhi geothermal demonstratedby Kingma and Hemley (1994) in Želd,aswellasfromlatePleistocene, thecourse of using a Ramanmicroprobe to >40,000BP, sintersof Umukuri (Campbell et characterizemoganite in microcrystalline quartz al.,2001)and Atiamuri, from tephra-rich sinter samples.Subseque ntly,Go ¨tze et al. (1998) remnantscapping hillocks in the Otamakokore employeda Ramanmicroprobe to explore the andWhirinaki stream valleys, and from the mid- distributionof moganite in agate while Allen et TertiaryWaitaia Sinter of Coromandel (Skinner, al.(1999)used Raman techniques successfully to 1976).Conventional X-ray powder diffraction characterizethe aggulinated silica (and titania) showedall these additional specimens to contain particlesof marine foraminifera tests. Although highproportions of opaline cristobalite (opal-CT

2 MICROTEXTURALCOMPONENTSOF SINTERS

and/oropal-C) and/ orquartz. Rodgers and non-crystalline,paracrystall ineand crystalline Cressey(2001) had demonstrated that moganite microtexturalcomponents. The crystalline silica wasintimately associated with microcrystalline phases,quartz, a-cristobalite, a-tridymiteand quartzin both the lower horizons of Umukuri, moganiteall possess distinctive, intense scattering quartz-richareas of the Otamakokore erosional bandsin the region 350 –550 cm – 1 as do the residuals,and some older Atiamuri sinters. betterordere dofthe paracr ystallineopals Referencesamples included an unlocate d, (Fig. 1). colourless,transparent quartz macrocrystal from Initialexperiments focused on collecting the Switzerlandand tridymite from Vechec Quarry, Ramanspectra of different textural portions of EasternSlovakia (AM D52115).A suitable eachsample from 150 to ~724 cm – 1,i.e.across cristobalitespecimen was unavailable and the an ~575 cm –1 widecounting window, using a Ramanspectral data given by Bates (1972) were DilorXY spectrometerof 500 mm focallength used.Sample numbers preŽ xed AU arethose of akinto that used by Kingma and Hemley (1994) theUniversit yofAuckland, Departmen tof andemployingsimilar experimental procedures to Geology,petrology collection; that preŽ xed AM thoseused by and Rodgers and Cressey (2001) isfrom the Australian Museum mineral collec- with Ar+ asthe exciting line operatin gat tion.Nomenclature of the silica phases follows 514.5nm. The Raman spectra were collected at Smith(1997, 1998). 298±3 K,usinga liquidnitrogen-cooled CCD detector.Beam power was 50 mW andspectral bandpass2.74 cm – 1,withan integration time of Experimental 120s forthe incremental window of ~600cm –1, Thenature of the present exercise was to examine witha slitof 100 cm –1.Anuncoated 650 thevariation in spectra obtained from a varietyof objectiveproved satisfactory for microsampling

–1 FIG.1. Laser Raman spectral scans from 150 to 724 cm of crystalline and paracrystalline silica phases. (a,b) scans perpendicular to 100 for alternative ac orientations of asingle quartz macrocrystal, Switzerland: (a) c parallel to microscopic E –W, (b) c parallel to microscopic N –S;(c) quartz+moganite, Umukuri sinter, AU48290; (d) tridymite with cristobalite, Vehec, AMD52115.

3 K. A.RODGERSAND W.A. HAMPTON

someof the coarser, crystalline materials but a collectand coadd Ž vespectral runs from each 6100provided a morerestricted sampling area sampledarea in order to improve signal to noise andgave less risk of counting photons scattered ratios. byfabric elemen tsadjace ntto that being Oneof the advantages of the Renishaw 1000 examined.With a magniŽcation of 6100, an systemis to allow direct CCD imagingof the area of ~2 mmindiameter can be sampled.While precisearea of eachspecimen, and hence to target thisrepresents a considerableimprovement on aspeciŽc sintertextural element (Fig. 2). Further, conventionalX-ray powder analysis it still fails to thequasi-confocal optics of the Renishaw model copeadequately with smaller and intergrown usedconsiderably reduces the effect of stray light fabricelemen tsthatmaybe <1 mm. The collectionscattered from neighbouring grains conventionalarrangement of stage and lens of suchas had affected some Dilor XY results. theOlympus microscope gave 180º scattering Imageresoluti onwas ~1 mm using a 620 geometry. objective. Fortysamples from Whenuaroa were analysed Inthe majority of cases no preparation was usinga RenishawRaman 1000 system with deep undertakenbeyond breaking or sawing a fresh depletionCCD. Thispermitted individual textural face.The same difŽ culties experie ncedby elementsof the sinter to be recognized and Rodgersand Cressey (2001) were encountered targetedfor mineralogical identiŽ cation to an whenattempts were made to examine polished extentthat was not possible with the Dilor XY. sinterspecimens. Variable results were obtained in Theavailable Raman range was 100 –4000 cm –1 analysingspeciŽ c texturalareas of polished thin usingan air-cooled argon 488 nm 25 mW (blue) sectionsdue to the high level of  uorescenceof laserand 200 –2500 cm – 1 usinga solid-state theparticular epoxy mounting medium used. 785nm 25 mW (red)laser; the resolution was Prismfaces of quartz were microsampled in a ~2 cm – 1.Whilethe redlaserproved less sensitive consistentorientation, where these were developed touorescence,the spectral cut-off at ~200cm –1 andaccessible among crystals growing in cavities, madeit desirable where possible to employ the allowinglike spectra to be compared with like bluelaser with a 25%reduct ionŽ lter.A spectra.Elsewhere the size of the microcrystals 1200l/ mm diffractiongrating was used with the and/orthemassive nature of thematrix aggregates redlaser and a 2400l/ mm withthe blue. Spectra precludedconsistently sampling crystals of the werecollected with a slitof 100 mm and an sameorientation. A consequenceof the lack of integrationtime of 10s underhigh gain. Spectral orientationis reected in variationsin therelative bandpasswas ~2 cm – 1.Theintensity of the intensitiesand/ orabsenceof someRaman bands in moganitescattering bands made it desirable to somespectra (e.g. Fig. 1 a,b).

FIG.2. Example of CCDimage and raw Raman spectrum obtained from Pliocene Plumduff specimen AU53504 by the Renishaw 1000 system. Image shows an inter-Žngering of blue-grey (dark) and light-brown ultraŽne-grained sinter. The laser beam diameter, i.e. sampled area, is delineated by the circle at the centre of the crosshairs. Selective sampling of the two sinter types shows both to consist of quartz+moganite. Q=quartz scattering band; M= moganite scattering band.

4 MICROTEXTURALCOMPONENTSOF SINTERS

Results at about 350 –355 cm –1,wereobserved (e.g. Ramansc atteringbehav iourof crystalline,paracrystalline AU48271,Fig. 3 e). andnon-c rystallinesil icas Thosesinters rich in a-cristobalite,i.e. opal-C- bearing,show a strongband at 421± 2 cm – 1 Manyof the raw Raman spectra presented here (Figs 1d, 3a,d,e, 4a).Cristobalitelacks the arenot of highquality. In a numberof respects, adjacentvery intense band at 403 cm – 1 and the theyrepresent the useful minimum of what can moderatelyintense bands at 356, 449 and beobtaine dfroma sinterusing a Raman 457 cm –1 of a-tridymite(Fig .1 d). Where microprobe.In part, this arises from the rough packetsof tridymite are present in the silica surfaces,manifold imperfection s,and small size lattice(opal-CT), these bands combine with that ofsome of the textural elements microprobed. at 422 cm – 1 togivea broaddistinctive scattering Somespectra include a highlevel of  uorescence patternin the middle of the scanned range thatdrownsthe Ra mansignal,wh ileth e (Fig. 3e).Therefore,where such a bandis scatteringbehaviour of the poorly crystalline absentand a strongband centred at ~420 cm – 1 anddisordered phases is akin to the limited isdetectedit is concluded that opal-C constitutes responseof these same phases to a primaryX-ray thebulk of the opaline silica phase present beam;in both cases the scattering bands tend to (Fig. 3a). bebroadandill-deŽned(e.g.Fig.1). Quartzand moganite both possess similar Nonetheless,the distinctive Raman scattering vibrationalspectra which re ect their sharing bandsof allcrystalline silica species common in manystructural elements (Kingma and Hemley, sintercan be distinguishedreadilyand the nature 1994).The most intense band in quartz lies at ofallsilica sinter phases present can be deduced 465 cm –1 whilethat of moganite is at 501 cm – 1 fromthe results, provided these are interpreted (Fig. 1c).However,moganite also possesses a judiciously,rel yingbothon th enegative moderatelyintense scattering band at 463 cm – 1 evidenceafforded by the absence of scattering andKingma and Hemley (1994) found it bandsof cry stallinesilicaphases and the necessaryto use the presence (or absence) of knowledgethat only a limitednumber of silica otherquartz modes such as thesharp A1 vibration phasesis present in any sinter specimen. at 355 cm –1 toascertainthe presence (or absence) Afeatureof theRaman spectra of themajority ofquartzin a sample.Rodgers and Cressey (2001) ofyounger (<10,000 BP), opal-Adominated madeuse of the non-fundamental, broad, low- sinterswas a highlevel of  uorescencethat frequency,vibrat ionalband presen tinthe effectivelyswamped any Raman photons. As a spectrumof quartz (205 cm – 1)andoccurring in consequence,despite protracted quench times themoganite spectrum at a slightlyhigher and/orpre-treatment of samples with 100 vol.% frequency(220 cm – 1)todiscriminate the two H2O2 todestroyany organics that may be present, phases.This band is absent in cristobalite and nodiagnostic Raman response was observed in tridymite.Given the spectral cut-off at 200cm –1, anysinter consisting solely of opal-A. Some of itwas not possible to use this yardstick when this uorescencebackground appears to residein employingthe red laser on the Renishaw 1000. thedisordered, hydrous nature of opal-A itself However,in two instances, spectra obtained by (Splett et al., 1997). theRenishaw spectrometer showed the 377 cm – 1 Manyolder Quaternary sinters that are also scatteringband of moganite. predominantlyopal-A have an appreci able, visibleorganiccontente.g.Omapere Occurrenceof silicaphases (~40,000BP). Thesealso show high count rates thatincrea sedrapidl yandun iformlywith Asanticipated, few of the younger (<20,000 BP) increasingwavenumber over the width of the sintersshowed any Raman scattering bands experimentalcounting window of the Dilor X-Y characteristicof any crystalline or better-ordered instrument.Microprobed portions of sinters rich paracrystallinesilica phases that commonly occur inopal-CT and having low but still detectable in sinters (cf. Herdianita et al., 2000a). Young uorescencegave a diffuse,ill-deŽ ned scattering (<40,000BP) sintersare composed principally of bandwhich mimics that seen in samples with non-crystallineopal-A and/ orparacrys talline negligible uorescence.Occasionally, following opal-CT. peroxidetreatment of some opal-CT-rich samples, Severalolder sinters (30-50,000 BP), identiŽed broad (100 –200 cm – 1)scatteringbands centred byHerdianita et al. (2000a)andCampbell et al.

5 K. A.RODGERSAND W.A. HAMPTON

FIG.3. Raw(unmanipulated) laser Raman microprobe spectra of late Quaternary sinters (Q=quartz; M=moganite; C=cristobalite): (a) pisoidal fabric AU48310, showing scattering broad bands of a-cristobalite; (b) monomineralic microcrystalline quartz sinter AU48290; (c) mottled palisade AU48305, showing quartz, cristobalite and moganite bands; (d) one microsampled spot of AU48310 showing bands of moganite and cristobalite and lacking aquartz band at 355 cm –1;(e) opaline sinter, AU48271, rich in opal-CT displaying broad scattering bands centred at ~200 cm –1 and 380 cm –1 following peroxide treatment.

(2001)as containing appreciable opal-C, showed mostbulk samples of these sinters. They can be welldeŽ ned, albeit broad bands of a-cristobalite seenunder SEM andhave been detected by at418 and 230 cm –1 (e.g.Umukuri pisoidal Ramanprobes of fractur es(e.g. Omapere , fabricsinter, AU48310; Fig. 3 a).Thebreadth of Otamakokore).Similarly, detrital quartz can be thesebands is taken as re ecting an inherently detectedin Raman-sampled spots of sintermatrix lowlevel of long-range ordering in the opaline whenit is present at levels insufŽ cient to be lattice (cf. Smith,1998). It is comparable to the detectedin XRD scansof bulk samples. broadeningseen in the X-ray diffraction (XRD) Inall sinters older than 50,000 BP, quartzis a linesof these same samples (Herdianita et al., major,if not dominant or even sole silica phase 2000b). present(Fig. 3 b).However,as reporte dby Inrare scans of some sinters rich in opal-A, a Rodgersand Cressey (2001), distinct moganite veryweak quartz band at ~465 cm – 1 occurs. On bands,both that at 501 cm –1 and a skewed theone hand this may re ect disseminated detrital 210 cm – 1 broadband,occur in the sp ectra quartzthat is widespread,although usually at low obtainedfrom many quartz-dominant horizons concentration,in sinters found throughout New ofNew Zealandsinters, particularly among those Zealand’s TaupoVolcanic Zone, where tephra is withmicroc rystallinequ artzmatric es,e .g. scatteredacross the sinter aprons following each UmukuriAU48290, Otamakoko reAU49673, eruption.On the other, the signals may arise from AtiamuriAU4724 0(Fig.1 c).Rodgersand tiny,very late-stage quartz crystals that occur Cressey(2001) failed to identify moganite in infrequentlyin fractures in some of these young anysinter older than 200,000 BP, butin the sinters.These crystals are generally not abundant presentstudy moganite was found to be unevenly enoughto be discerned in routine XRD scansof disseminatedthroughout specimens of the late

6 MICROTEXTURALCOMPONENTSOF SINTERS

FIG.4. CCDimages and raw Raman spectra of two microsampled spots of alepisphere in irregular fenestral sinter, AU53495, Plumduff, obtained using aRenishaw 1000 Raman microprobe system with ablue laser as the exciting line (Q,C,M as for Fig 3): (a) quartz with cristobalite; (b) quartz with signiŽcant moganite as indicated by the relative intensities of 465 and 501 cm –1 scattering bands.

PlioceneWhenuaroa sinter outcrops at Plumduff Theassociation of the scattering bands of andMount Mitchel lthatotherwi seconsist a-cristobalite(opal-C) with quartz and moganite predominantlyo fmicrocrystallinequartz. isnotuncommon in olderQuaternary sinters, e.g. Despitethe limitations outlined by Rodgers and intheUmukuri mottled palisade fabric AU48305 Cressey(2001) of using Raman spectra to (Fig. 3c).Inone sampled scan of AU48310 from estimaterelative proportions of different silica Umukuri,moganite bands were observed together phases,in general the ratio of the moganite with a-cristobalitewhere quartz appears to be 501 cm – 1 scatteringband to the composit e absent.Although the combined 465 cm –1 band is quartz+moganiteband at ~465 cm –1 in most quitesubdued, no scattering band arising from Whenuaroasamples is notably less than that quartzalone was observed in repeated and illustratedby Rodgers and Cressey (2001) from coaddedscans, such as that at 355 cm –1 (e.g. Quaternaryquartz-rich sinters of Umukuri and Fig. 3d).Thissample proved quite inhomoge- Otamakokore,and suggests the proportion of neousin respect of the distribution of its silica moganiteto quartz is lower at Whenuaroa, i.e. phases. <13%.However, in one example, AU53495 Rodgersand Cressey (2001) noted that, in (Fig. 4b),themoganite 501 cm –1 isexceptionally general,moganiteintheUmukuriand strongand the ratio to the composite 465 cm –1 Otamakokorelate Quaternary, quartz-rich sinters band 2 –4timesthat of the examples of Rodgers tendedto bepresentin thosesmaller, bipyramidal, andCressey. euhedralcrystals that line sinter cavities and that

7 K. A.RODGERSAND W.A. HAMPTON

FIG.5. CCDimages and raw Raman spectra of microsampled spots of two contrasting quartz microcrystals (Qand M as for Fig. 3): (a) AU53479, late stage crystal exhibiting numerous growth striae and growing perpendicular to cavity wall and showing scattering bands of quartz alone; (b) AU3501, tiny, euhedral crystal, lacking growth striae on prism faces and developed at low angle to cavity wall, that possesses scattering bands of both quartz and moganite. The opportunities for stray light collection scattered from neighbouring grains are minimized with the confocal optics in the Renishaw Raman microprobe used to collect these spectra particularly given the size of the laser spot (circled).

formon recrystallization of opaline sinter. They lucentblue-grey matrix but not present in any of failedto Ž nddeŽ nitive evidence of moganite cavity-liningcrystals microprobed that ranged amonglate-stage,  uid-depositedquartz crystals from 2 to >20 mm along c. formedsubsequent to therecrystallization process Incontrast to the other silica polymorphs, andwhich tend to belarger than those formed by b-cristobalitelacks any Raman scattering band recrystallization.In general, among the quartz- between300 and 700 cm – 1;itssingle Raman richsinters of Whenuaroa, the same pattern of activefundamental occurs at 777 cm – 1. It was not moganiteoccurrence was seen, proving it to be foundin anyof thesinters analysed, nor would it presentamong some individual quartz crystals but beexpected,even as ametastablephase, given its notothers (Fig. 5 a,b).Forexample, it was stabilityrange. Nonetheless, Keith et al. (1978, detectedat levels estimated to be <2% in a p.A20)reported " b-cristobalitedisorde red", 20 mmlongcrystal in AU53501 and a 10 mm togetherwith ``a-cristobalite,ordered ’’ among crystalinAU53504 ,butwas absent in a theolder sinters of Upper Geyser Basin ( cf. 1806120 mmcrystalin AU53479 and a 40 mm White et al., 1988,in the Norris Basin). Perhaps, crystalin AU53449. In AU53502, moganite was theseidentiŽ cations would be consideredtoday as dispersedthroughout the ultraŽ ne-grained trans- thoseof opal-CTand opal-C, respectively.

8 MICROTEXTURALCOMPONENTSOF SINTERS

FIG.6. Composite variable pressure SEMimage of an uncoated, rough-sawn, partially crystallized slab of Umukuri opaline palisade fabric, AU48305 ( cf. Campbell et al.,2001), showing raw Raman scattering spectral results from different areas of surface (Q,M,C as for Fig. 3): (A) massive microcrystalline layer at base of slab; (C) vitreous microcrystalline matrix; (D) vitreous palisade pillar; (H) dense white microcrystalline quartz layer; (J) yellowish intercavity matrix; (K)yellowish intercavity matrix; (M) yellowish interpalisade matrix; (O) yellow weathered matrix at top of slab. The extent of the different relative intensities of the scattering bands quartz, moganite and cristobalite in the different scans are taken as evidence of inhomogeneity in the sample’s surface. Note: base of sample is at top of page. Image by Chris Jones, NHM,London.

Texturalex amples sintersŽ ndsafŽ rmation in Fig. 6 wheredifferent Thepower and potential the Raman microprobe texturalelements of asingle,rough-cut sample of offersfor mineralogi caltextural analysis of mottledpalisade fabric (AU48305) from Umukuri

9 K. A.RODGERSAND W.A. HAMPTON

havebeen microprobed ( cf. Campbell et al., matrixand in clasts of both fenestral network and 2001).In interpreti ngthe results, allowance columnarfabric types, although it was not needsbe made for the transparen cyof the uniformlydistrib utedin any of these. For sampleand for photon scattering from areas example,in an example of irregular fenestral adjacentto those being probed, but the manner sinter,AU53495, moganite was discerned in a inwhich the mineralogy of the different fabric solidband of vitreous silica transecting the elementscan be identiŽ ed is clearlyevident. The samplebut not in the corroded, porous matrix microsampledareas of thespecimen varied in the surroundingthis band. Similarly, it was detected relativeintensity (and absence) of thecristobalite, inone vitreous white clot in recrystal lized quartzand moganite scattering bands as did columnarsinter AU53481 but was absent in a spectrafrom different sampling areas of the seeminglyidenticalclot~4mmdistant. matrixof some specimens in which only quartz Varicoloured,translucent, vitreous, chalcedonic andmoga nitewere presen t,e.g. Atiamu ri clastsin sinter breccias proved both moganite- AU47240(Fig. 7). bearingand moganite-free as, indeed, was the Asimilarpattern of microscopic inhomo- matrixof some breccias. For example, blue-grey geneityis evident in the considerably older, clastsof ultraŽ ne-grained sinter in AU53501 and quartz-rich,heavily recrystallized and mineralized AU53472contained moganite but it was absent Whenuaroasinter. Importantly, it isnot a pattern fromthe texturally similar, albeit light-brown, mademanifest by conventional X-ray powder matrixof thesesame rocks. In contrast, AU53504 analysis.The Raman microprobe failed to detect moganitewas present in both blue-grey clasts and anymoganite scattering in any recrystallized light-brownmatrix. examplesof primary sinter palisade ,tufted Inthe case of the Otamakokor emassive networkor Ž nely-laminatedfacies. It was, mottledsinter, AU49673, the Raman results however,recognized in portions of blue-grey demonstratethat the specimen is bimineralic and

2 FIG.7. Examples of raw Raman spectra from differing microsampled spots (500 mm area) of broken surface of recrystallized, quartz-rich sinter from Atiamuri, AU47240 (Q,M as for Fig. 3). The variation in the relative intensities of the principal scattering bands of quartz and moganite in the different spectra are taken as evidence of inhomogeneity in the sample’s surface.

10 MICROTEXTURALCOMPONENTSOF SINTERS

moreor lesshomogeneous (Fig. 8), in contrastto Discussion Atiamurisinter AU47240 (Fig. 7). Not only do thespectra of threedozen spots in a100mm 2 area Aprincipaladvantage of the Raman microprobe ofAU49673display the samescattering bands but asadeterminativetool for sinters lies in its ability therelative intensities of the quartz to moganite tomicrosample individual textural elements. bandsare more or less similar, regardless of the Coupledwith the high rate of spectral acquisition areaprobed. Paradoxically, in this instance the availablefrom the latest generation of instru- transparentnatureo fthesamplean dthe ments,the limited sample preparation, and the opportunitiesfor Raman scattering from areas abilityto directly target individual domains, the adjacentto thosebeing probed provide support for techniqueallows the twin paths of phase and deducinga uniformityof sample. texturalmaturation of sinters to be mapped Theirregular fenestrae sinter sample from simultaneously,at least in their later stages of Plumduff,AU53495, that proved high in moga- development.This contrasts with techniques such nitecontains abundant microbotryoidal masses of asconventional powder diffraction that allow for lepispheres,<50 mmindiameter. The Raman identiŽcationofonlybulk,albeitsmall spectraobtained from the lepispheric plates by (~400mg), samples that can not always be targetedRenishaw sampling using a longfocal clearlyrelated to speciŽ c microtexturalcompo- length 650objective, contains, along with the nents.Certainly, in sinter specimens containing scatteringbands of moganite and quartz, those of opal-A,and to a lesserextent opal-CT, Raman cristobalite(Fig. 4 a,b).Thisis the oldest sinter doesnot allow the full range of silica phases fromNew Zealandin which cristobalite has been presentto be unambiguously identiŽ ed, nor for identiŽed and it is notable that the characteristic thedegree of lattice order and disorder of the X-raypowder patterns of opal-CTor opal-C were differentsilica phases to be evaluated. On the notobserved in >90 traces obtained from a wide otherhand, in those samples lacking strong texturalrange of Whenuaroasinter specimens. uorescence,the absence of scattering bands of

2 FIG.8. Examples of raw Raman spectra from aseries of 30 different microsampled spots within a100 mm area of Otamakokore massive mottled sinter, AU49673. (Qand Mas for Fig. 3). The similarities in the relative intensities of the principal scattering bands of quartz and moganite in the different scans are taken as evidence of relative homogeneity in this portion of the specimen’s surface.

11 K. A.RODGERSAND W.A. HAMPTON wellcrystallized phases can be takenas indicating process– exactlywhat is seen in the Raman notonly the absence of these phases but also, evidence.However, the extent of inhomogeneity presumably,the presence of poorly crystalline or inthe samples analysed here and the relatively non-crystallinesilicas – giventhe limited number verysmall size of the Raman sampled areas are ofsilica phases present in sinters. such,that many thousands of measurements Ofparticular importance is the manner in wouldneed to be made before any distribution whichthe Raman microprobe frees the sinter patternof silica phases might be deduced that had workerfrom the need to rely on morphologyas a realmeaning at the outcrop, let alone in individual meansof ide ntifyingsilica phases at t he handspecimens. ultrastructurallevel. Indeed, the present results cautionagainst relying on phasemorphology as a Acknowledgements primarydeterminative tool. The habit of the Plumduffquartz+mogan ite+opal-Clepispheres Thisstudy is a contributionto theAuckland Sinter closelymatches examples typical of opal-CT Programme.It was undertaken in part with the morphologiesdescribed by Campbell et al. (2001, assistanceof the award to KAR ofa Visiting Fig. 6b and d)fromthe Umukuri sinter, and from Fellowshipby the University of Technology, elsewherein the Taupo Volcanic Zone (Rodgers, Sydney,andof a SeniorBursar ybythe 2000,20 01).Th ePlumduffex amplesar e MineralogicalSociety. The Williams family and presumablypartially pseudomorphous after opal- GRD MacraesLtd generously allowed access to CT. Theimplications this mixed assemblage Plumduffand Mount Mitchell .Fundswere holdsfor morphological-base didentiŽcations of contributedtoWAHfromanAusIMM silicaphases are also contained in the dual EducationEndowmen tTrustScholars hip,a mineralogyof the moganite+quartz crystals that Societyof Economic Geologists Foundation- exhibittypical quartz forms. Such implications BHPStudentResearchGrant,fromthe aresigniŽ cant when determining the onset and GeothermalInstitute Freeston Licensing Trust hencetiming of phase changes in sinters. anda Universityof Auckland Centennial Award. Perhapsthe most signiŽ cant result of both the Wethank Professor Evan Leitch (UTS, Sydney), presentstudy and that of Rodgers and Cressey DrsPat Browne and John Seakins (Auckland), (2001)is conŽ rmation that the silica phase DrsAllan Ball, Gordon Cressey and Chris Jones inhomogeneityexists at microstructural level. (NaturalHistory Museum, London), and Ross Campbell et al. (2001)described outcrops at Pogson(Australian Museum, Sydney) for assis- Umukurithat consist of intergrown, irregular tanceand/ orfruitfuldiscussions. Louise Cotterall patchymasses of quartz-rich and opal-C-rich assistedwith the Ž gures. sinter.Given the surŽ cial conditio nsunder whichsinters undergo their transformation, the References slightdifferences between these conditions and thosethat existed during a sinter’s original Allen, K., Roberts, S.and Murray, J.W. (1999) Marginal deposition,and the largely solid-state nature of marine agglutinated foraminifera; afŽnities for themetamorphosis, clear-cut, dramatic bound- mineral phases. Journal of Micropalaeontology , 18, ariesbetween phases and textures are not to be 183 –191. expected.However, the present results demon- Bates, J.B. (1972) Raman spectra of a and b cristobalite. stratethat such macrosco picinhomoge neity Journal of Chemical Physics , 57, 4042 –4047. merelyre ects an underlying inhomogene ity Campbell, K.A.,Sannazzaro, K.,Rodgers, K.A., pertainingat the microstructural level. Herdianita, N.R.and Browne, P.R.L.(2001) Theresult is in keeping with Landmesser’ s Sedimentary facies and mineralogy of the Late (1995)innovative diffusion model for silica Pleistocene Umukuri silica sinter, Taupo Volcanic transportand accumulation in sedimentary envir- Zone, NewZealand. Journal of Sedimentary onmentsand which, among other matters, would Research, 71, 727 –746. facilitatethe transformation process. As a resultof Farmer, J.D. (2000) Hydrothermal systems: doorways to theextreme sluggishness of the silica phase early biosphere evolution. GSA Today, 10, 1 –9. transformationprocesses at ambienttemperatures, Fournier, R.O.and Rowe, J.J. (1966) Estimation of adirectconsequence of Landmesser’ s modelis underground temperatures from the silica content of thata heterogeneityof productwill be expected to water from hot springs and wet-steam wells. obtainthroughout much of the transformation American Journal of Science , 264, 685 –697.

12 MICROTEXTURALCOMPONENTSOF SINTERS

Go¨tze, J., Nasdala, L., Kleeberg, R.and Wenzel, M. Wales, 134, 112 –113. (1998) Occurrence and distribution of "moganite" in Rodgers, K.A.and Cressey, G.(2001) The occurrence,

agate/chalcedony: acombinedmicro-Raman, detection and signiŽcance of moganite (SiO 2) among Rietveldandcathodoluminescencestudy. some silica sinters. Mineralogical Magazine , 65, Contributions to Mineralogy and Petrology , 133, 157 –167. 96 –105. Skinner, D.N.B.(1976) Sheet N40 and parts of Sheets Heaney, P.J. and Post, J.E. (1992) The widespread N35, N36 and N39 Northern Coromandel (1st distribution of anovel silica polymorph in micro- edition). Geological Map of NewZealand 1:63,360. crystalline quartz varieties. Science, 255, 441 –443. Map (1 sheet) and notes (28 p.) NewZealand Herdianita, N.R., Browne, P.R.L.,Rodgers, K.A.and Department of ScientiŽc and Industrial Research, Campbell, K.A.(2000 a)Mineralogical and morpho- Wellington, NewZealand. logical changes accompanying aging of siliceous Smith, D.K.(1997) Evaluation of the detectability and sinter and silica residue. Mineralium Deposita , 35, quantiŽcation of respirable crystalline silica by 48 –62 X-ray powder diffraction. Powder Diffraction , 12, Herdianita, N.R., Rodgers, K.A.and Browne, P.R.L. 200 –227. (2000b)Routine procedures to characterise the Smith, D.K.(1998) Opal, cristobalite, and tridymite: mineralogy of modern and ancient silica sinter noncrystallinity versus crystallinity, nomenclature of deposits. Geothermics , 29, 65 –81. the silica minerals and bibliography. Powder Keith, T.E.C.,White, D.E.and Beeson, M.H. (1978) Diffraction , 13, 2 –19. Hydrothermal alteration and self-sealing in Y-7 and Splett, A., Splett, Ch. and Pilz, W.(1997) Dynamics of Y-8 drill holes in the northern part of Upper Geyser the Raman background decay. Journal of Raman Basin, Yellowstone National Park, Wyoming. United Spectroscopy , 28, 481 –485. States Geological Survey Professional Paper , Walter, M.R., Des Marais, D., Farmer, J.D. and Hinman, 1054 –A, 26 pp. N.W. (1996) Lithofacies and biofacies of Mid- Kingma, K.J. and Hemley, R.J. (1994) Raman spectro- Paleozoicthermal springdepositsin the scopic study of microcrystalline silica. American Drummond Basin, Queensland, Australia. Palios, Mineralogist , 79, 269 –273. 11, 497 –518. Landmesser, M.(1995) Mobilita¨tdurch Metastabilita¨t: White, D.E., Hutchinson, R.A.and Keith, T.E.C.(1988) SiO2 Transport und Akkumulation beiniedrigen The geology and remarkable thermal activity of Temperaturen. Chemie der Erde , 55, 149 –176. Norris Geyser Basin, Yellowstone National Park, Rodgers, K.A.(2000) Research on silica sinters. Wyoming. USGeological Survey Professional Mineralogical Society Bulletin 126(April), 16 –17. Paper, 1456, 1 –84. Rodgers, K.A.(2001) Silica phases of NewZealand sinters.Geodiversity Symposium, Australian Museum,4December 2001. Journal and [Manuscript received 3May 2002: Proceedings of the Royal Society of NewSouth revised 7August 2002 ]

13