American Mineralogist, Volume 77, pages 314-328, 1992

Chemical controls on ferrierite crystallization during diagenesisof silicic pyroclastic rocks near Lovelock, Nevada

SrBpnBN B. Rrcn* Exxon Researchand Engineeringcompany, Annandale, New Jersey08801, U.S.A. Knrrn G. PIPrB 2180 SchroederWay, Sparks,Nevada 89431'U.S.A. Devro E. W. Ylucn.lN Exxon Researchand Engineeringcompany, Annandale, New Jersey08801, u.s.A.

Ansrru'cr The ferrierite, mordenite, and clinoptilolite have formed with smectite through hydration reactions in a sequenceof rhyolitic pyroclastic rocks near Lovelock, Nevada. Fiigh activities of silica and magnesia were required simultaneously for the large-scale crystallization of ferrierite in thesediagenetically altered rocks. Post-depositionaladdition of Mg coincided with ferrierite and smectiteformation. The sourceof the Mg is not known, but it could have originated either from a brackish, saline lake or from nearby leached .Thus, origins of the Lovelock ferrierite deposit may have involved both lacustrine deposition and an open ground water system for the transformation of vitric tuffs into zeolites. Ferrierite crystallized early, mainly in the pore space of the tuffs. Clinoptilolite and cristobalite grew later by replacementof glass shards. Orthoclase and mordenite crystal- lized somewhat later, orthoclase usually in intimate associationwith ferrierite in K-rich zonesand mordenite alone in open pore spaces.Ferrierite crystallizationproceeded through a mechanism of dissolution and reprecipitation on glass and clay surfaces' Significant variation in chemical composition of individual ferrierite crystalsoccurs on the scaleof a few cubic millimeters.

near INrnonucrroN Since its discovery in an outcrop ofolivine the shore of Kamloops Lake in British Columbia (Gra- Ferrierite is a medium-pore that has interesting ham, l9l8), the zeolite ferrierite has been observed in potential applications in petrochemical processing. nearly 30 localities. With the exception of a very few Klowledge of the conditions leading to its formation may occurrencesin altered volcanogenic sediments,these all provide insight into creating a simple, inexpensive syn- appear to be hydrothermal vug occurrences.In the de- thesis method. The Lovelock, Nevada, deposit, in which posit near Lovelock, discoveredin 1965 by P' E' Galli, ferrierite occurs in minable proportions, provides a nat- ferrierite formed through hydration reactions in a se- ural laboratory for studying the processesof zeolitization quence of rhyolitic pyroclastic rocks. A plzzle at Love- and the conditions under which ferrierite crystallizes. lock is the dominance of ferrierite over mordenite and What were the special conditions responsiblefor the un- clinoptilolite, two zeolites that are very common ln slm- usually large quantity offerrierite at Lovelock? From the ilar tuffaceoussettings in the western United States. In results of a study of ferrierite synthesis,Cormier and Sand fact, severalzeolites, including erionite, mordenite, chab- (1976) concludedthat metastability may be the key. Their azite, clinoptilolite, and analcime, are very common in conclusion, although soundly based on experiment and such deposits, whereasferrierite is only known to occur intuitively pleasingwhen applied to the Lovelock depos- in the Lovelock deposit and in smaller amounts in the it, nevertheless requires further examination through Stillwater Range near Fallon, Nevada. A Korean occur- careful study of the mineralogical and chemical relation- rence was interpreted by Noh and Kim (1986) to result ships in the rocks themselves.In this study we have com- from diagenesisof silicic vitric tuffs. bined electron petrographic characterization with bulk Reports describing the Lovelock zeolite deposit con- chemical and phaseassemblage data to interpret the or- tain widely divergent opinions about the genesisof the igins offerrierite in altered pyroclastic rocks. zeolites.Sand and Regis (1968) suggesteda hydrothermal origin for the ferrierite, but no supporting evidence for or sourc€of the fluids was given. * Presentaddress: Exxon Production ResearchCompany, P.O. estimatesof temperature Box 2189. Houston.Texas 77252-2189, U.S.A. Regis(1970) describedthe Lovelock and Stillwater Range 0003-004x/92l0304-03I 4$02.00 314 RICE ET AL.: FERRIERITE CRYSTALLIZATION 315

\ ll a

foE-l I ptioceneto OUATERNARY Sedimentaryand rclcanic rocks Hotocene I I Ferrieriterichtulf LOVELOCKZEOLITE DEPOSIT I T$ I mZ ptircene i and Z Mordenit+richtuff Tutfand sedimentaryrocks Basaltand andesite Miocene TERTIARY I tg zeolitictuff GTn;l -ll6mts 1 Mioceneto N ctass Votcanicrocks Oligocene J I Diatomite tl:KE,ll I Cretaceousto CRETACEOUSto CoreHole Locations plutonic n weldedrhyolite tufl a Metamorphicand rocks J Triassic TRIASSIC E Alluvium a FerrieriteDeposit Fig.2. Geologicmap of the Lovelockdeposit. Fig. 1. L,ocationmap for theLovelock zeolite deposit (Shep- pardet al.,1983). material. This sequenceagrees with the order of appear- occurrencesof central Nevada, giving compositions of ance of these zeolites deduced from the experiments of the ferrierite at each locality, and he also reported zona- Cormier and Sand (1976). The cores themselves,how- tion for ferrierite, mordenite, and clinoptilolite basedon ever, are more difficult to correlate. The discontinuities outcrop sampling. Low alkalinity conditions were sug- betweencores 2. 3. and 4 could be fault-bounded discon- gested.In a guidebook accompanyingthe zeolite fieldtrip tinuities or unconformities related to variations in the Zeo-Trip'83, Sheppardet al. (1983)described the Love- sedimentary environment such as erosional surfacesor lock occurrencein somewhat$eater detail and concluded pinch-outs. that the original pyroclastic material was deposited in a lake but that little was known about the fluids responsible Cores and outcrop samples for zeolitization. An examination of the chemical con- Four vertical core holes (Fig. 2), each 5 cm (2 in.) in trols on zeolite crystallization in the Lovelock deposit diameter, were drilled under contract for W. R. Grace that account for this unusual occurrenceis the subject of and Company in the early 1970s. Portions of the cores the present study.

B.q.crcnouNn Geologic setting of the Lovelock deposit The Lovelock zeolitedeposit occursin a group ofun- named sedimentary and volcanic rocks of Tertiary age MordenileHill (Miocene or Pliocene)in a small valley within the Trinity Range(T. 28 N., R. 30 E.) in PershingCounty, northwest T Nevada (Fig. l). Reconnaissancemapping placed the Lovelock deposit in the Upper Tertiary volcanic group JUm A welded ftyolite tutl (Tuv) of Stewart and Carlson (1978). The ferrierite de- D Finegla$ + smectite posit of the Stillwater Range also lies within the Upper E coaR gla$ + Eolile N Ferrlerhe+ smecllte group. The observablesurface features of the deposit and I Benlonile the core locations are shown in Figure 2. Z Mordenite+ lerrierite+ orthoclag VedicalEraggeEtion I Ferrierite+ orthoclase The stratigraphy based on the bulk chemistry, petrog- -10x I Oinoptilollre+ crlstoballte raphy, and phase-assemblagedata is shown in Figure 3. There is a zonation characteizedby mordenite at the top, Fig. 3. Stratigraphy ofthe Lovelock zeolite deposit. Layers grading downward into less mordenite and increasing A, B, and C are defined on the basis of composition (as in Fig. amounts of ferrierite, and finally a large amount of glassy I 4) and assemblage. 316 RICE ET AL.: FERRIERITE CRYSTALLIZATION

diffractometer with CuKa radiation. Scanswere run al2 per min from 2 to 60 20, at 40 kV and 30 mA. Phase identification, particularly for zeolites, was enhancedby a computer-based matching progam with a library of natural and synthetic zeolites as well as other silicates. The scanning electron microscope used was a JEOL 35C equipped with an LaB6 gun, a four-crystal wave- length spectrometer (WDX), adjunct energy-dispersive X-ray detector (EDX), and PGT SystemIV analysis sys- tem. Both normal secondaryelectron imaging (SEI) and backscatteredelectron imaging (BEI) were used. Electron microprobe analysis was performed by a method com- bining EDX and WDX, using orthoclase from Benson Mines, St. Lawrence County, New York (K, Al, Si), lab- radorite plagioclasefrom l,ake County, Oregon (Ca, Al, Si), and hornblende from Kakanui, New Zealand (Mg, Ti, Fe, Al), as standards.Alkali difusion during analysis of zeolites and glasswas corrected by monitoring counts for 30 s and extrapolating to zero time. Clay and zeolite analysestotaled less than l00o/oprimarily becauseof the presenceof HrO, both adsorbedand structural. Ferrierite typically has l0-l5o/o HrO by weight.Matrix corrections were calculated by the NBS Frame C program. The sec- ondary massspectrometer (SIMS) usedwas a Cameca 3f with a high-gain camera(20 million ASA) and a Crys- tal image processor(kta, 1985). Becausethe charging problems inherent with the imaging SIMS method are more severethan with SEM, SIMS samplesrequired em- bedding in a conductive mount. A Philips 420ST transmission electron microscope (TEM) with EDX and adjunct PGT System IV analysis Fig. 4. SEM micrographshowing (a) a fracturesection of systemwas usedfor characterizationofglassy and zeolitic glassin sampleSR75 and (b) the morphologyof smectiteon the tuffs and for identification ofzeolites by electron diffrac- glasssurface. tion. Chemical compositions of ferrierite crystals were determined using the Cliff-Lorimer thin-film method (Goldstein et al., 1977). are stored at the Nevada Bureau of Mines and Geology Rnsur-rs in Reno, Nevada. Samples were selectedfrom the four petrography cores,with particular attention paid to the best preserved Electron cores from hole nos. 2 ar;,d3. Unfortunately, the lower Vitric tufls. Vitric tufffrom 30.5-m depth in hole no. core from hole no. 3, below about 12.2 m, was inadver- 3 is relatively unaltered. Glass constitutes most of the tently discarded at W. R. Grace and Company. specimen, and the remainder is quartz, plagioclase,and Vitric and zeolitic tuffs were sampled in outcrop to phenocryst sanidine and Fe-Ti oxides. These are clearly augment the core samples. The ferrierite-rich outcrops air-fall tuffs because,in addition to good sorting (average were surprisingly difficult to sample becauseof their ex- shard size is 50-100 prm),there is a largeamount of in- treme toughness.They have a green color that is subtly terclast porosity and no evidence for either welding or different from the other outcrops of tuffs, most of which flow. are buff to tan. They are also locally strongly silicified, A fine-grained unaltered ash (SR73) outcrops at the althoughthe spatialrelationship between the opalinezones base of the welded tuff at the northern edge of the field and the ferrierite tuff was not clear in the available out- area. Theseare angular shardsexhibiting elongatedpores crops. There is banding in many of the zeolitic tuffs, and and occasional cuspate morphology and whose average some sedimentary features,such as ripple marks, pinch- grain size(5-10 pm) is very similar to the Hekla Ho teph- outs, and cross-beddingrelationships, were preservedde- ra layer. Heiken and Wohletz (1985) interpreted those to spite the zeolitization process. be of phreatoplinian origin, implying a highly vesicular magma resulting from mixing with surfaceHrO. Instrumental techniques Figure 4 shows fresh, fractured glass shards (50-100 Powder X-ray diffraction (PXD) was carried out using prmgrain size) from SR75 and SR76 with smectite rims, powder-pack methods with an automated SiemensD500 which have this featurein common with C3-30.5.Smec- RICE ET AL.: FERRIERITE CRYSTALLIZATION 5t I

tite, the predominant authigenic phasein the early stages of alteration, is present on most of the glasssurfaces ob- served. The morphology shown in Figure 4b is typically observed for smectite, althqugh at least in some in- stances,this corn-flake texture may be an artifact ofthe preparation for SEM. SIMS images from polished sectionsof specimen C3- 30.5 in Figure 5 illustrate severalsalient featuresofthese slightly altered shards. The K ion image indicates that alkalies have not been leachedfrom this glass,a conclu- sion also inferred from the composition of the glass(Ta- ble l) becauseNa (3.020loNarO) and K concentrations (5.25o/oKrO) are typical ofrhyolites with about 780/oSiOr. The Mg image and corroborating TEM examination clearly show that Mg is concentratedin the smectitephase rimming the glass. Therefore, the glass alteration here appearsto be limited to the surfaceof the shards. Cristobalite, clinoptilolite, and possibly amorphous sil- ica are pseudomorphicafter shards(Fig. 6) and thus form in replacementreactions in severalhole 2 tuffs. Ferrierite is present in only minor amounts and occurs as coatings on shards.Figure 7a shows in cross section a relict pum- ice shard that is now largely cristobalite. Much of the silica reported in this study is probably the opal-CT phase defined by Jonesand Segnit( I 97 I ), characterizedby lepi- spherescomposed of intersecting platelets of disordered a cristobalite. Becausethe 4.3-A tridymite peak charac- teristic of opal-CT was not always observed,we refer to it throughout as cristobalite. The pores, which appear nearly circular in this view, are a composite of authigenic orthoclaseand ferrierite, lined with smectite. The ortho- clase crystals, distinguishable in powder X-ray patterns from sanidine phenocrysts,form an elliptical pattern that suggestsa concentration gradient of K with ellipsoidal symmetry. Figure 7b shows a void outlined by smectite and ferrierite in succession.It is not known whether there is any direct relationship betweenthe zeolite and the clay in terms of nucleation, crystallographiccontrol, chemical exchange,or reaction. Probably the clay merely provided a substrate on which the ferrierite could grow because ferrierite was also observed to have grown directly on remnant glasssurfaces in specimensC2-3.7 and C4-8.5. Productsoccur at shard edgesand in poresin the hole-2 and hole-4 tuffs, which have carried the alteration process further than the early smectite alteration stagerepresent- ed by specimenC3-30.5. These additional crystalsat- tached to shards (Fig. 8a) have a morphology and com- position confirming their identity as ferrierite. They appear physically separatedfrom glassby a clay layer. Ferrierite prismatic tlpically developsin a habit normal to the shard Fig. 5. SIMS imagesof a polished section of vitric tuffsam- surfaces(Fig. 8b). There are very large, highly vesicular ple C3-30.5.The Si imageidentifies the shapeofthe glassshard. pumiceous grains and blocky, nonvesicular shards. Per- The Na image,which imitates Si here, confirms that alkalies are litic texture is very common in some of the blocky non- uniformly present in the glass and therefore leaching has not porous particles. occurred. The Mg is a constituent of the smectite coating the Some bentonitic beds are interspersedwith glassyand glass. zeolitic tuffs. In certain strata the smectite is pure enough to be consideredfor recovery,along with the zeolite, were the deposit to be mined commercially. 318 RICE ET AL.: FERRIERITE CRYSTALLIZATION

Tlale 1. Glasscompositions

74.2 73.5 74.4 77.1 74.8 76.2 75.7 sio, 75.0 't2.2 Al2o3 11.8 11.9 11.8 12.2 12.9 12.4 12.4 Mgo 000 0.00 0.00 0.00 0.00 0.00 0.05 0.14 CaO o.22 0.17 0.28 o.22 0.00 0.31 0.00 0.89 NarO 2.88 2.83 2.71 3.27 4.60 4.59 5.33 4.12 K"O 5.13 4.86 5.04 5.10 5.76 4.40 4.49 4.26 Tio, 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FeO 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Total 95.02 94.00 93.40 95.12 100.35 96.25 98.48 97.51 Remarks vesicular vesicular nonvesic. nonvesic. Sampleno. c3-30.5 c3-30.5 c3-30.5 c3-30.5 sR78 SR78 SR78 SR78

Ferrierite tuffs. These tuffs have been converted to as- from the ferrierite, also shows the vitroclastic texture. semblagesof ferrierite, authigenic orthoclase,cristobalite, Phenocryst sanidine contains significant concentrations and minor mordenite. Some of thesetufs are nearly pure of Na, whereasthe authigenic orthoclasecontains almost ferrierite. The chemical compositions of severalferrierite none (Table 3). This allows the two feldsparsto be easily samplesfrom Lovelock are given in Table 2. Like other differentiated by SIMS or EDX analysis when the grain ferrieritesamples (Wise and Tschernich,1976), these are size or textures are not definitive. magnesian. The tuffaceous ferrierite samples from the Mordenite tuffs. Although most zeolitic tuffs contain Lovelock, Stillwater, and South Korean occurrences, both ferrierite and mordenite, the predominanceof mor- however, have distinctively high K concentrations. denite in some tuffs made textural distinctions possible. There are severalmanifestations of the preservationof Close inspectionof Figures I2a and l2b revealsa fine, the microscopic vitroclastic texture. One of these takes lacelike network of authigenic orthoclaseand cristobalite, the form of ferrierite and authigenic potassium feldspar apparently positioned at former rims of glassshards. This in a dense,fine-grained intergrowth (Fig. 9) in which the microtexture was also observedin the ferrierite tuffs, but orthoclase inhabits the periphery of former shards. An- the contrast betweenzeolite-filled void spaceand the nar- other feature is the presencethroughout the tuffs ofshard- row network of orthoclaseand cristobalite was not so well shapedvoids. It is not always obvious whether particular developed as in the mordenite-rich tuffs. Note also the voids represent primary porosity or secondaryporosity presencein Figure l2b of a double-walled network main- introduced by the dissolution of glass.Those voids with ly composed of orthoclase.The spacebetween the walls definite cuspate pumice morphologies appear to be dis- appears to be void. The mordenite tuffs are in general solved shards.That early ferrierite crystalsgrew into pri- much more porous than ferrierite-rich tuffs becausethe mary pores is supportedby the observation in core 4 tuffs mordenite grows, with few exceptions, as bundles and of ferrierite crystal egowth on fairly fresh glass shards. radial sheavesof prismatic crystals.These (e.g., Fig. l2a) Where ferrierite was permitted to crystallize in an uncon- are morphologically indistinguishable from acicular fer- strained spacethe morphologiesrepresented in Figure l0 rierite. In contrast to the intimate intergrowth of ferrierite were observed. A third feature is the occurrenceofcris- with orthoclase, in which the individual crystallites are tobalite, which is commonly found as sphericalpolycrys- often not discernible by SEM, no similar textures were tals, often seenalong or adjacent to remnant vitric out- observedfor mordenite. lines. In Figure 6 the cristobalite spheresare just visible Typical Lovelock mordenite and clinoptilolite com- within the limbs of the relict shards.The K SIMS image positions are given in Table 4. The somewhat unusual in Figure I l, which clearly distinguishes the orthoclase composition for the clinoptilolite (high Si/Al and high

TABLE2. Ferrieritechemical analyses

sio, 69.7 69.9 69.3 72.9 70.4 70.1 76.2 73.5 73.7 69.6 't2.1 Al2o3 10.6 10.3 10.9 11.4 10.0 11.6 11.3 11.3 11.1 Mgo 0.48 1.98 2.51 3.13 1.61 1.95 1.74 1.89 2j9 2.48 '1.74 'l.72 CaO 0.99 1.46 1.06 0.61 0.74 2.78 2.89 2.10 Naro 1.18 0.30 o.20 0.59 0.01 0.01 0.00 0.08 0.30 0.30 KrO 4.11 4.12 4.01 3.64 5.05 5.46 1.70 1.97 2.56 2.96 0.68 0.48 0.2'l Tio, 0.40 0-01 0.27 0.45 0.04 0.00 0.70 't.21 FeO o.82 0.25 0.73 1.36 0.23 0.45 1.41 1.45 0.66 Total 89.02 87.90 89.45 94.55 87.89 90.80 96.14 93.71 93.83 89.01 Remarks matrix matrix matrix vug matrix matrix Sample no. KP52 KP52 KP52 KP52 SR64 SR64 c2-14.3 c2-'t4.3 c2-17.4 C2-17.4 RICE ET AL.: FERRIERITE CRYSTALLIZATION 3r9

Fig. 6. Backscatteredelectron image of a crosssection of b sampleC2-3.7 showing replacement of glassby clinoptilolite(CL) and cristobalite(CR). Ferrierite(FER) and orthoclase(OR) formedbetween the limbs of the shard.Some of the ferrierite has grown normal to the former glasssurface. Note also that cristobalitespheroids occur as discretecrystallites within the limbs.

Ca) makes a note concerning heulandite group nomen- clature worthwhile. The zeolitesheulandite and clinopti- lolite have the same framework structure and are distin- guished on the basis of composition [heulandite if Ca > (Na + K) accordingto Mason and Sand, 1960,or if Si/ Al < 4 according to Boles, 19721or structural stability upon heating to 450 "C (heulandite if the structure col- lapses,according to Mumpton, 1960).Our phase(Table 4, analyses3 and 4) could be termed heulandite on the Fig. 7. (a) Back-scatteredelectron image of sample C2-17'4 basis of cations becauseCa > (Na + K) or clinoptilolite orthoclase (OR), ferrierite (FER), and smectite (SM) (-4.8). showing on the basisof frameworkSi/Al ratio We did not filling the pores of a relict shard. The shard walls are the con- perform the heating test. We call this phaseclinoptilolite nected curvilinear features separatingrounded pores. Some of becauseheulandite rarely has a sedimentary occurrence, this wall material is now cristobalite (CR). (b) Fracture section whereasclinoptilolite and mordenite have a frequent as- of sampleC2-17.4 showingempty pore space(black) rimmed by sociationin alteredtuffs (e.g.,Gottardi and Galli, 1985), smectite and ferrierite. particularlyin high-silicaenvironments. of Lake Tecopa, a saline, alkaline lake. Other examples Parageneticsequence include Boles and Surdam (1979), Brey and Schmincke One of the first changesattending glass alteration at (1980), and Hay (1963). Smectite may not always precede Lovelock was the crystallization of smectite rich in Mg, zeolitization, but it clearly is an early stage ofhydration Ca, and Fe as an abundant shard-rimming phase(e.g., in specimensC3-100 and SR74-76).The appearanceof clay, and in particular smectite, is a frequently observed phe- TABLE3. Feldsparchemical analyses nomenon in altered glassesof a range of compositions. Smectite was reported, for instance, by Broxton et al. sio, 68.7 69.1 72.1 69.3 (1987)in alteredrhyolitic glassesat YuccaMountain, Ne- Al,o3 15.3 18.7 14.8 18.0 vada; by Keller et al. ( 197I ) in altered rhyolite near a hot Mgo o.28 0.80 0.68 0.00 spring in Michoacan, Mexico; and by Boles and Coombs CaO 0.81 0.32 0.58 0.44 Naro 0.25 0.00 0.23 3.10 (1975)as the first phaseofalteration in Triassictuffs of K"O 13.2 11.1 11.2 8.67 andesitic through rhyolitic composition. Davies et al. Tio, 0.00 0.00 0.00 0.01 0.00 0.00 0.65 (1979) reported glass grains in first FeO 0.00 smectite-coated the Total 98.41 100.00 99.62 100.18 alteration zone of Guatemalan andesitic volcaniclastics. Type orthoclase orthoclase orthoclase sanidine Sheppardand Gude (1968) reported smectiteas a rem- Sampleno. KP52 c2-'t4.3 c2-17A SR3 nant phase adjacent to altered glassy shards in the tuffs 320 RICE ET AL.: FERRIERITE CRYSTALLIZATION

Fig. 8. (a) Cross section of a single glassshard (G) with fer- Fig. 10. Fracturesection of sampleSR67 showing ferrierite rierite at its periphery. 0) Higher magnification image of fer- habits.(a) Radiatingbundle of fibers.(b) Boxyacicular crystals. rierite (FER) and smectite (SM) adjacent to the glass. Micro- analysisand morphology confirmed the identify of ferrierite. alteration of the unstable natural glasses.Hay (1963) showed that the early-formed clay in the John Day For- mation, resulting from percolatingground waters,was an important agent in altering the water chemistry, render- ing it favorable for zeolite crystallization by raising the pH. There are several reasonsfor suggestingthat ferrierite is probably the first zeolite to appear in the transforma- tion ofglass: (1) ferrierite occurson silica and clay sub- stratesin core 2 tuffs, in combination with orthoclaseand smaller amounts of clinoptilolite, (2) ferrierite coexists with glass and clay in core 4 tuffs, and (3) the vertical zonation within the deposit as a whole shows a progres- sion from glassto ferrierite to mordenite. At Lovelock, cristobalite and clinoptilolite cocrystal- lized as replacementsfor dissolving glass,probably after the onset of the crystallization of ferrierite, mordenite, and orthoclase.The occasionaldraping of mordenite on clinoptilolite (SRl.5) suggeststhaI, at leastin those par- Fig. 9. Backscatteredelectron image ofcross sectionofdensely ticular tuffs, mordenite postdated the clinoptilolite. intergrown ferrierite and orthoclase(ferrierite tuffSR67), the lat- Based on their coexistencein many tuffs, mordenite ter showing the relict shard position. growth occurred during or after ferrierite crystallization. RICE ET AL.: FERRIERITE CRYSTALLIZATION JZI

Fig. I l. SIMS images of a cross section of sample SR67. Most of the areasof high K concentration representorthoclase. The two grains in the bottom ofthe field are sanidine pyroclastic grains.

There is no convincingevidence that one crystallizedat centric to adjacent shard pore outlines, could indicate the expenseof the other. Both ferrierite and mordenite crystallization contemporaneouswith ferrierite. In these crystallizeinto openpore spacein the tuffs,but therewas instances,orthoclase crystallization occurred along zones no evidence from TEM characterizationthat these two of microchemicalenrichment of K and Al. Overgrowths zeolites crystallized in close proximity in the fashion and crack-fillingsof orthoclase-richmaterial suggestthat that ferrierite and orthoclasedid. The textural preference in those instancesit persistedafter the main episodeof of mordenite for filling pores produced by elimination of zeolitrzahon. shard grains, requiring as it does an advancedstage of Figure l3 schematicallysummarizes the paragenetic glassdissolution, is interpretedto occur later than the sequenceofauthigenic phasesin the Lovelock deposit. intimate growth of ferrierite and orthoclase. is minor to major phasein the hole no. Cristobalite a Bur,x crrynrICAL AND MTNERALOGICALTRENDS 2 tuffs as well as in some ferrierite tuffs. It also coexists with clinoptilolite and orthoclase,but there was no pet- Bulk chemistryof core hole no. 3 rographicevidence of the time relationshipof cristobalite Bulk chemical data for core samplesfrom hole no. 3 and ferrierite. fig. 1a)suggest that severaldistinct units arerepresented Orthoclaseappears to have been the last phaseto crys- in the 30-m thick deposit of rhyolite pyroclastics.Al- tallize. The orthoclasetextures in the ferrierite tuffs, such though the averagepyroclast grain size, as inferred from as its presenceat the peripheryof former shardsor con- the texturesin SEM images,seems to be relatively con- 322 RICE ET AL.: FERRIERITE CRYSTALLIZATION

Tlele 4. Mordenite(1 , 2) and clinoptilolite(3, 4) chemicalanal- yses

sio, oo.o 68.3 72.8 76.8 Alro3 10.2 10.8 13.2 13.5 Mgo 0.00 0.00 1.16 0.84 CaO 2.0s 2.17 4.00 4.71 Naro 0.24 0.97 0.45 0.20 KrO 0.84 1.10 0.96 0.78 Tio, 0.01 0.00 0.00 0.00 FeO 0.09 0.08 0.00 0.00 Total 80.02 83.32 92.59 96.87 Remarks vugfilling vugfilling rhombus rhombus

compositionof this clay is Mg rich, as distinct from later clays, which were detennined by TEM/EDX to be Fe rich. These later clays occur as thin alteration products on ferrierite laths and are not considered important in the context of ferrierite genesis.As shown in Table 5, layers A and B both have higher MgO than does layer C. Becausethe initial bulk chemicalcompositions of A, B, and C were probably comparable, zeolite-forming reac- tions must have involved some additional chemical transformation specificto the tufs shallowerthan 12.5 m. Bulk compositions of the upper zeolitic tuffs in hole no. 3 show decreasingalkali-metal concentrations with depth (Fig. l4). In the near-surfacetuffs of layer A the KrO concentration (6-80/o)is higher than in fresh rhyolite, but KrO gradually decreasesto a relatively low value in layer B. Na content also decreasesdownward over this interval, whereasCa increasesgradually. The SEM data support the presenceof K-enriched zoneswithin the layer A tuffs, and it may be that there is more authigenic feld- spar in the near-surfacetuffs. Many examplesof KrO en- Fig. 12. Crosssections of mordenitetuff SR3 showing(a) richment, such as authigenic orthoclase overgrowths on individualvoid space,outlined by orthoclasecrystallite network relict pyroclastic feldspars and feldspar-rich veinlets in and partiallyfilled with prismaticmordenite, (b) double-walled textureoforthoclase and minor cristobalite. the zeolitic tuffs, reflect the presenceof a K-rich fluid. The questionremains whether the K and Na wereadded from an external source or reflect variations in original stant over the entire length of core 3, the chemical vari- tuffchemistry. Fractionation of KrO is not uncommon in ations require the recognition of several distinct units. some ignimbrite flow sequences,such as the Bishop Tuff Phaseidentification basedon PXD helped in delineating (Hildreth, 1979)and the Valley of Ten ThousandSmokes the zones.From the surfaceto about l0 m is a ferrierite- (Mahood, 1986),in which increasinglyless silicic lavas rich zeolitic zone (layer A), containing some authigenic extrude onto earlier more silicic flows. However, if this orthoclaseand cristobalite but whose content ap- were the case, K2O and Si/Al values from the base of proachespure ferrierite in someplaces. These tuffs appear layer B to the top of layer A in hole no. 3 would be to have the same mineralogical characteristicsas the fer- expectedto decrease.They demonstrateprecisely the op- rierite-rich rocks that crop out in the southwestcorner of posite trend. It also seemsunlikely that concentrations the area.Between l0 and 12.5 m is an Mg-rich, low Si/ varying by a factor of 4 over just l5 m would result from Al zone (layer B) containing ferrierite and smectite. At continuous changesin the source chemistry or that an about the 12.5-m level there is a break in the dominant air-fall pyroclastic deposit would reliably record these assemblage.Below 12.5 m ferrierite is absentand, with changes.A more probable explanation is the addition of the exception of some clay-rich beds and pyroclastic frag- alkali-rich fluids or K-rich fluids that partially exchanged ments, only vitric ash (layer C) is present. Judging from with Na in the deposit, thus reducing the total Na con- the material recovered at 30 m, smectite rims probably centrations below typical values for rhyolites. occur on most of the glassshards below 12.5 m, even in those beds that are primarily vitric. This indicates that, Mass balance although smectite alteration preceded ferrierite forma- In order to understand the chemistry of zeolitization, tion, not all of the clay-altered tuffs were zeolitized. The particularly with respectto ferrierite crystallization, iI is RICE ET AL.: FERRIERITE CRYSTALLIZATION 323

magnesrum l-."""-_l ironsmectitel;--T smectite ';:,:"w

cristobalite

ctinoptitotiteW rordeniteW

opalineveins I I t216b 012 3 { 5 60 r 2 30 1 2 3 r0 2 4 6 a

Al2o3 MgO CaO Na2O KzO Time _-______> Fig. 14. Variationof elementconcentrations with depthin Fig. 13. Parageneticsequence ofauthigenic recog- hole no. 3 tufs. LayerA is predominantlyferrierite and ortho- nizedin the Lovelockdeposit. clase.Layer B is predominantlyferrierite and smectite.l,ayer C is glassand clay. necessaryto determine whether the reactions were iso- chemicalor required significantfluxes of material. Changes the vug occurrencesin andesite and basalt have linings in elementalconcentrations were calculatedassuming that of chalcedonyin intimate contact with ferrierite (Gottardi the composition of core 3 vitric tuffs of layer C was rep- and Galli, 1985).The presenceof cristobaliteimplies the resentative of the original tuffs from which the zeolites presenceoffree silica during the zeolite diagenesisin the formed. Layers A and B were consideredindividually. It Lovelock deposit. Abundant opaline rocks visible in out- is not known a priori that any of the components were crop are further evidence. truly immobile or insoluble during the hydration reac- The total Mg content of the original rhyolite, only about tions. The compositions were calculated on an H.O-free 0.50/oMgO, was insufficient to account for the vast quan- basis in order to determine the relative changein A and tities of ferrierite in the Lovelock deposit. The addition B from C. Rhyolites or obsidians usually contain only a of Mg in the form of smectite coatings on glass shards few percent HrO, whereas ferrierite tuffs have approxi- appears to be responsible for most of the Mg necessary mately 100/oHrO. It was also assumedconsistent with for ferrierite formation. metasomatism that the volume of the pyroclastic tufs remained approximatelyconstant during the process.This Ferrierite crystal chemistry is the metasomatic assumption. The following relation- Several populations of ferrierite samples were identi- ship was used to calculate A values: fied by electronmicroprobe analysis,namely, a group with Mg/Ca of 2.5 to 3.0 and a group with Mg/Ca of about Ao*iae: 100(oxide".po- oxide..p./oxide..p.) (l) 0.5. Both compositional groups are representedin KP52 in which oxideo is the weight percentof the oxide in layer (e.g.,Table 2, analysesI and 2) and SR64 tuffs. During A and po is the density of tuff layer A. The equivalent TEM/EDX work, substantialvariation in Si, Mg, and Ca equation for layer B relative to layer C was also applied. concentrationswas observed for crystals from tuff speci- There are at least two possibilities that explain the dif- mens as small as I mm3 (Fig. l5). The groupingof fer- ferencesbetween layers A, B, and C: (l) that they were rierite compositions into high and low Mg/Ca groups ap- separateand chemically different units to begin with, or (2) that concentration changeswere produced by the ad- dition or subtraction of material. Both may be correct. Tmle 5. Massbalance in holeno. 3 ferrieritetuffs. Layers A and B have lower Na contents than does C, Upper Lower which suggestsleaching. The fate of the Na is not clear, (A) (B) however. Becausethe K concentration ferrierite ferrierite Vitrictuff tends to be rela- zone zone hole3 tively high and that of the Na low, an alternative mech- oxide (7)* (4)-- (6)-. a (A),% a (B),% anism is coupling through exchangereactions. The ob- sio, 77.8 71.46 75.0 -2.O - 10.0 servations suggest that the reactions assisting zeolite Al2o3 10.3 16.49 12.73 -23 +22 crystallization were not isochemical.All Na.O 1.50 0.78 2.59 -45 -72 three layershave -4.5 -34 much KrO 5.39 3.74 5.39 higher Mg, higher Ca, and lower Na concentrations Mgo 1.06 3.57 0.63 +59 +435 than would be expectedfor a rhyolite. The large relative CaO 1.12 1.98 1.61 -34 - +16 -0.30 differences determined for A and B with respect to C Fe,O" 1.32 2.75 1.79 +45 Tio, 0.15 naf 0.15 0 naf suggestalso that thesewere depositedseparately. The Al-, Total 98.64 98.02 99.83 Fe-, and Mg-rich composition of layer B suggeststhat it 'Mass changes for layers A and B assuming C is the original com- originally had a larger mafic component than did layer A. position(dry basis)using Equation1 with bulk compositionsof core 3 tuffs. Becauseferrierite is a high-silicazeolite, a high activity " Number of samples averaged in parentheses. : of silica in the zeolite-forming fluids is required. Many of t The abbreviationna not analyzed. 324 RICE ET AL.: FERRIERITE CRYSTALLIZATION

LOVELOCKPHASE COMPOSITIONS Lovelockferrierite (KP52)

1.5 (J n v a)n oMg a)v 10 (tv o ca

00i- 30 tr Si/unitcell tro Fresh \-rl Fig. 15. Divalent cation concentrationsvs. Si for individual Na2O+ K2O CaO ferrierite crystalsfrom a small chip of ferrierite tuff(KP52).Data + KrO)-CaOcontents(wtYo) were obtained from TEM/EDX analyses,based on unit-cell con- Fig. 16. NormalizedMgO-(NarO tentsof 72 O atoms. for Lovelockzeolites and tuffs. The dashed field represents about 40 bulk analysesof tuffsfrom Lovelock.

pearsto be a consequenceofCa concentrationdifferences. indicate a fundamental role of Mg-bearing fluids in the Another grouping of ferrierite compositions occurs in the formation of ferrierite in the Lovelock deposit. core 2 tuffs but in smaller amounts than in core 3 tuffs. The cluster of three calcic ferrierite compositions in Fig- DrscussroN ure 16, outside the compositional field defined by bulk pyroclastic zeolitic tufl probably reflects in part a chemistry inher- Comparisonof Lovelock tuffs with other silicic ited from the original core 2 tuff. tuffs Compositions of glass, ferrierite, and smectite dem- Of the many potential controls on ultimate zeolite as- onstrate the chemical control of the precursor glass and semblages,one that is constrainedat Lovelock is the bulk added smectite on the resultant ferrierite (Fig. 16). The glassstarting composition, representedby the fresh glass ferrierite compositions (solid circles) reside between the analysesof C3-30.5, SR78, and C4 tuffs. The chemical alkali-rich glass and the (Mg + Ca)-rich smectite (open compositions representativeof the three layers A, B, and circle). Two inferencescan be drawn from this. First, the C in core 3 are given in Table 5. In Figures l7a-17d, ferrierite in the Lovelock deposit, although alkali-rich Lovelock compositions are compared with those of ap- compared with many other ferrierite compositions, has proximately 40 other pyroclasticglassy rhyolites from the too much Mg to be derived in large quantity from the literature. It was desirable to have the most direct com- pristine glassand HrO alone. All known natural ferrierite parison with like tuffs, namely, pyroclastic fall deposits sampleshave at least 0.50/oMgO. Second,the mordenite of silicic composition,rather than the whole rhyolitic suite. compositions(solid squaresin Fig. l6) and the clinoptilo- Some of these have zeolites associatedwith them, but lite compositions (open squares)are too calcic to be de- most do not. Chemical data were obtained for Lovelock rived from the glass + HrO + smectite. This is one key glasscompositions by electronmicroprobe analysis,SIMS to the subordination of mordenite and clinoptilolite rel- imaging, and TEM/EDX. Lovelock bulk ash (layer C) has ative to ferrierite in this locality becauseit is not likely approximately two times more MgO and about one and that either mordenite or clinoptilolite could accept such one-half times more CaO than other pyroclastic rhyolites large amounts of Mg and remain stable. with the same SiO, content (Figs. l7a, l7b). The MgO To sum up: Mg occurs at significant concentrationsin (<0.05 wto/o)and CaO (-0.2 wto/o)concentrations deter- all of the known natural ferrierite occurrences.The Mg mined for the interior of the Lovelock shards were right concentration per unit cell, based on 72 O atoms, ranges on the trend for a silica value of 780/0. from approximately 0.4 to nearly 3, averagingaround 2 When the compositions of layers A and B are com- (Sameshima,1986). The Lovelock ferrierite,although it pared with the silicic trend, it is found that MgO and has a small Mg content compared with many others, nev- CaO contents are, in general,much higher than expected erthelessrequired the addition of substantial Mg to the for silicic glassytuffof the samesilica content. K concen- silicic glass for development in large quantities. Mg has trations are typical for rhyolites, whereasNa concentra- a specific site in the ferrierite structure, for which it is tions are about a factor of 2 below normal. Thus, al- especially suited, and it is not easily removed by acid though layer C has apparently not been significantly leaching (Vaughan, 1966). These observations strongly leached ofNa, layers B and C probably have been, sug- RICE ET AL.: FERRIERITE CRYSTALLIZATION 325

z0 3.s7toB

1.5 a Mgo CaO .B 1.0 OA o a o .Co a a oa OA tlo oc 3.' 3 I .... aa a 'r'r . o '' d o a on lj'i...o c a ^\: 0.0 74 76 78 80 68 72 74 76 7A 80 SiO2 SiO2

o

O3

Na2O Kzo O a .oG a

. OA . o3

68 70 72 74 76 68 70 72 74 7A 80 SiO2 SiOz Fig. 17. Diagrams of variations in element concentrations for bulk analysesfrom silicic pyroclastic tuffs. Data taken from literature sourcesare solid circles; Lovelock layers A, B, and C, as well as the fresh Lovelock glass(G) in layer C, are represented by large open circles.

gesting that in addition to the clay formation, other hy- Lacustrine deposition could well be involved, as evi- dration reactions accompaniedzeolitization. dencedby the presenceof diatomite beds,authigenic feld- spars that commonly occur in saline, alkaline lake de- Geological setting posits (Hay, 1966), and the relatively sharp boundary Of the many sourcesof aqueousfluids in which zeolites between the zeolite tuffs and the underlying glassybeds. occur, those possibly important for the Lovelock deposit From SEM observations of morphological characteris- are hydrothermal, percolating ground water, and lacus- tics, the diatoms were identifiedas Melosirq crenulata or trine. There is an absenceof minerals diagnostic of hy- Melosira granulata. As far as the authors are aware,these drothermal activity. The widespread silicification does speciesare not constrained by salinity. If connate lake have the appearanceof a fused material, but this feature water is invoked as the sole agent of alteration, the pH neednot be a high-temperaturereaction becausethe silica of the water could not have exceededabout 9 without could be derived from feldspar-forming reactions at low dissolving much of the opaline silica. The fairly sharp temperature.The authigenicfeldspar itself, authigenic or- contact between vitric tuff and ferrierite tuff in hole no' thoclase, although perhaps representativeof the highest 3 could be interpreted several ways. One interpretation temperature or highest grade of alteration attained at would favor a lacustrine depositional environment be- Lovelock, need not be ofhydrothermal origin. For these causethe bottom ofthe lake would be representedby an reasons,a hydrothermal origin seemsunlikely. abrupt break. A fault juxtaposing relatively fresh tuffs 326 RICE ET AL.: FERRIERITE CRYSTALLIZATION

with highly zeolitized,tuffs could also explain the discon- Iijima (1971) noted that clinoptilolite analysesdeter- tinuity. The vertical changesin alkali concentrationsand mined by electron microprobe were anomalously high in the presenceofdiatomaceous beds could both be due to silica and attributed this to finely divided silica mixed the presenceof a relatively freshwaterlake. However, the with the zeolite. Severalsimilar observationswere found results of mass-balancecalculations described above, in the course of the presentwork as well. which show addition of both Mg and K and the removal Na was either leachedfrom the bedsor exchangedwith of Na, strongly argue for an open ground water system. K prior Io zeolitization. Perlite textures point to the like- lihood of Na leaching,since Na lossusually accompanies Chemical and textural developmentduring diagenesis the early stagesofglass hydration (Jezekand Noble, I 978). Becauseof the widespreadoccurrence of Mg-rich clay A transfer of Na could have had an important influence coatingson the glassshards and the relatively minor con- on the resultant assemblagebecause ferrierite, although centrations of alkaline-earth cations in the fresh glass,it almost invariably containing minor Na, does not prefer is inferred that Mg2* and Ca2* were added to the vitric it as mordenite does. Furthermore, although reactionsin pile by aqueousfluids sometime during or after deposi- the presenceof high c"" could have precipitated phases tion. This could have been accomplishedthrough either such as albite or analcime, such relations are not found the presenceof brackish connate HrO derived from a at Lovelock. Presumably because the prevailing silica saline lake or by leaching ofnearby basaltsand transport concentrations were high, mordenite is the main Na- of Mg and Ca to the Lovelock site. The smectite also had bearing phase. It is possible that low Na concentrations important implications for the textural development of were necessaryin addition to high Mg concentration in the deposit. The Lovelock zeolite tuffs, as do most other order that large quantities of ferrierite could crystallize. zeolitetuffs, preserveda vitroclastic texture inherited from The onset of ferrierite crystallization did not require the precursor ash. Textural observations made in this the completedissolution of the shardsbecause it initially study point to early smectite as a key factor in this pro- formed within primary pores and utilized glassand clays cess.The presenceof smectite at the peripheriesof shard as substrates.The coarsetuffsample 4-8.5 is an example grains was responsiblefor preservingthe vitroclastic tex- in which a mostly vitric tuff shows incipient ferrierite ture, whereasthe porous nature of the coating permitted crystallization (Fig. 7). Here, where pore spacewas am- glassdissolution to proceed.Stanley and Benson(1979) ple, the prismatic habit developed.The ferrierite-rich tuff observedhollow clay coatingsof smectite cementsin vit- SR67 provides an example of a casewhere the spacewas ric rocks from which detrital grains had been dissolved. more limited or in which the activity of K was sufficiently The observed variations in Mg content of the Lovelock high. As a consequence,finer, more massiveferrierite co- ferrierite samples noted earlier could be caused by the crystallizedwith orthoclase(Fig. 8). In this case,the habit proximity of the crystallizing ferrierite to smectite or by of ferrierite is not obvious. Figure 8 illustrates the chem- variation in smectite composition. ical heterogeneitywithin microscopic zones in many of The widespreadoccurrence ofauthigenic orthoclaseat the tuffs. The clinoptilolite and cristobalite replaced the Lovelock is a feature that was not fully recognized in glasswhile the K-rich phasesferrierite and orthoclaseto- previous descriptionsofthe deposit. According to a study gether fill the primary pore space. by Sheppardand Gude (1973), zonation of minerals often Ferrierite is a rarely occurring zeolite and is unusual occursin lacustrinezeolite depositswhere saline minerals for its Mg-rich composition. Studiesof the known natural are interbedded with zeolites. These typically exhibit a ferrierite compositions(Wise and Tschernich,1976; Sa- sequenceproceeding from glassunder freshwater condi- meshima, 1986)reveal that, without exception,ferrierite tions to zeolites under moderately saline conditions to is a high-silica zeolite with a narrow range of framework potassium feldsparand searlesiteunder highly saline con- Si/Al ratios (-4.5-6) and that Mg is an essentialingre- ditions. No saline minerals have been recognizedin the dient. The conditions for ferrierite crystallization in the deposit. Thus, rather than indicating the extent of dia- Lovelock deposit appear to require both high concentra- genesisin a highly saline environment, crystallization of tions of silica, supplied by the rhyolite, and of Mg. Haw- orthoclase in the Lovelock deposit may be more a con- kins and Roy (1963) pointed out that clay mineralswill sequenceofhigh K activity in the altering fluids. normally crystallize from Mg-bearing solutions formed Cristobalite and clinoptilolite formed as diageneticre- by glassdissolution. The fact that ferrierite is so abundant placementsof glass shards. One of the petrographic fea- in the Lovelock deposit must dependnot only on such tures observed(e.g., C2-17.3) was relict shardsreplaced influencesas pH, temperature, and composition but also by silica and discrete crystals of clinoptilolite. This re- on the nature of structural units present in the fluids. placement was probably the result of a dissolution-pre- Certainly the combination of high Mg and high Si was cipitation mechanism in which local shards were dis- important for ferrierite to crystallize. It is also possible solved, with or without the presenceofa gel phase, and that a templating processanalogous to that proposed for reprecipitatedwithin the samelocal volume. This process large organic cations in some synthetic zeolites (Barrer, could have occurred on a shard-sized scale becausethe l98l) was operative during diagenesisin the Lovelock dissolution of a rhyolite glass shard will yield a compo- deposit. As noted above, Mg has a special role in the sition similar to silicaand any of the zeolitesrepresented. ferrierite structure. The presenceof hydrated Mg , RICE ET AL.: FERRIERITE CRYSTALLIZATION 32'l

Mg(HrO)l+, in crystallizing fluids likely fosteredthe knit- mala. Society of Economic Paleontologistsand Mineralogists Special 281-306. ting together of the ferrierite tetrahedral framework. Be- Publication, 26, Goldstein,J.I., Costley,J.L., Lorimer, G.W., and Reed' S'J'B' (1977) K ions can reside indiscriminately in the fer- Elec- cause the Quantitative X-ray analysisin the electron microscope-Scanning rierite channels,responding to framework chargebut not tron Microscopy/ 1977, 1, 315-323. affectedby stringentsteric considerations, the K concen- Gottardi, G., and Galli, E. (1985) Natural zeolites.Minerals and Rocks trations are probably an accidentof the prevailing solu- 18, 409 p. Springer-Verlag,Berlin. R.P.D. (1918) On ferrierite, a new zeolitic mineral from British tion chemistry. Graham, Columbia; with notes on some other Canadian minerals. Transactions The differencesin ferrierite composition from one bed of the Royal Societyof Canada,3 (12)' 185-20 I to another are indicative of the local chemical control on Hawkins, D B, and Roy, R. (1963) Distribution of trace elementsbe- ferrierite formation and the lack of homogenization of tween claysand zeolitesformed by hydrothermal alteration ofsynthetic Cosmochimica Acta, 27 785-795. diagenetic fluids on a large scale. Which of the zeolites basalts.Geochimica et ' Hay, R.L. (1963) Stratigraphy and zeolitic diagenesisof the John Day crystallized was a function of the chemical composition Formation of Oregon.University of California Publicationsin Geolog- of the mineralizing fluid, and this, in turn, was controlled ical Sciences,42, 199-262. by the compositions of parent glassand the mineral frag- -(1966) Zeolites and zeolitic reactions in sedimentary rocks. Geo- p. ments in the pyroclastic deposit. The ferrierite chemical logical SocietyofAmerica SpecialPaper, 85, 130 and Wohletz, K. (1985) Volcanic ash,246 p' University of TEM/EDX also suggest(e.g., Fie. 15) Heiken. G.. data obtained by California Press,Berkeley, California. that the aqueouschemistry can vary on a very fine scale, Hildreth, W. (1979) The Bishop tuff Evidence for the origin of compo- perhapsapproaching the volume of a few pores.We know sitional zonation in magma chambers.Geological Society of America that pH can have alarge effect on the proportions ofAl SpecialPaper, 180, 43-75' (1971) and origin of clinoptilolite in the Naka- and Si speciesin aqueoussystems. For example,the ratio Iijima, A. Composition nosawatuffof Rumoi, Hokkaido. In Molecular Sieve Zeolites-I, Ad- of two critical ionic species,Si(OH)"/Al(OHh, may vary vancesin Chemistry Series,l0l, 334-341. by as much as a factor of 4 betweenpH 9 and l0 (Mariner Jezek,P A., and Noble, D.C. (1978) Natural hydration and ion exchange and Surdam, 1970).This diageneticsystem is clearlyone of obsidian: An electronmicroprobe study. American Mineralogist, 63, of several scalesof mass transfer through time. The first 266-273. E.R. (1971)The natureofopal I' Nomenclature formation, was the result of Jones,J.B., and Segnit, event, magnesium smectite and constituent phases.Joumal of the GeologicalSociety of Australia, widespread but heterogeneousinflux of fluids into the l8 (l), 57-68 tuffs. Crystallizarion of zeoliteswas apparently controlled Keller.W.D., Hanson,R.F., Huang,W.H.' and Cervantes,A' (1971)Se- on a much smallerscale. quential active alteration of rhyolitic volcanic rock to endellite and a precursorphase of it at a spring in Michoacan, Mexico. Clays and Clay Minerals, 19, l2l-127. AcrNowr,nocMENTS kta, D.P. (1935) A high-resolution, single ion sensitivity video system for secondaryion microscopy In A. Benninghoven,R.J' Colton, D'S work portion of the first author's Ph.D. dissertation This representsa Simons, and H.W. Werner, Eds., Secondaryion mass spectrometry wishes Exxon Researchand Engineer- at Lehigh University. He to thank (SIMS v), Springer Seriesin Chemical Physics,vol. 44' p' 232-234' for this researchand allowing its publication, and ing Co. supporting Springer-Verlag.New York A.J. in particular, for their encouragement. M.M.J. Treacy and Jacobson, Mahood, G. (1986) Waxing and waning magma chambers:A comparison grateful to Charles B. Sclar for his helpfulnessas the major He is also of zoning in ignimbrites and plutons. Fourteenth International Min- microscopywas provided by D.P. kta and W.B. Lamberti advisor. SIMS eralogicalAssociation Abstractswith Program, 163. EngineeringCo. and is geatly appreciated.Re- of Exxon Researchand Mariner, R.H., and Surdam, R.C. (1970) Alkalinity and formation of reviewerand editorial comments views by R.J. Iander and an anonymous zeolitesin saline alkaline lakes.Science, l7O'977-980- by D.L. Bish contributed to significantimprovements in the manuscnpt. Mason. B , and Sand, L.B. (1960) Clinoptilolite from Patagonia'The re- lationship betweenclinoptilolite and heulandite. American Mineralo- RrrnnsNcrs cITED gist,45,341-350. Mumpton, F.A. (l 960) Clinoptilolite redefined. American Mineralogist, Barrer,R.M. (1981)Z€olites and their synthesis.Zeolites, l, 130-140. 45,35r-369. (1972) properties, cell dimensions, and Boles,J.R. Composition, optical Noh. J.H., and Kim, S'J. (1986) Zeolites from Tertiary tuffaceousrocks group American Miner- thermal stability of some heulandite zeolites. in Yeongil area, Korea. In Y. Murakami, A' Iijima, and J'W' Ward' alogist,56, 1463-1493. Eds., New developmentsin zeolite scienceand technology' 28, p' 59- (1975)Mineral reactionsin zeolitic Triassic Boles,J.R., and Coombs,D S. 66. Proceedingsofthe Seventh International Zeolite Conference,Ko- Societyof America Bul- tuff, Hokonui Hills, New Zealand.Geological danshaand Elsevier, TokYo. letin, 86, 163-173. Regis,A.J. (1970) Occurrencesofferrierite in altered pyroclasticsin cen- Boles, J.R., and Surdam, R.C. (1979) Diagenesisof volcanogenic sedi- tral Nevada. Geological Society of America Abstracts with Program, ments in a Tertiary saline lake; Wagon Bed formation, Wyoming' 2, 661. American Journal of Science,279, 832-853. Sameshima,T. (1986) Ferrierite from Tapu, Coromandel Peninsula,New Brey, G., and Schmincke,H.-U. (1980)Origin and diagenesisof the Roque Z,ealand, and a crystal chemical study of known occurrences' Miner- Nublo breccia, Gran Canaria (Canary Islands)-petrology of Roque alogical Magazine, 50, 63-68. Nublo volcanics,II. Bulletin ofVolcanology, 43, 15-33. A.J. (1968) Ferrierite, Penhing County, Nevada. Broxton, D.E., Bish, D.L., and Warren, R.G. (1987) Distribution and Sand, L.B., and Regis, for 1966, Geological Society ofAmerica Special Paper, chemistry of diageneticminerals at Yucca Mountain, Nye County, Ne- In Abstracts vada. Clays and Clay Minerals, 35, 89-l 10. l0l,189. genesisof Cormier, W.E., and Sand, L.B. (1976) Synthesisand metastablephase Sheppard,R.A., and Gude, A.J., III (1968) Distribution and Inyo transformations of Na-, Na,K-, and K-ferrierites. American Mineral- authig,enic silicate minerats in tuffs of Pleistocene l-ake Tecopa, Paper,597, ogist,6l, 1259-1266. County, California. U.S. Geological Survey Professional Davies,D.K., Almon, W.R., Bonis, S.B.,and Hunter, B.E. (1979)De- 38 p. -(1973\Zqolites in tuff- position and diagenesisof Tertiary-Holocene volcaniclastics,Guate- and associatedauthigenic silicate minerals 328 RICE ET AL.: FERRIERITE CRYSTALLIZATION

aceousrocks of the Big Sandy formation, Mohave County, Arizona. Stewart, J.H., and Carlson, J.E. (1978) Geologic map of Nevada. U.S. U.S. Geological Survey ProfessionalPaper, 830, 36 p. Geological Survey, scale 1:500000. Sheppard,R.A., Gude, A.J., III, and Mumpton, F.A. (1983) Field trip Vaughan, P.A. (l%6) The crystal structure of the zeolite ferrierite. Acta stop 6, Invelock zeolite deposit,Lovelock, Nevada. In F.A. Mumpton, Crystallographica,2l,983-99O. Ed., International Committee on Natural ZeolitesZeo-Trip '83, p. 48- Wise, W.S., and Tschemich, R.W. (1976) Chemical composition of fer- 55. Allen Press,Larvrence, Kansas. rierite. Arnerican Mineralogist, 6I, 60-66. Stanley, K.O., and Benson, L.Y. (197q Early diagenesisof high plains Tertiary vitric and arkosic sanilstone, Wyoming and Nebraska. Society ofEconomic Paleontoldgistsand MineralogistsSpecial Publication,26, Maruscrrrr RrcErvEDFesnuA,ny 22, l99l 401-423- Mercuscnrpr AccEprEDNoverrlsen l. l99l