Sieve-Textured Plagioclase in Volcanic Rocks Produced by Rapid

American Mineralogist, Volume 77, pages 1242-1249, 1992

Sieve-texturedplagioclase in volcanic rocks producedby rapid decompression

SrnpHnN T. NBr,soN, Anr MoNrANrA.* Department of Earth and SpaceSciences, UCLA, Los Angeles,California 90024-1567,U.S.A.

Ansrnlcr Disequilibrium textures,particularly the coarsesieve texture ofplagioclase,are common in orogenic volcanic rocks. The textures are usually interpreted as resulting from magma mixing, but they may occur by rapid decompression,where heat loss is minor relative to the ascent rate. We conducted high-pressurepiston-cylinder experiments on an andesite to test this hypothesis. Experiments starting at 12 kbar, followed by isothermal pressure releasein increments of 2, 4, and 6 kbar, produce sieve textures in plagioclasevery much like those in many volcanic rocks. Therefore, the presenceof sieve-texturedplagioclase should not be taken as a priori evidencefor magma mixing. Many volcanic systemsprob- ably experienceconditions of decompressionsimilar to those simulated in this study, and decompressionis considered to be a simple mechanism to produce such textures, as it requires no addition of heat or mass.Rapid decompressionmay also operate in conjunc- tion with magmamixing.

INrnonucrroN One interpretation of coarse sieve textures similar to Plagioclasecommonly exhibits a variety of disequilib- those in Figure I is rapid skeletal growth resulting from rium textures in volcanic rocks, especially in orogenic undercooling(Kuo and Kirkpatrick, 1982).However, melt andesites.These textures often include combinations of inclusions are commonly elongateparallel to, or abutted complex zoning patterns and resorption features in pla- against, polysynthetic twin planes (Fig. l), or they cut gioclasethat record changingphysical conditions in mag- boundaries betweenindividual crystals in glomerocrysts. matic systems.Thus, the potential exists to learn much Given that glomerocrystsand polysynthetic twin planes about magmatic processesinvolved in their formation if form after crystallization (Dungan and Rhodes, 1978), the textures can be interpreted, with the caveat that er- the cross-cuttingrelationships indicate that thesetextures roneous inferencesmay result if a particular texture can must form by resorption rather than rapid crystallization. arise through more than one process. This study deals Two other mechanismshave been proposedto explain with the origin ofcoarse sieve-texturedvolcanic plagro- the origin of sieve textures in plagioclasefrom volcanic clase (Fig. l) in the light of new experimental observa- rocks. They are magma mixing (e.g.,Dungan and Rhodes, tions and petrographiccharacteristics of resorbednatural 1978;Tsuchiyama, 1985) and magmaticdecompression plagioclase. (e.g.,Vance, 1965; Stormet 1972; Nelson, 1989). The We interpret the texture of the plagioclasein Figure I possible effectsof magma mixing have been experimen- as representingthe remnants of an interconnecting net- tally investigated, although only under a limited set of work ofchannels, developedas the crystal underwent dis- conditions (Tsuchiyama,1985). To date, these experi- solution in responseto physical and chemical changesin ments have not produced textures like those in Figure l, the magma reservoir or conduit. Initially, dissolution al- although future experiments may produce them if they lowed diffusive and convective communication of the liq- are carried out under an expanded range of pressures, uid in the channelswith the bulk of the melt. However, temperatures, and compositions. Many studies have in many volcanic rocks, sieve-textured plagioclasehas shown the widespreadoccurrence of magma mixing as a euhedral rims that may result from subsequentcrystal petrogeneticprocess, and it is not our intent to dispute growth in responseto loss of volatiles or cooling prior to its importance. However, our results indicate that coarse eruption. If the rims encapsulatethe crystal, this may sieve textures may result from the rapid ascent of mag- leave the resorbed plagioclasewith entrained liquid in- mas; therefore,invoking magma mixing solely on the ba- clusions that have no obvious means of communication sis of coarsesieve-textured plagioclase may not be war- with the bulk of the magma. Upon extrusion, quenching ranted. In the absence of other criteria indicative of of the magma leavesthe network of channelsfilled with magma mixing, such as trapped liquid enclaves,disequi- microcrystalline or glassymelt inclusions. librium phenocryst assemblages(e.g., Mg-rich olivine + quartz),or other chemical and isotopic criteria, magmatic decompressionis a geologicallysimpler model, as it does * Presentaddress: P.O. Box I139, Pecos,New Mexico 87552, not require an open system.However, the two processes U.S.A. need not be mutually exclusive.Both mechanismsdesta- 0003-004x/9 2 / | 1| 2-r 242SO2.OO 1242 NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE t243 bilize relatively sodic plagioclase,and all magmasdecom- pressduring ascent.

Magma mixing Some of the relative merits of magma mixing and magmatic decompressionmodels can be evaluated by a re- view of pertinent phaseequilibria and experimental studies. Tsuchiyama's experiments model a mafic magma mixing with a more silicic, sodic plagioclase-bearingmelt, which superheatsand destabilizes the sodic plagioclase (Tsuchiyama,1985). Any calcicplagioclase in the mafic a,'a i magma would be supercooled,leaving a compositionally bimodal population of plagioclasecrystals. Mixing experimentsproduced dusty calcic mantles sur- rounding sodic cores(see Fig. 2 of Tsuchiyama, 1985), an effect produced by diffusional exchangebetween pla- f,n gioclaseand melt, assistedby the development of micro- a channelsin the partially resorbingmantle. However, Tsu- chiyama noted that his resultsdid not produce the coarse sieve textures seenin Figure l, which are similar to textures ascribed to magma mixing (see Fig. 5 of Dungan and Rhodes, 1978).Although it is apparentthat some disequilibrium texturesin lavas may result from the mixing of magmas, it seemsthat coarse sieve textures may also form by decompression.

Magrnatic decompression Plagioclasephase equilibria provide a framework for predicting the possible effectsof rapid magmatic decompression and for designing experiments to model it. In Fig. 1. Photomicrographof the startingmaterial containing Figure 2a, decompressionof the bulk composition, rep- large resorbedplagioclase phenocrysts typical of many inter- resented by the filled squares,from A to B drives the mediateand mafic lavas. Arrow indicatesa memberof a second systemtoward a more calcic plagioclasecomposition and populationof euhedral,often more calcic,plagioclase mrcro- a greater proportion of liquid, as representedby the rel- phenocrysts. ative lengths of the tie lines connecting A and B to the solidi and liquidi. Our experiments were designed to mimic this effect and to test whether rapid decompres- of diopside(Wyllie, 1963).Thus, additional components sion can produce resorption in volcanic plagioclasephe- and crystallizing phasesin natural systemsmay enhance nocrysts. The vigor with which a decompressingplagio- resorption effectscaused by decompression.Our experi- clase-bearingsystem begins to reequilibratedepends upon mental data and thoseof Meen (1987)confirm that, un- the magnitude and rate of pressurerelease. Decompres- der nearly isothermal conditions, increasingpressure de- sion experiments modeled ascent rates >10 m/s (d.P/dt creasesthe An content of plagioclasein multicomponent > 0.2 kbar/min) over absolutepressure intervals of 2,4, natural systems(Fig. 3). This agreeswith the observations and 6 kbar. of other workers(Green and Ringwood, 1967, 1968;Co- The compositional changesproduced by decompres- hen et al., 1967) and supports our model for decompression in the binary system(Fig. 2a) are not large.However, sion-induced dissolution of plagioclasephenocrysts. the effects of additional components in decompressing Although the curves in Figure 2 are for dry systems, systemscan be evaluated in Figure 2b. The curves rep- no magma is strictly anhydrous, making it necessaryto resent the plagioclaseloop with coprecipitating diopside evaluate briefly the effectsof volatiles. It is necessaryto in the system Di-Ab-An. For decompressionintervals of keep in mind that the processof magmatic decompres- equal magnitude, the simultaneous dissolution of diop- sion we describeis rapid, precluding a significantincrease side produces a greater change in the equilibrium com- in volatile pressuredue to crystallization of an anhydrous position of plagioclaseand a greaterproportion of liquid assemblage.Therefore, it is important to consider:(l) the at the final pressurethan in the system Ab-An (Fig. 2a). effects of decompression on vapor-undersaturated sys- The increased sensitivity of the system Di-Ab-An to tems, (2) the effectsof decompressionon vapor-saturated changesin pressureand temperatureresults from the low- systems,and (3) volatile loss. er slopesof the plagioclaseliquidus and solidus in both HrO lowers the temperatureof the liquidi and especial- P-X and T-X projections causedby the coprecipitation ly the solidi, but it does not significantly changethe over- 1244 NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE

IGL tr PL+CL tr PL+CPX+GL O PL+I,IT+GL

(! Plagioclase + Liquid : l.i PL+CPX+L

th th J H qJ 10 ti F< U) 1 atm a

l-i

20 oC T = 1230

15 Plagioclase+ Diopside 1100 1200 1300

H Temperature('C) 10 Plagioclase+ Diopside+Liquitl Fig.3. P-Tprojectionfor andesitemelts modified from Green o ( I 972).Experimental phase assemblages from this studyare also o plottedfor comparativepurposes. Symbols the sameas in Ta- H ble 2.

1 atm voirs. Most mafic-intermediate orogenic magmasare not o 'Lor" loo sufficiently HrO-rich (<3 wtolo)to exsolve a separateva- pil."''tol'o.rnrr" por phaseuntil they ascendto lessthan 2-3 km from the Earth's surface(Gil1, l98l). Therefore,as intermediate Fig.2. (a)P-Xsection of rheAb-An system at 1325.C (see magmasrise through the crust, they are neither sumcient- Morse,1980, Fig. 18.8).(b) P-X sectionof the pseudobinaryly HrO-rich to approach a negatively sloping solidus, nor plagioclaseloop at about 1230"C in the systemDi-Ab-An con- likely to lose volatiles in transit until they reach very structedby movingthe l-atm loop up 100"C to approximate shallow levels. its position at l0 kbar. This is a reasonableassumption, es- peciallyfor the An-rich sideof the diagram(see Morse, 1980, ExpnnrlrnxrAl pRocEDUREAND RESULTS Figs.18.6, 18.9). A givenisothermal pressure drop producesa greaterchange in the compositionand proportionof liquid in Methods the systemDi-Ab-An than in the systemAb-An. The starting material (Table l) was a high-K andesite from east-centralUtah containing phenocrystsof plagioclase, clinopyroxene, orthopyroxene, and olivine, but all slope of the P-X and T-X loops when projected from lackingprimary hydrousphases (Nelson, 1989). Liquidus HrO. Thus, moderate volatile contents do not affect the and solidus curvesare well constrainedbecause of a com- generalphase relationships describedin Figure 2, as long positional similarity to other experimentally investigated as the magma does not become vapor saturatedand de- andesites(Green and Ringwood, 1968; Green, 1972; gas. The presenceof HrO will assistdecompression-in- Meen, 1987)(Table l). High-pressureexperiments were duced resorption by reducing viscosity of the magma, conducted in a piston-cylinder apparatus with a furnace which permits increasedascent and diffusion rates. De- assembly2.54 cm in diameter,similar to that described compression of a vapor-saturated magma would hinder by Boettcheret al. (1981).The starting material was sealed resorption ifit causesthe systemto approacha negatively inside Pt capsules,which in turn were sealedwithin cap- sloping solidus in P-T space-a condition likely to pre- sules containing hematite. The HM buffer was used to vail at crustal pressures.On the other hand, the rapid loss minimize Fe loss to Pt and to maintain anhydrouscon- ofvolatiles during eruption will arrest resorption by ele- ditions. Nonetheless,moderate Fe loss did occur (4-4.5 vating the system'ssolidus temperature.Gradual volatile wtolo).Experiments at atmospheric pressureemployed a loss could induce a shift in plagioclasecomposition to- CO/CO, gas-mixing furnace near the value of the HM ward the Ab end-member and induce crystallization, buffer. Temperatures were monitored with Pt-PteoRh,o thereby tending to counteract the effects of decompres- thermocouples. sion. However, it is difficult to envision that this is an We conducted two sets of experiments. The first set important process,except in very shallow magma reser- establishedthe equilibrium phaseassemblages and com- NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE t245

TleLe 1. Comparisonof the compositionsof startingmaterials TABLE2. Resultsand conditionsof experimentset 1

Duration P (kbar) rfc) (h) sio, 56.08 55 58 56.40 59.72 Tio, 0.94 0.88 1.40 0.68 12 1250 24 GL Al,o3 16.98 17.39 16.60 16.82 12 1225 24 GL, PL, CPX FeO 9.14 8.28 8.70 6.30 12 1150 48 GL, PL, CPX MnO 0.13 na 0.10 0.13 10 1150 48 GL, PL, CPX Mgo 2.85 4.82 4.30 310 I 1150 48 GL, PL CaO 5.99 7.13 8.50 7.16 o 1150 48 GL, PL Na.O 3.36 3.81 3.00 3.99 1 atm 1150 96 GL, PL, MT KrO 3.30 2.11 1.00 1.28 P,O. 0.38 na na 0.20 Notei All exoeriments were conducted with the HM buffer to minimize Total 99.16 100.0 100.0 99.39 Fe loss to the capsuleand keep the assemblageanhydrous. Temperature was monitored using a Pt-PLoRhlothermocouple. Nofe.' na: data not available. Numbers 1-4 designate the following: ' GL: glass;PL: plagioclase;CPX: clinopyroxene;MT: magnetite. 1 : starting materialfor this study (sampleGey-2, Nelson, 1989); 2 : Meen (1987)sample 2085; 3: Greenand Ringwood(1968) basattic andesite; 4: Green(1972) sample 68-66. ent rates becauseof the Or component, which increases from 4 to 8 molo/obetween I atm and 12 kbar. Despite positions at atmosphericpressure and 6, 8, 10, and 12 considerablescatter in the data, it appearsthat the com- kbar (Table 2). All experiments were at 1250 "C for the position of plagioclase may change most rapidly as a first 24 h (above the liquidus at all pressures)to ensure function of pressurewhen coprecipitating with diopside complete melting of the crystalline starting material be- Gig. a). This observation is consistentwith the increased fore subsequentcrystallization; they were then rapidly pressuredependence of plagioclaseand liquid composi- cooled to ll50 "C (betweenthe liquidus and solidus at tions when coprecipitating with diopside in the system all pressures). Di-Ab-An (Fie. 2b) vs. the binary system (Fig. 2a). This We designedthe second set of experiments to model probably results from rapid chemical changesin the liq- magmatic decompression(Table 3). The experimentswere uid brought on by the coprecipitation or resorption of held at 12 kbar and 1250"Cfor 24 h to completelymelt clinopyroxene with plagioclase.Although both clinopy- the starting material. The temperature was rapidly roxene and plagioclaseare increasinglysoluble under iso- dropped isobaricallyto I 150'C for another24 h to par- thermal conditions as pressuredecreases, clinopyroxene tially crystallize the liquid. For eachexperiment, the pres- becomesunstable at pressuresless than about 8-10 kbar sure was then dropped either 2, 4, or 6 kbar and held for (Fig. 3). Therefore, at pressuresapproaching the stability 3-12 h. This was accomplishedfor 2- and 4-kbar inter- limit of clinopyroxene,the liquid becomesenriched in Ca vals by releasing piston pressure at a rate of approxi- as the ratio of clinopyroxene to plagioclaserapidly de- mately 0.2 kbar/min. Becauseof thermocouple failure, creases.This allows the crystallization of more calcic experiments at 6-kbar intervals were reloaded into new plagioclases. furnace assembliesand processedagain at 6 kbar. De- compression rates were limited by the speed at which Developmentof resorption textures piston pressurecould be releasedwithout thermocouple The results of the second set of experiments confirm failure. that coarse sieve textures in volcanic rocks can be gen-

Pressuredependence of plagioclasecomposition TABLE3. Resultsand conditions of experimentset 2

The first set of experiments establishesthe phaserela- P tionships as a function of pressure.However, phasecom- inter- Dura- val. tion" positions and the crystallization sequencealso dependon (kbar) (h) Effect on plagioclaseand other phases (Hamilton et al., 1964).Because the experimentswere "f", 2(12-10160 (12) Approx. 10% resorption. A few resorption carried out under very oxidizing conditions, our results channels oer unit volumei in PL. may reflect a different crystallization sequenceand range 4 (12-8) 60(12) Approx.30o/" resorption. up to tens of resorF of mineral stabilities from those experiencedin nature. tion channels per unit volume. 4 (12-8) 52(4) Approx. 10% resorption. Up to tens of resorp- Our experimentsare generallyconsistent with the exper- tion channels per unit volume. imentally determined phase relationshipsof andesitic 6 (12-6) 60(12) Nearly complete resorption of PL. New euhed- melts (Fig. 3). However, the liquidus temperature at 12 ral calcic PL. 6 (12-6) 54(6) Approx. 30% resorption. Up to hundredsof re- kbar of our starting material is somewhathigher than that sorption channels per unit volume. Pseudo- of Green (1972) and the stability of clinopyroxene in our morphic MT after CPX. 6 (12-6) s1 (3) Numerous resorption channels in PL. CPX be- experimentsis alsolimited to somewhathigher pressures comrngopaque. (Fig. 3). * pressures. plagioclase Numbers in oarenthesesindicate the initial and final Microprobe analysesshow that is more Ab ** Number in oarenthesesindicates the duration of the exoeriment at rich at increasedpressure (Fig. a). Plagioclasecomposi- the final oressure. tion varies from about An' to An' betweenatmospheric f Degree of resorption and number of channels estimated for a cubic unit volume100 pm in lengthper side. pressureand l2 kbar. Ab and An contents vary at differ- 1246 NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE

eratedby rapid decompression(Fig. 5). In terms of min- re$ion of clinopyroxene imum time and pressureincrements, sieve texturesap- stability pear after decompressionsof 2 kbar and 12 h, and after I-r 4-kbar decompressionsof just 4 h. The plagioclasecrystals contain a mosaic of melt inclusionsand dissolution channels strikingly similar to those in natural samples (compareFigs. I and 5). Ti The 6-kbar decompressionexperiments revealed a two- a a stage breakdown of clinopyroxene as the samples were to well below its stability limit. The break- ;-.i decompressed down of clinopyroxene, in a sequenceof experiments where postdecompressionconditions lasted 3, 6, and 12 h (Table 3), beganas metastableFe-Ti oxides nucleatein 30 40 50 60 70 the clinopyroxene, rendering it opaque. After 6 h, pseu- Mole Percent Anorthite domorphic aggregatesof oxide after clinopyroxene were dispersed throughout the liquid. After 12 h, essentially Fig. 4. Composition of plagioclaseexperiment products (open all crystalline phaseshad resorbed into the melt, and a squares)as a function of pressurein terms of mole percent An, plagioclase were grow- as determined by electron microprobe. Data from Meen (1987) few new, relatively calcic crystals (filled circles)are shown for comparative purposeswhere tie lines ing in the liquid. join plagioclasecompositions from four starting materialsunder The nature of resorption in plagioclasedepends on a nearly isothermal (- I 150 "C) conditions. Dashedline illustrates number of factors. The number of channelsper unit vol- that the change in plagioclase composition may vary more ume increasesrapidly with the magnitude of the pressure strongly with pressurewhen clinopyroxene is stable. interval, and the percentageofcrystal resorbed increases with time (Table 3). For decompressionintervals of 6 kbar, the density ofresorption channelsexceeds that which

Fig. 5. Backscatteredelectron images ofexperiment products produced by a 4-kbar pressuredrop of 12 h. Note the resorption in both plagi6sls5sand clinopyroxene. Symbols the same as in Table 2. l-arge scalebars are 100 pm. NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE 1247

is commonly found in orogenic lavas. The dissolution of nu.8" of .o.ditions wheredecomPression rate exceeds 10 m/s ffi clinopyroxenemay be important in the developmentof resorption textures in plagioclase,as its increasedsolu- bility and disappearanceat lower pressures(Fig. 3) cause rapid changesin the composition of the melt that help drive resorption. We envision that a range of disequilibrium textures could develop in plagioclasein responseto decompressionrates that vary from exceedinglyrapid to v the infinitely slow rate required to produce equilibrium melting. These textures might include reversezoning in addition to resorption. However, slow ascentrates would favor magmatic heat loss, which, in turn, would tend to counteract the effectsof decompression. 0 10 20 30 50 (m/s) DrscussroN Mean Flow Rate Our experimental procedure involves three assump- Fig. 6. Curvesof meanflow rate or ascentrate of magma tions in relation to the magmatic processesbeing mod- througha dikelikeconduit as a functionof conduitwidth and eled. First, decompressionrates and intervals are reason- magmaviscosity for a densitycontrast of 0.2g/cm3 between the able analoguesofnatural conditions.Second, heat lossis wall rock and magma(Kushiro, 1980, p. ll7). Approximate are shownfor comparison. insignificant compared with the decompressionrate, such valuesfor dry andesiteand basalt Note that ascentrates of >10 m/s (shadedregion) are easily that the processis nearly adiabatic. Third, over the pres- achievedfor a reasonablerange of geologicconditions in this is sure intervals considered, adiabatic decompression model. nearly isothermal-otherwise significantcooling will counteract the effects of decompression. On the other hand, raising the temperature of the system by magma pressuresless than 15 kbar, where plagioclaseis on the mixing will promote dissolution of plagioclase.However, liquidus and the liquidus is least sensitive to pressure, we will consider the caseof decompressionwithout mag- dT/dP is still about 5 "C/kbar (Green and Ringwood, ma mlxrng. I 968). In addition, Figure 2 showsthat changingpressure In regard to the third assumption, adiabatic cooling also changes the equilibrium plagioclase composition, must be less than the depressionof the plagioclaseliqui- providing a further driving force for resorption. There- dus causedby decompression.A liquid of composition fore, we employed isothermal experimentsas an approx- Ab3jAnrjDi33at 1425"C has an adiabaticgradient of - 1.5 imation of natural processes,as the effectof pressuremust 'Clkbar (calculatedfrom Rivers and Carmichael, 1987), be at least severaltimes that of the adiabatic gradient. cooling the liquid just 3-9 'C for the pressureintervals We produced strong resorption textures in plagioclase (2-6 kbar) in this study. In a crystal-bearingmagma, there for decompressionrates equivalent to rapid ascentrates is a second component to the total adiabatic gradient as (>10 m/s). Therefore,resorption must also occur for a latent heat is extracted from the liquid as the crystals range of decompression rates less than this. Previous resorb, resulting in further cooling ofthe system at equi- studies have determined magma ascentrates of up to 0.7 librium. Unlike the adiabatic gradient of a pure liquid, m/s for Mount St. Helens(Scandone and Malone, 1985) the extraction of latent heat is limited by the rate at which and 1.7 m/s beneathKilauea (Klein et al., 1987),whereas the crystals can resorb. Therefore, as long as resorption Spera(1980) calculated minimum ascentrates of 0.5 m/s is incomplete, a rapidly decompressedmagma is apt to basedupon the settling velocity of ultramafic nodules in be superheatedwith respectto its total adiabatic gradient. basaltic magma. These data indicate that ascentrates in Morse (1980) provided a convenientreview of perti- nature can be within at least an order of magnitude of nent experimental data regarding the relative effects of our experimentalrates. isothermal decompressionand adiabatic cooling. Albite Becausethe upper range of magmatic ascentrates may has a liquidus with a slope of > l0 "C/kbar, and the li- not be obvious to all, we constructed Figure 6 in order quidus of intermediate plagioclase(Anoo) is lowered by to evaluatepossible ascent rates in magmatic systemswith about 4 "C/kbar, about three times that of the liquid adi- establishedconduits. Figure 6 is intended to illustrate the abat. However, as shown above, the effect ofpressure is approximate scale(order of magnitude) of potential mag- enhancedin multicomponent systems,whereas adiabatic ma ascent rates: It is not a rigorous model. We assume gradients are not nearly as compositionally dependent. the flow of a Newtonian fluid through a dikelike conduit, Morse (1980)reviewed data showingthat the Di-An eu- where the viscosities and density contrast between mag- tectic is loweredon the order of l0 "C/kbar. Gill (198l) ma and wall rock used to derive the curves were treated noted that plagioclaseis the liquidus phasein almost all as constants,although they vary somewhatwith pressure andesites;thus, the whole rock liquidus representsa pla- (Kushiro, 1980).A model calculatedfor a tabular, rather gioclaseliquidus, albeit independent of composition. At than a cylindrical conduit, geometry provides a more 1248 NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE

positional profiles in the resorbed crystals are quite Or reverse, zon- Or complex, exhibiting normal, and oscillatory ing. The euhedral phenocrystsnever approach the size of the largeresorbed crystals (<0.5 mm), and usuallyexhib- t it normal zoning, although reverse and rare oscillatory zoning are also present.We summarize the compositions populations AN of crystal interiors of both in Figure 7 to help range of feldspar compositions interpret this sample. Despite a compositional overlap resorbedphenocrysts a between resorbed and unresorbed plagioclase,small eu- o o Q euhedralphenocrysts hedral plagioclasephenocrysts tend to range to significantly more calcic compositions. One analysis of re- o"k, sorbedplagioclase is distinctly An-rich and may represent Ab<-- \/ l/ -----+ An a Ca-rich sector in a zoned crystal. 30 40 50 60 70 Many of the compositional and textural aspectsof the phenocrystsmay be antecedentto those featuresimposed by decompression.Therefore, in light of our experiments, Fig. 7. Compositionsof coexistingeuhedral and resorbed phenocrystinteriors from the lava in Figurel. we outline a subsetof the characteristicsof the sample in Figure I that may result from rapid pressurerelease. We suggestthat the large resorbed plagioclasesamples rep- conservativeestimate ofascent rates becauseit takesinto resent partial crystallization of a magma at depth. Their accountthe largercomponent offrictional resistancebe- subsequentresorption was initiated by pressure release tween the magma and wall rock. accompanied by relatively little heat loss as the magma The detailsofthe calculationofthe curves(Fig. 6) are ascendedto a shallow magma reservoir. As a result of only modestly dependenton the model. Magma ascent this process,a coarsesieve texture was superimposedon ratesin conduits of varying geometry(e.g., Spera, 1980, the complex zoning patterns that had already been ac- p. 280-281; Kushiro, 1980,p. ll7) areproportional to quired by these crystals. Unfortunately, it is nearly im- the density contrast and inversely proportional to the vis- possible to quantify this process.The pressureat which cosity. However, ascentrates are also proportional to the the resorbedcrystals formed is difficult to estimate with- square of the conduit width or radius, depending upon out a suitablegeobarometer. Even if a suitablegeobarom- the geometry, causingthis term to dominate the model. eter existed, the problem would not be solved because As a result, the calculationsvary only by a factor of about uncertaintiesin calculatedpressures are often aslarge (-2- 2-3, given reasonablevariations in the density contrast 3 kbar) as the decompressioninterval required to pro- (0.2-0.3 g/cm') independentof conduit geometry.There- duce sieve textures. Likewise, the ascent rate is equally fore, it appears that magma ascent rates may be very difficult to estimate, as our experimentsdo not constrain rapid, approachingour decompressionrates over a range the relationship between resorption textures and the rate of geologically reasonable conditions. Magmas always of pressurerelease. However, if the equilibrium compo- contain a volatile component, and modest concentrations sition of plagioclaseimmediately prior to and after erup- of HrO (0.25 wto/o)and CO, (0.5 wto/o)can result in the tion could be determined, it might be possibleto estimate reduction of viscosity by an order of magnitude in silicate the minimum increment of pressure release from Fig- liquids (White and Montana, 1990),thereby enhancing ure 4. ascentrates. Although it is relatively easyto ascribe first-order features like coarsesieve textures to decompression,the ex- Observationsfor natural systems planation of other compositional and textural featuresis Our experimental results provide a basis for interpret- more tenuous. Perhapsthe nucleation ofeuhedral calcic ing magmaticprocesses in volcanic rockscontaining coarse microphenocrystcores was in responseto new conditions sieve-textured plagioclase.The phenocrysts in Figure I of equilibrium at a substantially reduced pressure,and are, on average,l0-20 times largerin sectionthan those normal zoning of these crystals may have been in re- in our experiments, which develop similar textures after sponseto subsequentcooling and evolution ofthe liquid, 4-12h. If resorption scaleslinearly with crystal size,these as reflected in their compositional variation (Fig. 7). On textures may develop within a few days for largevolcanic the other hand, the reverselyzoned euhedral crystalsare phenocrysts, following depressurization of at least 2-3 more problematic. They may reflect phenomenasuch as kbar. crystal growth in a continually ascendingmagma in which The lava in Figure I is strongly porphyritic, with pla- plagioclasecompositions are increasingly calcic (Fig. 2), gioclase phenocrysts occurring in two distinct popula- or crystallization under increasing HrO pressure (Jo- tions: (l) resorbedcrystals, and (2) euhedralphenocrysts. hannes,1978) caused by the formation ofan anhydrous The phenocrystsshow a fairly continuous size distribu- crystal assemblageafter decompressionor by a number tion up to a few millimeters, although the resorbedphe- of other phenomena,as reviewed by Gill (198l). It is nocrysts are always the largestcrystals in the rock. Com- difficult to interpret the presenceof both normal and re- NELSON AND MONTANA: SIEVE-TEXTURED PLAGIOCLASE 1249 verse zoning in the euhedral phenocrystsfrom the same neousrock suite. Contributions to Mineralogy and Petrology, I 8, 105- sample. Zoning in plagioclaseis a complex problem, un- 162. C.W., and Osborn, E.F. (1964) The solubility problematic Hamilton, D.L., Bumham, derscoringthe nature of inferring magmatic of water and effects of oxygen fugacity and water content on crystalli- processes from plagioclase phenocryst textures (Gill, zation in mafic magmas.Journal of Petrology, 5, 2l-39. l 98l). Johannes,W. (1978) Melting of plagioclasein the systemAb-An-H'O and In general,the nucleation of new calcic crystals,rather Q-Ab-An-HrO at P',o : 5 kbar, an equilibrium problem. Contribu- 295-303. than calcic overgrowths on the sodic plagioclase,is ne- tions to Mineralogy and Petrology,66, Klein, F.W., Koynagr, R.Y., Nakata, J.S., and Tanigawa, W'R. (1987) cessitated by the concomitant dissolution of the sodic The seismicity of Kilauea's magna system. U.S. Geological Survey crystals. Instancesmay occur in which resorption is ar- ProfessionalPaper, 1350 (2), l0l9-1185. rested as dissolution kinetics become sluggishwhen the Kushiro, I. (1980) Viscosity, density, and structure of silicate melts at system approaches its new equilibrium. The resorbed high pressures,and their petrologicalapplications. In R.B. Hargraves, processes,p. 93-12O. Princeton University plagio- Ed., Physics of magmatic crystals could be subsequentlyrimmed by calcic Press,Princeton, New Jersey. clase. Alternatively, degassingor cooling of the magma Kuo, L.C., and Kirkpatrick, R.J. (1982) Pre-eruptive history of phyric may also arrest resorption and allow the crystallization basalts from DSDP Legs 45 and 46: Evidence from morphology and of Ab-rich rims. A variety of processesmay operate in zoning patterns in plagioclase. Contributions to Mineralogy and Pe- combination at diferent times and placeswithin a mag- trology,79, 13-27. Meen, J.K. (1987) Formation of shoshonites from calcalkaline basalt ma body, giving rise to many of the complex textural magmas: Geochemical and experimental constraints from rhe type lo- featuresobserved in the plagioclaseof orogenic magmas. cality. Contributions to Mineralogy and Petrology,97 , 333-351. Morse, S.A. (1980) Basaltsand phasediagrams-An introduction to the AcxNowLnncMENTs quantitative useofphase diaS,ramsin igneouspetrology,356 p. Spring- er-Verlag,New York. We America for grant thank the Geological Society of a to S.T.N. and Nelson, S.T. (1989) Geologic map ofthe GeyserPeak quadrangle,Sevier the Division of Physical Sciencesat UCLA for financial support to A.M. and Wayne Counties, Utah Geologic report included). Utah Geological We Davidson for his careful reading of early thank our colleague Jon an and Mineral Survey Map Publication I14, scale l:24000. version of the manuscript, and Robert for assistancewith the elec- Jones Rivers, M.L., and Carmichael, I.S.E. (1987) Ultrasonic studiesof silicate tron microprobe. We also thank Kathy Cashman,Allen Glazner, Robert melts. Journal of GeophysicalResearch, 92, 9247-927 O. Hildebrandt, Myers, and Rosamond Kinzler for their constructive James Scandone,R., and Malone, S.D. (1985)Magma supply, magmadischarge, reviews.Institute ofGeophysics and PlanetaryPhysics contribution num- and readjustment of the feeding system of Mount St. Helens during ber 3438. 1980.Journal ofVolcanology and Geothermal Research,23,239-262. Spera, F.J. (1980) Aspects of magma transport. In R.B. Hargraves,Ed., RornneNces crrED Physics of magmatic processes,p. 261323. Princeton Univenity Press, Boettcher,A.L., Windom, K.8., Bohlen, S.R.,and Luth, R.W. (1981) Low Princeton, New Jersey. petrology Raton-Clayton vol- friction, anhydrous, low- to high-temperature furnace sample assembly Stormer, J.C. (1972) Mineralogy and ofthe of America for piston-cylinder apparatus. Reviews of Scientific Instruments, 52, canic field, northeastern New Mexico. Geological Society I 903-1904. Bulletin. 83. 3299-3322. plagioclasein the melt Cohen, L.H., Ito, K., and Kennedy, G.C. (1967) Melting and phasere- Tsuchiyama, A. (1985) Dissolution kinetics of and origin of dusty plagioclase in an- lations in an anhydrous basalt to 40 kilobars. American Journal of system diopside-albite-anorthite, Petrology,89, l-16. Science.265. 475-518. desites.Contributions to Mineralogy and plagioclase:Patchy zoning. Dungan, M.A., and Rhodes,M.J. (1978) Residualglasses and melt inclu- Vance, J.A. (1965) Zoning in igneous Journal -651. sions in basaltsfrom DSDP legs45 and 46:.Evidence for magma mix- of Geology,'l3, 637 (1990) HrO and CO, on the ing. Contributions to Mineralogy and Petrology,67,417-431. White, B.S., and Montana, A. The effect of liquid at high pressures. Journal of Geophysical Gill, J.B. (1981) Orogenic andesitesand plate tectonics,390 p. Springer- viscosity of sanidine Verlag, New York. Research,95, I 5683-15693. (1963) in occurring on liquidus and Green,D.H., and Ringwood, A.E. (1967) The genesisof basalticmagmas. Wyllie, P.J. Effecrsofchanges slope Mineralogical So- Contributions to Mineralogy and Petrology, I 5, 103- I 90. solidus paths in the system diopside-anorthite-albite. Green, T.H. (1972) Crystallization of calc-alkaline andesiteunder con- ciety ofAmerica SpecialPaper, l,204-212. trolled high-pressure hydrous conditions. Contributions to Mineralogy and Petrology,34, 150-166. MeNuscnrm RECETVEDDpcrrrrnen 30, l99l Green, T.H., and Ringwood, A.E. (1968) Genesisof the calc-alkalineig- M,c.Nuscnrpr AccEprED Jutv 13, 1992