American Mineralogist, Volume 81, pages 158-168, 1996

Morphology and maturation of melt inclusions in quartz phenocrysts from theBadlands rhyolite lava flow, southwestern Idaho

Cunrrs R. Mnvr.rv

Department of Geology, Box 871404, Arizona State University, Tempe, Arizona 85287-1404, U.S.A.

Arsrucr

Morphologies of primary rhyolitic melt inclusions in three successive populations ofquartz phenocrysts in a single volcanic unit illustrate the change of inclusion shape aftertrapping. The 14 Ma Badlands lava flow on the Owyhee Plateau of southwestern Idahocontains two magmas with three distinct quartz populations differing in size and mor-phology. In tiny quartz crystals that nucleated shortly before the eruption, large meltinclusions retain their initial, irregular shapes, whereas the smallest inclusions show maturenegative crystal shapes. Inclusions in a population of larger quartz crystals that nucleatedearlier all show either negative crystal shapes or faceted shapes transitional to negativecrystals. All inclusions in a third population of large corroded quartz crystals have maturenegative crystal shapes regardless of inclusion size. In silicic magmatic systems, irregularlyshaped inclusions in quartz may imply trapping just before eruption; analyses of suchinclusions should have the greatest likelihood of revealing the magmatic volatile condi-tions driving the eruption.

Electron microprobe analyses show that maturation had no effect on the compositionof the trapped melt. After eruption, however, slow cooling led to coarse crystallization andloss of silica from the melt; when these inclusions were revitrified in the laboratory, theydid not regain all lost silica and did not become completely homogeneous. Revitrificationof inclusions that cooled more quickly showed no such loss of silica to the host. Thus,given appropriate cooling conditions, even very old (Precambrian?) silicic melt inclusionsmay be suitable for microbeam analysis after any necessary revitrification in the labora-tory.

Irtlnoouctrox

Microbeam analytical techniques have dramatically in-creased the usefulness of igneous silicate melt inclusionsfor determining the original volatile contents, composi-tions, and temperatures of erupted magmas. To exploitfully the information recorded in melt inclusions, we needto understand the inclusions' probable morphology andcomposition at the time of their trapping and how thesemay have changed after entrapment. Such informationhas the potential to provide better constraints on mag-matic evolution and crystallization processes, triggeringof eruptions, and eruption dynamics.

Most of what we know about the processes of trappingof fluid and melt inclusions has been gained from micro-scopic study of natural inclusions and more directly bylaboratory observations of crystal growth and fluid-inclu-sion formation in relatively low-temperature aqueoussystems. Primary inclusions, which are trapped duringgrowth of the surrounding host crystal (see discussion inRoedder 1984), form because ofa variety ofconditions,including defects in crystal growth, such as kinking of anew growth layer over a planar crystal face (Sisson et al.1993), and localized temporary stagnation of growth

0003-004x/96l0 1 02-0 I 58$05.00

causing depressions or hollows that are covered by sub-sequent growth (Wilkins 1979). This is in contrast to sec-ondary inclusions, which are trapped, in either fractures(see Roedder 1984) or dissolved reentrants (Donaldsonand Henderson 1988), after the crystal has formed.

Primary fluid and melt inclusions are thought to betrapped only very rarely with negative crystal shapes(Roedder 1984). Sisson et al. (1993) observed irregularprimary fluid inclusions forming in laboratory experi-ments, but the temperatures and conditions involved intrapping of primary melt inclusions have so far precludeddirect observation of their formation. In the laboratory,synthetic, secondary inclusions of aqueous fluid, formedalong fractures in quartz, are commonly observed to neckdown into many smaller, isolated inclusions with moreregular shapes by local dissolution and reprecipitation ofqtartz (Shelton and Orville 1980; Roedder 1984). Like-wise, the regular shapes of most primary melt inclusionsare thought to result from posttrapping evolution or mat-uration of their morphology to minimize surface energy(Chaigneau et al. 1980; Beddoe-Stephens et al. 1983;Roedder 1984). This maturation often yields inclusionswith the spherical (lowest surface area per volume) ornegative crystal shapes (lowest surface energy) most often

1 5 8

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION 159

seen in igneous phenocrysts (Roedder 1984). Maturationoccurs only if temperature conditions remain suitable; forexample, Clocchiatti (197 5) reported that a rounded meltinclusion in quartz developed a negative crystal shapeafter it was held at 700-800 "C for six weeks. Experi-mental studies at 900'C and I GPa (Laporte and Provost1994) indicate that the equilibrium shape (no net growth,no net dissolution; see Wortis 1988) of a quartz crystalin a silicate liquid is a bipyramid with pristine, flat facesbut with smooth apices and interfacial edges. This is alsothe morphology of most melt inclusions with negativecrystal shapes in igneous quartz phenocrysts, i.e., smooth,lacking sharp angles and terminations.

It is generally assumed that maturation involves onlythe redistribution of material of the host crystal, with nonet loss or gain of chemical components from the inclu-sion fluid as the shape of the inclusion changes. This ismost easily shown for synthetic, secondary inclusions ofaqueous fluid, where the composition of the fluid is verydifferent from that ofthe host quarlz. In contrast, rhyo-litic melt inclusions have high SiO, contents in commonwith their quartz and feldspar hosts, and more mafic meltsshare more elements with their pyroxene and olivine hosts.

The initial shapes of igneous melt inclusions, and howthese then change with maturation, are nicely illustratedby rhyolitic melt inclusions in successively nucleatedpopulations of quartz phenocrysts in the Badlands lavaflow of southwestern Idaho.

Gnolocrclr, coNTExr oF sAMPLEs

The Badlands lava flow is a rhyolitic unit with a totalknown volume of about l5 km3 (Manley 1994, 1995).his associated with other large-volume rhyolite lavas andignimbrites erupted between 14 and 9 Ma (Bonnichsen1982a, 1982b: Ekren et al. I 984; Bonnichsen and Kauflman 1987) when the Yellowstone hotspot was locatedbeneath what is now the southwestern corner of Idaho,also known as the Owyhee Plateau.

The main events during the Badlands eruption, whichtapped two distinct types of magma, can be determinedfrom field relations in the unit's well-exposed vent area(Fig. l; Manley 1994, 1995). A dike at least 5-10 kmlong (in plan view) propagated to the surface, and the firstsurficial activity involved minor extrusion of aphyric lavaat the dike termination and sub-Plinian explosive erup-tions from the main eruptive fissures, which built up -50m thick ridges of nonwelded tephra representing mixedaphyric and phenocryst-rich magma. The narrow diketermination probably quickly became inactive because ofrapid cooling (e.g., Delaney and Pollard l98l). When ex-plosive activity ceased, lava effusion began. Near the diketermination, aphyric and phenocryst-rich magmas min-gled during ascent and emplacement; the majority of themultilobed l5 km3 Badlands lava is composed ofthe phe-nocryst-rich material.

The predominant Badlands magma, a phenocryst-rich(30 volo/o) rhyolite (-15 wto/o SiOr, Table 1), had an an-hydrous mineral assemblage of quartz * sanidine * pla-

Frcunp l. Perspective drawing of the exposed vent area ofthe Badlands lava flow, showing distribution of the aphyric andphenocryst-rich lithologies. Effusive aphyric lava is primarilyfound in the narrow dike termination and the portion ofthe flowfront closest to it. Cross-sectional views of the wedge-shapedtephra deposits are shown in black. The northeastern tephra de-posit was shoved aside by the weight of the spreading lava flow.Lines from boxes show locations ofsamples from which quartzphenocrysts were separated for this study. Also shown is theinferred preeruption geometry of layering in the Badlands mag-ma chamber, based on field and petrologic relations in the tephradeposits and lava flow. Two layers of aphyric magma overlie themain volume of phenocryst-rich magma. Eruptive tapping pre-sumably occurred through dikes that propagated upward fromthe top of the chamber ("footprint" of dike is shown on roof ofchamber; dike omitted for clarity).

gioclase + augite * orthopyroxene + magnetite + il-menite. The matrix melt was a silica-rich rhyolite (sample89-153-ts, Table l). Compositions of coexisting feldsparand pyroxene phenocrysts (Fuhrman and Lindsley 1988;Andersen et al. 1993) indicate a probable eruption tem-perature of about 830 + 30'C (Manley 1994).

The nearly aphyric magma was a crystal-poor (<0.5volo/o), silica-rich rhyolite (-77 wlo/o SiOr, Table l) witha mineral assemblage of small euhedral crystals of quartz* sanidine * plagioclase + biotite * hornblende + mag-netite + ilmenite (Manley 1994). Compositions of thecoexisting iron titanium oxide phenocrysts (see Andersenet al. 1993) indicate a probable eruption temperature ofabout 820 + 30'C (Manley 1994); compositions of co-existing feldspar phenocrysts do not converge to a mean-ingful temperature using the approach of Fuhrman andLindsley (1988).

The initial explosive phase of the eruption preferen-tially tapped the aphyric magma, but phenocrysts of allmagma types are represented in individual beds of thetephra. Although the later effusive phase was dominated

t = =

1 6 0 MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

Tlele 1. Compositions of bulk lava and matrix glass samples

Bulk samplesMatrix glasses

sio,Tio,Alro3FerO".MnOMgoCaONaroKrOPrO,

FTotal

HrO

75.10 .313

12.71 .950.031o.121.032.90c / Y

0.048

99.03

I O . I

0.1 16't2.31 . 1 10.0230.00o.623.595 3 10.0110.080 . 1 4

99.59n.a.

/ o,c0 .14

12.21 6 00.050 0 30.513.29J . Z C

0.030.080.32

98.381 .91"

Phenocryst-rich Aphyric

Sample 90-177R 90-188Material Vitrophyre ObsidianAnalysis

Phenocryst-rich Aphyric89-153-ts 90-188-G1

Matrix glass Obsidian17 27

76.40.08

12.31 .23

<0.05<0.01

0.553.715.26

<0.030.07n a o

100.650.101

Nofe; Analyses recalculated to 1 00% anhydrous; totals are those beforenormalization HrO analyses by secondary ion mass spectrometry (ionmicroprobe; see Hervig et al. 1989); all other oxides by wavelength-dis-persive X-ray fluorescence spectroscopy (bulk samples) or electron mi-croprobe (matrix glasses). All data in weight percent; n.a. : not analyzed.

. Total Fe as FerO..-. This is secondary water of meteoric origin in the hydrated (perlitic)

glass.t This is juvenile water of magmatic origin in the pristine glass (obsidian).

Frcunr 2. Transmittedlight photomicrographs of represen-tative grains from the three populations of quartz phenocrystsshown in the histogram in Figure 3. Note different scales. (a)Thin-section view of a large corroded quartz phenocryst in theBadlands lava-flow vitrophyre. The crystal shows curved frac-tures and embayments (lower left) filled with matrix glass. Inupper left is a large crystallized (black) melt inclusion with neg-ative crystal shape. Areas of speckled gray within the phenocrystwere plucked during thin-section preparation. Field of view is3.25 mm. (b) Bipyramidal intermediate euhedral quartz crystal,lying on its side. Note smooth apex, rough appearance offaces,and slight reentrant (out offocus) on face at top. Crystal was notacid cleaned. Field of view is I .625 mm. (c) Three bipyramidal,tiny euhedral quartz crystals. Note sharp interfacial edges andapices, and pristine faces; material on faces is adhering glassymatrix (crystals were not acid cleaned). Field of view is 0.65 mm.

by the phenocryst-rich magma, it began with effusion ofa small volume of the aphyric magma near the termina-tion ofthe dike. The field and petrographic relations in-dicate that a small body (a layer or a cupola) of the aphyr-ic magma must have overlain the phenocryst-rich magmain the chamber at the beginning of the eruption (Manley

1994, 1995). With the assumption that the chamber wastapped near its apex, mixing of the magma types duringthe initial explosive phase likely involved draw-up of theunderlying magma layers (i.e., Blake and Ivey 1986). Theapparent stratified chamber geometry and the aphyricmagma's low crystal content and major and trace elementcompositional similarity to the matrix liquid of the phe-

nocryst-rich magma (Manley 1994) imply that the aphyr-

ic magma evolved from the phenocryst-rich magma; thesmall euhedral phenocrysts in the aphyric magma nucle-ated and grew there; they were not inherited from thephenocryst-rich magma. The rise of an evolved, less-dense, HrO-rich melt along the magma-chamber wall(McBirney 1980; Turner and Gufstafson l98l), accu-mulating at the chamber top, is consistent with the fieldand petrologic constraints.

MBrrrons

Sample preparation

All quartz phenocrysts were collected from the Bad-lands rhyolite unit; they were separated from hydrated

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

TinyEuhedralOuartz

IntermediateEuhedralQuartz

Large Corroded Quartz

0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0 5 . 5

Crystal Long Dimension (mm)Frcunr 3. Histogram showing the sizes of quartz phenocrysts separated from various samples of the Badlands rhyolite unit.

The crystals fall into three discrete, nonoverlapping populations: tiny euhedral bipyramidal quartz (250-580 pm in length; n: 26crystals), intermediate euhedral bipyramidal quartz (750-1375 pm. n:24), and large corroded quartz (1750-5500 pm1' n:20nonfragmented crystals). The similar sizes and morphologies of the tiny and intermediate euhedral quartz demonstrate that theyare closely related.

1 6 1

. 0

1 098

L 7

96F c= 4Z g

2100

(perlitic) lava-flow vitrophyre, oxidized lava-flow cara-pace (surficial) pumice, and zeolitized ashy tephra by gen-tle crushing, followed by hand-picking. The crystals werecleaned by hand and in an ultrasonic water bath; no acidsor solvents were used. Separates of quartz were immersedin mineral oil and inspected for uncracked melt inclu-sions sufficiently large for microbeam analysis. Crystalswith uncracked devitrified melt inclusions were revitri-fied in the laboratory, generally at atmospheric pressureand -900 'C; crystals with glassy melt inclusions werenot heated. The crystals were then individually mountedin Araldite epoxy disks and polished with emery paperand diamond powder (both with distilled water) to ex-pose the inclusion glass.

Microbeam analyses

For all compositionally analyzed inclusions, ion micro-probe analysis was performed only after electron micro-probe analysis. The samples w€re coated with carbon pri-or to electron microprobe analysis; this coating was thenremoved by gentle polishing with <2 prm diamond pow-der and distilled water and replaced with a gold-palladi-um coating for ion microprobe analysis. The diameter ofthe primary ion beam was approximately 20 pm duringthe time of this study; other ion microprobe techniquesat Arizona State University are described in Hervig et al.(1989). Electron microprobe analyses were performed witha I 5 kV, l0 nA beam with a diameter of 20 pm to min-imize loss of Na.

Qulnrz popur,ATroNs AND MELT-INCLUSroNMORPHOLOGIES

Three distinct populations of quartz phenocrysts (Figs.2 and 3) were found in the tephras and lavas ofthe Bad-lands unit. The phenocryst-rich magma contained largecorroded and fractured quartz phenocrysts (up to manymillimeters in size), and different portions of the aphyric

magma grew two distinct populations of euhedral quartzphenocrysts of different sizes, intermediate (75G-l 37 5 pm)and tiny (250-580 pm)

Because all the host crystals are magmatic quartz, andall the trapped liquids are silica-rich rhyolites, chemicaland nucleation effects on melt-inclusion morphologiesacross the three host populations should be minimal. Thetwo populations of euhedral quartz crystallized from es-sentially the same melt at the same temperature and like-ly at very similar rates, implying that the melt inclusionsin both populations were probably trapped by the sameprocesses and originally had the same morphologies. Thegrowth history of the larger, corroded quartz crystals wasprobably similar, but there is no way to be certain of this.

Large corroded quartz crystals

The quartz phenocrysts in the phenocryst-rich portionofBadlands lava are spheroids ranging up to at least 5.5mm in length (Figs. 2a and 3); they are also found in thelithologically mixed tephra vented at the beginning of theeruption. These large quartz phenocrysts have been ex-tensively dissolved, and many show variously sized reen-trants or embayments (Donaldson and Henderson 1988)filled with melt (now glass). None of the embaymentsseem to have been resealed by later growth of the outercrystal surface. These crystals (as well as associated largefeldspars, often found in glomerocrysts) are highly frac-tured, but most fractures annealed or were never through-going, as the majority of crystals are intact. Dissolutionwas enhanced where these fractures intersect the crystalsurface, creating troughs that crisscross the surface; manyof the reentrants follow these fractures into the crystals.These observations imply that fracturing and dissolu-tion occurred significantly before the eruption and didnot result from the energetics of the eruption itself orfrom decompression (i.e., Nelson and Montana 1992)related to the rise of the masma toward the surface dur-rng eruptron.

t62 MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

All melt inclusions in the large corroded quartz phe-nocrysts have smooth, bipyramidal, negative crystalshapes, which often appear rhombus-shaped or nearlysquare in plan view under a microscope (see Fig. 4a andphotographs in Skirius et al. 1990). The maximum sizeof these inclusions is 190 x 180 pm. Many of the largestinclusions in these crystals are cut by fractures and haveleaked and devitrified as a result.

Intermediate euhedral quartz crystals

A second population of quartz phenocrysts, so far foundonly in the explosively erupted tephra dominated by theaphyric magma, is of euhedral bipyramidal crystals that

Frcunr 4. Melt inclusions in the intermediate and tiny eu-hedral quartz crystals. Transmitted-light photomicrographs ofinclusions in mounted and polished quartz crystals. All sampleswere coated with gold-palladium; the bright areas show wherethis coat was sputtered away by the primary beam of the ionmicroprobe. Field of view in each figure is 0.325 mm. (a) Largemelt inclusion of negative crystal shape, exposed at surface of asample of intermediate euhedral quartz. This inclusion wasquenched to glass during the eruption, and thus revitrificationwas not performed in the laboratory; the glass is brown, and thelighter area within the inclusion is the ion-beam scar throughthe gold-palladium coating. Smaller, well-faceted inclusions withnegative crystal shapes are just below plane of focus in upperright. The fractures were observed forming during sample pol-

ishing. Sample 225-Q3. (b) Two large irregular melt inclusionsexposed at surface of a sample of tiny euhedral quartz. Someslight faceting of the inclusion outline is apparent. The shape ofthe inclusion that is out offocus in upper left is shown in Figure5. Formation of the fractures was observed during sample pol-

ishing. Opposite crystal faces can be seen in the upper right andthe lower left. Melt inclusions were originally finely crystallineand were revitrified at 900 "C at atmospheric pressrue' Sample314-Q2. (c) One large "peanut shell"-shaped irregular melt in-

clusion is exposed (bright, ion-sputtered area in lower center) atsurface of a sample of tiny euhedral quartz (larger bright area isa trail of ion-probe scars on the quartz host). Smaller inclusionswith negative crystal shapes can be seen above and to the rightof the irregular inclusion. On the right is a partly annealed frac-ture associated with two ovoid, coarsely crystalline inclusionsthat did not successfully revitrify (far right, with large vaporbubble; top center, dark, out of focus). Two adjoining crystalfaces can be seen on left. Melt inclusion was originally finelycrystalline and was revitrified at 900 "C at atmospheric pressure.

See drawings in Figure 5. Sample 314-Q4.

(-

were partially dissolved or equilibrated (Fig. 2b); theyrange from 750 to 1375 pm in length (Fig. 3). Primarycrystal-growth faces are apparent on all these crystals, butthey have been slightly dissolved and are no longer pris-tine; apices and interfacial edges have been smoothed byeither dissolution (due to changing external conditions)or change in shape of the crystal as it began to minimizeits surface energy (Wortis 1988; Laporte and Provost1994). A few crystals show hollows at the centers of py-ramidal faces; though the locations ofthese features sug-gest they may have initiated by starvation of the centerof the crystal face during growth (Clocchiatti and Basset1978; Wilkins 1979), they appear to have been deepenedand roughened by the dissolution that affected the rest ofthe crystal's surfaces.

One-half dozen melt inclusions with shapes presum-ably transitional between irregular and negative crystalshapes (elongate, but with faint facets; Fig. 5) were notedin the intermediate euhedral qtrarlz phenocrysts; theserange from 50 x l0 pm to 190 x 80 pm in size (Fig. 6b).All other inclusions, ranging from 190 x 165 pm to 2.5x 2.5 pm (Fig. 6b), show the same smooth, bipyramidal,negative crystal shapes (Fig. 4a) seen in the large corrodedquartz crystals.

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION 163

FlcunB 5. Shapes of rhyolitic melt inclusions from the Bad-lands unit. Line drawings traced from photomicrographs oftheplan view (under a microscope) shapes of rhyolitic melt inclu-sions from two ofthe Badlands quartz phenocryst populations.Groups of inclusions from a common quartz crystal are indicat-ed by arrows; arrangement ofinclusions within a group reflectstheir relative positions in the crystal. Open arrows indicate in-clusions with negative crystal shape. Compositions of inclusions3 14-Q4- I and 3 l4-Q5- 1 are listed in Table 2. Transitional (elon-gate but partly faceted) inclusion was heated at 900 'C and at-mospheric pressure, but revitrification was incomplete, leavingseveral large vapor bubbles. All inclusions are shown at the samescale. Vapor bubbles are represented as black circles; othermarkings in these inclusions reflect incompletely revitrifieddaughter phases on walls.

Tiny euhedral quartz crystals

The third quartz population, found in the tephra andin surf,cial pumice and obsidian from the aphyric portionof the lava flow, is of small, fresh, euhedral bipyramidalcrystals (Fig.2c), which generally range from 250 to 580pm in length (Fig. 3). They show no evidence of disso-lution of the outer surfaces: faces, interfacial edges, andapices are all regular, pristine, and sharp, and the crystalsreflect light like faceted diamonds, which they resemble.As seen in the intermediate euhedral qvartz, some of thesetiny euhedral crystals show starvation hollows at the cen-ters of pyramidal faces, but in the fresh, tiny quartz crys-tals these hollows remain shallow and appear unmodifiedby any dissolution (see Fig. III-C, D of Clocchiatti andMervoyer 1976).

Melt inclusions in the tiny euhedral quartz have di-verse forms. The largest inclusions (from 30 to roughly120 trm in length) are highly irregular, rangrng from ovoidsthrough elongate "peanut shell" and "boomerang" shapes(Figs. 4b and 4c). The smallest melt inclusions have neg-ative crystal shapes (Figs. 4b and 5); a similar relation isseen in synthetically produced, secondary aqueous fluidinclusions in quartz, where in a given sample the smallestinclusions tend to be the most regularly shaped (Sternerand Bodnar 1984). Figure 6c shows that with decreasinginclusion size there is a distinct change in morphologyfrom irregular to negative crystal shape around 30 pm inlength. There are a few exceptions to this: six smallinclusions have spherical or irregular shapes, some ofwhich also show slight faceting, and three inclusions seemtransitional, partly faceted but with elongate shapes.

Discussion

The three populations of quartz phenocrysts found inproducts of the rhyolitic Badlands lava-flow eruption re-veal details of how crystal shapes and melt-inclusionmorphologies can change with time. These relations alsoprovide information on the state of the magma chamber,and the processes within it, before it was tapped by theeruption.

The large corroded quartz crystals remained at mag-matic temperature sufficiently long that all their melt in-

Irregular Melt Inclusions in Tiny Euhedral Quartz

314-Q4-1

)r3A_a+.\l\-J

314-Q5 />1 \-/

t / 3r4-Q3

314-Qs-1 |

Transitional Melt Inclusionin Intermediate Euhedral Quartz

0 l 0 0 p m

ff

clusions matured to negative crystal shapes. The size,abundance, and resorption features ofthe quartz and oth-er phenocrysts indicate that the phenocryst-rich magmahad an extended history, involving one or more changesin temperature, pressure, or composition conditions longbefore eruption. If resorption of the large corrode d qluartzcrystals was caused by a change in external melt com-

<,.,:"'.- o

r64 MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

150Ei

;oU'c 1 n n( D ' - -

EoE

(f) 50

L a r g eCor rodedQuar tz

#

tr Negative crystal shapes

Frcunr 6. Relations between melt-inclusion shape and in-clusion size. Plots of melt-inclusion size (long vs. short dimen-sion) showing the variation of inclusion shape with size. Diag-onal lines indicate long to short dimension ratios of l:1. (a) Meltinclusions in the large corroded quartz population. All inclusionshave negative crystal shapes. (b) Melt inclusions in the inter-mediate euhedral quartz population. Nearly all inclusions havenegative crystal shapes, and none are irregular. The transitionalinclusions are more strongly faceted than the transitional shapesin the tiny euhedral quartz crystals but are highly nonequidi-mensional. (c) Melt inclusions in the tiny euhedral quartz pop-

ulation, apparently growing just before the eruption. The largestinclusions are irregular in shape or transitional (partly facetedbut without the equidimensional negative crystal shape). Truebipyramidal, negative crystal shape is restricted to inclusions<30 pm in length; a few smaller inclusions of the other shapesare also <30 pm in size.

(-

position, the melt inclusions should not have been af-fected. If, on the other hand, resorption was caused by arise in temperature or drop in pressure (Nelson and Mon-tana 1992), the inclusions may have become irregular atthat time as well. Even if exterior conditions or lack ofsufficient time did not allow the crystals to return to anequilibrium morphology, their inclusions, smaller andisolated from exterior conditions, were nonetheless ableto mature back to negative crystal shapes.

The high degree of fracturing of the phenocrysts alsoraises the possibility that the magma had once nearly orcompletely solidified (during which time the phenocrystswere deformed) and was later remelted (e.9., Bacon et al.1989). Differences in trace element contents (Manley1994) between the melts (now matrix glasses) of the twolavas indicate that the phenocryst-rich magma was notproduced by syneruptive mixing of disrupted cumulatecrystals from the magma-chamber wall into the aphyricmagma, as has been noted by Nakada et al. (1994) for asuite of rhyodacite lavas elsewhere.

In the tiniest euhedral quartz crystals, which must bethe youngest of the quartz populations, the larger meltinclusions apparently did not have enough time to ma-ture before eruption quenching froze-in their irregular,presumably initial shapes. Only the smallest inclusionsin these tiny crystals had sufficient time at high temper-ature to mature to smooth, bipyramidal, negative crystalshapes.

Though the intermediate-sized euhedral quartz crystalsare clearly related by size and morphology to the tinyeuhedral crystals, even their very large melt inclusionsnonetheless had sufficient time at magmatic conditionsto mature.

The field and petrologic observations indicate that theBadlands magma chamber was stratified, with two layers(at least) of aphyric magma underlain by the phenocryst-rich magma (Fig. l). The fresh morphology of the tinyeuhedral quartz crystals show they were growing, in theaphyric magma closest to the chamber roof, shortly be-

^ 150E1

c

oc i n no ' - -

EoE

/ ! / ) 5 0

50 100 ' l 50

Long Dimension (pm)

100

Long Dimension (pm)

1 0 0

200

2001 5 0

150EI

oo5 100EoEo@ 5 0

1 5 0

ln te rmed ia teE u h e d r a IQuartz

trt r t r

A

trt r 6

^ Ao a

A Transitional shapestr Negative crystal shapes

TinyE u hedralQuartz 1 5

1 0

oo4e I

o lrregularshapes@- 60 | a Transitional shapes

tr Negative crystal shapes

Long Dimension (pm)

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION 165

fore being erupted; their presence in the tephra that erupt-ed earliest, which vented rapidly, shows that they did notnucleate in response to initiation oferuption. The inter-mediate euhedral quartz crystals ceased growth before theeruption and were probably in equilibrium in the melt;these crystals resided in nearly aphyric magma presum-ably below that containing the tiny euhedral crystals.

Though it is likely that all the quartz crystals attainedtheir present forms at depth in the chamber some timebefore the initiation of the Badlands eruption, there areno constraints on the absolute length of time any of thecrystal populations resided in the magma. The size andfreshness of the tiny euhedral quartz crystals and the ir-regular shape of their melt inclusions, however, suggestthat their residence time may have been as brief as sev-eral weeks to several months. If this is the case, the erup-tion intemrpted active differentiation in the Badlandsmagma chamber.

The distinct size ranges of the two populations of eu-hedral quartz and the apparent restriction of the popu-lations to different layers in the preeruptive magmachamber also hint that the process that produced theevolved aphyric magma was episodic rather than contin-uous.

Mrlr-rr.rclusroN coMposrrroNs

Large corroded quartz crystals

Compositions of two groups of melt inclusions fromthe large corroded quartz phenocrysts were analyzed: nat-ural glassy inclusions and coarsely crystallized inclusionsrevitrified in the laboratory. Whole and fragmented quartzphenocrysts from the Badlands tephra contain melt in-clusions that were quenched to glass during the explosiveeruption; no revitrification was needed to analyze theseinclusions. Inclusion glasses have silica contents rangingfrom 76.2 to 77 .l wto/o (normalized anhydrous, Table 2),which is very similar to the matrix melt of the pheno-cryst-rich lava (Table l), presumed to be little differentfrom the melt from which the inclusions were originallytrapped. The negative crystal shapes imply that silica wasredistributed within and around the inclusions, but thereis no compositional or textural evidence for growth ofquartz or other daughter minerals at the expense of in-clusion composition or volume.

Quartz phenocrysts from a sample of the basal vitro-phyre of the Badlands lava flow contain inclusions thatcrystallized coarsely during slow cooling of the lava. Theseinclusions were revitrified at temperatures from 900 to920 "C, at either atmospheric pressure or pressures of 230or 530 MPa (2.3 or 5.3 kbar), in an internally heated,argon gas pressure vessel (see Holloway 197 l), for 8-20h. They show variable glass compositions that range from67 .5 to 7 6.5 wto/o SiO, (normalized anhydrous), with oth-er elements affected accordingly (Table 2).

Intermediate euhedral quartz crystals

Inclusions in the intermediate euhedral crystals, foundonly in the tephra, were quenched during their explosive

eruption and thus remained glassy, though a discrete irontitanium oxide daughter-mineral grain grew in one inclu-sion (225-Q6-l), and several others developed incipientcrystallite groWh. A few of these inclusions were there-fore revitrified at 900'C and atmospheric pressure for 23h. No significant compositional differences are apparentbetween inclusions that were revitrified and those thatwere not. The inclusion glasses have normalized silicacontents of 75.9-78.0 wto/o (Table 2) and uniform con-centrations of other elements; they are essentially iden-tical to the melt from which the host crystals presumablygrew, now represented by the aphyric lava (90-188, Ta-ble l).

Growth of the iron titanium oxide daughter mineral ininclusion 225-Q6-l predictably lowered the Fe contentof the melt (Table 2). Loss of silica by crystallization alongthe wall of the host apparently occurred along with growthof the oxide grain because melt silica is low and otherelements show increases.

One of the inclusions of transitional shape was alsoanalyzed (Manley unpublished data); its composition dif-fers in no significant way from the other inclusion (ofnegative crystal shape) in the same host, nor from inclu-sions in the other hosts (Table 2).

Tiny euhedral quartz crystals

With one exception, the irregularly shaped inclusionsin the tiny euhedral quartz population show no evidenceof crystallization of discrete daughter-mineral grains. Alleight analyzed inclusions are from samples of lava-flowcarapace pumice that cooled slowly enough for the inclu-sion glass to become finely crystalline. Thus, the inclu-sions were revitrified at 900 .C for 22 h at atmosphericpressure. Inspection ofthe inclusions before and after re-vitrification showed that their shapes were not changedby the heating. Analyzed inclusions (33-120 pm in length)show normalized silica contents (Table 2) from roughly77 to 78 wto/o; this is identical to the melt from which thehost crystals grew (90-188, Table l).

One irregularly shaped melt inclusion,62 x 52 pm insize (inclusion 314-Q5-l), shows the efects of crystalli-zation of quartz along the wall of the inclusion. Afterrevitrification, the glass of this inclusion appeared ho-mogeneous and clear, though crystallized material wasvisible along the inclusion wall, and the vapor bubbleremained larger than average. Silica contents are low andvariable: two electron microprobe analyses of differentportions of the inclusion indicate 68.7 and 70.3 wto/o SiOr,with other oxides correspondingly higher (Table 2).

None of the inclusions with negative crystal shape inthe tiny euhedral crystals were sufficiently large to analyzeon the ion microprobe at the time of the study; thus, theywere also not analyzed by electron microprobe.

Discussion

The textural and compositional character of the Bad-lands melt inclusions clearly indicates that the trappedmelt of the inclusions is not measurablv modified as host

t 6 6 MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

TABLE 2. Compositions of silicate melt inclusions from the Badlands rhyolite

Inclusions of irregular shape

Tiny euhedral quartz

Finely crystallineRevitrified at 900'C

lnclusions of negative crystal shape

Intermediate euhedral quartz

GlassyNot revitrified

Brown/clear glass with oxide

Group size

SampleInclusion

Analysis

Clear glass

314o1 -1

1 5

31404-1

1 q

22503-1

1 7

227Q5-1

22

n : 1

22506-1

1 5

Growth on wall

n : 1

3 1 405-1

n : 2

sio,Tio,AlrosFerO."MnOMgoCaONaroKrOPrOuclF

TotalH.o

76.9o .12

1 1 . 01 .35

<0.05<0.01

0.495.014.65

<0.030 . 1 20.25

98.902.38

78.00 .12

10.81 . 1 5

<0.050.020.504.504.52

<0.030.100.35

97.511.56

68.70.08

16.40.98

<0.05<0.01

0.806.435.87

<0.030.130.60

100.75n.a.

70.30 .13

15.21 .28

<0.05<0.01

0.675.905.820.050.090.50

100.60n.a.

76.20 .15

12.41 .80

<0.050.020.573.285.27

<0.03n.a.0.28

99.522.09

76.30 .15

12.31 .58

<0.050.050.563.375.30

<0.03n.a.0.38

97.532.66

73.80.08

14.50.45

<0.05<0.01

0.273.966.060.07

n.a.0.82

1 02.1 31.08

/Vote.' Analyses recalculated to | 00% anhydrous; totals are those before normalization. HrO analyses by secondary ion mass spectrometry (SIMSor ion microprobe; see Hervig et al. 1989); all other oxides by electron microprobe. All data in weight percent; n.a. : not analyzed.

- Includes some compositionally identical glassy melt inclusions in large sanidine hosts..'Total Fe as FerO3.

quartz dissolves and reprecipitates during the change fromirregular to negative crystal shapes. In the Badlands in-clusions, growth of other daughter minerals after trappingdepletes the melt of those elements incorporated by thegrowing crystals and may also trigger loss of silica to theinclusion wall. Slow cooling from magmatic temperaturesof the inclusions in the large corrodedquarlz crystals fromthe interior of the lava flow caused coarse-grained crys-tallization of the melt and significant loss of silica to thehost quartz (see also Webster and Duffield 1991). Thislast process was responsible for the most serious changesto inclusion composition, as the silica that was lost to thequartz host is only very slowly remelted during revitrifi-cation of the inclusions in the laboratory. Even in revit-rified inclusions that appear clear and glassy, significantcompositional gradients can remain even after heating at920 "C for up to 20 h (cf. Skirius et al. 1990). Because ofsuch inhomogeneity, determining the original concentra-tions of other elements by correcting for silica loss (Web-ster and Duffield 1991) will not be completely successfulfor these inclusions.

CoNcr,usroNs

Primary inclusions of rhyolitic melt in three distinct,apparently successive populations of quarlz phenocrystsfrom the Badlands lava flow in southwestern Idaho illus-trate a change from irregular to mature, negative crystal

shapes. This process minimizes the surface energy of theinclusion by the solution and reprecipitation of quartzfrom the inclusion wall. In the quartz population growingjust before the eruption, only the smallest melt inclusions,which would mature most quickly, show negative crystalshapes. In these, the larger melt inclusions are irregular,with shapes ranging from ovoid to "boomerang," prob-ably the shapes these inclusions had at the time of trap-ping. Inclusions in a population of larger quartz crystals,which had nucleated earlier but ceased growth, all showeither negative crystal shapes or faceted but elongateshapes presumably transitional to negative crystal shapes.All inclusions in a third population of even larger, verycorroded quartz crystals, which spent significantly moretime at magmatic temperatures, have mature negativecrystal shapes regardless of inclusion size.

These considerations should apply to other silicic mag-matic systems as well. Inclusions with irregular or non-matured shapes may reveal phenocrysts that were form-ing just before their eruption or in which, in some cases,nucleation was triggered by, or along with, the eruptionitself. In ideal cases, melt-inclusion and host-crystal mor-phology relations might help reveal the course or thecauses ofthe eruption where other evidence has not beenpreserved or is inconclusive. For volcanic systems of moremafic compositions, gowth of phenocrysts and forma-tion of melt inclusions often continue during the eruptionand afterward, during emplacement of lava flows. For-

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION r67

f ABLE 2.-Continued

Inclusions ol negative crystal shape

Intermediate euhedral quartz Large corroded quartz

classyRevitrified at 900 €

Clear glass

n : 6

classyNot revitrified

Clear glass

n: 12'

Coarsely crystallineRevitrified at 900 to 920'C

Clear glass with remnant daughter crystals

n : ' 12

225Q 1 1 - 1

o

225Q12-2

1 2

225Q2-1

8

225Q1-3

164o3€!-1

I

164Q1-1

23

n : 1

164o40-1

75.90 .10

12.21 .45

<0.05<0.01

0.604.225.O2

<0.030 . 1 10.31

99.063.60

78.00.08

1 1 . 11.33

<0.05<0.01

0.513.924.70

<0.030.120.28

97.582.20

76.20 .13

12.71.54

<0.050.020.533.605.21

<0.030.09n.a.

96.713.06

77.10 .13

12.21 .41

<0.050.020.473.285.29

<0.030.07n.a.

96.852.66

b / . 5

0.2016.12.640.050.051.005.896.24

<0.03n.a.0.34

101.250.73

76.50 .10

10.82.74

<0.050.050.573.855.06

<0.03n.a.0.33

99.672.32

70.20 .15

15.31.63

<0.05<0.01

0.815.246.20

<0.03n.a.0.41

99.3r34.55

75.30 .15

12.01.98

<0.050.030.533.975.77

<0.03n.a.0.29

99.612.47

tuitous conditions and careful sampling may permit pre-and posteruptively trapped inclusions to be studied toreveal compositional changes accompanying eruption. Forsuch mafic magmas, in which inclusions would be trappedin phases other than qvarlz, relationships similar to thosedocumented here for rhyolitic quartz-hosted inclusionswould need to be confirmed.

Electron microprobe analyses of the various popula-tions of inclusions also confirm that maturation of inclu-sions has no effect on the composition of the trappedmelt, at least in these rhyolitic inclusions in quartz phe-nocrysts. After eruption, however, those inclusions thatcooled most slowly lost silica to quartz growth on theinclusion walls and to coarse crystallization in the bodyof the inclusion. Although such crystalline inclusions wererevitrified to clear glass in the laboratory, silica contentsremained low and variable, and other elements high, in-dicating that the inclusions did not regain all lost silicaand did not become homogeneous. Other inclusions,which cooled more quickly but nonetheless became finelycrystalline and needed to be revitrified to clear glass,showed no such loss of silica to the quartz host.

In the study of silicic melt inclusions from older erup-tions, the age of the eruption is much less important thanthe cooling rates of the inclusions immediately after. Ifthe inclusion cooled quickly once it reached the surface,and if it was not significantly reheated later, very old(perhaps even Precambrian) inclusions may be candi-dates for microbeam analysis.

AcxNowr,nocMENTs

Discussions about melt inclusions with Richard Hervig, Jacob Iow-enstern, and Kurt Roggensack were very helpful. I am grateful to JohnHolloway for allowing me to use laboratory equipment and sample prep-aration facilities, and to Ken Domanik for invaluable assistance rrith theinternally heated, gas pressure vessel during the high-pressure revitrifi-cation of certain of the melt inclusions. Richard Hervig provided advice,instruction, and assistance with the ion microprobe, and Jim Clark as-sisted with electron microprobe analyses Reviews by Eric Christiansen,Richard Hervig, and Virginia Sisson greatly improved the manuscript.Various portions ofthis research were supported by Geological Society ofAmerica research granl 4266-89, a Stanford School of Earth SciencesMcGee Fund grant, and NSF grants EAR-9018216 to R. Hervig andEAR-9105329 to Jonathan Fink.

RrrnnrNcrs crrEDAndersen, D.J., Lindsley, D.H., and Davidson, P.M. (I993) QUILF: A

PASCAL program to assess equilibria among Fe-Mg-Mn-Ti oxides,pyroxenes, olivine, and quartz. Computers and Geosciences, I 9, I 333-1 350.

Bacon, C.R., Adami, L.H., and Lanphere, M.A. (1989) Direct evidencefor the origin of low-'8O silicic magmas: Quenched samples of a magmachamber's partially-fused granitoid walls, Crater Lake, Oregon. Earthand Planetary Science Letters, 96,199-208.

Beddoe-Stephens,8., Aspden, J.A., and Shepherd, T.J. (1983) Glass in-clusions and melt compositions of the Toba Tuffs, northern Sumatra.Contributions to Mineralogy and Petrology, 83,278-287 .

Blake, S., and Ivey, G.N. (1986) Magma mixing and the dynamics ofwithdrawal from stratified reservoirs. Journal ofVolcanology and Geo-thermal Research, 22, l-31

Bonnichsen, B. (1982a) The Bruneau-Jarbidge Eruptive Center, south-western Idaho. In Idaho Bureau of Mines and Geology Bulletin, 26,237-254.

- (1982b) Rhyolite lava flows in the Bruneau-Jarbidge Eruptive Cen-

1 6 8

ter, southwestern Idaho. In Idaho Bureau of Mines and Geology Bul-letin. 26. 283-320.

Bonnichsen, B., and Kauffrnan, D.F. (1987) Physical features ofrhyolitelava flows in the Snake River Plain volcanic province, southwesternIdaho. In Geological Society of America Special Paper, 212, I I 9- 145.

Chaigneau, M, Massare, D., and Clocchiatti, R. (1980) Contribution el'6tude des inclusions vitreuses et des 6l6ments volatils contenus dansles ph6nocristaux de quartz de roches volcaniques acides. Bulletin Vol-canologique, 43, 233-240.

Clocchiatti, R. (1975) Les inclusions vitreuses des cristaux de quartz:Etude optique, thermo-optique et chimique. Applications g6ologiques.Societe G6ologique de France Memoires, LIV, Memoire 122,96 p.

Clocchiatti, R., and Mervoyer, B. (1976) Contribution d l'6tude des cris-taux de quartz de la Guadeloupe. Bulletin du Bureau de RecherchesGeologiques et Minieres, France, 2's6rie, sec. IV,4,311-324.

Clocchiatti, R., and Basset, A.M. (1978) Skeletal growth and melt inclu-sions in quartz crystals of rhyodacitic ignimbrites of the Valley of TenThousand Smokes, Katmai, Alaska. Eos, 59, 225.

Delaney, P T., and Pollard, D.D. (1981) Deformation of host rocks andflow of magma during growth of minette dikes and breccia-bearingintrusions near Ship Rock, New Mexico. U.S. Geological Survey Pro-fessional Paper, 1202,61 p.

Donaldson, C.H., and Henderson, C.M.B. (1988) A new interpretation ofround embayments in quartz crystals. MineralogicalMagazine, 52, 27 -

J J

Ekren, E.B., Mclntyre, D.H., and Bennett, E.H. (1984) High-temperature,large-volume, lavalike ash-flow tuffs without calderas in southwestemIdaho. U.S. Geological Survey Professional Paper, 1272,76 p.

Fuhrman, M.L., and Lindsley, D.H. (1988) Ternary-feldspar modelingand therrnometry. American Mineralogist, 7 3, 201-215.

Hervig, R.L., Dunbar, N., Westrich, H.R., and Kyle, P.R. (1989) Pre-eruptive water content of rhyolite magnas as determined by ion mi-croprobe analyses of melt inclusions in phenocrysts. Journal of Vol-canology and Geothermal Research, 36, 293-302.

Holloway, J R (1971) Internally heated pressure vessels. In G.C. Ulmer,Ed., Research techniques for high pressure and high temperature, p.217 -258. Springer-Verlag, New York.

Laporte, D., and Provost, A. (1994) The equilibrium crystal shape ofsilicates: Implications for the grain-scale distribution of partial melts.Eos. 75. 364

Manley, CR (1994) On voluminous rhyolite lava flows, 314 p. Ph.D.thesis, Arizona State University, Tempe.

MANLEY: MELT-INCLUSION MORPHOLOGY AND MATURATION

- (1995) How voluminous rhyolite lavas mimic rheornorphic ignim-

brites: Eruptive style, emplacement conditions, and formation of tuff-

like textures. Geology, 23, 349-352McBirney, A.T. (1980) Mixing and unmixing of magmas. Journal of Vol-

canology and Geothermal Research, 7, 357-37 l.Nakada, S., Bacon, C.R., and Gartner, A E (1994) Origin of phenocrysts

and compositional diversity in pre-Mazama rhyodacite lavas, CraterLake, Oregon. Journal of Petrology, 35, 127 -162.

Nelson, S T., and Montana, A. (1992) Sieve-textured plagioclase in vol-

canic rocks produced by rapid decompression American Mineralogist,77, 1242-t249

Roedder, E. (1984) Fluid inclusions. In Mineralogical Society ofAmericaReviews in Mineralogy, 12,644 p.

Shelton, K.L , and Orville, P.M. (1980) Formation of synthetic fluid in-

clusions in natural quartz American Mineralogist, 65,1233-1236.Sisson, V.8., Lovelace, R.W, Maze, W.B., and Bergman, S.C. (1993)

Direct observation ofprimary fluid-inclusion formation. Geology, 21,

7 5 t - 7 54 .Skirius, C.M., Peterson, J.W., and Anderson, A.T., Jr. (1990) Homoge-

nizing rhyolitic glass inclusions from the Bishop Tuff. American Min-eralogist , 75, t38l-1398.

Sterner, S.M., and Bodnar, R.J (1984) Synthetic fluid inclusions in nat-ural quartz: I. Compositional types synthesized and applications to

experimental geochemistry. Geochimica et Cosmochimica Acta, 48,

2659-2668.Turner, J.S., and Gufstafson, L.B. (1981) Fluid motions and composr-

tional gradients produced by crystallization or melting at Yertical

boundaries. Journal ofVolcanology and Geothermal Research, I 1, 93-r25.

Webster, J.D., and Duffield, W.A. (1991) Volatiles and lithophile ele-ments in Taylor Creek Rhyolite: Constraints from glass inclusion anal-ysis. American Mineralogist, 76, 1628-1645

Wilkins, R.W.T. (1979) Formation, modification and destruction of fluidinclusions: Notes for fluid inclusion workshop. Department ofGeology,Ia Trobe University, Ia Trobe, Bundoora, Victoria, Australia, Feb-ruary 1979 (mimeographed, 8 p.) (as referenced by Roedder 1984).

Wortis, M. (1988) Equilibrium crystal shapes and interfacial phase tran-sitions. In R. Vanselow and R.F. Howe, Eds., Chemistry and physics

of solid surfaces YlI, p. 367-405. Springer-Verlag, New York.

MnNuscnrrr REcEIVED Ocrosen 3, 1994Mer.ruscrrrr ACCEPTED Seprei"rsen l, 1995