The Stillwater Complex, Montana: a Subvolcanic Magma Chamber?

American Mineralogist, Volume 80, pages 1343-1346, 1995

The Stillwater Complex, Montana: A subvolcanicmagma chamber?

Rosar-rNo Turnrr.r, Hnnz U.S. GeologicalSurvey, M.S. 107,Reston, Virginia 22092,U.5.4.

Ansrucr Five magma types occur as sills in the footwall of the Stillwater Complex, all of them coeval with the Stillwater in age.Two of the magma types have compositions that suggest they are similar to the magmas from which the cumulates of the Stillwater's Basal and Ultramafic series crystallized. Melting experiments performed on samples of these two magma types and on a 50-50 mix of the two types show that the crystallization sequence inferred from the Ultramafic seriescumulates (olivine > orthopyroxene > plagioclase> augite) is matched only in the mixture of the two sill types at 1.5-3 kbar. Thus, the observed stratigraphy of the Basal and Ultramafic seriesmay result from mixing of two distinct magmas rather than fractionation of a single magma. The permissible pressure range of 1.5-3 kbar implies that the depth of the Stillwater magma chamber was similar to that of subvolcanic magma chamberssuch as that of Kilauea Volcano, which raisesthe possibility that the Stillwater magma body may also have been a subvolcanic reservoir.

INrnorucrroN Jackson,1961, 1970). With this assumptionin mind, and analyzedchill sam- One obstacle to deciphering the magmatic history of on the basisof a limited number of 40o/oof the Complex has layeredmafic intrusions is that it is not easyto identify ples, Hess (1960) inferred that recently,it has beenrec- the compositions of the parental magmasfrom which the beenremoved by erosion.More the Banded seriesmust cumulateswere precipitated. Most layeredcomplexes are ognized that many cumulates of magma different from that accompaniedby finer grainedsatellite intrusions, but the have been derived from a (seee.g., McCallum relationship between the compositions of the smaller which producedthe lower cumulates Hess'sapproach to estimating bodies and the compositionsof the magmasparental to et al., 1980),which makes lessviable. the cumulates is usually not straightforward(see, e.g., what has beenlost to erosion Sharpe,l98l; Cawthornet al., l98l). If the layeredin- trusion is itself complete,its bulk composition may be FoR THE inferred by summing up the individual layers,assuming ClNorurr PARENTAL MAGMAS that the intrusion representsa closedsystem. Incomplete B,lslr, AND ULTRAMAFTCSERTES sections, and the possibility of open-system behavior, The footwall of the Stillwater Complex contains a maf- make the problem considerably more difficult for most ic sill swarm, formerly included in the Basalseries. Though rntrusrons. now excluded(Zientek et al., 1985) from the Complex The Stillwater Complex of Montana is a large layered proper, thesesills are closelyassociated with the Complex mafic intrusion, which has lost its upper part to erosion. in space(Page, 1979; Zientek, 1983)and time (Premoet The preservedcumulates include the Basalseries (dom- al., 1990; DePaolo and Wasserburg,1979). The five in- inatedbycumulatesof orthopyroxenet plagioclase:see dependentmagma types found in the Basal sill swarm Hess, 1960; Page, 1979)followed by the Ultramafic series have been characteized petrographically,chemically, and (see,e.g., Jackson, l96l;Raedeke and McCallum, 1984). isotopically (Helz, 1985; Zientek, 1983; Lambert et al., The latter unit hasbeen subdivided (Zientek et al., 1985) 1994).Two are magnesian(Helz, 1985),enriched in light into the Peridotite zone (olivine + orthopyroxenecu- rare earth elements(LREE) (Helz, 1985; Lambert and mulatesoccurringincyclicunits)andtheBronzititezone Simmons, 1987), and have e*ovalues (Lambert et a1., (orthopyroxene cumulates). These mafic and ultramafic 1994) similar to those of the Ultramafic seriescumulates, layers are overlain by a thick section of plagioclase-rich characteristicsthat make them candidate parental mag- cumulates,the Bandedseries (Hess, 1960; McCallum et mas for the Basaland Ultramafic seriescumulates. Cal- a1.,1980).Theuppercontactof thestillwaterhasnot culationsusingthemethodofGhiorso(1985)suggested been preserved,however, so some fraction of the upper that a 50:50 mix of the two magnesiansill types might part of the Complex has been removed by erosion (Hess, have a more Stillwater-like crystallization sequencethan 1960).For many yearsthe Stillwatercumulates were con- either type alone.The compositionsof the two samples sideredto have beenproduced by the very efficient,closed- used as starting materials and that of the 50:50 mix are system fractionation of a single magma (Hess, 1960; shown in Table l. 0003404x/95lI l l 2-l 343$02.00 r343 t344 HELZ: STILLWATER COMPLEX

TABLE1 . Bulksilicate compositions of sillsused as experimental P= 1 5 kbar P=2530kbar startingmaterials graphrtepresenl graphitepresent 3 a 1 2 Magnesian Maficnorite. Mix 50:50-- gabbronoritef a sio, 53.10 51.90 50.70 a o 1200 Tio, 0.44 0.73 1.02 X Alr03 11.20 13.20 15.20 o > FeO l 14.70 12.72 10.74 F MnO 0.18 0.18 0.18 E Mgo 12.70 11.O2 9.33 q ilso 7.00 8.50 10.00 U F Naro 0.41 1.11 1.83 OLIVINE-OUT KrO 0.19 0.27 0.34 Pq a PrO. 0.08 0.09 0.10 Sum 100.00 99.72 99.45 S content 1.68 0.85 0.02 - Mafic norite sample 81VCZ-23. The sample consists ot orthopyroxene, plagioclase,minor augite and ilmenite,and trace amounts of chromite and apatite. lt also contains about4h sulfide.The S in the bulk analysis Magnesian Mix Mafic Magnesian Mix Malic was calculated as FeS, and omitted from the renormalizedanalysis, to gabbronorite norite gabbronorite norite give a clearerpicture of the bulk silicatecomposition. H2O was also omitted in the renormalization. Fig. 1. Experimentallydetermined phase relations of the .'Mixture of rock powders shown in columns 1 and g. The mixture compositionsshown in Table I, as a functionof temperature containsabout 2h sulfide.S in the bulk analysiswas handledas in col- andcomposition, at 1.5kbar (left) and 2.5-3.0kbar (right). The umn1. platinum the Magnesiangabbronorite sampleswere contained in Fe-saturated capsules; t sampleNB I8/374. The samplecontains or- graphite thopyroxene, plagioclase,augite, minor ilmenite, and trace amounts of /o, of the sampleswas definedby the + CO + CO, biotite,apatite, and sulfide. buffer.The shadedregion shows the locationof the crystallization sequenceolivine > orthopyroxene> plagioclase> augite. Bracketsaround a symbolindicate temperature uncertainties that includea contributionfrom a temperaturegradient; the uncer- ExpnnrprnrrrAt, RESULTSAND DrscussroN taintyin temperaturefrom othersources is +10'C. The "X" position The crystallizationsequences of thesethree composi- symbolson the2.5-3.0 kbar T-X sectionshow the of two experimentsperformed at 2.5-2.6kbar. Symbols marked tions at 1.5 and 3.0 kbar are shown in Figure l. The with "P" indicateCa-poor pyroxene that is pigeoniterather than sequenceof crystallizationfor the major silicateminerals orthopyroxene.All assemblagesin the maficnorite and the mix (olivine > orthopyroxene > plagioclase > augite) that containimmiscible sulfide liouid in additionto the ohasesin- matches the sequenceinferred for the lower parts of the dicated. Stillwater is observedin the 50:50 mix at 1.5 kbar and can be inferred to occur in slightly different mixing ratios at 3 kbar, as indicatedin Figure 1. Compositionsof the lower temperaturesat lower pressures.The desired se- nearJiquidus olivine and orthopyroxeneare Forr_roand quenceshould exist in the magnesiangabbronorite just Eflrn_r,,which match the observedcompositions of oliv- above3 kbar; however,olivine doesnot disappearinthat ine and orthopyroxenein the lowermost part of the Pe- composition until 7" < 1100 'C, far below the first ap- ridotite zone (Raedekeand McCallum, 1984;Loferski et pearanceof both plagioclaseand augite,which is incon- al., 1990).The principal differencebetween the experi- sistent with the phaserelations observedin the Stillwater mental resultsand the crystallizationsequence inferred cumulates. Only intermediate mixtures of the two liquid from the cumulatesis the late crystallizationof chromite typesexhibit the right crystallizationbehavior at reason- in the experimentsvs. its ubiquity in the Peridotitezone. able pressures. This results from the use of graphite to control f, during Figure 2 shows how the position of the olivine-ortho- the experiments and confirms earlier inferences(Mathez pyroxene boundary shifts with pressure,relative to the I et al., 1989) that although parts of the Stillwater Complex atm phaserelations, in the systemFo-An-SiOr. By 1.5 contain graphite, the fo, in the early stagesofits devel- kbar, the boundary shifts to near the position of the mix; opment was higher and evolved toward graphite satura- at 3.0 kbar it lies betweenthe mix and the magnesian tion with time. gabbronoritecompositions, essentially coinciding with the The T-X relations show that neither of the naturally enstatite-anorthitejoin,a position not reachedin the Fe- occurringliquids could serveas sole parent to the Basal free systemuntil l0 kbar (Senand Presnall,1984). This and Ultramafic seriesin this pressurerange. Olivine, the meansthat the boundary is no longera peritectic,but a characteristiccumulus mineral of the Peridotitezone, is cotectic boundary, and the reaction of olivine with the completely absent from the crystallization range of the liquid to produce orthopyroxeneno longer occursat near- mafic norite at and above 1.5 kbar. The magnesiangab- liquidus conditions.This shift in the position of the ol- bronorite by itself can be ruled out as a parent because ivine-orthopyroxene boundary places an upper limit on orthopyroxene in gabbronorite begins to crystallize only the pressure at which the Ultramafic series could have at the same temperature as plagioclaseat 3 kbar and at crystallized, as resorption of olivine next to orthopyrox- HELZ: STILLWATER COMPLEX r345

CaAl2Si206

T a m @ @ C I m

r p

summlr Stillwater magma chamber reservolr Fig. 3. Sketchshowing the relativesizes and approximate depthsof the summitreservoir of KilaueaVolcano and the Still- watermagma chamber. The vertical exaggeration is 2:1.

gle-magma models for the origin of the Basal and Ultra- mafic series.It offers an explanation for the Fe-rich na- io*rr.*'r./ ture of the first cumulus olivine (Forn) to appear at the baseofthe Peridotitezone, for the abrupt disappearance of olivine [still at its peakMg/(Mg * Fe) contentof Fo"] Mg2Si04 MgSiO3 50 Si02 at the top ofthe Peridotitezone, and for the fact that no Fig. 2. The planeforsterite(Fo)-anorthite(An)-SiO,, in weight major discontinuity in Mg/(Mg + Fe) in the cumulus percent.Short dashed lines are isothermsand light solid lines mineralsis observedat eitherstratigraphic position (Rae- arethe phaseboundaries at I atm, both as determinedby An- deke and McCallum, 1984);all these featuresare very dersen(1915). The compositionsin TableI areshown by the difficult to explain by fractionation of a single liquid. In black circles.The heavysolid lines show the position of the a magma-mixing model, the shift to olivine-dominant olivine-orthopyroxeneboundary for thesecompositions at 1.5 cumulates within the Peridotite zone would be produced and 3.0 kbar, as deducedfrom the experimentalresults shown by a changein the mixing ratio, and henceof SiO2activ- in Fig. 1. Notethat by 3.0kbar, this boundarycoincides with liquids in the recharge,from nearly 1000/o the compositionaljoin betweenenstatite (orthopyroxene) and ity, of the two anorthite(marked by longdashes). mafic norite to > 500/omagnesian gabbronorite, not by the arrival of less-differentiatedliquid. The lossof olivine at the top ofthe Peridotitezone would not reflectfraction- ene is observedin the Ultramafic series(Jackson, l96l; ation (for which there is no evidencein the Mg' of the Raedekeand McCallum, 1984).The presentresults sug- cumulates) but either a decreasein the fraction of the gestthat total pressureat the baseof the Stillwater cham- gabbronoriterelative to the mafic norite or its complete ber was <3.0 kbar durine formation of the Ultramafic loss as a component of the new magma entering the senes. chamber. This model is also more consistentwith the heterogeneity of REE patterns in Ultramafic series cu- INrpr,rc-c.troNs FoR THE STTLLwATERCoMpLEx mulates (Lambert and Simmons, 1987),and with their The Stillwater's inferred crystallization sequence is isotopic heterogeneity(Lambert et al., 1994), than mod- basedin part on the nature ofthe cyclic units in the Pe- els invoking a single parent for the Ultramafic series. ridotite zone(Jackson, 1961, I 970).The observationthat Lastly, it is consistentwith the observedpresence of two this sequenceis reproducedonly in a hybrid composition magnesiansill types in the footwall of the Complex,nei- (Fig. 1) at reasonablepressures suggests that the Perido- ther of which by itself hasthe right compositionto be the tite zone may be the product of mixing of the two distinct parental magma. The absenceof hybrids among the foot- liquids rather than the repeatedintrusion of only one. wall sills implies that the mixing took placein the Still- Comparison of the phaserelations calculatedfor the maf- water chamber itself. ic norite with those observedin Figure I suggeststhat The upper pressurelimit of 3 kbar found here is con- olivine replaces orthopyroxene on the liquidus of that sistent with P-7" conditions inferred from contact meta- composition somewherebetween 0.5 and 1.0 kbar, at morphic assemblagesin the footwall hornfelses(825 "C, which point the desired Stillwater crystallization se- 3 + 0.5 kbar; Vaniman et al., 1980;Labotka, 1985), if querce would be producedin that one liquid. However, one assumesthose conditions representreequilibration the orthopyroxene-richBasal zone cumulatesand those during post-Stillwater cooling of the contact aureole; the of the Bronzitite zone could be derived from the mafic pressurein the footwall at the end of the developmentof norite magmatype aloneat 1.5-3.0 kbar. To producethe the StillwaterComplex can only havebeen higher than it olivine cumulatesof the Peridotite zone at the samepres- was during the development of the Ultramafic series.The sures that obtained during crystallization of the under- inferred pressure of formation of the Ultramafic series lying Basal zone and the overlying Bronzitite zone re- (1.5-3.0 kbar) requiresthat the magmachamber was rel- quires the presenceof a second,less siliceous magma. atively shallow (approximately3-9 km). This is similar The two-liquid model has severaladvantages over sln- to the 2-6 km depth of the main part of the summit 1346 HELZ: STILLWATER COMPLEX

reservoir of Kilauea Volcano (Ryan, 1981), one of the Stillwater Complex: Assemblagesand conditions of metamorphism. In Montana SpecialPublication,92,70-76. most active volcanoesin the world. The BureauofMines and Geology cross-sectionallambert, D.D., and Simmons, E.C. (1987) Magma evolution in the Still- area of Kilauea's reservoir is inferred to be about 2 x 2 water Complex, Montana: I. Rare-earthelement evidencefor the for- km (Ryan, l98l ), whereasthat of the Stillwateris at least mation of the Ultramafic series.American Journal of Science,287, 40 x 20 km (Hess,1960; Brodzowski, 1985), as shown t-Jz. in Figure 3. Given that the Stillwater Complex was two Lambert,D.D., Walker,R.J., Morgan, J W., Shirey,S.8., Carlson, R.W., Zientek,M L., Lipin, B.R., Koski, M.S., and Cooper,R.L. (1994)Re- orders of magnitude larger than Kilauea's reservoir, and Os and Sm-Nd isotope geochemistryof the Stillwater Complex, Mon- that its inferred depth was not much greater, it seems tana: Implications for the petrogenesisof the J-M reef. Journal of Pe- quite possiblethat the Stillwaterwas a subvolcanicmag- trology,35, 1717 -17 53. ma chamber,as inferred by Lipin (1993).It presumably Lipin, B.R. (1993) Pressureincrease, the formation of chromite seams, and the development of the Ultramafic seriesin the Stillwater Com- underlay a sizeableplateau of mafic lavas. If so, it seems plex, Montana, Journal of Petrology, 34, 955-976 Iikely that much of the missing material of the complex Loferski,P.J., Lipin, B.R, and Cooper,R.W. (1990)Petrology of chro- would not be eroded cumulates but a lone-vanishedvol- mite-bearing rocks from the lowermost cyclic units in the Stillwater canlc carapace. Complex, Montana. U.S. Geological Survey Bulletin, 1674-E, l-28. Mathez,E.A., Dietrich, V J., Holloway,J.R., and BoudreauA.E. (1989) AcrtvowlnocMENTs Carbon distribution in the Stillwater Complex and evolution of vapor during crystallization of Stillwater and Bushveld magmas. Joumal of I wish to thank D.L. Hamilton, Senior Lecturer in Geologyat the Uni- Petrology,30, 153-173. versity of Manchester,in whose laboratory these experimentswere per- McCallum, I S., Raedeke,L.D, and Mathez,E.A. (1980)Investigations formed. I am also grateful for the assistanceof B Kjaersgaardand J. of the Stillwater Complex: Pan I Stratigraphy and structure of the Small for their assistance.The paper benefitted from reviews by B.R. Bandedzone. American Journal ofScience, 280-4. 59-87. Lipin, J.S.Huebner, J.S. Pallister, G.K. Czamanske,S B. Shirev.and M.L. Page,N.J. (1979) Stillwater Complex, Montana: Structure, mineralogy, Zientek petrologyofthe Basalzone with emphasison the occurrenceofsulfides U.S Geological Survey ProfessionalPaper, 1038, 69 p. Rnrnnnncns crrED Premo,W.R., Helz,R.T., Zientek,M.L., and l-angston,R B (1990)U-Pb and Nd-Sm agesfor the Stillwater Complex and its associatedsills and Andersen, O (1915) The system anorthite-forstente-silica. American dikes, Beartooth Mountains, Montana: Identification of a parent mag- Joumal ofScience(Fourth Series),39, 4O7-454. ma?Geology, 18, 1065-1068. Brodzowski, R.A. (1985) Cumulate xenoliths in the Lodgepole, Enos Raedeke,L.D , and McCallum, I.S. (1984) Investigationsin the Stillwater Mountain and Susie Peak intrusions: A guide. In Montana Bureau of Complex: Part II. Petrology and petrogenesisofthe Ultramafic series. Mines and Geology SpecialPublication, 92, 368-372. Joumal of Petrology, 25, 395-420. Cawthorn, R.G., Davies, G., Clubley-Armstrong,A., and McCanhy, T.S. Ryan, M.P. (1981) Modeling the three-dimensionalstructure of macro- (1981)Sills associatedwirh the BushveldComplex, South Africa: An scopicmagma transport systems:Application to Kilauea Volcano, Ha- estimateof the parentalmagma composition. Lithos 14, l-16. waii. Journal of GeophysicalResearch, 86, 7 | | | -7 129. DePaolo, D J., and Wasserburg,G.J (1979) Sm-Nd age of the Srillwater Sen, G., and Presnall, D C. (1984) Liquidus phase relationships on rhe Complex and the mantle evolution curve for neodymium. Geochimica join anorthite-forsterite-quartzat l0 kbar with applications to basalt et CosmochimicaActa. 43. 999-1008. petrogenesisContributions to Mineralogy and Petrology,85, 404-408. Ghiorso, M S. (1985) Chemical mass transfer in magmatic processes:I. Sharpe, M.R. (1981) The chronology of magma influxes to the eastern Thermodynamic relations and numerical algorithms. Contributions to compartment of the Bushveld Complex as exemplified by its maryinal Mineralogy and Petrology, 90, l2l-l4l border groups Journal of the GeologicalSociety of London, I 38, 307- Helz, R.T. (1985) Composition of fine-grainedmafic rocks from sills and J ZO- dikes associatedwith the Stillwater Complex, Montana. In Montana Vaniman, D.T., Papike,J.J., and Labotka,T. (1980)Contact-metamor- Bureau of Mines and Geology SpecialPublication, 92, 97-l 17. phic effectsofthe Stillwater Complex, Montana: The concordant iron Hess, H.H. (1960) Stillwater igneouscomplex, Montana: A quantitative formation. American Mineralogist, 65, 1087- l 102. mineralogicalstudy. GeologicalSociety of America Memoir, 80, 230 p. Zientek, M.L. (1983)Petrogenesis of the Basalzone of the Stillwater Com- Jackson, E.D. (1961) Primary textures and mineral associationsin the plex, Montana Ph.D. thesis,Stanford University, Stanford,California. Ultramafic zone of the Stillwater Complex, Montana. U.S Geological Zientek, M.L., Czamanske,G.K., and Irvine, T.N. (1985)Stratigraphy SurveyProfessional Paper, 358, 106p. and nomenclature for the Stillwater Complex. In Montana Bureau of -(1970) The cyclic unit in layered intrusions: A comparison of re- Mines and GeologySpecial Publication, 92, 2l-32. petitive strati$aphy in the Ultramafic parts of the Stillwater, Muskox, Great Dyke and Bushveld Complexes.Geological Society ofSouth Af- rica, SpecialPublication l, 391-424. MaNuscnrrr RECETvEDJuNr 24, 1995 Labotka, T C (1985) Petrogenesisofthe metamorphrc rocks beneaththe Merrrscnrrr ACCEITEDSppre^{sen 20, 1995