Borisova_etal_J. Petrology-2012.pdf

A NewView on the Petrogenesis of the OmanOphiolite Chromitites from Microanalysesof Chromite-hosted Inclusions

ANASTASSIAY. BORISOVA1,2*, GEORGES CEULENEER1,VADIM S. KAMENETSKY3, SHOJI ARAI4, FRE DE RIC BE JINA5,6,BE NE DICTE ABILY1, ILYA N. BINDEMAN7, MIREILLE POLVE 1,PHILIPPE DE PARSEVAL1, THIERRYAIGOUY1 ANDGLEB S. POKROVSKI1

1GE OSCIENCES ENVIRONNEMENT TOULOUSE, GET-UMR 5563, OBSERVATOIRE MIDI-PYRE NE ES, UNIVERSITE

TOULOUSE III/CNRS/IRD/CNES, 14 AVENUE E. BELIN, 31400 TOULOUSE, FRANCE2GEOLOGICAL DEPARTMENT, LOMONOSOV MOSCOW STATE UNIVERSITY, VOROBIEVI GORI, 119991, MOSCOW, RUSSIA3CODES, ARC CENTRE OF EXCELLENCE IN ORE DEPOSITS, UNIVERSITY OF TASMANIA, HOBART, TAS. 7001,

AUSTRALIA4DEPARTMENT OF EARTH SCIENCES, KANAZAWA UNIVERSITY, KANAZAWA 920-1192, JAPAN5UNIVERSITE DE TOULOUSE III, UPS-OMP, IRAP, TOULOUSE, FRANCE6CNRS, INSTITUT DE RECHERCHE EN ASTROPHYSIQUE ET PLANE TOLOGIE, IRAP, 14 AVENUE E. BELIN,

31400 TOULOUSE, FRANCE7GEOLOGICAL SCIENCES, UNIVERSITY OF OREGON, 1275 E 13TH STREET, EUGENE, OR, USA

RECEIVED NOVEMBER 2, 2011; ACCEPTED JULY 16, 2012ADVANCE ACCESS PUBLICATION SEPTEMBER 12, 2012

To decipher the petrogenesis of chromitites from the MohoTransitionZone of the Cretaceous Oman ophiolite, we carried out detailed scan-ning electron microscope and electron microprobe investigations of500 silicate and chromite inclusions and their chromite hosts, andoxygen isotope measurements of seven chromite and olivine fractionsfrom nodular, disseminated, and stratiform ore bodies and associatedhost dunites of the Maqsad area, Southern Oman. The results,coupled with laboratory homogenization experiments, allow severalmultiphase and microcrystal types of the chromite-hosted inclusionsto be distinguished. The multiphase inclusions are composed ofmicron-size (1^50 mm) silicates (with rare sulphides) entrappedin high cr-number [100Cr/(CrAl) up to 80] chromite.The highcr-number chromite coronas and inclusions are reduced (oxygen fuga-city, fO2, of 3 log units below the quartz^fayalite^magnetitebuffer, QFM). The reduced chromites, which crystallized between600 and 9508C at subsolidus conditions, were overgrown by more

oxidized host chromite (fO2QFM) in association with microcrys-tal inclusions of silicates (plagioclase An86, clinopyroxene, and par-gasite) that were formed between 950 and 10508C at 200MPafrom a hydrous hybrid mid-ocean ridge basalt (MORB) melt.Chromium concentration profiles through the chromite coronas, inclu-sions, and host chromites indicate non-equilibrium fractional crystal-lization of the chromitite system at fast cooling rates (up to 018Ca1). Oxygen isotope compositions of the chromite grains imply in-volvement of a mantle protolith (e.g. serpentinite and serpentinizedperidotite) altered by seawater-derived hydrothermal fluids in anoceanic setting. Our findings are consistent with a three-stage modelof chromite formation involving (1) mantle protolith alterationby seawater-derived hydrothermal fluids yielding serpentinitesand serpentinized harzburgites, which were probably the initialsource of chromium, (2) subsolidus crystallization owing to pro-grade metamorphism, followed by (3) assimilation and fractional

*Corresponding author. Present address: GET, UMR 5563,14 AvenueE. Belin, 31400 Toulouse, France. Telephone: 33(0)5 61 33 26 31.Fax:33(0)5 6133 25 60. E-mail: [email protected]

The Author 2012. Published by Oxford University Press. All rightsreserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 53 NUMBER12 PAGES 2411^2440 2012 doi:10.1093/petrology/egs054 at Biblio Planets on N

ovember 22, 2012

http://petrology.oxfordjournals.org/D

ownloaded from

crystallization of chromite from water-saturated MORB.This studysuggests that the metamorphic protolith assimilation occurring atthe Moho level may dramatically affect MORB magma chemistryand lead to the formation of economic chromium deposits.

KEY WORDS: Oman ophiolite; chromitite; chromian spinel; MORB;inclusions; Moho Transition Zone; hydrothermal fluid; serpentinites;protolith; oxygen isotopes; AFC

I NTRODUCTIONChromian spinel in the mantle^crust transition zone(referred to here as the Moho Transition Zone or MTZ)plays an important role in the global budget of chromiumin the Earths lithosphere. The chemical composition ofchromian spinel and chromite-hosted inclusions is variable,reflecting the wide range of physicochemical and geolo-gical conditions that can exist along the interface betweenthe crust and upper mantle (e.g. Dick & Bullen, 1984;Roeder & Reynolds, 1991; Sack & Ghiorso, 1991;Kamenetsky, 1996; Borisova et al., 1997, 2002, 2008;Kamenetsky et al., 2001; Arai et al., 2006). Chromian spinelis a main economic Cr reserve on Earth and may formlarge chromitite ore bodies whose petrogenesis remains asubject for debate. The pioneering model of McDonald(1965) on the Bushveld chromitites invoked the formationof Cr-rich immiscible liquids, forming at temperaturesabove 14008C and even as high as 20008C (Jackson, 1966).However, this model cannot be applied to chromititesin ophiolites or layered intrusions where hydrous silicateminerals, crystallizing at much lower temperatures, aresystematically found as inclusions in chromite (e.g.Lorand & Ceuleneer, 1989; Melcher et al., 1997; Li et al.,2005; Arai et al., 2006; Page et al., 2008; Rollinson, 2008).Other models were subsequently proposed to explain chro-mitite petrogenesis based on experimental data (e.g.Roeder & Reynolds, 1991), chromite-hosted inclusion data(e.g. Melcher et al., 1997; Schiano et al., 1997; Li et al., 2005;Spandler et al., 2005; Borisova et al., 2008), or both (e.g.Irvine, 1975; Matveev & Ballhaus, 2002). Most modelsinvoke a magmatic system, based on experimental data onbasaltic systems (e.g. Irvine, 1975; Murck & Campbell,1986; Roeder & Reynolds, 1991; Matveev & Ballhaus,2002). For example, Irvine (1975) suggested either crustalcontamination or silicate melt mixing to stabilize chromiteon the liquidus in such a system. Roeder & Reynolds(1991) suggested a highly reduced basaltic magma as thechromium source. Matveev & Ballhaus (2002) proposed amagmatic system saturated with an aqueous fluid phasewith physical separation of chromite from olivine. Thesemodels consider either open-system interactions ofolivine-saturated melts with felsic melts (Irvine,1975) or in-congruent dissolution of crustal rocks with the formationof chromite-saturated melts (Be dard et al., 2000). Hydrous

magmas are indeed favourable for abundant chromitecrystallization because the presence of water in a maficmelt widens the stability domain of spinel, especiallyunder oxidizing conditions (Ford et al., 1972; Feig et al.,2006) and stabilizes chromite on the liquidus (Nicholson& Mathez, 1991). However, a major shortcoming of mostprevious models is the lack of a satisfactory explanationfor the source of the Cr capable of accounting for theformation in a mid-oceanic ridge setting of huge chromititeore bodies reaching tens of thousand cubic meters (e.g.Lorand & Ceuleneer, 1989).Most researchers acknowledge the role of aqueous

fluids or hydrous silicate melts in the petrogenesis of theOman chromitites (Johan et al., 1983; Auge , 1987; Lorand& Ceuleneer, 1989; Leblanc & Ceuleneer, 1992; Schianoet al., 1997; Arai et al., 2004, 2006; Ahmed et al., 2006;Rollinson, 2008). The source of water and its involvementin the Maqsad MTZ chromitite petrogenesis is, however,controversial.Water might originate from either (1) hydro-thermal fluid sources (Auge & Johan,1988) or (2) dehydra-tion of a subducting slab (Schiano et al., 1997; Ahmedet al., 2006; Rollinson, 2008) or (3) re-melting of hydratedlithosphere in a spreading-ridge setting in response to epi-sodic emplacement of mantle diapirs at the MTZ (Benoitet al., 1999; Python et al., 2007).Studies of inclusions trapped in chromian spinel crystals

may provide key information on the petrogenesis of chro-mitite ore bodies. Chromian spinel is an ideal microcon-tainer for inclusions owing to its resistance to alteration(e.g. Shimizu et al., 2001; Arai et al., 2006, 2011). In contrastto chromites disseminated in volcanic rocks (Kamenetsky,1996; Borisova et al., 2001; Kamenetsky et al., 2001, 2002),chromian spinel forming chromitite ores contains abun-dant inclusions of fluid, hydrous and anhydrous silicates,sulphides, and platinum group minerals (Auge , 1987; Auge& Johan, 1988; Lorand & Ceuleneer, 1989; Melcher et al.1997; Kamenetsky et al., 2001, 2002; Arai et al., 2004;Spandler et al., 2005; Ahmed et al., 2006; Borisova et al.,2008). Inclusions of hydrous silicate minerals have beenfound in chromian spinel from chromitites in different tec-tonic settings, ranging from subduction zones to spreadingridges, and in their fossil equivalents exposed in ophiolites(e.g. Lorand & Ceuleneer, 1989; Melcher et al., 1997; Araiet al., 2006; Rollinson, 2008). However, the origin of thesehydrous silicate minerals and the role of aqueous fluidsand/or hydrous melts in chromite petrogenesis remainunclear.We carried out a systematic investigation by scanning

electron microscope and electron microprobe of about500 chromite-hosted inclusions and their host chromites aswell as oxygen isotope measurements of chromite and oliv-ine grain fractions from four nodular, disseminated, andstratiform chromitites and three associated host duniteslocated at different levels of the MTZ in the Oman

JOURNAL OF PETROLOGY VOLUME 53 NUMBER 12 DECEMBER 2012

2412

at Biblio Planets on Novem

ber 22, 2012http://petrology.oxfordjournals.org/

Dow

nloaded from

ophiolite, Maqsad area. We also performed homogeniza-tion experiments on the chromite-hosted inclusions,and analysed numerous elemental profiles through inclu-sion^host interfaces. The new data are evaluated withexisting petrogenetic models and imply that metamorphicrecrystallization of a mantle protolith altered by seawater-derived hydrothermal fluids is likely to be the key processresponsible for the formation of chromitite ore in an ocea-nic MTZ setting.

GEOLOGICAL BACKGROUNDAND SAMPLESThe Maqsad area of the Oman ophioliteThe Oman ophiolite is an 95^100Ma fragment of theTethyan oceanic lithosphere obducted onto Arabian con-tinental lithosphere (Coleman, 1981; Tilton et al., 1981;Tippit et al., 1981). It was detached during an intra-oceanicthrusting event that occurred 95 Myr ago; that is, soonafter its igneous accretion (Ghent & Stout, 1981; Boudieret al., 1985, 1988; Montigny et al., 1988). The final emplace-ment onto the Oman margin was achieved duringMaastrichtian times (70 Ma) (Glennie et al., 1973;Coleman, 1981). The present exposure of the Oman ophio-lite results from the recent (Miocene) uplift of the Omanrange (Glennie et al., 1973). The detailed context of thepetrogenesis and emplacement of the Oman ophiolite isstill debated. It may have evolved in a mid-ocean ridge set-ting (e.g. Coleman, 1981) or in a marginal, possiblyback-arc, setting (e.g. Pearce et al., 1981; Searle & Cox,1999; Tamura & Arai, 2006).

Maqsad mantle diapirThe Maqsad area is located in the southeastern part of theOman ophiolite (Fig. 1). Its structural and igneous evolu-tion was conditioned by the emplacement of a mantlediapir at Moho level below a spreading axis (Jousselinet al., 1998; Rabinowicz & Ceuleneer, 2005). The Maqsadmantle diapir is less than 10 km in cross-section but consti-tutes the central part of a much larger structure: a formerspreading segment, about 80 km in length parallel to thepaleo-spreading axis (which trends NW^SE). Abundantprimitive melts percolated through the mantle peridotitein the diapir and are now manifested by troctolitic andgabbroic impregnations and segregations (Python &Ceuleneer, 2003; and references therein). The parentalmelts of the Maqsad troctolites and gabbros were similarto mid-ocean ridge basalt (MORB) as attested by the crys-tallization sequence and mineral composition (Kelemenet al., 1995; Python et al., 2008; and references therein).Their isotopic signature is typically that of present-dayIndian MORB, pointing to a common mantle source(Benoit et al., 1999; and references therein).

The crustal section at the top of the Maqsad diapir ismade of flat-lying, slightly dipping to the SE, interlayeredcumulates such as dunites, troctolites, olivine gabbrosand gabbros, presenting variably developed modal gradedbedding. The thickness of this gabbroic unit reaches about2 km. The overlying sheeted dykes have a NW^SE azi-muth. The regularity of the gabbro layering is locally per-turbed by syn-magmatic normal faults (Abily et al., 2011).The petrological nature of MORB-derived segregationsin the mantle peridotite that were processed in theMaqsad diapir contrasts with the more common situationobserved in Oman where mantle dykes are essentiallyfilled with pyroxenite and high-Mg gabbronorite (Python& Ceuleneer, 2003). The parental melts of the Oman pyr-oxenites and gabbronorites are interpreted to have beenformed by low-pressure hydrous re-melting of the hydratedoceanic lithosphere in a mid-ocean ridge setting (Benoitet al., 1999; Cle net et al., 2010). Such pyroxenites and gab-bronorites are absent from the Maqsad diapir but occur asintrusions in the mantle peridotite and as layered lowercrustal cumulates 10^15 km away from the Maqsad diapir.These pyroxenites and gabbronorites are possibly relics ofshallow lithospheric melts that formed in the early stagesof the opening of the Maqsad basin, and that were eventu-ally transported passively into an off-axis position bysea-floor spreading (Benoit et al., 1999; Cle net et al., 2010).

Signatures of hydrothermal activityEvidence for deep and very hot (up to at least 9008C)hydrothermal activity is strong in the Maqsad area. It ismanifested by veins (usually ridge-parallel) filled withhydrothermal diopside and anorthite (Python et al., 2007),peridotites transformed into assemblages of talc and carbo-nates, and igneous minerals (cumulates from the lowercrust) partially transformed into amphibole and greenspinel in the vicinity of syn-magmatic faults (Abily et al.,2011). These are the footprints of transient interaction pro-cesses between the hot basaltic melts percolating throughthe Maqsad diapir and filling the magma chamber andhydrothermal fluids reaching the Moho level. More clas-sical lower-temperature minerals and their mineral assem-blages (serpentine, chlorite, tremolite, sulphide, sulphate,carbonate and zeolite) are also abundant in faults andcracks withn the Maqsad mantle and lower crustal sections(Abily et al., 2011; and references therein). This hydrother-mal activity has led locally to the formation of small plagi-ogranitic pods (Amri et al., 1996, see also Koepke et al.,2007).

Maqsad MohoTranzition Zone (MTZ)The transition between the layered gabbroic cumulatesand the underlying mantle harzburgites is represented byan 300m thick dunitic transition zone. The dunitic rocksare particularly well developed above the center of theMaqsad mantle diapir (Fig. 2). Chromitite ore bodies,

BORISOVA et al. OMANOPHIOLITE CHROMITITE PETROGENESIS

2413



Dow

nloaded from

which are the focus of the present study, are abundantin the Maqsad dunitic transition zone (Leblanc &Ceuleneer, 1992). They were neither deformed nor trans-posed by mantle solid-state flow, unlike most of the chro-mitite ore bodies cropping out in the mantle harzburgites.This good preservation of the Maqsad chromitite bodiesoffers a unique opportunity to study in detail the magmaticprocesses related to chromite ore genesis. Our studyfocuses on the undeformed chromitites, which are amongthe most spectacular manifestations of complex processesin the MTZ. The chromitite samples investigated comefrom three ore bodies located at various levels in theMTZ above the Maqsad diapir (Fig. 2).

Host dunitesThe host dunite samples are predominantly composed ofolivine (495% of olivine, Fo 875^905 mol %), scattered

chromian spinel (cr-number varies from 32 to 55) andinterstitial diopside (Abily & Ceuleneer, in press). Theinvestigated samples are weakly altered (30% serpen-tine) dunites (11TUF1: 23861686N, 578575844E, 883m;11TUF2: 23861596N, 57858378E, 898m) anddiopside-bearing dunite (11TUF7: 2386954N, 578582826E, 977m) hosting the chromitites at the Maqsad MTZ.The dunite samples consist of partly serpentinized olivine(Fo 87^89 mol %), scattered chromite (cr-number50^54) and interstitial diopside (mg-number 96).

Stratiform chromititeSample 04 OM 31E is from a stratiform chromitite bodyat the base of the MTZ (23807.055N, 57856.618E, 950m).This body is about 45m thick and consists of interlayeredhorizons of massive ore (490% chromite) up to 3m inthickness and of disseminated ore. The massive ore is a

Fig. 1. Map of the Oman ophiolite showing the location of the Maqsad study area. Troctolites are MORB cumulates developed above andwithin the Maqsad diapir surface area (1000 km2). There are numerous occurrence of evolved lithospheric cumulates (gabbronorites and pyr-oxenites) near the Maqsad area (marked by orange and green shading). Modified after Python & Ceuleneer (2003).


2414



Dow

nloaded from

compact assemblage of euhedral to subhedral chromitegrains 100^200 mm in size; the interstitial silicate mineralsare olivine, diopside, pargasite and, rarely, enstatite.The disseminated ore exhibits antinodular textures (nodu-lar aggregates of olivine grains embedded in a matrix ofchromian spinel) or, more commonly, chain textures (sub-hedral single olivine grains with interstitial chromianspinel). These textures suggest that a partly consolidateddunitic host was invaded and partly disaggregated by ahydrated melt that precipitated chromian spinel ( othersilicate minerals) (e.g. Be dard et al., 2007). The chromianspinel to olivine ratio exhibits gradual variations, defininggraded bedding. This indicates that the stratiform chromi-tite formed a suspension at some stage of its evolution.Chromite grains contain silicate and chromite inclusions(Table 1).

Disseminated chromititeSample 04 OM 52D from a disseminated chromitite of themedian part of the dunitic MTZ (23805.703N, 57857.675E,997m) where irregular patches of interstitial chromianspinel (locally concentrated to form lenses of massive ore)are abundant and interpreted as former magmatic impreg-nations (Leblanc & Ceuleneer, 1992). The silicate mineralsare olivine, clinopyroxene, pargasite, and plagioclase.

Nodular chromititeTwo samples of nodular chromitite [83 OG 68(1) and83 OG 68(2)] come from an 3m thick and 35mhigh dyke located at the top of the MTZ (23806.013N,57858.601E, 1225m), several tens of meters below thelayered gabbros. Troctolitic and gabbroic sills are presentin the dunite that hosts the dyke. The dyke is vertical andstrikes parallel to the paleo-ridge axis (NW^SE). The con-tact between the chromite ore filling the dyke and thehost dunite is sharp. The texture of the ore is nodular.Chromitite nodules may reach 2 cm in size. They consistof sintered aggregates of minute (a few hundred micro-metres) chromite grains containing silicate inclusions(Table 1). A nucleus rich in silicates (olivine, clinopyroxene,plagioclase) is often present in the core of the nodules. Atthe base of the dyke [83 OG 68(1)], the nodules are tightlypacked and flattened. The size and abundance of the nod-ules decrease progressively upward, and the texture of theore becomes disseminated [single chromian spinel grainsembedded in a silicate matrix, 83 OG 68(2)] at the topof the dyke. This chromitite ore body is pervasivelyaffected by hydrothermal alteration, primary magmaticminerals from the silicate matrix (olivine, clinopyroxeneand plagioclase) being partly transformed into low-temperature hydrous minerals such as serpentine, chlorite,prehnite and zeolites (Leblanc & Ceuleneer, 1992).Carbonates are also commonly found as alterationproducts.

Sample preparation and analysesWe selected samples that had escaped pervasive trans-formation of the primary silicate assemblage intolow-temperature hydrothermal phases. The freshest sam-ples still contain some minor secondary minerals (5%)such as tremolite, hydrogrossular, talc, chlorite, carbonateand serpentine (in dunites) and sulphides scattered be-tween and inside the primary grains along cracks andreplacing the primary silicate phases. A database of allthe studied silicate and chromite phases, including thehost chromite and minerals of the chromite-hosted inclu-sions (Table 1), is available in the Electronic Appendix(EA; available for downloading at http://www.petrology.oxfordjournals.org).Different inclusion types were identified based on de-

tailed backscattered electron images of about 200 inclu-sions in four samples of nodular, disseminated andstratiform chromitite using a scanning electron microscope[JEOL JSM-6360 LV at Geosciences EnvironmentToulouse (GET), France and JEOL JSM-6480LV withEDS INCA-Energy 350 at Geological Department ofMoscow State University, Moscow, Russia]. Major andminor element concentrations in the minerals of the inclu-sions and their host spinel were determined by electronmicroprobe using a CAMECA SX50 microprobe (GET)

Fig. 2. Schematic geological cross-section of the Moho TransitionZone (MTZ) in the Maqsad area of the Oman ophiolite. Chromititeore bodies are localized in the host dunite layer (300m thickness).


2415



Dow

nloaded from

with SAMX automation and wavelength-dispersive spec-trometry (WDS). Analyses were conducted using an accel-erating voltage of 15 kV, a beam current of 10 nA forsilicate minerals and 20 nA for chromian spinels, and abeam size of 2 mm. The applied acquisition conditions aretypical for standard electron microprobe analysis (e.g.Borisova et al., 2010), allowing low detection limits (300^700 ppm for most major and minor elements in chromiteand 1200 ppm for Ni) and high internal precision (RSD3% for major Mg, Al, Cr and Fe, and 30% for minorMn andTi in chromite). The compositions are reported inFigs 3^6, Table 2 and the Electronic Appendix. The ferric

iron content of the chromite was determined by assumingideal stoichiometry, XY2O4, where X (Fe2, Mg, Ni,Mn, Co, Zn) and Y (Cr3, Fe3, Al). Titanium isassumed to be present as an ulvo spinel component, andCr as Cr3. Iron is subdivided into ferrous and ferric to sat-isfy the condition nY 2 nX, where nY is the totalnumber of trivalent cations and nX is the total number ofdivalent cations per unit cell. Possible Cr2 contents inchromite are negligible because only Cr3 is a major com-ponent stabilizing chromian spinel (Hanson & Jones,1998).The large size of the chromite crystals and aggregates

makes them ideal targets for high-precision analyses by

Table 1: Types of chromite-hosted inclusions of nodular, disseminated and stratiform chromitites of the Maqsad MTZ(Oman ophiolite)

Type Shape Mineralogical

composition

Features of inclusion Sample number Figure Participating

melt or fluid

Multiphase assemblage

1 Rounded or

lens-shaped

CrNa Parg EnstCpxSulph asco-entrapped phases*

CrNa Parg overgrowths on

Enst; associated with Chr

coronas

Nodular: 83 OG 68(2) 4b Aqueous fluid

2 Rounded or

octahedral

shape

Aspid EnstCrNaPargSulph

Aspid is the predominant

mineral in inclusion; Aspid

overgrowths on Enst and

CrNa Parg; associated with

Chr coronas

Nodular: 83 OG 68(1)

Disseminated: 04 OM 52D

4d, 5a Aqueous fluid

2a Octahedral

shape

Aspid/Phlog/CrNa

Parg EnstSulph,rare Serp, altered

spinel

Aspid/Phlog/CrNa Parg are

associated with Chr coronas;

predominant type in the

stratiform chromitite 04

OM 31E


Stratiform: 04 OM 31E

4c, 5b and f Aqueous fluid

Microcrystal(s)

3 Euhedral

Anhedral

EnstCpx, Fo Euhedral, anhedral as asso-ciated with Chr coronas

Nodular: 83 OG 68(2)


4e, 5e Aqueous fluid

4 Anhedral Cpx/PlagCrNa Parg Microcrystals are isolated orassociated; Trem as replacing

Plag

Nodular: 83 OG 68(1) 83

OG 68(2) Disseminated:

04 OM 52D

4f Hydrous hybrid

MORB

5 Anhedral

Euhedral

HgrossCrNa Parg Isolated crystals; Hgross asreplacing Plag

Nodular: 83 OG 68(2) 4g Hydrous hybrid

MORB

6 Euhedral Chr Isolated primary crystals;

frequent association with

hydrous silicate

micro-inclusions

Disseminated: 04 OM 52D 4h, 5c and d Aqueous fluid

Aspid, aspidolite; Cpx, clinopyroxene (between augite and diopside composition); Plag, plagioclase; Chr, chromite withhigh cr-number (chromite phase visible with optical microscopy in type 2a and invisible in type 2 inclusions); CrNa Parg,CrNa pargasite; Enst, enstatite; Fo, forsterite; Hgross, hydrogrossular; Phlog, phlogopite (in stratiform chromitites);Sulph, sulphides; Trem, tremolite replacing plagioclase; Serp, serpentine (in stratiform chromitites).*FeNi and OsRuIr sulphide microphases (lauriteerlichmanite group) are identified in the type 1, 2 and 2a multiphaseinclusions.


2416



Dow

nloaded from

Fig. 3. (a^c) Cr-number [100Cr/(CrAl)],TiO2 and Fe2O3 contents vs mg-number [100 Mg/(MgFe2)] of the host chromite, chromite cor-onas and inclusions in nodular and disseminated chromitites of the Maqsad MTZ of the Oman ophiolite. Chromite data are summarized inTable 2 and the Electronic Appendix.The data for Haylayn metamorphic harzburgitescorrespond to chromites from the serpentinized perido-tites of the Haylayn area (G. Ceuleneer, unpublished data). The typical Maqsad MTZ chromite compositions are according to G. Ceuleneer(unpublished data). The MORB field corresponds to chromite compositions from http://www.petdb.org (Lehnert et al., 2000) and data sum-marized by Kamenetsky et al. (2001, 2002); Arc boninites are chromite compositions from Sigurdsson et al. (1993) and Kamenetsky et al. (2001,2002) and the chromite database of Barnes & Roeder (2001). (d^f) Cr-number [100Cr/(CrAl)], TiO2 and Fe2O3 contents vs mg-number[100 Mg/(MgFe2)] of the host chromite and chromite coronas in the Maqsad stratiform chromitites of the Oman ophiolite. The MORBfield and Arc boninitic seriesor Arc boninites field are shown for comparison. All data sources are as in (a^c).


2417



Dow

nloaded from

Fig. 3. (Continued)


2418



Dow

nloaded from

Fig. 4. Backscattered electron (BSE) images and X-ray map (using wavelength-dispersive spectroscopy;WDS) of chromite-hosted inclusions inchromitites of the Maqsad MTZ. Cr^Na Parg, Cr^Na pargasite; Aspidolite is Na phlogopite; En, enstatite; Cpx, clinopyroxene; Grt, hydrogros-sular; Chr, high cr-number chromite coronas and inclusion. (a) BSE image showing the distribution of multiphase and microcrystal inclusionsin the entire host chromite grain [83 OG 68(2) nodular chromitite]; (b) type 1 multiphase inclusion containing enstatite, clinopyroxene andCr^Na pargasite [83 OG 68(2) nodular chromitite]; (c) type 2a coexisting inclusions consisting of single-phase Cr^Na pargasite and aspidoliteenveloped by high cr-number chromite coronas (04 OM 31E stratiform chromitite); (d) type 2 multiphase inclusion containing Cr^Na pargasiteand aspidolite [83 OG 68(1) nodular chromitite]; (e) type 3 inclusion Mg X-ray map (Ka) of enstatite and clinopyroxene [83 OG 68(2) nodularchromitite]; (f) type 4 microcrystal inclusion containing clinopyroxene and pargasite (04 OM 52D disseminated chromitite); (g) type 5 micro-crystal inclusion with hydrogrossular (Grt) and Cr^Na pargasite [83 OG 68(2) nodular chromitite]; (h) type 6 microcrystal inclusion of highcr-number chromite (04 OM 52D disseminated chromitite).


2419



Dow

nloaded from

Fig. 5. (a, b) Cr-number and mg-number profiles through chromian spinel adjacent to examples of (a) type 2 and (b) 2a inclusions in theMaqsad MTZ disseminated and stratiform chromitites. X is the measured length of the profile of the type 2 inclusion [see equation (1) inthe main text] and is obtained by finding the intersection between the estimate of the cr-number in the core of the grain and the tangent tothe concentration profile near the corona^host interface (shown by the grey lines). Data are given in Electronic Appendix. (c, d) BSE imageof an example of a chromite inclusion in host chromite of the disseminated chromitite 04 OM 52D. (c) Cr-number, mg-number and (d) Fe2/Fe3 in chromian spinel and melt calculated for profile A^B through the type 6 inclusion. Data are given in the Electronic Appendix. Xis the length measured on the profiles of type 6 inclusion [see equation (1) in the main text] and is obtained by finding the intersection be-tween the estimate of the cr-number in the core of the grain and the tangent to the concentration profile near the inclusion^host interface(shown by the grey lines). (e) BSE image of olivine (Fo95) in a type 3 inclusion overgrown by the high cr-number chromite phase (Chr) andthe host chromite in the disseminated chromitite 04 OM 52D. The transition zone between the olivine grain (Fo94^95; CaO up to 003 wt %and Cr2O3001wt % at olivine grain center) is enriched in Cr (up to Cr2O301^03wt %) at the contact with the high cr-number chro-mite. Cr2O3 (wt%) and mg-number in chromian spinel (and Fo mol%, in olivine inclusion) are illustrated for profile A^B. Data are given inthe Electronic Appendix. (f) BSE image of an example of altered spinel (cr-number 30) and pargasite in a type 2a inclusion overgrown bythe high cr-number chromite phase and the host chromite in the disseminated chromitite 04 OM 52D. Cr-number and mg-number in spinelsare illustrated for profile A^B. Data are given in the Electronic Appendix.


2420



Dow

nloaded from

laser fluorination for oxygen isotopes (Bindeman, 2008).Grain fractions of four chromite samples, treated in dilutedHCl, rinsed with H2O and dried, were selected foroxygen isotope analysis. To separate the primary olivinegrains from the secondary serpentine, olivine fractionsfrom three dunite samples were treated chemically atGET according to the leaching method of Snow et al.(1994). Three leaches were made on each crushed dunitesample, using leaching solution of 62N HCl, 5% HF. Thefirst leach was cold for 5min, the second was in an ultra-sonic bath for 10min, and the third was in the ultrasonicbath for 10min followed by 10min at 1258C. The sampleswere then dried and the fresh olivine cores were selectedfor laser fluorination analysis at the University of Oregon

(USA). Samples were heated with an infrared laser(96 mm, CO2) in the presence of purified BrF5 to releaseoxygen. The generated O2 gas was purified in a series ofcryogenic traps held at liquid nitrogen temperature, andthen a mercury diffusion pump was used to remove any re-maining traces of BrF5. Oxygen was converted to CO2gas by reacting with graphite at high T (14508C), theyields were measured, and CO2 was analyzed by isotoperatio mass spectrometry (IRMS) in a dual inlet mode.Four to seven garnet standards (d18O 652, Universityof Oregon garnet from Gore Mt, NY, USA named asUOG garnet) were analyzed together with the samplesduring each of seven analytical sessions. Day-to-day d18Ovariability of standards ranged from 01 to 03 lighter

Fig. 5. (Continued)


2421



Dow

nloaded from

than their reference values and the measurements of un-knowns were adjusted accordingly. The precision on stand-ards is better than 01 of 1s standard deviation.Homogenization experiments on selected inclusions

were conducted at the University of Tasmania (Hobart,Australia) to study the silicate inclusion transformationsupon heating. Hand-picked chromite grains from the nodu-larchromitite sample (83OG68-1)wereplaced inaplatinumcontainer and heated in a vertical furnace at 130058C for3min under high-purity argon gas flow. These conditionswere found to be suitable for complete melting of the multi-phase inclusions entrapped in chromite. Heating was fol-lowed by quenching in water. Then the grains werehand-picked and mounted in epoxy and polished to exposethe inclusions to the surface. Silicate glasses (silicate inclu-sions were converted to homogeneous melt at 13008C) inmore than 60 experimentally homogenized inclusions wereanalysed at standard conditions (e.g. Kamenetsky et al.,

2001) for major elements using a Cameca SX-50 microprobe(UniversityofTasmania, Australia).

RESULTSChromite host, coronas, and inclusionsThe compositions of the host chromian spinel (up to70^80% of the chromite volume) are reported in Figs 3and 5, Table 2, and the Electronic Appendix.The chromites in nodular sample 83 OG 68(1) havemg-numbers (54^63), cr-numbers (50^54), TiO2 (06^09wt %) and Fe2O3 (02^60wt %) contents typical ofthe Maqsad MTZ nodular chromitites. Host chromianspinel in both disseminated and stratiform samples (04OM 52D and 04 OM 31E) is also typical of the highmg-number chromites of the Maqsad MTZ. The associ-ation of chromite (cr-number 50^62; mg-number54^68) with olivine (Fo88^91) from the dunite hosting the

Fig. 5. (Continued)


2422



Dow

nloaded from

Fig. 5. (Continued)


2423



Dow

nloaded from

Fig. 5. (Continued)


2424



Dow

nloaded from

Fig. 6. (a^d) Major elements: (a) SiO2, (b) TiO2, (c) Cr2O3, (d) FeO contents vs Al2O3 contents in silicate and chromite phases of multiphaseand microcrystal inclusions from the Maqsad MTZ nodular [83 OG 68(1,2)], disseminated (04 OM 52D) and stratiform (04 OM 31E) chromi-tites. Grey stars indicate the glass compositions of experimentally homogenized inclusions at 13008C (this study). Lines show the compositionaltrends for melts obtained in homogenization experiments at 1050^14008C (open stars with corresponding homogenization temperature;Schiano et al., 1997).Chromite profiles indicateTiO2, FeO and Al2O3 contents in chromites adjacent to the multiphase inclusions.The high con-centrations of FeO (up to 14wt %) and Cr2O3 (1^3wt %) in the experimentally homogenized inclusions are probably controlled by a strongdiffusive Fe input from the adjacent chromite during the laboratory experiment and the initially high Cr contents in hydrous silicate minerals,respectively (see Electronic Appendix). The hydrous minerals and experimental melt compositions are recalculated on a water-free basis. In(b) the Melt Field indicates the calculated TiO2 and Al2O3 contents in equilibrium melts based on the compositions of the host chromite andchromite coronas and inclusions (as well as the equilibrium chromite^melt partitioning; see text). TiO2 melt contents (52wt %) calculatedbased on the composition of the type 4 clinopyroxenes [and using experimental KCpx=meltd clinopyroxene^mafic melt partition coefficients from027 to 043 of Dunn (1987), Hart & Dunn (1993) and Skulski et al. (1994); and TiO2Cpx 02^05 wt %] are lower than those measured in thehomogenized glasses. The type 1, 2, 4 and 5 pargasite gives TiO2 melt512wt % [as calculated based on the experimental K

Amph=meltd amphi-

bole^mafic melt partition coefficient range 095^129 of Adam & Green (1994) and LaTourette et al. (1995); and TiO2Parg15^35wt %],which is lower compared with the homogenized glasses.The type 2 phlogopite givesTiO2 melt512wt % [calculated based on the experimentalKPhlog=meltd phlogopite^mafic melt partition coefficient range 098^177 for phlogopite of Adam & Green (1994) and LaTourette et al. (1995);and TiO2Phlog15^25wt %], which is lower than that measured in the homogenized glasses. All data are given in Table 2 and theElectronic Appendix.


2425



Dow

nloaded from

Maqsad chromitites corresponds to the MORB field in theolivine^spinel mantle array (Arai et al., 2011; and refer-ences therein). It should be noted that the mg- andcr-numbers in the host chromite of the Maqsad MTZ chro-mitites are typical of MORB (Fig. 3), whereas the TiO2and Fe2O3 contents plot in the intermediate field betweenthose of the typical MORB and the hydrothermallyaltered harzburgites.Typical compositional profiles through chromite cor-

onas and inclusions (up to 10^15% of the chromitevolume) are shown in Fig. 5. The chromite coronas innodular, disseminated and stratiform chromitites havehigh cr-numbers (68^80) and low mg-numbers (50^58). Asimilar range of high cr-numbers (73^80) and lowmg-numbers (54^60) is observed in the chromite inclu-sions (Type 6) in the host chromite of the disseminatedchromitites (Table 2 and Electronic Appendix). The simi-lar idiomorphic and compositional features of the high

cr-number chromite coronas and inclusions (Figs 3^5)suggest that they were formed in a common process.High cr-numbers and low TiO2 contents in the chromitecoronas and inclusions resemble those of arc boninites(Fig. 3). In contrast, the Maqsad MTZ chromite coronasand inclusions have much lower Fe2O3 contents thanthose of chromites from typical arc boninites andMORB. The Maqsad high cr-number chromites are alsodifferent from the chromites described in anomalousMORB-related plutonic rocks from Ocean DrillingProgram (ODP) Holes 735B (Southwest Indian Ridge)and 895 (Hess Deep) (Arai & Takemoto, 2007;Kamenetsky & Gurenko, 2007), which have higher TiO2and Fe2O3 contents. It should be noted that the extremelylow TiO2 and Fe2O3 contents in the Maqsad highcr-number chromites approach those of chromites in theHaylayn hydrothermally altered harzburgites of theOman ophiolite (Fig. 3).

Fig. 6. (Continued)


2426



Dow

nloaded from

Chromite-hosted inclusion types andmineral associationsThe Maqsad MTZ chromite-hosted multiphase andmicrocrystal inclusions consist of silicate, sulphide andchromite of 1^50 mm in size. Silicates make up to 20% ofthe host chromite grain surface (Fig. 4a). Six main typesand one sub-type of inclusion are distinguished based oninclusion shape (rounded to octahedral, anhedral to euhe-dral), mineral assemblages and composition; they

are summarized in Table 1 and illustrated in Figs 4b^hand 5a^f.

Type 1, 2 and 2a multiphase inclusionsType 1, 2 and 2a multiphase inclusions are composed ofhydrous and anhydrous silicates and minor sulphides,and are always associated with high cr-number coronas(Table 1). Type 1 contains abundant Cr^Na pargasite, andtypes 2 and 2a contain aspidolite (Na phlogopite). In type

Table 2: The Maqsad MTZ chromite inclusions and their host chromite compositions

Label: 04 OM

52Di-incl-3

04 OM

52Di-incl-4

04 OM

52Di-incl-5

04 OM

52Di-incl-8

04 OM

52Di-incl-15

04 OM

52Di-incl-16

04 OM

52Di-incl-18

04 OM

52Di-incl-19

04 OM

52Di-incl-20

Analysis number: 50 51 52 59 85 86 88 91 92

TiO2 009 009 005 008 009 014 014 D.L. 013

Al2O3 1129 1152 1222 1026 1178 1350 1159 1082 1156

Cr2O3 5826 5732 5693 5970 5636 5437 5643 5709 5685

FeOtot 1738 1788 1734 1752 1854 1814 1848 1797 1800

MnO 024 036 036 023 023 027 032 015 032

MgO 1198 1187 1287 1088 1137 1183 1121 1184 1184

NiO 008 0011 D.L. 002 012 013 009 008 D.L.

Sum 9951 9934 10033 9871 9861 9866 9853 9869 9886

FeO* 1511 1567 1489 1655 1605 1583 1599 1581 1550

Fe2O3 253 246 272 107 277 257 277 241 278

Fe2/Fe3spinel 663 708 609 645 684 642 730 620

Fe2/Fe3melt 3342 3641 2989 3225 3481 3208 3791 3064

Al2O3 melt 1088 1097 1124 1046 1107 1171 1100 1069 1099

TiO2 melt 017 017 010 015 017 023 023 5003 022

100Mg/(Mg Fe2) 5857 5745 6063 5394 5580 5712 5553 5717 5765100Cr/(CrAl) 7759 7696 7577 7962 7625 7299 7657 7797 7674

Label: 04 OM

52Di-host-1

04 OM

52Di-host-2

04 OM

52Di-host-3

04 OM

52Di-host-4

04 OM

52Di-host-5

04 OM

52Di-host-6

04 OM

52Di-host-8

04 OM

52Di-host-9

04 OM

52Di-host-10

Analysis number: 53 54 57 58 65 66 89 90 96

TiO2 021 018 023 021 013 008 018 029 019

Al2O3 2259 2257 2243 2270 2022 2278 2260 2248 2237

Cr2O3 4568 4546 4564 4613 4862 4506 4514 4463 4511

FeO* 1265 1211 1219 1228 1275 1289 1249 1201 1201

Fe2O3 293 363 369 367 333 307 375 395 360

MgO 1475 1501 1513 1537 1443 1440 1168 1503 1505

Fe2/Fe3spinel 479 371 367 372 426 466 371 338 371

Fe2/Fe3melt 2185 1565 1541 1571 1873 2108 1562 1383 1564

100Mg/(Mg Fe2) 6752 6884 6886 6906 6686 6657 6783 6904 6908100Cr/(CrAl) 5758 5747 5773 5770 6174 5703 5727 5713 5750

*FeO, Fe2O3, Fe2/Fe3 in spinel are calculated according to spinel stoichiometry; Al2O3 and Fe

2/Fe3 in melt arecalculated after Maurel & Maurel (1982a, 1982b, 1982c). TiO2 in melt is calculated after Kamenetsky et al. (2001) andRollinson (2008) using the equation for Arc. SiO2 contents are below 05wt %. D.L., values below detection limit.


2427



Dow

nloaded from

1, 2 and 2a inclusions, Cr^Na pargasite surrounds enstatitelocated in the inclusion core and is associated with aspido-lite; it is occasionally co-entrapped with clinopyroxene(Fig. 4b^d). Subtype 2a inclusions are distinguished fromthe main type 2 by euhedral coronas of high cr-numberchromite overgrowing pargasite, aspidolite or phlogopite(Fig. 4c). The chromite-hosted micas have variable Na2Oand K2O contents ranging from aspidolite to phlogopite(phlogopite occurs only in the type 2a inclusions ofthe stratiform chromitite; Electronic Appendix). Rare in-clusions of an altered spinel and serpentine assemblageare also associated with high cr-number coronas (Fig. 5f,Table 1).

Type 3, 4, 5 and 6 microcrystal inclusionsType 3, 5 and 6 inclusions are euhedral microcrystals ofsilicates and high cr-number chromite. Type 3 inclusionscontain an association of clinopyroxene with enstatite [lim-ited by a single crystal in nodular chromitite, sample 83OG 68(2); Fig. 4e] associated with the host chromite. Incontrast, the high cr-number chromite phases are foundadjacent to type 3 forsterite grains (Fig. 5e), suggestingthat the olivine is associated with the high cr-number chro-mite rather than with the host chromite. In addition, type4 anhedral inclusions associated with the host chromitecontain clinopyroxene and bytownite partially trans-formed into secondary phases (Table 1), sometimes asso-ciated with Cr^Na pargasite (Fig. 4f). In addition,anhedral type 5 inclusions, also associated with the hostchromite, are composed of hydrogarnet and occur aloneor in association with Cr^Na pargasite (Fig. 4g). Type 6 in-clusions are euhedral microcrystals of high cr-numberchromite (Figs 4h and 5c, d). Their euhedral shape indi-cates a primary origin and an early entrapment into thehost chromite.

Silicate mineral compositionThe compositions of Cr^Na pargasite, aspidolite, phlogo-pite, olivine, clinopyroxene and enstatite are summarizedin Fig. 6 and the complete dataset is provided in theElectronic Appendix. The composition of the chromite-hosted plagioclase (An86) is in the range of that inorthopyroxene-rich, high-Mg gabbronorites depleted in in-compatible trace elements drilled from the Mid-AtlanticRidge at Deep Sea Drilling Project (DSDP) Site 334(Nonnotte et al., 2005), of those from the periphery of theMaqsad diapir (Benoit et al., 1999), and of those of theMaqsad mantle plagiogranites (Amri et al., 1996). In con-trast, the chromite-hosted Fe^Mg silicates have highermg-numbers (90^97) and Cr2O3 (05^4wt %) than (1)those from the troctolitic and gabbroic cumulates, occur-ring in the crustal section of the Maqsad area and fillingformer melt channels and dykes in the Maqsad mantlediapir (e.g. Python & Ceuleneer, 2003), (2) those from theMaqsad mantle plagiogranites (Amri et al., 1996), and (3)

those from the DSDP Site 334 gabbronorites (Nonnotteet al., 2005). The Maqsad MTZ clinopyroxenes (varyingfrom augite to diopside in composition), Cr^Na pargasite,and micas are variable in TiO2 (typically 02^05wt %,15^35wt %, and 17^40wt %, respectively, Fig. 6).

Composition of homogenized inclusionsThe glasses obtained from the homogenization experi-ments at 130058C show large variations in the concen-trations of major elements, especially Al2O3 (10^18wt %),SiO2 (44^56wt %) and CaO (4^10wt %) (Fig. 6;Electronic Appendix). These compositions are similar tothose of polymineral solid inclusions from the chromititedyke homogenized at 13008C by Schiano et al. (1997).In the latter study, the inclusions were homogenized at1050^14008C. They show that compositional trends in allhomogenized inclusions are controlled by incongruentmelting with formation of olivine at 1150^12008C.Intra-inclusion homogeneity of the quenched glasses wasobserved by Schiano et al. (1997) only at 13008C. Theaverage TiO2melt, Al2O3melt, FeOmelt and MgOmelt con-tents in the homogenized inclusions are 26 04wt %,142wt %, 11118wt %, and 10521wt %, respect-ively, whereas the corresponding host chromite containsaverage TiO2Chr, Al2O3Chr, FeOChr and MgOChr contentsof 04 005wt %, 231wt %, 1207wt %, and154 05wt %, respectively (for 60 pairs of the homo-genized inclusion^host chromite, see ElectronicAppendix). The corresponding partition coefficients(KChr=meltd ) between the host chromite and the melts homo-genized at 13008C are: Kd Ti 015003; Kd Al1702 and Kd Fe/Mg (FeOChrMgOmelt)/(FeOmeltMgOChr) 0701 (see Electronic Appendix). The partition coeffi-cient KChr=meltd for Ti is far from the equilibrium chro-mite^melt partitioning summarized by Kamenetsky et al.(2001), which is 08 at TiO2Chr 04wt %. In contrast,Al2O3melt contents (average 142wt %) and FeO/MgOmelt ratios (average 1101) in the homogenizedmelts are in accord with equilibrium with the host chro-mite (Electronic Appendix). The high concentrations ofFeO (up to 14wt %) in the homogenized inclusionsheated to 13008C would require a strong diffusive Fe inputfrom the chromite coronas or host during the experiment(e.g. Kamenetsky et al., 2001; Danyushevsky et al., 2002).The elevated Cr2O3 contents (average 1805wt %) inthe quenched glasses are probably due to the initialhigh Cr2O3 contents in the hydrous silicate minerals andadditional chromite dissolution in the melt (e.g. Murck &Campbell, 1986).

Oxygen isotope signatures of chromitesand olivines from the Maqsad MTZWe performed the first measurements of the oxygen isotopecomposition of chromite and associated olivine grains inthe MTZ of the Oman ophiolite. The d18O values of


2428



Dow

nloaded from

chromite grains (Maqsad chromitites) vary from 32 to37 (Table 4). The d18O values of single olivine grains inthe associated Maqsad dunites vary from 52 to 58,which is similar to and slightly higher than the518028% reported for olivines from mantle peridotites(Mattey et al., 1994), perhaps reflecting protolith seawateralteration at low temperatures.

DISCUSSIONWas a silicate melt involved in the MaqsadMoho Transition Zone petrogenesis?In this section we use the composition of the Maqsad MTZchromites, together with the new data from this study andpublished experiments on the homogenization of thechromite-hosted multiphase inclusions, to better constrainthe origin of the chromite-hosted inclusions. The knownsensitivity of the Cr/Al ratio in chromian spinel to wateractivity and the composition of the original melt (Irvine,1975; Onuma & Tohara, 1981; Dick & Bullen, 1984; Roeder& Reynolds, 1991) can be used to better constrain theseparameters during chromite crystallization. Calculationsof the melt composition using the analysed chromitecompositions and reported experimental chromite^meltpartitioning coefficients (Maurel & Maurel, 1982a, 1982b,1982c; Kamenetsky et al., 2001; see Table 2 and ElectronicAppendix) imply that the melts in equilibrium with thehost chromite of the nodular, disseminated, and stratiformchromitites had high Al2O3 melt (145wt % on average)and TiO2 melt (08wt %) contents, FeO/MgOmelt of 1,and high mg-numbers (65), consistent with primitive tho-leiitic MORB-like compositions (Fig. 6b; ElectronicAppendix). The calculated melts in equilibrium with thehigh cr-number chromite coronas and inclusions hadlower Al2O3 melt (10^11wt %) and TiO2 melt (03wt %)contents, but similar mg-numbers of 65 (Table 2). TheTiO2 melt contents are much lower than those measured inthe homogenized glasses (2^4wt %, Fig. 6b). Moreover,the temperature stability domain of amphibole and micasand associated chromites below 10508C (see discussionbelow; Table 3) indicates that the observed homogeniza-tion temperature of 13008C for the multiphase inclusionsis much higher than the true temperature of the MaqsadMTZ chromitite crystallization. Therefore, neither hostchromite, high cr-number inclusions and coronas, nor sili-cate minerals in the multiphase inclusions could have crys-tallized from the experimentally produced melts. Thelarge inter-inclusion compositional variation of the meltscan be explained only by heterogeneous entrapment ofsilicates (i.e. variable proportions of enstatite, pargasite,aspidolite and/or phlogopite). The euhedral shapes of thehigh cr-number coronas and inclusions and their predom-inance over the silicate phase in the multiphase inclusions(Figs 4c and 5b) indicate that the high cr-number chromite

crystallized early, either as microcrystals (Figs 4h and 5c,d) or as chromite overgrowths on the silicate crystals ortheir assemblages (Fig. 5e and f) rather than as a daughterphase from the homogenized melt as previously suggestedby Schiano et al. (1997) and Borisova et al. (2008).In the framework of an entirely magmatic scenario,

non-equilibrium (fractional) crystallization of the hostchromite accounts for the preservation of early-stagephases (high cr-number chromite). However, our hom-ogenization experiments provide no evidence either for en-trapment of a melt phase or for a melt^crystal reaction(e.g. incongruent reaction with melt formation), whichcould be responsible for the Maqsad chromite crystalliza-tion. We did not find any geological evidence for a closeassociation of the Maqsad chromitites with pyroxenitesor anorthosites. Therefore, a chromitite origin by incon-gruent dissolution of pyroxene or feldspar into an oliv-ine chromite-saturated melt (Be dard & He bert, 1998;Be dard et al., 2000) is also unlikely for the Maqsad MTZchromitites. Because the experimentally homogenizedmelts are well explained by heterogeneous trapping of sili-cates, the Maqsad MTZ chromites could not have crystal-lized from an arc boninite melt, in contrast to thesuggestion of Schiano et al. (1997).

Constraints on T^P^aH2O conditions of theMaqsad MTZ chromitite crystallizationThe mineral assemblages of the chromite-hosted inclu-sions, together with available experimental and geologicaldata on mineral phase equilibrium, provide new con-straints on the chromite crystallization sequence and theT^P^aH2O conditions during chromitite crystallization(Table 3).

High cr-number chromite corona and inclusioncrystallizationThe first group of constraints stems from the multiphaseinclusions (Types 1, 2 and 2a) and euhedral type 3 inclu-sions associated with the high cr-number chromite coronas.The high mg-numbers of the chromite-hosted Fe^Mg sili-cates are consistent with high water activity (aH2O1) inthe crystallizing melt (e.g. Koepke et al., 2009) or its crys-tallization directly from a high-temperature aqueous fluid(e.g. Python et al., 2007). Phase-equilibria experiments(e.g. Presnall, 1995; Feig et al., 2006) indicate that, in thepresence of an H2O or H2O^CO2 phase, pargasiticamphibole forms from a silicate melt or fluid between 700and 10008C and P 50MPa. Above 900^10008C, chro-mian pargasite transforms into olivine, chromite and amelt (Johan & Le Bel, 1978), although pargasite may bestable at higherT in the presence of high Fand Cl contents(e.g. Presnall, 1995). It is well known that a miscibility gapexists in the aspidolite^phlogopite solid solution at200MPa and 7008C (e.g. Costa et al., 2001, and referencestherein). Thus, where aspidolite (Na end-member),


2429



Dow

nloaded from

phlogopite (K end-member) and Cr^Na pargasite are allpresent in type 2a chromite-hosted inclusions (Figs 4c and5b), all phases could have crystallized from a hydrother-mal fluid at temperatures as low as 7008C and pressuresof 200MPa. The association of clinopyroxene with ensta-tite and pargasite found in one inclusion of type 1 (Fig.4b) and in type 3 (Fig. 4e) suggests that these phases crys-tallized simultaneously and were entrapped near thetwo-pyroxene closure temperature of 860^9008C (ob-tained according to Wells, 1977). It should be noted thatthe Maqsad MTZ chromite-hosted Cr^Na pargasite andaspidolite contain high Na2O and Cl concentrations reach-ing 6 and 04wt %, respectively. Therefore, the assem-blage of type 1^3 inclusions with high cr-number

Table 3: T^P^fO2 conditions and model stages responsible for the Maqsad MTZ chromitite petrogenesis

Model stages: Serpentinized mantle

protolith

Prograde metamorphism Assimilation by MORB

Stage number: I II III

Altered spinel Serpentine assemblage High cr-number chromite Forsterite Phlogopite Aspidolite Clinopyroxene Enstatite* CrNa pargasite Plagioclase Host chromite Inclusion types: 2a 1, 2, 2a, 3, 6 4, 5

Pressurey (MPa): 200? 200 200Temperaturey (8C): 600 600950 (10) 9501050 (10)Redox conditionsz (log fO2, bars) 14 to 12 11Oxygen buffer: QFM 27 to QFM at 6009508C QFM 04 at 1000 158COlivinechromite assemblagez The Maqsad MTZ type 3 inclusion (Fo9596)

and incongruent olivine (Fo8591)Highcr-number chromite coronas and inclusions

Olivine (Fo8891) from the Maqsad

MTZ dunites Host chromitefrom the Maqsad MTZ

chromitites

*Rare presence of enstatite in chromitite matrix.yTemperatures were estimated based on the available experimental data on silicate mineral phase equilibrium (see text).The pressure of 200MPa used for the calculation (see below) is based on the thickness of the geological sequence of theMaqsad area (see text).zOxygen fugacity and closure temperature for the olivinechromite assemblages were calculated following ONeill &Wall (1987) and Ballhaus et al. (1991) at 200MPa. The log units of fO2 are relative to the quartzfayalitemagnetite(QFM) oxygen buffer.Composition of incongruent olivine appearing during homogenization experiment on polymineralic inclusions is afterSchiano et al. (1997).Composition of olivine Fo8891 from the Maqsad MTZ dunites, which are the host rocks for the investigated MTZchromitites, is according to Abily & Ceuleneer (in press). indicates presence of the phase or assemblage.

Table 4: Oxygen isotope composition d18O () of theMaqsad MTZ chromitites and host dunites

Host dunites Olivines* Chromitites Chromites

11 TUF1 58 01 83 0G 68 (1) 32 0111 TUF2 54 01 83 0G 68 (2) 38 0111 TUF7 52 01 04 OM 52D 36 01

04 OM 31E 35 01

*Oxygen isotope composition of the mineral fractions mea-sured at the University of Oregon (USA).


2430



Dow

nloaded from

chromite (Tables 1 and 3) may have crystallized in thepresence of an aqueous NaCl-bearing fluid. Our estimatesof closure temperature for olivine inclusions and highcr-number chromite fix the lowest temperature of thehigh cr-number chromite crystallization at 600^6308C[obtained using equations of ONeill & Wall (1987) andBallhaus et al. (1990, 1991) at 200MPa, Table 3]. Thus, aTof 600^9508C and P of 200MPa for high cr-numberchromite crystallization corresponds to subsolidus condi-tions for a hydrous mafic melt (similar to primitiveMORB; Feig et al., 2006) and implies metamorphic recrys-tallization in the presence of an NaCl-bearing fluid.

Host chromite crystallizationThe second group of constraints stems from anhedral type4 and 5 inclusions.The ore bodies were emplaced and crys-tallized at the base of the oceanic crust whose stratigraphicthickness does not exceed 6 km. This places a major geolo-gical constraint for the maximum lithostatic pressure ofcrystallization of 200MPa. Moreover, there is abundantcrystallization of MORB-type tholeiites along thelow-pressure olivine^plagioclase cotectic (troctolitic cumu-lates) evolving to the olivine^plagioclase^clinopyroxenecotectic as temperature decreases (olivine gabbro cumu-lates) in impregnations, sills, and dykes adjacent to thechromitite ore bodies (Korenaga & Kelemen,1997, and ref-erences therein). This provides further evidence forlow-pressure conditions in the Maqsad MTZ. Type 4 and5 inclusions, containing plagioclase (An86) in associationwith anhedral pargasite (Fig. 4f and g), could be at equilib-rium at 800^11208C and 200^300MPa according to thegeothermometer of Holland & Blundy (1994). To avoidcomplete destabilization of the chromite-hosted pargasite,the chromite crystallization should occur in the pargasitestability field (below 10708C; Muntener et al., 2001).Indeed, Feig et al. (2006) demonstrated that H2O-saturatedMORB melts crystallize pargasite below 10208C in equilib-rium with chromite at pressures of 500MPa. Thus,our findings of the same assemblage of chromite with par-gasite, clinopyroxene and plagioclase at lower pressures(200MPa) in the Maqsad MTZ offer two possibilities: (1)pargasite crystallization from a fluid-saturated (5wt %H2O at 200MPa) mafic melt (magmatic scenario), or(2) direct pargasite precipitation from an aqueous fluid orrecrystallization from another silicate mineral in the pres-ence of fluid (subsolidus scenario). We prefer the first hy-pothesis for the host chromite crystallization because, inthe case of a water-saturated MORB system at 200MPa,hydrous MORB melt is present above 9508C (Berndtet al., 2005; Feig et al., 2006), and the host chromite showscompositions similar to MORB chromites (Fig. 3).However, the second subsolidus scenario cannot be com-pletely excluded. Thus, the host chromite may crystallizeat 950^10508C from a water-saturated hybrid MORBmelt (see discussion below). Therefore, P of 20050MPa

andT from 600 to 10508C are likely to be the range of con-ditions of the Maqsad MTZ chromite formation in awater-bearing system.

Constraints on redox conditions andmantle sourceThe major challenge in understanding chromitite forma-tion is to identify a mechanism that concentrates Cr giventhat it has a low solubility in basaltic melts (Roeder &Reynolds, 1991) and aqueous fluids at redox conditionscorresponding to the quartz^fayalite^magnetite buffer(QFM) (see below). Our data suggest that the MaqsadMTZ chromitites crystallized in the presence of an aque-ous fluid and/or a hydrous melt sharing many characteris-tics with MORB. Because MORB magmas are generallydry (e.g. Danyushevsky et al., 1996; Asimov & Langmuir,2003), these conditions require an external source ofwater. The strong decrease in Fe2/Fe3 in the profilesfrom chromite inclusions to the host chromite (Fig. 5d) in-dicates a sharp change in temperature and/or oxygen fuga-city upon the chromite crystallization [fO2 is calculatedbased on the olivine^chromite equilibrium of ONeill &Wall (1987) and Ballhaus et al. (1990, 1991) at 200MPa,Table 3]. Initially high Fe2/Fe3 ratios (up to 16) aresimilar to those of chromite from the Oman metamorphicharzburgites (Fe2/Fe312^16), whereas Fe2/Fe3values of 3^5 are typical magmatic values (e.g.Kamenetsky et al., 2001). Three independent argumentsprovide evidence for reducing conditions during the earlystages of the chromite crystallization in the presence of ahigh-temperature fluid: (1) extremely high Fe2/Fe3

ratios in early (high cr-number) chromite (Table 2, Fig.5d); (2) elevated Cr contents (probably in the form ofCr2) in the silicate-phase inclusions (Fig. 6); (3) lowoxygen fugacity inferred from the olivine^chromite equi-librium. The reducing conditions (from QFM ^ 27 toQFM at 600^10008C, Table 3) evidenced for the chromitecrystallization strongly support the derivation of the chro-mite grains from serpentinized oceanic lithosphericmantle rather than from more oxidized fluids from thesubducting slab, at least in the Maqsad area of the Omanophiolite. Only serpentinized oceanic mantle can providehighly reducing fluids as evidenced by native iron in theserpentinized peridotites from the Maqsad area (Lorand,1987). Methane-rich fluids are observed as olivine-hostedinclusions in hydrated harzburgite (Sachan et al., 2007).The initial stages of dehydration of methane-bearing ser-pentinized mantle could result in the liberation of suchaqueous CH4-bearing fluids. Highly reduced C^O^Hfluids have been documented in chromite-hosted inclusionsin the Kempirsai ophiolites from the mantle^crust transi-tion zone (Melcher et al., 1997), and highly reducing condi-tions have also been inferred for the Thetford Minesophiolite (Page et al., 2008). Our estimates of the redox con-ditions of the chromite-hosted fluid, based on the


2431



Dow

nloaded from

Kempirsai fluid inclusion data (Melcher et al., 1997) usingthe SUPCRT thermodynamic database (Sverjensky et al.,1997), yield oxygen fugacity between QFM ^ 5 and QFM^ 2 at 10008C and 200MPa. The Cr2O3 (08^20wt %)and 100Mg/(Mg Fe2) (92^96) values in the MaqsadMTZ type 2 and 3 pyroxenes and forsterite inclusions aremuch higher than those typical of the Maqsad harzburg-ites (08wt % and 91, respectively; G. Ceuleneer, un-published data), suggesting that these phases are notxenocrysts derived from water-poor harzburgitic mantle.The Maqsad MTZ assemblage of high cr-number chro-mite (low Fe2O3) with forsterite (Fo95^96), pyroxene(mg-number 92^94), and pargasite (mg-number 90^94) inclusions (Tables 2 and 3; Electronic Appendix) sug-gests their origin to be from a highly refractory mantlesource, probably hydrated harzburgite.

Constraints on Cr mobility inhigh-temperature fluidsThe hydrous silicates of the chromite-hosted inclusionsare either magmatic or high-temperature hydrothermalphases as indicated by their high TiO2 (05wt %) andCr2O3 (15wt %) concentrations, because both Ti4and Cr3 are generally not mobile in low- to moderate-temperature aqueous fluids (4508C; e.g. Crerar et al.,1985). In addition, highT and reducing conditions favourlower oxidation states of multivalent elements (Feand Cr), which are far more soluble.Thus, the high Cr con-tent (up to 4wt % Cr2O3) in the Maqsad MTZchromite-hosted silicates (Fig. 6c) might be due to abun-dant Cr2 in the fluid^melt system. By analogy with Fe2,Cr2 and its Cl complexes are expected to be far more sol-uble than Cr3 species (e.g. Crerar et al., 1985; Pokrovskiet al., 2003). Equilibrium calculations using thermo-dynamic data for Cr hydroxide complexes from theSUPCRT database (Shock et al., 1997, http://geopig.asu.edu/index.html#) also indicate that the stability field ofCr2 species in aqueous solution under the redox condi-tions of the QFM buffer widens with increasing tempera-ture. The elevated Cr/Al ratios in boninitic melts andhydrothermal fluids may thus reflect high concentrationsof Cr2 in the form of chloride complexes in aqueous fluidunder reducing conditions that significantly enhance thesolubility of Cr3-bearing minerals, whereas the solubilityof Al3- and Ti4-bearing solid phases is unaffected bythe redox potential. In addition, dissolved silica may alsoinfluence Cr^Al^Ti fractionation during fluid^rock inter-action.The high silica contents, evident from silicate inclu-sions, may promote formation of soluble metal^silicatecomplexes as demonstrated by experimental studies forAl, Ti and similar metals (Fe, Ga) both in moderateT^Phydrothermal solutions (Pokrovski et al., 2002; and refer-ences therein) and in high T^P subduction-zone fluids(Manning et al., 2008; and references therein). Such com-plexes may significantly enhance Cr transport by

hydrothermal fluids. Such fluids may be responsible forthe metamorphic recrystallization at subsolidus conditionsof mineral phases rich in Cr.An alternative or additional form of chromium trans-

port and accumulation at subsolidus conditions may beso-called crystallosols containing finely crystallized chro-mian spinel (Pushkarev et al., 2007). The discovery inour work of micrometre-size multiphase silicates and asso-ciated high cr-number chromites in the Maqsad chromi-tites may represent such micron-size crystallosols. Thissubstance, when assimilated by the oxidized MORB melt,may trigger the host chromite crystallization.

Constraints on the Maqsad chromititecooling rate from elemental diffusionprofilesThe Maqsad MTZ chromite-hosted Fe^Mg silicates (for-sterite, pyroxene, amphibole and mica) of type 1^5 inclu-sions with elevated mg-numbers (90^97) have high MgO(16wt %) and low FeO (5wt %) contents (Fig. 6d,Electronic Appendix) compared with the lower MgO(12^15wt %) and higher FeO (15^20wt %) of the hostchromites. This was previously explained by subsolidusFe^Mg exchange between the Fe^Mg silicates and hostchromite (e.g. silicate enrichment in Mg and depletion inFe after entrapment owing to diffusive re-equilibration;Johan & Le Bel, 1978; Johan et al., 1983). However, this isunlikely for the following reasons: (1) the profiles throughthe host-chromites to silicate inclusions show no evidencefor Fe^Mg exchange between the chromite and silicates(e.g. no forsterite content variation is observed in the4100 mm olivine associated with high cr-number chromiteand host chromite, Fig. 5e); (2) the cooling rates (seebelow) are fast enough to inhibit diffusion across the sili-cate^chromite interface; (3) the Fe^Mg silicate mineralcompositions do not show any dependence on the inclusionsize (e.g. all silicates are systematically and homogeneouslydepleted in FeO, which is below 5wt %; Fig. 6d). Allthese observations suggest originally high MgO and lowFeO contents and high mg-numbers in the Maqsad MTZchromite-hosted silicates.We estimated closure temperatures for diffusive Fe^Mg

exchange between the Maqsad MTZ host chromite andthe Maqsad MTZ olivine of the host dunite to be1000158C at 200MPa (see Table 3). This is much higherthan the closure temperature of 681448C proposed forslowly cooling ophiolitic peridotites (Kamenetsky et al.,2001). The chromitite cooling rates may be estimated fromthe elemental profiles through the chromite-hosted inclu-sions (Fig. 5) using the diffusion models of Ganguly et al.(1994) and Be jina et al. (2009).The available data on Cr dif-fusion in magnesiochromite (Suzuki et al., 2008; and refer-ences therein) were used to estimate the cooling rate (s)


2432



Dow

nloaded from

from Cr concentration profiles using the following equa-tion (Ganguly et al., 1994; Be jina et al., 2009):

s 4DToRT2o =Ex2c 1where D(To) is the Cr diffusion coefficient in magnesio-chromite (cm2 s1) at To, xc 2(Dt)050564X, whereX is measured on the profiles of type 2 and 6 inclusions(Fig. 5a, c) and xc is the characteristic distance (mm) of dif-fusion, R is the ideal gas constant (83144 J mol1 K1),Tois the maximum temperature of crystallization (1273^1323K), and E is the activation energy (J mol1) of Crdiffusion in magnesiochromite (cm2 s1). In equation (1),xc is the only parameter that is extracted from the concen-tration profiles. On all the Cr concentration profiles ana-lysed, we found xc to be between 4 and 10 mm. Using thesevalues, equation (1) gives a rough estimate for s between0003 and 028C a1. The uncertainties of the estimationshave been discussed by Be jina et al. (2009). A more precisedetermination of s would require the use of a multicompo-nent diffusion approach, but diffusion coefficients of mostelements in chromites are lacking at present. Our calcu-lated values approach the highest values reported formid-ocean ridge crust (0018C a1, Schmitt et al., 2011).Moreover, our values are close to the highest values re-ported for the Oman lower oceanic crust (18C a1,VanTongeren et al., 2008). The high cooling rates inferredin our study strongly support the involvement of an aque-ous fluid that was cooler than the chromite-bearingsystem. The preservation of both the Maqsad chromititetextures (e.g. Leblanc & Ceuleneer, 1992) and the sharpzonation of the Maqsad chromite grains provides a furthersupport for rapid cooling in the MTZ. The fast coolingrates can be explained by active hydrothermal circulationat the Moho level of the oceanic lithosphere.

Moho Transition Zone chromititepetrogenesis: magmatic versussubsolidus modelTo unravel the complexity of the MTZ chromitite petro-genesis and the role of differentiation processes we discussbelow all the new data obtained in this study in the lightof existing models of chromitite origin in ophiolite massifs.

Magmatic versus subsolidusscenariosTwo main types of petrogenetic model for chromititepetrogenesis exist. The most popular magmatic modelsuggests massive chromite crystallization from chromite-saturated magmas (e.g. Auge , 1987). This and similarmodels require oversaturation in chromium in a hydroussilicate melt owing to a sharp temperature drop or oxida-tion of the magmatic system in minichambers (Lagoet al., 1982) or crustal assimilation by the olivine-saturatedmelt (Irvine, 1975). Entrapment of silicate melt intochromite is advocated by all these models (Lorand &

Ceuleneer, 1989; Leblanc & Ceuleneer, 1992; Schianoet al., 1997).We found two types of evidence for the possiblepresence of a melt in our system. (1) The host chromitecomposition is consistent with its crystallization from aMORB melt. (2) The type 4 and 5 plagioclase^pargasiteinclusions are in an assemblage with the host chromiteand were formed at temperatures within the stabilitydomain of hydrous MORB melts (950^10508C at 200MPa, Table 3). The inclusions were possibly entrapped inthe host chromite during magmatic crystallization (incase of water-saturated MORB melt at 200MPa, Feiget al., 2006). Far above 11008C, an efficient process ofre-equilibration of the high cr-number chromite with theresidual melt may occur by cationic diffusion (Scowenet al., 1991). However, the maximum temperature of10508C (Table 3) estimated here makes such a process forspinel^melt re-equilibration inefficient. In fact, the preser-vation of high cr-number chromite inclusions indicates ex-tremely sluggish rates of spinel dissolution (Brearley &Scarfe, 1986) and the lack of equilibration in a mafic meltowing to fast cooling and efficient fractional separationof chromite. The fast cooling inferred in this study helps toexplain the preservation of the chromite-hosted hydroussilicates (phlogopite, aspidolite, pargasite). However,the Maqsad chromite-hosted inclusions themselves provideevidence for subsolidus conditions rather than for amolten or partially molten system, which would supportthe second group of models discussed below.The second group of models suggests chromite crystal-

lization at subsolidus conditions based on the stabilityfield of the pargasite inclusions, which is below 10708C(e.g. Muntener et al., 2001). These models state that thepresence of sodium-rich hydrous phases indicates a highNa activity in the system and implies a temperature limitcorresponding to the maximum thermal stability of parga-site (Auge & Johan, 1988). Our investigation of theMaqsad chromitite inclusions provides several lines of evi-dence in favour of such subsolidus petrogenesis (below9508C). (1) In our case we have no evidence for a silicatemelt in the chromitite system, as the multiphase inclusionsare low-temperature silicates heterogeneously entrappedby the growing chromite rather than crystallized as daugh-ter phases from an entrapped melt. (2) Because coronasare typical textures of metamorphic crystallization andrarely for magmatic systems (Deer et al.,1997), the presenceof high cr-number chromite crystallized below 9508Cprobably indicates an early metamorphic stage of the chro-mite growth. (3) It helps explain the very good preserva-tion of an association of hydrous minerals such asaspidolite and phlogopite (which are not stable above7008C) and pargasite (which is not stable above 10508C).Such preservation of multiphase inclusions would indicatesubsolidus rather than magmatic conditions to avoid thecomplete destabilization and/or melting of entrapped in


2433



Dow

nloaded from

chromite aspidolite and pargasite phases. Figure 5e and fdemonstrates textural relationships between silicatephases, high cr-number and host chromites that are nottypical for magmatic systems. It can be seen that thehigh-Mg silicate and altered spinel phases were overgrownby high cr-number chromites similar to some hydrother-mal or metamorphic systems.

Three-stage model for the Maqsad MTZ chromititepetrogenesisSome of our data are consistent with the magmatic pro-cesses, whereas others support the subsolidus modelsdescribed above. We believe that both types of model canbe reconciled within a single scenario involving threestages: (1) retrograde metamorphism (serpentinization) ofa mantle protolith that was the primary source of chro-mium and water; (2) prograde metamorphism of this

serpentinized protolith (or serpentinite dehydration); (3)magmatic assimilation by MORB (Table 3; Figs 7 and 8).Although the first stage is generally poorly preserved, it

is evidenced by rare inclusions of serpentine and alteredspinel in stratiform and disseminated chromitites (Fig. 5f;Tables 1 and 3), which are typical of serpentinized perido-tites formed by hydration of the upper mantle at 6008C(e.g. Li & Lee, 2006). Moreover, serpentinites and serpen-tinized harzburgites are enriched in chromium with con-centrations higher than 3000 ppm Cr (Li & Lee, 2006)and, therefore, may constitute an abundant chromiumsource. Thus, deep harzburgites serpentinized as a resultof interaction with seawater-derived hydrothermal fluidsmay be a potential source for the MTZ chromitites.The second stage is prograde metamorphic recrystal-

lization of this hydrous mantle protolith as recorded bythe multiphase type 1, 2 and 2a and microcrystal type 3

Fig. 7. Phase diagram model of Irvine (1975) applied to the petrogenetic model of the Maqsad MTZ chromitites formed through assimilation ofa metamorphic mantle protolith by olivine-saturated MORB and fractional crystallization.The assimilation (b^c) is shown as a distinct processfrom the crystallization (a^b, c^d, and d^e). In reality, however, chromite could crystallize while the metamorphic protolith was being continu-ously assimilated. The composition of the serpentinized harzburgites corresponds to those of Li & Lee (2006).


2434



Dow

nloaded from

and 6 inclusions (Table 3). High-Mg and -Cr silicate min-erals in multiphase inclusions, pyroxenes and forsteritemicrocrystals in equilibrium with high cr-number chro-mite may be formed during this stage as a result of dehy-dration of the high-Mg and -Cr serpentinized protolithupon prograde metamorphism above 6008C. For example,the forsterite microcrystals may represent the dehydratedserpentine assemblage (e.g. Khedr & Arai, 2012). TheMaqsad diapir (Fig. 8) channelled a large amount ofrising MORB magma that produces the heat source in thecontact aureole and the wall-rock serpentine dehydrationduring the second stage metamorphism. In addition tothe mineral composition and inclusion data discussedabove, our analyses of the oxygen isotope composition ofthe chromite ore bodies also support this scenario. Thed18O values measured in the chromites (32^38) areconsistent with the involvement of hydrothermal fluidsderived either from seawater directly or from prograde de-hydration of the protolith, which was originally altered byseawater-derived hydrothermal fluids during the initialstage (e.g. Alt et al., 1996; Fruh-Green et al., 2003). Indeed,calculation of the isotope composition of the aqueous fluidusing the equilibrium 18O values for the fractionation be-tween chromite and water at 10008C (Zheng, 1991) andthe measured d18O of chromite (32^38) yields a

d18OH2O value of 79^85 for the aqueous fluid, typicalfor Cretaceous seawater-derived hydrothermal fluids asso-ciated with the Oman ophiolites (Gregory & Taylor, 1981).The third stage, illustrated in Figs 7 and 8, occurred in

the presence of a hydrous hybrid MORB melt as recordedby crystallization of the host chromite and late anhedraltype 4 and 5 inclusions. The petrogenesis of the MTZchromitites and host dunites during this stage is in accord-ance with Irvines (1975) model applied to illustrate ourmodel of assimilation of the metamorphic protolith into aMORB melt (Fig. 7). The resultant hydrous, hybridMORB melt (enriched in Cr and H2O) becames oversatu-rated in chromite, leading to abundant fractional crystal-lization of chromite in the MTZ. At this stage, thechromite grains could crystallize while the metamorphicprotolith was being continuously assimilated. Cooling wasrapid owing to the involvement of hydrothermal fluidsand fractional crystallization was efficient. The observedtextures of the MTZ chromitites suggest a melt- orfluid-saturated medium during the third stage and cumu-lative, possibly, convective control on the distribution ofthe minerals inside the crystallizing dykes (see Leblanc &Ceuleneer, 1992). Once formed, the host chromite crystalsmay be accumulated gravitationally in an ore body owingto their high density (4500 kg m3; e.g. Mungall &

Fig. 8. Three-stage model of chromitite formation through assimilation of a metamorphic mantle protolith by MORB. Stages correspond tothose inTable 3. Stage I is not shown in the figure. Stage II: during the uprise of an asthenospheric diapir, the overlying lithosphere is heatedand serpentinized peridotite experiences prograde metamorphism. Reduced hydrothermal fluids (probably rich in Cr2) and the high Mg^Cr silicate minerals of the multiphase and microcrystal inclusions are generated in the thermal field of mantle upwelling above 6008C. StageIII: the metamorphic phases generated during stage II are assimilated by primitive MORB melts (initially poor in Cr and H2O, and oxidized).The resultant hydrous hybrid MORB melt (enriched in Mg, Cr and H2O) becomes oversaturated in chromite, leading to abundant chromiteand chromite^olivine fractional crystallization in the MTZ.The process is responsible for the petrogenesis of the MTZ chromitites and dunites.The cooling was rapid owing to hydrothermal fluid circulation. After Benoit et al. (1999), Bosch et al. (2004) and Python et al. (2007).


2435



Dow

nloaded from

Naldrett, 2008) and may be sintered (Hulbert & vonGruenewaldt, 1985). The latter process may explain the ab-sence of entrapped basaltic glass in the host chromite. Ouroxygen isotope data also provide independent support forthe third magmatic stage. The measured d18O values inchromite (32^38) in the Maqsad chromitites (Table 4)and associated olivine (52^58 in the Maqsad dunites)yield an average chromiteôlivine oxygen isotopic fraction-ation (18O) value of 20. This small fractionation isconsistent with equilibrium isotope fractionation at hightemperatures of 10508C, calculated using a combinationof available d18O fractionation in the pairs chromite^water (Zheng, 1991), basalt^water and olivine^basalt(Zhao & Zheng, 2003). If the total range of 18O (oliv-ine^spinel) 14^26 that we observe between chromi-tites and associated dunite is considered, this wouldcorrespond to temperatures higher that 6008C. This indi-cates that although the involved protolith had a highd18O, perhaps because of alteration by seawater duringthe first stage at lower temperature, the later progrademetamorphism and magmatic stages variably equilibratedolivine and chromite with respect to their d18O values.The proposed model for MTZ chromitite genesis ex-

tends our knowledge of physicochemical processes occur-ring during MORB transport at Moho-level depths. Themetamorphic mantle protolith would supply H2O, S, Cland Na to the primitive MORB melts (Be dard et al., 2000;and references therein). The saturation of the MORBmelts with a CÔ^H^Cl^S fluid would have played a keyrole in the MTZ chromitite formation. The presence of adeep metamorphic source and its impact on the MORBmelt chemistry is thus consistent with the growing evi-dence for (1) oceanic lithosphere assimilation by MORB,and the petrogenesis of evolved lithospheric magmas inthe mid-ocean ridges (e.g. Nonnotte et al., 2005;Kamenetsky & Gurenko, 2007) and in geodynamicallysimilar regions such as Iceland (Bindeman et al., 2012), (2)abundance of MORB enriched in CÔ^H fluids (e.g.Javoy & Pineau, 1991), and (3) ubiquity of chromitite oresfound in the Moho Transition Zone of ophiolite massifsover the world.

CONCLUSIONSThe results of systematic analyses, carried out in this study,of chromite-hosted inclusions as well as oxygen isotopes ofchromite grain fractions from nodular, disseminated andstratiform chromitites and associated host dunites that arelocated at different levels of the MTZ in the Oman ophio-lite (Maqsad area) are consistent with a three-stagemodel of the chromitite petrogenesis in which most ofthe Maqsad MTZ chromitites crystallized at temperaturesbetween 6008C and 10508C and pressures of 200MPa.In the first stage, a mantle protolith altered byseawater-derived hydrothermal fluids yielded serpentinites

and serpentinized harzburgites, which were probably themain source of chromium for chromitite ore at the Moholevel and explain the systematically high Mg/Fe ratios ofthe silicate minerals found in the ophiolite chromitites.The presence of such hydrous mantle protolith thus elimin-ates the existing controversy on the source of Cr and H2Oand its role in the chromitite petrogenesis. The secondstage involved subsolidus (metamorphic) crystallization at600^9508C. It was followed by a third stage of assimilationand fractional crystallization of chromite from water-saturated MORB at 950^10508C in the oceanic ridge set-ting. The assimilation of this protolith at Moho-leveldepths dramatically affected the MORB magma chemistryand favored Cr concentration into economic ores. Ourmodel is supported by the growing evidence for oceaniclithosphere assimilation by MORB, abundance of hydrousMORB enriched in CÔ^H fluids and ubiquity of chromi-tite ore deposits in the Moho Transition Zone observed inophiolite massifs world-wide.

ACKNOWLEDGEMENTSWe dedicate this work to the memory of Leonid V.Dmitriev, who was a great personality and scientist prob-ing the ocean depths in numerous Russian Academy ofScience, Deep Sea Drilling Program (DSDP) and OceanDrilling Program cruises. We thank editors G. Woernerand M.Wilson, and H. Rollinson, J. Natland, R. Almeev,J. Be dard, F. Melcher, F. Costa and several anonymous ref-erees for editorial corrections and very constructive re-views. M. Gre goire, K. Kouzmanov, C. Li and M. Pythonare acknowledged for their help during preparation of themanuscript. Ch. Cavare-Hester is thanked by A.Y.B. forgreat help in figure design. We are grateful to S. Matveevfor providing us with unpublished data. We thank J. F.Mena for sample preparation. V. Yapaskurt is thanked forX-ray imaging (Moscow State University). C. Ballhaus, B.Bazylev, M. Benoit, A. Bychkov, A. Delacour, F. Faure, S.Feig, D. Jousselin, J. Koepke, G. Nikolaev, M. andD. Ohnenstetter, M. Pichavant, M. Portnyagin,C. Ramboz, L. Reisberg, O. Safonov, B. Scaillet, P.Schiano, K. Schmulovitch, M. Toplis, L. Truche and M.Yudovskaya are thanked for insightful discussions. We areextremely grateful to J. Prunier for assistance during theleaching procedure at GET (France), and to J. Palandrifor oxygen isotope measurements at the University ofOregon (USA).

FUNDINGThis work was supported by a CNRS grant Poste Rougeto A.Y.B. and French^Japanese linkage grant PHC-Sakura21225NG. The work was supported by grants fromRFBR (10-05-00254 -, Russia, 2010^2012), INSU(CHROMAN, France, 2010^2011) and OMP(Observatoire Midi-Pyrenees, France, 2011^2012).V.K. was


2436



Dow

nloaded from

supported by the Australian Research Council, theUniversity of Tasmania and the Russian Ministry ofScience and Education (grant 2012-1.5-12-000-1006-001).

SUPPLEMENTARY DATASupplementary data for this paper are available at Journalof Petrology online.

REFERENCESAbily, B., Ceuleneer, G. & Launeau, P. (2011). Synmagmatic normal

faulting in the lower crust: Evidence from the Oman ophiolite.Geology 39, 391^394.

Abily, B. & Ceuleneer, G. (in press). The dunitic mantle/crust transi-tion zone in the Oman ophiolite: Residue of melt rock interaction,cumulates from high-MgO melts or both? Geology, in press.

Adam, J. & Green, T. H. (1994). The effect of pressure and tempera-ture on the partitioning of Ti, Sr and REE between amphibole,clinopyroxene and basanitic melts. Chemical Geology 117, 219^233.

Ahmed, A. H., Hahghj, K., Kelemen, P. B., Hart, S. R. & Arai, S.(2006). Osmium isotope systematics of the Proterozoic andPhanerozoic ophiolitic chromitites: In situ ion probe analysis of pri-mary Os-rich PGM. Earth and Planetary Science Letters 245, 777^791.

Alt, J. C., Laverne, C., Vanko, D. A., Tartarotti, P., Teagle, D. A. H.,Bach, W., Zuleger, E., Erzinger, J., Honnorez, J., Pezard, P. A.,Becker, K., Salisbury, M. H. & Wilkens, R. H. (1996).Hydrothermal alteratio

Documents

Borisova_etal_J. Petrology-2012.pdf