A LA–ICP–MS EVALUATION OF Zr RESERVOIRS IN COMMON …grupo179/pdf/Bea2006.pdf · Hf in...

693

§ E-mail address:[email protected]

The Canadian MineralogistVol.44,pp.693-714(2006)

A LA–ICP–MS EVALUATION OF Zr RESERVOIRS IN COMMON CRUSTAL ROCKS: IMPLICATIONS FOR Zr AND Hf GEOCHEMISTRY,

AND ZIRCON-FORMING PROCESSES

FernandoBea§,PilarMonTeroandMiguelorTega

Department of Mineralogy and Petrology, Campus Fuentenueva, University of Granada, E–18002 Granada, Spain

aBsTracT

Theresultsof~4000LA–ICP–MSanalysesin152thinsectionsfromcommoncrustalrocksrevealthatseveralrock-formingmineralscontaintenstoafewthousandppmofZr.ThehighestconcentrationsofZrarefoundinxenotime,followedbytitanite,ilmenite,rutile,allanite,amphibole,clinopyroxene,garnet,magnetiteand,lesscommonly,plagioclase,K-feldsparandorthopy-roxene.Olivine,cordierite,biotite,muscovite,apatite,epidoteandmonazitehavelowlevelsofZr(<5ppm,generally<1ppm).ThemineralswiththehighestKD

Hf/KDZraretitanite(2.5),orthopyroxene(2.0),amphiboleandclinopyroxene(1.8),andepidote

andrutile(1.6–1.7).Ilmenite,magnetite,thefeldsparsandapatitehaveKDHf/KD

Zr≈1,andvalueslessthanonewerefoundinxenotimeandzircon(0.8),garnet(0.7),andallanite(0.6).Themostimportantimplicationsoftheseresultsfollow.First, thegrowth of a Zr-bearing phase during partial melting does not influence the Zr concentration of the melt, but increases the fraction ofzirconthatcanbedissolvedatagiventemperature.Thisacceleratesthedisappearanceofzirconfromtheprotolithor,inmeltsalreadysegregated,thedissolutionofinheritedzircon.Theeffectwillbemoremarkedinmetaluminousmagmasprecipitatingamphibole and titanite than in any other type of magma. Second, the presence of Zr-bearing phases has little influence of the zircon-saturation“geothermometer”.Itmaycausesomewhathigher(20–30°C)resultsinmetaluminousrocks,buthasnoeffectonperaluminousrocks.Third,theuptakeofZrbymajormineralsincrystallizingmagmasmayreduceboththeconcentrationofZrinthemeltavailabletoformzirconandthetemperatureatwhichzirconbeginstoprecipitate.Mineral–meltreactionsinvolvingZr-bearingphasesmayleadtozircongrainswithcomplicatedpatternsofzoningandtexturallydiscordantzones,apparentlydiachronous.Fourth,higher-than-chondriteZr/Hffractionatesmayarisefromtitanite,amphiboleorclinopyroxenefractionation,but this requires very little or no crystallization of zircon. Significantly lower-than-chondrite Zr/Hf magmas only result from zirconfractionation.Lastly,twonewexamplesofmineralreactionsthatinvolvetheformationofamass-balancingaccessoryphase,usefulforhigh-resolutiongeochronology,aredescribed:theformationofxenotimeasaproductofthebreakdownofgarnetinamphibolite-grademetapelites,andthesubsolidusgrowthofanewrimonzirconincludedinZr-bearingfeldspars.

Keywords:zircon,xenotime,mineralcomposition,saturationtemperature,Zr,Zr/Hf,geochronology,geochemistry.

soMMaire

Lesrésultatsd’environ4000analyseseffectuéesparlatechniqueLA–ICP–MSsur152lamesmincesderochesbanalesdelacroûtemontrentqueplusieursdesminérauxformantcesrochescontiennentdesdizaines,voiredesmilliersdeppmdeZr.LesconcentrationslesplusélevéesdeZrsontdanslexénotime,suividelatitanite,l’ilménite,lerutile,l’allanite,l’amphibole,le clinopyroxène, le grenat, la magnétite et, moins couramment, le plagioclase, le feldspath potassique et l’orthopyroxène.L’olivine,lacordiérite,labiotite,lamuscovite,l’apatite,l’épidoteetlamonazitepossèdentdefaiblesteneursenZr(<5ppm,etengénéral,<1ppm).LesminérauxayantlesvaleurslesplusélevéesdeKD

Hf/KDZrsonttitanite(2.5),orthopyroxène(2.0),

amphiboleetclinopyroxène(1.8),etépidoteetrutile(1.6–1.7).L’ilménite,lamagnétite, lesfeldspathsetl’apatitepossèdentunevaleurKD

Hf/KDZrvoisinede1,etdesvaleursinférieuressonttypiquesduxénotimeetduzircon(0.8),dugrenat(0.7),et

del’allanite(0.6).Lesimplicationslesplusimportantesdecesrésultatssuivent.1)Lacroissanced’unephaseporteusedeZrau cours de l’anatexie n’influence pas la concentration de Zr dans la fraction liquide, mais augmente la fraction de zircon qui peutdissoudreàunetempératuredonnée.Ceciapoureffetd’accélérerladisparitionduzirconduprotolitheou,dansunliquidedéjàséparé,ladissolutionduzirconhérité.Ceteffetseraplusmarquédanslesmagmasmétalumineuxprécipitantl’amphiboleet la titanite que dans tout autre type de magma. 2) La présence de phases porteuses de Zr exerce très peu d’influence sur le “géothermomètre”fondésurlasaturationenzircon.Ellepeutêtrelacausedetempératureslégèrementplusélevées(20–30°C)danslesrochesmétalumineuses,maisn’aaucuneffetdanslecasdesrocheshyperalumineuses.3)L’incorporationdeZrdanslesprincipauxminérauxcristallisantdansunmagmapeutréduireàlafoislaconcentrationdeZrpourformerlezircondansleliquideetlatempératureàlaquellelezirconcommenceàcristalliser.LesréactionsentreminéraletliquideimpliquantlesphasesporteusesdeZrpourraientmeneràdesgrainsdezirconayantdestracéscompliquésdezonationetàdeszonesdiscordantes,

694 ThecanadianMineralogisT

apparemmentdiachrones.4)DesvaleursdeZr/Hfplusélevéesquelesvaleurschondritiquespeuventrésulterd’unfractionnementdetitanite,d’amphiboleoudeclinopyroxène,maiscecirequierttrèspeu(ouaucune)cristallisationdezircon.LesmagmasayantunevaleurZr/Hfplusfaiblequelavaleurchondritiquenepeuventrésulterquedufractionnementduzircon.5)Nousdécrivonsdeuxnouveauxexemplesderéactionsminéralesimpliquantlaformationd’unephaseaccessoireselonlebilandesmasses,cequis’avèretrèsutiledansledomainedelagéochronologieàrésolutionélevée:laformationduxénotimecommeproduitdedéstabilisationdugrenatdanslesmétapélitesaufacièsamphibolite,etlasurcroissancesubsolidussurlesgrainsdezirconinclusdansunfeldspathporteurdeZr.

(TraduitparlaRédaction)

Mots-clés:zircon,xénotime,compositiondesminéraux,températuredesaturation,Zr,Zr/Hf,géochronologie,géochimie.

– [Ca,Th,Ce]Zr[Ti,Nb]2O7, restricted to nepheline-normativerocks).But thesemineralsarenot theonlyrepositoriesofZr;severalsilicatesandoxideshavealsobeen found to contain small but significant concentra-tions of this element (e.g., Pearce 1990, Fraser et al.1997,Vannucciet al. 1998,Seifert&Kramer2003).Upon breakdown, some of them may even releasesufficient Zr to form new crystals of zircon (Bingen et al.2001,Degelinget al. 2001).

The partitioning of Zr and Hf in rock-formingminerals is not yetwell known,despite theprofoundinfluence it may have on the geochemistry of these elements and zircon-forming processes. Historically,this is attributable to analytical problems relatedwith the difficulties in avoiding zircon (or xenotime, baddeleyite, etc.) inclusions in mineral-concentrateanalysis, the lackofsensitivityof theelectronmicro-probe, and the limited availability of ion-microprobeinstruments. Now, however, these problems may beovercome with the laser-ablation inductively coupledplasma–massspectrometry(LA–ICP–MS)technique,because it can easily perform rapid, sensitive andprecisespot-analysesofZrandHfinmineralsofrockthin-sectionsatamoderatecost.

saMPlesandMeThods

Oursamplecollectioncomprisesatotalof152thinsectionsofrepresentativeexamplesofcommonigneousandmetamorphic rocks.These include:peraluminousS-type granitic rocks (32 sections), appinitic ultramafic and mafic rocks (7), amphibolite-grade metapelites (5) andmigmatites(13),andalkaline(camptonitic)lampro-phyres (6) from central Spain; calc-alkaline graniticrocks (32), ophiolitic gabbros and ultramafic rocks(12),amphibolites(3)andeclogites(4)fromtheUrals,Russia; amphibolite- to granulite-grade metapelites(14), mafic granulites (4), and peridotites (5) fromthe Ivrea–Verbano zone, northwestern Italy; alkalinegabbros(3),migmatites(3)andcalc-alkalinegneisses(3)fromKola,Russia;andamphibolites(3)andeclo-gites(3)fromtheBeticCordilleras,southeasternSpain.Most of these rocks are of upper Carboniferous tolowerPermianage;exceptionsarethemiddleProtero-zoicmigmatitesandgneissesofKola,andtheAlpine

inTroducTion

The importance of the geochemistry of zirconiumandhafnium,thecompanionelementofZr,invariablycamouflaged in Zr minerals owing to its lower abun-dance and almost identical ionic radius, stems fromtwofacts.First,Zrisanessentialstructuralcomponentofzircon(ZrSiO4),aubiquitousaccessoryphasewithoutstanding petrogenetic and geochronological appli-cations(Hanchar&Hoskin2003).Second,ZrandHfconcentrations and the ratio Zr/Hf find application as indicators of magmatic and metallogenetic processes(e.g., Linnen 1998, David et al. 2000, Kamber &Collerson2000).

BackgroundinForMaTion

In thispaper,wepresent themost relevant resultsof a survey to determine the concentration of Zr andHfinrock-formingmineralsofcommoncrustalrocks.We first summarize results of about 4000 analysesfrom152thinsectionsofawidevarietyofigneousandmetamorphicrocksfromIberia,theUrals,theAlpsandtheKolaPeninsula,andofabout500analysesofzircongrainsfrommineralconcentratesfromarepresentativeselection of these samples.We then discuss how thepresenceofZr-bearingmineralsmayaffectthezirconsaturationinmagmasand,therefore,zirconinheritanceandcrystallization,and theevolutionofZr/Hfduringmagmaticdifferentiation.ThepotentialofbreakdownreactionsofZr-bearingmineralstoproducenewzircon(oranyotheraccessoryphase)usefulforgeochronologyisalsoconsidered.

There are 99 known Zr minerals: 73 silicates, 14oxides,sixphosphates,threeboratesandthreecarbon-ates(seehttp://www.MinDat.org/index.php),butmostof them are restricted to peralkaline silica-undersatu-ratedand-oversaturatedrocks,whicharevolumetricallyinsignificant at a crustal scale. In the great majority of commoncrustalrocks,frommetapelites,gneissesandgraniticrocktoeclogites,ophiolitesandalkalibasalts,the only primary Zr minerals are zircon (by far themostabundant),baddeleyite(ZrO2,restrictedtoolivine-normativerocksincludingmembersofcalc-alkalineandtholeiitic suites) and zirconolite–zirkelite (CaZrTi2O7

zrreservoirsincoMMoncrusTalrocks 695

rocksoftheBeticCordilleras.Peralkalinerocks,bothsilica-saturatedand-undersaturated,weredeliberatelyexcluded because they play a minor role in crustalevolutionandhaveadistinct,andhighlycomplicated,Zrgeochemistrycontrolledbytheformationofalkalizirconosilicate complexes. Such rocks will form thebasis of a future study. Accordingly, the optimum fields for application of our data are subduction- and colli-sion-related magmatism, and medium- to high-grademetamorphismandanatexisofpelitesandgraywackes.A table with the localities, rock name, the nature ofthe Zr-saturated phase, and the summary statisticsforthemineralsanalyzedineachsection,isavailablefrom the authors as a SYLK file, and was submitted to theDepositoryofUnpublishedData,CISTI,NationalResearch Council of Canada, Ottawa, Ontario K1A0S2,Canada.

Laser-ablationanalysesweredoneonsections60–70mm thick; this thickness was such that the laser didnotpiercethesample(under theconditionsdescribedbelow), but still permitted good observations underthe optical microscope. Sections were studied first by optical methods and back-scattered-electron imagesacquired with a scanning electron microscope (BSE–SEM) todetermine inclusion-freeareas,subsequentlyanalyzedbyEDAXformajorelements.

The laser-ablation inductively coupled plasma–massspectrometry (LA–ICP–MS)systemconsistedof a torch-shielded quadrupolarAgilent–7500 spec-trometer and a 213 nm Nd–YAG Mercantek laserunit.AblationwasdoneinaHeatmosphere.Thelaserbeam was set either at a 95-mm-sided square sectionoracircularsection120mmindiameter,dependingonZrconcentrations.Foranalysisofxenotime ingarnetaureoles,thelaserbeamwassetto20mm.Thelaser-repetitionratewas10Hz.Spotswerepre-ablatedfor60secondsusingalaseroutputenergyof50%.Ablationlasted60seconds,withalaseroutputenergyof75%,withthesamplestagemovingupward5mmevery20seconds to keep the laser focused and maximize theablation efficiency. Silicon or Ca (for apatite), Ti (for rutileandilmenite),Fe(formagnetite)orP(forxeno-timeandmonazite),previouslydeterminedbyEDAX,wereusedasinternalstandards.TheglassNIST–610,whichcontains415ppmZrand419ppmHf (Pearceet al.1997),wasusedasanexternal standard.EveryanalyticalsessionstartedandendedbyanalyzingNIST–610,whichwasalsomeasuredevery6 to8spots.Toimprovedetectionlimits,blanks(i.e., analyseswiththelaserenergysettozero)wererecordedbeforeeachspotand subtracted from the analytical signal. Data werereduced using home-made software (freely availablefromF.Bea)writteninSTATAprogramminglanguage(Statacorp.2005).Thissoftwarepermitsoutlierstobeidentified and discarded, blank subtraction, drift correc-tion,correctionforaninternalstandardandconversionto concentration units based on an external standard.The precision, calculated as the coefficient of variation

(100*standard deviation/average) on 10 replicates ofNIST–610measuredineverysession,wasabout±3%forZr andHf.Oneexperiment to estimate thepreci-sionat lowerlevelsofconcentration,runwithalaserbeam120mmindiameterontwoglassespreparedforthisstudywithlevelsofconcentrationof~1ppmand~0.1 ppm, yielded values of ±10% and ±17%.Withtheconditionsdescribed,thedetectionlimitforZrandHfwasabout0.05ppmfor95-mm-squaredspots,andabout0.01ppmforcircularspots120mmindiameter.These figures do not apply to rutile, ilmenite or magne-titeanalysesbecausetheinadequacyoftheNIST–610(aborosilicateglass)asanexternalstandardforFe–Tioxides,andthefactthateitherFeandTi,usedasinternalstandards,arepronetooxide-ionisobaricinterferencesintheICP–MSsystem.Theresultsforthesemineralscanbeaffectedbysystematicerrorsandsoshouldbeusedwithcaution.Topreventcontaminationresultingfrom minute inclusions of zircon, or other Zr-richminerals, those analyses that in 90Zr/29Si versus timeplotsyieldedspikeshigherthanthreetimesthesignal-ratiostandard-deviationwerediscarded.

ThelevelofZrinzirconwasdeterminedbyeitherelectron microprobe or EDAX–SEM using the SPI®AS2230–AB (ZrO2 79.93%, HfO2 1.78%, Y2O318.25%)asexternalstandard.TheZr:Hfratioinzircon,however,wasdirectlydeterminedbyLA–ICP–MSbymeasuring 91Zr/178Hf under the same conditions asbefore,usingahome-madeglassofeudialiticcomposi-tion(98200ppmZr,1910ppmHf,andZr/Hf=51.4)as an external standard.The precision estimated on10replicatesmeasuredineachrunwasusuallybetterthan2%.

zirconiuMaBundancesinrock-ForMingMinerals

Results of LA–ICP–MS analyses of minerals aresummarized inTable1.As the frequency-distributionpatternsofZrineachmineralarecommonlystronglyasymmetrical,themodeisalsoreportedinadditiontothemeanandthestandarddeviation.AsHfwasfoundtobenearorbelowdetectionlimitinmanyinstances,we report inTable 1 only the Zr/Hf mode for eachmineralineachrocktype.Inaddition,thevariabilityinZr/Hf is displayed by kernel-density plots (a refinement of ahistogramor frequencyplot) calculatedwith thegaussian kernel function (Statacorp. 2005).As Zr/Hfshows little or no dependence on the nature of thehost rock (Table1),onlyonekernel-densityplotwasrequiredforeachmineral.Themostimportantobserva-tionsareasfollows.

Olivine

The concentration of Zr in olivine from tholeiiticgabbrosandperidotitesinthisstudyisinvariablybelowthe detection limit (0.01 ppm). Olivine from alkaline


basalts, however, has detectable Zr and yielded anaverageof0.06ppm(Fig.1A).Thehighestconcentra-tionsofZr(mean1.5ppm)werefoundinolivinefromtheIberiancortlandites(appiniticamphibole–phlogopiteperidotites). The Zr:Hf ratio, difficult to determineowing to the low concentrations of Hf, seems to bearound42(Fig.1B),slightlyhigherthanthechondriticratioof37.1(accordingtoMcDonough&Sun1995).

Orthopyroxene

The concentration of Zr in orthopyroxene in thisstudy is highlyvariable (Fig. 2A).The lowest valueswerefoundinUralianharzburgites,whichhaveamodeof0.02ppmandameanof0.06ppm(Table1).Tholei-itic gabbros show an asymmetrical distribution, withthemodeat1.3ppmandtheaverageat3.2ppm.Asinolivine,theorthopyroxenefromtheIberiancortlanditesistherichestinZroftheigneoussamples,andshowsanearlysymmetricaldistribution,withmeanandmodeequal at 4.0 ppm. Orthopyroxene from mafic granulites haslowconcentrationsofZr,usuallyintherange<0.01to 1.5 ppm. Orthopyroxene from felsic granulites, incontrast, is Zr-rich (also reported by Degeling et al.2001), and yielded an asymmetrical distribution with


Fig.1. A)AverageconcentrationofZrinolivine,sortedbyrocktypes.Notethelowbutperceptibleconcentrationinalkalibasaltsandthehighervaluesinzirconinappinites.B)Kernel-densityplot(seetext)showingthedistributionofZr/Hfvaluesinolivine,whichshowsapeakat42;thechondriticratioisshownat37.1,accordingtoMcDo-nough&Sun(1995).

Fig. 2. A)Average concentration of Zr in orthopyroxene, sorted by rock types.Thehighestconcentrationsarefoundinfelsicgranulites.B)DistributionofZr/Hfvaluesinorthopyroxene.Themodeisaround19,notablylowerthanthechondriticratio.


themodeat21ppmand themeanat24ppm.ValuesofZr/HfinZr-bearingorthopyroxenearesystematicallylowerthanthechondriticratio,withthemostfrequentvalueplacedaround19(Fig.2B).

Clinopyroxene

Clinopyroxene crystals analyzed in this studycontaindetectableamountsofZr.Thelowestconcen-trations were found in clinopyroxene from tholeiiticgabbros, which yielded a symmetrical distributioncenteredat3ppmwithameanof2.9ppm(Fig.3A,Table1).Augite fromcalc-alkalinegabbrosyieldedasimilarpatternofdistribution,butcenteredat4.1ppm(mean=4.2ppm).Chromium-richdiopsidefrombothUralianpyroxenitesandAlpinelherzolitesfromIvreahassomewhathigherZrcontents,displayinganasym-metricaldistributionwiththemodeat6.5ppmandthemean at 5.1 ppm.The concentration of Zr increasesnotably in clinopyroxene of alkaline gabbros andlamprophyres(generallytitanianaugite),whichhasZrintherange8–110ppm,withthemodeat21ppmandthemeanat31ppm.ThedistributionofZrcontentsinomphaciteineclogiticrocksisasymmetricallytailedtothehighend,withamodeat1.3ppmandthemeanat8.9ppm.ThehighestconcentrationsofZrwerefoundin Mg-rich augite from the mafic granulites of the Ivrea Zone,withameanat137ppmandamodeat158ppm.ValuesofZr/Hfinclinopyroxenearenotablylowerthan

thechondritic ratio,andclusteraround21,nearly thesamevalueasinorthopyroxene(Fig.3B).

Amphibole

Among coexisting major minerals, amphibolecommonlyhasthehighestconcentrationofZr(Fig.4A,Table 1). In this study, the lowest values were foundin edenite of tholeiitic gabbros and diabases, whichshow a quasi-normal distribution centered around 8ppm(withameanof8.3ppmandamodeat7.2ppm).Edenite to pargasite from calc-alkaline granitic rockscommonlyhaveelevatedZrconcentrations,especiallyinthelesssilicicvarieties.Ingabbrosanddiorites,theseamphiboleshaveameanof154ppmandamodeof137ppm,whereas in tonalites,granodioritesandgranites,they yielded a mean of 41 ppm (mode = 53 ppm).KaersutiticamphibolefromappiniticcortlanditesisalsorichinZr,withameanof163ppmandamodeof153ppm.Thisamphiboleisrimmedbyalow-Caamphibole(Beaet al.1999)withmuchlowerZrcontents(mean= 17 ppm, mode = 10.5 ppm).The kaersutite fromcamptoniticlamprophyreshaslowerZrthaninappin-ites,andyieldedanasymmetricaldistributiontailedtothelowend,withamodeat95ppmandameanat67ppm.The arfvedsonitic amphibole of alkali gabbrosfromKolayieldedameanof187ppmandamodeof196ppm.Theamphibolefromperidotitesdisplaysanasymmetric distribution tailed to the low end, with a

Fig.3. A)AverageconcentrationofZrinclinopyroxene,sortedbyrocktypes.Thehighestvalues are found in mafic granulites. B) Distribution of Zr/Hf values in clinopyroxene. Itclustersaround20,thesamevalueasorthopyroxene.


modeof39ppmandameanof28ppm.TheamphibolefromamphiboliteshasnearlythesameconcentrationsandpatternsofdistributionofZrasthatfromtholeiiticgabbros. In mafic granulites, however, the Zr in amphi-bolemaybehigher,withvaluescommonlyabove100ppm(mode=107ppm,mean=75ppm).Bycontrast,theamphibolefromeclogiteshasthelowestZrcontents,yieldinganasymmetricaldistributiontailedtothehighend,withamodeof1.5ppmandameanof6.3ppm.TheZr/Hfvalue isalso lower than inchondrite,withthemostcommonvaluearound21,thesameasthetwopyroxenes(Fig.4B).

Biotite and muscovite

TheconcentrationofZrinthetwomicasinthisstudyis,inmostcases,lowerthan1ppmanddoesnotshowperceptible variations from one rock type to another.Bothshownotablyasymmetricaldistributionstailedtothehighend(Fig.5),withmodesat0.6ppm(biotite)and 0.9 ppm (muscovite), and equal mean values of1.8ppm.TheZr/Hfvalueinbothmicasisaround10,thoughthisvalueshouldbeusedcautiouslybecauseHfisgenerallybelowornearthedetectionlimit.

Garnet

For the samples in this study, garnet invariablycontains a few ppm of Zr (Fig. 6A, Table 1). Inmetasedimentary rocks, the lowest concentrations ofZr were found in garnet from low-grade metapelites,andthehighestappearingarnetfrommigmatitesandfelsic granulites, where most values are in the range20–80ppm,withameanof45ppmandamodeof33ppm.IngarnetinmetapeliticrocksoftheIvreaZone,theconcentrationofZrincreaseswiththemetamorphicgrade (Fig. 7), whereasY and the HREE decrease(Beaet al. 1997, Bea &Montero 1999).Villaseca et al.(2003)reportedgarnetgrainsfromfelsicgranulites(found as xenoliths in Iberian lamprophyres) with Zrconcentrationsashighas450ppm,butwewereunableto reproduce these results in the samematerials fromneighboring areas. Garnet of mafic granulites alsocontainelevatedZrcontents;theyyieldanasymmetricaldistribution,withthemodeat12ppmandthemeanat24ppm.Inmostcases,thegarnetinamphiboliteshaslowZr,commonlyintherange0.5–3ppm,butrarelyitmaycontainuptoseveraltensofppmZr.Thedistribu-tionishighlyasymmetrical,withthemodeat1.9ppm

Fig.4. A)AverageconcentrationofZrinamphibole,sortedbyrocktypes.Calc-alkaline(1)representamphibolefromgabbroanddiorite;calc-alkaline(2)representamphibolefromtonalitetogranite.B)DistributionofZr/Hfvaluesinamphibole.Notethatthemostcommonlyencounteredvalueinamphibole,Zr/Hf≈20,isthesameasinthetwopyroxenesandnotablylowerthanthechondriticvalue.


Fig. 5. Kernel-density plots of the Zr concentration in biotite and muscovite. In bothcases, themodeis lowerthan1ppm.Zirconiumconcentrationshigherthan10ppmmayberegardedasspurious,andprobablycausedbyinclusions.Hfisbelow,detectionlimit.

Fig.6. A)AverageconcentrationofZringarnet,sortedbyrocktypes.B)DistributionofZr/Hfvaluesingarnet.NotethatitisoneofthefewmineralswithZr/Hfhigherthanthevalueinchondrite.


and themean at 7.7ppm.This situation is paralleledby garnet in eclogites, but at slightly higher values,withamodeat2.1ppmandameanat12ppm.Lastly,almandinefromhighlyevolvedperaluminousgranitesalsoshowsanasymmetricaldistribution,withthemodeat3ppmandthemeanat8.3ppm.TheZr/Hfvalueingarnetishigherthanchondrite,andclustersaround51(Fig.6B).

Cordierite

Cordierite is one of the major minerals that ispoorestinZr.Inthisstudy,mostvaluesarebelow0.1ppm,withthemodeat0.03ppmandthemeanat0.29ppm(Table1).

Plagioclase

Plagioclase from magmatic rocks in this studyusually contains detectable, though highly variable,concentrations of Zr (Table 1, Fig. 8A). In tholeiiticgabbros, most values are close to zero, but there arealso a fewoutlierswithZr in the range1 to10ppmZr. Something similar is observed in calc-alkalinerocks(mean=4.6ppm,mode=0.4ppm),thoughthefrequencyandtheZrcontentsoftheoutliersincrease,sothatcrystalswith20to80ppmZrrepresentabout5%ofthetotal.Inperaluminousgraniticrocks,thefractionofplagioclasecrystalswithnear-zeroconcentrationsofZrdecreases,yieldingahighlyasymmetricaldistribu-

tionwiththemodeat1.1ppmbutthemeanat10ppm.Thistendencyisstillmoremarkedinplagioclasefrommigmatites (mode = 4.5 ppm), which commonly hasconcentrations higher than 20 ppm. Plagioclase fromappiniticrocks,whetherfromcortlandite,kentallenite,appiniteortonalite,istherichestinZr,anditsfrequencycurve is totally different from the other rock-types;here the curve is just slightly asymmetrical, tailed tothelowerend,withamodeat55ppmandameanat41 ppm. By contrast, metamorphic plagioclase hasnear-zeroZr;infelsicgranulites,wefoundafewgrainswithtenorsoppmZr(seealsoVillasecaet al.2003);in mafic granulites from the Ivrea Zone, the plagioclase commonly has Zr in the range 0–5 ppm.The valueofZr/Hf inplagioclasegrainshavingHf>0.1variesbetween 20 and 60, with the mode at ~40, which issimilartothechondriticvalue(Fig.8B).

K-feldspar

K-feldsparfromthegraniticandmigmatiticsamplesexaminedinthisstudyalsocontainsdetectableamountsofZr (Table1).That fromperaluminousgraniteshasan asymmetrical distribution tailed to the high end(Fig.9A)withamodeat3ppmandameanat10ppm.K-feldspar from calc-alkaline granitic rocks is richerin Zr, with equal mode and mean at about 11 ppm.ThehighestZrcontentsarefoundinK-feldsparfromgranodioritefromappiniticsuites,whichhasanearlysymmetrical distribution with a mode at 27 ppm and

Fig. 7. Kernel-density curve of Zr contents in garnet in metapelitic rocks from theKinzigiteFormationofIvrea–Verbano;Beaet al.(1997)andBea&Montero(1999)provideddescriptionsof thesamples.NotehowtheZrcontents increasewithmeta-morphicgrade.


Fig.8. A)AverageconcentrationofZrinplagioclase,sortedbyrocktypes.MagmaticplagioclaseusuallyhasdetectablethoughhighlyvariableamountsofZr.ComparemodeandmeanvaluesinTable1.B)DistributionofZr/HfvaluesinplagioclasecrystalswithHfabove0.1%.Notethatitcoincideswiththechondriticvalue.

Fig.9. A)AverageconcentrationofZrinK-feldspar,sortedbyrocktypes.Asinplagio-clase,magmaticK-feldsparusuallyhasdetectablethoughhighlyvariableamountsofZr.ComparemodeandmeanvaluesinTable1.B)DistributionofZr/HfvaluesinK-feldsparwithHfabove0.1%.Notethatitcoincideswiththechondriticvalue.


a mean at 26 ppm.The K-feldspar from migmatitesalso has an elevated Zr content, and yields a near-symmetricaldistributionwithamodeat13ppmandameanat19ppm.Lastly,mostK-feldspargrainsfrommetapelites and felsic granulites contain near-zeroZr,butoutlierswitha few tensofppmarecommon,especiallyingranulites.Asinplagioclase,Zr/HfinK-feldsparwithHf>0.1ppm isclose to thechondriticvalue(Fig.9B).

Titanite

Titaniteinthisstudycontainsbetweenafewhundredto a few thousand ppm Zr.The highest values werefound in grains from alkaline gabbro (mean = 2103ppm,mode=1550ppm)andcalc-alkalinegabbroanddiorite(mean=2023ppm,mode=1970ppm)(Table1,Fig.10A).Asobservedwithamphibole, titanite fromthemorefelsiccalc-alkalinerocks,tonalite,granodioriteandgranite,havelessZr,withameanof893ppmandamodeof610ppm.Remarkably,intheappiniticrocksofIberia,wheretexturalevidenceindicatesthattitaniteisalatemineral,probablyarisingbyreactionbetweenkaersutiteandliquid, theconcentrationofZr ismuchlower,withameanof33ppmandamodeof27ppm.ThevalueofZr/Hfintitaniteisconsistentlycloseto15(Fig. 10B), the lowest among rock-forming minerals,which reflects the compatibility of Hf in this mineral.

Apatite

The concentration of Zr in apatite is invariablyvery low (Table 1). The highest values are foundin peraluminous granites (Fig. 11A); it is probable,however, thatvalueshigherthan10ppmmightresultfrommicrometricinclusionsofxenotimeinapatite,assuggestedbythegoodcorrespondencebetweenZrandYspikes in the time-resolvedsignals shownby theseanalyses.Apatitefromcalc-alkaline(mean=2.5ppm,mode=1.1ppm)andalkalinerocks(mean=3.3,mode=1.3ppm) rarelyhaveZrhigher than2–3ppm.Thesameistrueofapatitefromleucosomesofmetapeliticmigmatites.TheZr/Hfvalueofapatiteclustersaround38(Fig.11B).

Epidote

The concentration of Zr in magmatic epidote incalc-alkalinegranitic rocks in this study is invariablylow,withameanof5.9ppmandamodeof4.9ppm(Table1).TheZr/Hfvalueclustersaround20.

Rutile

Thismineralcommonlycontainsmorethan100ppmZr(Table1;seealsoZacket al.2002).Inthisstudy,the

Fig.10. A)AverageconcentrationofZrintitanite,sortedbyrocktypes.Calc-alkaline(1)representstitanitefromgabbroanddiorite;calc-alkaline(2)representstitanitefromtonalite to granite.The low concentration in titanite from appinitic rocks might berelatedtothelate-magmaticoriginoftitaniteintheserocks.B)DistributionofZr/Hfvaluesintitanite.Notethatitisthelowestamongrock-formingminerals.


highestconcentrationswere found in felsicgranulites(mean = 234 ppm), and the lowest, in mafic granulites (mean = 123 ppm) and eclogites (mean = 139 ppm).TheZr/Hfvalueclustersaround23.

Ilmenite and magnetite

It was difficult to get reliable analyses of ilmenite andmagnetiteforZrandHfbecauseinadditiontotheanalyticalproblemsmentionedinthemethodssection,these minerals are almost invariably too fine grained.

IlmenitecanhavehighZrcontents.Pearce(1990),forexample,reportedconcentrationsofZrinilmenitefrommafic alkaline rocks as high as 3850 ppm. Bingen et al.(2001)showedhowilmeniteisthesourceofZrduringhigh-grade metamorphism of mafic rocks, and Jang & Naslund(2003)determinedanaverageof485ppmwithZr/Hf=36 in theSkaergaard intrusion.The analysesobtainedinthiswork(Table1,Fig.12A)indicatethatZr in ilmenite decreases dramatically with the silicacontentof the rock.Whereascalc-alkalinegabbroanddioriteyieldedilmenitewithameanof677ppm,relatedtonaliteandgranodioriteyieldedilmenitewith105ppm.Inperaluminousgranitesandmigmatites,ilmeniteseemstohavenotably lowerZr,with amean around28–30ppm.TheZr:Hfratioclustersaround37–40(Fig.12B)andis,therefore,similartothechondriticratio.

Magnetitegenerallycontains lessZr than ilmeniteandhasasimilarZr/Hf(Table1).Thatfromsamples

ofalkaligabbroyieldedanaverageof47ppm,whereascalc-alkalinegraniticrocksyielded7ppm.TheZr/Hfvalueclustersaround31.

Monazite

TheconcentrationofZrinmonaziteanalyzedinthisstudyis,inmostcases,below10ppm.Thedistributionis highly asymmetrical, with a mode at 0.8 ppm anda mean at 4.5 ppm (Table 1).As the highest valuesweremostlyfoundinmonazitefromxenotime-bearingrocks,wesuggestthattheymayresultfrominclusionsof thatmineral,so that theactualconcentrationofZrinmonaziteisprobablybelow1ppm.

Allanite

Primaryallaniteinthisstudycontainselevated,buthighlyvariable,Zr.Theaverageofourdeterminationsfromallrocktypesis135ppm(Table1),butvaluesupto400ppmarecommon.TheZr/Hfvalueishigherthanthechondriticvalueandclustersaround61.

Xenotime

Zirconandxenotimeareisotypic,sothateachmayaccommodate elevated concentrations of the othercation, thus forming intermediate zircon–xenotimephases(e.g.,Bea1996).Forthisreason,xenotime[in

Fig.11. A)AverageconcentrationofZrinapatite,sortedbyrocktypes.B)Distributionof Zr/Hf values in apatite.The most commonly encountered value is close to thechondriticvalue.


allcasesxenotime-(Y)]hasahighlyvariable levelofZr,withmostdeterminationsinthisstudybetween0.05wt.%and20wt.%.TheZr/Hfvalueisalmostidenticaltothatinzircon,clusteringaround45.

Zircon

As would be expected, the concentration of Zrin zircon is, in general, quite constant, close to thestoichiometric value of 49.76%.The average of 511determinationsonzircongrainsfromawidevarietyofigneous and metamorphic rocks is 49.07%; significantly lower values were found only in strongly metamictgrains,owingtothepresenceofH2O,andingrainswithanelevatedfractionofdissolvedhafnonorxenotime.Values of Zr/Hf in this study vary between 20 and70,with themostcommonlyencounteredvalueat47(Fig.13),which isnotablyhigher than the chondriticvalue. Values below 20 are mostly found in zirconfromhighlyfractionatedsegregations,especiallyfromperaluminousmagmas.

discussion:iMPlicaTionsoFzrParTiTioninginMajorMinerals

Influence on zircon saturation in melts

Intheirexperimentalmodelrelatingtheconcentra-tion of Zr in a melt equilibrated with zircon to themajor-elementcompositionandtemperatureofthemelt,

Fig.12. A)AverageconcentrationofZrinilmenite,sortedbyrocktypes.B)DistributionofZr/Hfvaluesinapatite.Themostcommonlyencounteredisclosetothechondriticvalue.

Fig. 13. Kernel-density plot of Zr/Hf in 511 zircon grainsfromawidevarietyofigneousrocksfromtheUralsandIberia.Notehowthemostcommonlyencounteredvalueisaround47,notablyhigherthanthechondriticvalue.


Watson&Harrison(1983)(hereafterWH)assumedthatZrisaccommodatedonlybyzircon.Tounderstandhowthe partitioning of this element in other phases mayaffecttheWHmodel,wewillconsiderthefollowing.

Duringpartialmeltingofazircon-bearingprotolith,theformationofaZr-bearingphasedoesnotnecessarilyhave to influence the Zr concentration in the melt. The meltprovidesavehicleforZrtransportfromzircontothenewlyformedphase,butitwillholdZrconcentra-tionatthezircon-saturationlevelpredictedbytheWHmodelaslongasenoughzirconisavailablefordissolu-tion.TheformationofaZr-bearingphasebeforemeltsegregationwillacceleratethedisappearanceofzirconfromtheprotolithand,oncethemeltissegregated,will

enhancethedissolutionofinheritedzircon.Tokeepthemelt at zircon-saturation level, the precipitation of 1wt.%ofgarnet(50ppmZr),amphibole(150ppmZr)ortitanite(2000ppmZr)wouldrequirethedissolutionof0.0001wt.%,0.0003wt.%and0.004wt.%ofzirconrespectively, figures that gain relevance if one considers thatthemodalfractionofzirconinmostcrustalrocksisrarelyhigherthan0.05–0.07%(~250–350ppmZr).SincethemajormineralsthataremostenrichedinZrareamphiboleand titanite, thiseffectwould result inmetaluminousmagmashaving less zircon inheritancethan peraluminous magmas if they were derivedfrom protoliths with equal zircon contents at similartemperatures.

The temperatureobtainedwith theWHmodel forthe specific composition of a given rock (assuming it matches the composition of the liquid, which rarelyhappens) is known as the zircon-saturation tempera-ture(ZST)forthatrock.ThoughtheZSTisnotarealgeothermometer,whenappliedtograniticrocks,itmaybe regarded as a useful estimate of the temperatureof themagmaat the source (e.g.,Hanchar&Watson2003, Miller et al. 2003). How the presence of Zr-bearingphasesotherthanzirconaffectsthecalculationof the ZST will depend on the Zr concentration andmodalfractionofthesephases.LookingatTable1,itisevident that theZr-bearingphasesofperaluminousrocks, namely garnet and perhaps the feldspars, willhavenegligibleeffects.Inmetaluminousrocks,ontheother hand, the crystallization of titanite and amphi-bole may result in slightly elevated zircon-saturationtemperatures. For example, the precipitation of 10%amphibole(150ppmZr)and1%titanite(2000ppmZr)could increase theconcentrationofZr in the rockby~50ppmabovethezircon-saturationlevel,butthisonlyrepresentsanincrementoftheZSTof~20°C.

Notably, the uptake of Zr by crystallizing majorminerals may have a great influence on the morphology and modal abundance of zircon formed duringmagmatic crystallization. To illustrate this, let usconsiderthefollowingexampleofcrystallizationofametaluminousgranodioriticmagma(Fig.14)modeledwith the MelTs software (Ghiorso & Sack 1995,Asimow & Ghiorso 1998).The input for MelTs wasas follows: (1) Magma composition: SiO2= 61.34%,TiO2=1.09%,Al2O3=16.13%,Fe2O3=0.9%,FeO=3.88%,MgO=1.93%,CaO=4.16%,Na2O=3.46%,K2O=2.67%,H2O=4.55%;thiscorrespondstotheinitialmagmainferredfortheStepninskPlutonoftheUralides(Beaet al.2005),forwhichtheZrpartitioningamongmajormineralswasdeterminedinthiswork.(2)Temperature:from945°C(liquidus)downto655°Cinintervals of 10°C. (3) Pressure: 5 kbar. (4)The f(O2)constraint: Qtz–Fa–Mgt; (5) Solid phases allowed toprecipitate: titanite, amphibole, hornblende, biotite,muscovite,feldspar,quartz,“ortho-oxide”,andapatite(correspondingtothemodalmineralogyofStepninsk).

Fig. 14. Simulation of the evolution of the concentrationof Zr in a metaluminous granodioritic melt.The modalcompositionoftheprecipitateandthemajor-elementcom-positionoftheresidualliquidwithdecreasingtemperaturewere calculated with the software MELTS (Asimow &Ghiorso1998,Ghiorsoet al.2002);seethetextfordetails.Thezircon-saturationcurvewascalculatedfromtheabovedata and theWatson & Harrison (1983) model.KD cur-ves were calculated with the trace-element fractionationequation for fractional crystallization and the indicatedbulk-partition coefficients (KD); the curveKD=0 repre-sents theabsenceofZr-bearingmajorphases.ThecurveA represents the evolution of the bulk partition-coefficient calculatedfromthemodalcompositionoftheprecipitate,asprovidedbyMelTs,andthefollowingZrmineral/meltpartition coefficients: KD

amp/melt = 0.5, KDcpx/melt = 0.25,

KDttn/melt=10,KD

grt/melt=0.15,estimatedbyassumingaKD

amp/melt = 0.5, the central value of the amphibole rim–meltconcentrationratiosdeterminedbySisson(1994)oncalc-alkalinecompositions,andtheaverageZrconcentra-tionratiosbetweencoexistingamphibole,clinopyroxene,andtitaniteoftheMainSeriesofStepninskpluton(Beaet al.2005),asdeterminedinthiswork.


ThedetailedMelTsoutputisavailablefromtheauthorsuponrequest.

ApplyingtheWHexpressiontotheliquidcompo-sitionforeachcrystallization interval,asprovidedbyMelTs, the zircon-saturation curve was first calculated. Then, assuming 100 ppm as the initial Zr concentra-tionof themelt, andusing theMelTs output and thetrace-elementfractionationequationforRayleighcrys-tallization, the evolution of the Zr concentration wascomputed in fractionated liquids as a function of thetemperature.ThesecalculationsweredoneforZrbulksolid–melt partition coefficients (KD)equalto0(noZr-bearingphases),0.5,1and1.5,andfortheKDresultingfromthemodalfractionofmineralsintheprecipitate,providedbyMelTs,andthefollowingZrmineral–meltpartition coefficients: KD

amp/melt=0.5,KDcpx/melt=0.25,

KDttn/melt=10,KD

grt/melt=0.15.ThesewereestimatedbyassumingaKD

amp/melt=0.5,thecentralvalueoftheamphibole rim /melt concentration ratiosdeterminedby Sisson (1994) on calc-alkaline compositions, andthe averageZr concentration ratios amongcoexistingamphibole, clinopyroxene, and titanite of the MainSeries of Stepninsk pluton, as determined in thiswork.

Figure14showstheresultsofthissimulation;itisevidentthatthecrystallizationoftheZr-bearingphasesin the proportions foreseen by MelTs (curveA) willcause the liquid tosaturate inzirconata temperature30°ClowerthanwhentheprecipitateisformedbyZr-freeminerals(KD = 0) and, significantly, the amount of Zravailableforzirconwillbeabout33%less.Besidesthe obvious effect of decreasing the modal fractionof this mineral, this effect would probably lead to azircon population composed of smaller grains owingto the well-known effects of decreasing temperatureon nucleation and growing rates (Kirkpatrick 1981).

An interesting situation, represented by curve B inFigure14, occurs where early-formed Zr-bearingminerals react with late liquids to produce Zr-freephases, for example the transformation of amphiboleto biotite and the resorption of early titanite.The Zraccommodated in these minerals is released to themelt, which may then become locally oversaturated.Depending on the environment, this may lead eithertoovergrowthsor toanewgenerationofzircon.Theopposite situation will occur where a Zr-poor phase,suchasolivineororthopyroxene,reactswiththemelttoformamphibole.Ifzirconwasalreadyprecipitating,thepartitioningofZrintothenewlyformedamphibolewilldecrease theconcentrationofZr tosubsaturationand so may lead to the resorption of zircon crystals.Thecomplex interplayamongmelt,Zr-bearingmajorsilicatesandzirconmaybeoneofthemaincausesofthe appearance of texturally discordant zones, appar-entlydiachronous,commonlyfoundinmagmaticzircon(e.g.,Corfuet al.2003).

Applications to the ratio Zr/Hf during petrogenetic processes

The KDHf/KD

Zr value for each mineral can beestimated from its Zr/Hf mode (Table 1) assuming achondritic ratio (Zr/Hf = 37.1; McDonough & Sun1995)forthemagmas(orthebulkrockinthecaseofmetamorphicminerals)fromwhichtheywerederived.The resulting values (Table 2) almost exactly matchthe data for the few minerals already experimentallystudied(seesummaryinLinnen&Keppler2002).Forthis reason, the whole dataset in Table 2 can confidently beusedforcalculatingtheeffectofZr-bearingmineralsonZr/Hfduringpetrogeneticprocesses.

The minerals with the highest partition-coefficient of HfrelativetoZraretitanite(KD

Hf/KDZr=2.5),orthopy-

roxene(2.0),amphiboleandclinopyroxene(1.8),andepidote and rutile (1.6–1.7). Ilmenite, magnetite, thefeldspars and apatite have KD

Hf/KDZr approximately

equalto1.Valueslessthanonewerefoundinxenotimeandzircon (0.8),garnet (0.7),andallanite (0.6).AlsoconsideringthemodalabundanceandconcentrationofZrintheseminerals,itfollowsthatthebestcandidatesto have a strong influence on Zr/Hf are, in the order, zircon, titanite, amphibole, clinopyroxeneandgarnet;the influence of the other minerals would be negligible inmostcases.

Figure 15 shows how the joint crystallization ofzirconpluseachoftheaforementionedmineralsaffectsZr/Hfinresidualmelts.Theeffectsofmajormineralsweresimulatedwiththeequationdescribingthetrace-element fractionation for fractional crystallization,assumingZr/Hf=37.1and100ppmZr;theKD

Zrvaluesareaverageestimatesfromtheliterature;theKD

HfwerecalculatedfromtheseandtheKD

Hf/KDZrshowninTable

2. The effects of zircon were simply calculated bysubtractingZrandHffromthemeltintheproportion


47:1 according to the amount of zircon crystallized.Thecurvesforeachmineralrepresentdifferentzircon:mineralmodalratios.

The calculation reveals that the most effectivemineraltocausehigher-than-chondriteZr/Hfvaluesinfractionatesistitanite,followedbyamphiboleandclino-pyroxene. Nevertheless, the effects of these mineralscan be easily compensated by zircon. Clinopyroxeneandamphibolerequireazircon:mineralmodalratioaslowas0.0002to0.0003,buttitaniterequiresamodalratio of about 0.01 (Fig. 16). Garnet fractionation,on theotherhand,has little effect; the crystallizationof 40 wt.% modal garnet moves Zr/Hf from 37.1 to~ 35.These results do not change very much if thecalculations are done for higher concentrations of Zrin the initialmagmaorreasonablydifferentpartition- coefficients for Zr.

Itseems,therefore,thatclinopyroxeneoramphibolefractionation can increase Zr/Hf in fractionates onlywhere the precipitation of zircon is inhibited or kepttoaminimum.Themostfavorablesituationoccursinmafic alkaline magmas, owing to their low Si and high Naactivities;astrikingexample,showninFigure16A,isdisplayedbythecamptoniticlamprophyresofcentralIberia(Scarrowet al.,submitted).Thesehaveabout300ppmZrbutcontainvery little, ifany,zirconorotherZr-saturatedaccessoryphase,sothatthefractionationwas caused first by titanian augite + kaersutite and then by kaersutite alone.This caused the smooth increaseofZr/Hffrom37to50withaslopechangeatSiO2±48coincidingwiththedisappearanceoftitanianaugitephenocrysts.

Thefractionationoftitanite,commonlyaccompaniedbyamphibole,mightbeaneffectivewayofincreasing

Fig.15. NumericalsimulationoftheeffectsofthesimultaneouscrystallizationofzirconandoneZr-bearingmajorphaseontheZr:Hfratiooffractionatedmelts.Theeffectsofmajormineralswerecalculatedwiththetrace-elementfractionationequationforfractionalcrystallization.TheKD

Zrareaverageestimatesfromtheliterature;theKDHfwerecalculatedfromtheseandthe

Zr :Hf partition coefficient ratios shown in Table 2. The effects of zircon were calculated by subtraction of Zr and Hf (47:1) accordingtotheamountofzirconcrystallized.Thecurvesforeachmineralrepresentdifferentzircon:mineralmodalratios.Notethesmallproportionofzirconrequiredtocompensatetheeffectsofamphiboleandclinopyroxene.


Zr/Hfinfractionates,butthisrarelyoccurswithouttheprecipitation of sufficient zircon to compensate, or even reverse,thetrend.Agoodexampleisdisplayedbytheabove-mentionedStepninskPlutonintheSouthUrals,Russia (Beaet al.2005),composedofgabbro-dioriteto leucogranite forming a continuous differentiationseries.Here,despitestrongfractionationofamphibole+ titanite, the Zr/Hf ratio remains “noisily” around 38 duringdifferentiationfromgabbro-diorite(SiO2≈52%)togranodiorites(SiO2≈68%),compensatedbyzirconfractionation.Inmoreevolvedcompositions,amphibolewas totally replaced by biotite, and titanite stoppedcrystallizing,sothattheonlyphasefractionatingZrandHfwaszircon,thuscausingtherapiddecreaseofZr/Hffrom~36to~24(Fig.16B).

Thesecalculationsalso reveal that themosteffec-tive way for magmas to get lower-than-chondritevaluesofZr/Hfinmagmasiszirconfractionation,thussupportingtheconclusionsofLinnen&Keppler(2002).Neithergarnet,owingtoits lowconcentrationsofZr,norxenotimeorallanite,owingtolowmodalfractionand late crystallization, seem capable of producingperceptibledifferencesinZr/Hfduringmagmaticevolu-tion(Fig.15).Zirconfractionation,however,doesnotinvariablyleadtolower-than-chondriteZr/Hffraction-ates(e.g.,Dostal&Chatterjee2000),thoughthisseems

tobethemostcommoncaseinsilicicmetaluminousandperaluminousmelts.

Applications to geochronology

A mineral reaction in which the reactants containmore Zr than can be accommodated in the productswould also produce a Zr-rich accessory as a mass-balancing phase.The same applies toY, REE or anyother trace element that, being an essential structuralcomponentofanaccessory,isalsopartitionedinmajorminerals(e.g.,Pan1997).Datingtheso-formedacces-soryphasesdatesthereactionthatformedthem,permit-ting high-resolution geochronology. Examples havebeendocumentedbyFraseret al.(1997)andDegelinget al.(2001),whofoundnewzirconformedduringtheretrogression of Zr-bearing granulitic garnet to a Zr-freeassemblage,andBingenet al.(2001),whofoundbaddeleyite, srilankite ([Ti,Zr]O2) and zircon formedafter Zr-rich ilmenite in mafic granulites.

Twoadditionalexampleswerefoundinthesamplecollectioninvestigatedhere:theformationofxenotime,insteadofzircon,asaproductofgarnetretrogressionin amphibolite-grade migmatites, and the overgrowthofnewrimsoverzircongrainsincludedinZr-richK-feldsparmegacrysts.

Fig.16. EvolutionofZr/Hf in twodifferentiatedseries.A)Camptonitic lamprophyresfromcentralSpain;Zr/Hf increasesowing to zircon-absent fractionationof titanianaugite + kaersutite. The slope change at SiO2≈48correspondstothedisappearanceoftitanianaugite.B)MainSeriesoftheStepninskpluton.Thefractionationoftitanite+ amphibole is compensated by zircon up to SiO2≈68%,whenthesemineralsstopcrystallizing.Then, zircon fractionation dominates the system and quickly reducesZr/Hffrom36to24.


The first process was identified in the Peña Negra migmatites, where the scarce crystals of garnet areusually rimmed by well-developed cordierite + biotite + quartz coronas (Pereira & Bea 1994), probably developed by the reaction K-feldspar + garnet + melt = biotite + cordierite + quartz (Thompson 1982). Garnet has about 500 ppmY, 250 ppm HREE, 250 ppm P,and 56 ppm Zr, whereas the retrogression products,cordierite,biotiteandquartz,havenegligibleamountsofall theseelements(Table3).Thecoronas(Fig.17)containnozircon,butabundantsmallgrainsofxeno-time ([Y,HREE]PO4), a mineral that never appearsincludedingarnetandisuncommon,butstillfound,inotherareasofthesamethinsection.Xenotimeinsteadof zircon, therefore, seems to be the mass-balancingaccessory of the reaction.This can be easily under-stoodbecauseinthetrace-elementleftoveraftergarnetbreakdown, Y+ HREE are far more abundant than Zr, and there is enough P to precipitate all Y+ HREE as a phosphate.Inaddition,xenotime,whichformsasolidsolution with zircon (Vlasov 1966), will dissolve thesmall fraction of ZrSiO4 formed during the reaction,inhibiting the formation of zircon as an independentphase. This mechanism is reflected by the fact that xeno-timefromtheretrogressioncoronasisnotablyricherinZr,butpoorerinThandU,thanxenotimefromotherareas (Table3).A simplemass-balance indicates thatthe volume ratio garnet/xenotime is about 1000, i.e.,

about1%of the area formerlyoccupiedbygarnet inthinsection,roughlyabout0.3–0.5%ofthecorona,maycurrently be occupied by xenotime. These figures are ingoodagreementwiththerelativemodalproportionsfoundinthinsections(e.g.,Fig.17A).

Theformationofeitherxenotimeorzirconduringthe retrogression of garnet will depend on the trace-elementcompositionofthegarnet.Inmetapelites,theconcentration ofY, HREE and P in garnet decreaseswith increasing metamorphic grade (Bea et al. 1997,Bea&Montero1999,andunpublisheddata).Itseems,therefore, that the breakdown of amphibolite-gradegarnet would tend to form xenotime, as in the Peña Negramigmatites,whereasthebreakdownofgranulite-gradegarnetwouldtendtoformzircon,asdocumentedby Fraser et al. (1997) and Degeling et al. (2001).Xenotimeisalsousefulforradiometricdating(Fletcheret al.2004),sothatthisdifferencedoesnotaffectthepotential of garnet-breakdown reactions for geochro-nology.Buttheformationofeitherzirconorxenotimebythismechanismisprobablylimitedtometapelites.The retrogression of garnet from eclogites or mafic granuliteshardlyrequiresamass-balancingaccessorybecauseitcontainslittleZr(YandHREE)(Table1)andotherZr–Y–HREE-bearingphases,suchasamphibole,frequentlyappearasproductsofthereaction.Asnotedby Bingen et al. (2001), reactions involving Zr-richilmenitearemuchmorepromisingfortheserocks.

Fig.17. SEMimageoftheretrogessionaureoleofagarnetfromametapelitemigmatizedatamphibolite-gradeconditions.Qtz:quartz;Crd:cordierite;Grt:garnet;Bt:biotite;Mnz:monazite,Themass-balancingaccessoryproductofthereactionisZr-richxeno-timeratherthanzircon.


Another zircon-forming reaction is the transfor-mation of Zr-rich feldspars to Zr-free feldspars pluszircon. This phenomenon is well represented in avolcaniclastic formationofcentral Iberia, theOllodeSapo(Frog’sEye)Formation(DiazMonteset al.2004),afelsicgneisswithhuge(upto10cm)megacrystsofK-feldspar of magmatic origin within a medium- tofine-grained groundmass. The megacrysts have variable and irregularlydistributedZr, from0 to30ppm,andscatteredinclusionsofzircon.Some,butnotall,inclu-sionsshowalow-Urimwithamarkedzonalstructureand variable thickness (Fig. 18). Rim morphologysuggests highly different growth-rates over the samefaceoftheinclusion,exactlywhatonewouldexpectifthediffusionofZrtowardtheinclusionhadbeenchan-neledthroughfavorablestructures(cracks,microveins,etc.)ofthehostcrystal.OngoingU–Pbion-microprobeand LA–ICP–MS dating reveals that whereas thecoresyieldedagesclusteringaround490–485Ma,theprobableageoftheOllodeSapoFormation,therimsyieldedabnormal agesof460–420Ma,younger thantheminimumstratigraphicageoftheformation,whichis well known because it is discordantly overlain bytheArenig(488–478Ma)ArmoricanQuartzite.Zircongrains separated from the groundmass, on the otherhand,neithershowrimswithcomparablemorphologynoryieldedU–Pbagesyoungerthan485Ma.Itseems,therefore, that the rimsofyoungerzircon included intheZr-richK-feldsparmegacrystsgrewatasubsolidusstagefromZrsuppliedbythehostfeldspar,althoughthefactorthattriggeredthegrowthofthenewrimsisstillpoorlyknown.Itcouldhaveoccurredduringmetamor-phism,deformationor simplyorderingof theZr-richfeldsparuponcooling.Onceproperlyunderstood,thismechanismmayprovideausefultoolfortrackingthepost-crystallizationhistoryofmagmaticrocks.

Looking atTable 1, and considering the levels oftheREEincommonminerals(e.g.,Bea1996),itseemsthat other reactions could also form mass-balancingaccessory phases useful for geochronology, such aszircon,xenotimeandmonazite.Intheauthors’opinion,reactionsinvolvingthetransformationofamphiboletobiotite in magmatic and late-magmatic systems, andthe breakdown of titanite at either metamorphic ormagmatic conditions, are potentially interesting anddeserve further study.The knowledge of the role ofaccessory phases, either as reactants or products, inmajormineral–mineralandmineral–melt reactions, isstill in its infancy; however, it seems a promising field thatcanhaveagreatimpactnotonlyonhigh-resolutiongeochronology but, presumably, also on the reactionkineticsandthestabilityofmajorphases.

suMMaryandconclusions

Several rock-forming minerals contain significant concentrations of Zr and Hf (Table 1).The highestvalues are found in xenotime, followed by titanite,ilmenite, rutile, allanite, amphibole, clinopyroxene,garnet, magnetite and, occasionally, plagioclase, K-feldspar and orthopyroxene; Degeling et al. (2001)also found elevated Zr concentrations in osumilite.Olivine,cordierite,biotite,muscovite,apatite,epidoteandmonazite,ontheotherhand,havelowabundancesofZr,inmostcaseslessthan1ppm,andthoseofHf,nearorbelow thedetection limit.ThepartitioningofHfrelativetoZrinrock-formingmineralsalsoshowslargedifferences(Table2).ThehighestKD

Hf/KDZrwere

foundintitanite(KDHf/KD

Zr=2.5),orthopyroxene(2.0),amphibole and clinopyroxene (1.8), and epidote andrutile(1.6–1.7).Ilmenite,magnetite,thefeldsparsandapatitehaveKD

Hf/KDZr≈1.Valueslessthanonewere


found inxenotimeandzircon (0.8), garnet (0.7), andallanite(0.6).Themostrelevantimplicationsofthesedataareasfollows:

1)ThegrowthofaZr-bearingphaseduringpartialmelting does not influence the Zr concentration of the meltpredictedbytheWatson&Harrison(1983)model,butincreasestheamountofzirconthatcanbedissolvedatagiventemperature.

2)Beforemeltsegregation,thiswillacceleratethedisappearanceofzirconfromtheprotolith.Inalreadysegregated melts, it will enhance the dissolution ofinherited zircon.This effect will be more marked inmetaluminous magmas precipitating amphibole andtitanitethaninanyothertypeofmagma.

3)TheeffectofZr-bearingphasesontheWHzirconsaturation temperature is negligible in peraluminousrocks,butcanbeofsomeimportanceinmetaluminousrocks,increasingthecalculatedZSTby20to30°Catmost.

4)TheuptakeofZrbycrystallizingmajormineralsinfluences the morphology and modal abundance of magmaticzircon.ThecrystallizationandresorptionofZr-bearingmajorsilicatesaffectthegrowthofmagmaticzircon, and may be one of the main causes of theappearance of texturally discordant zones, apparentlydiachronous,commonlyfoundinthismineral.

5)A higher-than-chondrite Zr/Hf value in a meltcanbecausedbyfractionationoftitanite,amphiboleorclinopyroxene,butthisonlyiftheprecipitationofzirconisinhibitedorkepttoaminimum.Ontheotherhand,themosteffectivewayforamagmatogetalower-than-chondriteZr/Hfvalueistofractionatezircon.

In addition, we found two new examples of reac-tions that produce a mass-balancing accessory usefulfor geochronology: the formation of xenotime as aproduct of garnet retrogression in amphibolite-grademigmatites,andtheovergrowthofnewrimsonzircongrainsincludedinZr-richK-feldsparmegacrysts.

Theformationofxenotimeinsteadofzirconoccursbecauseinthetrace-elementleftoveraftergarnetbreak-down, Y+ HREE are far more abundant than Zr, and there is enough P to precipitate all Y+ HREE as xeno-time.Inmetapelites,theformationofeitherxenotimeorzircondependsonthetrace-elementbudgetofgarnet,whichisrelatedtometamorphicgrade(Beaet al.1997,Bea&Montero1999).Amphibolite-gradegarnetwouldtendtoformxenotime,whereasgranulite-gradegarnetwould tend to form zircon. Garnet from mafic granulites andeclogites,ontheotherhand,canhardlybeexpectedtoformeitherxenotimeorzircon.

ThegrowthofnewrimsoverzircongrainsincludedinZr-richK-feldsparmegacrystswasfoundinafelsicgneiss.These rimsyieldU–Pbagesyounger than theminimum stratigraphic age of the gneiss, and theirmorphologysuggestsgrowthowingtoZrreleasedbythe feldspar and channeled to the inclusion throughcracks, microveins, etc. in the host crystal. Thismechanism,onceproperlyunderstood,mayprovideausefultoolfortrackingthepost-crystallizationhistoryofmagmaticrocks.

Other potentially interesting reactions deservingstudyarethetransformationofamphiboletobiotiteinmagmatic and late-magmatic systems, and thebreak-down of titanite at either metamorphic or magmaticconditions.

acknowledgeMenTs

TheauthorsaregreatlyindebtedtoC.F.Miller,R.Linnenandananonymousrefereefortheirconstructivesuggestions and comments, which greatly improvedthe manuscript, and to O. Cazalla, J.F. Molina and J.H. Scarrow for their help during this work. J.H. Scarrow also improved the English. This project has beenfinancially supported by the Spanish CICYT project BTE2002–04618–C02–01, andby theNATO Collab-orativeGrantEST–CLG–978997.

reFerences

asiMow,P.d.&ghiorso,M.s. (1998): Algorithmic modifi-cations extending MELTS to calculate subsolidus phaserelations. Am. Mineral.83,1127-1131.

Fig. 18. Cathodoluminescence image of a zircon includedinaZr-bearingK-feldsparmegacrystofthevolcaniclasticOllo de Sapo Formation, Spain.The highly luminescentrim(owingtolowU)yieldedaU–Pb(LA–ICP–MS)ageof420Ma,younger than theminimumstratigraphicageof theFormation,which is coveredbyArenig (488–478Ma)sediments.Thecore,however,yielded490Ma,closetorealageoftheOllodeSapoFormation(unpubl.data).Therimmorphologysuggestshighlydifferentgrowth-ratesoverthesamefaceoftheinclusion;thediffusionofZrtotheinclusionwaspresumablychannelizedthroughfavora-blestructures(cracks,microveins,etc.)ofthehostcrystal.These features indicate it grew in subsolidus conditionsfromZrreleasedbythefeldspar.


Bea,F.(1996):ResidenceofREE,Y,ThandUingranitesandcrustalprotoliths;implicationsforthechemistryofcrustalmelts. J. Petrol.37,521-552.

Bea, F., FershTaTer, g.B., MonTero, P., sMirnov,v.n. &Molina, j.F. (2005): Deformation-driven differentiationofgraniticmagma, theStepninskplutonof theUralides,Russia. Lithos81,209-233.

Bea,F.&MonTero,P.(1999):Behaviorofaccessoryphasesand redistribution of Zr, REE, Y, Th, and U duringmetamorphism and partial melting of metapelites in thelowercrust,anexamplefromtheKinzigiteFormationofIvrea–Verbano,NWItaly. Geochim. Cosmochim. Acta63,1133-1153.

Bea, F., MonTero, P., garuTi, g. & zaccarini, F. (1997):Pressure-dependenceofrareearthelementdistributioninamphibolite-andgranulite-gradegarnets.ALA–ICP–MSstudy. Geostand. Newslett.21,253-270.

Bea,F.,MonTero,P.&Molina, j.F. (1999): Mafic precur-sors, peraluminous granitoids, and late lamprophyres intheAvilabatholith,amodelforthegenerationofVariscanbatholithsinIberia. J. Geol.107,399-419.

Bingen, B.,ausTrheiM, h. &wesTinghouse, M. (2001):Ilmenite as a source for zirconium during high-grademetamorphism?TexturalevidencefromtheCaledonidesofwesternNorwayandimplicationsforzircongeochronol-ogy. J. Petrol.42,355-375.

corFu, F., hanchar, j.M., hoskin, P.w.o. & kinny, P.d.(2003):Atlasofzircontextures.In Zircon (J.M. Hanchar & P.W.O. Hoskin, eds.). Rev. Mineral. Geochem. 53,468-500.

david,k.,schiano,P.&allègre,c.j.(2000):AssessmentoftheZr/HFfractionationinoceanicbasaltsandcontinentalmaterialsduringpetrogeneticprocesses. Earth Planet. Sci. Lett.178,285-301.

degeling,h.,eggins,s.M.&ellis,d.j.(2001):Zrbudgetsfor metamorphic reactions, and the formation of zirconfromgarnetbreakdown. Mineral. Mag.65,749-758.

diaz MonTes,a., navidad, M., gonz‡lez lodeiro, F. &MarT’nez caTal‡n, j.r. (2004): El Ollo de Sapo. In Geolog’a de Espa–a. (J.A. Vera, ed.) Soc. Geol. Espa–a. & Inst.Geol.Min.Espa–a,Madrid(69-72).

dosTal,j.&chaTTerjee,a.k.(2000):ContrastingbehaviorofNb/TaandZr/Hfratiosinaperaluminousgraniticpluton(NovaScotia,Canada). Chem. Geol.163,207-218.

FleTcher, i.r., McnaughTon, n.j.,aleinikoFF, j.a., ras-Mussen, B. & kaMo, s.l. (2004): Improved calibrationproceduresandnewstandardsforU–PbandTh–PbdatingofPhanerozoicxenotimebyionmicroprobe.Chem. Geol.209,295-314.

Fraser,g.,ellis,d.&eggins,s.(1997):Zirconiumabun-dance in granulite-facies minerals, with implications for

zircon geochronology in high-grade rocks. Geology 25,607-610.

ghiorso,M.s.,hirschMann,M.M.,reiners,P.w.&kress,v.c.i.i.i.(2002):ArevisionofMELTSaimedatimprovingcalculationofphaserelationsandmajorelementpartition-inginvolvedinpartialmeltingofthemantleatpressuresupto3GPa.Geochemistry, Geophysics, Geosystems3(5),10.1029/2001GC000217.

ghiorso,M.s.&sack,r.o. (1995):Chemicalmass trans-fer in magmatic processes. IV.A revised and internallyconsistent thermodynamic model for the interpolationand extrapolation of liquid solid equilibria in magmaticsystemsatelevated temperaturesandpressures. Contrib. Mineral. Petrol.119,197-212.

hanchar,j.M.&hoskin,P.w.o.,eds.(2003):Zircon.Rev. Mineral. Geochem.53.

hanchar, j.M. &waTson, e.B. (2003): Zircon saturationthermometry.In Zircon (J.M. Hanchar & P.W.O. Hoskin, eds.).Rev. Mineral. Geochem.53,89-112.

jang,y.d.&naslund,h.r.(2003):MajorandtraceelementvariationinilmeniteintheSkaergaardIntrusion,petrologicimplications. Chem. Geol.193,109-125.

kaMBer,B.s.&collerson,k.d.(2000):Zr/Nbsystematicsof ocean island basalts reassessed – the case for binarymixing. J. Petrol.41,1007-1021.

kirkPaTrick,r.j.(1981):Kineticsofcrystallizationofigneousrocks.InKineticsofGeochemicalProcesses(A.C.Lasaga& R.J. Kirkpatrick, eds.). Rev. Mineral.8,321-396.

linnen, r.l. (1998):The solubility of Nb–Ta–Zr–Hf–W ingranitic melts with Li and Li+F, Constraints for mineral-izationinraremetalgranitesandpegmatites. Econ. Geol.93,1013-1025.

linnen,r.l.&kePPler,h.(2002):Meltcompositioncontrolof Zr/Hf fractionation in magmatic processes. Geochim. Cosmochim. Acta66,3293-3301.

Mcdonough,w.F.&sun,s.s. (1995):ThecompositionoftheEarth. Chem. Geol.120,223-253.

Miller, c.F., Mcdowell, s.M. & MaPes, r.w. (2003):Hot and cold granites? Implications of zircon saturationtemperatures and preservation of inheritance. Geology31,529-532.

Pan, yuanMing (1997): Zircon- and monazite-formingmetamorphic reactions at Manitouwadge, Ontario. Can. Mineral.35,105-118.

Pearce, n.j.g. (1990): Zirconium and niobium-bearingilmenitesfromtheIgalikodykeswarm,SouthGreenland. Mineral. Mag.54,585-588.

Pearce, n.j.g., Perkins, w.T., wesTgaTe, j.a., gorTon,M.P.,jackson,s.e.,neal,c.r.&chenery,s.P.(1997):Acompilationofnewandpublishedmajorandtraceele-


mentdataforNISTSRM610andNISTSRM612glassreferencematerials. Geostand. Newslett.21,115-144.

Pereira,M.d.&Bea,F.(1994):Cordierite-producingreac-tions at the Peña Negra complex, Avila batholith, central Spain.Thekeyroleofcordieriteinlow-pressureanatexis. Can. Mineral.32,763-780.

seiFerT,w. & kraMer,w. (2003):Accessory titanite, animportant carrier of zirconium in lamprophyres. Lithos71,81-98.

sisson,T.w. (1994): Hornblende–melt trace-element parti-tioning measured by ion microprobe. Chem. Geol. 117,331-344.

sTaTacorP.(2005):Stata Statistical Software, Release 9.Col-legeStation,TX,StataCorpLP.

ThoMPson,a.B.(1982):Dehydrationmeltingofpeliticrocksand thegenerationofH2O–undersaturatedgranitic liq-uids. Am. J. Sci.282,1567-1595.

vannucci,r.,BoTTazzi,P.,wulFFPedersen,e.&neuMann,e.r.(1998):PartitioningofREE,Y,Sr,ZrandTibetween

clinopyroxene and silicate melts in the mantle under LaPalma(CanaryIslands),implicationsforthenatureofthemetasomaticagents. Earth Planet. Sci. Lett.158,39-51.

villaseca,c.,roMera,c.M.,delarosa,j.&BarBero,l.(2003):ResidenceandredistributionofREE,Y,Zr,ThandUduringgranulite-faciesmetamorphism,behaviourofaccessoryandmajorphasesinperaluminousgranulitesofcentralSpain.Chem. Geol.200,293-323.

vlasov,k.a.(1966):Geochemistry and Mineralogy of Rare Elements and Genetic Types of Their Deposits.IsraelPro-gram for Scientific Translation, Jerusalem, Israel.

waTson, e.B. & harrison,T.M. (1983): Zircon saturationrevisited, temperature and composition effects in a vari-ety of crustal magma types. Earth Planet. Sci. Lett. 64,295-304.

zack,T.,kronz,a.,Foley,s.F.&rivers,T.(2002):Traceelementabundances inrutilesfromeclogitesandassoci-atedmicaschist.Chem. Geol.184,97-122.

Received October 25, 2004, revised manuscript accepted February 25, 2006.

A LA–ICP–MS EVALUATION OF Zr RESERVOIRS IN COMMON …grupo179/pdf/Bea2006.pdf · Hf in...

Documents

Transcript of A LA–ICP–MS EVALUATION OF Zr RESERVOIRS IN COMMON …grupo179/pdf/Bea2006.pdf · Hf in...