Fluid mixing as primary trigger for cassiterite deposition: Evidence from in situ δ<sup>18</sup>O-δ<sup>11</sup>B analysis of tourmaline from the world-class San Rafael tin (-copper) deposit, Peru

(1)

Article

Reference

Fluid mixing as primary trigger for cassiterite deposition: Evidence from in situ δ

¹⁸

O-δ

¹¹

B analysis of tourmaline from the world-class San

Rafael tin (-copper) deposit, Peru

HARLAUX, Matthieu, et al .

Abstract

We present a high-resolution in situ study of oxygen and boron isotopes measured in tourmaline from the world-class San Rafael Sn (-Cu) deposit (Central Andean tin belt, Peru) aiming to trace major fluid processes at the magmatic-hydrothermal transition leading to the precipitation of cassiterite. Our results show that late-magmatic and pre-ore hydrothermal tourmaline has similar values of δ18O (from 10.6‰ to 14.1‰) and δ11B (from −11.5‰ to

−6.9‰). The observed δ18O and δ11B variations are dominantly driven by Rayleigh fractionation, reflecting tourmaline crystallization in a continuously evolving magmatic-hydrothermal system. In contrast, syn-ore hydrothermal tourmaline intergrown with cassiterite has lower δ18O values (from 4.9‰ to 10.2‰) and in part higher δ11B values (from

−9.9‰ to −5.4‰) than late-magmatic and pre-ore hydrothermal tourmaline, indicating important contribution of meteoric groundwater to the hydrothermal system during ore deposition. Quantitative geochemical modeling demonstrates that the δ18O-δ11B composition of syn-ore tourmaline records variable degrees of mixing of a [...]

HARLAUX, Matthieu, et al . Fluid mixing as primary trigger for cassiterite deposition: Evidence from in situ δ

¹⁸

O-δ

¹¹

B analysis of tourmaline from the world-class San Rafael tin (-copper) deposit, Peru. Earth and Planetary Science Letters , 2021, vol. 563, no. 116889

DOI : 10.1016/j.epsl.2021.116889

Available at:

http://archive-ouverte.unige.ch/unige:150678

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Earth and Planetary Science Letters 563 (2021) 116889

Contents lists available atScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Fluid mixing as primary trigger for cassiterite deposition: Evidence from in situ δ ¹⁸ O- δ ¹¹ B analysis of tourmaline from the world-class San Rafael tin (-copper) deposit, Peru

Matthieu Harlaux

^a^,^b^,^∗

, Kalin Kouzmanov

^a

, Stefano Gialli

^a

, Katharina Marger

^c

, Anne-Sophie Bouvier

^c

, Lukas P. Baumgartner

^c

, Andrea Rielli

^d

, Andrea Dini

^d

, Alain Chauvet

^e

, Miroslav Kalinaj

^f

, Lluís Fontboté

^a

aDepartmentofEarthSciences,UniversityofGeneva,1205Geneva,Switzerland bNevadaBureauofMinesandGeology,UniversityofNevada,Reno,NV89557,USA cInstituteofEarthSciences,UniversityofLausanne,1015Lausanne,Switzerland dIstitutodiGeoscienzeeGeorisorse,CNR,56124Pisa,Italy

eGéosciencesMontpellier,UMR5243-CNRS,34095Montpellier,France fMinsurS.A.,Jr.LorenzoBernini149,SanBorja,Lima27,Peru

a rt i c l e i n f o a b s t r a c t

Articlehistory:

Received16November2020 Receivedinrevisedform7March2021 Accepted11March2021

Availableonlinexxxx Editor: R.Hickey-Vargas

Keywords:

tourmaline oxygenisotopes boronisotopes tindeposit hydrothermalﬂuids ﬂuidmixing

We present a high-resolution in situ study of oxygen and boron isotopes measured in tourmaline from the world-class San Rafael Sn (–Cu) deposit (Central Andean tin belt, Peru) aiming to trace majorﬂuidprocessesatthemagmatic-hydrothermaltransitionleadingtotheprecipitationofcassiterite.

Ourresultsshow thatlate-magmatic and pre-orehydrothermaltourmalinehassimilar values ofδ¹⁸O (from10.6^to^14.1⁾ ând δ¹¹B (from−^11.5^to −^6.9^). ^The ôbservedδ¹⁸Oand δ¹¹B variations aredominantly drivenbyRayleighfractionation, reflectingtourmalinecrystallizationinacontinuously evolvingmagmatic-hydrothermalsystem.Incontrast,syn-orehydrothermaltourmalineintergrownwith cassiteritehaslowerδ¹⁸Ovalues(from4.9^to^10.2⁾ândⁱⁿ^part^higher δ¹¹Bvalues(from−^9.9 to−^5.4⁾^thanlate-magmaticandpre-orehydrothermaltourmaline,indicatingimportantcontribution ofmeteoricgroundwater tothe hydrothermalsystemduring oredeposition. Quantitativegeochemical modelingdemonstratesthattheδ¹⁸O-δ¹¹Bcompositionofsyn-oretourmalinerecordsvariabledegrees ofmixingofahotSn-richmagmaticbrinewithmeteoricwatersthatpartiallyexchangedwiththehost rocks.Theseresultsprovidethusdirectinsituisotopicevidenceoffluidmixingasamajormechanism triggeringcassiteritedeposition.Further,thisworkshowsthatcombinedinsituδ¹⁸Oandδ¹¹Banalysesof tourmalineisapowerfulapproachforunderstandingfluidprocessesindynamicmagmatic-hydrothermal environments.

©²⁰²¹^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

TinandWmineralizationisspatiallyandgeneticallyassociated withreducedgranitoids thatwere generatedin orogenicbeltsby melting of tectonically thickened sedimentary sequences (Romer and Kroner, 2016). Cassiterite ± wolframite-bearing quartz vein systems are a speciﬁc type of Sn-W deposits that are emplaced intheupperpartsofevolvedgraniticintrusionsorextendinginto surroundingcountryrocks( ˇCernýetal.,2005).Whilethemineral- izingﬂuidsandmetalsaregenerallyinterpretedtobeofmagmatic

*

Correspondingauthorat:NevadaBureauofMinesandGeology,Universityof Nevada,Reno,NV89557,USA.

E-mailaddress:mharlaux@unr.edu(M. Harlaux).

origin (Audétat et al., 2000a,b; Hulsbosch et al., 2016; Harlaux et al., 2018), the precipitation mechanism(s) of ore minerals re- main(s)debated. Cooling, boiling,fluid-rock interaction,andmix- ing with external fluids have been proposed as major processes controlling cassiterite deposition (Heinrich, 1990). Among these processes,fluid mixingis considered asthe mostlikely effective mechanism based on mineral geochemistry, fluid inclusion, and stableisotopedata(Audétatetal., 2000a,b;Vallanceetal., 2001;

Fekete et al., 2016; Van Daele et al., 2018; Legros et al., 2019;

Hongetal.,2020). Korgesetal.(2018) challengedthismodeland suggesteddepressurizationandboilingofmagmaticﬂuidasmajor factorsforSn-Wdepositformation.

Inthepresentpaper,we reporthigh-resolution insituoxygen andboronisotopemeasurementsbysecondaryionmassspectrom- https://doi.org/10.1016/j.epsl.2021.116889

0012-821X/©²⁰²¹^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY-NC-ND^license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

(3)

M. Harlaux, K. Kouzmanov, S. Gialli et al. Earth and Planetary Science Letters 563 (2021) 116889

Fig. 1.A:Longitudinalcross-sectionoftheSanRafaelSn(–Cu)deposit,Peru.BtoG:Petrographicfeaturesofthestudiedtourmalinesamples(picturesfromHarlauxet al.,2020).B:Late-magmaticquartz-tourmalinenodule(Tur1)inmicrogranite.C:Late-magmatictourmaline(Tur1)disseminatedinmicrogranite.D:Pre-orehydrothermal tourmaline(Tur2)partiallyreplacingaK-feldsparphenocrystinmegacrysticgranite.E:Pre-orequartz-tourmalinevein(Tur2)surroundedbyalterationhalocrosscuttingthe megacrysticgranite.F:Syn-oretourmaline(Tur3)veinletsandovergrowthsoverprintingapre-orequartz-tourmalinevein(Tur2).G:Syn-oretourmaline(Tur3)intergrown with cassiterite.Abbreviations:Ab= albite,Bt= biotite,Cst =cassiterite,Kfs= K-feldspar,Plg =plagioclase,Qtz =quartz,Ser= sericite,Tur= tourmaline.(For interpretationofthecolorsintheﬁgure(s),thereaderisreferredtothewebversionofthisarticle.)

etry (SIMS)intourmalinefromtheSanRafaeldeposit(Peru),one of the world’s largest and richest vein-type Sn deposits. Due to its chemical stability and refractory nature, tourmaline is a ro- bust isotopic monitorof ore-forming processes andfluid sources (Slack and Trumbull, 2011; Codeço et al., 2019; Trumbull et al., 2020;Hong etal., 2020). Theaimofthisstudy isto usetheiso- topiccompositionoftourmalinetotracefluidprocessesoccurring atthemagmatic-hydrothermaltransitionleadingtocassiteritede- position. Usingquantitative geochemical modeling,we show that insituδ¹⁸O-δ¹¹Banalysesontourmalineareindicativeoffluidex- solutionintheearlystagesofthemagmatic-hydrothermalsystem and, subsequently,mixingwithmodifiedmeteoric waterstrigger- ingcassiteritedeposition.

2. Geologicalbackground

Theworld-classSanRafaelSn(–Cu)deposit(>1MtSnataver- agegradeof2%)islocatedinthenorthernpartoftheCentralAn- deantinbeltextendingfromsoutheastPerutoBoliviaandnorth- ern Argentina. This classic metallogenic province hosts hundreds

ofSn-Wdeposits,whichcontainwidespreadtourmalinealteration andhighBcontents(Lehmannetal., 2000).MineralizationatSan Rafaelconsistsofanorthwest-trendingcassiterite-quartz-chlorite- sulﬁdevein-brecciasystemhostedbyalateOligocene(ca.25Ma) peraluminousgranitic complexandby Ordovician metasediments of the Sandia Formation (Fig. 1A; Kontak and Clark, 2002; Mly- narczyk et al., 2003). The San Rafaeldeposit is characterized by abundant tourmaline, both ofmagmatic andhydrothermalorigin (Kontak and Clark, 2002; Mlynarczyk and Williams-Jones, 2006).

Three major tourmaline (Tur) generations were identified atSan Rafael(Harlaux et al., 2020): (i) Tur 1 is found in peraluminous granitesasquartz-tourmalinenodulesanddisseminations(Fig.1B- C)andis interpreted to be late-magmatic; it is texturallyhomo- geneous andhas draviticcomposition, closeto the schorl-dravite limit, with Fe/(Fe + ^Mg) ^ratios ôf ^0.36-0.52; ⁽ⁱⁱ⁾ ^Tur ² îs ^pre- ore hydrothermal tourmaline crosscutting the granites and host rocks and formed during post-magmatic alteration and veining (Fig. 1D-E); it shows oscillatory zoning at the microscopic scale andhasintermediate compositionsrangingfromdravitetoschorl, with Fe/(Fe + ^Mg) ^ratios ^between ^0.01 ând ^0.83; ⁽ⁱⁱⁱ⁾ ^Tur ³ îs

2

(4)

syn-orehydrothermaltourmalineformingwidespreadmicroscopic veinlets and overgrowths, partly crosscutting the previous tourmaline generations (Fig. 1F-G); it has schorl to foitite compositions, withFe/(Fe + ^Mg) = ^0.48-0.94, ând îs ^locally întergrown withcassiteriteandchlorite ofthemainorestage.Inthepresent study, we selected representative samples covering the different generations of magmatic to hydrothermal tourmaline (Tur 1 to Tur 3). These samples were mostly collected from underground workingsintheSanRafaelmine(between3,610 and4,475 masl) andweretexturallyandchemicallycharacterizedbyHarlauxetal.

(2020).

3. Analyticalmethods 3.1. Oxygenisotopeanalysis

Oxygen isotope analyses of tourmaline and quartz were per- formedattheSwissSIMSionprobenationalfacilityattheUniver- sity of Lausanne (Switzerland) using a Cameca IMS 1280-HR instrument.Atotalof180¹⁸O/¹⁶Oisotopicratioanalyseswere performed ontourmalineandquartz chipsplacedinindiummounts together with grainsof three tourmaline (UNIL-T2, UNIL-T6, and IAEA-B-4; Marger et al., 2019) and one quartz (UNIL-Q1; Seitz et al., 2017) reference materials. The complete SIMS dataset for O isotopes is reported in Supplementary Table S1. Prior to the SIMSsession,themountswerecleanedwithdryethanolandwere driedinan ovenat60^◦Cfor24 h,andthen coatedwitha35-40 nm-thicklayer ofgold.Oxygen isotopemeasurements were done witha 10 kVCs⁺ primary beam anda ∼² ^nA ^current, ^resulting in a ∼^{15 μm} ^beam ^size. ¹⁶Ô ând ¹⁸Ô ^secondary îons, âcceler- ated at10kV, were measured ata mass resolving powerM/M of 2400 using an entrance slit open at 122 μm and the multicollection exit slit 1, and were counted in multicollection mode on Faraday cups (L’2 with 10¹¹ resistor for ¹⁶Oand H’2with 10¹⁰ resistor for ¹⁸O). The background of the Faraday cups was calibrated at the beginning of each day session, and a nu- clearmagneticresonancewasusedtolockthemagneticfield.Each calculated isotopic ratio ¹⁸O/¹⁶O corresponds to 5 measurement blocks of4 cyclesper block, resultingin 20cycles ofacquisition with countingtimes of5 s per cycle on both massessimultane- ously. Consequently,each analysistakes ∼^3.5^min,încluding ^pre- sputtering (30 s)and automated centeringof secondary ions(60 s).All data havebeen obtainedduring the sameSIMS session in March2018. Thereferencematerials weremeasured betweenev- ery5 to8unknowns formonitoring theinstrumentstability and analytical accuracy. The average reproducibility on the reference materials wasbetter than0.4^(2SD)ândînternal êrror^forêach analysislowerthan0.2(2SD).Aftercorrectionoftheinstrumen- talmassfractionation(IMF),themeasured¹⁸O/¹⁶Oisotopicratios were correctedfortourmalinematrixeffectbasedontheMg/(Mg + ^Fe) ând Âl(Y) ^contents ^determined independently by electron microprobe (Marger et al., 2019; Harlaux et al., 2020). The average IMF for all sessions of O isotope measurement was about 3.6 ± ^0.2^forÎAEA-B4 ^schorl, ^2.3± ^0.3 ^forÛNIL-T2 ^dravite, 3.6 ± ^0.3 ^for ÛNIL-T6 êlbaite, ând ^5.8 ± ^0.1 ^for ÛNIL-Q1 quartz. The IMF dependence on tourmaline chemistry (between schorl anddravite) is therefore 1.3 ^for Ô îsotopes. ^Results âre reported in δ¹⁸O = 1000 × [(¹⁸O/¹⁶Osample)/(¹⁸O/¹⁶OV-SMOW) – 1)] in ^relative ^to^the ^Vienna ^Standard ^MeanÔcean ^Water^(V- SMOW).

3.2. Boronisotopeanalysis

Boron isotope analyses of tourmaline were performed at the SwissSIMSionprobenationalfacilityattheUniversityofLausanne (Switzerland)usingaCamecaIMS 1280-HRinstrument.Atotalof

166¹¹B/¹⁰B isotopic ratioanalyses were performed onthe same tourmalinechipsanalyzedforOisotopes.Each indiummountin- cluded grains of three tourmaline reference materials (UNIL-T2, UNIL-T6,and IAEA-B-4; Marger et al., 2020). The complete SIMS datasetforBisotopesisreportedinSupplementaryTableS1.Sam- ple preparation was the same as for O isotope analysis. Boron isotopemeasurementsweredonewitha13kVO⁻duoplasmatron primarybeamanda∼^5-6^nA^current,^resultingⁱⁿâ∼^{20 μm}^beam size.¹⁰Band¹¹Bsecondaryions,acceleratedat10kV, weremea- suredata massresolving powerM/M of2400(entranceslit at 100 μmandexitslit1ofthemulticollection)andwerecountedin multicollectionmodeonFaradaycups(L’2for¹⁰BandH’2for¹¹B, bothsetwith10¹¹resistors).Typicalcountrateobtainedon¹⁰B was2.5×¹⁰⁶^cps^(count^per^second)ând⁹×¹⁰⁶^cps^for¹¹^B.^The background ofthe Faraday cups was calibrated at the beginning ofeach daysession,and anuclear magnetic resonancewas used to lockthe magnetic field. Eachcalculated isotopic ratio ¹¹B/¹⁰B corresponds to3 measurement blocksof10 cyclesper block, re- sultingin30cyclesofacquisition withcountingtimesof5s per cycleoneach mass.Eachanalysistakes∼^4.5^min,încluding^pre- sputtering(50 s) andautomated centering ofsecondary ions (60 s).Thissettingallowedanaveragereproducibilityonthereference materialsbetterthan0.6^(2SD)ândînternalêrror^forêachânal- ysislower than 0.4 ^(2SD). Âll ^data ^have^been ôbtained ^during thesame SIMS session inNovember 2018. The referencemateri- alsweremeasuredbetweenevery2to4unknownsformonitoring the instrument stability and analytical accuracy. After correction of the IMF, the measured ¹¹B/¹⁰B isotopic ratios were corrected fortourmalinematrix effectbasedon thewt.% FeO+ MnO con- tentdeterminedindependentlybyelectronmicroprobe(Margeret al., 2020; Harlaux etal., 2020). The average IMF forall sessions of B isotope measurement was about 20.6 ± ^0.7 ^for ÎAEA-B4 schorl and about24.1 ± ^0.9 ^for ÛNIL-T2 ^dravite. ^The ÎMF ^dependence on tourmaline chemistry (between schorl and dravite) is therefore3.5 ^for^B îsotopes. ^Results âre ^reportedⁱⁿ δ¹¹B = 1000× ^[(¹¹^B/¹⁰^Bsample)/(¹¹B/¹⁰BNIST SRM 951) –1] in ^relative^to theNISTSRM951.

4. Results

4.1. Oxygenandboronisotopiccompositionsoftourmaline

The δ¹⁸O and δ¹¹B compositions of tourmaline measured by SIMS arereported inSupplementary TableS1. Late-magmatic Tur 1hasanear-constantδ¹⁸Oisotopiccomposition fallinginarange between10.6ând^11.6^,^whereasîtsδ¹¹Bcompositionismore variable ranging between−^11.1 and −^7.8 (Fig. 2A). No sig- nificant core-rim isotopic zoning is observed. Pre-ore hydrother- malTur 2 hasmore variable valuesof δ¹⁸Oof 10.7 ^to ^14.1 and δ¹¹B of −^11.5 ^to −^6.9^, overlapping with those of Tur 1. In detail, metasediment-hosted Tur 2 shows higher δ¹⁸O values (13.6-14.1⁾ ^than granite-hosted Tur 2 (10.7-11.9⁾ ^while their δ¹¹B compositionsoverlap.In asingle oscillatory-zoned Tur 2 crystal, isotopic variations of 1.0 ⁱⁿ δ¹⁸O and3.5 ⁱⁿ δ¹¹B arereachedoverallfromthebaseofthecrystaltothetip(Fig.3).

Thesevariations outlinepseudo-periodiccyclesatthemicroscopic scalesuggestingadynamicenvironmentlikelycausedbyrepeated ﬂuid pulses. Syn-ore hydrothermal Tur 3 has lighter δ¹⁸O values ranging from 4.9 ^to ^10.2 ^and ⁱⁿ ^part ^heavier δ¹¹B values from −^9.9 ^to −^5.4^. ^No ^relation ^between ^sample ^loca- tion in the deposit and isotopic composition of tourmaline was found.

4.2.Quartz-tourmalineoxygenisotopeequilibriumtemperatures The δ¹⁸Ocompositions ofmagmatic andhydrothermal quartz showingequilibriumtextureswithtourmalinewerealsomeasured

(5)

Fig. 2.A:Plotofδ¹⁸Ovs.δ¹¹Bcompositionofmagmaticandhydrothermaltourmaline(Tur1toTur3)fromtheSanRafaeldepositdeterminedbysecondaryionmass spectrometry(SIMS).B:Calculatedδ¹⁸Ovs.δ¹¹Bcompositionofwaterequilibratedwithtourmalinebyassumingcrystallizationtemperaturesof650^◦CforTur1,550^◦C forTur2,and350^◦CforTur3.ThefieldforS-typegranitesisfromTrumbullandSlack(2018) andHarrisetal.(1997).Calculatedδ¹⁸Ovaluesforwaterequilibratedwith mineralsfromthepre-orestage(Qtz,Tur)at550^◦Candthemainorestage(Qtz,Cst,Chl)at350^◦Careshown.C:Modelingofδ¹⁸Ovs.δ¹¹Bcompositionoftourmaline calculatedfordifferentfluidevolutionscenarios:coolingofasingle-phasemagmaticfluid(inred),Rayleighfractionationofasingle-phasemagmaticfluid(inblue),fluid-rock interactionbetweenmagmaticvaporandthehostmetasediments(ingreen),andfluidmixingbetweenaSn-richmagmaticbrineandmodifiedmeteoricwaters(inblackand grey).Seetextfordetails.

by SIMS in order to determine crystallization temperatures. The analyzed quartz-tourmaline pairs show primary intergrowth tex- tures and lack hydrothermal alteration features, as revealed by preliminary cathodoluminescencequartz imaging.Results ofδ¹⁸O compositionsofquartzandtourmalinearereportedintheSupple- mentaryTable S2andin Fig.4.Isotope equilibriumtemperatures forquartz-tourmaline pairswere calculatedusingtheequation of Matthews et al. (2003), which is the most appropriate for es- timating temperatures in peraluminous granitic systems (Marger et al., 2019). Quartz-tourmaline δ¹⁸O compositions yield equilib- rium temperatures of 615 ± ³⁸^◦^C ^for ^Tur ¹ ând ⁵³⁰ ± ²⁵^◦^C for Tur 2 (Fig. 4). This indicates temperatures of 500^◦ to 600^◦C forthe magmatic-hydrothermal transition,which isin agreement withtheearliestrecordedfluidsatSanRafaelasrevealedbypre- vious fluidinclusionstudies onmagmaticquartzphenocrystsand pre-orequartz-tourmaline veins(KontakandClark, 2002; Wagner etal.,2009).

5. Discussion

5.1. Originandevolutionoftheboron-richhydrothermalﬂuids

Theuppervaluesforthequartz-tourmaline isotopethermome- ter indicate crystallization temperatures of about 650^◦C for Tur 1 and 550^◦C for Tur 2. No quartz directly intergrown with Tur 3 was found, butprevious works reported homogenization temperatures of about 350^◦C for primary ﬂuid inclusions hosted in cassiterite (Wagner et al., 2009). Because Tur 3 co-precipitated withcassiterite,an averagevalue of 350^◦Cis assumedtobe the besttemperature proxyforits formation.Consideringcrystalliza- tiontemperaturesof650^◦CforTur1,550^◦CforTur2,and350^◦C forTur3,we calculatedthe δ¹⁸Oandδ¹¹B compositionof water inequilibriumwithtourmaline(Fig.2B)usingtheOandBisotope fractionationfactorsofZheng (1993) and Meyeretal.(2008),re- spectively. Changingthe temperature oftourmaline formation by

±⁵⁰^◦^C^would^shift^the^calculated^range^of^watercompositionsby 4

(6)

Fig. 3.Insituδ¹⁸Oand δ¹¹Bvariationsinasingleoscillatory-zonedtourmalinecrystal(Tur2)fromagranite-hostedpre-orequartz-tourmalinevein.A:Backscattered electron(BSE)imageoftourmalineshowingprimarygrowthzoning.B:δ¹⁸Oandδ¹¹Bvariationsalongatransectperpendiculartogrowthbanding(errorbarsare2σ).

ChemicalvariationsoftheFe/(Fe+Mg)ratioreﬂecttheoscillatoryzoningvisibleontheBSEimage(datafromHarlauxetal.,2020).

amaximumof± ^1.0ⁱⁿδ¹⁸Oand±^0.5ⁱⁿδ¹¹B.Theisotopic compositionsofwaterequilibratedwithgranite-hostedTur1and Tur2 overlapovera rangeof δ¹⁸Ovaluesfrom 10.9 ^to^12.1 andδ¹¹B valuesfrom −^10.0to −^5.3. Theseδ¹⁸Ovalues are consistent withprevious bulk δ¹⁸O analyses of quartz and tourmaline separates from the pre-ore stage equilibratedwith water at550^◦C (SupplementaryTable S3),while the δ¹¹B valuesfall in the δ¹¹Brangereported frommelt inclusionsinBolivian Snpor- phyries (Wittenbrink et al., 2009). The values of δ¹⁸O = ^11.5 and δ¹¹B = −^10.0 âllow ^defining â ^single ^fluid êndmember (“fluid A” in Fig. 2B) that fits the typical isotopic compositional range of a magmatic fluid derived from a S-type granite (Har- ris et al., 1997; Trumbull andSlack, 2018). The magmatic origin ofthe fluid Ais supportedby the high-salinity(40-60 wt.%NaCl eq) and high-temperatures (350-550^◦C) of primary multiphase aqueousinclusionscoexistingwithlow-salinityvaporinclusionsin magmaticquartz phenocrysts andquartz-tourmaline veinsatSan Rafael, whichare interpreted tobe formed underlithostaticfluid pressureof0.8to1.0kbar(KontakandClark,2002;Wagneretal., 2009).

Consideringvaluesofδ¹⁸O= ^11.5 ândδ¹¹B = −^10.0 ^for the magmatic brine andusing liquid-vapor isotopic fractionation factorsextrapolatedto 550^◦C(Shmulovich etal., 1999; Liebscher etal., 2005),thecoexistingvaporwouldhaveδ¹⁸O= ^11.3ând δ¹¹B = −^8.1^. Âssuming â ^20:80 ^proportionôf ^liquid ⁽⁴⁵ ^wt.%

NaCl) and vapor (1 wt.% NaCl), the single-phase magmatic fluid wouldhaveaninitialsalinityofca.10wt.%NaClandisotopicval- uesofδ¹⁸O=^11.4ândδ¹¹B= −^8.5^.^Theîncreasing^trendⁱⁿ δ¹¹Batnear-constantδ¹⁸Ovaluesobservedforgranite-hostedTur 1andTur2maythereforereflecttheprogressivedegassingofthe silicatemelt andphase separation ofthe single-phasemagmatic fluid,similar towhat istraditionallyobserved forHisotope fractionation (Hedenquist and Lowenstern, 1994). The isotopic com- positionofwaterinequilibriumwithmetasediment-hosted Tur2 differsfromtheir granite-hostedequivalentbyhigherδ¹⁸Ovalues between13.7ând^14.3^while^their δ¹¹Bvaluesof −^6.8^to

−^5.8 ôverlap.^Theseδ¹⁸O-δ¹¹B compositions areinterpreted to represent a fluid endmember with δ¹⁸O = ^14.0 ând δ¹¹B =

−^6.0 ^that ^possibly corresponds to B-rich magmatic vapor that ascended toward thetop of the intrusion upon phase separation andinteractedwiththehostmetamorphic rocks, whilethemag-

(7)

Fig. 4.Plotofδ¹⁸Ovaluesforquartz(Qtz)andtourmaline(Tur)pairsofmagmatic and hydrothermalorigin fromthe SanRafaeldeposit.Isothermsareplottedus- ingthequartz-tourmalineoxygenisotopefractionationequationofMatthewsetal.

(2003).

maticbrineremaineddominantlyintheupperpartofthegranitic stock. The isotopic composition of water equilibrated with Tur 3 plots along an array ranging from δ¹⁸O = ^8.4 ^and δ¹¹B =

−^6.7^toδ¹⁸O= ^3.2ândδ¹¹B = −^2.1^(Fig. ^2B).^This^range ofδ¹⁸Ovaluescoversthepreviousbulkδ¹⁸Ocompositionsofore- stage quartz, cassiterite, and chlorite equilibrated with water at 350^◦C (Supplementary Table S3). The observed δ¹⁸O-δ¹¹B array suggestsamixingprocess betweena magmaticfluidandanother fluidendmemberofexternalorigin,havinglighterδ¹⁸Oandheav- ierδ¹¹Bcomposition.

5.2. Quantitativemodelingofﬂuidevolution

Based on the tourmaline O and B isotopic compositions, we tested differentscenarios considering cooling, Rayleigh fractionation, fluid-rock interaction, and fluid mixing as major processes controlling cassiterite deposition. The calculationsare detailed in the Supplementary Material andthe resultsare reported inSup- plementaryTablesS4toS9.Thecalculatedδ¹⁸Oandδ¹¹Bcomposi- tionsoftourmalineforeachscenario(Fig.2C)werethencompared withtheonesmeasured bySIMS (Fig. 2A). Theisotopiccomposi- tion of Tur2 was modeled assuming that thestarting magmatic fluid is single-phaseand hasa temperature of550^◦C, a 10 wt.%

NaCl salinity, and isotopic values of δ¹⁸O = ^11.4 ^and δ¹¹B =

−^8.5^.^Coolingôf^this^magmatic^fluid^down^to²⁵⁰^◦^C^resultsⁱⁿâ trendofincreasingδ¹⁸Ovaluesanddecreasingδ¹¹Bvalues,which doesnot reproducethe measuredisotopic compositionsofTur2.

Rayleigh fractionationdriven by tourmalinecrystallization froma single-phasemagmaticfluidoramagmaticbrine(“fluidA”),atde- creasing temperatures from550^◦ to 500^◦C, resultsin a trend of near-constant δ¹⁸O and increase in δ¹¹B values.This reproduces closelytheisotopiccompositionalrangeofgranite-hostedTur2for 10% to 90% fractionation ofthe initial magmatic fluid.Fluid-rock interaction betweenan ascendingB-rich magmaticvapor andthe host metasediments was modeled for different water/rock ratios (0.1to10)at500^◦Cbyassumingisotopicexchangewithmuscovite (Zheng, 1993; Wunderet al., 2005), which isthe mostabundant andthemainB-carriermineralinmetasedimentsfromtheSandia Formation.We assumedthatthe magmaticvaporhasa tempera- tureof500^◦C,a1 wt.%NaClsalinity,andisotopicvaluesofδ¹⁸O

= ^11.3 ^and δ¹¹B = −^8.1^. ^A concentration of 3,000 ppm B was estimated for the magmatic vapor based on measurements ofindividual vaporinclusionsfromgranite-related Sn-Wdeposits (Audétat et al., 2000a,b). For the metasediments, an initial δ¹⁸O

=^14.0^wasâssumed^basedôn^theâverageÔîsotopic^composi- tionsofEarlyPaleozoicshales(Bindemanetal.,2016),andvalues of δ¹¹B = −^11.0 ând ^B = ¹⁵⁰^ppm ^were ^taken ^fromâverage bulkanalysesofrepresentativemetasedimentsamplesoftheSan- diaFormation(SupplementaryTableS10).Resultsofthefluid-rock interaction modeling reproduce the measured isotopic composi- tionsofmetasediment-hostedTur2formoderatewater/rockratios of0.1-0.3.

The isotopic composition of Tur3 was modeled using a fluid mixing scenario between a magmatic brine and modified meteoric waters.For the magmatic brine, we assumeda temperature of 550^◦C, δ¹⁸O = ^11.5^, ând ^variable δ¹¹B (−^10.0, −^8.5, ând

−^7.0⁾ ⁱⁿôrder ^to^cover^the ^rangeôf ³^variation ⁱⁿδ¹¹B ob- servedforTur1toTur2(Fig.2A). Aconcentrationof1,500 ppm B was assumed for the magmatic fluid based on measurements ofindividual brine inclusionsfrom granite-related Sn-Wdeposits (Audétatet al., 2000a,b; Thomas et al., 2003). In contrast to the pre-ore hydrothermalstage yielding quartz-tourmaline veins and breccias, the salinity of the ore-forming magmatic brine is un- known. Therefore, we considered different initial fluid salinities (45,35,and30wt.%NaCl)forthecalculations.Stableisotopemea- surements of volcanic glasses in the Peruvian Eastern Cordillera yield estimates of δ¹⁸O = −^8.0 ând δD = −⁵⁴ ^for âncient meteoricwatersduringthelate Oligocene(ca. 25Ma),i.e., coeval withtheformationoftheSanRafaeldeposit(Sundelletal.,2019).

UsingtheBmeteoricwaterlineofRose-Kogaetal.(2006),wees- timateda δ¹¹B = ^30.0 ^for^the ^lateÔligocene ^meteoric ^fluidⁱⁿ the PeruvianEastern Cordillera, which falls in the typical values ofcontinental rainwaters (Gaillardet andLemarchand, 2018). Ad- ditionally, we assumed a zero salinity and a B concentration of 0.1ppm forthemeteoricfluid.Thecompositionoflate Oligocene meteoric waters(δ¹⁸O= −^8.0^, δ¹¹B = ^30.0^, ^B = ^0.1^ppm) afterinteraction withtheSandiametasediments(δ¹⁸O= ^14.0^, δ¹¹B = −^11.0^,^B = ¹⁵⁰^ppm)^was êstimatedûsing â^modelôf sequential B extraction and addition to downwelling percolating water(SupplementaryMaterial). Atadepthof4km(correspond- ingtoalithostaticpressureof1kbarandatemperatureof120^◦C), the modified meteoric water after exchange with the metasedi- mentswould contain600ppm B andwouldhaveisotopicvalues ofδ¹⁸O= −^3.3ând δ¹¹B = ^16.0^forâ ^moderate^water/rock ratioof 0.5. The calculated B concentrationsand δ¹¹Bvalues are consistentwithmeasurementsofmoderngeothermalwatersfrom meteoric-dominated hydrothermalsystems, such as Larderello in Italy (Duchi et al., 1992; Pennisi et al., 2001). The fluid mixing modeling betweena magmatic brine andmodified meteoric wa- tersresultsinasystematictrendofdecreasingδ¹⁸Oandincreasing δ¹¹B, whichreproduces thefull rangeofisotopiccompositions of Tur3fordegreesofmixingbetween5%and60%ofmodifiedme- teoricwater.Whiledifferentinitialsalinities(45, 35,and30wt.%

NaCl)ofthemagmaticbrineyieldsimilarmixingcurves,theﬂuid salinity changes the fraction of meteoric water that is required toreproduce themeasured δ¹⁸Oandδ¹¹B compositionsofTur3 (SupplementaryMaterial). Fluidmixing modeling combined with salinityestimatesfromprimaryﬂuidinclusionshostedinore-stage minerals(12-22wt.%NaCleq;KontakandClark,2002;Wagneret al.,2009)indicatethatthemagmaticbrineinvolvedintheprecip- itationofTur3hadasalinityofca.30wt.%NaCl.

5.3.Geneticmodelandtinmineralization

Experimental work showed that high B contents (>1 wt.%

B2O3) in peraluminous graniticmelts lower solidustemperatures 6

(8)

Fig. 5.FluidevolutionmodelfortheSanRafaelSn(–Cu)deposit.A:Crystallizationoflate-magmaticdisseminatedtourmalineandquartz-tourmalinenodules(Tur1)was followedbyexsolutionofaB-richsingle-phasemagmaticfluid.B:PhaseseparationoftheB-richsingle-phasemagmaticfluidintolow-salinityvapor(V)andhigh-salinity liquid(L)yieldedtheformationofpre-orequartz-tourmalineveinsandbreccias(Tur2)cuttingthegraniteandthehostmetasediments.C:MixingofaSn-richmagmatic brinewithmodifiedmeteoricwaterscirculatingintheupperpartsofthegraniteresultedinthedepositionofcassiterite(Cst)togetherwithquartz(Qtz),chlorite(Chl),and syn-oretourmaline(Tur3)inveinsandbreccias.

to <680^◦C at 1 kbar, reduce melt viscosity, and increase water solubility (Dingwell etal., 1996). The SanRafael granite is char- acterizedby amoderately fractionatedperaluminouscomposition and whole-rock B content of 60 to 160 ppm (Kontak andClark, 2002; Mlynarczyk et al., 2003; Harlaux et al., 2020). This im- plies that the San Rafaelsilicate melt had initially much higher B contentsowingto elevatedB fluid/meltpartitioncoefficientsin graniticsystems(Hervigetal.,2002;Thomasetal.,2003;Schatzet al.,2004;Fiedrichetal.,2020).CrystallizationofTur1atca.650^◦C isinterpreted tohaveoccurredduring thelate-magmaticstage of the SanRafaelintrusion prior orconcomitantly to theexsolution of a B-rich single-phasemagmaticfluid (Fig. 5A). The magmatic- hydrothermaltransitionoccurredbetween500^◦ and600^◦Casin- dicated by temperature estimates from quartz-tourmaline pairs, but the single-phase magmatic fluid likely exsolved at temperatures close to 600^◦C. During water exsolution, the immiscible volatilephase preferentiallypartitionsBandbecomesisotopically heavier relative to the granitic melt (Maner and London, 2018).

Thisispossiblyreflectedby theincreasing trendinδ¹¹Bobserved forgranite-hostedTur1andTur2.Theoverpressuredsingle-phase magmatic fluid triggered hydrofracturing of the roof of thecrys- tallizingmagmachamberresultinginphaseseparationintovapor (ca. 1wt.%NaCl) andbrine(ca.45 wt.%NaCl)andthe formation ofTur2inveinsandbreccias(Fig.5B).Thislikelyoccurredunder dominant lithostatic pressure conditions as reported by previous fluid inclusion studies (Kontak and Clark, 2002; Wagner et al., 2009). Phase separation alsoresulted in geochemical decoupling, whereSnpreferentiallypartitionsintothe brinewhereasBparti- tionsintothevapor(Heinrichetal.,1999;Audétatetal.,2000a,b;

Schatzetal.,2004;FoustoukosandSeyfried,2007;Fiedrichetal., 2020). Basedon their similarδ¹⁸O-δ¹¹Bvalues,we interpretthat Tur1andTur2crystallizedinacontinuouslyevolvingmagmatic- hydrothermalsysteminwhichisotopicvariationsweredominantly drivenbyRayleighfractionation.

The δ¹⁸O-δ¹¹B array observedfor Tur3 reflects infiltration of modified meteoric waters into the hydrothermal system during the main orestage. Mostlikely, thisoccurredduring a transition from lithostatic to hydrostatic pressure conditions causedby the opening of dilatational fault jogs during sinistral-normal strike- slip faulting (Mlynarczyket al., 2003; Wagner etal., 2009). This

processcausedthe mixingandthecoolingofa hotSn-rich magmatic brine (ca. 30 wt.% NaCl) with modified meteoric waters (Fig.5C),resultingindestabilizationofSn-chloridecomplexesand the consecutive precipitation of cassiterite (Schmidt, 2018). This interpretation is consistent with the evolution of the San Rafael magmatic-hydrothermal system from reducing toward relatively more oxidizing conditions as deduced from the changing textu- ral and compositional features of Tur 1 to Tur3 (Harlaux et al., 2020). The Sn-rich magmatic brine was produced during a new hydrothermal activity in the magma reservoir. We propose that thisoccurredeitherduring the fluid exsolutionfroman underly- ing residual granitic magma, or during the mixing of a residual brine fromthe pre-ore stage, temporally storedat depth, witha new pulseof low-salinity magmatic fluid, as proposed for other oredeposits(Kouzmanovetal.,2010;Rottieretal.,2018).

6. Conclusions

High-resolution in situ δ¹⁸O-δ¹¹B analysis of tourmaline has allowed tracing the fluid evolution of the San Rafael magmatic- hydrothermalsystem. Ourresults showthat late-magmatic Tur1 andpre-ore hydrothermalTur2 formed ina continuously evolving magmatic-hydrothermal system, whereas syn-ore hydrother- malTur3recordsvariabledegreesofmixingbetweenSn-richmagmatic brine and modified meteoric waters. Fluid mixing related tolithostatic-hydrostaticpressuretransitionandinfiltrationofme- teoricwaters into the hydrothermal systemwas the key mechanism triggering cassiterite deposition in the San RafaelSn (–Cu) deposit. Furthermore, thisstudy demonstrates that the combina- tionofinsituδ¹⁸Oandδ¹¹Banalysesoftourmalineisapowerful approachforunderstandingfluidprocessesindynamicmagmatic- hydrothermalenvironments.

CRediTauthorshipcontributionstatement

MatthieuHarlaux:Writing - originaldraftpreparation,Review andediting,Dataacquisitionandtreatment,Dataanalysisandin- terpretation.KalinKouzmanov:Reviewandediting,Data analysis andinterpretation, Supervision, Conceptualization. StefanoGialli:

Reviewand editing.KatharinaMarger: Reviewandediting, Data

Fluid mixing as primary trigger for cassiterite deposition: Evidence from in situ δ<sup>18</sup>O-δ<sup>11</sup>B analysis of tourmaline from the world-class San Rafael tin (-copper) deposit, Peru

Article

Reference