0361-0128/09/3809/223-26 223

IntroductionHYDROTHERMAL TIN LODES represent a major class of granite-related tin deposits characterized by very high grade (about1–5 wt % Sn) and high tonnage, which makes them an idealtarget in the exploration for economically viable primary tinresources (Taylor, 1979; Menzie et al., 1988). However, de-spite their having been mined since antiquity and being welldescribed in the literature, there is still considerable uncer-tainty about key issues related to the genesis of these deposits,such as the source of the tin, the origin of the ore fluids, andthe mechanism(s) of ore deposition. Although it is generallybelieved that the bulk of the metal is derived directly from anevolved granitic melt by being partitioned into a magmatic

aqueous phase, the leaching of crystallized magmatic miner-als by later fluids may be another significant metal source(Taylor, 1979; Lehmann, 1990). In addition, some metal couldhave been derived from the sequences of metasedimentaryrocks, which host the granitic plutons (Williamson et al.,2000; LeBoutillier et al., 2002). A variety of ore fluids havebeen invoked, with emphasis on primary magmatic fluids, butalso including meteoric waters, basinal brines (formation wa-ters), metamorphic fluids, and their mixtures. With regard tothe meteoric fluids, the timing of their circulation and inter-action with magmatic fluids may be relatively early or occuronly in the waning stages of the hydrothermal system(Collins, 1981; Primmer, 1985; Sun and Eadington, 1987;Alderton and Harmon, 1991; Sheppard, 1994; Linnen andWilliams-Jones, 1995; Wilkinson et al., 1995; Smith et al.,

Stable Isotope Constraints on Ore Formation at the San Rafael Tin-Copper Deposit, Southeast Peru

THOMAS WAGNER,†

Institute of Isotope Geochemistry and Mineral Resources, ETH Zurich, NW F 82.4, Clausiusstrasse 25, CH-8092 Zürich, Switzerland

MICHAEL S. J. MLYNARCZYK,Rathdowney Resources Ltd., Killaderry, Daingean, County Offaly, Ireland

ANTHONY E. WILLIAMS-JONES,Department of Earth and Planetary Sciences, McGill University, 3450 University Street, Montréal QC H3A 2A7, Canada

AND ADRIAN J. BOYCE

Scottish Universities Environmental Research Centre (SUERC), East Kilbride, Glasgow, G75 0QF, Scotland, United Kingdom

AbstractThe San Rafael tin-copper deposit in the Eastern Cordillera of the Peruvian Central Andes is the world’s

largest hydrothermal tin lode, with a total resource of about 1 million metric tons metal, at an average grade of4.7 wt percent Sn. The mineralization is of the cassiterite-sulfide type and occurs in a vertically extensive vein-breccia system centered on a shallow-level, late Oligocene granitoid stock. The tin ores form cassiterite-quartz-chlorite–bearing veins and breccias hosted by several large fault-jogs at depth in the lode. By contrast, the cop-per ores, which contain disseminated acicular cassiterite, are localized in the upper part of the system. Bothore types are associated with a very distinctive, strong chloritic alteration, which was preceded by intense seric-itization, tourmalinization, and tourmaline veining. The δ34S values of the sulfides range between 2 and 6 permil, and vary very little with location in the deposit. This indicates that the hydrothermal system was large, witha relatively homogeneous source of sulfur, likely of magmatic origin. This is confirmed by stability relationshipsof ore minerals, which indicate that the ore fluids were initially reducing. Microthermometric studies of fluidinclusions in cassiterite, quartz, tourmaline, and fluorite show that the fluids responsible for the early, barrenstage were hot, hypersaline brines (380°–540°C, 34–62% NaCl equiv), whereas the ore-stage fluids had mod-erate to low salinity (0–21 wt % NaCl equiv), and were of moderate temperature (290°–380°C). In addition tothe marked dilution of the ore fluids with evolution of the hydrothermal system, they became progressivelymore oxidizing, as inferred by the local association of minor hematite with cassiterite and the ubiquitous re-placement of pyrrhotite by pyrite and marcasite. The δ18O values of the fluid decreased systematically withtime, as indicated by the δ18O values of different generations of tourmaline, cassiterite, and quartz. This evo-lution was paralleled by an increase in the δD values of the fluid, inferred from the δD values of tourmalineand chlorite. This trend is consistent with mixing of the ore fluids with a cooler fluid that had substantially lowerδ18O, and cannot be explained by fluid boiling. Based on structural evidence for an opening of the vein systemand a transition from lithostatic to hydrostatic conditions at the onset of mineralization, we infer that ore de-position was caused by an influx of hot groundwater of meteoric origin which mixed repeatedly with tin-bear-ing magmatic brines. The oxidation, dilution, cooling, and acid neutralization of the ore fluids destabilized chlo-ride complexes of tin and triggered the large-scale precipitation of cassiterite.

† Corresponding author: e-mail: thomas.wagner@erdw.ethz.ch

©2009 Society of Economic Geologists, Inc.Economic Geology, v. 104, pp. 223–248

1996; Walshe et al., 1996; Jackson et al., 2000). Finally, thefollowing mechanisms have been proposed for the precipita-tion of cassiterite: cooling, pressure decrease, wall-rock alter-ation, boiling, and fluid mixing (Kelly and Turneaure, 1970;Eadington, 1985; Heinrich and Eadington, 1986; Heinrich,1990, 1995; Linnen and Williams-Jones, 1995). Among these,mixing between magmatic and meteoric waters is thought tobe the most effective means of producing the unusually highore grades of some deposits (Heinrich, 1990). However, thishypothesis has not been properly tested because the contri-bution of meteoric water to hydrothermal systems depositinglode tin mineralization has been poorly documented.

San Rafael, the world’s richest primary Sn-Cu deposit(Kontak and Clark, 2002; Mlynarczyk et al., 2003), which islocated in the high Andes of the Eastern Cordillera of south-eastern Peru and is currently exploited by MINSUR S.A., isan ideal natural laboratory in which to investigate the variousissues of lode tin genesis raised above. With a total resourceof about 1 million metric tons (Mt) Sn, and an average gradeof 4.7 wt percent Sn, the San Rafael lode, which has been ex-posed by mining activity over a vertical extent of more than1,300 m, allows an unprecedented opportunity to advance ourunderstanding of the formation of this important deposittype. In addition, the deposit is very young (about 22–24 Ma;Clark et al., 1983; Kontak et al., 1987; Kontak and Clark,

2002) and, therefore, was not affected by any later geologicprocesses. Following a study of the deposit geology (Mlynar-czyk et al., 2003) and its alteration (Mlynarczyk and Williams-Jones, 2006; Mlynarczyk et al., 2009, submitted), a compre-hensive stable isotope (S, O, H, C) investigation of ore andgangue minerals and a preliminary fluid inclusion study wereperformed to evaluate the source of the fluids and metals, aswell as the timing and the mechanism of ore deposition. Ourdata indicate that ore deposition was caused by an influx ofhot groundwaters of meteoric origin and their mixing with thetin-bearing magmatic brines.

Geologic Setting

Regional geology

The San Rafael deposit lies in the northwesternmost exten-sion of the Central Andean Sn-W(-Ag) metallogenic province(e.g., Turneaure, 1960a, b; Kelly and Turneaure, 1970;McBride et al., 1983; Mlynarczyk and Williams-Jones, 2005)and crops out on the glaciated slopes of the CordilleraCarabaya, which forms part of the Eastern Cordillera ofsouthern Peru (Fig. 1). The geology of the region is domi-nated by a thick sequence of Lower Paleozoic, marine clasticmetasedimentary rocks, which experienced extensive crustaldeformation in the Cenozoic. The unexposed Precambrian

224 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 224

N

SAN RAFAEL

JORGE

GUILLERMO

VICENTE

CATALINA

VICTO

RIA

PEDRO

MA

RIA

NO

ANDES

PERUANOS

MA

RIA

EL

EN

A

MARIANELA

500 m

Lima

PA

CIF

IC O

CE

AN

PERU

SAN RAFAEL

Quartzites (SandÌa Fm.)

Slates and hornfelses (SandÌa Fm.)

Phyllites (SandÌa Fm.)

Meta-sediments (Upper Ordovician):

Fine-grained granitoidsDyke of granitoid porphyry

Coarse-grained granitoids

Tourmalinized granite

Granitoids (Late Oligocene):

Access to the mine

Major faults

Mineralized veins

Structures:

FIG. 1. Geologic map of the western part of the San Rafael district (modified after Arenas, 1980).

gneissic basement is covered by a 10- to 15-km-thick se-quence of Ordovician to Siluro-Devonian metapelites andmetapsammites (the San José, Sandía, and Ananea Forma-tions), followed by about 3 to 4 km of late Paleozoic psam-mites and carbonates (Mississippian Ambo Group, Pennsyl-vanian Tarma Group, and Permian Copacabana Group) and 3km of mid-Permian to Triassic red beds and intercalated vol-canics (Mitu Group). These rocks are, in turn, overlain byroughly 1 km of Cretaceous psammites and carbonates (Co-tacucho Group) and 800 m of Miocene-Pliocene felsic ign-imbrites and red beds (Crucero Supergroup). A detailed re-view of the geology of the area is provided by Laubacher(1978), Clark et al. (1990), and Sandeman et al. (1996).Granitic and granodioritic plutons (some of which host signif-icant Sn-W-base metal mineralization) were emplaced be-tween the late Devonian to early Carboniferous (Kontak andClark, 1988) and the late Tertiary (Kontak et al., 1987; Kon-tak and Clark, 2002). Examples of these are the large Permo-Triassic granitic batholiths, which host the Sarita polymetallicdeposit, and the middle- to late-Tertiary granitic stocks,which host the San Rafael tin-copper and Palca XI tungstendeposits (Clark et al., 1983; Kontak and Clark, 1988; Kontaket al., 1990). The peraluminous Tertiary granitic intrusionsare thought to originate from deep-crustal partial melting ofmetasediments related to magmatic underplating (Sandemanet al., 1995; Kontak and Clark, 2002). In addition to grani-toids, there are also subordinate peralkaline syenites and gab-bro-diorites. However, most rocks of mafic composition arevolcanic and form localized occurrences of shoshonite,minette, absarokite, and high K calc-alkaline basalt (Kontaket al., 1986; Sandeman et al., 1995).

Deposit geology

The bulk of the tin in the San Rafael-Quenamari miningdistrict is hosted by a single, large vein-breccia system calledthe San Rafael lode. This lode is part of an array of a fewdozen subparallel veins that host Sn-(W)-Cu-Zn-Pb-Ag min-eralization. Typically, the veins are planar, 500 to 3,500 mlong, have a northwestern strike and dip moderately tosteeply to the northeast. They are centered on a small (about15 km2), high-level, late Oligocene granitoid pluton and strad-dle the contact with the surrounding Lower Paleozoic meta-sedimentary rocks, which consist dominantly of Ordovicianphyllites and quartzites of the Sandía Formation (Fig. 1). Thepluton is a polyphase intrusion composed of coarse- tomedium-grained K-feldspar megacrystic, cordierite-biotite-(±garnet-sillimanite) granite, leucogranite and granodiorite,with local minor enclaves of tonalite and quartz diorite. Themagma was strongly peraluminous, S-type in character, andclearly formed due to the partial melting of metasediments(Mlynarczyk et al., 2003). However, it was not geochemicallyevolved, and the different igneous phases do not represent afractional crystallization series (Dolejs et al., 2009, submit-ted). The pluton is elongated to the northeast and corre-sponds to a prominent topographic high (Nevado Quenamari,5,300 m), but is only exposed locally, although its horizontalextent is clearly evident from the distribution of the overlyingcontact metamorphic rocks (hornfelses and slates). The avail-able geochronological data (Kontak and Clark, 2002) demon-strate that the pluton was emplaced at 24.6 ± 0.2 (U-Pb zircon

age) to 24.7 ± 0.2 Ma (U-Pb monazite age). Ar/Ar data showthat the upper part of the stock had cooled down to the clo-sure temperature of biotite (about 300°–350°C: Harrison etal., 1985) at about 23.7 ± 0.2 Ma. The main Sn-Cu mineral-ization appears to be slightly younger than this cooling age,based on Ar/Ar ages ranging between 22.7 ± 0.7 and 21.9 ±0.2 Ma, obtained from hydrothermal adularia and muscovite(Clark et al., 1983; Kontak et al., 1987; Kontak and Clark,2002).

The San Rafael lode crops out on the eastern flanks of theglacier-capped Quenamari mountain, at an altitude between4,500 and 5,100 m, and has a length of over 3 km. It has beenexposed by mining over a vertical extent of about 1,300 m(from 5,100 to 3,800 m above sea level). Current mining ac-tivity is restricted to the lower 500 m of the deposit. Theupper part of the lode is hosted by metasedimentary rocks,attains a width of up to 2 m, and contains appreciable chal-copyrite and subordinate needle-tin cassiterite. By contrast,the lower part of the lode is hosted by granitic rocks, opensinto a series of subvertical shoots up to 50 m wide, and con-tains a large volume of high-grade cassiterite ore (Fig. 2).The ore shoots are composed of sets of veins that have thick-nesses in the range of 5 to 30 cm. Quartz and chlorite are theprincipal gangue minerals, and are common in both theupper and lower parts of the deposit. The lode has an aver-age strike of 330°, dips 40° to 75°NE, and is texturally quitecomplex, having undergone multiple episodes of vein re-opening and brecciation. Mlynarczyk et al. (2003), who pro-vided a detailed description of the deposit geology, proposedthat the large ore shoots at depth in the lode represent faultjogs formed during sinistral-normal strike-slip movementalong the plane of the lode, which was synchronous with cir-culation of the ore fluids.

Alteration and Vein ParagenesisA detailed campaign of core-logging and underground

mapping in the San Rafael lode outlined as many as 15 dis-tinct vein types, which were arranged chronologically, basedon crosscutting relationships (Mlynarczyk et al., 2009, sub-mitted). The mineralogy of these veins and the associatedwall-rock alteration are summarized briefly below, within theframework of four principal paragenetic stages, adopted froma subdivision originally proposed by Palma (1981).

(I) Early, barren tourmaline stage

The early, barren stage was initiated by pervasive sericiticalteration, followed by strong tourmalinization, which pro-duced wide alteration halos around early tourmaline-quartzveins and tourmaline stringers. These early veins, which cutthrough large parts of the granitic pluton, are typically sealedand lack cassiterite. Locally, however, they host abundant ar-senopyrite. The tourmaline is mainly dravite and Fe-richdravite, but other varieties (foitite, Mg foitite) were also ob-served. A detailed description of the mineralogy, chemistry,and temporal relationships of the different varieties of veinand alteration tourmaline can be found in Mlynarczyk andWilliams-Jones (2006).

In addition to the tourmaline-rich veins, laterally extensive(up to several meters wide) veins of hydrothermal micro-breccia are very common. These are composed of strongly

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 225

0361-0128/98/000/000-00 $6.00 225

tourmalinized rock flour and subordinate quartz, generallyoccur in series, and contain the bulk of the tourmaline in thedeposit. Commonly, they are crosscut by a dense swarm ofyounger quartz veins and are associated with subordinate sili-cification which, locally, could have been coeval with tourma-line precipitation. Both the tourmaline-rich veins and micro-breccia veins occur throughout the vertical extent of the lode,although they are more abundant in its central and upperparts. Petrographic observations indicate that reopening of theearly tourmaline veins was quite common, and that there weremany episodes of brecciation. It is also apparent that some ofthe major tourmaline-bearing veins, which clearly follow thelode along strike and were responsible for establishing its ini-tial structural framework, subsequently reopened during theore stage (Mlynarczyk et al., 2003). The principal chemicalchanges associated with the strong sericite-tourmaline alter-ation were a nearly complete removal of alkalis and alkali-earth elements (Na, K, Ca, Ba, Rb, Sr, Cs, Li, and Cl) from thewall rock and a marked addition of B, Mg, and Fe, and possi-bly Sn, W, and In, though the latter could represent a lateoverprint (Mlynarczyk et al., 2009, submitted).

(II) Main cassiterite stage

The initiation of the ore stage was related to a transitionfrom a relatively closed to an open vein system, as all ore andlater vein types are characterized by textures consistent withfilling of open fractures. The earliest phase of the ore stage ismarked by a rare type of tourmaline-cassiterite-chlorite-ar-senopyrite vein, in which an obvious but continuous transitionin mineral chemistry is observed. An early variety of orange-colored dravite was progressively replaced by green-coloredschorl (a strongly Fe-rich tourmaline), and was accompaniedby simultaneous precipitation of cassiterite and chlorite, fol-lowed by massive arsenopyrite (Mlynarczyk and Williams-Jones, 2006).

The majority of the veins and breccias of the main tin orestage, however, are characterized by a simple mineralogy that

consists of cassiterite, chlorite, and quartz (Fig. 3a). Cassi-terite occurs mainly as massive, dark-brown, locally botryoidallayers (up to 7 cm wide) or as beige-colored, collomorphicwood tin. Chlorite is dark green and moderately to stronglyenriched in iron, corresponding compositionally to ripidoliteand daphnite (Mlynarczyk et al., 2009, submitted). Locally,the cassiterite-chlorite-quartz veins also contain minor pro-portions of Fe-rich wolframite or arsenopyrite (Fig. 3b),which is the essential sulfide mineral associated with thisstage. The larger, high-grade veins are complexly banded,with multiple generations of alternating quartz, cassiterite,and chlorite, that commonly record a consistent temporal se-quence of mineralization that can be distinguished on thebasis of crystal morphology and color (Fig. 4). Some of theseveins have quartz-rich centers with a high proportion of sul-fides (e.g., chalopyrite, arsenopyrite, pyrite) that were formedduring the later sulfide stage, as deduced from crosscuttingvein relationships. Finally, traces of hematite are associatedwith chlorite and cassiterite in the brecciated ores at about anelevation of 4,400 m. No tourmaline occurs in the cassiterite-chlorite-quartz veins, but hairline veinlets of a late, Fe-richvariety of dark-blue or dark-green tourmaline crosscut theearly tourmaline veins and microbreccias of the barren stage.The ubiquitous cassiterite-chlorite-quartz veins are sur-rounded by wide halos of strong chloritic alteration and dis-play evidence of frequent reopening (crack-and-seal textures)and cockscomb textures, producing a characteristic recurrentalternation of cassiterite, chlorite, and quartz layers.

The bulk of the tin ore is restricted to the lower, granite-hosted part of the San Rafael lode, especially the wide, sub-vertical fault-jogs, which form elongated sinuous zones up to50 m wide. The highest ore grades (up to 45 wt % Sn) are as-sociated with several-meter-wide breccia zones, which eithercontain fragments of cassiterite veins cemented by chloriteand quartz, or fragments of chloritized wall rock cementedby cassiterite and quartz. Very high grade cassiterite miner-alization also occurs in the major (1–2 m wide) footwall and

226 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 226

NWSE Nevado Quenamari

Copper stopes

Tin orebodies

500m

5000 m

4500 m

4000 m a.s.l.

Ore ShootContact Orebody

R119R113

R123R55 R57R511

R399

R137R65 R28 R27

R404R573

R165R708R709

R726R725

R428R429

D3-39D3-69

D3-27A13

R17

R729

FIG. 2. Longitudinal section through the San Rafael lode, showing the location of hand samples collected during thisstudy. The location of drill core samples (elevations below about 4,300 m) is indicated in Tables 1 to 3.

hanging-wall veins of the lode structure, as well as in loca-tions where veins are deflected, branch, or intersect. Itshould be emphasized that the bulk of the tin is contained inthe veins, veinlets, and breccias, as opposed to being dissem-inated in the chloritized wall rock (Mlynarczyk et al., 2003,2009, submitted).

The entire structure of the San Rafael lode is enveloped bya 2- to 10-m-wide zone of very strong, texturally destructive,chloritic alteration, which affected the host rocks during thisand the following stage. Chloritization is the dominant type ofalteration in the deposit, affecting its entire vertical extentand strongly overprinting all earlier alteration types. Locally,chloritization is associated with silicification, but the lattermay have been developed late. Studies of the whole-rockchemistry indicate that chloritization leached most of the con-tent of alkali and alkali-earth elements from the granitic wallrocks (Na, K, Ca, Ba, Rb, Sr, Cs, and F, Cl, S) and introducedlarge amounts of Fe, H2O, Mn, Sn, W, In, and Mg (Mlynar-czyk et al., 2009).

(III) Sulfide stage

The bulk of the sulfide ore is restricted to the middle andupper parts of the San Rafael lode, and is most strongly con-centrated where the lode straddles the intrusion-slate contactand within the slates. The sulfides clearly postdated cassi-terite deposition, and are characterized by a great mineralog-ical diversity. The dominant sulfide mineral is chalcopyrite(Fig. 3c), which is associated with subordinate amounts ofneedle-tin cassiterite and stannite, yielding a low grade (0.5–1wt %) tin ore. Other sulfides, which locally are very abundant,are pyrrhotite, arsenopyrite, pyrite, sphalerite, and galena. Inaddition, hematite, marcasite, bismuthinite, and native bis-muth have also been observed. The main gangue minerals arequartz, chlorite, and fluorite. The minerals that formed dur-ing the sulfide stage occur either in distinct quartz-sulfideveins (commonly with a marked predominance of eitherquartz or sulfides) or as a late filling (or reopening) of stage IIcassiterite-quartz-chlorite veins. The structural orientation of

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 227

0361-0128/98/000/000-00 $6.00 227

a

qzchl

castou

5.5 cm

cp

qz

c 3.5 cm d

cpsp

3.5 cm

qz

b

qz

cas

caschl

cp spapy

2.5 cm

FIG. 3. Photographs of representative ore hand specimens. a. Example of typical tin ore from a fault-jog in the lower partof the deposit. Layers of massive, dark-brown cassiterite (cas) overgrown on fragments of early quartz (qz)-tourmaline (tou)breccia and chloritized (chl) wall rock (sample SAR-R464, contact orebody, level 4350 m). b. Complex, open space-filling;quartz-cassiterite-sulfide vein (the arrow indicates the direction of growth). Near the strongly chloritized margin of the veinthere are alternating layers of cassiterite, chlorite, and quartz. The center of the vein consists of large quartz crystals cappedby arsenopyrite (apy), with minor grains of sphalerite (sp) and chalcopyrite (cp) (sample SAR-R511, ore shoot, level 4310 m).c. Typical copper ore from the upper part of the deposit, composed dominantly of chalcopyrite, with subordinate quartz andpyrite (sample SAR-R710, San Rafael lode, level 4820 m). d. Broken fragments of sphalerite and chalcopyrite, cemented byquartz, in the central part of a quartz-chalcopyrite vein (sample SAR-R722, San Rafael lode, level 4730 m).

the veins formed during the sulfide stage is the same as thatof stage II veins, suggesting a common mechanism of forma-tion. Moreover, the sulfide stage veins are associated with thesame, pervasive chloritic alteration as the stage II veins (Mly-narczyk et al., 2003). In addition, the lode hosts breccias,which either contain sulfide fragments (Fig. 3d) or are ce-mented by sulfides. It also hosts large (10–30 cm) vugs filledby pyrite, marcasite, and rare hematite. We infer from the oc-currence of cubes of pyrite (several centimeters in diameter)

inside large vugs in massive pyrrhotite ore, and siderite vein-lets cutting across pyrrhotite, that most of the pyrite, marca-site, and siderite originated by hypogene replacement ofpyrrhotite.

Mineralogical and textural observations suggest that thesulfide ores of San Rafael formed at comparatively high tem-peratures, as indicated by the widespread presence of well-developed sphalerite stars in chalcopyrite (Wiggins and Craig,1980; Kojima and Sugaki, 1984; Sugaki et al., 1990) and com-mon sphalerite blebs enclosed in other sulfides. Examples ofsome of these ore textures are shown in Figure 5. However,although specific paragenetic sequences can be determinedfor individual samples, these parageneses generally cannot beextended to the entire deposit, as most sulfides (aside frompyrite and marcasite) occur in multiple generations and arebroadly coeval.

(IV) Late, barren quartz-carbonate stage

The last stage of veining is represented by numerous bar-ren quartz ± carbonate (calcite, siderite) veins with traceamounts of chalcopyrite and sphalerite but no cassiterite.They have narrow, strongly chloritized margins. These veinscrosscut all other types of veins and breccias, locally produc-ing dense stockworks, and are commonly quite thick (0.2–1m), implying massive precipitation of quartz in the waningstages of the hydrothermal system. On this basis, the bulk ofthe silicification observed in the deposit is tentatively assignedto this stage.

In addition to the four hypogene stages, a fifth, supergenestage is well developed in the uppermost, copper-rich part ofthe lode. This resulted in the alteration of the primary ores toFe oxides and hydroxides (limonite) and secondary Cu min-erals such as covellite and chalcocite. No secondary enrich-ment of tin minerals was observed.

Fluid inclusionsWe carried out a reconnaissance fluid inclusion study of the

San Rafael deposit in order to provide important constraintson the temperature and composition of the ore-forming flu-ids. The minerals studied were quartz, cassiterite, tourmaline,and fluorite, for which over two dozen fluid inclusion sectionswere made. Microthermometric measurements were per-formed at McGill University on a Linkam THMSG-600 fluidinclusion stage, using synthetic fluid inclusions (SYNFLINC)as temperature standards (calibration temperatures: –56.6°,–21.2°, –10.7°, 0.0°, and 374.1°C). Heating rates were0.1°C/min when phase transitions were approached. Errorswere ±0.2° and ±5°C for final ice melting and homogeniza-tion temperatures, respectively. The total salinity of the inclu-sion fluids (expressed as wt % NaCl equiv) was calculatedusing the equation of Bodnar (1993) for low- and moderate-salinity inclusions, and the equation of Sterner et al. (1988)for hypersaline inclusions. Fluid inclusion densities and iso-chores were calculated using the computer package FLUIDS(Bakker, 2003).

Fluid inclusion petrography

The studied samples contain abundant fluid inclusions,ranging in diameter between 2 and 200 µm, many of whichclearly represent different generations. Representative fluid

228 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 228

3 cm

3.3

2.23.0

2.7

3.4

10.6

3.3

2.910.6

9.7

10.14.6

qz

chl

cas

py

cp

R 119

10.8

9.8qz

cas

qz

cas3

qz4

cas2

qz1

qz2qz3

cas4

qz5

qz6

qz7

cas5

qz8

FIG. 4. Photograph of a large, banded cassiterite-quartz-chlorite-chal-copyrite vein (sample R 119, contact orebody, level 4310 m). Also shown arethe locations of crystals analyzed for their oxygen isotope composition. Thearrow indicates the direction of vein growth; different generations of quartzand cassiterite are identified by labels at the left side of the figure. Note thatthe oldest cassiterite generation (cas1) is not shown in the photograph, but ispresent in the sample. The minerals analyzed are quartz (white label), cassi-terite (yellow), chlorite (green) and wolframite (pink).

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 229

0361-0128/98/000/000-00 $6.00 229

a

cas

chl qz

500 µm

cp

py

sp

qz

h 300 µm

c

cas

cp

py

py

po

stn

sph

chl 300 µm

cp

py

sp

qz

f 200 µm

cp

py

stnsp

qz

g 80 µm

b

cas

chl

cp

qz

580 µm

cp

mc

po

py

apy

sp

gne 580 µm

d

cas

cp

po

apy

py

sp

chl

460 µm

FIG. 5. Photomicrographs in reflected light showing representative textures of ore assemblages. a. Alternating layers ofchlorite (chl), cassiterite (cas) and quartz (qz), typical of the rich tin ores hosted by the fault-jogs in the deeper parts of thedeposit (sample R119, vein selvage, contact orebody, level 4310 m). b. Aggregates of needle-tin cassiterite crystals (seen incross section), intergrown with chlorite, quartz, and chalcopyrite (cp) in ore from the upper part of the deposit (sample R725,San Rafael lode, level 4730 m). c. Needle-tin cassiterite, rimmed and crosscut by stannite (stn), associated with chlorite, chal-copyrite, and pyrite (py), which contains inclusions of sphalerite (sp) and pyrrhotite (po) (sample R119, center of the vein,contact orebody, level 4310 m). d. Euhedral arsenopyrite (apy) overgrown by chlorite, chalcopyrite, and cassiterite. The hostchalcopyrite contains abundant blebs of pyrrhotite (locally replaced by pyrite) and veinlets of sphalerite (sample R55, oreshoot, level 4330 m). e. Arsenopyrite replaced by sphalerite and galena (gn), and overgrown by chalcopyrite and pyrrhotite.Pyrrhotite is partly replaced by pyrite and marcasite (mc) (sample A1-1, orebody 150, level 4100 m). f. Sphalerite with abun-dant chalcopyrite inclusions (chalcopyrite disease), veined by coarse-grained chalcopyrite, quartz, and pyrite. The latter sul-fides contain fine sphalerite inclusions (sample R28, ore shoot, level 4330 m). g. Sphalerite rimmed by stannite, enclosed inquartz, and neighboring chalcopyrite and pyrite (sample R119, center of the vein, contact orebody, level 4310 m). h. Euhe-dral crystals of late pyrite in small vugs hosted by chalcopyrite, and crosscut by late sphalerite veinlets (copper ore, charac-teristic of the upper levels of the deposit. sample R710, San Rafael lode, level 4820 m).

inclusion assemblages are shown in Fig. 6. The vast majorityof the inclusions are liquid-vapor (LV) and liquid-vapor-solid(LVS) types, with the latter type commonly containing morethan one solid. On the basis of EDS analyses of opened LVSinclusions, these solids are (1) a ubiquitous, highly birefrin-gent mineral, which in some inclusions forms bundles andcould be a hydrous phyllosilicate (chlorite or phengiticsericite-illite), (2) aggregates of radiating birefringent crystals,(3) transparent, platy solids, some of which are birefringent(probable Fe-Al-Na-Ca-bearing chlorides and carbonates),(4) single tiny specks of an equant opaque mineral (likelychalcopyrite), (5) cubes of halite (in hypersaline inclusions),(6) relatively uncommon, finely prismatic crystals of cassi-terite (needle-tin variety), (7) anhedral grains of cassiterite(occurring also as solid inclusions), and (8) bladed crystals ofcalcite (in late quartz-calcite veins). With the exception ofhalite, these minerals do not redissolve on heating and likelyrepresent accidently trapped crystals. In addition to the solid-bearing inclusions, vapor-rich and vapor-only fluid inclusionsare locally present, but these are restricted to the early, bar-ren tourmaline stage (I). No aqueous-carbonic inclusionswere observed in the minerals studied and preliminary gaschromatographic analyses of inclusion fluids indicate very lowproportions of CO2 and CH4.

Because multiple generations of secondary inclusions arevery abundant in most samples and their trails are locally verydense, it appears possible that some of the original primaryfluid inclusions were subsequently refilled. In addition, someof the secondary fractures in the host minerals (especiallyquartz) have healed very well, producing a scattered distribu-tion of those inclusions, which texturally appear isolated but

are in fact secondary in origin. As a result, clusters of fluid in-clusions displaying highly variable salinities, likely unrelated toprimary entrapment, are quite common. We have addressedthese issues by a particularly careful selection of primary andpseudosecondary fluid inclusions, where we adhered strictly tothe criteria for these fluid inclusion types established by Roed-der (1984) and Goldstein and Reynolds (1994). Moreover, inestablishing the final data set reported in this paper, we re-stricted ourselves to fluid inclusion assemblages, i.e., groups oftexturally well-constrained fluid inclusions that yielded highlyconsistent microthermometric results.

Microthermometric results

Several fluid inclusion assemblages (in the sense of Gold-stein and Reynolds, 1994) in each sample were analyzed mi-crothermometrically using the cycling technique method fordetermination of Tmice and Th (Haynes, 1985). The prelimi-nary results of fluid inclusion microthermometry are plottedon a homogenization temperature (Th) versus salinity diagram(Fig. 7). Although NaCl is the dominant solute, cassiterite-hosted primary and pseudosecondary inclusions contain ap-preciable CaCl2 and lesser amounts of KCl. This was inferredfrom the microthermometric measurements (temperature ofinitial ice melting between –60 and –50°C, temperature ofhydrohalite dissolution close to –30°C) and, more directly, byEDS analysis of fluid inclusion decrepitate mounds. In addi-tion, FeCl2 was also present in the ore fluids, as demonstratedby comparatively low but consistent amounts of FeCl2 de-tected in the analyzed decrepitates.

Stage I quartz- and tourmaline-hosted inclusions have asalinity range of 34 to 62 wt percent NaCl equiv (with most

230 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 230

d 50 µmc 25 µm

b 30 µma 30 µm

FIG. 6. Photomicrographs of representative fluid inclusion assemblages from a complex cassiterite-quartz-sulfide vein(sample R-119, contact orebody, level 4310 m). a. Cluster of primary fluid inclusions in stage II quartz. b. Primary fluid in-clusions in stage II cassiterite developed along growth zones. c. Large primary fluid inclusion in stage II quartz. d. Denselydistributed very low salinity secondary fluid inclusions in stage IV quartz.

data clustering around 41–56 wt %) and homogenize by dis-appearance of either halite or vapor in the range 240° to535°C, but mainly between 340 and 475°C. Based on texturalevidence, it is inferred that the inclusions, which have a veryhigh salinity and homogenization temperature, formed due tolocal boiling of the early fluids, although necking down aftera phase change could also explain the highly variable phaseratios observed within fluid inclusion assemblages. The dataobtained are in broad agreement with those of Kontak andClark (2002).

Stage II, cassiterite-hosted, primary and pseudosecondaryfluid inclusions were measured in several generations of cas-siterite (cas1 to cas3, Fig. 4), and have a very narrow range insalinity (17–21 wt % NaCl equiv) and Th (354°–361°C).Quartz-hosted inclusions from the same stage exhibit consis-tently lower temperature and salinity, which vary dependingon the generation of the host quartz (11–16 wt % NaCl equivand 265°–305°C for qz1 and qz2, and 2.5–5.5 wt % NaClequiv and 265°–295°C for the later qz3 and qz4, Fig. 4). Thesalinity of the fluids which precipitated stage II quartz is likelyto have spanned the entire range between 2.5 and 16 wt per-cent NaCl equiv, as suggested by the behavior of fluidstrapped in secondary inclusions. However, the most remark-able feature of the quartz-hosted inclusions from this stage isa population of inclusions of very low salinity (Tmice as high as–1.3°C), which are quite similar in composition and tempera-ture to the fluids that precipitated the barren quartz fromstage IV (Fig. 7). In addition, the character of the ore veins,in which bands of cassiterite, chlorite, and quartz repetitivelyalternate, and the consistently lower homogenization temper-ature and salinity of the fluid inclusions trapped in quartz

bands relative to those in cassiterite layers indicate marked,periodic fluctuations in the composition of the fluids circulat-ing in the vein system.

Primary inclusions in stage III quartz and fluorite are char-acterized by a salinity in the range of 0.5–13 wt % NaCl equiv(mostly 3–6 wt %) and a Th in the range of 265°–340°C. Sig-nificantly, primary and pseudosecondary inclusions in quartzassociated with chalcopyrite in the upper zone of the deposit(sample R709, elev 4,820 m) have, on average, a Th lower byabout 40°C than inclusions hosted by a very similar variety ofquartz associated with the chalcopyrite-rich centers ofdeeper-seated stage II-III veins (sample R119, elev 4,310 m).Another important observation is the lower Th limit of about265°C consistently displayed by all (primary-secondary) stageII and stage III inclusions. Fluid inclusions hosted by stage IVquartz have the lowest salinity (0–2.5 wt % NaCl equiv) andhomogenization temperature (235°–265°C).

When the microthermometric data for all four parageneticstages are considered jointly (Fig. 7), it is evident that thesets of fluid temperature and salinity recorded for each stageform a broad linear trend of salinity with homogenizationtemperature. Quartz-hosted primary-pseudosecondary stageII inclusions form two populations at lower and salinity thanstage II cassiterite- (the main cassiterite stage) hosted inclu-sions, whereas stage III quartz-fluorite-hosted inclusionshave higher Th values than stage II quartz-hosted inclusions.Overall, the fluid inclusion data point to repeated fluid mix-ing during the evolution of the San Rafael deposit, with thehotter and more saline fluids of the main cassiterite stagemore closely reflecting the composition of the primary orefluid.

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 231

0361-0128/98/000/000-00 $6.00 231

Stage II cassiterite

Stage I quartz / tourmaline

Stage III quartz / fluorite

Stage IV quartz

Stage II quartz

episodic boilin

g

neck

ing

dow

n

30 40 50 60 70200 10

300

400

500

600

200

100

Total salinity (wt.% NaCl eq.)

Hom

ogen

izat

ion

tem

per

atur

e (°

C)

Primary / pseudosecondary LV inclusionsSecondary LV inclusions

LVH inclusions (homogenize by halite dissolution)

LVH inclusions (homogenize by vapor disappearance)

halite

satu

ratio

n cu

rve

FIG. 7. Homogenization temperature versus salinity of fluid inclusions from San Rafael ores. The host minerals investi-gated were quartz, fluorite, cassiterite, and tourmaline. Abbreviations: L = liquid, V = vapor, H = halite.

Mineral GeothermometryMlynarczyk et al. (2009) have tested two empirical chlorite

thermometers (Cathelineau, 1988; Jowett, 1991) on differentsamples of alteration and vein chlorite from stage II of theparagenesis. The chlorite temperatures obtained are in therange of 335°–415°C. The thermometer of Jowett (1991)yields values about 5°–15°C higher than those from that ofCathelineau (1988), and the values for alteration chlorite are,on average, roughly 50°C higher than those for vein chlorite.The theoretical geothermometer of Walshe (1986) yieldstemperatures in the range 295° to 320°C for the alterationchlorite (Mlynarczyk et al., 2009). These temperatures areconsistently below the Th values of most stage II fluid inclu-sions. We have not further considered these results, becausethe thermodynamic model of Walshe (1986) has been shownto be inconsistent with more recent experimental and theo-retical studies (Holland et al., 1998; Vidal et al., 2001; Aja,2002; Aja and Dyar, 2002;). Overall, the results suggest thatchlorite (and, by inference, the associated cassiterite) formedin the temperature range 350°–420°C.

The stage III sulfide ores commonly contain microscopicovergrowths of stannite on sphalerite (Fig. 5g), as well as lesscommon intergrowths between these two minerals. It hasbeen shown that the equilibrium compositions of sphaleriteand stannite can be used successfully as a geothermometer(Nekrasov et al., 1976, 1979; Shimitzu and Shikazono, 1985;Bortnikov et al., 1990). Carefully selected sphalerite-stanniteintergrowths in representative ore samples were analyzed byelectron microprobe at McGill University. Representative re-sults and average data are given in Table 1. The equation ofNakamura and Shima (1982) was used to calculate the for-mation temperature of the intergrowths, which is interpretedto represent the lower temperature limit of stage III ore for-mation. The resulting temperatures are generally in the range270° to 290°C, i.e., they are slightly lower than fluid inclusionhomogenization temperatures from this stage.

Stable Isotope StudiesIn order to test our hypothesis that fluid mixing was the

dominant ore-forming process, we performed a comprehen-sive stable isotope study on the main ore and gangue miner-als. Representative samples from the principal mineralizationstages (stages I–IV) of the San Rafael deposit were selected

for sulfur, oxygen, hydrogen, and carbon isotope analysis. Thesamples cover a wide range of textures of hydrothermal sul-fides, oxides, and silicates, and permit characterization of theisotopic evolution of the system along the paragenetic se-quence. In addition to samples that cover the different min-eralization stages, we also performed selected small-scalestudies of individual veins employing a high sample density toresolve isotopic effects at the micro- and mesoscales.

Analytical techniques

Separates of hydrothermal carbonate (siderite, calcite),oxide, and silicate minerals (quartz, cassiterite, wolframite,chlorite, tourmaline) were prepared by careful handpickingunder a binocular microscope, followed by cleaning in doublydistilled water. Sulfide minerals (pyrite, pyrrhotite, arsenopy-rite, chalcopyrite, sphalerite, galena) were analyzed by in situlaser combustion from standard polished blocks. Whole-rockoxygen isotope compositions were analyzed from rock pow-ders prepared by standard crushing and milling techniques.All mineral separates and whole-rock samples were dried at80°C for at least 6 hours prior to isotopic analysis. The sulfur,oxygen, and carbon isotope analysis was conducted in the sta-ble isotope lab at SUERC (East Kilbride, UK), whereas hy-drogen isotope analysis was carried out at the isotope labora-tory of the Department of Geological Sciences at QueensUniversity (Kingston, Canada).

Sulfide minerals were combusted in an oxygen atmosphereusing a SPECTRON LASERS 902Q CW Nd:YAG laser (1 Wpower); typical spot diameters in this study were 200 to 400µm. Laser extraction was followed by cryogenic purificationof the SO2 gas and subsequent on-line mass-spectrometricanalysis. Details of the laser extraction technique, calibration,and correction procedures are provided by Kelley and Fallick(1990), Kelley et al. (1992) and Wagner et al. (2002). Lasercalibration data for arsenopyrite were taken from the study ofWagner et al. (2004). Reproducibility of the analytical results,and mass spectrometer calibration, were monitored throughreplicate measurements of international standards NBS-123(δ34SV-CDT: 17.1‰) and IAEA-S-3 (δ34SV-CDT: –31.0‰), aswell as the internal lab standard CP-1 (δ34SV-CDT: –4.6‰).The analytical precision (1σ) was about ±0.2 per mil. All sul-fur isotope compositions are reported in standard delta nota-tion relative to V-CDT.

232 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 232

TABLE 1. Composition of Sphalerite-Stannite Intergrowths and Equilibrium Temperatures Calculated Using the Geothermometer of Nakamura and Shima (1982)

Sample no. Location Mineral Cu (wt%) Fe (wt%) Zn (wt%) Sn (wt%) S (wt%) Total XFe/XZn T (°C)

R27 Ore shoot 4330 m Sphalerite 0.13 11.14 54.55 0.04 33.19 99.05 0.20 282Stannite 28.34 14.94 2.11 25.56 30.13 101.08 7.08

R119 Contact orebody 4310 m Sphalerite 0.20 11.79 53.69 0.01 33.45 99.14 0.22 277Stannite 27.93 15.24 1.80 24.99 30.02 99.98 8.47

R447 Contact orebody 4270 m Sphalerite 0.51 11.47 52.80 0.02 33.05 97.85 0.22 269Stannite 28.25 14.40 1.43 26.51 29.62 100.21 10.07

R404 S contact orebody 4310 m Sphalerite 0.10 12.70 52.83 0.01 34.20 99.84 0.24 305Stannite 28.53 13.75 2.59 26.85 30.41 102.13 5.31

A1-1 No. 150 orebody 4100 m Sphalerite 0.82 13.69 50.6 0.00 33.91 99.02 0.27 288Stannite 28.27 14.23 1.69 25.4 29.52 99.11 8.42

R65 Ore shoot 4390 m Sphalerite 0.42 11.32 53.55 0.07 33.61 98.97 0.21 269Stannite 28.52 13.95 1.42 26.32 30.10 100.31 9.82

Oxygen was extracted from mineral separates and whole-rock powders by reacting 1 to 5 mg of sample with purifiedchlorine trifluoride in a laser fluorination system, based ontechniques of Sharp (1990) and Mattey and Macpherson(1993). The oxygen was converted to CO2 by reaction with aheated graphite rod; the isotopic composition of the cryo-genically purified CO2 gas was measured on-line with a VGPRISM III mass spectrometer. Analytical precision was con-trolled through replicate measurements of the internal labo-ratory standard SES-2 (δ18OV-SMOW: 10.2‰) during thecourse of the study. The latter was calibrated against interna-tional standards IAEA-NBS-28 (δ18OV-SMOW: 9.6‰) andIAEA-NBS-30 (δ18OV-SMOW: 5.1‰). Precision (1σ) was foundto be better than ±0.2 per mil for the whole dataset. Some ad-ditional tourmaline samples were analyzed by conventionaltechniques following the procedures of Clayton and Mayeda(1963). All O isotope data are reported relative to V-SMOW.

Hydrothermal carbonates were analyzed for their C andO isotope composition using an Analytical PrecisionAP2003 continuous-flow mass spectrometer, equippedwith an automated carbonate preparation system. About 1mg of sample powder was placed in a 6 ml vacutainer, thensealed and loaded onto the autosampling unit. Each sam-ple was flushed with helium, then a predeterminedamount of 93 percent phosphoric acid was injected intoeach tube, following the procedure of McCrea (1950). Theacid reaction was conducted at a temperature of 70° ±0.1°C; reaction times were 24 hours for pure calcite sam-ples and 120 hours for all other carbonates. After comple-tion of the reaction, the samples were transferred to theprocessing system and analyzed with an AP2003 massspectrometer. Oxygen isotope data for siderite were cor-rected using an acid fractionation factor calculated fromRosenbaum and Sheppard (1986). Reproducibility of the

analytical results was monitored through replicate measure-ments of the internal Mab2b standard (δ13CV-PDB: 2.48‰ ;δ18OV-PDB: –2.40‰) before and after each batch of samples.Accuracy was controlled by replicate measurements of inter-national standards IAEA-CO-1 (δ13CV-PDB: 2.48‰; δ18OV-PDB:–2.44) and IAEA-NBS-19 (δ13CV-PDB: 1.95‰; δ18OV-PDB:–2.20 ‰). External precision (1σ) was found to be better than± 0.2 ‰ for both carbon and oxygen isotope compositions.Carbon and oxygen isotope data are reported relative to V-PDB and V-SMOW, respectively.

Hydrogen isotope analysis of chlorite and tourmaline wascarried out by elemental analysis continuous-flow isotoperatio mass spectrometry (CF-IRMS), using a high-tempera-ture reduction method (Sharp et al., 2001). This method in-volves passing the sample in an He stream through a carbon-packed furnace heated to 1450°C, and releasing the water,which is then reduced to H2. Hydrogen gas is purified bypassing it through a 5A molecular sieve gas chromatographiccolumn, and subsequently analyzed with a Finnigan MatDelta XL Plus gas source mass spectrometer. Precision (1σ)was found to be better than ±3 per mil. The hydrogen isotopedata are reported relative to V-SMOW.

Sulfur isotope data

A total of 64 in situ laser sulfur isotope analyses were per-formed on polished blocks from 20 samples, spanning boththe paragenesis and the vertical extent of the deposit. Themajority of the samples are from stage III (sulfide stage),where sulfides are abundant and generally several sulfideminerals are present, in contrast to stages I and II for whichsulfides are scarce and are represented only by arsenopyrite.The sulfides analyzed were arsenopyrite, chalcopyrite,pyrrhotite, pyrite, sphalerite, and galena. Typically, severalcoexisting sulfides were analyzed in each sample (Fig. 8). The

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 233

0361-0128/98/000/000-00 $6.00 233

qz

R 404

5.95.3

4.23.8

gn

py

sp

5.4

3.6

qz

tou3.7

apy

D3-27

sp

4.83.7

4.0

5.0

qz

J 410

po

pypo

po

1.4

6.3

3.0

2.45.1

1.24.8

5.4

R 428

sp

cp

qz

gn

sp

gn

gn

sp

cp

5.0

5.5

5.1

qz

cp

py

R 119

cp

py

4.4

3.9

qz

cp

R 511a

apy apy

apy

FIG. 8. Images showing millimeter-scale textural relationships and sulfur isotope data for representative polished sections.The average diameter of the polished blocks is 2.5 cm. (apy = arsenopyrite, cp = chalcopyrite, gn = galena, po = pyrrhotite,py = pyrite, qz = quartz, sp = sphalerite, tou = tourmaline).

234 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 234

TABLE 2. Sulfur Isotope Data for the San Rafael Sn-Cu Deposit (n = 64)

Sample no. Mineral Textural relationships δ34SV-CDT (‰)

Early, barren stageD3-39-1 Arsenopyrite Subhedral, in early tourmaline-quartz vein (DDH no. 3 to the Mariano vein, 39 m) 2.0D3-39-2 Arsenopyrite Subhedral, in early tourmaline-quartz vein (DDH no. 3 to the Mariano vein, 39 m) 3.0D3-27-1 Arsenopyrite Subhedral, in very early quartz veinlet (DDH no. 3 to the Mariano vein, 27 m) 3.6D3-27-2 Sphalerite Anhedral, with arsenopyrite, in early quartz veinlet (DDH no. 3 to the Mariano vein, 27 m) 5.4D3-27-3 Arsenopyrite Subhedral, with tourmaline, in early quartz veinlet (DDH no. 3 to the Mariano vein, 27 m) 3.7D8-50-1 Arsenopyrite Early veinlets in slate, associated with tourmaline and quartz (DDH #8, 50.5 m) 3.8D8-50-2 Arsenopyrite Early veinlets in slate, associated with tourmaline and quartz (DDH #8, 50.5 m) 3.7D3-182b-1 Arsenopyrite Early zone, massive, in tourmaline-cassiterite vein (DDH no. 3 to the Mariano vein, 182.5 m) 2.4D3-182b-2 Arsenopyrite Late zone, massive, in tourmaline-cassiterite vein (DDH no. 3 to the Mariano vein, 182.5 m) 6.2D3-182b-3 Arsenopyrite Late zone, massive, in tourmaline-cassiterite vein (DDH no. 3 to the Mariano vein, 182.5 m) 3.6

Sulfide ore stage511b-1 Arsenopyrite Large subhedral crystals (early zone) center of quartz-chlorite-cassiterite-arsenopyrite vein 3.9511b-2 Sphalerite Anhedral grain, with arsenopyrite, center of quartz-chlorite-cassiterite-arsenopyrite vein 4.2511a-1 Arsenopyrite Large subhedral crystals (latest zone) center of quartz-chlorite-cassiterite-arsenopyrite vein 3.9511a-2 Arsenopyrite Large subhedral crystals (late zone) center of quartz-chlorite-cassiterite-arsenopyrite vein 4.4A13-1 Galena Inclusions in massive pyrite ore 5.4A13-2 Pyrite Coarse-grained massive ore 4.0R729-1 Pyrite Narrow veinlet in chalcopyrite ore 3.8R729-2 Chalcopyrite Coarse-grained, massive ore 2.3R17-1 Chalcopyrite Vug fillings in quartz vein 4.2R17-2 Pyrite Anhedral grains in quartz 4.7R573-1 Chalcopyrite Rim around pyrrhotite 3.7R573-2 Pyrrhotite Massive vein, with quartz 3.0R573-3 Pyrrhotite Massive vein, with quartz 3.1R404-1 Pyrite Fine-grained, replacing pyrrhotite, in quartz vein 3.8R404-2 Pyrite Euhedral, replacing fine pyrite in quartz vein 4.2R404-3 Sphalerite Subhedral, in quartz vein 5.9R404-4 Galena Subhedral, in quartz vein 5.3R165-1 Chalcopyrite Coarse-grained massive ore 4.8R165-2 Pyrite Euhedral crystals in chalcopyrite ore 4.9R165-3 Chalcopyrite Coarse-grained massive ore 3.9G4-1 Arsenopyrite Euhedral crystals in massive chalcopyrite ore (San German vein, level 4650 m) 3.9G4-2 Pyrrhotite Veinlets crosscutting chalcopyrite (San German vein, level 4650 m) 3.4G4-3 Chalcopyrite Coarse-grained massive ore (San German vein, level 4650 m) 3.7G4-4 Pyrrhotite Veinlets crosscutting chalcopyrite (San German vein, level 4650 m) 4.0R428-1 Galena Subhedral inclusions in sphalerite, in quartz-chalcopyrite-sphalerite vein 1.4R428-2 Chalcopyrite Coarse-grained, overgrowing quartz, in quartz-chalcopyrite-sphalerite vein 3.0R428-3 Sphalerite Coarse-grained, overgrowing chalcopyrite, in quartz-chalcopyrite-sphalerite vein 6.3R428-4 Galena Subhedral inclusions in sphalerite, in quartz-chalcopyrite-sphalerite vein 5.1R428-5 Sphalerite Coarse-grained, overgrowing chalcopyrite, in quartz-chalcopyrite-sphalerite vein 5.4R428-6 Sphalerite Coarse-grained, overgrowing chalcopyrite, in quartz-chalcopyrite-sphalerite vein 4.8R428-7 Galena Subhedral inclusions in sphalerite, in quartz-chalcopyrite-sphalerite vein 1.2R428-8 Galena Subhedral inclusions in sphalerite, in quartz-chalcopyrite-sphalerite vein 2.4R726-1 Galena Anhderal inclusions in quartz, in quartz-chalcopyrite vein 3.1R726-2 Sphalerite Subhedral inclusions in quartz, in quartz-chalcopyrite vein 5.6R726-3 Chalcopyrite Massive, coarse-grained ore, with quartz 5.1R119-1 Pyrite Veinlets crosscutting chalcopyrite 5.1R119-2 Chalcopyrite Massive, in the central part of cassiterite-quartz-chlorite-chalcopyrite vein 5.5R119-3 Chalcopyrite Massive, in the central part of cassiterite-quartz-chlorite-chalcopyrite vein 5.0R65b-1 Pyrite Massive, replacing pyrrhotite, in chalcopyrite-arsenopyrite-pyrite vein 3.4R65b-2 Pyrite Massive, replacing pyrrhotite, in chalcopyrite-arsenopyrite-pyrite vein 3.8R65b-3 Pyrite Fine-grained, replacing pyrrhotite, in chalcopyrite-arsenopyrite-pyrite vein 4.5R55-1 Arsenopyrite Euhedral crystal in chalcopyrite, in quartz-chalcopyrite-pyrrhotite-arsenopyrite vein 5.0R55-2 Pyrrhotite Anhedral inclusions in chalcopyrite, in quartz-chalcopyrite-pyrrhotite-arsenopyrite vein 5.8R55-3 Chalcopyrite Coarse-grained, massive ore, in quartz-chalcopyrite-pyrrhotite-arsenopyrite vein 4.9R55-4 Pyrrhotite Anhedral inclusions in chalcopyrite, in quartz-chalcopyrite-pyrrhotite-arsenopyrite vein 5.6R55-5 Arsenopyrite Euhedral crystal in chalcopyrite, in quartz-chalcopyrite-pyrrhotite-arsenopyrite vein 5.1R28a-1 Galena Anhedral crystals in quartz-chalcopyrite-pyrite-sphalerite-galena vein 3.3R28a-2 Sphalerite Anhedral crystals in quartz-chalcopyrite-pyrite-sphalerite-galena vein 3.9R28a-3 Sphalerite Anhedral crystals in quartz-chalcopyrite-pyrite-sphalerite-galena vein 4.5R28a-4 Galena Anhedral crystals in quartz-chalcopyrite-pyrite-sphalerite-galena vein 3.7J410-1 Pyrrhotite Inclusions in large pyrite crystal (Jorge vein, level 4000 m) 3.7J410-2 Pyrite Large, blocky crystal replacing pyrrhotite (Jorge vein, level 4000 m) 4.8J410-3 Pyrrhotite Coarse-grained, massive ore (Jorge vein, level 4000 m) 4.0J410-4 Pyrrhotite Coarse-grained, massive ore (Jorge vein, level 4000 m) 5.0

Notes: All samples were analyzed in situ by a laser combustion system; unless indicated otherwise, the samples are from the San Rafael vein and theirlocation is shown in Figure 2

results of the analyses are listed and illustrated in Table 2 andFigure 9, respectively, and show that the sulfide ores have arelatively uniform δ34S composition, ranging between 2 and 6per mil. Although the range of δ34S values is narrow, as mightbe expected from a deposit associated with a S-type granitoid(Ohmoto and Goldhaber, 1997), minerals within individualsamples commonly display variable δ34S values. For example,the δ34S values of arsenopyrite in sample D3-182b vary withinthe same growth band from 3.6 to 6.2 per mil, whereas theδ34S of galena in sample R428 varies from 1.2 to 5.1 per milwithin only a few millimeters (Fig. 8; Table 2). There is, how-ever, little variation in the δ34S values of the sulfide mineralswith respect to their location in the deposit. Arsenopyrite (theonly sulfide that occurs throughout the entire paragenesis)displays a distinct trend of increasing δ34S values from stage I(2.0–3.8‰) to stage III (3.8–5.1%).

Oxygen isotope data

Thirty-nine oxygen isotope analyses were performed oncarefully handpicked mineral separates of oxides and silicatesfrom 17 samples, including a complexly banded, 5-cm-thick,quartz-chlorite-cassiterite-wolframite-chalcopyrite-pyritevein, which was sampled in detail (Fig. 4). In addition, 5whole-rock samples of granitoids (ranging from fresh tostrongly chloritized) and 12 vein carbonates (see sectionbelow) were analyzed. The oxide and silicate minerals ana-lyzed were tourmaline, quartz, chlorite, cassiterite, and wol-framite, and the results are summarized in Figure 10 andlisted in Table 3. The δ18O values for the individual mineralsare all positive and have relatively narrow ranges of 9.7 to11.9 per mil for tourmaline, 9.7 to 14.4 per mil for quartz, 3.9to 4.6 per mil for chlorite, 1.8 to 3.4 per mil for cassiterite,and 1.7 to 3.0 per mil for wolframite. The transition from theearly, barren stage I to the tin-rich stage II is characterized bya marked decrease in the δ18O values of tourmaline. Withinthe same vein the δ18O value of early, orange-colored tour-maline is 11.7 per mil, whereas the succeeding dark-greentourmaline that is associated with abundant cassiterite andchlorite (the first to form in the paragenesis) has a δ18O valueof 9.7 per mil (sample D3-182, Table 3). In the subsequentstages (II and III), the δ18O values of cassiterite and quartzare variable, but nevertheless provide evidence for an overalldecrease of δ18O with time, which for the ore vein investi-gated in detail corresponds to a change from 3.4 to 2.7 per milfor cassiterite, and from 10.6 to 9.8 per mil for quartz (sampleR 119, Fig. 4; Table 3). The latest cassiterite in the paragene-sis (stage III needle-tin cassiterite, intergrown with chalcopy-rite from the upper zone of the deposit), has an even lowerδ18O value of 1.8 per mil (sample R725, Table 3). By contrast,quartz from the paragenetically latest stage (IV) has markedlyhigher δ18O values, in the range of 11.3 to 14.4 per mil.Whole-rock δ18O values of fresh granitic rocks range from 9.6(biotite-granodiorite) to 10.9 per mil (leucogranite). A se-quence of four slabs of granodiorite with increasing chloriticalteration was investigated in detail and shows a systematicdecrease in δ18O with alteration, down to 8.0 per mil for thestrongly chloritized granodiorite and 4.9 per mil for the chlo-ritite, which occupies the center of the vein (Table 3).

Although a lack of complete isotopic equilibrium betweencoexisting quartz and cassiterite has been reported for a num-ber of hydrothermal tin lodes (Alderton, 1989), several iso-topic geothermometers were tested on the silicate-oxide as-semblages (quartz, cassiterite, wolframite, chlorite) from thebanded ore vein investigated in detail (sample R119, Fig. 4).Because quartz and cassiterite could not always be separatedfrom the same growth band, we calculated oxygen isotopetemperatures based on data for pairs of minerals intimatelyintergrown in a single band, as well as for pairs where inwhich the minerals were in adjacent bands. Overall, the geo-thermometers (quartz-cassiterite, quartz-wolframite, quartz-chlorite), which are all based on carefully conducted experi-ments (Matsuhisa et al., 1979; Zhang et al., 1994; Cole andRipley, 1999) give average equilibrium temperatures in therange of 370° to 450°C, about 40° to 60 °C higher than thefluid inclusion homogenization temperatures. The error on

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 235

0361-0128/98/000/000-00 $6.00 235

Freq

uenc

y

δ34SV-CDT (‰)

8

6

4

2

00 1 2 3 4 5 6 7 8 9 10

Arsenopyritea

6

4

2

00 1 2 3 4 5 6 7 8 9 10

Chalcopyriteb

6

4

2

00 1 2 3 4 5 6 7 8 9 10

Pyrrhotitec

6

4

2

00 1 2 3 4 5 6 7 8 9 10

d Pyrite

6

4

2

00 1 2 3 4 5 6 7 8 9 10

e Sphalerite

6

4

2

00 1 2 3 4 5 6 7 8 9 10

f Galena

FIG. 9. Histograms displaying the sulfur isotope composition of SanRafael sulfides.

the calculated equilibrium temperatures, considering the an-alytical uncertainty of the δ18O values, is on the order of about±30°C. When different stages of the vein evolution are com-pared, the calculated oxygen isotope temperatures (e.g., thequartz-cassiterite pair for which most data are available) showa general decrease from the early stage II assemblage (450°C)to the later complex cassiterite-quartz-wolframite-chlorite as-semblage (400°–420°C).

Hydrogen isotope data

Four tourmaline and two chlorite samples were analyzedfor their hydrogen isotope composition and the results arelisted in Table 4. The δD values of tourmaline lie in a narrowrange between –75 and –73 per mil and are essentially iden-tical within errors. The δD values of the two chlorite samplesare –88 and –81 per mil.

Carbon and oxygen isotope data of carbonates

The oxygen and carbon isotope data for late-stage vein car-bonates (5 samples of calcite and 7 of siderite) are shown inFigure 11 and presented in Table 5. The δ18O and δ13C val-ues of calcite samples are in the range of 8.6 to 14.2 per miland of –8.5 to –3.2 per mil, respectively. The data for sideriteare more variable, and are in the range of 13.8 to 23.1 per miland –6.5 to 4.3 per mil, respectively (Fig. 11). In two samples,

two successive generations of siderite were studied, but didnot yield a consistent trend.

Discussion

Conditions of ore deposition

Based on our geologic and textural observations, the hy-drothermal history of the ore-forming system at San Rafaelcomprised two contrasting episodes. The first of these (stageI) was characterized by the formation of numerous tourma-line-quartz (± arsenopyrite) veins, veinlets, and stringers,with which sericitic and tourmaline alteration are associated.It is important to note that these structures are typically quitenarrow, locally discontinuous, and invariably sealed, likely in-dicating pressures approaching lithostatic. Stratigraphic re-construction based on regional geologic data (e.g., Clark etal., 1990) suggests that about 2.5 km of rock has been erodedsince the emplacement of the deposit (at about 25 Ma). Thistranslates into a lithostatic fluid pressure of 700 to 800 bars,which is consistent with pressure estimates given by Kontakand Clark (2002). Fluid inclusion data demonstrate that thefluids responsible for this early, barren stage were hot, hyper-saline brines (fluid trapping temperatures of 380°–540°Cbased on isochoric projection of fluid inclusion Th values to700–800 bars, salinity of 34–62 wt % NaCl equiv) and the

236 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 236

6

4

2

0

6

4

2

0

6

4

2

0

6

4

2

0

6

4

2

0

Tourmalinea

Freq

uenc

y

δ18OV-SMOW (‰)

0 1 2 3 4 5 6 7 8 9 10 1512 141311 16

0 1 2 3 4 5 6 7 8 9 10 1512 141311 16

0 1 2 3 4 5 6 7 8 9 10 1512 141311 16

0 1 2 3 4 5 6 7 8 9 10 1512 141311 16

0 1 2 3 4 5 6 7 8 9 10 1512 141311 16

b

c

d

e

Quartz

Chlorite

Cassiterite

Wolframite

FIG. 10. Histograms displaying the oxygen isotope composition of San Rafael silicate and oxide minerals.

alteration minerals that formed indicate that they were acidic(Mlynarczyk et al., 2009, submitted).

By contrast, the subsequent stages (i.e., tin-mineralizedstage II, copper-mineralized stage III, and late, barren quartzstage IV) are all associated with a distinctive, strong chloriticalteration (and locally, silicification). The veins invariably have

an open fracture-filling character, with ubiquitous vugs, likelyindicative of hydrostatic conditions. Evidence for multiplevein-opening events suggests that the pressure must havefluctuated from slightly above lithostatic (700–800 bars) tonear hydrostatic (200–300 bars) during ore deposition (as-suming a roughly similar depth to that during stage I). The

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 237

0361-0128/98/000/000-00 $6.00 237

TABLE 3. Oxygen Isotope Data for Quartz, Cassiterite, Wolframite, Chlorite, Tourmaline, and Whole-Rock Samples from the San Rafael Sn-Cu Deposit (n = 44)

Sample no. Mineral Textural relationships δ18OV-SMOW (‰)

Early, barren stageD5-70 Tourmaline Tourmaline vein, euhedral, prismatic tourmaline (DDH # 5 to the San Rafael vein, 70 m) 11.5D5-94 tou Tourmaline Quartz-tourmaline vein (DDH #5 to the San Rafael vein, 94 m) 11.9D5-94 qz Quartz Quartz-tourmaline vein (DDH #5 to the San Rafael vein, 94 m) 12.3D1-280 Tourmaline Tourmaline-quartz vein (DDH #1 to the San Rafael vein, 280 m) 10.8D3-69 Quartz Quartz-tourmaline veinlet: the quartz is very early and borders the tourmaline vein 12.0

(DDH no. 3 to the Mariano vein, 69 m)D3-182 a Tourmaline Tourmaline-cassiterite-chlorite-arsenopyrite vein (early, orange tourmaline) 11.7

(DDH no. 3 to the Mariano vein, 182.5 m)

Main ore stage (tin and copper)D3-182 b Tourmaline Tourmaline-cassiterite-chlorite-arsenopyrite vein (late, green tourmaline, associated with 9.7

cassiterite) (DDH no. 3 to the Mariano vein, 182.5 m)D3-182 b Cassiterite Tourmaline-cassiterite-chlorite-arsenopyrite vein (earliest cassiterite in the deposit) 2.0

(DDH no. 3 to the Mariano vein, 182.5 m)R119 Banded quartz-chlorite-cassiterite-wolframite-chalcopyrite-pyrite vein (Fig. 5)R119 qz1 Quartz Euhedral quartz associated with chlorite, in between cassiterite bands (Qtz IB) 10.6R119 qz2 Quartz Large, euhedral quartz crystals, preceding the deposition of massive cassiterite (Qtz IV) 10.6R119 qz3 Quartz Small, euhedral quartz crystals, with chlorite and needle-tin cassiterite (Qtz VI) 10.1R119 qz4 Quartz Large, euhedral quartz crystals, preceding needle-tin cassiterite deposition (Qtz VII) 9.7R119 qz5 Quartz Large, subhedral quartz crystals, following needle-tin cassiterite deposition (Qtz VIII) 10.8R119 qz6 Quartz Very late, barren, discordant quartz-chlorite veinlet (Qtz X) 9.8R119 cas1 Cassiterite Earliest layer of massive, dark-brown cassiterite (Cas I) 3.4R119 cas2 Cassiterite Earliest zone of a 7-cm-thick layer of dark-brown, massive cassiterite (Cas IIIA) 2.9R119 cas3 Cassiterite Middle zone of a 7-cm-thick layer of dark-brown, massive cassiterite (Cas IIIB) 3.3R119 cas4 Cassiterite Latest zone of a 7-cm-thick layer of dark-brown, massive cassiterite (Cas IIIC) 3.3R119 cas5 Cassiterite Early needle-tin cassiterite (Cas IV) 2.2R119 cas6 Cassiterite Late, needle-tin cassiterite, associated with chalcopyrite (Cas V) 2.7R119 chl Chlorite Chlorite, associated with quartz and needle-tin cassiterite 4.6R119 wol Wolframite Wolframite, associated with chlorite and quartz 3.0R113 Wolframite Same occurrence as R119 wol 2.1R137 Cassiterite, quartz and chlorite-cemented breccia of chloritized wall-rock fragmentsR137 qz1 Quartz Euhedral quartz preceding cassiterite deposition 10.8R137 cas Cassiterite Layer of massive, dark-brown cassiterite 2.7R137 qz2 Quartz Euhedral quartz following cassiterite deposition 11.2D4 cas1 Cassiterite Early layers of light-colored wood-tin cassiterite (San Rafael Lode, elev 4500 m) 2.3D4 cas2 Cassiterite Late layers of dark wood-tin cassiterite (San Rafael Lode, elevation 4500 m) 3.4R511 chl Chlorite Cassiterite-chlorite-quartz-arsenopyrite vein 3.9R511 qz Quartz Large, euhedral quartz crystals from a cassiterite-chlorite-quartz-arsenopyrite vein 12.2D1-219 wol Wolframite Quartz-chlorite-cassiterite-wolframite vein (DDH no. 1 to the San Rafael vein, 219 m) 1.7D1-219 qz Quartz Quartz-chlorite-cassiterite-wolframite vein (DDH no. 1 to the San Rafael vein, 219 m) 10.1R725 Cassiterite Needle tin cassiterite-chalcopyrite-quartz vein (tin ore from the upper zone) 1.8R429 Quartz Anhedral quartz from a large quartz-sphalerite vein 9.8R708 Quartz Subhedral quartz from a major quartz-chalcopyrite-fluorite-siderite vein 11.1

Late, barren stageR399 qz1 Quartz Late-stage, very large crystal of euhedral vug quartz (core) 14.0R399 qz2 Quartz Late-stage,very large crystal of euhedral vug quartz (outer zone) 13.1R709 Quartz Late-stage, barren, very large quartz vein 14.4D5M-155 Quartz Quartz-calcite fill of a reopened tourmaline vein (DDH no. 5 to the Mariano vein, 155 m) 11.3

Whole-rock samplesI20 Whole rock Fresh alkali-feldspar cordierite-biotite leucogranite (Ramp, elev 3930 m) 10.9D1M-61d Whole rock Fresh biotite-granodiorite (DDH no. 1 to the Mariano vein, 61 m) 9.6D1M-61c Whole rock Mildly altered biotite-granodiorite (DDH no. 1 to the Mariano vein, 61 m) 9.6D1M-61b Whole rock Strongly chloritized biotite-granodiorite (DDH no. 1 to the Mariano vein, 61 m) 8.0D1M-61a Whole rock Chloritite adjacent to the center of a chloritic vein (DDH no. 1 to the Mariano vein, 61 m) 4.9

Notes: Unless indicated otherwise, the samples are from the San Rafael vein and their location is shown in Figure 2

fluids circulating in the system were much cooler (260°–380°C, calculated from isochoric projection of fluid inclusionTh values) and had a moderate to low salinity (0–21 wt %NaCl equiv). During the main cassiterite stage, fluid temper-atures were in the range of 370°–380°C, whereas quartz wasdeposited at consistently lower temperatures of 290°–330°C.Because the vein textures indicate multiple repetition of cas-siterite and quartz deposition, the observed temperature dif-ference between cassiterite- and quartz-hosted fluid inclu-sions likely reflects a sequence of distinct pulses of hottermagmatic and cooler meteoric fluids that mixed repeatedly ina hydrologically very dynamic system.

The quartz-cassiterite oxygen isotope temperatures calcu-lated for this stage (400°–420°C) are in reasonable agree-ment with the temperature estimate obtained from pressure-corrected fluid inclusion data. However, as it was commonlynecessary to base the isotopic temperatures on cassiteriteand quartz from adjacent layers, which as noted aboveformed at different temperatures, the isotopic temperatures

are considered less reliable than the pressure-corrected fluidinclusion temperatures.

The transition between the early, barren stage I and thesubsequent tin and copper stages is recorded by rare tourma-line-cassiterite-chlorite-arsenopyrite veins, in which the earlyand main stage mineral assemblages overlap, indicating thatstage II may have occurred very soon after stage I (Mlynar-czyk and Williams-Jones, 2006). We emphasize that theseveins are of the open fracture-filling type and are surroundedby envelopes of strong chloritic alteration, identical to theore-stage veins. This observation is very important, as it im-plies that the onset of ore deposition (stage II) and themarked change in vein and alteration mineralogy were di-rectly related to a change in structural style, i.e., a transitionfrom a closed vein system (lithostatic conditions) to an openvein system (hydrostatic conditions).

Although the tin stage (stage II) and the copper stage (stageIII) were part of the same, protracted hydrothermal event,and were associated with the same chloritic alteration, thephysico-chemical conditions of the ore fluid were markedlydifferent, and, therefore, resulted in the formation of con-trasting mineral assemblages. In order to constrain the depo-sitional conditions for each stage, we have thermodynamicallymodeled them in fO2-pH and fO2-aΣS space, using phase dia-grams combining aqueous and mineral equilibria in the Cu-Fe-Sn-As-S-O-H system (Fig. 12). In addition, we have mod-eled the speciation of tin in the system Sn-Na-Cl-O-H, toconstrain the total solubility of Sn as a function of pH, and theoxidation state of the hydrothermal fluid (Fig. 13). The detailsof the calculation methods and the source of thermodynamicdata used are listed in the Appendix.

As discussed above, the temperature and pressure duringcassiterite precipitation (stage II) were approximately 370°to 380°C and 200 to 300 bars, respectively, and the fluid hada salinity of about 20 wt percent NaCl equiv. The equilibriummineral assemblage was very simple and consisted of cassi-terite, Fe-rich chlorite (daphnite), and quartz. Subordinatearsenopyrite, which occurs locally, was the only sulfide pre-sent, and in one sample minor hematite was observed togrow coevally with cassiterite and chlorite (Mlynarczyk et al.,2009, submitted). Based on the observed mineral assem-blages and the activity diagrams (Fig. 12), it is concluded thatthe conditions during precipitation of cassiterite evolved

238 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 238

TABLE 4. Hydrogen Isotope Data for Tourmaline and Chlorite from the San Rafael Sn-Cu Deposit (n = 6)

Sample no. Mineral Textural relationships δDV-SMOW (‰)

Early, barren stageD5-70 Tourmaline Tourmaline vein, euhedral, prismatic tourmaline (DDH no. 5 to the San Rafael vein, 70 m) –75D1-280 Tourmaline Tourmaline-quartz vein (DDH #1 to the San Rafael vein, 280 m) –75D3-182 a Tourmaline Tourmaline-cassiterite-chlorite-arsenopyrite vein (early, orange tourmaline) –74

(DDH no. 3 to the Mariano vein, 182.5 m)

Main ore stageD3-182 b Tourmaline Tourmaline-cassiterite-chlorite-arsenopyrite vein (late, green tourmaline, associated with –73

cassiterite) - DDH no. 3 to the Mariano vein, 182.5 mR119 chl Chlorite Associated with quartz and needle-tin cassiterite, from a complexly banded quartz-chlorite –88

-cassiterite-wolframite-chalcopyrite-pyrite vein (Fig. 5)R511 chl Chlorite Cassiterite-chlorite-quartz-arsenopyrite vein –81

Notes: Unless indicated otherwise, the samples are from the San Rafael vein and their location is shown in Figure 2

5 10 15 20 25

0

2

4

6

-2

-4

-6

-8

-10

SideriteCalcite

18OV-SMOW (‰)

13C

V-P

DB (‰

)

FIG. 11. A plot of δ13C versus δ18O for different carbonates from the SanRafael deposit. The arrows link successive generations of siderite from thesame sample. The two data points in the lower right corner of the diagramcorrespond to samples from the uppermost part of the deposit (elev 4820 m);the remaining samples originate from or below an elevation of 4330 m.

from moderately reduced (log fO2 about –28 to –26, upperpart of the arsenopyrite stability field) to more oxidizing (logfO2 above –25, magnetite-hematite boundary). The pHevolved from acidic to more alkaline, with the final pH at theend of massive cassiterite precipitation being slightly aboveabout 4. At this pH value, the cassiterite solubility contoursflatten out considerably (Fig. 13), i.e., the solubility decreasesonly very slightly upon further increase in pH. The sulfur con-centration in the fluid is difficult to constrain, but a minimumvalue of aΣS of about 0.005 to 0.01 can be estimated from thestability field of arsenopyrite (Fig. 12b).

During the sulfide stage (stage III), the temperature andpressure were approximately 300°C (based on fluid inclusiontrapping temperatures and stannite-sphalerite thermometry)and 200 to 300 bars (assuming hydrostatic conditions and aroughly similar depth to that during the previous mineraliza-tion stages), and the fluid salinity was around 5 wt percentNaCl equiv. The mineral assemblage consists mainly of chal-copyrite, pyrrhotite, Fe-rich chlorite (daphnite), quartz, andminor cassiterite, but locally other sulfides (e.g., arsenopyrite,sphalerite, galena, pyrite) are also present. Figure 12 showsthe most relevant mineral stability relations, as well as the sul-fur isotope contours that were calculated for chalcopyrite (seebelow). It can be deduced from the stability of chlorite (i.e.,the daphnite end member as calculated from the measuredchlorite composition) and pyrrhotite that during most of stageIII, the oxidation state was lower (log fO2 below –35, upperdaphnite stability limit in Fig. 12c) than during stage II, whilethe pH was likely similar (around 4). It should be noted, how-ever, that in many locations in the deposit, pyrite is an abun-dant sulfide of stage III mineralization, and pyrrhotite dis-plays an incipient or advanced replacement by pyrite, marcasite,or rare hematite. Although in some case this replacement of

pyrrhotite was clearly late, the common presence of pyrite instage III ores indicates that there must have been large fluc-tuations in fO2, and that the hydrothermal environment be-came progressively more oxidizing (i.e., moved from thepyrrhotite to the pyrite stability field in Fig. 12c) with time.Furthermore, the ubiquitous replacement of pyrrhotite bypyrite might indicate that the sulfidation state of the systemincreased with time (Fig. 12d), a feature that has been ob-served for many magmatic-hydrothermal systems (Einaudi etal., 2003). Additional information about the oxidation stateand the fluid evolution during stage III mineralization comesfrom the sulfur isotope data. The calculated sulfur isotopecontours show that an increase in oxidation state above a logfO2 of –31 to –29 would result in a considerable shift in theδ34S of precipitated sulfides towards more negative values(Fig. 12). From the relatively narrow range in δ34S values ofthe San Rafael sulfides it can be concluded that although theconditions during stage III became more oxidizing with time,they did not reach as high as the pyrite-hematite boundary(Fig. 12c).

Source of sulfur

The δ34S values of sulfide minerals from the San Rafael de-posit range between 2 and 6 per mil, and show relatively lit-tle variation with respect to the location in the deposit (seeabove). These rather uniform values point to a large-scale hy-drothermal system with a homogeneous source of sulfur,which was likely of magmatic origin. As argued by Hattori andKeith (2001) for porphyry systems, the very narrow range inδ34S values displayed by giant, granite-related deposits is bestexplained by a single magmatic source of sulfur, rather thanby homogenization of the sulfur from a variety of sources inlocal country rocks having diverse S isotope compositions. It

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 239

0361-0128/98/000/000-00 $6.00 239

TABLE 5. Carbon and Oxygen Isotope Data for Vein Carbonates from the San Rafael Sn-Cu Deposit (n = 12)

Sample no. Mineral Textural relationships δ13CV-PDB (‰) δ18OV-SMOW (‰)

Late, barren stageD1-118 Calcite Quartz-calcite-sphalerite-chalcopyrite infill of a reopened tourmaline vein –8.5 8.6

(DDH no. 1 to the San Rafael vein, 118.5 m)D5-137 Calcite Quartz-calcite infill of a reopened tourmaline vein –4.0 9.4

(DDH no. 5 to the San Rafael vein, 137.9 m)D5-165 Calcite Quartz-calcite infill of a reopened tourmaline vein –7.8 9.6

(DDH no. 5 to the San Rafael vein, 165.8 m)D4-230 Calcite Quartz-calcite-arsenopyrite infill of a reopened tourmaline vein –3.2 14.2

(DDH no. 4 to the San Rafael vein, 230.5 m)D5M-155 Calcite Quartz-calcite infill of a reopened tourmaline vein –3.5 9.1

(DDH no. 5 to the Mariano vein, 155 m)R27 sid1 Siderite Early, anhedral, reddish, opaque siderite from a quartz- –0.3 18.8

sphalerite-pyrite-marcasite-chalcopyrite-galena-siderite veinR27 sid2 Siderite Late, euhedral, greenish, translucent siderite from a quartz- 0.0 15.7

sphalerite-pyrite-marcasite-chalcopyrite-galena-siderite veinR28 sid Siderite Late, euhedral, greenish, translucent siderite from a quartz- –0.3 13.8

sphalerite-pyrite-marcasite-chalcopyrite-galena-siderite veinR57 sid Siderite Late, orange, translucent siderite from a quartz-cassiterite- –0.3 17.7

chlorite-siderite veinR123 sid Siderite Greenish, translucent siderite from a quartz-fluorite-pyrite- 4.3 15.7

marcasite-siderite veinR708 sid2 Siderite Early, anhedral, reddish, opaque siderite from a quartz- –6.5 23.2

chalcopyrite-fluorite-siderite veinR708 sid1 Siderite Late, euhedral, greenish, translucent siderite from a quartz- –4.9 22.0

chalcopyrite-siderite veinlet in chloritized wall rock

Notes: Unless indicated otherwise, the samples are from the San Rafael vein and their location is shown in Figure 2

240 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 240

0 2 4 6 8 10 12

-18

-20

-22

-24

-26

-28

-30

-32

-34

HSO4-

HS-H2S

SO42-

pyrite

pyrrhotite loeas

asploe

magnetite

hematite

Log

fO2

pH

Stability field of arsenopyrite

aH2O=0.873 (20wt.% NaCl eq.)

P=300 barQtz saturation

T=380 C a

aΣS = 0.01

-4 -3 -2 -1 0

-18

-20

-22

-24

-26

-28

-30

-32

-34

HSO4-

H2S

hematite

Log aΣS

magnetite

pyrrhotite

pyrite

asp

asp

loe

loe

as

as

as

asp

Stability field of arsenopyrite

Log

fO2

aH2O=0.873 (20wt.% NaCl eq.)pH=4

P=300 barQtz saturation

T=380 C b

0 2 4 6 8 10 12

-24

-26

-28

-30

-32

-34

-36

-38

-40

HSO4-

HS-H2S

SO42-

dap

kfs

cp

cp

msprl

hematite

pH

magnetitepyrrhotite

pyrite

Stability field of chalcopyrite

adap=0.18

ams=0.517

+2.8

+3.

0

+2.

9

0.0-5.0

-10.0-15.0

-17.5

bn+py

bn+po

Log

fO2

aH2O=0.973 (5 wt.% NaCl eq.)

aK+= 0.005

P=300 barQtz saturation

T=300 C c

aΣS = 0.01δ34Stot=+3.0‰

-5 -4 -3 -2 -1

-24

-26

-28

-30

-32

-34

-36

-38

-40

HSO4-

hematite

Log aΣS

magnetite

pyrrhotite

pyrite

Stability field of chalcopyrite

H2S

cp

cp

bn+py

bn+po

+2.8

0.0-5.0

-10.0-15.0

-17.5

Log

fO2

aH2O=0.973 (5 wt.% NaCl eq.)pH=4

P=300 barQtz saturation

T=300 C d

δ34Stot=+3.0‰

FIG. 12. Log fO2-pH and log fO2-aΣS diagrams showing stability relationships in the system Fe-O-S, for stage II (oxide as-semblage, diagrams a and b) and stage III (sulfide assemblage, diagrams c and d) of the paragenesis. The temperature andfluid salinity for which the diagrams were drawn were constrained from fluid inclusion microthermometry, whereas pressurewas estimated. Diagrams a and b also show stability relationships for the Fe-As-S system (asp = arsenopyrite, as = native ar-senic, loe = loellingite). Diagrams c and d show the stability relationships in the Cu-Fe-S system (cp = chalcopyrite, bn =bornite, py = pyrite) and of the principal alteration minerals (dap = daphnite, prl = pyrophyllite, ms = muscovite, kfs = Kfeldspar). The activities of alteration minerals were calculated from microprobe data (see Appendix for details), and thechemical reactions were balanced assuming conservation of Al and quartz saturation. Diagrams c and d (sulfide stage) alsoshow superimposed S isotope contours, calculated for chalcopyrite, assuming aΣS = 0.01 and δ34Stotal= 3.0 ‰, inferred fromthe S isotope composition of the sulfides from this stage (see text for details).

should be noted though that the San Rafael sulfides haveslightly higher δ34S values than some other tin deposits (Kon-tak, 1990), implying an initial relative enrichment in the 34Scontent of the sulfur source. This is not unexpected for S-typegranitic magmas, which acquire most of their sulfur throughassimilation of country rocks. Thus, for such plutons, the δ34Svalues of ore sulfides are very similar to those of local igneousrocks and local country rocks, and reliable discrimination be-tween magmatic and sedimentary sulfur is not possible(Ohmoto and Goldhaber, 1997).

Source of ore fluids

We have calculated the oxygen and hydrogen isotope com-position of the ore fluids (Fig. 14) during different mineral-ization stages to constrain the most likely fluid sources for theSan Rafael deposit. The isotopic composition of the early flu-ids, which were in equilibrium with stage I tourmaline, is veryclose to magmatic water values, which, considering the hightemperature and salinity of these fluids, supports the idea thatthey were magmatic in origin. The water in equilibrium withlate, Fe-rich tourmaline, which formed at the onset of cassi-terite precipitation (single analysis in the upper right cornerof the magmatic water box, Fig. 14) however, was enriched indeuterium and had isotopically lighter oxygen. Compared tothe tourmaline data, the chlorite data demonstrate a fluidevolution characterized by a marked decrease in δ18O and aslight increase in δD with time. The water in equilibrium withore-stage chlorite had δ18O lower by about 7 per mil and δDhigher by 10 per mil than that of the fluid in equilibrium withearly tourmaline (Fig. 14). It should be noted, however, thatthe δD of water in equilibrium with chlorite cannot be pre-cisely constrained due to the lack of accurate fractionationfactors (Graham et al., 1987). The trend of decreasing fluidδ18O with time is, nevertheless, clearly indicated by the

oxygen isotope compositions of successive generations ofquartz and cassiterite (see earlier section), and also by miner-als which formed late in the paragenesis, such as calcite andsiderite.

The decrease in ore fluid δ18O with time is paralleled bymarked corresponding decreases in fluid salinity and temper-ature, as indicated by the fluid inclusion data (Fig. 7). Thesetrends, which appear broadly aligned, could represent mixingof the early, hot, hypersaline (presumably magmatic) brinewith a relatively warm fluid that had consistently lower δ18Oand δD values. Considering the shallow level of emplacementof the San Rafael pluton and fluid inclusion evidence for verylow salinity fluids (this could not be produced by simple boil-ing of a magmatic brine; Kontak and Clark, 2002) circulatingin the system, a good candidate for this external fluid wouldbe heated groundwater of meteoric origin that had partly ex-changed its oxygen isotope composition with the host rocks.

Unfortunately, it is currently not possible to properly con-strain the isotopic composition of the late Oligocene, Andeanmeteoric waters, coeval with formation of the deposit, to fur-ther substantiate this hypothesis. This is because of the very

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 241

0361-0128/98/000/000-00 $6.00 241

3.0 3.5 4.0 4.5 5.0 5.5 6.0

10-3

10-5

10-4

10-2

10-1

1.0To

tal S

n so

lub

ility

(mol

/kg)

pH

log fO2 = - 32

- 31

- 30

- 29

- 28

- 27

- 26- 25- 24- 23

- 22

FIG. 13. Total solubility of tin (mol/kg) in the system Sn-Na-Cl-O-H cal-culated as a function of pH and fO2 at 380°C and 300 bar (the inferred con-ditions for stage II). For details of the calculations, see the Appendix.

18OOre fluid (‰)

200 10-10-20

qz caswol

chl

tou

18OV-SMOW (‰)

SMOW

Met

eoric

wat

er li

ne

Magmatic water box

Tertiary Andeanmeteoric water ? Groundwater

Present-dayAndeanmeteoric water

ChloriteTourmaline

200 10-10-20

0

-100

-120

-80

-60

-40

-20

20

DV-

SM

OW (‰

)

FIG. 14. Plot of δ18O versus δD for tourmaline (open squares) and chlo-rite (open circles) from the San Rafael deposit and calculated water in iso-topic equilibrium with these minerals (black squares and circles). The thickdashed trendline from tourmaline to chlorite represents the inferred evolu-tion of fluid (water) during the formation of the deposit. The lower diagram(δ18O axis) shows the distribution of the calculated δ18O values for water inequilibrium with the different ore and gangue minerals (cas = cassiterite, chl= chlorite, qz = quartz, tou = tourmaline, wol = wolframite). The calculationsapplied isotopic fractionation equations that are mostly based on well-con-strained experiments in the appropriate temperature range (Matsuhisa et al.,1979; Graham et al., 1984, 1987; Carothers et al., 1988; Kotzer et al., 1993;Zheng, 1993; Zhang et al., 1994; Cole and Ripley, 1999).

large uncertainty concerning the rate of the uplift in this partof the Andes during the last 25 m.y. It is generally assumedthat the elevation of the Central Andes in the late Oligocenewas about a third of their current elevation (Gregory-Wodz-icki, 2000; Anders et al., 2002), implying that ancient mete-oric waters must have had considerably higher δ18O and δDvalues than their present-day counterparts, but the actual val-ues can only be roughly estimated (Fig. 14). Applying bothclosed- and open-system scenarios (Taylor, 1977, 1997) foroxygen isotope exchange between water of meteoric originand typical Andean granites (e.g., Longstaffe et al., 1983), thecalculated δ18O values of hot groundwater at temperatures of200° to 250°C (assuming moderate fluid/rock ratios between0.1 and 1.0) would be in the range between –2 and 2 per mil.

Ore deposition processes

The rich tin ores of the San Rafael deposit testify to an un-usually effective mechanism of ore deposition, which causedsubstantial supersaturation of tin and focused mineralizationinto several large fault-jogs, at depth in the lode (Mlynarczyket al., 2003). Experimental studies and chemical modelinghave shown that in natural hydrothermal systems the solubil-ity of tin is highest in hot, reduced, saline, and acidic solutions(Fig. 13), and that the bulk of the tin is transported as stan-nous (Sn2+) chloride complexes (Eugster and Wilson, 1985;Pabalan, 1986; Taylor and Wall, 1993; Müller and Seward,2001). The precipitation of cassiterite (SnO2), in which tin isin the tetravalent state, can, therefore, be induced by an in-crease in fO2 as well as by decreases in temperature and ligand(chloride) ion activity and an increase in pH. Possible mech-anisms for tin ore formation could include reaction of the orefluids with the host rocks, boiling, redox-coupled precipita-tion, mixing of the ore fluids with fluids of a markedly differ-ent composition, or a combination of some of the above (Ead-ington, 1985; Heinrich, 1990, 1995).

The fluid-rock interaction, which accompanied ore deposi-tion at San Rafael, produced a wide envelope of very strongchloritization but is not considered to have influenced ore for-mation as mass-balance calculations show that such alterationincreases the acidity of the fluid (Mlynarczyk et al., 2009, sub-mitted), leading to an increase in cassiterite solubility (Fig.13). Redox-coupled precipitation (e.g., coupled precipitationof arsenopyrite and cassiterite) is also not considered to haveplayed a significant role because it would have led to consid-erable heterogeneity in the compositions of the ore and alter-ation minerals, which was clearly not the case.

On the other hand, structural evidence for a transition froma closed to an open vein system, and a concomitant changefrom lithostatic to hydrostatic conditions, implies that boilingmay have taken place. Boiling would have oxidized the orefluid, decreased its temperature and significantly increased itspH due to the loss of acidic volatiles. Thus, it would have pro-moted the precipitation of cassiterite. However, vapor-richfluid inclusions are rare and are only found in stage 1 veinsthat formed prior to tin mineralization. Thus, if boiling oc-curred during the main ore-forming events, it was probably oflimited extent. By contrast, the coincidence of the ore stagewith the opening of the vein system and the widespreadappearance of cooler, much more dilute fluids (down to al-most zero salinity), as well as the ensuing periodic fluctuation

of fluid salinity and temperature (above 360°C and about 21wt % NaCl equiv during formation of massive layers of cassi-terite, but below 300°C and 2–16 wt % NaCl equiv duringformation of intervening quartz layers) is readily explained byincursions of heated groundwaters of meteoric origin whichmixed with magmatic fluids (Mlynarczyk et al., 2003). Amodel invoking the mixing of magmatic fluids with watershaving lower δ18O and δD could also explain the observedsystematic decrease of cassiterite and quartz δ18O values withtime. Assuming that hot groundwaters of meteoric originwould not have had sufficient time to fully equilibrate withthe host rocks, they likely would be cooler and much more ox-idizing than the magmatic fluids. Numerical simulation ofcomplex fluid-granite equilibria (Dolejs and Wagner, 2008)has shown that rock-buffered low-salinity aqueous fluids willbecome much more oxidizing as temperature decreases from400° to 200°C. Therefore, mixing of hot groundwaters withtin-bearing magmatic fluids would have oxidized the fluid sys-tem, increased the overall pH, and decreased temperatureand ligand activity, all of which would have destabilized tinchloride complexes and triggered cassiterite precipitation.

A further evaluation of boiling and fluid mixing scenariosand their relative importance is possible from the evolution ofthe oxygen isotope composition of the ore fluid. Boiling in-creases the δ18O of the remaining liquid because of preferen-tial partitioning of the light isotope into the vapor, and this isreflected in the paragenesis by a gradual increase in δ18O withtime. By contrast, mixing of magmatic waters with groundwa-ters of meteoric origin (isotopically light) should lead to pro-gressive decreases in the δ18O values of the ore fluid. Itshould be noted, however, that the temperature decrease as-sociated with mixing will increase the isotopic fractionationbetween the ore fluid and the precipitating minerals, which,in the case of the oxygen isotope composition of cassiteriteand wolframite, will partly offset the effect of addition of low-δ18O groundwaters. Because of such complexities, quantita-tive modeling of boiling and fluid mixing mechanisms of tinore formation are required to discriminate between them.

Modeling of fluid evolution scenarios

To test the hypothesis of fluid mixing and compare it to oneof boiling of the ore fluid, the oxygen isotope composition ofsilicate and oxide minerals precipitating by each mechanismwas calculated as a function of temperature. Because theavailable experimental studies of isotopic liquid-vapor parti-tioning in the H2O-NaCl system are restricted to conditionsbelow the critical point of pure water (Horita et al., 1993,1995), it was not possible to adequately model boiling for thetemperature-composition space at San Rafael. Based on boil-ing models for low-salinity fluids (e.g., Truesdell and Nathen-son, 1977; Wagner et al., 2005), it can be predicted that boil-ing would produce a progressive increase in mineral δ18Ovalues. This would be the consequence of preferential parti-tioning of the lighter 16O isotope into the vapor phase (Horitaand Wesolowski, 1994), yielding an increasingly 18O-richresidual brine. Furthermore, the effect would be enhancedby the cooling of the residual brine as a consequence of en-thalpy transfer from the liquid to the vapor that is required tomaintain the enthalpy balance. Hence, the first-order effectof boiling on the δ18O of the precipitated minerals is broadly

242 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 242

similar to that of fluid cooling and can therefore be simulatedby the latter.

Based on the fluid inclusion data and the measured oxygenisotope data for stage I silicates, the parameters of the start-ing magmatic fluid were assumed to be a temperature of500°C, a salinity of 45 wt percent NaCl equiv, and a δ18Ovalue of 11 per mil. One model considered that this fluidcooled due to boiling down to a temperature of 200°C,whereas the other considered its mixing with a cooler, dilutefluid (230°C, zero salinity), assuming three possible δ18Ocompositions of 2.0, 0.0, –2.0 percent representing the com-position of hot groundwaters of meteoric origin that had ex-changed their oxygen with their host rocks. A stepwise de-crease of ore fluid temperature took place in each model andin the case of mixing was accompanied by a decrease in salin-ity and fluid δ18O, proportional to the aliquot of diluting fluidadded. For simplicity, a linear relationship between the tem-perature and composition of the fluid mixture was assumed,although it is actually the enthalpy, not the temperature, thatvaries linearly with composition during mixing (e.g., Reed andSpycher, 1984; Spycher and Reed, 1989). The influence offluid salinity on the concentration of oxygen was taken intoaccount (Schwinn et al., 2006) because the salinity of themagmatic end member was quite high but, in the absence ofpertinent experimental data, the mineral-water fractionationfactors were assumed to be salinity independent. The latterwere calculated for every temperature step (using the equa-tions of Matsuhisa et al., 1979; Zhang et al., 1994; and Coleand Ripley, 1999), enabling the δ18O composition of quartz,cassiterite, chlorite, and wolframite in equilibrium with thefluid to be determined.

Selected modeling results are presented in Table 6 and thecalculated changes in the δ18O values of quartz and cassiterite

with time are shown in Figure 15, together with the measuredranges of δ18O values and fluid inclusion trapping tempera-tures for ore stage quartz and cassiterite. From the calcula-tions, it is evident that boiling and fluid mixing have a con-trasting influence on the oxygen isotope composition of theprecipitating minerals. Whereas mixing with hot, low-δ18Ofluid produces a systematic decrease in mineral δ18O values(followed by a minor increase below 250°C), boiling results ina systematic increase of the δ18O values of the oxides and sil-icates with time, a scenario clearly not supported by the par-agenetic information and the isotopic data for San Rafael. Inaddition to the good agreement between the predicted andobserved trend of decreasing mineral δ18O values, the mixingmodel also reproduces closely the relatively narrow range ofmineral δ18O compositions observed in the deposit (Fig. 15).It is, however, noteworthy that this is only the case when thediluting fluid is assumed to be relatively hot (230°C), as mix-ing of the magmatic brine with cold (25°–100°C), dilute waterwould drive the mineral δ18O compositions toward very highvalues.

The clear lack of a trend of increasing mineral δ18O valuesat San Rafael (apart from that observed in the lower-temper-ature stage IV, which is consistent with the mixing model),therefore, precludes a significant role for boiling in cassiteritedeposition. By contrast, the mixing of a magmatic brine withheated, dilute groundwaters of meteoric origin reproducesremarkably well the observed δ18O composition of the oreand gangue minerals. In our model, fluid mixing was simu-lated as a continuous process to capture the first-order ef-fects, whereas the complex banding of the veins with multiplegenerations of the main ore and gangue minerals suggeststhat fluid mixing occurred repeatedly in a hydrologically verydynamic system.

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 243

0361-0128/98/000/000-00 $6.00 243

TABLE 6a. Calculated Oxygen Isotope Composition of Quartz, Cassiterite, Chlorite and Wolframite, in Equilibrium with a Magmatic Brine (T = 500°C, δ18O = 11.0‰), which is Mixing with Meteoric Water (T = 230°C, δ18O = 2.0‰)

Tmix (°C) δ18O (fluid) δ18O (mineral)F = massFraction brine Quartz Cassiterite Chlorite Wolframite

1.00 500 11.0 13.3 6.7 10.1 7.80.95 487 10.4 13.0 6.2 9.5 7.30.90 473 9.9 12.6 5.7 9.0 6.80.85 460 9.4 12.3 5.2 8.5 6.30.80 446 8.9 12.0 4.7 8.0 5.80.75 433 8.4 11.8 4.3 7.5 5.40.70 419 7.9 11.6 3.9 7.0 5.00.65 406 7.4 11.4 3.5 6.6 4.60.60 392 6.9 11.2 3.1 6.2 4.20.55 379 6.5 11.0 2.8 5.8 3.80.50 365 6.0 10.9 2.6 5.4 3.50.45 352 5.6 10.9 2.3 5.1 3.20.40 338 5.1 10.8 2.1 4.8 2.90.35 325 4.7 10.8 2.0 4.5 2.60.30 311 4.3 10.8 1.9 4.3 2.40.25 298 3.9 10.9 1.9 4.1 2.10.20 284 3.5 11.0 1.9 4.0 1.90.15 271 3.1 11.2 2.0 3.9 1.80.10 257 2.7 11.4 2.1 3.9 1.60.05 244 2.4 11.6 2.4 3.9 1.50.00 230 2.0 12.0 2.7 4.0 1.5

Concluding RemarksThe hydrothermal history of the San Rafael vein system

began with hot, acidic, saline and reducing fluids, which hadan isotopic composition very close to that of typical magmaticfluids. These fluids likely exsolved from a late granitic melt re-lated to the San Rafael igneous center and produced exten-sive sericitic and tourmaline alteration of the wall rocks (stage

I). The modeling of tin solubility suggests that these fluidscould have transported high concentrations of tin, but theydid not deposit any cassiterite.

The onset of cassiterite deposition came with a majorchange in the plumbing of the hydrothermal system, inferredto correspond to a change from lithostatic to hydrostatic pres-sure conditions, upon a major reopening of the San Rafaellode. The change in structural style, indicated by the ubiqui-tous open fracture-filling character of stage II-IV veins, wasassociated with a drop in temperature and major changes influid chemistry. This is evident from the strong chloritic al-teration and the massive deposition of cassiterite and, subse-quently, sulfide minerals. The ore fluids were considerablyless saline and somewhat cooler than the early brines, weremore oxidizing, and had lower δ18O and δD values.

The precipitation of cassiterite was most likely caused by amarked increase in the oxidation state of the ore fluid (sup-ported by evidence of rare hematite), an increase in pH, andto a lesser extent, decreases in temperature and chloride ac-tivity. The two depositional mechanisms that could explain thegeologic evidence are boiling of the magmatic brine or its mix-ing with cooler, oxidizing meteoric waters. Quantitative mod-eling of mineral-water oxygen fractionation predicts the ob-served decrease of fluid δ18O values with time for the case offluid mixing, whereas this modeling suggests that boiling willproduce the opposite trend in δ18O values. Also, the general

244 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 244

250 500400 450300 350

Boiling (cooling)

Mixing

Stage II-III cassiterite

02

-2

4

-4

68

10

141618

12

+2.0 ‰

0.0 ‰

-2.0 ‰

TOre fluid (°C)

b

18O

Cas

site

rite

(‰)

250 500400 450300 350

Boiling (cooling)

Mixing

Stage II-III quartz

02468

10

1416182022

12

18O

Qua

rtz

(‰)

+2.0 ‰

0.0 ‰

-2.0 ‰

TOre fluid (°C)

18O=+11‰

High-T fluid:

45 wt.% NaCl eq.500 COLow-T fluid:

0 wt.% NaCl eq.230 CO

a

Stage IV quartz

FIG. 15. Diagrams showing the effect of boiling (cooling) and mixing ofmagmatic and meteoric water derived fluids on the oxygen isotope composi-tion of quartz (a) and cassiterite (b). Because no experimental data are avail-able for isotopic liquid-vapor fractionation in the H2O-NaCl system at condi-tions above the critical point of water, the boiling model has only consideredthe cooling effect (resulting from enthalpy transfer from the liquid to thevapor). Therefore, the boiling curves are labeled as boiling (cooling). Thestarting conditions for the magmatic brine were a temperature of 500ºC, asalinity of 45 wt percent NaCl equiv, and a δ18O value of 11 per mil. Mixinglines were calculated assuming that groundwater had a temperature of230ºC, zero salinity and δ18O values between –2.0 and 2.0 per mil (labels onthe mixing lines).

TABLE 6b. Calculated Oxygen Isotope Composition of Quartz, Cassiterite, Chlorite and Wolframite, in Equilibrium with a Magmatic Brine (T = 500°C, δ18O = 11.0‰), Which is Progressively Cooling

δ18O (fluid) δ18O (mineral)

T (°C) Quartz Cassiterite Chlorite Wolframite

500 11.0 13.3 6.7 10.1 7.8480 11.0 13.6 6.8 10.1 7.8460 11.0 13.9 6.8 10.1 7.9440 11.0 14.3 6.9 10.1 8.0420 11.0 14.7 7.0 10.2 8.1400 11.0 15.1 7.1 10.2 8.2380 11.0 15.6 7.4 10.3 8.4360 11.0 16.1 7.6 10.5 8.5340 11.0 16.7 8.0 10.7 8.7320 11.0 17.3 8.4 10.9 8.9300 11.0 18.0 8.9 11.2 9.2280 11.0 18.7 9.5 11.6 9.5260 11.0 19.6 10.3 12.0 9.8240 11.0 20.5 11.2 12.7 10.3230 11.0 21.0 11.8 13.0 10.5

evolution of the hydrothermal system toward domination bystrongly dilute, cooler fluids is consistent with an incursion ofhot groundwaters of meteoric origin. As the bulk of the tinore at San Rafael is restricted to large fault-jogs at depth inthe lode, we conclude that these jogs provided sites favorablefor repeated episodic mixing between magmatic and meteoricfluids, which resulted in unusually efficient, structurally fo-cused cassiterite deposition.

AcknowledgmentsThe authors wish to thank Ing. Fausto Zavaleta Cruzado,

General Manager of MINSUR S.A., Ing. Luis Alva Florianand Ing. Otto Velarde Junes, successive managers of the SanRafael mine, Ing. Julver Alvarez Romero, Ing. Pastor LuqueMalagá, Ing. Ladislao Guillén Cardenas, and Ing. Luis San-talla Medrano, mine geologists, as well as other staff of theSan Rafael mine, for logistic support and helpful collabora-tion during the field work. The assistance of Harold E. Wallerand Nestor Roldan from EMINASA and Mario ArenasFigueroa, consulting geologist, were much appreciated. Theauthors would also like to thank the reviewers, B. Lehmannand A. Campbell, as well as associate editor J. Muntean andEditor L. Meinert for their constructive comments, whichhelped to improve this paper. The research was funded byNSERC and FQRNT grants to AEW-J and NERC support ofthe Isotope Community Support Facility at SUERC.

November 6, 2008; February 18, 2009

REFERENCESAja, S.U., 2002, The stability of Fe-Mg chlorites in hydrothermal solutions.

II. Thermodynamic properties: Clays and Clay Minerals, v. 50, p. 591–600.Aja, S.U., and Dyar, D.M., 2002, The stability of Fe-Mg chlorites in hy-

drothermal solutions. I. Results of experimental investigations: AppliedGeochemistry, v. 17, p. 1219–1239.

Alderton, D.H.M., 1989, Oxygen isotope fractionation between cassiteriteand water: Mineralogical Magazine, v. 53, p. 373–376.

Alderton, D.H.M., and Harmon, R.S., 1991, Fluid inclusion and stable iso-tope evidence for the origin of mineralizing fluids in southwest England:Mineralogical Magazine, v. 55, p. 605–611.

Anders, M.H., Gregory-Wodzicki, K.M., and Spiegelman, M., 2002, A criti-cal evaluation of late Tertiary accelerated uplift rates for the EasternCordillera, Central Andes of Bolivia: Journal of Geology, v. 110, p. 89–100.

Arenas, M.J., 1980, Mapa geologico superficial del Distrito Minero SanRafael, Puno: Unpublished Report MINSUR Archives.

Bakker, R.J., 2003, Package FLUIDS. 1. Computer programs for analysis offluid inclusion data and for modelling bulk fluid properties: Chemical Ge-ology, v. 194, p. 3–23.

Barton, P.B., 1969, Thermochemical study of the system Fe-As-S: Geochim-ica et Cosmochimica Acta, v. 33, p. 841–857.

Bodnar, R.J., 1993, Revised equation and table for determining the freezingpoint depression of H2O-NaCl solutions: Geochimica et CosmochimicaActa, v. 57, p. 683–684.

Bortnikov, N.S., Zaozerina, O.N., Genkin, A.D., and Muravitskaya, G.N.,1990, Stannite-sphalerite intergrowths—possible indicators of conditionsof ore deposition: International Geology Review, v. 32, p. 1132–1144.

Cathelineau, M., 1988, Cation site occupancy in chlorites and illites as a func-tion of temperature: Clay Minerals, v. 23, p. 471–485.

Carothers, W.W., Adami, L.H., and Rosenbauer, R.J., 1988, Experimental oxy-gen isotope fractionation between siderite-water and phosphoric acid liber-ated CO2-siderite: Geochimica et Cosmochimica Acta, v. 52, p. 2445–2450.

Clark, A.H., Palma, V.V., Archibald, D.A., Farrar, E., Arenas, M.J., andRobertson, R.C.R., 1983, Occurrence and age of tin mineralization in theCordillera Oriental, Southern Peru: ECONOMIC GEOLOGY, v. 78, p. 514–520.

Clark, A.H., Farrar, E., Kontak, D.J., Langridge, R.J., Arenas Figueroa, M.J.,France, L.J., McBride, S.L., Woodman, P.L., Wasteneys, H.A., Sandeman,H.A., and Archibald, D.A., 1990, Geologic and geochronologic constraints

on the metallogenic evolution of the Andes of southeastern Peru: ECO-NOMIC GEOLOGY, v. 85, p. 1520–1583.

Clayton, R.N., and Mayeda, T.K., 1963, The use of bromine pentafluoride inthe extraction of oxygen from oxides and silicates for isotopic analysis:Geochimica et Cosmochimica Acta, v. 27, p. 43–52.

Cole, D.R., and Ripley, E.M., 1999, Oxygen isotope fractionation betweenchlorite and water from 170 to 350°C: A preliminary assessment based onpartial exchange and fluid/rock experiments: Geochimica et CosmochimicaActa, v. 63, p. 449–457.

Collins, P.L.F., 1981, The geology and genesis of the Cleveland tin deposit,Western Tasmania: Fluid inclusion and stable isotope studies: ECONOMICGEOLOGY, v. 76, p. 365-392.

Dolejs, D., and Wagner, T., 2008, Thermodynamic modeling of non-idealmineral-fluid equilibria in the system Si-Al-Fe-Mg-Ca-Na-K-H-O-Cl at el-evated temperatures and pressures: Implications for hydrothermal masstransfer in granitic rocks. Geochimica et Cosmochimica Acta, v. 72, p.526–554.

Dolejs, D., Mlynarczyk, M.S.J., and Williams-Jones, A.E., 2009, S-type pera-luminous granitic rocks associated with a world-class tin deposit: a petro-genetic study of the San Rafael stock, southeastern Peru. J. Petrol. (sub-mitted)

Eadington, P.J., 1985, The solubility of cassiterite in hydrothermal solutionsin relation to some lithological and mineral associations of tin ores, in Tay-lor, R.P., and Strong, D.F., eds., Recent advances in the geology of granite-related mineral deposits: Canadian Institution of Mining and Metallurgy,Special Volume 39, p. 25–32.

Einaudi, M.T., Hedenquist, J.W., and Inan, E.E., 2003, Sulfidation state offluids in active and extinct hydrothermal systems: Transitions from por-phyry to epithermal environments: SEG Special Publications, v. 10, p.285–313.

Eugster, H.P., and Wilson, G.A., 1985, Transport and deposition of ore-form-ing elements in hydrothermal systems associated with granites, in Halls, C.,ed., High heat production (HHP) granites, hydrothermal circulation andore genesis: Institution of Mining and Metallurgy Conference, London, p.87–98.

Goldstein, R.H., and Reynolds, T.J., 1994, Systematics of fluid inclusions indiagenetic minerals: Society of Sedimentary Geology Short Course, v. 31,199 p.

Graham, C.M., Atkinson, J., and Harmon, R.S., 1984, Hydrogen isotope frac-tionation in the system chlorite-water [abs.]: Natural Environment Re-search Council (NERC) Publication Series D, no. 25, p. 139.

Graham, C.M., Viglino, J.A., and Harmon, R.S., 1987, Experimental study ofhydrogen isotope exchange between aluminous chlorite and water and ofhydrogen diffusion in chlorite: American Mineralogist, v. 72, p. 566–579.

Gregory-Wodzicki, K.M., 2000, Andean paleoelevation estimates: A reviewand critique: Geological Society of America Bulletin, v. 112, p. 1091–1105.

Grønvold, F., and Stølen, S., 1992, Thermodynamics of iron sulfides. II. Heatcapacity and thermodynamic properties of FeS and of Fe0.875S at tempera-tures from 298.15 K to 1000 K, of Fe0.98S from 298.15 K to 800 K, and ofFe0.89S from 298.15 K to about 650 K. Thermodynamics of formation: Jour-nal of Chemical Thermodynamics, v. 24, p. 913–936.

Harrison, T.M., Duncan, I., McDougall, I., 1985, Diffusion of 40Ar in biotite:Temperature, pressure and compositional effects: Geochimica et Cos-mochimica Acta, v. 49, p. 2461–2468.

Haynes, F.M., 1985, Determination of fluid inclusion compositions by se-quential freezing: ECONOMIC GEOLOGY, v. 80, p. 1436–1439.

Hattori, K.H., and Keith, J.D., 2001, Contribution of mafic melt to porphyrycopper mineralization: Evidence from Mount Pinatubo, Philippines, andBingham Canyon, Utah, USA: Mineralium Deposita, v. 36, p. 799–806.

Heinrich, C.A., 1990, The chemistry of hydrothermal tin (-tungsten) ore de-position: ECONOMIC GEOLOGY, v. 85, p. 457–481.

——1995, Geochemical evolution and hydrothermal mineral deposition inSn (-W-base metal) and other granite-related ore systems: some conclu-sions from Australian examples, in Thompson, J.F.H., ed., Magmas, fluidsand ore deposits: Mineralogical Association of Canada (MAC) ShortCourse, v. 23, p. 203–220.

Heinrich, C.A., and Eadington, P.J., 1986, Thermodynamic predictions of thehydrothermal chemistry of arsenic, and their significance for the parage-netic sequence of some cassiterite-arsenopyrite-base metal sulfide de-posits: ECONOMIC GEOLOGY, v. 81, p. 511–529.

Holland, T.J.B., and Powell, R., 1998, An internally consistent thermody-namic data set for phases of petrological interest: Journal of MetamorphicGeology, v. 16, p. 309–343.

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 245

0361-0128/98/000/000-00 $6.00 245

Holland, T., Baker, J., and Powell, R., 1998, Mixing properties and activity-composition relationships of chlorites in the system MgO-FeO-Al2O3-SiO2-H2O: European Journal of Mineralogy, v. 10, p. 395–406.

Horita, J., and Wesolowski, D.J., 1994, Liquid-vapor fractionation of oxygenand hydrogen isotopes of water from the freezing point to the critical tem-perature: Geochimica et Cosmochimica Acta, v. 58, p. 3425–3447.

Horita, J., Wesolowski, D.J., and Cole, D.R., 1993, The activity-compositionrelationship of oxygen and hydrogen isotopes in aqueous salt solutions. I.Vapor-liquid water equilibration of single salt solutions from 50 to 100°C:Geochimica et Cosmochimica Acta, v. 57, p. 2797–2817.

Horita, J., Cole, D.R., and Wesolowski, D.J., 1995, The activity-compositionrelationship of oxygen and hydrogen isotopes in aqueous salt solutions. III.Vapor-liquid equilibration of NaCl solutions to 350 °C: Geochimica et Cos-mochimica Acta, v. 59, p. 1139–1151.

Jackson, P., Changkakoti, A., Krouse, H.R., and Gray, J., 2000, The origin ofgreisen fluids of the Foley’s zone, Cleveland tin deposit, Tasmania, Aus-tralia: ECONOMIC GEOLOGY, v. 95, p. 227–236.

Johnson, J.W., Oelkers, E.H., and Helgeson, H.C., 1992, SUPCRT92: A soft-ware package for calculating the standard molal thermodynamic propertiesof minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0to 1000 °C: Computers and Geosciences, v. 18, p. 899–947.

Jowett, E.C., 1991, Fitting iron and magnesium into the hydrothermal chlo-rite geothermometer: GAC-MAC-SEG Joint Meeting, Program with Ab-stract, p. A62.

Kelley, S.P., and Fallick, A.E., 1990, High precision spatially resolved analy-sis of δ34S in sulphides using as laser extraction technique: Geochimica etCosmochimica Acta, v. 54, p. 883–888.

Kelley, S.P., Fallick, A.E., McConville, P., and Boyce, A.J., 1992, High preci-sion, high spatial resolution analysis of sulfur isotopes by laser combustionof natural sulfide minerals: Scanning Microscopy, v. 6, p. 129–138.

Kelly, W.C., and Turneaure, F.S., 1970, Mineralogy, paragenesis and geot-hermometry of tin and tungsten deposits of the eastern Andes, Bolivia:ECONOMIC GEOLOGY, v. 65, p. 609–680.

Kojima, S., and Sugaki, A., 1984, Phase relations in the Cu-Fe-Zn-S systembetween 500° and 300°C under hydrothermal conditions: ECONOMIC GE-OLOGY, v. 80, p. 158–171.

Kontak, D.J., 1990, A sulfur isotope study of main-stage tin and base metalmineralization at East Kemptville tin deposit, Yarmouth County, Nova Sco-tia, Canada: evidence for magmatic origin of metals and sulfur: EconomicGeology, v. 85, p. 399-407.

Kontak, D.J., and Clark, A.H., 1988, Exploration criteria for tin and tungstenmineralization in the Cordillera Oriental of southeastern Peru, in Taylor,R.P., and Strong, D.F., eds., Recent advances in the geology of granite-re-lated mineral deposits: Canadian Institution of Mining and Metallurgy,Special Volume 39, p. 157–169.

——2002, Genesis of the giant, bonanza San Rafael lode tin deposit, Peru:origin and significance of pervasive alteration: Economic Geology, v. 97, p.1741-1777.

Kontak, D.J., Clark, A.H., Farrar, E., Pearce, T.H., Strong, D.F., and Baads-gaard, H., 1986, Petrogenesis of a Neogene shoshonite suite, Cerro Moro-moroni, Puno, southeastern Peru: Canadian Mineralogist, v. 24, p.117–135.

Kontak, D.J., Clark, A.H., Farrare, E., Archibald, D.A., and Baadsgaard, H.,1987, Geochronological data for tertiary granites of the southeast Peru seg-ment of the Central Andean tin belt: ECONOMIC GEOLOGY, v. 82, p.1611–1618.

Kontak, D.J., Cumming, G.L., Krstic, D., Clark, A.H., and Farrar, E., 1990,Isotopic composition of lead in ore deposits of the Cordillera Oriental,southeastern Peru: ECONOMIC GEOLOGY, v. 85, p. 1584–1603.

Kotzer, T.G., Kyser, T.K., King, R.W., and Kerrich, R., 1993, An empiricaloxygen- and hydrogen-isotope geothermometer for quartz-tourmaline andtourmaline-water: Geochimica et Cosmochimica Acta, v. 57, p. 3421–3426.

Laubacher, G., 1978, Estudio geologico de la region norte del Lago Titicaca:Peru, Instituto Geologia y Mineria, v. 5, 120 p.

LeBoutillier, N.G., Camm, G.S., Shail, R.K., Bromley, A.V., Jewson, C., andHoppe, N., 2002, Tourmaline-quartz-cassiterite mineralization of theLand’s End granite at Nanjizal, west Cornwall: Proceedings of the UssherSociety, v. 10, p. 312–318.

Lehmann, B., 1990, Metallogeny of tin: Lecture Notes in Earth Sciences, v.32, Berlin,Springer-Verlag, 211 p.

Linnen, R.L., and Williams-Jones, A.E., 1995, Genesis of a magmatic meta-morphic hydrothermal system: The Sn-W polymetallic deposits at Pilok,Thailand: ECONOMIC GEOLOGY, v. 90, p. 1148–1166.

Longstaffe, F.J., Clark, A.H., McNutt, R.H., and Zentilli, M., 1983, Oxygenisotopic compositions of Central Andean plutonic and volcanic rocks, lati-tudes 26°-29° south: Earth and Planetary Science Letters, v. 64, p. 9–18.

Matsuhisa, Y., Goldsmith, J.R., and Clayton, R.N., 1979, Oxygen isotopicfractionation in the system quartz-albite-anorthite-water: Geochimica etCosmochimica Acta, v. 43, p. 1131–1140.

Mattey, D.P., and Macpherson, C.G., 1993, High-precision oxygen isotopemicroanalysis of ferromagnesian minerals by laser fluorination: ChemicalGeology, v. 105, p. 305–318.

McBride, S.L., Robertson, R.C.R., Clark, A.H., Farrar, E., 1983, Magmaticand metallogenic episodes in the northern tin belt, Cordillera Real, Bolivia:International Journal of Earth Sciences, v. 72, p. 685–713.

McCrea, J.M., 1950, On the isotopic chemistry of carbonates and apalaeotemperature scale: Journal of Chemical Physics, v. 18, p. 849–857.

Menzie, W.D., Reed, B.L., and Singer, D.A., 1988, Models of grades and ton-nages of some lode tin deposits, in Hutchison, C.S., ed., Geology of tin de-posits in Asia and the Pacific; mineral concentrations and hydrocarbon ac-cumulations in the ESCAP region: Selected papers from the InternationalSymposium on the Geology of Tin Deposits, Nanning, China, October1984, p. 73–88.

Migdisov, A.A., Williams-Jones, A.E., Lakshtanov, L.Z., and Alekhin, Y.V.,2002, Estimates of the second dissociation constant of H2S from the surfacesulfidation of crystalline sulfur: Geochimica et Cosmochimica Acta, v. 66, p.1713–1725.

Mlynarczyk, M.S.J., and Williams-Jones, A.E., 2005, The role of collisionaltectonics in the metallogeny of the Central Andean tin belt: Earth andPlanetary Science Letters, v. 240, p. 656–667.

——2006, Zoned tourmaline associated with cassiterite: implications forfluid evolution and tin mineralization in the San Rafael Sn-Cu deposit, SEPeru: Canadian Mineralogist, v. 44, p. 347–365.

Mlynarczyk, M.S.J., Sherlock, R.L., and Williams-Jones, A.E., 2003, SanRafael, Peru: geology and structure of the worlds richest tin lode: Mineral-ium Deposita, v. 38, p. 555–567.

Mlynarczyk, M.S.J., Wagner, T., Williams-Jones, A.E., and Dolejs, D., 2009,Geology and geochemistry of alteration at the San Rafael Sn-Cu deposit,SE Peru: ECONOMIC GEOLOGY (in review)

Müller, B., and Seward, T.M., 2001, Spectrophotometric determination ofthe stability of tin (II) chloride complexes in aqueous solution up to 300ºC:Geochimica et Cosmochimica Acta, v. 65, p. 4187–4199.

Nakamura, Y., and Shima, H., 1982, Fe and Zn partitioning between spha-lerite and stannite [abs.]: Joint Meeting of the Society of Mining Geologyof Japan, Association of Mineralogy, Petrology and Economic Geology, andMineralogical Society of Japan, p. A8.

Nekrasov, I.Y., Sorokin, V.I., and Osadchii, E.G., 1976, Partition of iron andzinc between sphalerite and stannite at T = 300 to 500°C and P = 1kb: Dok-lady Earth Sciences, v. 226, p. 136–138.

——1979, Fe and Zn partitioning between stannite and sphalerite and its ap-plication in geothermometry: Physics and Chemistry of the Earth, v. 11, p.739–742.

Oelkers, E.H., and Helgeson, H.C., 1990, Triple-ion anions and polynuclearcompexing in supercritical electrolyte solutions: Geochimica et Cos-mochimica Acta, v. 54, p. 727–738.

Ohmoto, H., 1972, Systematics of sulfur and carbon isotopes in hydrothermalore deposits: ECONOMIC GEOLOGY, v. 67, p. 551–578.

Ohmoto, H., and Goldhaber, M.B., 1997, Sulfur and carbon isotopes, inBarnes, H.L., ed., Geochemistry of hydrothermal ore deposits, 3rd ed:New York, Wiley, p. 517–611.

Pabalan, R.T., 1986, Solubility of cassiterite (SnO2) in NaCl solutions from200°C–350°C, with geologic applications: Unpublished Ph.D. thesis, Penn-sylvania State University, 141 p.

Palma, V.V., 1981, The San Rafael tin-copper deposit, Puno, SE Peru: Un-published M.Sc. thesis, Kingston, Canada, Queen’s University, 235 p.

Pashinkin, A.C., Muratova, V.A., Moiseyev, N.V., and Bazhenov, J.V., 1991,Heat capacity and thermodynamic functions of iron diarsenide in the Trange 5 K to 300 K: Journal of Chemical Thermodynamics, v. 23, p.827–830.

Primmer, T.J., 1985, Discussion on the possible contribution of metamorphicwater to the mineralising fluid of south-west England: Preliminary stableisotope evidence: Proceedings of the Ussher Society, v. 6, p. 224–228.

Reed, M.H., and Spycher, N.F., 1984, Calculation of pH and mineral equi-libria in hydrothermal waters with application to geothermometry andstudies of boiling and dilution: Geochimica et Cosmochimica Acta, v. 48, v.1479–1492.

246 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 246

Robie, R.A., and Hemingway, B.S., 1995, Thermodynamic properties of min-erals and related substances at 298.15 K and 1 Bar (105 Pascals) pressureand at higher temperatures: U.S. Geological Survey Monograph, v. 2131,461 p.

Robie, R.A., Seal, R.R. II, and Hemingway, B.S., 1994, Heat capacity and en-tropy of bornite (Cu5FeS4) between 6 and 760 K and the thermodynamicproperties of phases in the system Cu-Fe-S: Canadian Mineralogist, v. 32,p. 945–956.

Roedder, E., 1984, Fluid inclusions: Reviews in Mineralogy, v. 12, 646 p.Rosenbaum, J., and Sheppard, S.M.F., 1986, An isotopic study of siderites,

dolomites and ankerites at high temperatures: Geochimica et Cosmochim-ica Acta, v. 50, p. 1147–1150.

Ryzhenko, B.N., Shvarov, Y.V., and Kovalenko, N.I., 1997, The Sn-Cl-F-C-S-H-O-Na system: Thermodynamic properties of components within theconditions of the Earth’s crust: Geochemistry International, v. 35, p.1016–1020.

Sandeman, H.A., Clark, A.H., and Farrar, E., 1995, An integrated tectono-magmatic model for the evolution of the Southern Peruvian Andes (13-20°S) since 55 Ma: International Geology Review, v. 37, p. 1039–1073.

Sandeman, H.A., Clark, A.H., Farrar, E., and Arroyo-Pauca, G., 1996, A crit-ical appraisal of the Cayconi Formation, Crucero Basin, southeastern Peru:Journal of South American Earth Sciences, v. 9, p. 381–392.

Schwinn, G., Wagner, T., Markl, G., and Baatartsogt, B., 2006, Quantificationof mixing processes in ore-forming hydrothermal systems by combinationof stable isotope and fluid inclusion analyses: Geochimica et Cosmochim-ica Acta, v. 70, p. 965–982.

Sharp, Z.D., 1990, A laser-based microanalytical method for the in situ de-termination of oxygen isotope ratios of silicates and oxides: Geochimica etCosmochimica Acta, v. 54, p. 1353–1357.

Sharp, Z.D., Atudorei, V., and Durakiewicz, T., 2001, A rapid method for de-termination of hydrogen and oxygen isotope ratios from water and hydrousminerals: Chemical Geology, v. 178, p. 197–210.

Sheppard, S.M.F., 1994, Stable isotope and fluid inclusion evidence for theorigin and evolution of hercynian mineralizing fluids, in Seltmann, R.,Kämpf, H., and Möller, P., eds., Metallogeny of collisional orogens: CzechGeological Survey, Prague, p. 49–60.

Shimitzu, M., and Shikazono, N., 1985, Iron and zinc partitioning betweencoexisting stannite and sphalerite: A possible indicator of temperature andsulfur fugacity: Mineralium Deposita, v. 20, p. 314–320.

Shock, E.L., Oelkers, E.H., Johnson, J.W., Sverjensky, D.A., and Helgeson,H.C., 1992, Calculation of the thermodynamic properties of aqueousspecies at high pressures and temperatures: Journal of the Chemical Soci-ety Faraday Transactions, v. 88, p. 803–826.

Shock, E.L., Sassani, D.C., Willis, M., and Sverjensky, D.A., 1997, Inorganicspecies in geological fluids: Correlations among standard molal thermody-namic properties of aqueous ions and hydroxide complexes: Geochimica etCosmochimica Acta, v. 61, p. 907–950.

Shvarov, Y.V., 1978, Minimization of the thermodynamic potential of an openchemical system: Geochemistry International, v. 15, p. 200–203.

——1981, A general equilibrium criterion for an isobaric-isothermal modelof a chemical system: Geochemistry International, v. 18, p. 38–45.

Shvarov, Y.V., and Bastrakov, E., 1999, HCh: A software package for geo-chemical equilibrium modeling. User’s guide: Australian Geological SurveyOrganization, 61 p.

Smith, M., Banks, D.A., Yardley, B.W.D., and Boyce, A., 1996, Fluid inclu-sion and stable isotope constraints on the genesis of the Cligga Head Sn-Wdeposit, S.W. England: European Journal of Mineralogy, v. 8, p. 961–974.

Spycher, N., and Reed, M.H., 1989, Evolution of a Broadlands-type epither-mal ore fluid along alternative P-T paths: Implications for the transport anddeposition of base, precious, and volatile metals: ECONOMIC GEOLOGY, v.84, p. 328–359.

Sterner, S.M., Hall, D.L., and Bodnar, R.J., 1988, Synthetic fluid inclusions.V. Solubility relations in the system NaCl-KCl-H2O under vapor-saturatedconditions: Geochimica et Cosmochimica Acta, v. 52, p. 989–1005.

Sugaki, A., Kitakaze, A., and Kojima, S., 1990, Sphalerite stars in chalcopy-rite; are they always the results of an unmixing process; discussion: Miner-alium Deposita, v. 25, p. 82–83.

Sun, S., and Eadington, P.J., 1987, Oxygen isotope evidence for the mixing ofmagmatic and meteoric waters during tin mineralization in the Mole gran-ite, New South Wales, Australia: ECONOMIC GEOLOGY, v. 82, p. 43–52.

Sverjensky, D.A., Shock, E.L., and Helgeson, H.C., 1997, Prediction of thethermodynamic properties of aqueous metal complexes to 1000°C and 5kb: Geochimica et Cosmochimica Acta, v. 61, p. 1359–1412.

Taylor, H.P., 1977, Water/rock interactions and the origin of H2O in graniticbatholiths: Journal of the Geological Society London, v. 133, p. 509–558.

——1997, Oxygen and hydrogen isotope relationships in hydrothermal min-eral deposits, in Barnes, H.L., ed., Geochemistry of hydrothermal ore de-posits, 3rd ed.: New York, Wiley, p. 229–302.

Taylor, J.R., and Wall, V.J., 1993, Cassiterite solubility, tin speciation andtransport in a magmatic aqueous phase: ECONOMIC GEOLOGY, v. 88, p.437–460.

Taylor, R.G., 1979, Geology of tin deposits: Developments in Economic Ge-ology, v. 11, 543 p.

Truesdell, A.H., and Nathenson, M., 1977, The effects of boiling and dilutionon the isotopic compositions of Yellowstone thermal waters: Journal ofGeophysical Research, v. 82, p. 3694–3704.

Turneaure, F.S., 1960a, A comprarative study of major ore deposits in Bolivia:ECONOMIC GEOLOGY, v. 55, p. 217–254.

——1960b, A comparative study of major ore deposits in Bolivia. Part II:ECONOMIC GEOLOGY, v. 55, p. 574–606.

Vidal, O., Parra, T., and Trottet, F., 2001, A thermodynamic model for Fe-Mgaluminous chlorite using data from phase equilibrium experiments and nat-ural pelitic assemblages in the 100-600°C, 1-25 kb range: American Jour-nal of Science, v. 301, p. 557–592.

Wagner, T., Boyce, A.J., and Fallick, A.E., 2002, Laser combustion analysis ofδ34S of sulfosalt minerals: Determination of the fractionation systematicsand some crystal-chemical considerations: Geochimica et CosmochimicaActa, v. 66, p. 2855–2863.

Wagner, T., Boyce, A.J., Jonsson, E., and Fallick, A.E., 2004, Laser micro-probe sulphur isotope analysis of arsenopyrite: experimental calibrationand application to the Boliden Au-Cu-As massive sulphide deposit: OreGeology Reviews, v. 25, p. 311–325.

Wagner, T., Williams-Jones, A.E., and Boyce, A.J., 2005, Stable isotope-basedmodeling of the origin and genesis of an unusual Au-Ag-Sn-W epithermalsystem at Cirotan, Indonesia: Chemical Geology, v. 219, p. 237–260.

Walshe, J.L., 1986, A six-component chlorite solid-solution model and theconditions of chlorite formation in hydrothermal and geothermal systems:ECONOMIC GEOLOGY, v. 81, p. 681–703.

Walshe, J.L., Halley, S.W., Anderson, J.A., and Harrold, B.P., 1996, The in-terplay of groundwater and magmatic fluids in the formation of the cassi-terite-sulfide deposits of western Tasmania: Ore Geology Reviews, v. 10, p.367–387.

Wiggins, L.B., and Craig, J.R., 1980, Reconnaissance of the Cu-Fe-Zn-S sys-tem: sphalerite phase relationship: ECONOMIC GEOLOGY, v. 75, p. 742–751.

Williamson, B.J., Spratt, J., Adams, J.T., Tindle, A.G., and Stanley, C.J., 2000,Geochemical constraints from zoned hydrothermal tourmalines on fluidevolution and tin mineralization: an example from fault breccias at Roche,SW England: Journal of Petrology, v. 41, p. 1439–1453.

Wilkinson, J.J., Jenkin, G.R.T., Fallick, A.E., and Foster, R.P., 1995, Oxygenand hydrogen isotopic evolution of Variscan crustal fluids, south Cornwall,U.K: Chemical Geology, v. 123, p. 239–254.

Zhang, L., Liu, J., Chen, Z., and Zhou, H., 1994, Experimental investigationsof oxygen isotope fractionation in cassiterite and wolframite: ECONOMICGEOLOGY, v. 89, p. 150–157.

Zhang, X., and Spry, P.G., 1994, FO2PH: a quickBASIC program to calculatemineral stabilities and sulphur isotope contours in log fO2-pH space: Min-eralogy and Petrology, v. 50, p. 287–291.

Zheng, Y.F., 1993, Calculation of oxygen isotope fractionation in hydroxyl-bearing silicates: Earth and Planetary Science Letters, v. 120, p. 247–263.

STABLE ISOTOPE CONSTRAINTS—SAN RAFAEL SN-CU DEPOSIT, SE PERU 247

0361-0128/98/000/000-00 $6.00 247

Thermodynamic CalculationsWe modeled the depositional conditions prevalent during

the main cassiterite stage and the late sulfide stage at SanRafael by constructing phase diagrams showing aqueous andmineral equilibria in the Cu-Fe-Sn-As-S-O-H system andcontours of sulfur isotope fractionation. In addition, we car-ried out a series of speciation calculations in the model systemSn-Na-Cl-O-H in order to obtain the total solubility of Sn asa function of pH and the oxidation state of the hydrothermalfluid. All the calculations were carried out with the HCh soft-ware package (Shvarov and Bastrakov, 1999), which modelsheterogeneous equilibria and reaction progress by minimiza-tion of the Gibbs free energy of the total system (Shvarov,1978, 1981). Thermodynamic data for most aqueous specieswere taken from the SUPCRT92 database and subsequentupdates (Johnson et al., 1992; Shock et al., 1997; Sverjenskyet al., 1997). Data for a number of aqueous Sn species camefrom Ryzhenko et al. (1997). Thermodynamic data for rock-forming silicate and oxide minerals were taken from the in-ternally consistent dataset of Holland and Powell (1998).

The data for cassiterite, pyrite, chalcopyrite, bornite, andpyrrhotite (not contained in this dataset) were compiled fromRobie and Hemingway (1995). As heat capacity functions forpyrrhotite and bornite are not given in Robie and Hemingway(1995), the experimentally determined heat capacity data

from the original sources (pyrrhotite: Grønvold and Stølen,1992; bornite: Robie et al., 1994) were fitted with a four-termpolynominal of the form used by Holland and Powell (1998).It should be noted that the entropy values for bornite andpyrrhotite given in Robie and Hemingway (1995) also origi-nate from Grønvold and Stølen (1992) and Robie et al.(1994), thereby ensuring internal consistency of these datasets.Data for calculating the stability fields of arsenic phases (ar-senopyrite, loellingite, native arsenic) were compiled fromRobie and Hemingway (1995) and Barton (1969). The heatcapacity function for loellingite was fitted to the original datagiven in Pashinkin et al. (1991), which were the source of theentropy data tabulated in Robie and Hemingway (1995). Allcalculations of individual activity coefficients of aqueousspecies applied an extended Debye-Hückel model using theb-gamma equation for NaCl as the background electrolyte(Oelkers and Helgeson, 1990; Shock et al., 1992). The set ofequations of Zhang and Spry (1994) and the most recent setof isotopic fractionation factors given in Ohmoto and Gold-haber (1997) were used to calculate the sulfur isotope con-tours. The model of Zhang and Spry (1994), which excludesthe aqueous S2– species, is preferred over the original formal-ism of Ohmoto (1972), because the second dissociation con-stant of H2S is too small for S2– to be a significant species atgeologically realistic values of pH (Migdisov et al., 2002).

248 WAGNER ET AL.

0361-0128/98/000/000-00 $6.00 248

APPENDIX