Geochimica et Cosmochimica Acta, Vol. 69, No. 22, pp. 5233–5246, 2005Copyright © 2005 Elsevier Ltd

Printed in the USA. All rights reserved

doi:10.1016/j.gca.2005.06.022

Cu isotopic fractionation in the supergene environment with and without bacteria

RYAN MATHUR,1,* JOAQUIN RUIZ,1 SPENCER TITLEY,2 LAURA LIERMANN,3 HEATHER BUSS,3 and SUSAN BRANTLEY3

1Department of Geology, Juniata College, Huntingdon, Pennsylvania 16652, USA2Department of Geosciences, University of Arizona, Tucson, Arizona 85721, USA

3Department of Geosciences, Pennsylvania State University, State College, University Park, Pennsylvania 16802, USA

(Received February 28, 2005; accepted in revised form June 21, 2005)

Abstract—The isotopic composition of dissolved Cu and solid Cu-rich minerals [�65Cu (‰) � (65Cu/63Cusample/

65Cu/63Custd) - 1)*1000] were monitored in batch oxidative dissolution experiments with andwithout Thiobacillus ferrooxidans. Aqueous copper in leach fluids released during abiotic oxidation of bothchalcocite and chalcopyrite was isotopically heavier (�65Cu � 5.34‰ and �65Cu � 1.90‰, respectively,[�0.16 at 2�]) than the initial starting material (�65Cu � 2.60 � 0.16‰ and �65Cu � 0.58 � 0.16‰,respectively). Isotopic mass balance between the starting material, aqueous copper, and secondary mineralsprecipitated in these experiments explains the heavier isotopic values of aqueous copper. In contrast, aqueouscopper from leached chalcocite and chalcopyrite inoculated with Thiobacillus ferrooxidans was isotopicallysimilar to the starting material. The lack of fractionation of the aqueous copper in the biotic experiments canbest be explained by assuming a sink for isotopically heavy copper present in the bacteria cells with �65Cu � 5.59� 0.16‰. Consistent with this inference, amorphous Cu-Fe oxide minerals are observed surrounding cellmembranes of Thiobacillus grown in the presence of dissolved Cu and Fe.

Extrapolating these experiments to natural supergene environments implies that release of isotopicallyheavy aqueous Cu from oxidative leach caps, especially under abiotic conditions, should result in precipitatesin underlying enrichment blankets that are isotopically heavy. Where iron-oxidizing cells are involved,isotopically heavy oxidized Cu entrained in cellular material may become associated with leach caps, causingthe released aqueous Cu to be less isotopically enriched in the heavy isotope than predicted for the abioticsystem. Rayleigh fractionation trends with fractionation factors calculated from our experiments for bothbiotic and abiotic conditions are consistent with large numbers of individual abiotic or biotic leaching events,explaining the supergene chalcocites in the Morenci and Silver Bell porphyry copper deposits. Copyright

0016-7037/05 $30.00 � .00

© 2005 Elsevier Ltd

1. INTRODUCTION

Recent studies have demonstrated that biologic and kineticprocesses cause measurable isotopic shifts for stable transitionmetal isotopes (Shields et al., 1965; Halliday et al., 1998; Galeet al., 1999; Maréchal et al., 1999; Anbar et al., 2000; Zhu etal., 2000; Brantley et al., 2001; Bullen et al., 2001; Matthews etal., 2001; Zhu et al., 2002; Ellis et al., 2002; Beard et al., 2003;Anbar, 2004). In natural samples, copper exhibits the greatestvariation in isotopic composition (up to 9‰) of the transitionmetals (Johnson et al., 2003; Larson et al., 2003).

An ideal natural location to observe the fractionation oftransition metal isotopes is the supergene enrichment environ-ment. The supergene enrichment process involves leaching ofmetal-rich igneous rocks that contain significant concentrationsof sulfide minerals. The interaction of pyrite and similar min-erals with low-temperature aqueous fluids in the so-calledleached capping generates dilute sulfuric acid, which in turnfurther dissolves the sulfide minerals releasing aqueous oxi-dized metals. These metals subsequently precipitate in deeperhorizons after oxidants are consumed, and sulfides determinethe redox state (e.g., Brimhall, 1980; Guilbert and Park, 1986;Titley, 1995). The process naturally concentrates metals inwhat is termed the enriched supergene blanket below the watertable. As erosion progresses and/or climate changes, water

* Author to whom correspondence should be addressed(mathur@juniata.edu).

5233

table fluctuations can lead to subsequent leaching of the en-riched supergene blanket and reprecipitation in new enrichmentblankets at depth (Alpers and Brimhall, 1989). The dominantCu-containing sulfide that leaches in the oxidative vadose zoneto initiate the supergene leach cycle is chalcopyrite (CuFeS2),whereas the dominant precipitate in enrichment blankets ischalcocite (Cu2S).

The behavior and distribution of copper during enrichment iscomplex. The oxidative dissolution process does not com-pletely remove all of the Cu in the leach cap as evidenced byCu occurring in both primary chalcopyrite and secondary sul-fate minerals such as chalcanthite, brochantite, and antlerite inthe leach cap. In contrast, chalcocite (Cu2S) and other relatedcopper sulfides dominate the mineralogy in enrichment blan-kets; chalcocite probably forms by the progressive replacementof chalcopyrite (CuFeS2) and pyrite (FeS2) (Brimhall et al.,1985). Stokes (1970) demonstrated the importance of pyrite asa template for precipitating chalcocite, although Sillitoe et al.(1996) found bacteria forms in etched chalcocite from ancientsupergene systems and argued that nanobacteria promoted thefixation of copper in the supergene environment.

To begin to understand the biologic, kinetic, and equilibriumchemical behavior of copper during this process, we designedseveral batch leach experiments to compare with samples takenfrom ancient supergene environments. The experiments allowus to compare the nature of both redox and precipitation reac-tions working simultaneously during the biotic and abiotic

leaching of chalcopyrite and chalcocite.

5234 R. Mathur et al.

1.1. Chemistry of Leaching

In an effort to understand the biotic and abiotic mechanismsthat may cause Cu fractionation, it is worthwhile to comparethe process of industrial leaching of copper ores to the super-gene environment (Dold and Fonbonte, 2001; Dold, 2003). Theindustrial process involves placing large quantities of crushedchalcocite ore (�2 wt% Cu) into piles called leach pads. Anetwork of plumbing distributes dilute sulfuric acid on top ofthe leach pads. Chemically resistant pads underneath the pilesof ore collect the percolated copper-rich solution. Electrolyticprocesses are then used to extract copper from the solution.

Oxidative leaching of chalcopyrite with ferric sulfuric acidsresults in the release of Cu2� to solution. Other reactionsoccurring on the surface of the chalcopyrite involve the pre-cipitation of native sulfur, polysulfides, and/or iron hydroxides(Dutrizac et al. 1969; Hackel et al., 1995; Stott et al., 2000).Mineral precipitates are thought to armor the surface of chal-copyrite and cause the relatively slow kinetics associated withleaching of chalcopyrite.

The extraction of Cu from chalcocite (Cu2S) by ferric ion-containing sulfuric acid also causes the release of aqueousCu2�. As copper is lost, chalcocite is replaced sequentially bydigenite (Cu1.96S), anilite (Cu1.75S), geerite (Cu1.6S), spionko-pite (Cu1.4S), yarrowite (Cu1.1S), and covellite (CuS) (Goble,1985; Whiteside and Goble, 1986). The proportion of Cu2�

increases in the precipitated mineral phase until all of the Cu inthe final product, covellite, is thought to be Cu2�. However,Brelle et al. (2000) analyzed nanoparticles of CuxS phases anddiscovered that covellite precipitated from copper-rich solu-tions contains as much as 33% Cu1�. Thus, the leachingprocess generally does not completely oxidize all of the copperin the system. A summary reaction describing formation ofcovellite from chalcocite during leaching is (Lizama, 2001):

Cu2S � 2Fe(aq)3� ? CuS � Cu(aq)

2� � 2Fe(aq)2� (1)

As described in this reaction, Fe3� acts as the oxidant for theCu mineral. Aqueous ferric ions can also oxidize the S insulfide minerals, as summarized in this composite reaction forpyrite oxidation (e.g., Singer and Stumm, 1968; Amend andShock, 2001):

FeS2 � 14Fe(aq)3� � 8H2O ? 15Fe(aq)

2� � 2SO4(aq)� � 16H(aq)

�

(2)

If similar sulfur-oxidizing reactions are written for coppersulfide oxidation, mineral precipitates with Cu2� can includesuch minerals as chalcanthite (CuSO4·5H2O), brochantite(Cu4SO4(OH)6), and antlerite (Cu3SO4(OH4)). The chemistryof precipitates and aqueous leach fluids during supergeneleaching thus depends upon the chemistry and extent of leach-ing and transport.

Ample evidence suggests that acidiphilic organisms existand thrive in leach pads and in natural supergene-leachingenvironments. Enders (2000) and Hong et al. (2000) discoveredThiobacillus ferrooxidans in the fluids of currently active su-pergene zones of a porphyry copper deposit and Carlin golddeposit, respectively. This species catalyzes the oxidation ofsulfur and/or metal, which in turn leads to the solubilization of

copper (Colmer and Hinkle, 1947). Several studies confirm that

the rates and mechanisms of the oxidation in the presence of T.ferrooxidans (e.g., Breed and Hansford, 1999; Hansford andVargas, 2001; Crundell, 2003) affect the release of copper inindustrial leach pad environments and enhance the copperyields. Singer and Stumm (1968) suggests that the rate-deter-mining step of pyrite oxidation is oxidation of ferrous to ferriciron in solution under low pH conditions. T. ferrooxidans isthought to catalyze this reaction.

This brief overview of the controls of copper sulfide leachingprovides fundamental information of the multiple and inter-twined biologic and kinetic mechanisms that may influencecopper isotope fractionation.

2. EXPERIMENTAL DESIGN

The experimental design is composed of two 30-day batch experi-ments, one containing copper sulfides to monitor leaching of mineralsand one with bacteria and copper-rich medium to monitor copperuptake into the cells without sulfide minerals present. A 24-h dissolu-tion experiment containing copper sulfide minerals was also run toinvestigate the effect of partial dissolution. Other measurements ofcopper isotopes from completely dissolved copper-rich minerals asso-ciated with the hypogene and supergene mineralization of porphyrycopper deposits sampled from many localities around the world wereconducted.

2.1. Leach Experiments

In the 30-day leach experiments, copper was leached from chalcopy-rite from the El Teniente porphyry copper mine (e.g., Skewes andStern, 1994; Maksaev et al., 2004) and from chalcocite purified fromseveral ores in the Morenci porphyry deposit (described in Enders[2000]). The samples, exhibiting grain sizes varying from 1- to 2-mmgranules to 1.5- to 2-mm very coarse sand, were homogenized beforeuse. As shown by X-ray diffraction (XRD), the chalcopyrite containedonly chalcopyrite, whereas the chalcocite contained two observablephases, chalcocite and quartz. However, the amount of quartz is min-imal (�10%) as indicated by XRD and optical petrography.

Chalcopyrite and chalcocite were leached in 250 mL flasks contain-ing dilute sulfuric acid (pH 2.3) medium chosen to enhance the growthof T. ferrooxidans (13598 from The American Type Culture Collection[ATCC] Manassas, VA). The medium contained 0.8 g l-1 (NH4)2SO4,2.0 g l-1 MgSO4 · 7H2O, 0.4 g l-1 K2HPO4, 20.0 g l-1 FeSO4 · 7H2O,and 5.0 mL l-1 Wolfe’s mineral solution (Wolin et al., 1963). The pHof the solution was adjusted to 2.3 with sulfuric acid, and the mediumwas filter-sterilized (4.5 micron). A 1.56 g sample of either chalcopyriteor chalcocite was autoclaved in each of 12 flasks (six of each mineral),to which 200 mL of the sterile medium was aseptically transferred. Theexperimental conditions were (1) chalcocite � medium (n � 3), (2)chalcocite � medium � bacteria (n � 3), (3) chalcopyrite � medium(n � 3), and (4) chalcopyrite � medium � bacteria (n � 3); where n� number of replicates of each condition (Table 1). Where indicated,flasks were inoculated with 100 �L of T. ferrooxidans in late-log tostationary phase of growth. The inoculated experiments are termedbiotic and the experiments which were not inoculated are termedabiotic.

The two sets of six flasks were capped with sterilized plastic capsand set in a shaker-incubator (New Brunswick Scientific) at a constanttemperature of 25°C. Approximately every 7 to 10 days, one inoculatedand one noninoculated flask was removed from each set. The solutionsfrom these flasks were filtered through a 0.2 �m filter and the remnantmineral grains were separated and rinsed ultrasonically three times inacetone. During the experiment, yellow-orange minerals encrusted themineral surfaces. The fine-grained coatings could not be separated fromthe copper mineral residues.

The dried solid mineral grains were separated into two aliquots. A0.1 g aliquot was separated for copper isotope analysis; this portion wasplaced in a 5 mL Teflon beaker and dissolved in concentrated nitric

acid (8 N). Approximately 0.2 g of the other aliquot was powdered forXRD analysis on a Scintag Pad V X-ray powder diffractometer. XRD

eous an

5235Cu isotopes in the supergene environment

scans were completed in slow, step-scan mode for precision analysis.Some minerals might be present but not detected if they occur at lessthan 5 vol% of the material analyzed or are amorphous.

In the inoculated experiments, evidence of cell growth was deter-mined by a characteristic color change in the medium (from a yellow-brown to a reddish-brown, which was not observed to occur in theabsence of cells), and by occasional cell count estimates under a lightmicroscope.

2.2. Bacterial Uptake Experiments

The bacterial uptake experiments were designed to monitor the Cucontent and Cu isotopic composition of the T. ferrooxidans. Thisexperiment was conducted because a Percol density gradient separationof T. ferrooxidans from the mineral grains in the preceding experimentwas not successful. This organism has been renamed to Acidiphiousferrooxidans (Kelly and Wood, 2002).

To run the bacterial uptake experiments, an aliquot from a dissolved,pure (99.99%) copper nugget (2.45 g, from the Phelps Dodge’sMorenci processing plant) in 10 mL of concentrated nitric acid (8 N)was mixed into the Thiobacillus medium to a final concentration of 50mM copper. Nielsen and Beck (1972), Leduc et al. (1997), Boyer et al.(1998), Gericke and Pinches (1999) demonstrated that Thiobacillus canexist in solutions with as much as 400 mM copper.

Six flasks were set up with 150 mL of this copper-rich medium,inoculated with Thiobacillus as described above, and set on a rotatingshaker for 30 days. Three flasks were used to determine the copperisotopic composition of the bacteria. Two flasks were used to determinethe concentration of copper in the bacteria, and one flask was used forimaging by transmission electron microscopy (TEM) and scanningelectron microscopy (SEM). Bacteria were also analyzed using energy-dispersive X-ray spectroscopy (EDS). The Cu concentration of themedium for this experiment remained the same after 30 days ofbacterial growth (initial [Cu] � 50 mM, final [Cu] � 50 mM). Thisconcentration represented the approximate midway point between thetwo final aqueous copper concentrations of the chalcopyrite and chal-cocite biotic leach experiments after 30 days ([Cu] � 110 mM and 14mM, respectively).

Two of the three flasks were combined for copper isotope analysis to

Table 1. Results of 30 leach expe

SampleCollection

day �65Cumin‰* �65Cuaq‰* p

Chalcocite 2.60 –

Abiotic7 1.90 5.34 1

14 1.99 4.68 230 1.67 4.54 2

Biotic7 2.34 2.96 1

14 2.75 2.63 330 0.69 2.81 3

Chalcopyrite 0.58 –

Abiotic7 0.37 1.90

14 0.27 1.9030 0.59 1.51

Biotic7 0.69 0.79

14 0.33 0.8630 0.29 0.81

* �65Cu reported in per mill with respect to the 976NIST standard,** % of Cu in starting material leached.***�aq-min�(�65Cuaq � 1000)/(�65Cumin � 1000), where aq � aqu

increase the cell mass and entrained total copper. The cells werepelleted by centrifugation and washed with 18 � water three times to

ensure that the medium was completely rinsed from the pellet. Thebacteria pellets were dried, weighed, and acid-digested for copperanalysis.

Cells were also pelleted and washed as described above from theflask set up for TEM and EDS analysis. Single drops of the pellet wereplaced on Formvar carbon-coated glow-discharged nickel TEM gridsand incubated for 3 min; after wicking away excess solution, 2%aqueous uranyl acetate was added, then the samples were incubated for30 s in the dark, wicked again, and dried overnight before TEMviewing.

2.3. Partial Dissolution Experiments

In the 24-h partial dissolution experiments, copper was leachedincompletely from chalcopyrite and bornite (Cu5FeS4). The chalcopy-rite sample was a mineral concentrate from the Ertsberg District, IrianJaya (described in Mathur et al. [2000]). The bornite sample was fromCollahuasi, Chile (described in Masterman et al. [2004]). The sampleshave a grain size similar to the El Teniente and Morenci samplesdiscussed above, and no other phases were observed in the XRDpatterns of the samples.

Instead of using sulfuric acid to leach chalcocite and chalcopyrite(which is known to produce phase changes during partial leaching),nitric acid was used to leach the chalcopyrite and bornite. Linge (1976)and others have noticed that leaching chalcopyrite with nitric acid doesnot produce significant mineralogical phase changes as found whenleaching chalcocite. With this in mind, 0.2 g of copper sulfide wasplaced (pure chalcopyrite or bornite) in 10 mL of heated 8N nitric acidovernight. After leaching, the remaining solid materials were dried,weighed, and dissolved in 5 mL Teflon beakers. XRD analyses ofpowdered samples were used to identify mineralogical changes.

2.4. Copper Isotopic Composition of Minerals From PorphyryCopper Deposits

The copper isotopic composition of hypogene chalcopyrites fromveins that contained high-temperature alteration silicate minerals (suchas biotite and potassium feldspar) from porphyry copper depositsaround the world (Table 2) were analyzed. In addition, supergene

with chalcocite and chalcopyrite.

Feppm % Cu leached**

Phases identified byXRD �aq-min**

– – chalcocite � quartz –

3620 21.4 digenite � quartz 1.00343730 35.7 covellite � quartz 1.00271870 36.4 covellite � quartz 1.0029

2820 21.5 digenite � quartz 1.000692 53.8 covellite � quartz 0.9999

1100 56.7 covellite � quartz 1.0021

– – chalcopyrite –

4170 11.3 chalcopyrite 1.00153430 15.8 chalcopyrite 1.00162000 24.0 chalcopyrite 1.0009

2120 10.2 chalcopyrite 1.00012050 14.5 chalcopyrite 1.00052130 16.7 chalcopyrite 1.0005

onstrated in Eqn. 1, with 2� � 0.16‰.

d min � mineral remaining at end of experiment.

riments

Cupm

–

320200240

320310490

–

301420634

272386444

as dem

chalcocites taken from enrichment blankets from Morenci and SilverBell porphyry copper deposits in Arizona were analyzed. For each

5236 R. Mathur et al.

analysis, �0.2 g of each copper mineral were completely dissolved inheated 8N nitric acid overnight and analyzed isotopically (next sec-tion).

3. ANALYTICAL METHODS

3.1. Concentrations of Cu and Fe

Concentrations of Cu and Fe were measured by high resolutioninductively-coupled-plasma mass spectrometry (HR-ICP-MS; FinniganMAT Element I). All solution samples were acidified and diluted in 2%nitric acid for chemical analysis. Fe and Cu concentrations were de-termined by standard calibrations with the instrument in medium res-olution and indium was used as an internal standard.

Cell pellets from the 30-day biotic experiment were rinsed four timeswith 18 � water to remove the copper-rich medium and then driedovernight. The dried pellet was weighed and dissolved in 2 mL of7NHCl � 0.01% peroxide, and copper was separated from the pelletthrough wet ion-exchange column chemistry (described in detail be-low).

3.2. Copper Isotope Analysis

Sample preparation for Cu isotope analysis on the Micromass Iso-probe (multicollector inductively-coupled-plasma mass spectrometer[MC-ICPMS] at the University of Arizona) varied for the three types ofsamples analyzed: fluid digestates from copper minerals containing�10 mM Cu, leach fluids with aqueous copper �0.15 mM Cu, andfluids from bacteria that contained 30 mM copper.

Fluid digestates and leach fluids from experiments with �10 mM Cudid not require ion chromatography for chemical separation. All of thecopper analyses for these samples were centrifuged and diluted with2% nitric acid to 0.008 mM Cu for analysis. Because copper is thedominant element in solution, the dilution eliminated any possiblematrix effects from other ions in solution (Zhu et al., 2000).

Table 2. �65Cu‰ from high-temperature copper sulfides.

Sample name/Location Phase �65Cu‰* Session**

Tt-1, El Teniente bornite 0.37 1Tt-1, El Teniente bornite 0.37 1Tt-1, El Teniente bornite 0.41 2Tt-2, El Teniente chalcopyrite 0.01 1Tt-2, El Teniente chalcopyrite �0.07 2Tt-3, El Teniente chalcopyrite 0.31 1Tt-4, El Teniente chalcopyrite 0.21 1Tt-5, El Teniente chalcopyrite �0.70 1Tt-6, El Teniente chalcopyrite �0.15 1Tt-7, El Teniente chalcopyrite �0.51 1Tt-8, El Teniente chalcopyrite 0.58 1El-1, El Salvador chalcopyrite 0.81 1Esc-1, Escondida chalcopyrite 0.42 1C-1, Chuquicamata chalcopyrite �0.09 1C-1, Chuquicamata chalcopyrite 0.17 2C-2, Chuquicamata chalcopyrite �0.11 1C-2, Chuquicamata chalcopyrite �0.03 2Coya-1, Collahuasi chalcopyrite 0.38 1Coya-2, Collahuasi bornite 0.64 1M-1, Mocha chalcopyrite �0.06 1T-1, Toquepala chalcopyrite �0.44 1Ers-1, Ertsberg chalcopyrite 0.17 1Gras-1, Grasberg chalcopyrite �0.18 1Pan-1, Panguna chalcopyrite �0.39 1

Notice that all values are within the reported 2� � 0.16‰.* �65Cu reported in per mill with respect to the 976NIST standard, as

demonstrated in Eqn. 1, with 2� � 0.16‰.** Indicates within-run and between-session variability of the Cu

ratio measured.

In contrast, a wet ion exchange chromatography (IEC) procedurewas used to separate Cu from the solutions derived from bacteria

pellets. IEC is necessary because the concentration of other ions in thepellet is high and could cause matrix effects when analyzing for copper.For example, test solutions were analyzed that contained proportions ofthe NIST copper isotope standard mixed with ICPMS standard solu-tions of Si and Zn at various molar ratios. Si and Zn were investigatedbecause Si impurities may have resulted due to the quartz in thechalcocite sample, whereas Zn is added to all analyses for mass biascorrection. The standard solutions analyzed contained 0.008 mM Cu(500ppb in 10 mL aliquot), and the other elements were added to thesolutions to attain different proportions of Cu:element to test for amatrix effect. Figure 1 demonstrates that matrix effects are not commonwhen Si or Zn are present at twice the molar concentration of thecopper because the error associated with the standard analysis does notchange significantly. However, when the proportions of Cu:element aregreater than a factor of two, it is observed that the analyzed value forthe test solution lay outside the 2� range for the standard.

To avoid these and other matrix effects, IEC was conducted on thesolutions of dissolved bacteria pellets. Similar types and concentrationsof acids for column elution (7N HCl � 0.001%H2O2) and resin (AGMP-1 anion exchange resin) were used as in Maréchal et al. (1999).The resins were rinsed and decanted five times in 18 � water to removeparticle fines, then settled in 18 � water. The volume of resin used was1.6 mL, and we followed the purification protocol for copper outlinedby Maréchal et al. (1999). The IEC procedure was tested with a mixedsolution prepared to contain 0.87 mM iron and 1.9 mM copper. Onehundred � 11% recovery of copper and no iron were found in the IECseparation of this solution. In these experiments, the copper startedcoming off the resin at the 8th mL rather than the 10th mL; therefore,copper elutions were collected between the 8th and 18th mL of elutionfrom the column (as noted in Table 2).

It has been well-documented that the ion-exchange resin can con-tribute to fractionation of transition metal isotopes because of fraction-ation between adsorbed metal on the ion exchange sites within the resinand the aqueous ions in eluted fluids (Anbar et al., 2000; Maréchal andAlbaréde, 2002). To be certain that the resin used for separation did notcause fractionation of copper in the T. ferrooxidans pellet samples, weanalyzed the Cu isotopic composition of a high-concentration copperfluid we prepared and that was processed through the IEC. Withoutusing the IEC chemistry to separate copper (in other words, simplydiluting the high-concentration copper sample), the high-concentrationcopper fluid was measured to have a �65Cu � 5.3 � 0.16‰ (abiotic day7; Table 1). Consistent with this, the same undiluted sample aftertreatment by IEC yielded a value for �65Cu within error of the dilutedsample (abiotic day 7; Table 3). Thus, at the percent recovery level of

0.5

0.4

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

-0.4

-0.5

2 4 6 8

500ppb NIST 976 Cu std + 500ppb Zn std

Day of Cu std analysis

65

�C

u

500ppb NIST 976 Cu std + 1ppm Zn std 500ppb NIST 976 Cu std + 2 ppm Si std 500ppb NIST 976 Cu std + 4 ppm Si std

Fig. 1. Measured �65Cu for a Cu standard spiked with Zn and Si,plotted as a function of the analytical sessions over 153 days. Observed2� � 0.16‰ for n � 76. Matrix effects (measured values outside 2�,as indicated by open symbols) were observed for solutions containing

65

�2 ppm of Zn or Si in solution. � Cu presented as per mill withrespect to the 976NIST copper standard.

5237Cu isotopes in the supergene environment

our IEC treatment, the chromatography does not produce isotopicfractionation in these samples.

Once the high-copper samples were diluted and the copper in thebacteria pellets was purified with IEC, the solutions were injected intothe MC-ICPMS using a microconcentric nebulizer to increase sensi-tivity. All solutions were diluted and prepared so that the concentrationof copper in the solutions was 500 ppb Cu. The nebulizer flow wasadjusted so that the intensity of the 63Cu beam remained constant at 2V. Because the signals were so large, our “blank” or background peaksof copper at 4 mV are insignificant when considering the overall error.

Two sets of 20 ratios were collected for every sample measured. Allratios are reported here according to the following equation:

�65Cu‰ ���65 Cu

63 Cusample

65 Cu

63 Custd �� 1� 1000 (3)

where the standard (std) was the NIST 976 Cu standard. Results arereported as an average for each run. The standard deviation for mea-sured �65Cu within each run was less than 0.01‰.

However, the major concern surrounding isotope data obtained dur-ing an analytical session is the measurement error associated with massfractionation within the instrument due to variations in operating con-ditions (Maréchal et al., 1999). To constrain the errors associated withcopper isotope analyses on our instrument, we compared all of thecopper ratios to the NIST 976 copper standard (Eqn. 3) using standard-sample-standard bracketing. The 2� error for the variation of thestandard for eight analytical sessions over 5 months was observed to be�0.16‰ (Fig. 1). Calculated errors using the Zn doping (Maréchal etal., 1999) technique to correct for mass bias produced similar errors.

To test long-term reproducibility further, both fluid digestates fromcopper minerals and leach fluids with aqueous copper were analyzed ondifferent run days. Results from these tests (Table 2) demonstrate thatthe isotopic ratios of samples are reproducible within the reported error.The errors for these measurements are larger than those presented inMaréchal et al. (1999) and Zhu et al. (2000, 2002). These workersreport errors of �0.1‰ and �0.02‰, respectively, for long-termreproducibility; however, our errors are small compared to the isotopicshifts observed for experiments reported here.

To demonstrate the reliability of the measurements further, weanalyzed high-temperature, hypogene copper-rich minerals (chalcopy-rite and bornite) collected from porphyry copper deposits around theworld (Table 2; Fig. 2). These samples were dissolved in a similarmanner as the solid copper mineral precipitates. As demonstrated byMaréchal et al. (1999), Zhu et al. (2000, 2002), Larson et al. (2003),and Graham et al. (2004), copper minerals that form in high-tempera-ture environments do not demonstrate significant copper isotopic frac-tionation when compared to the NIST 976 Cu standard; deviation fromthe NIST 976 standard of these copper-rich minerals is approximately�1‰. Figure 2 demonstrates that the hypogene minerals analyzed in

Table 3. �65Cu‰ for column elutions.

Elluant Volume Matrix analyzed* % Cu** �65Cu‰

Rinse 21mL 0.5 NHNO3 bd nmRinse 6mL 7NHCl � 0.01% H2O2 bd nmSample 2mL 7NHCl � 0.01% H2O2 bd nmRinse 8mL 7NHCl � 0.01% H2O2 bd 11.65Rinse 2mL MO water bd nmCollect 3mL 7NHCl � 0.01% H2O2 21.0 5.42Collect 7mL 7NHCl � 0.01% H2O2 79.1 5.42Rinse 2mL MO water bd nmRinse 7mL 7NHCl � 0.01% H2O2 bd nm

bd � below detection limit; nm � not measured.* Indicates the normality of acids used.** Indicates % of total Cu eluted from the column.

this contribution are in accordance with previous copper isotope mea-surements of similar materials.

4. RESULTS

4.1. Results of 30-Day Leach Experiments

XRD patterns do not reveal any phase changes for thechalcopyrite 30-day batch experiments, whereas diffractionpatterns from the chalcocite leach experiments reveal that chal-cocite (Cu2S) changed to digenite (Cu9S5) in 7 days and tocovellite (CuS) after 14 days. Table 1 summarizes the phasesobserved from the XRD patterns. XRD does not identify amor-phous precipitates. For instance, although yellow-orange coat-ings were observed on the residual minerals throughout theleach period for all the 30-day batch experiments with coppersulfide minerals, the XRD did not reveal the presence of sec-ondary precipitates of crystalline iron oxide.

The fraction of the total starting Cu present as aqueous Cu(Table 1) increased with time for all experiments, but washigher for chalcocite than chalcopyrite at every leach period. Inchalcopyrite experiments, aqueous concentrations (Fig. 3) ofcopper were lower in the biotic as opposed to abiotic experi-ments. In contrast, in the chalcocite experiments the concen-trations of aqueous copper in the biotic experiments weregreater than the abiotic experiments (Table 1; Fig. 3). Theconcentration of iron in leach solutions for the biotic andabiotic experiments with chalcopyrite and chalcocite generallydecreased with time (Table 1; Fig. 3).

After the 30-day batch experiments with chalcocite, mineralsremaining in both the abiotic and biotic experiments show�65Cu values less than the starting materials. In contrast, the�65Cu values of the solid material from the experiments withchalcopyrite remained within 2� of the starting material in boththe abiotic and biotic leach experiments.

Copper isotopic values of the aqueous Cu released duringabiotic leaching of both chalcopyrite and chalcocite are heavierthan the starting material and decrease with time (Fig. 4). Incontrast, the fluids from biotic leach experiments with chal-copyrite and chalcocite remain at relatively constant �65Cuvalues throughout the leach cycle. These constant values are, ingeneral, isotopically indistinguishable from the values in start-ing materials.

4.2. Bacterial Uptake Experiment

In the 30-day biotic uptake experiments, the dissolved cop-per in the Cu-rich medium has a measured �65Cu of 2.55 �0.16‰ at the start of the experiment and 2.50 � 0.16‰ at theend of the experiment. The bacteria pellets centrifuged fromthese solutions have significantly heavier isotope values thanthe original copper medium in which they grew (5.59 � 0.16‰average value for flasks 1 and 2; Table 4). The dried pelletcombined from two flasks by centrifugation for this experimentweighed 0.24 g. The calculated concentration of Cu in thepellet is 19 g Cu/kg dry weight bacteria.

TEM images and EDS spectra of bacteria collected from thisexperiment also reveal that T. ferrooxidans cells contain ele-vated concentrations of copper (Fig. 5). In contrast, EDS ele-ment maps of the mineral precipitates and polymeric materialthat formed in the medium during the experiment, when ob-served under the SEM and TEM, did not reveal copper-rich

areas. No diffraction patterns indicating crystallinity were de-

5238 R. Mathur et al.

tected in the particulates associated with cells despite severalattempts to collect them.

4.3. Partial Dissolution Experiments

During the 24-h partial dissolution experiments, �30% ofeach sample dissolved (measured residue minerals weighed0.15 g). The mineral composition as identified by XRD did notchange after partial acid digestion for both experiments. Partialdissolution of the copper minerals (Table 5) in nitric acid didnot produce copper isotopic shifts in the leach fluid in compar-ison to complete dissolution.

5. DISCUSSION

5.1. Mass Balance

To highlight mechanisms, a mass balance for each of thereactions is calculated using

�65Cumino � �65Cuaqfaq � �65Cuminfmin � �65Cubfb (4)

Here, �65Cumino describes the isotopic composition of the start-

ing mineral, �65Cumin describe isotopic composition of residualminerals, the subscript aq refers to Cu in aqueous leach fluid,and subscript b refers to Cu in the bacteria pellet. In addition,fi � ratio of the mass of copper in phase i divided by the totalmass of copper in the starting material (0.53 g Cu for chal-copyrite and 1.09 g Cu for chalcocite).

Equation 4 balances the measured values of faq, fmin,�65Cumin, and �65Cuaq within error for the abiotic leach exper-iments for both chalcocite and chalcopyrite (Table 6). Thisabiotic mass balance suggests sources and sinks for heavy andlight reservoirs where isotopically heavy aqueous Cu is insolution, and isotopically light Cu is precipitated out of solu-tion.

Equation 4 does not balance the measured values of faq, fmin,�65Cumin, and �65Cuaq for the biotic leach experiments withchalcocite (Table 6). To understand how the bacterial interac-tion in the experiment may cause the imbalance, we isolated the

Fig. 2. Plot of �65Cu‰ from high-temperature copper sulfide min-eralization from various porphyry copper deposits. ¦ � chalcopyritesamples and � � bornite samples, and the open square is average�65Cu‰ � 0.09 (solid blank line � 2� � 0.8, dashed lines) of allcopper sulfides. �65Cu presented as per mill with respect to the976NIST copper standard.

bacteria in separate experiments (termed the 30-day bacterial

uptake experiment) because we could not separate cells fromresidual mineral powders. This comparison between the uptakeexperiment and the leach experiment is an imperfect one,because the final concentration of aqueous Cu in the cellularuptake experiment (50 mM) differed from the final Cu concen-tration in the biotic chalcocite (114 mM) and chalcopyrite (14mM) experiments. Nonetheless, the bacteria uptake experimentmost likely best models the biotic chalcocite experiment. Thebacterial uptake experiments indicated that the copper concen-tration is 19g Cu/kg bacteria pellet when T. ferrooxidans isgrown for 30 days in copper-rich growth medium, and thepellet is isotopically heavy (�65Cub � 5.59 � 0.16‰). Ifhypothesized that this mass of copper (0.005 g) and the isotopiccomposition of Cu is also entrained in the bacteria in the 30-daybiotic chalcocite experiments (and an average of the threebiotic chalcocite experiments fb � 0.01) we calculate that�65Cubfb values that are within mass balance for the reportedvalues except for the last chalcocite biotic experiment (Table6). This deviation can be explained if more cellular mass can beassumed in the chalcocite biotic 30-day experiment than mea-sured in the uptake experiments. Therefore, an assumed largermass of bacteria could account for the missing heavy copper.Thus, the results are well explained by a scenario in whichbacterial uptake of Cu in the 30-day mineral leaching experi-ments were similar to the mineral-absent bacterial uptake ex-periment in amount and isotopic composition (�65Cub � 5.59� 0.16‰).

For Eqn. 2 of the biotic chalcopyrite experiment, we inferthat less Cu was taken up into T. ferrooxidans grown withchalcopyrite. This inference is consistent with the lower extentof oxidation in the chalcopyrite compared to chalcocite exper-iments and the significantly lower values of dissolved Cu in thepresence of the chalcopyrite at 30 days (Fig. 3). However,Figure 4 clearly demonstrates that the isotopic value of aqueousCu released during abiotic chalcopyrite oxidation differs fromthe biotic release. We can clarify this observation by the fol-lowing model calculations: If we assume that biotic uptake inthe chalcopyrite experiment can explain why the concentrationof Cu in solution at the end of the biotic experiment is lowerthan Cu concentration at the end of the abiotic experiment, wecalculate that 0.005 g of Cu (average of the three biotic chal-copyrite experiments fb � 0.06) was taken up by Thiobacilluscells. Assuming further that the isotopic composition of thecellularly incorporated Cu was �65Cub � 5.59 (as in the cellularuptake experiment, Table 5), the value of �65Cub fb for thebiotic chalcopyrite experiment is calculated to be 0.3, whichbalances Eqn. 4 within error. This calculation demonstrateswhy the abiotic release of Cu from chalcopyrite results inisotopically heavy Cu in solution, whereas the biotic release ofCu results in isotopically indistinguishable Cu from the startingchalcopyrite in solution.

5.2. Mechanisms for Copper Isotope Fractionation

In this discussion we argue for the simplest mechanism forthe cause of isotope fractionation that is consistent with ourresults, although clearly other mechanisms may be possible.Fractionation of isotopes such as those presented (Table 1)could occur during one or more fractionation steps, followed by

separation of the fractionated Cu reservoirs. Because of the

day ba

5239Cu isotopes in the supergene environment

complexity of our experiments, we consider the followingpossible mechanisms as both fractionating and separating steps:(1) release of copper from sulfide minerals, (2) adsorption orprecipitation of copper on or with iron oxides, (3) precipitationof copper as a secondary sulfide, and (4) adsorption or uptakeof copper on, or by the cells.

Mechanism 1, fractionation during dissolution, is not themost likely mechanism for fractionation of copper because weobserved no copper isotope fractionation during the 24-h partialdissolution experiments.

Mechanism 2 relies upon incorporation or adsorption ofcopper into or onto the iron oxides that precipitate during theexperiment. From a crystallographic standpoint, copper doesnot easily substitute for Fe in the iron oxide crystal structure(Waychunas, 1991). However, adsorption of Cu to the surfaceof iron oxides is a possibility. Many studies have examinedadsorption of metals onto iron oxides (e.g., Parkman et al.,1999; Jang et al., 2003 and references therein). For example,

Fig. 3. Plot of the concentration of iron and copper in th30-day batch experiments with chalcocite, (c) abiotic 30-batch experiments with chalcopyrite.

Christl and Kretzschmar (1999) examined the role of copper

adsorption at the hematite-water interface. They found that atpH �4, copper does not adsorb onto the surface of hematiteregardless of the concentration of copper in the solution (20–100 mM). Given that our experiments were run at pH 2 to 3,their results may suggest little to no copper adsorption onto ironoxides. Consistent with this, data shown in Figure 3 documentthat Cu and Fe behave differently in the experiments, and weobserved that suspended mineral precipitates observed underEDS did not contain significant Cu. Therefore, adsorption ofCu onto iron oxide surfaces most likely does not account for thefractionation of copper isotopes in these experiments.

Mechanism 3, the precipitation of new copper minerals,could explain the variations seen in both the abiotic 30-daybatch experiments with chalcopyrite and chalcocite. The massbalance documented between the fluid and the residue in theabiotic experiments ((Eqn. 4); Table 6) is consistent with thisidea. One possible explanation for the observed fractionation isthat the average copper coordination for Cu in the chalcocite-

iotic 30-day batch experiments with chalcocite, (b) biotictch experiments with chalcopyrite, and (d) biotic 30-day

e (a) ab

to-covellite series of copper sulfide transformations correlates

ols witd as pe

5240 R. Mathur et al.

with the copper isotope variation of acid-sulfate leach solutionsin abiotic leach experiments of chalcocite. They suggest thatthe change in copper bonding in the Cu mineral during acid-sulfate leaching causes an equilibrium copper isotope effect.Because similar phase changes occurred during our batch leachexperiments, crystallographic changes in mineral residues andCu isotopic differences in these minerals could explain ourobservations. Rouxel et al. (2004) also demonstrated that oxi-dized aqueous copper acquires a 3‰ heavier copper isotopicvalue during remobilization of copper in hydrothermal fields inthe Mid-Atlantic Ridge, and they attributed the fractionation tocrystallographic changes in the copper sulfide mineralogy.

Assuming mechanism 3 is correct, the definition of a frac-tionation factor, �aq-min

�aq�min � (�65Cuaq � 1000) ⁄ (�65Cumin � 1000) (5)

Fig. 4. Plot of �65Cu‰ in residual minerals and copperchalcocite, (b) biotic 30-day batch experiments with chalco(d) biotic 30-day batch experiments with chalcopyrite. Thicsulfide material. Symbols with fill indicate fluids and symb2� error (0.16‰) of the starting material. �65Cu presente

The fractionation factor is determined at isotopic equilibrium.

In our experiments, we did not prove isotopic equilibrium.However, to understand our system, we have calculated valuesfor these fractionation factors for every time point in ourexperiment (Table 1); these values may or may not representequilibrium values of �aq-min and are similar to values pre-sented by Ehrlich et al. (2004) and Zhu et al. (2002). Zhu et al.(2002) conducted an abiotic reduction experiment in dilutenitric acid where Cu mineral was precipitated from HNO3-KIsolution containing 0.05 M Cu (NO3)2. The experiment pro-duced isotopically light precipitates containing Cu� (average�65Cumin � �3.4 � 0.01‰) and slightly isotopically heavierfluids (average �65Cu of 0.53 � 0.01‰). They calculatedvalues of �aq-min of �1.004 for their experiments, and theyargued their values represented equilibrium.

For the chalcopyrite experiments, XRD did not reveal thepresence of newly formed sulfide minerals after 30-day, and the

uid vs. time for (a) abiotic 30-day batch experiments with) abiotic 30-day batch experiments with chalcopyrite, andline indicates the copper isotopic signature of the starting

hout fill indicate residual materials. Dashed lines indicater mill with respect to the 976NIST copper standard.

leach flcite, (ck black

�65Cu value of the residual material does not differ from the

5241Cu isotopes in the supergene environment

starting material. However, in these experiments, aqueous cop-per is isotopically heavier than the starting material. We inferthat other copper phases are present in trace quantities in theseexperiments, consistent with the observation that powder dif-fraction cannot detect phases �5%, or that there are amorphousphases not detected by XRD. For example, Hackel et al. (1995)identified, through the use of Auger electron spectroscopy andX-ray photoelectron spectroscopy, the presence of metal-richsulfide minerals on the surface of sulfuric-acid leached chal-copyrite. Todd et al. (2003) further examined chalcopyritesurfaces with Cu and Fe edge spectroscopy and found that inlow pH solutions, the chalcopyrite surface is characterized by alayer that is metal-deficient compared to the bulk stoichiome-try. Given that these experiments are relatively similar in de-sign, small quantities of metal-bearing and isotopically distinctsulfide minerals may have formed on the surface of the chal-copyrite. Therefore, the copper isotope fractionation recordedin the leach fluids from the chalcopyrite could be evidence thatcopper-rich sulfide minerals precipitated on the surface of thechalcopyrite.

The fourth mechanism, cellular uptake or adsorption, is notneeded to explain mass balance within error as exemplified inEqn. 4 for the biotic chalcopyrite experiment. However, thedifference between the isotopic values of aqueous Cu releasedduring biotic and abiotic oxidation of chalcopyrite is bestexplained by cellular uptake of �0.0045 g of Cu with �Cu �5.59 � 0.16‰, as discussed earlier. Furthermore, this mecha-nism explains the results for both the biotic 30-day batchexperiments with chalcocite and the bacterial uptake experi-ment. If the bacteria are not considered in the mass balanceexpression 4 for the biotic 30-day batch experiments withchalcocite, mass balance cannot be achieved. The bacteria aretherefore inferred to be a sink for the heavy copper. Specifi-cally, in the final collection of the biotic batch experiments withchalcocite, there must be at least 0.2 g of bacteria that contains19g Cu/kg bacteria pellet present for mass balance. Our exper-iments are consistent with 0.24 g dry weight for the bacteria inthe chalcocite experiment (Table 5). As shown in Figure 5,each bacterium is observed to be coated by Cu and Fe-contain-ing particles, presumed to be oxides. Interestingly, these Cuoxides must be isotopically heavy (Table 5), in contrast to thenewly precipitated Cu-sulfide residual minerals in abiotic ex-

Table 4. Results from 30-day batch experiments with copper-richmedium.

Sample number Description �65Cu‰*

1 flask 1 � 2** 5.545.65

2 flask 3*** 4.164.31

3 starting medium 2.712.72

4 Cu nugget 2.55

* �65Cu reported in per mill with respect to the 976NIST standard, asdemonstrated in Eqn. 1, with 2� � 0.16‰, during one session.

** Bacteria pellet from combination of two flasks to increase con-centration of copper in the pellet.

*** Bacteria pellet from one flask.

periments (Table 1), which are isotopically light.

Further clarification as to how the bacteria interact withcopper to produce these shifts involves understanding how theorganism uses copper. At least four biologic pathways can beenvisioned in which the copper isotopes could be fractionatedby bacteria. First, Cu uptake into a biofilm could cause therelative enrichment of one isotope. Kinzler et al. (2003) de-scribes the complex processes that occur within biofilms andpoints out that trace metals are often complexed to polymers inbiofilms. Secondly, the uptake of Cu through the cell mem-brane could selectively favor one isotope either due to transportdown concentration gradients or via carriers. Third, the copperfractionation could be related to active sites on enzymes in thebacteria where different coordination sites within these organicmolecules favor one isotope or another. It has recently beendemonstrated, for example, that different oxidation states of Cuare favored by differently structured copper-bearing proteins(Peariso et al., 2003). Conceivably, the selection of Cu� orCu2� by different enzymes could produce the isotopic shifts.Finally, cells can cause precipitation of metals at their outermembrane surface.

Given the high concentration of copper measured in the T.ferrooxidans, fractionation by cellular intake is not reasonablebecause the observed concentration of copper is so high as topresumably be toxic. No metal precipitates were observedwithin polymeric material associated with cells. A more likelyexplanation for Cu uptake by cells is precipitation of Cu min-erals on external membranes as suggested by Figure 5. Perhapscopper precipitation on membranes may represent a process bywhich the organism protects itself against high concentrationsof potentially toxic metals. The inference that mineral precip-itates around cell membranes are oxides rather than sulfidesmay further suggest that Cu entrained in secondary oxideminerals is isotopically heavy as compared to Cu in secondarysulfide minerals. In these experiments, precipitation of isotopi-cally heavy Cu-Fe amorphous nanoparticulates are an impor-tant vital effect that contributes to the isotopic value of Cureleased to solution during sulfide oxidation.

5.3. Supergene Environments

During supergene leaching, Cu is leached from chalcopyritein the oxidizing vadose zone and becomes concentrated asprecipitated chalcocite and covellite in a reducing environmentbelow the water table. Cu from the chalcocite/covellite can inturn be leached (as uplift occurs and exposes the inner portionsof the system) and precipitated again as a new, lower, enrich-ment blanket as chalcocite and covellite. These experiments arethought to roughly mimic conditions found in natural super-gene environments. As demonstrated by Bladh (1982), fluidsexisting in ambient atmospheric conditions provide enough O2

(fO2 � 10-2 atm.) to generate dilute sulfuric acid of low enoughpH to dissolve and mobilize metals in sulfide minerals.

As originally suggested by Shields et al. (1965) and con-firmed by the results presented here, the precipitated copperminerals in the enrichment blanket should acquire a heavycopper isotopic value during multiple leach events with orwithout bacteria (Fig. 4). In each leaching event, chalcocite willdissolve oxidatively to release isotopically heavy copper fromthe solid reservoir into solution. If bacteria are present, uptake

of Cu as nano-oxides will drive aqueous Cu back toward

ims mo

5242 R. Mathur et al.

starting material values. However, with or without bacteria, thechalcocite transforms into isotopically lighter covellite duringleaching. The aqueous copper is either lost from the system orprecipitates in the enrichment blanket as copper supergene

Fig. 5. TEM image and EDS spectra of transition metarims on the outer cell walls of the bacteria. These black r

minerals. As multiple supergene leaching events transpire, su-

pergene minerals become progressively enriched in the heavierisotope as compared to the originally emplaced hypogene cop-per (�65Cu � 0‰, Fig. 2).

A cross-sectional model of the copper isotope composition

ed in cell pellets of T. ferrooxidans. Note the thin blackst likely indicate amorphous copper-iron oxide minerals.

ls imag

of supergene minerals within an enrichment blanket is not

5243Cu isotopes in the supergene environment

possible to construct given the limited nature of these experi-ments. The copper isotopic composition of copper mineralsthroughout the enrichment blanket is most likely variable giventhe differences seen in the digenite and covellite residues fromthe leach experiments. However, the copper isotopic composi-tion of a processed copper nugget from Morenci (�65Cu � 2.5)has the same copper isotope value as the chalcocite sampledfrom the enrichment blanket. This most likely indicates theaverage copper isotope signature from all of the mined super-gene ore at Morenci is relatively similar to the copper isotopesignature of chalcocite sample used for the leach experiments.Therefore, single analyses of chalcocite may provide someinsight into the overall leaching process that has occurred in thegeological past.

To quantify this relationship, we constructed a Rayleighdistillation model using the calculated fractionation factorsfrom the abiotic and biotic leach experiments to predict how thecopper isotopic composition of supergene minerals wouldchange during sequential leaching and reworking of copper inthe Morenci and Silver Bell supergene environments (Fig. 6).

Table 5. Results from 24-hour partial dissolution experiments withcopper sulfide minerals.

Sample name Phase Amount dissolved �65Cu‰*

Ertzberg Chalcopyrite complete dissolution 0.17Etzberg Leach Chalcopyrite partial dissolution 0.21Collahuasi Bornite complete dissolution 0.64Collahuasi Leach Bornite partial dissolution 0.62

* �65Cu reported in per mil with respect to the 976NIST standard, asdemonstrated in Eqn. 1, with 2� � 0.16‰.

Table 6. Mass balance for co

Sample Leach day �65Cuaqfaq

Chalcocite

Abiotic

7 1.2914 1.6530 1.65

Biotic

7 0.6414 1.4230 1.59

Chalcopyrite

Abiotic

7 0.2214 0.3030 0.36

Biotic

7 0.0814 0.1330 0.14

* �65Cu reported in per mill with respect to the 976 NIST standard.

The following equations describe the change in isotopicvalue of Cu in the mineral and aqueous species during leachingof chalcopyrite as a function of extent of reaction (Faure andMensing, 2005). The chalcopyrite is assumed to be of knownisotopic value at time zero, �65Cumin

o :

�65Cumin � (�65Cumino � 103)fmin

��1 � 103 (6)

�65Cuaq � (�65Cumino � 103)�fmin

�-1 � 103 (7)

The physical model is based on average � values derived fromthe abiotic and biotic leach experiments (Table 1) and thefollowing assumptions:

1. The hypogene chalcopyrite is emplaced with a value of�65Cumin

o � 0‰.2. The only source of copper is chalcopyrite. No further copper

is added to the system, which is open only to oxygen, carbondioxide, and water.

3. The fractionation factor for leaching chalcopyrite,�aq-chalcopyrite, is 1.0012 for the abiotic model and 1.0004 forthe biotic model. These values represent values averaged forall measurements for each chalcopyrite experiment (Table1). Whereas in the biotic case the value is determined fromdifferences well within the margin of error of measurement,we include that calculation here simply to exemplify ourbest estimate of how such fractionation factors would allowboth aqueous and mineral Cu to evolve over time for thebiologic system.

4. During the first leaching event, 65% of the copper is leachedand 35% remains in minerals in the leach cap. These ap-proximate proportions were determined by examining howcopper is currently distributed in Morenci. At Morenci,

ts involved in leach cycle*.

�65Cuminfmin �65Cubfb �65Cumino

1.44 0 2.741.28 0 2.931.06 0 2.72

1.84 0.2 2.671.27 0.2 2.880.30 0.2 2.09

0.33 0 0.540.23 0 0.530.45 0 0.81

0.62 0.2 0.700.29 0.2 0.610.24 0.2 0.58

mponen

5244 R. Mathur et al.

several enrichment blankets are mined; the average enrich-ment blanket is 130 m thick and contains 0.47 wt% Cu. Theleach cap is 82 m thick and contains 0.1 wt% Cu (Enders,2000). We assume that, after the first leach event, �65% ofthe total copper within the deposit is now present in theenrichment blanket. Of course, as pointed out by Enders(2000), lateral transport of copper into exotic peripherydeposits may also have occurred; therefore, the system maynot have been closed with regard to total movement ofcopper. Nonetheless, our model assumes a closed system.

5. Of the 65% of the total copper that is removed from theleach cap, we assume all of the copper precipitates aschalcocite in the enrichment blanket. The assumed value of�aq-chalcocite is 1.003 for the abiotic model and 1.0009 for thebiotic model (Table 1).

6. During each of the subsequent leach events, all of the copperis derived from chalcocite in the last-formed blanket. Foreach leach event, 65% of the copper in the blanket isremoved and is precipitated quantitatively in the nextblanket.

To test these models, we compared the predicted copperisotopic composition of chalcocite derived from the model tothe copper isotopic composition of one sample of chalcocitetaken from one enrichment blanket from the Morenci andSilver Bell porphyry copper deposits, respectively. Results offield observations and study of enrichment blankets of depositsin the American Southwest reveal a complex process of uplifthistory and related lowering of the water table (Titley, 1983).These coupled events have resulted in the formation, destruc-tion, and reprecipitation of chalcocite enrichment blankets overtime.

The copper isotopic compositions of the two chalcocites areplotted on Figure 6 along with the predicted changes of thecopper isotopic ratios from the Rayleigh fractionation models.The overlap of the modeled copper isotopic composition ofchalcocite with the copper isotopic composition of the chalcoc-ite from Morenci suggests that multiple leaching events were

Fig. 6. Predicted leach events using Rayleigh distillation models.Solid lines indicate copper isotope composition of the chalcocites fromMorenci and Silver Bell porphyry copper deposits in Arizona. Eachfilled symbol on the graph represents the predicted �65Cu value ofchalcocite after one leaching/precipitation event as supergene enrich-ment progresses (see text). Circles indicate abiotic processes andsquares indicate biotic processes. �65Cu presented as per mill withrespect to the 976NIST copper standard.

needed to explain the isotopic systematics. If the system was

abiotic, 10 events are consistent with our data, whereas 27leaching events would have been needed if Thiobacillus hadbeen important in the supergene system. Strong geologic evi-dence based on mineral textures and radiometric dating areconsistent with multiple supergene leaching events at this site(Titley and Marozas, 1995; Enders, 2000); however, theseauthors could only suggest that the number of leach periodswas greater than two or three.

The heavy copper isotopic composition found in both en-richment blanket chalcocites and these calculated high numberof leach events could be consistent with enrichment occurringincrementally rather than as discrete enrichment/leachingevents separated in time. In other words, the water table mayhave slowly dropped and the copper isotope composition of thechalcocites may now reflect the slow evolution of the migrationof enrichment to its current level. There is no geologic evidenceto support this idea, in fact Cook (1994) dated alunites fromsupergene blankets across the southwestern United States andinterpreted the scatter of alunite ages to represent several epi-sodes of enrichment throughout the geologic past. This evi-dence indicates that there are distinct intervals of enrichmentand that the enrichment process does not appear to be a con-tinual process once erosion is initiated.

Field studies of oxidized sulfide ores and outcrop features atSilver Bell (Lopez and Titley, 1995) have revealed the presenceof at least two former dated enrichment blankets above thepresent enriched Silver Bell ores. Those superjacent blanketshave lost most of their copper through oxidation, leaching, andmovement to the new level in the North Silver Bell deposit. Thesample of chalcocite from Silver Bell (�65Cu � 6.54‰) con-tains copper derived by oxidation and leaching of Cu from themultiple enrichment blankets above it. Thus, although ourmodel is too simplistic, it provides a basis for future modeling,and it emphasizes that the isotopic values of supergene miner-als are consistent with many stages of leaching and enrichment.

An essential prediction from this model is that the “lighter”copper isotope values must exist somewhere in the copperdeposit. There are two possible reservoirs for this light copper:(1) The light copper could be in the leach cap. Currently theleach cap is �82 m thick and contains �0.1% copper (Enders,2000) and could be used to make the mass balance. However,there has been significant erosion in the deposit (upwards of 1Km; Enders, 2000), and it would difficult to place an exactconstraint on the amount of missing copper. (2) The coppercould have moved laterally and essentially “leaked” out of thesystem and formed what is termed “exotic” copper deposits.One or a combination of both these sources could provide anexplanation for where copper minerals with light copper iso-tope values exist.

Future models must incorporate more complexities. For ex-ample, during the sequential leaching events, copper fromprimary chalcopyrite in the ore deposit is probably added to thesystem. In contrast, in our simplified model, 35% of the originalchalcopyrite was left behind, unleached. If additional hypogenechalcopyrite were considered in the model, this source wouldsignificantly change the overall copper isotopic composition ofthe fluid by lowering its copper isotopic composition. There-fore, the values modeled here represent minimum copper iso-tope values per leach cycle.

In addition, copper-rich phases other than chalcopyrite exist

5245Cu isotopes in the supergene environment

in hypogene ores. It is possible that minerals such as bornite,primary chalcocite, and others could yield different fraction-ation factors. Furthermore, copper not only precipitates ascopper sulfide phases in the supergene system, but also precip-itates as sulfates and oxides (Titley, 1978). Copper leachedfrom primary chalcopyrite is also observed to precipitate aschalcocite and other phases, even in the leach cap. Thesecopper precipitates are not accounted for in our model butcould be incorporated in future models.

Despite the simplistic nature of our model, our calculationsconfirm that multiple leaching and precipitation events musthave occurred at both Morenci and Silver Bell deposits. IfThiobacillus was involved in the leaching, more leach eventsare warranted to explain the data. Our experiments furtherdocument the possibility of a vital effect during oxidativedissolution of Cu sulfides; Thiobacillus promotes the formationof isotopically heavy Cu nano-oxide particles that may accu-mulate in the leach caps. If this is true, it may be possible tofind such particulate material (e.g., Sillitoe et al., 1996) andisotopic values of biotic and abiotic reactions during leaching.Further research into the systematics of Cu isotopes shouldrefine such models further.

6. CONCLUSIONS

The experiments conducted in this study demonstrate thatduring the oxidation of copper via abiotic leaching, Cu releasedto solution is isotopically heavy because Cu mineral precipi-tates entrain isotopically light Cu. Interestingly, Thiobacilluscells promote the formation of isotopically heavy Cu oxidenanoparticles during such leaching. This is the first study todocument the complexity of copper isotopic shifts during abi-otic and biotic supergene processes. The results suggest thatsupergene minerals should possess heavier copper isotope val-ues than their parent material. Further experiments, mineralog-ical observations, and modeling using isotopes of Cu and otherelements will undoubtedly provide insights into the supergeneprocess.

Acknowledgments—We would like to thank Art Rose, Steve Young,Bryn Kimball, Henry Ehrlich, and Richard Thompson for their insight-ful comments and discussion concerning this project. We would alsolike to thank Maria Orosz and Lawrence Mutti for their help with thediffraction of minerals and Ed Ripley and two anonymous reviewersfor their helpful comments. We would finally like to thank PhelpsDodge for providing the chalcocite samples and John Uhrie for assis-tance in completing this project.

Associate editor: E. M. Ripley

REFERENCES

Alpers C. and Brimhall G (1989) Paleohydrologic evolution and geo-chemical dynamics of cumulative supergene metal enrichment atLa Escondida, Atacama Desert, Northern Chile. Econ. Geol. 84,229–257.

Amend J. P. and Shock E. L. (2001) Energetics of overall metabolicreactions in thermophilic and hyperthermophilic Archaea and Bac-teria. FEMS Microbiology Rev. 25, 175–243.

Anbar A. (2004) Molybdenum stable isotopes: Observation, interpre-tation, and directions. In Geochemistry of NonTraditional Stable

Isotopes, Reviews in Mineralogy (ed. C. M. Johnson et al.), Vol. 55,pp. 429–450, Mineralogical Society of America.

Anbar A., Roe J. Barling J., and Nealson K. (2000) Nonbiologicalfractionation of iron isotopes. Science 288, 126–128.

Beard B., Johnson C., Skulan J., Nealson K., Cox L., and Sun H. (2003)Application of Fe isotopes to tracing the geochemical and biolog-ical cycling of Fe. Chem. Geol. 195, 87–117.

Bladh K. W. (1982) The formation of goethite, jarosite and aluniteduring the weathering of sulfide-bearing felsic rocks. Econ. Geol.77, 176–184.

Boyer A., Magnin J-P., and Ozil P. (1998) Copper ion removal byThiobacillus ferrooxidans biomass. Biotechnology Letters 20, 187–190.

Brantley S., Liermann L., and Bullen T. (2001) Fractionation of Feisotopes by soil microbes and organic acids. Geology 29, 535–538.

Breed A. W. and Hansford G. S. (1999) Studies on the mechanism andkinetics of bioleaching. Miner. Eng. 12, 383–392.

Brelle M. C., Torres-Martinez C. Z., McNulty J. C., Mehra R. K., andZhang J. Z. (2000) Synthesis and characteristics of CuxS nanopar-ticles and nature of infrared band and charge carrier dynamics. PureAppl. Chem. 72, 101–117.

Brimhall G. H. (1980) Deep hypogene oxidation of porphyry copperpotassium-silicate protore at Butte, Montana; a theoretical evalua-tion of the copper remobilization hypothesis. Econ. Geol. 75, 384–409.

Brimhall G. H., Alpers C. N., and Cunningham A. B. (1985) Analysisof supergene ore-forming processes and ground-water solute trans-port using mass balance principles. Econ. Geol. 80, 1227–1256.

Bullen T., White A., Child C., Vivit D., and Schulz M. (2001) Dem-onstration of significant abiotic iron isotope fractionation in nature.Geology 29, 699–702.

Christl I. and Kretzschmar R. (1999) Competitive sorption of copperand lead at the oxide-water interface; implications for surface sitedensity. Geochim. Cosmochim. Acta 63, 2929–2938.

Colmer A. R. and Hinkle M. E. (1947) The role of microorganisms inacid mine drainage: A preliminary report. Science 106, 253–256.

Cook S. (1994) The Geologic History of Supergene Enrichment in thePorphyry Copper Deposits in Southwestern North America. Ph.D.Thesis, University of Arizona.

Crundell F. (2003) How do bacteria interact with minerals. Hydromet-allurgy 71, 75–81.

Dold B. (2003) Enrichment processing in oxidizing sulfide mine tail-ings: Lessons for supergene ore formation. SGA News 16, 7–15.

Dold B. and Fonbonte L. (2001) Element cycling and secondarymineralogy in porphyry copper tailings as a function of climate,primary mineralogy and mineral processing. J. Geochem. Explor.74, 3–55.

Dutrizac J., Mac Donald R., and Ingraham T. (1969) The kinetics ofdissolution of synthetic chalcopyrite in aqueous acidic sulfate so-lutions. Trans. Metall. Soc. AIME 245, 955–959.

Ehrlich S., Butler I., Halicz Richard D., Oldroyd A., and Matthews A.(2004) Experimental study of the copper fractionation betweenaqueous (CuII) and colellite (CuS). Chem. Geol. 209, 259–269.

Ellis A. S., Johnson T. M., and Bullen T. D. (2002) Cr isotopes and thefate of hexavalent Cr in the environment. Science 265, 2026–2060.

Enders S. (2000) The Evolution of Supergene Enrichment in theMorenci Porphyry Copper Deposit, Greenlee County, Arizona.Ph.D. Thesis, University of Arizona.

Faure G. and Mensing T. M. (2005) Isotopes: Principles and Applica-tions, 3rd ed Wiley.

Gale N., Woodhead A., Stos-Gale Z., Walder A., and Bowen I. (1999)Natural variations detected in the isotopic composition of copper:possible applications to archaeology and geochemistry. Int. J. MassSpectrom. 184, 1–9.

Gericke M. and Pinches A. (1999) Bioleaching of copper sulfideconcentrate using extreme thermophilic bacteria. Miner. Eng. 12,893–904.

Goble R. J. (1985) The relationship between crystal structure, bondingand cell dimensions in the copper sulfides. Canadian Mineralogist23, 51–76.

Graham S., Pearson N., Jackson S., Griffin W., and Reilly S. (2004)Tracing Cu and Fe from source to porphyry: in situ determinationof Cu and Fe isotope ratios in sulfide from the Grasberg Cu-Au

deposit. Chem. Geol. 207, 147–169.

5246 R. Mathur et al.

Guilbert J. M. and Park C. F. (1986) The Geology of Ore Deposits.W. H. Freeman.

Hackel R., Dreisinger, D., Peters E., and King J. (1995) Passivation ofchalcopyrite during the oxidative leaching in sulfate media. Hydro-metallurgy 39, 25–48.

Halliday A., Lee D., Christensen J., Rehkamper M., Li W., Luo X.,Hall C., Ballenstine C., Pettle T., and Stirling C. (1998) Applica-tions of multi collector ICPMS to cosmochemistry, geochemistryand paleoceanography. Geochim. Cosmochim. Acta 62, 919–940.

Hansford G. S. and Vargas J. (2001) Chemical and electrochemicalbasis of bioleaching processes. Hydrometallurgy 59, 135–145.

Hong L., Aiyun Z., and Shen A. (2000) The main microorganism groupin supergene zones of the Mingshan gold deposit, northwesternGuangxi and its metallogenic significance. Diqiu Huaxue �Geochimica 29, 50–55.

Jang J. H., Dempsey B. A., Catchen G. C., and Burgos W. D. (2003)Effects of Zn (II), Cu(II), Mn (II), Fe(II), NO3 and SO4 at pH 6.5and 8.5 on transformations of hydrous ferric oxide as evidenced byMössbauer spectroscopy Colloids Surf., A 221, 55–68.

Johnson C., Beard B., Albaréde F., Eds. (2003) Reviews in Mineralogyand Geochemistry, Geochemistry of Non-Traditional Stable Iso-topes, Vol. 55. Mineralogical Society of America, GeochemicalSociety.

Kelly D. P. and Wood A. P. (2000) Reclassification of some species ofThiobacillus to the newly designated genera Acidithiobacillus gen.nov., Halothiobacillus gen. nov. and hermithiobacillus gen. nov Int.J. Syst. Evol. Microbiol. 50, 511–516.

Kinzler K., Gehrke T., Telegdi J., and Sand W. (2003) Bioleaching—aresult of interfacial processes caused by extracellular polymericsubstances (EPS). Hydrometallurgy 71, 83–88.

Larson P., Maher K., Ramos F., Chang Z., Gaspar M., and Meinert L.(2003) Copper isotope ratios in magmatic and hydrothermal ore-forming environments. Chem. Geol. 201, 337–350.

Leduc L. G., Ferroni G. D., and Trevors J. T. (1997) Resistance toheavy metals in different strains of Thiobacillus ferrooxidans.World J. Microbiol. Biotechnol. 13, 453–455.

Linge H. (1976) A study of chalcopyrite dissolution in acid ferricnitrate by potentiometric titration. Hydrometallurgy 2, 51–64.

Lizama H. (2001) Copper bioleaching behavior in an aerated heap. Int.J. Mineral Processing 62, 257–269.

Lopez J. A. and Titley S. (1995) Outcrop and capping characteristics ofthe supergene sulfide enrichment at North Silver Bell, PimaCounty, Arizona. In Porphyry Copper Deposits of the AmericanCordillera, Arizona Geological Society, Digest 20 (eds. R. H.Sillitoe, J. Perello, and C. E. Vidal), pp. 424–435.

Maksaev V., Munizaga F., McWilliams M., Mathur R., Ruiz J. andFanning M. (2004) Superimposed mineralization produced super-giant Cu-Mo porphyry deposits. In SEG Special Publication 11:Andean Metallogeny: New Discoveries, Concepts and Updates (ed.R. H. Sillitoe et al.), pp 15–54. Society of Economic Geologists.

Maréchal C. and Albaréde F. (2002) Ion exchange fractionation ofcopper and zinc isotopes. Geochim. Cosmochim. Acta 66, 1499–1509.

Maréchal C., Telouk P., and Albarede F. (1999) Precise analysis ofcopper and zinc isotopic compositions by plasma-source spectrom-etry. Chem. Geol. 156, 251–273.

Masterman G. J., Cook D. R., Berry R. F., Clark A. H., ArchibaldD. A., Mathur R., Walshe J. L., and Durán M. (2004) 40Ar-39Ar andRe-Os geochronology of porphyry copper-molybdenum depositsand related copper-silver veins in the Collahuasi District, NorthernChile. Econ. Geol. 99, 673–690.

Matthews A., Zhu X., and O’Nions K. (2001) Kinetic iron stableisotope fractionation between iron (-II) and (-III) complexes insolution. Earth Planet. Sci. Lett. 192, 81–92.

Mathur R., Ruiz J., Titley, S., Gibbins S., and Margomoto W. (2000)

Different crustal sources for Au-rich and Au-poor ores of the

Grasberg Cu-Au porphyry deposit. Earth Planet. Sci. Lett. 183,7–14.

Nielsen A. and Beck J. (1972) Chalcocite oxidation and coupled carbondioxide fixation by Thiobacillus ferroxidans. Science 175, 1124–1126.

Parkman R. H., Charnock J. M., Byran N. D., Livens F. R., andVaughan D. J. (1999) Reactions of copper and cadmium ions inaqueous solution with goethite, lepidocrocite, mackinawite andpyrite. American Mineral. 84, 407–419.

Peariso K., Huffman D., Penner-Hanh J. E., and O’Halloran T. V.(2003) The Pc°C copper resistance protein coordinates Cu (I) vianovel S-methionine interactions. J. Am. Chem. Soc. 125, 342–343.

Rouxel O., Fouquet Y., and Ludden, J. (2004) Copper isotope system-atics of the Lucky Strike, Rainbow and Logatev seafloor hydro-thermal fields on the Mid Atlantic Ridge. Econ. Geol. 99, 585–600.

Shields W. R., Goldich S. S., and Garner E. L. (1965) Natural varia-tions in the abundance ratio and the atomic weight of copper. J.Geophys. Res. 70, 479–491.

Sillitoe R., Folk R., and Saric N. (1996) Bacteria as mediators of coppersulfide enrichment during weathering. Science 272, 1153–1155.

Singer P. C. and Stumm W. (1968) Kinetics of the oxidation of ferrousiron. Second Symposium on Coal Mine Research. Mellon Institute.

Skewes M. A. and Stern C. R. (1994) Tectonic trigger for the formationof late Miocene Cu-rich breccia pipes in the Andes of Central Chile.Geology 22, 551–554.

Stokes N. (1970) Experiments on the action of various solutions onpyrite and marcasite. Econ. Geol. 2, 14–23.

Stott M., Wating H., Franzmann P., and Sutton D. (2000) The role ofiron hydroxyl precipitates in the passivation of chalcopyrite duringbioleaching. Miner. Eng. 13, 1117–1127.

Titley S. (1978) Geologic history, hypogene features and processes ofsecondary sulfide enrichment at Plesyuni copper prospect, NewBritain, Papua New Guinea. Econ. Geol. 73, 768–784.

Titley S. (1983) Advances in Geology of the Porphyry Copper Depos-its. University of Arizona Press.

Titley S. (1995) The style and progress of mineralization and alterationin porphyry copper systems. In Advances in Geology of the Por-phyry Copper Deposits, Southwestern North America (ed. S. Tit-ley), pp. 93–115. University of Arizona Press.

Titley S. and Marozas D. (1995) Processes and products of supergenecopper enrichment. In Advances in Geology of the Porphyry Cop-per Deposits, Southwestern North America (ed. S. Titley), pp.156–168, University of Arizona Press.

Todd E. C., Sherman D. M., and Puton J. A. (2003) Surface oxidationof chalcopyrite under ambient atmospheric and aqueous (pH2–10)conditions, Cu, Fe, L and O K edge X-ray spectroscopy. Geochim.Cosmochim. Acta 67, 2137–2146.

Waychunas G. A. (1991) Crystal chemistry of oxides and oxyhydrox-ides. In Oxide Minerals; Petrologic and Magnetic Significance (ed.D. H. Lindsley), Reviews in Mineralogy, Vol. 25, pp. 11–68.Mineralogical Society of America.

Whiteside L. S. and Goble R. J. (1986) Structural and compositionalchanges in copper sulfides during leaching and dissolution. Cana-dian Mineralogist 24, 247–258.

Wolin E. A., Wolin M. J., and Wolfe R. S. (1963) Formation ofmethane by bacterial extracts. J. Biol. Chem. 238, 2882–2886.

Zhu X., O’Nions K., Guo Y., Belshaw N., and Rickard D. (2000)Determination of natural Cu-isotope variation by plasma-sourcemass spectrometry: implications for use in geochemical tracers.Chem. Geol. 163, 139–149.

Zhu X., Gou Y., Williams R. J. P., O’Nions R. K., Matthews A.,Belshaw N. S., Canters G. W., deWaal E. C., Weser U., BurgessB. K., and Salvato B. (2002) Mass fractionation processes of

transition metal isotopes. Earth Planet. Sci. Lett. 200, 47–62.