Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

Geochimica et Cosmochimica Acta 97 (2012) 247–265

Iron, zinc, magnesium and uranium isotopic fractionationduring continental crust differentiation: The tale

from migmatites, granitoids, and pegmatites

Myriam Telus a,⇑, Nicolas Dauphas a, Frederic Moynier b, Franc�ois L.H. Tissot a,Fang-Zhen Teng c, Peter I. Nabelek d, Paul R. Craddock a, Lee A. Groat e

a Origins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, Chicago, IL 60637, USAb Department of Earth and Planetary Sciences and McDonnell Center for the Space Sciences, Washington University in St. Louis, St. Louis,

MO 63130, USAc Department of Geosciences & Arkansas Center for Space and Planetary Sciences, University of Arkansas, Fayetteville, AR 72701, USA

d Department of Geological Sciences, University of Missouri-Columbia, Columbia, MO 65211, USAe Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, BC, Canada V6T 1Z4

Received 27 September 2011; accepted in revised form 14 August 2012; Available online 25 August 2012

Abstract

The causes of some stable isotopic variations in felsic rocks are not well understood. In particular, the origin of the heavyFe isotopic compositions (i.e., high d56Fe values, deviation in & of the 56Fe/54Fe ratio relative to IRMM-014) of granites withSiO2 > 70 wt.% compared with less silicic rocks is still debated. It has been interpreted to reflect isotopic fractionation duringlate stage aqueous fluid exsolution, magma differentiation, partial melting, or Soret (thermal) diffusion. The present studyaddresses this issue by comparing the Fe isotopic compositions of a large range of differentiated crustal rocks (whole rocksof migmatites, granitoids, and pegmatites; mineral separates) with the isotopic compositions of Zn, Mg and U. The samplesinclude granites, migmatites and pegmatites from the Black Hills, South Dakota (USA), as well as I-, S-, and A-type grani-toids from Lachlan Fold Belt (Australia). The nature of the protolith (i.e., I- or S-type) does not influence the Fe isotopiccomposition of granitoids. Leucosomes (partial melts in migmatites) tend to have higher d56Fe values than melanosomes (meltresidues) indicating that partial melting of continental crust material can possibly fractionate Fe isotopes. No clear positivecorrelation is found between the isotopic compositions of Mg, U and Fe, which rules out the process of Soret diffusion in thesystems studied here. Zinc isotopes were measured to trace fluid exsolution because Zn can easily be mobilized by aqueousfluids as chloride complexes. Pegmatites and some granitic rocks with high d56Fe values also have high d66Zn values. In addi-tion, high-SiO2 granites show a large dispersion in the Zn/Fe ratio that cannot easily be explained by magma differentiationalone. These results suggest that fluid exsolution is responsible for some of the Fe isotopic fractionation documented in felsicrocks and in particular in pegmatites. However, some granites with high d56Fe values have unfractionated d66Zn values andwere presumably poor in fluids (e.g., A-type). For these samples, iron isotopic fractionation during magma differentiation is aviable interpretation. Equilibrium Fe isotopic fractionation factors between silicic melts and minerals remain to be character-ized to quantitatively assess the role of fractional crystallization on iron isotopes in granitoids.� 2012 Elsevier Ltd. All rights reserved.

0016-7037/$ - see front matter � 2012 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.gca.2012.08.024

⇑ Corresponding author. Present Address: Hawai‘i Institute ofGeophysics and Planetology, School of Ocean and Earth Scienceand Technology, University of Hawai‘i at Manoa, Honolulu, HI96822, USA. Tel.: +1 239 247 0122; fax: +1 808 956 6322.

E-mail address: telus@higp.hawaii.edu (M. Telus).

1. INTRODUCTION

With its complex redox chemistry, iron is a key elementin biological and geochemical processes and its isotopicvariations can be used to probe these interactions. Initial

248 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

studies found all terrestrial igneous rocks to have homoge-neous Fe isotopic composition (Beard and Johnson, 1999,2004; Beard et al., 2003), suggesting that Fe isotopes couldbe discriminate indicators of biological activity and lowtemperature aqueous processes, which can significantlyfractionate Fe isotopes (Johnson et al., 2004; Dauphasand Rouxel, 2006; Anbar and Rouxel, 2007). However,growing evidence suggests that igneous processes such aspartial melting, magmatic differentiation, and low tempera-ture aqueous fluid exsolution can also impart measurableiron isotopic variations among bulk igneous rocks (Wil-liams et al., 2004, 2005, 2009; Poitrasson and Freydier,2005; Weyer et al., 2005; Dauphas and Rouxel, 2006; Poi-trasson, 2006; Schoenberg and von Blanckenburg, 2006;Schuessler et al., 2007, 2009; Weyer and Ionov, 2007; Hei-mann et al., 2008; Teng et al., 2008; Dauphas et al., 2009,2010). This opens the prospect of using iron isotopes as pet-rogenetic tracers of mantle and crustal differentiation.

Poitrasson and Freydier (2005) observed that severalgranites were isotopically heavier than the average d56Fecomposition of igneous rocks and identified a positive cor-relation between d56Fe and SiO2 wt.%. They suggested thatfluids exsolved from the magma were responsible forremoving isotopically light Fe from the crystallizing melt,resulting in a heavy granitic magma body. Heimann et al.(2008) found that magnetite from evolved granitic rockshad high d56Fe values while Fe-silicates had d56Fe valuesclose to the average d56Fe composition of igneous rocks,which they attributed to isotopic exchange between magne-tite and fluid rich in ferrous chloride. Thus, these authorsalso concluded that fluid exsolution was responsible forhigh d56Fe values in evolved granites and rhyolites. Schu-essler et al. (2009) found similar iron isotopic trends forFe isotopes, but did not detect measurable variation inthe isotopic composition of Li, which is easily mobilizedby fluids. Based on the absence of a correlation betweenLi and Fe isotopic compositions, they ascribed these trendsto fractional crystallization of a Fe-bearing phase that pref-erentially incorporated light Fe isotopes; thus, continuallyenriching the residual melt in the heavier isotopes of iron.Similar results were found for a basaltic system at KilaueaIki lava lake (Teng et al., 2008). In the latter study, the lightFe isotopic composition reflected diffusion-driven kineticisotope fractionation during Fe–Mg exchange in olivine(Teng et al., 2011; Sio et al., 2011).

The present study addresses the range of hypotheses putforth to explain these iron isotope variations by combiningFe, Zn, Mg and U isotopic data of migmatites, granitoids,and pegmatites to develop a more complete view of Fe iso-topic fractionation during the formation and differentiationof felsic rocks. Particular attention is directed towards thefollowing questions:

Does partial melting fractionate iron isotopes? This hasbeen documented in mantle peridotites and arc lavasthrough correlations between d56Fe values and proxies ofdegree of partial melting (Weyer and Ionov, 2007; Dauphaset al., 2009). It is not known however whether iron isotopesare fractionated in the process of generating felsic melts.Migmatites may provide the opportunity to directly answerthis question, as they represent highly metamorphosed

rocks arrested in the process of generating granitic melts(Brown, 1994).

Does the nature of the protolith influence the iron isoto-pic composition of felsic rocks? It was suggested that duringpartial melting of the mantle, the iron isotopic compositionof magmas was related to the degree of partial melting, theFe3+/Fe2+ ratio, and oxygen buffering capacity (Dauphaset al., 2009). Is that true for felsic rocks? Do granites thatwere derived from sources with elevated Fe3+/Fe2+ ratioshave higher d56Fe? To address this question, a comparisonis made between S- and I-type granitic rocks (Chappell andWhite, 2001) that are derived from sedimentary and igne-ous sources, which differ in their mineral assemblages andFe3+/Fe2+ ratios (Flood and Shaw, 1975; Whalen andChappell, 1988).

Is magmatic differentiation (fractional crystallizationand restite unmixing), fluid exsolution or Soret (thermal)diffusion responsible for the correlation between iron isoto-pic composition and silica content? Magmatic differentia-tion should be supported by fractional crystallizationmodels (Schoenberg and von Blanckenburg, 2006). Thefluid exsolution hypothesis (Poitrasson and Freydier,2005; Heimann et al., 2008) requires that significantamounts of isotopically light iron be removed from themagma body. Pegmatites, which represent the latest stagesof granitic magma evolution, should be particularly sensi-tive to such processes (Walker et al., 1989). Another meansof testing the fluid exsolution hypothesis is to study the iso-topic compositions of elements, such as Zn, that are mobilein such fluids (Nakano and Urabe, 1989; Shinohara, 1994),and to search for correlations with iron isotopic variations.This is tested by comparing the Zn and Fe isotopic compo-sitions of felsic rocks.

Finally, an alternative model to the standard view ofgranite formation was presented by Lundstrom (2009),who suggested that some granites form by a top-down ther-mal migration process. The model predicts that non-tradi-tional stable isotope systems such as Fe, Mg and Ushould show signatures of thermal diffusion. To addressthis hypothesis, the isotopic compositions of Mg, U andFe were measured in the same granitic rocks. Magnesium,U and Fe isotopes show strong and correlated mass depen-dent fractionation during experiments of thermal diffusionin silicate melts (Richter et al., 2008, 2009; Huang et al.,2010; Lacks et al., 2012), thus providing a straightforwardmeans of identifying Soret effect in geological processes(Dauphas et al., 2010).

2. SAMPLE SELECTION

2.1. I-, S- & A- type granites: bulk rocks and mineral

separates

To determine whether the Fe isotopic compositions ofgranites depend on their source material, granitic rocksfrom the Silurian-Devonian Lachlan Fold Belt (LFB,Southeastern Australia) and Proterozoic Harney PeakGranite (Black Hills, South Dakota) were analyzed for theirFe isotopic compositions. These samples have been wellcharacterized (Hine et al., 1978; Chappell, 1984; Nabelek

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 249

et al., 1992; Chappell and White, 1993; Sha and Chappell,1999) and are divided into three groups according to theirsource rock (Chappell and White, 1974; Loiselle andWones, 1979). I-type granites (nine samples; I1, I2, I3, I4,I5, I6, I7, I8, and I9) are derived from partial melting ofinfracrustal metaigneous rocks. They are hornblende-bear-ing and have relatively high Ca and Na contents. S-typegranites (11 samples; S1, S2, S3, S4, S5, S6, S7, S8, S9,HP49-A, and HP21-C) are derived from partial melting ofsupracrustal sedimentary source material that has been ex-posed to chemical weathering. They are generally more fel-sic than I-types and have low Ca and Na content relative toI-type granites due to aqueous removal of these elementsduring weathering of feldspars into clay minerals (Chappelland White, 2001). Another important difference between I-and S- type granites is that they tend to differ in their Fe3+/Fe2+ ratios. S-type granites generally have lower Fe3+/(Fe3++Fe2+) ratios (�0.16) than I-types (�0.30) (Whalenand Chappell, 1988). The metasedimentary sources of S-type granites are generally rich in graphite (Flood andShaw, 1975), which explains the more reducing conditionsof their formation (French and Eugster, 1965). Classifica-tion of A-type granite (two samples; A1 and A2) is morecomplex as it also involves considerations of the tectonicsetting (Loiselle and Wones, 1979). Whalen et al. (1987),Eby (1990) and Bonin (2007) provide detailed descriptionsof the chemical and petrographic characteristics of A-typegranites. They form in anorogenic environments, often intensional regimes, are mildly alkaline, and their parentalmagmas are inferred to have been relatively anhydrous.There is considerable uncertainty regarding their originand a unique petrogenetic model may not be appropriate.Some of the major models that have been proposed includemelting of relatively anhydrous granulitic meta-igneous res-tite left from a prior episode of melt extraction (Collinset al., 1982; Clemens et al., 1986; Whalen et al., 1987), melt-ing of granulitic crust metasomatized by mantle-derivedalkaline fluids (Martin, 2006), derivation from basalticmagma by magmatic differentiation (Loiselle and Wones,1979; Turner et al., 1992), and partial melting of crustaltonalitic-granodioritic rocks (Anderson, 1983; Creaseret al., 1991). The chemical compositions of most samplesstudied here are available in the literature (Hine et al.,1978; Chappell, 1984; Chappell and White, 1992, 1993;Nabelek et al., 1992; King et al., 1997; Sha and Chappell,1999). However, when not available (four out of 22 samplesstudied; I4, I6, I7 and S7), the chemical compositions weremeasured at Service d’Analyse des Roches et des Mineraux,Nancy, France, using established protocols (Carignan et al.,2001). The SiO2 concentrations of bulk samples range from56 to 77 wt.%. In order to better understand the processesthat control Fe isotopic fractionation in felsic rocks, min-eral separates (magnetite, biotite and hornblende) fromLFB granitoids (samples I1, I2, I3, I5 and A2) were alsomeasured.

2.2. Migmatites and pegmatites

Migmatites are generally found in high-grade metamor-phic terrains and are made up of leucosomes (L) that are

dominated by large quartz crystals and are thought to rep-resent partial melts, and melanosomes (M) that are domi-nated by biotite and sillimanite and are thought torepresent the residues of partial melting. Although not sam-pled for this study, migmatites also consist of non-migma-tized (non-separated) portions, called mesosomes, whichconsist of varying portions of quartz, biotite, sillimaniteand plagioclase. Schists, which are compositionally verysimilar to mesosomes, were analyzed for this study. Theseschists are associated with both migmatites and granitesof the Black Hills, South Dakota (Nabelek, 1999). Migma-tites are important for studying melt segregation and pro-cesses governing granite genesis and differentiation. Theyrepresent rocks in which melt has segregated on a localscale but has not escaped from the system. Melting inmigmatites begins at grain boundaries and as melting pro-gresses, melt pockets combine to form a network of meltchannels, which subsequently can migrate to produce gra-nitic bodies (Sawyer, 1991, 1994; Brown, 1994, 2002).

A total of 11 migmatite components (five leucosome–melanosome pairs and one unpaired leucosome) and twoschists from the Black Hills region of South Dakota werestudied. Melanosomes are very similar in chemical compo-sition, while leucosomes show more variations (e.g., 115-1-L and 131-1-L have very high concentrations of K2O andNa2O and 118-1-L has high concentrations of CaO andP2O5 compared to other leucosomes). The bulk SiO2 con-tent ranges from 72 to 79 wt.% for leucosomes, from 47to 62 wt.% for melanosomes and from 63 to 69 wt.% forschists. The FeO concentrations range from 0.3 to0.6 wt.% for leucosomes, and from 8 to 17 wt.% for mela-nosomes. Schists fall in between with �5.5 wt.% FeO.

Pegmatites are coarse-grained rocks, typically associatedspatially and compositionally with granites. Large quartz,feldspar and muscovite crystals make up the bulk of mostpegmatites. They have been interpreted as the products ofextreme crystal fractionation of granitic magmas (London,1996, 2005). They provide insight to the final stages of gran-ite differentiation, which is heavily influenced by fluid exso-lution processes (Norton and Redden, 1990). Suchprocesses have been suggested as the major mechanismfor producing variations in the Fe isotopic composition ofevolved felsic rocks (Poitrasson and Freydier, 2005; Hei-mann et al., 2008).

To address the influence of fluid exsolution on the Feisotopic composition, we studied four samples from simplepegmatites (81BH 5-1, 81BH 6-3, 81BH 6-4, WC-9) andthree wall zone samples from the more complex, zoned,Tin Mountain pegmatite (81BH9-2, 81BH10-3, 82BH43-1), which are closely associated (�12 km southwest) withthe Proterozoic Harney Peak Granite and migmatites fromthe Black Hills (Walker, 1984; Teng et al., 2006). The TinMountain pegmatite consists of five zones (wall zone, firstintermediate zone, second intermediate zone, third interme-diate zone, and core), which are characterized by their dis-tinct mineral assemblages, with the wall zone crystallizingfirst. Oxygen isotopic thermometry yields crystallizationtemperatures ranging from >600 �C in the outer regionsto 500 �C in the core (Walker et al., 1986), while some tem-peratures as low as 350 �C have been reported (Sirbescu

250 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

and Nabelek, 2003). Trace-element data (Rb, Li, Sr, andBa) suggest two possible origins for the Tin Mountain Peg-matite: low degree partial melting of metasedimentaryrocks that experienced moderate fractional or equilibriumcrystallization, or derivation from the Harney Peak Granitethrough multiple stages of crystal-liquid fractional crystalli-zation (Walker et al., 1989).

Pegmatites from Little Nahanni, Logan Mountains,Northwest Territories, Canada (ca. 82 Ma) were also stud-ied (six samples; P389704, P389910, P389913, P389726,P389739, and P389734). They have been characterized indetail by Groat et al. (2003). These samples are rare-ele-ment pegmatites, which generally have lower crystallizationtemperatures compared to simple and zoned pegmatites.The nearest granitic intrusion is 4.7 km2, 84–87 Myr and�15 km away. Rare-element granitic pegmatites typicallycontain high concentrations of H2O, Li, B, F, and P. Theysolidify under chemical and physical conditions that differfrom predictable paths of crystallization for normal felsicmelts. The mechanism for the formation of these pegmatitesis not well understood, but extreme fractional crystalliza-tion is the preferred interpretation (Cerny, 1985, 1991).Chemical evolution of these pegmatites shows systematictrends of Mg and Fe decreasing, while Al, Mn, Rb andCs increase from wall to core (Groat et al., 2003). LittleNahanni pegmatites are divided into spodumene-bearingand spodumene-free, but apart from this distinction, theyare similar in their major element composition to each otherand to pegmatites from the Black Hills. The SiO2 concen-trations of all pegmatites measured for this study rangefrom 68 to 74 wt.% and FeO concentrations range from0.3 to 2.2 wt.%.

The major element compositions for all samples ana-lyzed in this study are compiled in Table 1. A total alkalivs. SiO2 diagram (Fig. 1) provides an illustration of thecompositional diversity of the samples studied.

3. ANALYTICAL METHODS

Samples were crushed to a fine powder in an agate mor-tar. Mafic minerals from samples I1, I2, I3, I5 and A2 wereseparated using a combination of magnetic separation,manual extraction and density separation for Fe isotopicanalysis. Mineral mode compositions were estimated byanalyzing thin sections of these rocks with an optical micro-scope. A chart was used as a visual aid to help determinethe percentages of the Fe-bearing minerals in each thin sec-tion. The modal abundances range from 1–3& for magne-tite, 6–20% for biotite and 10–18% for hornblende. Thesenumbers are rough estimates. The samples were crushed,sieved under water, and fractions between 125 and250 lm were collected. Magnetite was separated by cover-ing a hand magnet with weighing paper and rastering itacross the grains. A Frantz magnetic separator was usedon each sample to separate mafic from non-mafic mineralfractions. Biotite was then separated from the mafic miner-als by simply pouring out the sample on a sheet of paperand continuously tapping and rotating the paper at variousangles until biotite was all that remained since its sheet sil-icate crystal structure allows it to stick to the sheet of paper.

Pure hornblende was collected through heavy liquid separa-tion and was then handpicked from the high-density frac-tion under an optical microscope.

3.1. Sample preparation and instrument analysis

The samples were prepared for iron isotope analysesaccording to the procedure described in Dauphas et al.(2004, 2009). Sizes of the sample powder aliquots rangedfrom 10 to 20 mg for geostandards, from 20 to 30 mg forplutonic rocks, from 21 to 35 mg for migmatites, from 18to 70 mg for pegmatites, and from 2 to 9 mg for the mineralseparates. Blanks and sets of geostandards were measuredtogether with samples. The blank contribution (<40 ng) toiron isotope analyses was always negligible. After completedissolution by step-wise acid digestion, the samples wereloaded in HCl 6 M on columns filled with AG1-X8 anionexchange resin. After elution of matrix elements and directisobars in 8 ml of HCl 6 M, iron was recovered from the re-sin in 9 ml total volume of HCl 0.4 M. This purificationstep was repeated a second time. All Fe isotope measure-ments were collected using the Neptune MC-ICPMS atthe Origins Laboratory of the University of Chicago. Thesamples were introduced into the mass spectrometer usinga 100 ll/min Teflon nebulizer inside a quartz cyclonic-scottspray chamber. The measurements were performed in med-ium resolution mode to resolve interferences of ArN, ArO,and ArOH on 54Fe, 56Fe, and 57Fe. Instrumental mass frac-tionation was corrected for using standard-sample bracket-ing. Iron isotopic variations are expressed in d-notation,

diFe ¼ ðiFe=54FeÞsample=ðiFe=54FeÞIRMM�014 � 1h i

� 103;

where i refers to mass 56 or 57 and IRMM-014 is a refer-ence material with near-chondritic iron isotopic composi-tion (Dauphas and Rouxel, 2006; Craddock andDauphas, 2010). Typical precisions on d56Fe and d57Feare ±0.03 and ±0.05& (95% confidence interval), respec-tively. Dauphas et al. (2009) showed that the Fe isotopicmeasurements are accurate at this level. Several geostan-dards were also analyzed and these have iron isotopic com-positions that are indistinguishable from recommendedvalues (Craddock and Dauphas, 2010).

For zinc, the experimental procedures were adaptedfrom those described by Moynier et al. (2006). Powderedsamples weighing 200–300 mg were leached in water for5 min in an ultrasonic bath. The leaching solution was re-moved and the residue was dissolved for 48 h in HNO3/HF at 130 �C. The samples were loaded in 1.5 M HBr onAG-1X8 200–400 mesh and Zn was extracted in 0.5 MHNO3. The same process was repeated to further purifyZn. Zinc isotopic compositions were measured on a NuPlasma High Resolution MC-ICP-MS at the Ecole Nor-male Superieure de Lyon (France), as described in Mare-chal et al. (1999). The total blank is �10 ng, which isnegligible compared to the total amount of Zn in the sam-ples (>10 lg). Isotope ratios are expressed as relative devi-ations, in d-notation, from the Zn JMC-Lyon standard as:

diZn ¼ ðiZn=64ZnÞsample=ðiZn=64ZnÞJMC�Lyon � 1h i

� 103;

Table 1Iron isotope and major element compositions of geostandards, granitoids, pegmatites and migmatites.

Sample type Name d56Fe(&) d57Fe(&) Averages of replicate

analyzes*

Major element compositions (wt.%)

d56Fe(&) d57Fe(&) SiO2 Al2O3 Fe2O3 FeO MnO MgO CaO Na2O K2O TiO2 P2O5 H2O+ H2O� CO2 References

Granitoids

Tonalite, Tuross Head I1 0.123 ± 0.020 0.184 ± 0.026 60.13 16.42 2.04 3.73 0.11 3.27 5.91 3.94 1.41 0.99 0.28 1.03 0.18 0.28 Sha and Chappell

(1999)

Granodiorite, Yurammie I2 0.081 ± 0.021 0.120 ± 0.038 65.27 15.22 1.76 2.83 0.09 2.01 4.40 2.74 3.43 0.58 0.19 1.26 0.09 0.20 Sha and Chappell

(1999)

Granodiorite, Glenbog I3 0.112 ± 0.021 0.167 ± 0.028 67.39 14.48 1.58 3.02 0.09 1.72 4.32 2.63 2.99 0.53 0.11 0.95 0.13 0.08 Sha and Chappell

(1999)

Adamellite, Wallagaraugh I4 0.193 ± 0.021 0.287 ± 0.030 74.87 12.97 1.04 0.60 0.05 0.08 0.38 3.63 4.71 0.06 0.00 This study

Tonalite, Jindabyne I5 0.067 ± 0.020 0.129 ± 0.026 62.29 16.46 1.85 3.41 0.10 2.85 6.05 2.60 2.01 0.54 0.11 1.20 0.20 0.10 Sha and Chappell

(1999)

Qtz. Diorite, Clear Hills I6 0.064 ± 0.021 0.109 ± 0.030 56.20 13.13 8.18 6.23 0.15 8.20 7.18 2.04 2.09 0.84 0.24 This study

Adamellite, Eugowra I7 0.121 ± 0.022 0.194 ± 0.041 69.51 15.34 2.53 1.22 0.06 0.59 2.56 4.81 2.57 0.42 0.10 This study

___, Lysterfield I8 0.125 ± 0.020 0.193 ± 0.025 63.73 15.14 0.87 3.77 0.07 2.40 3.74 2.65 4.08 0.88 0.24 1.02 0.20 0.32 Chappell and White,

1992

Dacite, Kadoona I9 0.112 ± 0.020 0.147 ± 0.031 66.17 14.52 1.80 3.40 0.09 2.19 4.72 2.46 2.74 0.59 0.11 0.68 0.16 0.31 Chappell and White

(1992)

Granodiorite, Cooma S1 0.099 ± 0.022 0.133 ± 0.034 72.00 13.72 0.59 3.48 0.06 1.76 0.95 1.49 3.73 0.54 0.13 1.28 0.14 0.13 Chappell (1984)

Granodiorite, Cowra S2 0.097 ± 0.020 0.114 ± 0.029 67.87 14.59 0.81 4.02 0.11 2.13 2.37 1.80 3.60 0.68 0.15 1.98 0.26 0.13 Chappell and White

(1993)

Granodiorite, Jillamatong S3 0.114 ± 0.020 0.150 ± 0.029 67.68 14.70 0.68 4.03 0.07 2.22 2.26 1.92 3.60 0.64 0.15 1.51 0.22 0.12 Chappell (1984)

Adamellite, Minnegans S4 0.155 ± 0.033 0.215 ± 0.059 68.72 14.36 0.81 3.36 0.06 1.95 1.97 1.99 3.94 0.57 0.16 1.65 0.23 0.12 Hine et al. (1978)

Granodiorite, Dalgety S5 0.113 ± 0.020 0.182 ± 0.026 68.21 14.25 0.93 3.10 0.07 1.68 3.11 2.23 3.77 0.53 0.13 1.28 0.12 0.29 Chappell (1984)

Adamellite, Numbla S6 0.139 ± 0.021 0.185 ± 0.038 73.48 13.41 0.44 1.47 0.04 0.71 1.70 2.63 4.63 0.22 0.08 0.74 0.14 0.11 Chappell (1984)

___, Koetong S7 0.112 ± 0.022 0.170 ± 0.032 73.65 14.00 1.60 0.98 0.03 0.32 0.80 3.17 4.36 0.17 0.27 This study

Granite, Strathbogie S8 0.152 ± 0.020 0.219 ± 0.025 73.65 13.08 0.16 2.27 0.05 0.61 1.20 2.58 4.60 0.36 0.17 0.90 0.16 0.09 Chappell and White

(1992)

Dacite, Hawkins S9 0.072 ± 0.021 0.122 ± 0.030 68.05 14.41 0.86 3.57 0.06 2.05 2.68 2.03 3.58 0.61 0.15 1.21 0.15 0.24 Sha and Chappell

(1999)

Granite, Watergums A1 0.148 ± 0.022 0.227 ± 0.034 73.60 12.72 1.99 0.82 0.04 0.20 0.83 3.62 4.11 0.38 0.08 0.73 0.29 0.29 King et al. (1997)

Granite, Mumbella A2 0.305 ± 0.022 0.467 ± 0.032 77.00 11.83 0.40 1.05 0.04 0.04 0.61 3.06 4.98 0.15 0.02 0.47 0.18 0.07 King et al. (1997)

Granite, Black Hills, S.

Dakota, USA

HP49-A 0.161 ± 0.052 0.270 ± 0.055 72.40 16.10 – 1.31 0.13 0.15 0.57 7.31 0.11 0.02 0.22 Nabelek et al. (1992)

Granite, Black Hills, S.

Dakota, USA

HP21-C 0.240 ± 0.041 0.368 ± 0.047 75.10 13.50 – 1.52 0.02 0.41 0.76 2.81 3.96 0.16 0.18 Nabelek et al. (1992)

Pegmatites

Simple Pegmatites

Black Hills, S. Dakota, USA 81BH 5-1 0.270 ± 0.054 0.330 ± 0.105 74.45 15.85 1.52 – 0.04 0.45 0.56 5.75 1.39 0.06 0.09 Walker (1984)

Black Hills, S. Dakota, USA 81BH 6-3 0.151 ± 0.054 0.262 ± 0.105 72.21 9.40 1.03 – 0.13 0.00 0.40 – 4.08 0.01 – Walker (1984)

Black Hills, S. Dakota USA 81BH 6-4 0.290 ± 0.054 0.377 ± 0.105 71.99 15.10 0.12 – 0.00 0.38 0.09 2.42 10.10 – – Walker (1984)

Black Hills, S. Dakota, USA WC-9 0.209 ± 0.054 0.305 ± 0.105 73.34 12.60 0.40 0.01 0.00 0.36 2.83 6.90 0.06 0.50 Walker (1984)

Tin Mountain Pegmatites

Black Hills, S. Dakota, USA 81BH 9-2 0.197 ± 0.054 0.268 ± 0.105 74.40 13.54 – 2.20 0.05 0.02 0.64 2.68 2.22 0.02 0.56 Walker et al. (1989)

Black Hills, S. Dakota USA 81BH 10-3 0.221 ± 0.054 0.375 ± 0.105 68.32 17.54 – 1.54 0.26 0.01 0.93 9.13 0.39 0.02 1.12 Walker et al. (1989)

Black Hills, S. Dakota, USA 82BH 43-1 0.385 ± 0.054 0.562 ± 0.105 70.78 15.67 – 0.51 0.05 0.01 0.61 8.01 2.01 0.08 0.50 Walker et al. (1989)

Rare-element Pegmatites

Little Nahanni, Canada P389734 0.189 ± 0.039 0.285 ± 0.060 72.31 16.73 0.27 0.26 0.23 0 0.08 3.24 3.92 0.45 1.03 Groat et al. (2003)

Little Nahanni, Canada P389910 0.061 ± 0.039 0.101 ± 0.059 0.053 ± 0.027 0.096 ± 0.042 70.76 17.56 0.28 0.26 0.15 0.01 0.61 4.77 3.58 0.64 0.78 Groat et al. (2003)

(continued on next page)

M.

Telu

set

al./G

eoch

imica

etC

osm

och

imica

Acta

97(2012)

247–265251

Table 1 (Continued)

Sample type Name d56Fe(&) d57Fe(&) Averages of replicate analyzes* Major element compositions (wt.%)

d56Fe(&) d57Fe(&) SiO2 Al2O3 Fe2O3 FeO MnO MgO CaO N O K2O TiO2 P2O5 H2O+ H2O� CO2 References

0.046 ± 0.039 0.091 ± 0.060

Little Nahanni, Canada P389913 0.160 ± 0.039 0.225 ± 0.059 0.137 ± 0.027 0.206 ± 0.042 67.16 19.42 0.46 0.32 0.10 0.10 1.07 5 3.20 0.74 1.30 Groat et al.

(2003)

0.115 ± 0.039 0.185 ± 0.060

Little Nahanni, Canada P389704 0.202 ± 0.041 0.317 ± 0.047 0.155 ± 0.027 0.261 ± 0.036 74.32 16.71 0.15 0.13 0.14 0.01 0.42 3 2.32 0.00 0.26 0.71 Groat et al.

(2003)

(Spodumene-bearing) 0.118 ± 0.036 0.184 ± 0.055

Little Nahanni, Canada P389726 �0.083 ± 0.039 �0.113 ± 0.059 �0.067 ± 0.027 �0.105 ± 0.040 70.67 17.24 0.27 0.26 0.21 0.04 0.63 4 3.22 0.91 0.91 Groat et al.

(2003)

(Spodumene-bearing) �0.052 ± 0.036 �0.097 ± 0.055

Little Nahanni, Canada P389739 0.215 ± 0.052 0.307 ± 0.055 71.77 17.36 0.28 0.26 0.16 0.00 0.65 4 2.26 0.89 0.92 Groat et al.

(2003)

(Spodumene-bearing)

Migmatites

Leucosomes, Black Hills, S.

Dakota, USA

115-1L 0.246 ± 0.031 0.372 ± 0.045 72.46 14.85 0.68 0.01 0.22 0.40 1.62 8 0.06 0.41 0.59 Nabelek

(1999)

Melanosomes, Black Hills, S.

Dakota, USA

115-1M 0.138 ± 0.031 0.205 ± 0.045 61.86 20.67 7.65 0.11 2.55 0.18 0.23 4 0.69 0.14 1.25 Nabelek

(1999)

Leucosomes, Black Hills, S.

Dakota, USA

118-1L 0.272 ± 0.029 0.416 ± 0.050 74.91 14.55 0.39 0.10 0.07 4.85 0.06 0 0.01 3.84 0.33 Nabelek

(1999)

Melanosomes, Black Hills, S.

Dakota, USA

118-1M 0.077 ± 0.029 0.153 ± 0.050 59.24 13.92 12.53 0.15 4.23 0.36 0.12 5 1.35 0.27 0.88 Nabelek

(1999)

Leucosomes, Black Hills, S.

Dakota, USA

131-1B-L 0.273 ± 0.031 0.395 ± 0.045 0.275 ± 0.025 0.399 ± 0.032 76.40 13.82 0.55 0.01 0.15 1.00 3.64 3 0.05 0.09 0.55 Nabelek

(1999)

0.279 ± 0.041 0.403 ± 0.047

Melanosomes, Black Hills, S.

Dakota, USA

131-1B-M 0.237 ± 0.031 0.329 ± 0.045 61.75 20.21 8.30 0.08 2.90 0.11 0.32 4 0.93 0.06 0.12 Nabelek

(1999)

Leucosomes, Black Hills, S.

Dakota, USA

127-1L 0.355 ± 0.028 0.517 ± 0.052 79.03 18.59 0.39 0.01 0.04 0.26 0.42 0 0.02 0.16 0.54 Nabelek

(1999)

Melanosomes, Black Hills, S.

Dakota, USA

127-1M 0.302 ± 0.028 0.456 ± 0.052 46.86 16.83 16.82 0.17 5.39 0.06 0.16 7 2.20 0.09 1.17 Nabelek

(1999)

Leucosomes, Black Hills, S.

Dakota, USA

129-1B-L 0.283 ± 0.028 0.389 ± 0.052 79.45 16.52 0.54 0.01 0.16 0.17 0.67 2 0.05 0.04 0.40 Nabelek

(1999)

Melanosomes, Black Hills, S.

Dakota, USA

129-1B-M 0.186 ± 0.028 0.270 ± 0.052 55.56 20.65 9.62 0.11 3.59 0.33 0.87 5 1.09 0.10 1.52 Nabelek

(1999)

Leucosomes, Black Hills, S.

Dakota, USA

118-2 0.480 ± 0.029 0.738 ± 0.050 78.74 18.35 0.80 0.02 0.22 0.42 0.00 0 0.06 0.37 0.65 Nabelek

(1999)

No melanosome pair

Schists, Black Hills, S.

Dakota, USA

84-1 0.146 ± 0.031 0.253 ± 0.060 69.34 14.67 5.14 0.06 1.75 0.65 1.73 3 0.7 0.16 1.24 Nabelek

(1999)

Schists, Black Hills, S.

Dakota, USA

157-1 0.158 ± 0.028 0.213 ± 0.052 63.89 17.39 6.14 0.08 2.15 0.76 1.90 4 0.8 0.17 1.44 Nabelek

(1999)

Notes: Error bars on d56Fe and d57Fe are 95% confidence intervals.* Replicates are for newly digested samples.

252M

.T

elus

etal./

Geo

chim

icaet

Co

smo

chim

icaA

cta97

(2012)247–265

a2

.41

.66

.66

.17

.43

.10

.23

.67

.24

.00

.67

.64

.32

.19

.37

.57

.31

I-typeS-typeA-type

Pegmatites

LeucosomesMelanosomesSchists

0

4

8

12

45 55 65 75 85SiO2 (wt.%)

Na 2

O +

K2O

(wt.%

)

Basalt Basaltic andesite

Andesite Dacite

Rhyolite

Basaltic trachy-andesite

Trachyte

Trachy-andesite

Foidite

Migmatites

Granitoids

Fig. 1. Total alkali vs. SiO2 wt.% illustrates the compositionalrange of samples analyzed for this study. Dashed lines connectleucosomes to their associated melanosomes.

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 253

where i refers to mass 66 or 68. External reproducibilities(2SD) of ±0.09& for d66Zn and ±0.3& for d68Zn are esti-mated from 10 replicate measurements of one sample thathad been split into 10 separate aliquots and measured inde-pendently by MC-ICP-MS (Herzog et al., 2009).

Magnesium isotopic analyses were performed on a NuPlasma MC-ICP-MC at the Isotope Laboratory of the Uni-versity of Arkansas, Fayetteville, following established pro-cedures (Yang et al., 2009; Li et al., 2010; Teng et al.,2010b). The digested samples were taken up in 1 MHNO3 for column purification. Separation of Mg wasachieved by cation exchange chromatography with Bio-Rad 200–400 mesh AG50 W-X8 resin in 1 M HNO3. Thisstep was repeated twice for each sample. The proceduralblank was less than 10 ng (i.e., less than 0.1% of Mg loadedon the column). Geostandards were processed with samplesto verify that the column purification does not introducespurious effects (Teng et al., 2010a). Magnesium isotopiccompositions were analyzed by the standard bracketingmethod and data are reported in standard d-notation inpermil relative to DSM3, a synthetic standard:

diMg ¼ ðiMg=24MgÞsample=ðiMg=24MgÞDSM3 � 1h i

� 1000;

where i refers to mass 25 or 26. The long-term precision,based on replicate analyses of synthetic solutions, mineralsand rock standards, is ±0.07& (2SD) (Yang et al., 2009;Teng et al., 2010a).

Samples were prepared for uranium isotope analysesaccording to the procedure outlined by Tissot and Dauphas(2011, 2012) and Weyer et al. (2008). The samples weredouble-spiked prior to digestion using commercially avail-able IRMM-3636 (�49.51% of 236U and �50.46% of233U). The amount of spike added is such that the ratio Us-

pike/Usample is �3% for each sample. In granitic rocks, Ucan be hosted in accessory minerals that are notoriously dif-ficult to digest, such as zircons. To ensure that the U iso-tope measurements are representative of the bulk rocks,the samples were digested in several steps: (1) two 24 hdigestions in HF/HNO3 2:1 on hot plates, (2) a 5 day diges-tion in Parr Bombs in HF/HNO3 2:1 at 180 �C, and (3) two

24 h digestions in HCl/HNO3 2:1 on hot plates. After dry-ing, the samples were dissolved in HNO3 3 M to be loadedon U/Teva resin (Eichrom). Most matrix elements wereeluted in 40 ml HNO3 3 M. Neptunium and Th were subse-quently removed with 20 ml HCl 5 M + 1 M oxalic acid(H2C2O4) followed by 10 ml HCl 5 M to rinse off the oxalicacid. Uranium was finally eluted in 25 ml HCl 0.05 M andthe procedure was repeated twice to ensure complete matrixremoval. All U isotope measurements were performed onthe Neptune MC-ICPMS at the Origins Laboratory ofthe University of Chicago, using a desolvating nebulizerAridus II. Uranium isotope analyses were done after theinstrument was upgraded with an OnTool Booster 150 Jetpump (Pfeiffer). Enhanced signal stability was achieved byplacing a spray chamber between the Aridus II and theMC-ICPMS. Correction of isotopic mass fractionationintroduced during chemical separation and mass spectrom-etry was done using a 233U/236U double spike (IRMM-3636). An additional correction was applied by bracketingsamples with standard measurements also spiked withIRMM-3636 at the same level as the samples. The proce-dural blank was estimated to be 0.02–0.05 ng U (0.01% ofsample uranium). Contributions of the 238U tail onto236U, 235U and 234U were estimated to be, respectively,0.6 � 10�6, 0.25 � 10�6 and 0.1 � 10�6 of the 238U intensityand were corrected for. The 238U tail corrections were sig-nificant only for 234U and 236U. Ratios are reported ind238U notation relative to the U standard CRM-112a (alsonamed SRM960, NBL112-a or CRM-145).

d238U ¼ ð238U=235UÞsample=ð238U=235UÞCRM�112a � 1h i

� 103:

The typical precision on d238U is ±0.05& (2SD). Twogranite geostandards were analyzed (G-2 and JG-1; Table 2)and show reasonable agreement with the data reported byWeyer et al. (2008). Comparison of a large record of geo-standards analyzed at the Origins Laboratory shows no sys-tematic bias when compared to literature data.

3.2. Replicates and geostandards

Samples that were especially difficult to dissolve werereplicated to ensure reproducibility of d56Fe measurements.Replicates were done mostly for pegmatite samples. Out offive replicate analyses, only one (P389704) has a replicatevalue that does not agree within error (i.e., d56Fe =+0.202 ± 0.041& vs. +0.118 ± 0.036&). The reason forthis is unclear but could be due to incomplete digestion.The averages of the replicate analyses are used in the figures(Table 1).

Accuracy and precision of sample preparation and iso-topic measurements were also evaluated by analyzing sev-eral international standards, such as AC-E, BE-N,BHVO-1, BHVO-2, DNC-1, DR-N, GA, and GSN.BHVO-1 was measured during each run and the replicatecompositions agree within analytical uncertainty. The mea-sured d56Fe values of the standards (Table 2) agree withinerror with previously published values (Craddock and Dau-phas, 2010 and references therein).

Table 2Iron , Zn, Mg and U isotopic compositions of geostandards.

Averages of replicateanalyses*

Sample type Name d56Fe (&)a d57Fe (&) d56Fe (&) d57Fe (&) d66Zn (&)b d68Zn (&) d25Mg (&)c d26Mg (&) U (ppm) d238U (&)d

Granite, Ailsa Craig island,Scotland

AC-E 0.311 ± 0.027 0.471 ± 0.037 0.318 ± 0.025 0.479 ± 0.0350.355 ± 0.065 0.556 ± 0.115

Basalt, Essey-la-Cote,France

BE-N 0.162 ± 0.039 0.242 ± 0.060 0.146 ± 0.024 0.210 ± 0.036 0.466 ± 0.09 0.849 ± 0.300.135 ± 0.031 0.192 ± 0.045

Basalt, Hawaii, USA BHVO-1 0.091 ± 0.028 0.150 ± 0.040 0.118 ± 0.013 0.170 ± 0.0190.104 ± 0.029 0.141 ± 0.0440.133 ± 0.052 0.162 ± 0.0550.133 ± 0.031 0.194 ± 0.0450.139 ± 0.036 0.192 ± 0.0550.137 ± 0.033 0.194 ± 0.046

Basalt, Hawaii, USA BHVO-2 0.120 ± 0.052 0.173 ± 0.055 0.294 ± 0.09 0.601 ± 0.30Dolerite, North Carolina,USA

DNC-1 0.065 ± 0.039 0.106 ± 0.059

Diorite, Neuntelsein, France DR-N 0.126 ± 0.039 0.202 ± 0.059 0.288 ± 0.09 0.502 ± 0.30Calc-alkaline granite, VosgesMountains,France

GA 0.137 ± 0.031 0.195 ± 0.045 0.267 ± 0.09 0.585 ± 0.30

Granite, France GSN 0.152 ± 0.041 0.239 ± 0.047Trachyte, Mont Dore,France

MDO-G 0.344 ± 0.09 0.773 ± 0.30

Serpentinite, ____ UB-N 0.380 ± 0.09 0.726 ± 0.30Pitscurrie Microgabrro,Scotland

PM-S 0.444 ± 0.09 0.936 ± 0.30

Sill Dolerite, England WS-E 0.173 ± 0.09 0.372 ± 0.30synthetic standardMg:Fe:Al:Ca:Na:K:Ti =1:1:1:1:1:1:0.1

DSM-3 �0.01 ± 0.07 �0.01 ± 0.09

Olivine, Kilbourne Hole,New Mexico, USA

KHsetp22

�0.18 ± 0.07 �0.33 ± 0.09

Granite, Rhode Island, USA G2 1.86 ± 0.01 �0.11 ± 0.10Sori Granodiorite, Japan JG-1 4.00 ± 0.02 �0.30 ± 0.04

a Error bars on d56Fe and d57Fe are 95% confidence intervals.b External reproducibilities (2SD) of ± 0.09& for d66Zn and ±0.3& for d68Zn are estimated from 10 replicate measurements of one sample that had been split into 10 separate aliquots and

measured independently by Herzog et al. (2009).c External reproducibilities (2SD) of 0.06& for d25Mg and 0.07& for d26Mg is based on replicate analyses of synthetic solutions, mineral and rock standards reported in Yang et al. (2009) and

Teng et al. (2010a). 2SD = 2 times the standard deviation of the population of 4 repeat measurements of a sample solution.d External reproducibilities (2SD) of 0.05& for d238U is based on numerous (>250) replicate measurements of the bracketing standard spiked at the same level as the samples.

* Replicates are for newly digested sample.

254M

.T

elus

etal./

Geo

chim

icaet

Co

smo

chim

icaA

cta97

(2012)247–265

-0.1

0.1

0.3

0.5

45 55 65 75 85SiO2 (wt.%)

δ56Fe

bulk

(‰)

Granitoids

I-typeS-typeA-type

Black HillsLittle Nahanni

LeucosomesMelanosomesSchists

Volcanic Rocks

MORBs Literature data

Granitoids

Pegmatites

Migmatites

Fig. 2. d56Fe vs. SiO2 wt.% of migmatites, granitoids and pegma-tites measured in this study compared to results of previous studiesfor silicic plutonic and volcanic rocks (Poitrasson and Freydier,2005; Heimann et al., 2008; Sossi et al., 2012). The shaded arearepresents the d56Fe value of terrestrial basalts (MORBs) of+0.110 ± 0.018& (Dauphas et al., 2009). Granitic rocks becomeenriched in 56Fe with increasing SiO2 wt.%, which is in agreementwith previous work. The d56Fe values of leucosomes are consis-tently heavier than average granites and MORBs, while the d56Fevalues of melanosomes are scattered. Dashed lines connectleucosomes to their associated melanosomes. Schists are slightlyheavier than MORBs, but similar in composition to averagegranitic rocks.

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 255

4. RESULTS

4.1. Iron isotopic compositions of felsic rocks

The Fe isotopic compositions of 22 granitic rocks (I-, S-,and A-type), 11 migmatite components (six leucosomes andfive melanosomes), and 13 samples of pegmatites are re-ported in Table 1 and plotted in Fig. 2. The results for geo-standards are reported in Table 2 and 14 mineral separates(five biotite, four amphibole, and five magnetite) are re-ported in Table 3. All samples follow mass-dependent frac-tionation, i.e., d57Fe = (+1.49 ± 0.09) � d56Fe.

4.1.1. Granitoids

The d56Fe values of granitic rocks measured in this studyrange from +0.064 ± 0.021& to +0.305 ± 0.022& (95%confidence intervals). The range of d56Fe values for I-typegranitoids is from +0.064 ± 0.021& to +0.193 ± 0.021&.S-type granitoids have a similar range of d56Fe values (from+0.072 ± 0.021& to +0.240 ± 0.041&). Poitrasson andFreydier (2005) and Heimann et al. (2008) found that plu-tonic and volcanic rocks with >70 wt.% SiO2 had highd56Fe values. This study obtained similar results (Fig. 2)although some granitoids between 70 and 75 wt.% SiO2

did not show the expected enrichment in heavy Fe isotopes.The d56Fe values of most of the granitic rocks measured inthis study lie within the range of MORBs (+0.110 ±0.018&; Dauphas et al., 2009) (Fig. 2).

4.1.2. Pegmatites

The iron isotopic compositions of pegmatites from theBlack Hills are heavier than the average igneous rock com-position with d56Fe values ranging from +0.151 ± 0.054&

to +0.385 ± 0.054& (Fig. 2). The “rare-element” pegma-tites from Little Nahanni have a range of d56Fe valuesbetween �0.067 ± 0.027& and +0.215 ± 0.052&. It com-prises the only pegmatite sample with a negative d56Fe va-lue. They have been classified as spodumene-bearing andnon-spodumene-bearing (Groat et al., 2003; see Table 1),but this classification does not seem to influence the d56Fecomposition. Rare-element granitic pegmatites, like thosefrom Little Nahanni, contain high concentrations of vola-tile elements, such as H2O, Li, B, F, and P (Groat et al.,2003), which may explain the scatter in their d56Fe compo-sitions (Fig. 2).

Pegmatites from both the Black Hills and Little Nahan-ni show no obvious correlation between d56Fe andSiO2 wt.%, or any other major oxides or trace elements.

4.1.3. Migmatites

The Fe isotopic compositions of thirteen migmatitecomponents from the Black Hills, South Dakota were mea-sured (Fig. 2): six leucosomes, five melanosomes and twoschists. The d56Fe values of the leucosomes range from+0.246 ± 0.031& to +0.480 ± 0.029&. Melanosomes haved56Fe compositions ranging from +0.077 ± 0.029& to+0.302 ± 0.028&. The two schists have an intermediated56Fe composition of +0.153 ± 0.021&. Ten of the 13migmatite components are leucosomes–melanosomes pairs(i.e., the leucosomes and melanosomes are spatially relatedto each other). The differences between d56Fe values forassociated melanosomes and leucosomes range from+0.038 ± 0.056& to +0.196 ± 0.058&. The Fe isotopiccompositions of leucosomes are systematically heavier thanmelanosomes compositions. Three of the five pairs do notoverlap in composition. Leucosome d56Fe values overlapwith the heaviest granite and pegmatites. Unlike granitoidsand pegmatites, none of the leucosome Fe isotopic compo-sitions overlap with MORBs. Melanosome Fe isotopiccompositions vary widely, some are indistinguishable fromMORBs and others are comparable to the heaviest pegma-tites and granitoids. Schists are slightly heavier than MOR-Bs (Fig 2).

4.1.4. Mineral separates

Magnetite was found to have the highest d56Fe compo-sition with values that range from +0.214 ± 0.026& to+0.645 ± 0.026& (Fig. 3). Without considering A2, whichhas an anomalously high d56Fe value for biotite(+0.249 ± 0.029&), the range for biotite is from�0.022 ± 0.023& to +0.056 ± 0.029&. The fractionationbetween magnetite and biotite (D56Femgt-bt) ranges between+0.234 ± 0.032& and +0.396 ± 0.037&. Hornblended56Fe values range from �0.001 ± 0.023& to+0.076 ± 0.029& and are very similar in compositions tobiotite. Iron isotopic fractionation between biotite andamphibole (D56Febt-amph) ranges between �0.048 ±0.041& and +0.030 ± 0.041&. The iron isotope measure-ments of mineral separates agree with previous results ofHeimann et al. (2008).

To assess whether all of the major Fe-bearing mineralsin each sample were accounted for, the bulk d56Fe value cal-culated from mineral mode analysis and the Fe isotopic

Table 3Fe isotopic compositions of mineral separates from granitoids.

Sample type Name d56Fe(&)(mineral)

d57Fe(&)(mineral)

d56Fe(&) bulk(measured)

Density* (gm/cm3)

FeConcentration(g/g)

Mineral mode(%)**

Fraction of Fe(mineral)

d56Fe(&) bulk(calculated)

Amphibole,Tonalite

I1-Amph 0.053 ± 0.023 0.086 ± 0.036 0.123 ± 0.02 3.3 0.107 18 0.272 0.147 ± 0.040

Biotite, Tonalite I1-Bt 0.045 ± 0.023 0.073 ± 0.036 3.0 0.129 18 0.301Magnetite,Tonalite

I1-Mgt 0.278 ± 0.023 0.423 ± 0.036 5.2 0.636 3 0.427

Amphibole,Granodiorite

I2-Amph �0.001 ± 0.023 0.001 ± 0.036 0.081 ± 0.021 3.3 0.113 10 0.215 0.093 ± 0.042

Biotite,Granodiorite

I2-Bt �0.022 ± 0.023 �0.038 ± 0.036 3.0 0.121 15 0.316

Magnetite,Granodiorite

I2-Mgt 0.214 ± 0.026 0.324 ± 0.041 5.2 0.520 3 0.469

Amphibole,Granite

A2-Bt 0.249 ± 0.026 0.380 ± 0.041 0.305 ± 0.022 3.0 0.107 6 0.455 0.465 ± 0.037

Magnetite, Granite A2-Mgt 0.645 ± 0.026 0.945 ± 0.041 5.2 0.444 1 0.545Amphibole,Tonalite

I3-Amph 0.026 ± 0.029 0.051 ± 0.036 0.112 ± 0.021 3.3 0.124 16 0.338 0.163 ± 0.050

Biotite, Tonalite I3-Bt 0.056 ± 0.029 0.071 ± 0.036 3.0 0.141 16 0.352Magnetite,Tonalite

I3-Mgt 0.433 ± 0.029 0.642 ± 0.036 5.2 0.573 2 0.310

Amphibole,Tonalite

I5-Amph 0.076 ± 0.029 0.116 ± 0.036 0.067 ± 0.02 3.3 0.123 16 0.440 0.079 ± 0.050

Biotite, Tonalite I5-Bt 0.028 ± 0.029 0.016 ± 0.036 3.0 0.108 20 0.443Magnetite,Tonalite

I5-Mgt 0.283 ± 0.029 0.409 ± 0.041 5.2 0.331 1 0.118

Notes: Bulk Fe isotopic data calculated using the fraction of Fe and the d56Fe for each mineral. Error bars on d56Fe and d57Fe are 95% confidence intervals.* Averages from Nesse (2000).** Mineral modes were determined by examining thin sections under an optical microscope and estimating what percentages of the minerals occupy the thin sections.

256M

.T

elus

etal./

Geo

chim

icaet

Co

smo

chim

icaA

cta97

(2012)247–265

-0.2

0.0

0.2

0.4

0.6

0.8

55 60 65 70 75 80SiO2 (wt.%)

δ56Fe

Amphibole

Amphibole

Biotite

Biotite

Magnetite

Magnetite

This Study

Heimann et al. (2008)

MORBs

of bulk rocks

of m

iner

al s

epar

ates

(‰)

Fig. 3. d56Fe of mineral separates vs. SiO2 wt.% of bulk rocks. Fe-silicates (biotite and amphibole) generally have lighter or indistin-guishable iron isotopic compositions compared to MORBs, whilemagnetite compositions are significantly heavier.

Table 4Zinc isotopic compositions of granitoids, pegmatites, and migma-tites (relative to JMC-Lyon standard).

Sample type Name d66Zn (&) d68Zn (&)

Granitoids

I1 0.247 ± 0.09 0.473 ± 0.30I2 0.192 ± 0.09 0.506 ± 0.30I4 0.266 ± 0.09 0.682 ± 0.30I6 0.171 ± 0.09 0.317 ± 0.30I7 0.119 ± 0.09 0.426 ± 0.30I8 0.179 ± 0.09 0.331 ± 0.30S1 0.130 ± 0.09 0.357 ± 0.30S2 0.122 ± 0.09 0.226 ± 0.30S3 0.330 ± 0.09 0.664 ± 0.30S4 0.224 ± 0.09 0.506 ± 0.30S5 0.325 ± 0.09 0.677 ± 0.30S6 0.209 ± 0.09 0.464 ± 0.30S7 0.357 ± 0.09 0.755 ± 0.30S8 0.355 ± 0.09 0.722 ± 0.30S9 0.303 ± 0.09 0.665 ± 0.30HP49-A 0.486 ± 0.09 0.918 ± 0.30HP21-C 0.342 ± 0.09 0.675 ± 0.30A1 0.269 ± 0.09 0.482 ± 0.30A2 0.233 ± 0.09 0.462 ± 0.30

Pegmatites

81BH 5-1 0.673 ± 0.09 1.318 ± 0.3081BH 9-2 0.875 ± 0.09 1.655 ± 0.3081BH 10-3 0.752 ± 0.09 1.422 ± 0.3082BH 43-1 0.631 ± 0.09 1.178 ± 0.30WC-9 0.534 ± 0.09 1.026 ± 0.30

Migmatites

115-L 0.258 ± 0.09 0.481 ± 0.30131-1B-L 0.238 ± 0.09 0.432 ± 0.30131-1B-M 0.179 ± 0.09 0.357 ± 0.30

Notes: External reproducibilities (2SD) of ±0.09& for d66Zn and±0.3& for d68Zn are estimated from 10 replicate measurements ofone sample that had been split into 10 separate aliquots andmeasured independently by Herzog et al. (2009).

0.0

0.2

0.4

0.6

0.8

1.0

55 60 65 70 75 80SiO2 (wt.%)

δ66Zn

(‰)

A-type

I-type

S-type

Pegmatites-Black Hills

MelanosomeLeucosomes

ig. 4. d66Zn vs. SiO2 wt.%. There is no clear correlation, but thereeems to be more dispersion in the Zn isotopic composition forore silicic magmas. The average pegmatite Zn isotopic compo-

ition is heavier than the average value for granitic rocks andigmatites by �0.4&.

Table 5Magnesium isotope compositions of granitoids (relative to DSM-3standard).

Sample type Name d25Mg (&) d26Mg(&)

GranitoidsI4 �0.216 ± 0.070 �0.446 ± 0.092I8 �0.129 ± 0.070 �0.216 ± 0.092S4 �0.123 ± 0.070 �0.251 ± 0.092S6 �0.107 ± 0.070 �0.201 ± 0.092S8 �0.104 ± 0.070 �0.209 ± 0.092HP21C �0.121 ± 0.070 �0.218 ± 0.092HP49A �0.388 ± 0.070 �0.720 ± 0.092A1 �0.104 ± 0.070 �0.214 ± 0.092A2 0.014 ± 0.070 0.026 ± 0.092

Notes: External reproducibilities (2SD) of 0.06& for d25Mg and0.07& for d26Mg and are based on replicate analyses of syntheticsolutions, minerals and rock standards reported in Yang et al.(2009) and Teng et al. (2010a). 2SD = 2 times the standard devi-ation of the population of four repeat measurements of a samplesolution.

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 257

compositions of mineral separates were compared with themeasured bulk rock compositions. The mineral mode, the

Fsmsm

density of each mineral and the Fe concentration of eachmineral give the fraction of Fe locked in each mineral x,(f Fe_x). A simple mass balance equation, 56Febulk =d56Femgtf Fe_mgt + d56Feamphf Fe_amph + d56Febtf Fe_bt, wasused to calculate the d56Fe bulk value from the composi-tions of the minerals separates (mgt refers to magnetite,amph refers to amphibole, and bt refers to biotite). Mostof the calculated bulk d56Fe measurements are within therange of the measured d56Fe values (Table 3). Predicted val-ues that are slightly inconsistent with measured valuesprobably reflect inaccuracy in mineral mode estimationusing visual charts (see “Methods” for the technique useto estimate the mineral modes).

4.2. Zinc isotopic compositions of felsic rocks

Zinc isotopic compositions were determined for 19granitoids, three migmatite components, five pegmatites

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.1 0.2 0.3 0.4 0.5δ56Fe (‰)

I-typeS-typeA-typePegmatites

MelanosomeLeucosome

δ66Zn

(‰) Black Hills

Fig. 5. d66Zn vs. d56Fe. Pegmatites and some granitic rocks havehigh d66Zn and d56Fe values. Since Zn is easily mobilized byaqueous fluids, this correlation suggests that fluid exsolution maybe responsible for the iron isotope variations measured in somefelsic rocks, as suggested by Poitrasson and Freydier (2005) andHeimann et al. (2008). However, some samples (e.g., A-typegranites and migmatites) have high d56Fe values, but unfraction-ated d66Zn values. For these samples, fractional crystallization isthe most likely cause of iron isotopic fractionation.

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

60 65 70 75 80SiO2 (wt.%)

δ26M

g (‰

)

I-type

I-type

S-type

S-type

A-type

A-type

MORBs

Literature

This Study

Fig. 6. d26Mg vs. SiO2 wt.% for granitic rocks compared toprevious studies by Shen et al. (2009), Li et al. (2010) and Liuet al. (2010). The d26Mg values for the I- and S-type granitoidsfrom this study are indistinguishable from MORBs (�0.25 ±0.07&; Teng et al., 2010a,b). Sample A2 and many other A-typegranites (Li et al., 2010) have significantly higher d26Mg values thanMORBS.

Table 6Uranium concentrations and isotope compositions of granitoids(relative to CRM-112a standard).

Sample type Name U(ppm) d238U(&)

GranitoidsI1 1.518 ± 0.010 �0.363 ± 0.033I2 2.065 ± 0.015 �0.272 ± 0.033I3 2.527 ± 0.017 �0.228 ± 0.033I4 7.109 ± 0.064 �0.250 ± 0.049I5 1.530 ± 0.007 �0.231 ± 0.030I6 2.076 ± 0.014 �0.270 ± 0.037I7 3.731 ± 0.026 �0.510 ± 0.039I8 7.595 ± 0.063 �0.385 ± 0.039I9 3.944 ± 0.027 �0.320 ± 0.039

S1 3.455 ± 0.025 �0.234 ± 0.049S1* 3.497 ± 0.025 �0.255 ± 0.033S2 2.911 ± 0.021 �0.306 ± 0.033S3 3.433 ± 0.025 �0.272 ± 0.049S4 3.343 ± 0.023 �0.300 ± 0.037S5 3.008 ± 0.021 �0.237 ± 0.037S6 4.272 ± 0.034 �0.401 ± 0.039S7 7.314 ± 0.070 �0.335 ± 0.039S8 7.533 ± 0.062 �0.285 ± 0.039S9 3.881 ± 0.028 �0.282 ± 0.039

A1 5.913 ± 0.047 �0.328 ± 0.049A2 7.208 ± 0.055 �0.340 ± 0.049

Notes: External reproducibilities (2SD) of 0.05& for d238U is basedon numerous (>250) replicate measurements of the bracketingstandard spiked at the same level as the samples.* Denotes replicates on a newly digested sample.

0

0.5

1

1.5

2

2.5

0.01 0.1 1 10 100 1000

D Zn/D

Fe

DFe (solid/liquid)

Pyroxene

Amphibole

Biotite

Olivine

Magnetite

Ilmenite

Plagioclase

Fig. 7. Solid/melt distribution coefficients of Fe and Zn inpyroxene, amphibole, biotite, olivine, magnetite, ilmenite, andplagioclase (Ewart and Griffin, 1994). Zinc is similar to Fe in radiusand valence. In addition, fractionation of Fe and Zn in mineralscommon in granitic rocks such as biotite, amphibole, and oxides issmall, so limited fractionation of the Zn/Fe ratio is expected duringmagmatic differentiation.

258 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

(Table 4, d68Zn = (+1.80 ± 0.32) � d66Zn), and eight geo-standards (Table 2). The range in d66Zn for granitic rocksis between +0.12& and +0.49& (2SD) and the range forpegmatites is from +0.53& to +0.88&. The Zn isotopiccompositions of migmatites (average +0.23 ± 0.05&) over-lap with those of granitic rocks. Pegmatites are heavier thangranitoids and migmatites by an average of �+0.4&. Thereis no clear correlation between d66Zn and SiO2 wt.% as ob-served for d56Fe (Fig. 4). There is however a relationshipbetween d66Zn and d56Fe for pegmatites (Fig. 5) in so muchthat samples with high d56Fe values also have high d66Znvalues. For granites and migmatites, the systematics is lessclear with some samples showing slightly heavy d66Zn val-ues, but for the most part, high d56Fe values are not asso-ciated with high d66Zn values.

4.3. Magnesium isotopic data

The magnesium isotopic compositions of eight graniticrocks (two I-types, five S-types and two A-types) were ana-lyzed and are reported in Table 5. Results for the standards

0

20

40

60

80

100

120

140

160

55 60 65 70 75 80SiO2 (wt.%)

Zn/F

e ( 1

04 , ato

mic

ratio

)

Magma differentiation

Fractional crystallization of a hydrous dioritic melt

Fractional crystallization of an anhydrous dioritic melt

Flui

d ex

solu

tion

Fig. 8. Zn/Fe atomic ratio (�104) vs. SiO2 wt.% for granitic rocks(data is from GEOROC database, http://georoc.mpch-main-z.gwdg.de/georoc/) and magma differentiation model results (greenand purple circles). The dashed line represents the average value forperidotites (Le Roux et al., 2010). Little fractionation is expectedbetween Zn and Fe during magmatic differentiation. However,strong fractionation can occur in exsolved aqueous fluids (Holland,1972; Khitarov et al., 1982; Urabe, 1987; Simon et al., 2004; Zajaczet al., 2008). The large dispersion in Zn/Fe of highly silicic magmas(i.e., SiO2 > 70 wt.%) is interpreted to reflect Fe–Zn mobilization aschloride complexes during fluid exsolution. Trajectories of Zn/Fevs. SiO2 wt.% during magma differentiation were calculated usingRhyolite Melts (1300–950 �C; 5 kbars) for fractional crystallizationof a hydrous (1.36 wt.% H2O; green circles) and anhydrous dioriticmelt (purple circles) with an initial Zn concentration of 50 ppm.(For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 259

are reported in Table 2. Expectedly, the data follow mass-dependent fractionation, d25Mg = (+0.530 ± 0.146) �d26Mg. The average d26Mg compositions for A-, S- and I-types are �0.094 ± 0.065&, �0.320 ± 0.041&, and�0.331 ± 0.065&, respectively (2SD). They are indistin-guishable from MORBs (�0.25 ± 0.07&; Teng et al.,2010a), except sample A2 (+0.026 ± 0.092), which is en-riched in the heavier isotopes of Mg relative to MORBs

A B

Fig. 9. Zn vs. SiO2 wt.% for granites from the GEOROC database (A)reliable because the dispersion in the data does not depend on the Zn co

by �+0.2& (Fig. 6). These results agree with those re-ported by Shen et al. (2009), Liu et al. (2010) and Liet al. (2010).

4.4. Uranium isotopic data

The uranium isotopic compositions of 22 granitoids (I-,S-, and A-type) are reported in Table 6 and results for thestandards are reported in Table 2. The typical uncertaintyfor measurements of granitic rocks is �0.04& (95% confi-dence interval). For the most part granitoids show little iso-topic variations with most of the samples between �0.4&

and �0.2& in d238U. I-, S- and A-type granites have differ-ent protoliths and different modes of formation; howeverthere is no significant difference in d238U between thesethree granite types. Individual granitic rocks show varia-tions in d238U. In particular, one I-type rock (I7) has a verylow isotopic composition of �0.510 ± 0.039&. The averageU isotopic composition of all the granitoids is �0.305 ±0.143& (2SD). These isotopic compositions agree well withpreviously documented U isotopic composition of igneousrocks, including basalts (Weyer et al., 2008; Tissot andDauphas, 2012). There is no correlation between U isotopevariations and proxies of felsic magma evolution, such asSiO2 content.

5. DISCUSSION

Previous work has suggested that Fe isotopes fractionateduring partial melting (e.g., Weyer and Ionov, 2007; Dau-phas et al., 2009), fractional crystallization (e.g., Schoen-berg and von Blanckenburg, 2006; Teng et al., 2008;Schoenberg et al., 2009), fluid exsolution (Poitrasson andFreydier, 2005; Heimann et al., 2008), and also as a resultof Soret (thermal) diffusion in plutons (Lundstrom, 2009).Here we discuss the possible origins of Fe isotopic varia-tions documented in felsic rocks and how our results canshed new light on this issue.

The fractional crystallization model has difficultiesexplaining why felsic rocks become heavy when a significantFe-bearing phase crystallizing is an oxide (magnetite or

and Zn/Fe ratio vs. Zn (B). The GEOROC Zn data is consideredncentration.

1.0-2.0-3.0-4.0-6.0- 5.0-

B

δ56Fe

(‰)

δ26Mg (‰)

Richter et al. (2009)and Lacks et al. (2012)I-typeS-typeA-type

A

Lacks et al. (2012)

0.0

0.1

0.2

0.3

0.4δ

56Fe (‰)

δ 238U (‰)

0.0

0.1

0.2

0.3

0.4

-1 -0.5 0 0.5

Fig. 10. Expected isotopic fractionation of d56Fe–d26Mg (A), and d56Fe–d238U (B) produced by thermal (Soret) diffusion in a silicate melt(dashed lines; Richter et al., 2009 and Lacks et al., 2012). Soret diffusion has been proposed as a possible mechanism for fractionating ironisotopes in felsic rocks (Lundstrom, 2009). The data do not follow the correlations expected for Soret diffusion.

260 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

ilmenite). Indeed, magnetite and ilmenite are expected tohave heavy rather than light Fe isotopic compositions rela-tive to silicate phases (Polyakov et al., 2007; Shahar et al.,2008; Craddock et al., 2010), which should drive the systemtowards lower d56Fe values, opposite to what is observed.However, in order to definitely rule out this model, the re-duced partition function ratios for iron isotopes in silicatemagmas remain to be determined. The Fe isotopic compo-sition of most of the granitic rocks measured in this studydid not show significant deviation from the igneous average(+0.09&; Beard et al., 2003) (relative to IRMM-014), ex-cept for the most felsic granites (>70 wt.% SiO2; Fig. 2).The d56Fe values of I- and S- type granites overlap, so theredox state of the source rock cannot be the main controlon iron isotopic variations in these samples.

5.1. Iron isotopic fractionation during partial melting

Migmatites are important samples for addressing thequestion of Fe isotopic fractionation during partial meltingof granitic source material because they represent melts ar-rested in the process of migrating from their source.Although leucosomes form by partial melting, their Fe iso-topic compositions can also be affected by fractional crys-tallization of the partial melts. Leucosomes measured inthis study have d56Fe values heavier by �+0.2& relativeto schists, which are most representative of migmatite pro-toliths, and also relative to the average composition of bulkgranitic rocks. In terms of mass-balance, most of the iron isexpected to stay in melanosomes, which are the residues ofpartial melting and are rich in mafic minerals. Melano-somes are, therefore, expected to have an isotopic composi-tion that is similar to that of the bulk protolith. Allmelanosomes have lighter Fe isotopic compositions thantheir leucosome complements (Fig. 2). Of the five leuco-some–melanosome pairs that were analyzed, at least twomelanosomes have Fe isotopic compositions identical tothat of schists, whereas two others have heavy d56Fe valuesby up to +0.2&. This enrichment is difficult to explain byclosed system behavior. Instead, the broadly positive corre-lation between the Fe isotopic compositions of leucosome–

melanosome pairs (Fig. 2) suggests that some of the mela-nosomes have equilibrated with isotopically evolved leuco-somes. Re-equilibration has been shown to also affect themajor and trace element compositions of leucosomes(Fourcade et al., 1992; Johannes et al., 1995). Despite thispossible overprint, the Fe isotopic data preserve evidencethat partial melting of crustal lithologies yields melts withfractionated Fe isotopic compositions, consistent with theinterpretations of Weyer and Ionov (2007) and Dauphaset al. (2009). A remaining issue is the extent to which frac-tional crystallization and reequilibration of the leucosomesduring cooling affected iron isotopes.

5.2. Iron isotopic fractionation during fluid exsolution and

fractional crystallization

Pegmatites are evolved melts that may be particularlysensitive to fluid exsolution. For example, Shearer et al.(1986) documented extensive interaction between pegma-tite-derived fluids and wallrock in the Black Hills, SouthDakota. The d56Fe values of pegmatites are similar or hea-vier than the igneous average (except for sample P389726),indicating that pegmatites are not the predominant reser-voir for isotopically light Fe. Examining the fractionationbetween zinc and iron provides a method of further evalu-ating the effect of fluid exsolution on Fe isotopes. Zinc is adivalent element that is similar in ionic radius to Fe2+

(0.74 A) and hence also similar in behavior. We measuredthe zinc isotopic compositions of many of our samplesand found in pegmatites and some granites that highd56Fe were accompanied by high d66Zn values (Fig. 5),implying that fluid exsolution is indeed an important sourceof Fe isotopic variations in felsic rocks as suggested by Poi-trasson and Freydier (2005) and Heimann et al. (2008).However, the d56Fe values for most granitoids and in par-ticular those of A-type, which formed under relatively dryconditions (Collins et al., 1982), do not correlate with theirZn isotopic compositions. For these samples that do notshow fractionated d66Zn values, the most likely cause ofiron isotopic fractionation is fractional crystallization. Asimilar conclusion was reached by Sossi et al. (2012) based

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 261

on the argument that A-type granites were unlikely to haveexperienced significant iron mobilization by fluid exsolu-tion, yet these granites often have high d56Fe values. Ironisotopic fractionation during magma differentiation is aprocess that could be tested by modeling. Unfortunately,most equilibrium iron isotopic fractionation factors be-tween crystallizing minerals and melts are unknown, whichshould be the focus of future studies.

Mafic magmas show limited fractionation of Zn from Fe(Le Roux et al., 2010; Lee et al., 2010). Felsic magma differ-entiation is also not expected to induce large fractionationof Zn/Fe as the minerals involved have similar mineral/meltdistribution coefficients (Fig. 7). The average DZn/DFe ra-tios (where D, the distribution coefficient between the solidand the liquid, is calculated as the ratio of the concentrationof the element in the solid and in the liquid, Csolid/Cliquid)for the minerals involved are �0.5 for pyroxene, �0.3 foramphibole, �0.6 for biotite, �0.4 for olivine, �0.5 for mag-netite, �0.2 for ilmenite, and �1.0 for feldspar (Ewart andGriffin, 1994). All DZn/DFe ratios (except for feldspar) arelower than one, meaning that the Zn/Fe ratio of the meltshould increase during felsic magma differentiation. Onthe other hand, Zn and Fe can be significantly fractionatedby complexation in chlorine-bearing magmatic fluids (Hol-land, 1972; Khitarov et al., 1982; Urabe, 1987; Simon et al.,2004; Zajacz et al., 2008). Fig. 8 shows a compilation of Zn/Fe ratios in granitoids from the GEOROC database (http://georoc.mpch-mainz.gwdg.de/georoc/). As discussed, littlefractionation of Zn/Fe ratios is expected during graniticmagma evolution and only a small increase in Zn/Fe vs.SiO2 is observed in the data. However, a notable featureis the large dispersion in Zn/Fe ratio of high-SiO2 granites,which spans two orders of magnitude for the most silicicsamples (>70 wt.% SiO2).

We modeled fractional crystallization of hydrous(1.36 wt.% H2O) and anhydrous felsic melts to examinewhether magmatic differentiation could explain the largedispersion in Zn/Fe ratios of granites. We used Rhyolite

Melts, a program that enables calculations of fractionalcrystallization of hydrous silicic melts under a range ofpressures and temperatures (Gualda et al., 2012; http://melts.ofm-research.org/). The initial composition of themelt is assumed to be dioritic (�66 wt.% SiO2) with initialZn and Fe contents of 50 ppm and 2.84 wt.%, respectively.The values for DFe for orthopyroxene, clinopyroxene, mag-netite and other oxides were determined by calculating theratio of the concentration of Fe in the minerals to that inthe melt and values for DZn for each mineral were estimatedby multiplying the DFe values by published DZn/DFe ratios(Ewart and Griffin, 1994). The concentration of Fe for themelt during fractional crystallization was computed by theMelts program, whereas changes in the concentration of Znfor the melt were determined by numerically modeling theincremental removal of Zn incorporated in the mineralsthat crystalized from the melt. Model results are plottedin Fig. 8 (green and purple circles) and indicate that frac-tional crystallization can explain the correlation betweenZn/Fe and SiO2 wt.%, but it cannot readily explain thelarge dispersion of Zn/Fe ratios observed for granites. Thisdispersion could potentially result from differences in the

quality of Zn and Fe analyses reported in the GEOROCdatabase. However, plots of Zn vs. SiO2 content (Fig. 9A)and Zn/Fe vs. Zn (Fig. 9B) indicate that the Zn concentra-tion data appear to be reliable. Thus, the dispersion in Zn/Fe ratios is likely produced by mobilization of Zn and pos-sibly Fe in exsolved fluids.

During late stage crystallization of a granitic melt, Fe-chlorides can form and subsequently be lost in exsolved flu-ids (Simon et al., 2004; Zajacz et al., 2008), which couldpossibly leave the remaining silicate melt enriched in heavyiron isotopes. This mechanism has been proposed as amechanism for enrichment of heavy Fe isotopes in evolvedgranitic rocks (Poitrasson and Freydier, 2005; Heimannet al., 2008). Heimann et al. (2008) presented a quantitativeassessment of this scenario based on solubility data andpredicted equilibrium fractionation factors for iron in min-erals and fluids. A significant result of the present study isthat in pegmatites and some granitic rocks, Zn isotopesshow a moderate correlation with Fe isotopes (Fig. 5). Thisobservation, together with the observed scatter in Zn/Fe ra-tio for granitic rocks (Fig. 8), suggests that fluid exsolutionis responsible in part for the isotopic fractionation of ironin felsic rocks. With the fluid exsolution model, a questionis posed regarding the fate of the isotopically light exsolvedreservoir. Significant amounts of iron must be removed toexplain the heavy iron isotopic composition of pegmatitesand some granitic rocks. Such a reservoir has not been con-clusively identified but it is worth noting that the LittleNahanni pegmatites, which are from volatile-rich sources,have d56Fe compositions that are highly variable, includinglight Fe isotopic compositions (�0.07&; Fig. 2). Furtherwork on fluid inclusions is needed to ascertain the natureof the putative complementary reservoir to isotopically hea-vy iron in pegmatites and some granites.

5.3. Demonstration that Soret effect is absent in the studied

granitoids from Mg, U and Fe isotope systematics

Magnesium isotopic compositions of granitic rocks areimportant to compare with Fe isotopic data because Mgisotopes do not fractionate during partial melting or mag-matic differentiation, but are fractionated significantly dur-ing diffusion and weathering (Teng et al., 2007, 2010b;Richter et al., 2008; Shen et al., 2009; Lundstrom, 2009;Liu et al., 2010). The Lachlan Fold Belt granitoids mea-sured in this study are fresh samples and their Mg isotopiccompositions were not affected by weathering. We mea-sured Mg isotopic data for granitic rocks with the highestd56Fe values to determine whether Soret diffusion and/orderivation from sedimentary protoliths are influencing thed56Fe composition of these felsic rocks. Little variation ind26Mg among the different types of granitic rocks wasfound. The one exception is sample A2, which has ad26Mg composition of +0.026 ± 0.092&, significantly hea-vier than MORBs. Sample A2 also has the highest d56Fe of+0.305 ± 0.022&. Heterogeneous d26Mg values for A-typegranites have also been reported by Li et al. (2010). The Mgisotopic variations of A-type granites possibly reflect chem-ical heterogeneities inherited from the source (Shen et al.,2009; Li et al., 2010; Liu et al., 2010). The reason for hea-

262 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

vier Mg and Fe isotopic compositions in this sample is un-clear. Soret (thermal) diffusion has been proposed as amechanism for producing heavy d26Mg and d56Fe composi-tions in plutons (Lundstrom, 2009). Based on the experi-mental results from Richter et al. (2009) and Lacks et al.(2012), Fe and Mg isotopes from a source that has experi-enced Soret diffusion in a thermal gradient should followthe trend illustrated in Fig. 10A; however, the granitic rocksthat were measured for both d56Fe and d26Mg do notclearly follow the expected correlation.

Further evidence that Soret diffusion played no role inthe isotopic fractionation documented here comes from Uand Fe isotope measurements. One of our motivations toinvestigate the U isotopic composition of granitic rockswas to assess whether the U isotopic compositions of zirconscould be related to the nature of the protolith (i.e., sedimen-tary vs. igneous) with potential application to Hadean zir-cons, which provide the only record of Earth’s evolutionduring that eon. We did not find any evidence for that; Uisotope variations in granites are not related to the natureof their protoliths. Another reason for measuring U isotopesis that Soret diffusion could fractionate U isotopes in silicatemelts (Lacks et al., 2012). Such fractionation would producecorrelated Fe–U isotopic variations (the slope betweend56Fe and d238U in laboratory experiments is 3.18). Asshown in Fig. 10B, we do not see any correlation in the sam-ple set studied that would suggest that Soret diffusion waspresent. If anything, the variations would define a trend thatis orthogonal to the trend predicted for Soret diffusion. Weconclude, based on the absence of correlation between Mg,U and Fe isotopic compositions, that the granitoids investi-gated here (I-, S-, and A-type granitoids from Lachlan FoldBelt) were not influenced by Soret effect.

6. CONCLUSIONS

The d56Fe values of granitoids are positively correlatedwith SiO2 wt.%, consistent with previous studies (Poitras-son and Freydier, 2005; Heimann et al., 2008). Severalmechanisms for enriching the melt in heavy Fe isotopeshave been proposed including: partial melting, fractionalcrystallization, fluid exsolution and thermal migration (Sor-et diffusion). This work sheds new light on the mechanismsgoverning the isotopic fractionation of iron and other non-traditional stable isotope systems in the crust:

1. The enriched Fe isotopic compositions of leucosomesprovide evidence that partial melting of the crust canfractionate the isotopes of Fe by at least 0.2& in d56Fe(Fig. 2). However, the extent to which the measured leu-cosomes were affected by fractional crystallization andreequilibration is unknown.

2. The nature and redox state of the protolith do notappear to influence the d56Fe and d238U compositionsof granitic rocks.

3. Poitrasson and Freydier (2005) and Heimann et al.(2008) suggested that exsolved fluids could have removedisotopically light iron and hence explain the heavy ironisotopic composition of some felsic rocks. This ideawas tested by measuring the isotopic variations of an ele-

ment sensitive to fluid exsolution, Zn, in the same sam-ples that were analyzed for iron. Pegmatites and somegranitoids show high d56Fe values that are accompaniedby high d66Zn values (Fig. 5), so fluid exsolution is mostlikely responsible for the iron isotope variations mea-sured in these samples. This is supported by the observa-tion that the Zn/Fe ratio shows increasing scatter inmore silicic granites, which cannot easily be explainedby magma differentiation and could be due to fluid exso-lution instead (Fig. 8).

4. Fluid exsolution alone cannot explain the high d56Fe val-ues measured in all granitoids, as many of these samplesshow no variations in d66Zn. Fractional crystallization isthe most likely cause of iron isotope variations in thesesamples. This is particularly relevant to A-type granites,which presumably formed in relatively dry conditionsand are not expected to have experienced significant fluidexsolution.

5. Finally, d26Mg and d238U results for granitic rocks sug-gest that thermal migration (Soret diffusion), which hasbeen proposed as a mechanism of granitic magma differ-entiation (Lundstrom, 2009), can be ruled out in oursample set because the d26Mg, d238U and d56Fe do notfollow the correlations expected for Soret diffusion(Richter et al., 2009; Lacks et al., 2012) (Fig. 10).

Our results on a limited, albeit representative, sample setsupport fluid exsolution and fractional crystallization as theprimary mechanisms for the d56Fe variations of felsic rocks.Equilibrium fractionation factors between silicate melts andminerals have to be characterized to quantitatively assessthe role of fractional crystallization on iron isotopes ingranitoids.

ACKNOWLEDGEMENTS

We thank Alfred T. Anderson for valuable assistance withmineral separation; Mark S. Ghiorso and Julia E. Hammer forassisting with the Melts calculations; Richard J. Walker forproviding the Black Hills pegmatite samples. We also thankFrancis Albarede and Philippe Telouk for the generous accessto the Nu Plasma mass-spectrometer in Lyon for the Zn isotopicmeasurements. Drafts of this manuscript have benefitted fromthoughtful comments by Alfred T. Anderson and Cin-Ty A.Lee. Comprehensive reviews by three anonymous referees helpedimprove the manuscript. This work was supported by NASA(NNX09AG59G, NNX12AH60G), NSF (EAR-0820807) and aPackard Fellowship to N.D., by NSF (EAR-0838227 and EAR-1056713) and Arkansas Space Grant Consortium (SW19002) toF.Z.T, and by NASA Grants (NNX12AD88G andNNX12AH70G) to F.M.

REFERENCES

Anbar A. D. and Rouxel O. (2007) Metal stable isotopes inpaleoceanography. Annu. Rev. Earth Planet. Sci. 35, 717–746.

Anderson J. L. (1983) Proterozoic anorogenic granite plutonism ofNorth America. In Proterozoic Geology: Selected Papers from

an International Proterozoic Symposium, vol. 161 (eds. J. G.Medaris, Jr., C. W. Byers, D. M. Mickelson and W. C. Shanks).Geological Society of America Memoir, pp. 133–154.

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 263

Beard B. L. and Johnson C. M. (1999) High precision iron isotopemeasurements of terrestrial and lunar materials. Geochim.

Cosmochim. Acta 63, 1653–1660.

Beard B. L. and Johnson C. M. (2004) Inter-mineral Fe isotopevariations in mantle-derived rocks and implications for the Fegeochemical cycle. Geochim. Cosmochim. Acta 68, 4727–4743.

Beard B. L., Johnson C. M., Skulan J. L., Nealson K. H., Cox L.and Sun H. (2003) Application of Fe isotopes to tracing thegeochemical and biological cycling of Fe. Chem. Geol. 195, 87–

117.

Bonin B. (2007) A-type granites and related rocks: evolution of aconcept, problems and prospects. Lithos 97, 1–29.

Brown M. (1994) The generation, segregation, ascent and emplace-ment of granite magma – the migmatite-to-crustally-derived-granite connection in thickened orogens. Earth Sci. Rev. 36, 83–

130.

Brown M. (2002) Retrograde processes in migmatites and granu-lites revisited. J. Metamorphic Geol. 20, 25–40.

Carignan J., Hild P. and Mevelle G., et al. (2001) Routine analysisof trace elements in geological samples using flow injection andlow pressure on-line liquid chromatograph couples to ICP-MS:a study of geochemical reference materials BR, DR-N, UB-N,AN-G and GH. Geostand. Newsl. 25, 187–198.

Cerny P. (1985) Extreme fractionation in rare-element graniticpegmatites: selected examples of data and mechanisms. Can.

Mineral. 23, 381–421.

Cerny P. (1991) Rare-element granite pegmatites: part II. Regionalto global environments and petrogenesis. Geosci. Can. 18, 68–

81.

Chappell B. W. and White A. J. R. (1974) Two contrasting granitetypes. Pacific Geol. 8, 173–174.

Chappell B. W. (1984) Source rocks of I-type and S-type granites inthe Lachlan Fold Belt, Southeastern Australia. Philos. Trans. R.

Soc. London, A 310, 693–707.

Chappell B. W. and White A. J. R. (1992) I- and S-type granites inthe Lachlan Fold Belt, southeastern Australia. Phil. Trans. R.

Soc. Edinb. Earth Sci. 83, 1–26.

Chappell B. W. and White A. J. R. (1993) The Cowra Granodiorite

and Its Enclaves. IAVCEI Canberra Excursion Guide.Chappell B. W. and White A. J. R. (2001) Two contrasting granite

types: 25 years later. Aust. J. Earth Sci. 48, 489–499.

Clemens J. D., Holloway J. R. and White A. J. R. (1986) Origin ofan A-type granite: experimental constraints. Am. Mineral. 71,

317–324.

Collins W. J., Beams S. D., White A. J. R. and Chappell B. W.(1982) Nature and origin of A-type granites with particularreference to Southeastern Australia. Contrib. Mineral. Petrol.

80, 189–200.

Craddock P. R. and Dauphas N. (2010) Iron isotopic compositionsof reference materials, geostandards and chondrites. Geostand.

Geoanal. Res. 35, 101–123.

Craddock P. R., Dauphas N. and Clayton R. N. (2010) Mineral-ogical control on iron isotopic fractionation during lunardifferentiation and magmatism. LPSC 41, #1230.

Creaser R. A., Price R. C. and Wormald R. J. (1991) A-typegranites revisited: assessment of a residual-source model.Geology 19, 163–166.

Dauphas N., Janney P. E., Mendybaev R. A., Wadhwa M., RichterF. M., Davis A. M., van Zuilen M., Hines R. and Foley C. N.(2004) Chromatographic separation and multicollection-ICPMS analysis of iron. Investigating mass-dependent and -independent isotope effects. Anal. Chem. 76, 5855–5863.

Dauphas N. and Rouxel O. (2006) Mass spectrometry and naturalvariations of iron isotopes. Mass Spectrom. Rev. 25, 515–550.

Dauphas N., Craddock P. R., Asimow P. D., Bennet V. C.,Nutman A. P. and Ohnenstetter D. (2009) Iron isotopes may

reveal the redox conditions of mantle melting from Archean topresent. Earth Planet. Sci. Lett. 288, 255–267.

Dauphas N., Teng F.-Z. and Arndt N. (2010) Magnesium and ironisotopes in 2.7 Ga Alexo komatiites: mantle signatures, noevidence for Soret diffusion, and identification of diffusivetransport in zoned olivine. Geochim. Cosmochim. Acta 74,

3274–3291.

Eby G. N. (1990) The A-type granitoids: a review of theiroccurrence and chemical characteristics and speculations ontheir petrogenesis. Lithos 26, 115–134.

Ewart L. and Griffin W. L. (1994) Application of proton-microprobe data to trace-element partitioning in volcanicrocks. Chem. Geol. 117, 251–284.

Flood R. H. and Shaw S. E. (1975) A cordierite-bearing granitesuite from the New England Batholith, Australia. Contrib.

Mineral. Petrol. 52, 157–164.

French B. M. and Eugster H. P. (1965) Experimental control ofoxygen fugacities by graphite–gas equilibriums. J. Geophys.

Res. 70, 1529–1539.

Fourcade S., Martin H. and Dars J. D. (1992) Chemical-exchangein migmatites during cooling. Lithos 28, 43–53.

Groat L. A., Muljia T., Mauthner M. H. F., Ercit T. S., RaudseppM., Gault R. A. and Rollo H. A. (2003) Geology andmineralogy of the Little Nahanni rare-element granitic pegma-tites, Northwest Territories. Can. Mineral. 41, 139–160.

Gualda G. A. R., Ghiorso M. S., Lemons R. V. and Carley T. L.(2012) Rhyolite-MELTS: a modified calibration of MELTSoptimized for silica-rich, fluid-bearing magmatic systems. J.

Petrol. 53, 875–890.

Heimann A., Beard B. L. and Johnson C. M. (2008) The role ofvolatile exsolution and sub-solidus fluid/rock interactions inproducing high 56Fe/54Fe ratios in siliceous igneous rocks.Geochim. Cosmochim. Acta 72, 4379–4396.

Herzog G. F., Moynier F., Albarede F. and Berezhnoy A. A.(2009) Isotopic and elemental abundances of copper and zinc inlunar samples, Zagami, Pele’s hairs, and a terrestrial basalt.Geochim. Cosmochim. Acta 73, 5884–5904.

Hine R., Williams I. S., Chappell B. W. and White A. J. R. (1978)Contrasts between I- and S-type granitoids of the KoscuiskoBatholith. J. Geol. Soc. Aust. 25, 219–234.

Holland H. D. (1972) Granites, solutions, and base metal deposits.Econ. Geol. 67, 281–301.

Huang F., Chakraborty P., Lundstrom C. C., Holmden C.,Glessner J. J. G., Kieffer S. W. and Lesher C. E. (2010) Isotopefractionation in silicate melts by thermal diffusion. Nature 464,

386-U89.

Johnson C. M., Beard B. L., Roden E. E., Newman D. K. andNealson K. H. (2004) Isotopic constraints on biogeochemicalcycling of Fe. Geochem. Non-Traditional Stable Isotopes 55,

359–408.

Johannes W., Holtz F. and Moller P. (1995) REE distribution insome layered migmatites: constraints on their petrogenesis.Lithos 35, 139–152.

Khitarov N. I., Malinin S. D., Lebedev E. B. and Shibieva N. P.(1982) The distribution of Zn, Cu, Pb and Mo between a fluidphase and a silicate melt of granitic composition at hightemperature and pressure. Geochem. Int. 19, 123–136.

King P. I., White A. J. R. and Chappell B. W. (1997) Character-ization and origin of aluminous A type granites of the LachlanFold Belt, Southeastern Australia. J. Petrol. 36, 371–391.

LacksD.J.,GauravGoel,Bopp,IV, C.J.,VanOrmanJ. A.,LesherC.E. and Lundstrom C. C. (2012) Isotope fractionation by thermaldiffusion in silicate melts. Phys. Rev. Lett. 108, 065901, 1-5.

Lee C.-T. A., Luffi P., Le Roux V., Dasgupta R., Albarede F. andLeeman W. P. (2010) The redox state of arc mantle using Zn/Fesystematics. Nature 468, 681–685.

264 M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265

Le Roux V., Lee C.-T. A. and Turner S. J. (2010) Zn/Fesystematics in mafic and ultramafic systems: implications fordetecting major element heterogeneities in the Earth’s mantle.Geochim. Cosmochim. Acta 74, 2779–2796.

Li W. Y., Teng F.-Z., Ke S., Rudnick R. L., Gao S., Wu F. Y. andChappell B. W. (2010) Heterogeneous magnesium isotopiccomposition of the upper continental crust. Geochim. Cosmo-

chim. Acta 74, 6867–6884.

Loiselle M. C. and Wones D. R. (1979) Characteristics and originof anorogenic granites. Geol. Soc. Am. Abstr. Programs 11, 468.

London D. (1996) Granitic pegmatites. Phil. Trans. R. Soc. Edinb.

Earth Sci. 87, 305–319.

London D. (2005) Granitic pegmatites: an assessment of currentconcepts and directions for the future. Lithos 80, 281–303.

Lundstrom C. (2009) Hypothesis for the origin of convergentmargin granitoids and Earth’s continental crust by thermalmigration zone refining. Geochim. Cosmochim. Acta 73, 5709–

5729.

Liu S.-A., Teng F.-Z., He Y., Ke S. and Li S. (2010) Investigation omagnesium isotope fractionation during granite differentiation:implication for Mg isotopic composition of the continentalcrust. Earth Planet. Sci. Lett. 297, 646–654.

Marechal C. N., Telouk P. and Albarede F. (1999) Precise analysisof copper and zinc isotopic composition by plasma-source massspectrometry. Chem. Geol. 156, 251–273.

Martin R. F. (2006) A-type granites of crustal origin ultimatelyresult from open-system fenitization-type reactions in anextensional environment. Lithos 91, 125–136.

Moynier F., Albarede F. and Herzog G. F. (2006) Isotopiccomposition of zinc, copper and iron in lunar samples.Geochim. Cosmochim. Acta 70, 6103–6117.

Nabelek P. I., Russ-Nabelek C. and Denison J. R. (1992) Thegeneration and crystallization conditions of the ProterozoicHarney Peak Leucogranite, Black Hills, South Dakota, USA:Petrologic and geochemical constraints. Contrib. Mineral.

Petrol. 110, 173–191.

Nabelek P. I. (1999) Trace element distribution among rock-forming minerals in Black Hills migmatites, South Dakota: acase for solid-state equilibrium. Am. Mineral. 84, 1256–1269.

Nakano T. and Urabe T. (1989) Calculated compositions of fluidsreleased from a crystallizing granitic melt: importance ofpressure on the genesis of ore forming fluid. Geochem. J. 23,

307–319.

Nesse W. D. (2000) Introduction to Mineralogy. Oxford UniversityPress, New York, USA, p. 415.

Norton J. J. and Redden J. A. (1990) Relations of zoned pegmatitesto other pegmatites, granites, and metamorphic rocks in thesouthern Black Hills, South Dakota. Am. Mineral. 75, 631–655.

Poitrasson F. (2006) On the iron isotope homogeneity level of thecontinental crust. Chem. Geol. 235, 195–200.

Poitrasson F. and Freydier R. (2005) Heavy iron isotope compo-sition of granites determined by high resolution MC-ICP-MS.Chem. Geol. 222, 132–147.

Polyakov V. B., Clayton R. N., Horita J. and Mineev S. D. (2007)Equilibrium iron isotope fractionation factors of minerals:reevaluation from the data of nuclear inelastic resonant X-rayscattering and Mossbauer spectroscopy. Geochim. Cosmochim.

Acta 71, 3800–3846.

Richter F. M., Watson E. B., Mendybaev R. A., Teng F.-Z. andJanney P. E. (2008) Magnesium isotope fractionation in silicatemelts by chemical and thermal diffusion. Geochim. Cosmochim.

Acta 72, 206–220.

Richter F. M., Dauphas N. and Teng F.-Z. (2009) Non-traditionalfractionation of non-traditional isotopes: evaporation, chemicaldiffusion and Soret diffusion. Chem. Geol. 258, 92–103.

Sawyer E. W. (1991) Disequilibrium melting and the rate of melt-residuum separation during migmatization of mafic rocks fromthe Grenville Front Quebec. J. Petrol. 32, 701–738.

Sawyer E. W. (1994) Melt segregation in the continental crust.Geology 22, 1019–1022.

Schoenberg R. and von Blanckenburg F. (2006) Modes ofplanetary-scale Fe isotope fractionation. Earth Planet. Sci.

Lett. 252, 342–359.

Schoenberg R., Marks M. A. W., Schuessler J. A. and vonBlanckenburg F. (2009) Fe isotope systematics of coexistingamphibole and pyroxene in the alkaline igneous rock suite ofthe Ilimaussa Complex, South Greenland. Chem. Geol. 258, 65–

77.

Schuessler J. A., Schoenberg R., Behrens H. and von BlanckenburgF. (2007) The experimental calibration of the iron isotopefractionation factor between pyrrhotite and peralkaline rhyo-litic melt. Geochim. Cosmochim. Acta 71, 417–433.

Schuessler J. A., Schoenberg R. and Sigmarsson O. (2009) Iron andlithium isotope systematic of the Hekla volcano, Iceland –evidence for Fe isotope fractionation during magma differen-tiation. Chem. Geol. 258, 78–91.

Sha L. K. and Chappell B. W. (1999) Apatite chemical composi-tion, determined by electron microprobe and laser-ablationinductively coupled plasma mass spectrometry, as a probe intogranite petrogenesis. Geochim. Cosmochim. Acta 63, 3861–3881.

Shahar A., Young E. D. and Manning C. E. (2008) Equilibriumhigh-temperature Fe isotope fractionation between fayalite andmagnetite: an experimental approach. Earth. Planet. Sci. Lett.

268, 330–338.

Shearer C. K., Papike J. J., Simon S. B. and Laul J. C. (1986)Pegmatite-wallrock interactions, Black Hills, South Dakota:interaction between pegmatite-derived fluids and quartz-micaschist wallrock. Am. Mineral. 71, 518–539.

Shen B., Jacobsen B., Lee C.-T. A., Yin Q. Z. and Morton D.(2009) The Mg isotopic systematic of granitoids in continentalarcs and implications for the role of chemical weathering incrust formation. Proc. Natl. Acad. Sci. 106, 20652–20657.

Shinohara H. (1994) Exsolution of immiscible vapor and liquidphases from a crystallizing silicate melt: implications forchlorine and metal transport. Geochim. Cosmochim. Acta 58,

5215–5221.

Simon A. C., Pettke T., Candela P. A., Piccoli P. M. and HeinrichC. A. (2004) Magnetite solubility and iron transport inmagmatic-hydrothermal environments. Geochim. Cosmochim.

Acta 68, 4905–4914.

Sio K., Dauphas N., Teng F-Z., Helz R. T. and Chaussidon M.(2011) In-situ Fe-Mg isotopic analysis of zoned olivines.Mineral. Mag. 75, 1884 (conference abstract).

Sirbescu M. L. C. and Nabelek P. I. (2003) Crystallizationconditions and evolution of magmatic fluids in the HarneyPeak Granite and associated pegmatites, Black Hills, SouthDakota – evidence from fluid inclusions. Geochim. Cosmochim.

Acta 67, 2443–2465.

Sossi P. A., Foden J. D. and Halverson G. P. (2012) Redox-controlled iron isotope fractionation during magmatic differ-entiation: an example from the Red Hill intrusion. Contrib.

Mineral. Petrol.. http://dx.doi.org/10.1007/s00410-012-0769.Teng F.-Z., McDonough W. F., Rudnick R. L., Walker R. J. and

Sirbescu M. L. C. (2006) Lithium isotopic systematics ofgranites and pegmatites from the Black Hills, South Dakota.Am. Mineral. 91, 1488–1498.

Teng F.-Z., Wadwha M. and Helz R. T. (2007) Investigation ofmagnesium isotope fractionation during basalt differentiation:implications for a chondritic composition of the terrestrialmantle. Earth Planet. Sci. Lett. 261, 84–92.

M. Telus et al. / Geochimica et Cosmochimica Acta 97 (2012) 247–265 265

Teng F.-Z., Dauphas N. and Helz R. T. (2008) Iron isotopefractionation during magmatic differentiation in Kilauea IkiLava Lake. Science 320, 1620–1622.

Teng F.-Z., Li W. Y., Ke S., Marty B., Dauphas N., Huang S., WuF. Y. and Pourmand A. (2010a) Magnesium isotopic compo-sition of the Earth and chondrites. Geochim. Cosmochim. Acta

74, 4150–4166.

Teng F.-Z., Li W. Y., Rudnick R. L. and Gardner L. R. (2010b)Contrasting behavior of lithium and magnesium isotope frac-tionation during continental weathering. Earth Planet. Sci.

Lett. 300, 63–71.

Teng F.-Z., Dauphas N., Huang S. and Helz R. T. (2011)Diffusion-driven kinetic magnesium and iron isotope fraction-ation in Hawaiian basalts and olivines. Earth Planet. Sci. Lett.

301, 317–324.

Tissot F. and Dauphas N. (2011) Development of high precision238U/235U ratio measurements for cosmochemical applications.Lunar Planet. Sci. 42, #1082.

Tissot F. L. H. and Dauphas N. (2012) 238U/235U ratios ofanagrams: angrites and granites. Lunar Planet. Sci. 43, #1981.

Turner S. P., Foden J. D. and Morrison R. S. (1992) Derivation ofsome A-type magmas by fractionation of basaltic magma- anexample from the Padthaway Ridge, South Australia. Lithos

28, 151–179.

Urabe T. (1987) The effect of pressure on the partitioning ratios oflead and zinc vapor and rhyolite melts. Econ. Geol. 82, 1049–

1052.

Walker R.J. (1984) Origin of the Tin Mountain Pegmatites, BlackHills, South Dakota. Ph. D. thesis, State University of NewYork at Stony Brook.

Walker R. J., Hanson G. N., Papike J. J., O’Neil J. R. and Laul J.C. (1986) Internal evolution of the Tin Mountain pegmatite,Black Hills, South Dakota. Am. Mineral. 71, 440–459.

Walker R. J., Hanson G. N. and Papike J. J. (1989) Trace elementconstraints on pegmatite genesis – Tin Mountain Pegmatite,Black Hills, South Dakota. Contrib. Mineral. Petrol. 101, 290–

300.

Weyer S. and Ionov D. A. (2007) Partial melting and meltpercolation in the mantle: The message from Fe isotopes. Earth

Planet. Sci. Lett. 259, 119–133.

Weyer S., Anbar A. D., Brey G. P., Munker C., Mezger K. andWoodland A. B. (2005) Iron isotope fractionation duringplanetary differentiation. Earth Planet. Sci. Lett. 240, 251–264.

Weyer S., Anbar A. D., Gerdes A., Gordon G. W., Algeo T. J. andBoyle E. A. (2008) Natural fractionation of 238U/235U.Geochim. Cosmochim Acta 72, 345–359.

Whalen J. B., Currie K. L. and Chappell B. W. (1987) A-typegranites: geochemical characteristics, discrimination and petro-genesis. Contrib. Mineral. Petrol. 95, 407–419.

Whalen J. B. and Chappell B. W. (1988) Opaque mineralogy andmafic mineral chemistry of I- and S-type granites of the LachlanFold Belt, southeast Australia. Am. Mineral. 73, 281–296.

Williams H. M., McCammon C. A., Peslier A. H., Halliday A. N.,Teutsch N., Levasseur S. and Burg J. P. (2004) Iron isotopefractionation and the oxygen fugacity of the mantle. Science

304, 1656–1659.

Williams H. M., Peslier A. H., McCammon C. A., Halliday A. N.,Levasseur S., Teutsch N. and Burg J. P. (2005) Systematic ironisotope variations in mantle rocks and minerals: the effect ofpartial melting and oxygen fugacity. Earth Planet. Sci. Lett.

235, 435–452.

Williams H. M., Nielsen S. G., Renac C., Griffin W. L., O’Reilly S.Y., McCammon C. A., Pearson N., Viljoen F., Alt J. C. andHalliday A. N. (2009) Fractionation of oxygen and ironisotopes by partial melting processes: implications for theinterpretation of stable isotope signatures in mafic rocks. Earth

Planet. Sci. Lett. 283, 156–166.

Yang W., Teng F.-Z. and Zhang H.-F. (2009) Chondritic magne-sium isotopic composition of the terrestrial mantle: a case studyof peridotites xenoliths from the North China craton. Earth

Planet. Sci. Lett. 288, 475–482.

Zajacz Z., Halter W. E., Pettke T. and Guillong M. (2008)Determination of fluid/melt partition coefficients by LA-ICPMS analysis of co-existing fluid and silicate melt inclusions:controls on element partitioning. Geochim. Cosmochim. Acta

72, 2169–2197.

Associate editor: Richard J. Walker