identification of chemical sedimentary protoliths using

tters 254 (2007) 358–376www.elsevier.com/locate/epsl

Earth and Planetary Science Le

Identification of chemical sedimentary protoliths using iron isotopesin the N3750 Ma Nuvvuagittuq supracrustal belt, Canada

Nicolas Dauphas a,b,⁎, Nicole L. Cates c, Stephen J. Mojzsis c, Vincent Busigny a,b

a Origins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago,5734 South Ellis Avenue, Chicago IL 60637, USA

b Department of Geology, The Field Museum, 1400 South Lake Shore Drive, Chicago IL 60605, USAc Department of Geological Sciences, Center for Astrobiology, University of Colorado, Boulder, Colorado 80309-0399, USA

Received 15 June 2006; received in revised form 31 October 2006; accepted 25 November 2006

Available onlin

Editor: H. Elderfield

e 16 January 2007

Abstract

An Eoarchean supracrustal belt dated at ca. 3750 Ma was recently identified in the Innuksuac Complex, northern Québec(Canada). Rocks from the Nuvvuagittuq locality include mafic and ultramafic amphibolites, quartz–biotite and pelitic schists,orthogneisses, and banded quartz–magnetite–amphibole/pyroxene rocks of probable chemical sedimentary origin. The purportedmetasediments are enriched in the heavy isotopes of Fe by approximately 0.3‰/amu relative to IRMM-014. They also have highFe/Ti ratios, up to 100× that of associated amphibolite units. These signatures demonstrate that quartz–magnetite–amphibole/pyroxene rocks from Nuvvuagittuq are chemical sediments (e.g. banded iron-formations, BIFs) formed by precipitation ofdissolved ferrous iron in a marine setting. All units were metamorphosed to upper amphibolite facies, which partly homogenized Feisotopes. Variable Fe isotope compositions of bulk quartz–magnetite rocks are interpreted to reflect binary mixing between primaryoxides and carbonates. Mixing relationships with major element chemistry (Ca/Fe, Mg/Fe, and Mn/Fe) are used to estimate the Feisotope composition of the primary Fe-oxide phase (0.3 to 0.4‰/amu) and the chemistry of the carbonate (siderite and ankerite).Iron isotopes can thus be used to constrain the primary mineralogy of Fe-rich chemical sedimentary precipitates beforemetamorphism. The possible presence of siderite in the primary mineral assemblage supports deposition under high PCO2. Wedeveloped an isotope distillation model that includes two possible abiotic oxidation paths, homogeneous and heterogeneous. Theisotopic composition of Fe in the precursor phase of magnetite in BIFs can be explained by partial oxidation through oxygenic oranoxygenic photosynthesis of Fe from a hydrothermal source.© 2006 Elsevier B.V. All rights reserved.

Keywords: Archean; Sediment; BIF; Iron; Isotopes; Metamorphism

⁎ Corresponding author. Origins Laboratory, Department of theGeophysical Sciences and Enrico Fermi Institute, The University ofChicago, 5734 South Ellis Avenue, Chicago IL 60637, USA. Tel.: +1773 7022930.

E-mail address: [email protected] (N. Dauphas).

0012-821X/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.epsl.2006.11.042

1. Introduction

No terrestrial rocks formed before ca. 4.03 Ga areknown to have survived subsequent crustal recyclingprocesses [1]. The only direct witnesses of the earliesttimes are tiny zircon grains accumulated in youngerdetrital sediments [2]. In this respect, the large exposures

mailto:[email protected]

http://dx.doi.org/10.1016/j.epsl.2006.11.042

359N. Dauphas et al. / Earth and Planetary Science Letters 254 (2007) 358–376

of Eoarchean (3.6–3.8 Ga) supracrustal rocks pre-served in the North Atlantic province of the NorthAmerican craton (Isua supracrustal belt, ISB; and Akiliaassociation, AA of southern West Greenland; Nulliakassemblage of northeast Labrador, Canada) are impor-tant [3–9]. The protoliths of some of these formed asmarine sediments and convey crucial information on theearly history of our planet. Was the Earth habitable by3.8 Ga? Had life already emerged? If so, what was thenature of the biosphere? How was the chemistry of theatmosphere–ocean system different? What was the fluxof extraterrestrial matter to the Earth? A principal issuewith these samples is that until recently, they werethought to be exclusive to West Greenland and parts ofLabrador, and it was unclear whether they offered anunbiased perspective of the Earth at that time.

Fig. 1. Map of the Ungava Peninsula, Northern Québec. The NuvvuagittuqInukjuak lithotectonic domain of the Northeastern Superior Province (Canadexposures of Eoarchean supracrustal rocks in SW Greenland (Isua supracruassemblage). Map modified from [73,74].

In 2002, David et al. reported the discovery of a newEoarchean volcano–sedimentary (supracrustal) se-quence in northern Québec along the eastern shore ofHudson Bay (Ungava Peninsula, Canada) [10]. TheNuvvuagittuq supracrustal belt (NSB) is located inthe Inukjuak lithotectonic domain of the NortheasternSuperior Province (Minto block, Canadian Shield,Fig. 1). Other supracrustals in the area comprise theInnuksuac Complex. David and coworkers [10,11]determined the age of a felsic unit from the NSB as3825±16 Ma based on U–Pb TIMS geochronology ofzircons, after rejection of an outlier (3805±77 Ma if it isincluded [12]). Zircon morphologies and trace elementgeochemistry were used to suggest the host rock was avolcanic tuff and that the zircons were not xenocrystic[13]. More recently, Cates and Mojzsis [12] reported an

supracrustal belt (NSB, marked with a white star) is located in theian Shield). The inset map also shows (marked with black stars) otherstal belt ISB and Akilia Association AA) and NE Labrador (Nulliak

360 N. Dauphas et al. / Earth and Planetary Science Letters 254 (2007) 358–376

age of 3758 +51/−47 Ma for the same unit based on ionmicroprobe U–Pb zircon geochronology. For compar-ison, in West Greenland the ISB is also 3.7–3.8 Ga [3,4]and some components of the AA are older than 3.83 Ga[4–7]. The mineral phases observed in the NSB indicatethat it was metamorphosed to upper-amphibolite/lower-granulite facies [12,13], similar to or higher than what isdocumented for the ISB (N550 °C 5 kb) [14].

Archean sedimentary rocks show variable Fe isotopecompositions at mineral and bulk sample scales (from−1.75 to +0.75‰/amu relative to the reference materialIRMM-014) [15–19]. This is in contrast with valuesmeasured in modern igneous rocks formed in differenttectonic settings, which cluster narrowly around+0.04‰/amu [20]. The measured variations may bedue to isotope fractionation during chemical precipita-tion in the oceans, biological reworking of Fe aftersediment deposition, and abiotic reactions duringdiagenesis. The banded quartz–magnetite and quartz–amphibole/pyroxene rocks (purported BIFs) from theca. 3.8 Ga ISB and AA supracrustal units in WestGreenland are enriched in the heavy isotopes of Ferelative to surrounding igneous rocks (from +0.1 to+0.5‰/amu) [17]. This feature may be the imprint ofpartial oxidation in the water column, which is knownexperimentally to enrich the oxidized species in theheavy isotopes of Fe [21–24]. Units of likely chemicalsedimentary origin have been identified in the NSB[10–13]. The presence of such rocks offers us theexciting possibility to probe Earth's surface chemistry at3.7–3.8 Ga from a different perspective than thatprovided by the West Greenland samples. The goals ofthis study are to: (i) establish the origin of the bandedquartz±magnetite±amphibole/pyroxene rocks found inthe NSB; (ii) search for possible petrogenetic relation-ships with supracrustal rocks from the coeval WestGreenland units; (iii) evaluate the effect of metamor-phism on Fe isotopes; and (iv) place further constrainson the mechanisms responsible for Fe precipitation inthe Archean oceans.

2. Sample selection and description

Sample selection was guided by detailed (1:50 scale)mapping of a 130×40 m shoreline exposure in the NSB[12]. The documented supracrustal sequence includesmafic and ultramafic amphibolite units, polymict(conglomeratic) quartz–biotite schists, pelitic to psam-mitic schists, orthogneisses, and banded quartz–mag-netite+ ferruginous quartz–amphibole/pyroxene rocks.The supracrustal lithologies have been multiply de-formed in several metamorphic events, and are intruded

by tonalite–trondhjemite–granodiorite (TTG) composi-tion orthogneisses, including 3.76 Ga trondhjemiticgneisses, some with clear cross-cutting relationships tothe paragneisses (see [12] for detailed descriptions ofthese units).

2.1. Mafic and ultramafic units (IN05013, 19, 45, 47)

Mafic amphibolites have tholeiitic compositions withflat chondrite-normalized rare earth element (REE)patterns [12] and radiogenic initial Nd isotope composi-tions indicative of a depleted mantle source [25]. Majormineral phases are amphibole, plagioclase, garnet, andquartz. Textures range from massive to banded.Ultramafic lenses of basaltic komatiite composition areintercalated with the amphibolite units (IN05047).

2.2. Quartz–biotite schists (polymict conglomerate?)(IN05004, 05, 20, 37)

Rocks that contain highly-strained polycrystallinequartz embedded in a biotite–clinozoisite matrix havetrace element geochemistry consistent with a detritalorigin. Due to the high strain, protolith assignment tothis unit is uncertain.

2.3. Pelitic schists (IN05050, 53)

A pelitic–psammitic quartzite unit preserves finelaminations a few millimetres to a few centimetres thick.Pelitic layers are dominantly biotite, with minorclinozoisite, carbonate and disseminated quartz, where-as quartzite layers are dominantly fine-grained quartz inaggregates of grains with (annealed) triple-junctions+minor biotite, clinozoisite and carbonate.

2.4.Quartz–magnetite and ferruginous quartz–amphibole/pyroxene units (IN05007, 08, 09, 10, 48)

The best preserved examples that escaped silicamobility and strain retain bands of magnetite alternatingwith bands of quartz and amphibole which resemblebanded iron-formation. In more metamorphosed sam-ples (IN05048), pyroxenes rather than amphiboles arepresent. REE patterns, including positive Eu anomalies,are consistent with an origin from hydrothermal sedi-ments for this unit [12,26]. Because sample IN05007 isthe best-preserved quartz–magnetite rock analyzed inthis study, it was selected for detailed petrographic andFe isotope characterizations (Figs. 2 and 3). The hand-specimen (approximately 2.8×3.7×1.8 cm in size),consists of parallel, partly anastomosing, bands of

Fig. 2. Petrography and chemistry of silicates and carbonates in banded quartz–magnetite sample IN05007. Top panel: SEM photomicrographshowing textural relationships between ankerite (ank), quartz (qtz), actinolite (act), and cummingtonite (cum). Bottom panel: chemical compositions ofcarbonates and amphiboles plotted in ternary diagrams. The grey areas are the convex hulls enclosing values measured in BIFs from the ISB [38,44].The metamorphic reaction responsible for the observed paragenesis is, primary carbonates+qtz+H2O→cum+act+ank+cal+CO2 (Eq. (4)) [41].


magnetite (1–4 mm in thickness) alternating with bandsof quartz and amphibole. Although the rock has beenstrongly deformed, these layers may represent relictsedimentary banding. Carbonates are also present andcrosscut in some places the layering. Image analysis of a0.5 mm2 area reveals that these carbonates consist of20% ankerite (Ca0.495Mg0.239Fe0.191Mn0.075) and 80%calcite (Ca0.910Mg0.015Fe0.030Mn0.045). The width of thesolvus between these two phases can be used as athermometer [27] and generates a temperature of 380 °C,much lower than inferred peak metamorphic conditionsof N550 °C. This can be explained by equilibration onthe retrograde path or mobilization by fluids. Theamphiboles consist predominantly (N99%) of cumming-tonite (Na2O 0.64, MgO 11.28, Al2O3 0.29, SiO2 53.77,CaO 0.96, MnO 2.30, FeO 30.78 wt.%) with minoractinolite (Na2O 0.55, MgO 12.02, Al2O3 0.99, SiO2

54.50, CaO 11.53, MnO 0.75, FeO 19.61 wt.%).Disseminated pyrite is also present in IN05007. Overall,the petrography and chemistry of this sample (also see

[12]) is very similar to what has been described forbanded iron-formations from the ISB [28].

3. Analytical methods

Large specimens were cut or broken to offer freshsurfaces and minimize the influence of alteration. Aftercleaning with acetone, the bulk samples (typically∼1 g)were reduced to powder in an agate mortar. Sampling ofindividual magnetite bands in IN05007 was done using amicromill apparatus (MicroMill, New Wave Research)equipped with a tungsten carbide mill bit (Brasseler,scriber H1621.11). The section was extensively charac-terized before microsampling using X-ray and BSEimaging on a JEOL JSM-5800LV SEM (Figs. 2 and 3).At each of the 25 sample points, the following sequencewas followed. A drop of MilliQ water was deposited atthe surface of the polished sample. The computer-controlled bit was moved within the water droplet to millan array of holes (∼18), each approximately 100 μm in

Fig. 3. Chemical and isotopic mapping of IN05007. Left panel: False color X-ray mosaic (assembled from ∼4×1000 frames), with Ca in cyan (C),S in magenta (M), Fe in yellow (Y), and Si in black (K). The whole section is 10 cm2. Magnetite is concentrated in parallel layers (yellow). The vein(cyan) that crosscuts the layering in the top-left corner consists of calcite and ankerite. Pyrite (magenta) is disseminated in magnetite layers.Cummingtonite (light dark yellow–grey) and quartz (dark grey) are the most abundant silicate phases. Actinolite is present as tiny domains (green–grey) associated with cummingtonite. Right panel: FFe values (‰/amu relative to IRMM-014) in individual bands of magnetite (the mill locations areshown on the schematic map) see Table 1. Distance is taken from the bottom of the section. The grey band is the 95% confidence interval for a bulksample measurement. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


diameter and 500 μm in depth (assuming a cylindricalgeometry, this corresponds to a total of ∼270 μg Femilled in magnetite). The sample slurry containing 100–500 μg of Fe was transferred in a Teflon beaker using amicropipette. It was then digested following the sameprotocol as for bulk samples described hereafter.

The protocol for sample dissolution, purification, andisotope analysis of Fe has been described elsewhere[29,30], and is only briefly reviewed here. Powderaliquots weighing less than 10 mg were dissolved inTeflon beakers in 1 mL HF, 0.5 mL HNO3, and a fewdrops of HClO4 at∼100 °C for 5–10 h (unless otherwisenoted, all acids listed are concentrated solutions).The solutions were then evaporated to dryness on hotplates under heat lamps and the salts were dissolved in0.25 mL HNO3, 0.75 mL HCl, and a few drops ofHClO4, evaporated, dissolved in 0.5 mL HNO3, 1 mLHCl, and a few drops of HClO4, evaporated again, andfinally dissolved in 0.5 mL HCl 6 M. At this stage, the

samples were ready for loading on anion exchange resin(typically 50–500 μg Fe).

Iron was separated from matrix elements and directisobars (54Cr and 58Ni can interfere on 54Fe and 58Fe,respectively) by anion exchange chromatography. Afterloading of sample solutions onto Bio-Rad Poly-Prepcolumns filled with 1 mL of AG1-X8 200–400 meshresin, matrix elements were eluted with 8 mL of HCl 6M,and Fe was recovered using 9 mL of HCl 0.4 M. Thissequence was repeated. The yield is close to 100% and noisotope fractionation is introduced by the chemistry [29].The blank of the whole procedure is b40 ng Fe (b0.1% ofthe amount of Fe loaded on the column).

Iron isotopic compositions were measured on anIsoprobe (Micromass) multi-collector inductively cou-pled plasmamass spectrometer (MC-ICPMS) installed atthe Field Museum. The samples (2 ppm Fe in HNO3

0.45 M) were introduced into the mass spectrometerusing a Cetac Aridus desolvating system. Instrumental


mass fractionation (∼50‰/amu) was corrected for usingstandard bracketing and time interpolation with thereference material IRMM-014 (5.845% 54Fe, 91.754%56Fe, 2.1191% 57Fe, and 0.2819% 58Fe) [31]. Isotopiccompositions are calculated using the δ notation, averagedover 2–8 replicate analyses of the same solution,

di=jFe ¼ ½ðiFe=jFeÞSample=ðiFe=jFeÞIRMM�014−1��103:

ð1ÞBecause different laboratories measure different iso-

tope ratios, we prefer to use theFFe notation [30], which isexpressed in permil per atomic mass unit (‰/amu)deviation relative to the composition of the standard,

FFe ¼ di=jFe=ði−jÞ: ð2Þ

If isotope ratios do not depart significantly fromlinear mass fractionation over the range of variationsinvestigated, then the FFe notation should give identicalvalues for any pair of isotopes considered (in thisstudy 54Fe and 56Fe). Error bars are calculated based onmultiple standard analyses and are reported as 95%confidence intervals.

At the beginning of each session of analyses, severalgeostandards and igneous rocks were measured to testthe stability of the instrumental mass bias and confirmthat there had been no problem during sample digestionand chromatography (Table 1). The samples T4D2#1(basalt from Loihi, Hawaii), Payun (pseudomorph ofhematite after magnetite from Payun Matra volcano,Argentina), and Naoned#1 (Eclogite from St-Philbert-de-Bouaine, France) have Fe isotope compositions(0.041±0.041, −0.027±0.032, and 0.045±0.088‰/amu, respectively) identical to earlier results reportedfor these samples (0.072±0.051‰/amu for T4D2#1[17] and 0.004±0.030‰/amu for Payun [32]) or forigneous rocks (0.039±0.008‰/amu [20,30]). Eightreplicate analyses (including dissolution and chroma-tography) of the geostandard IF-G (a 3.7–3.8 Ga BIFfrom the ISB [33]), give a weighted average of 0.288±0.019‰/amu, indistinguishable from the recommendedvalue of 0.316±0.010‰/amu [30]. The external repro-ducibility of these 8 replicate measurements is 0.09‰/amu (2σ), which is in good agreement with internalprecisions obtained for individual analyses (from 0.04 to0.10‰/amu).

4. Results

Mafic and ultramafic samples from the Nuvvuagittuqsupracrustal belt (n=4) have Fe isotopic compositions

very close to IRMM-014 (1/σ2 weighted average−0.030±0.018‰/amu or −0.006±0.022‰/amu ifIN05045, a banded amphibolite with a low Fe isotopiccomposition, is excluded, Table 1). This is similar tovalues reported for ≥3.83 Ga amphibolite rocks [7]from the Akilia association (+0.011±0.010‰/amu)[17] and carbonaceous chondrites (−0.012±0.010‰/amu, Table 1) [30]. Beard et al. [20] showed that modern(b0.1 Ga) igneous rocks sampled in a variety of tectonicsettings have almost homogeneous Fe isotopic compo-sitions, around +0.04‰/amu. This is slightly heavierthan the values found for Eoarchean samples. Variationsof −0.15 to +0.20‰/amu have been documented incarbonatites (Dauphas et al., in prep) and granitoids[34]. The values measured in mafic and ultramafic rocksfrom NSB fall well within this range. There are fewmeasurements published in the literature with which3.8 Ga mafic and ultramafic rocks from the NSB can becompared. Poitrasson et al. [35] reported a value of+0.019±0.008‰/amu for a 3.5 Ga komatiite fromBarberton (South Africa), which is close to the valuesreported here.

All quartz–biotite schists of possible conglomeratic(detrital) origin (n=4) have indistinguishable Fe isotopiccompositions, averaging −0.026±0.025‰/amu. This isidentical to the signature of igneous rocks from the NSB.

The pelitic schists have light Fe isotopic compositions(n=2, −0.118±0.070 and –0.240±0.070‰/amu) rela-tive to IRMM-014 and igneous rocks. Similar signatureshave been documented for other Archean pelitic rocks.Yamaguchi et al. [16] measured the Fe isotope composi-tions of well-preserved shales and greywackes from drillcores ranging in age from 2.20 to 3.25 Ga. Samples withsmall amounts of organic carbon, carbonates, andsulfides, have Fe isotopic compositions close to thesignature measured in igneous rocks and suspended riverloads. In contrast, most of the samples that containsignificant siderite, organic carbon, and magnetite havenegative isotopic compositions, down to −1‰/amu.Negative FFe values are also found in well-preservedBIFs from the Kaapvaal craton, especially in pyrite andFe-carbonate rich layers (down to −1.20‰/amu [15]).

The banded quartz–magnetite rocks of likely BIFprotolith (n=5) are, for the majority, enriched in theheavy isotope of Fe (around +0.3‰/amu). The onlyexception is IN05048, a highly deformed quartz–pyroxene rock within the finely banded quartz–magne-tite unit, with a FFe of +0.038±0.023‰/amu. Theseheavy Fe values are close to those reported for bulkferruginous quartzites from supracrustals rocks of theAkilia association and Isua (from +0.1 to +0.5‰/amu[17]). TheMean SquareWeighted Deviate (MSWD, also

Table 1Chemical and isotopic compositions of geostandards, meteorites, and supracrustal rocks from the NSB

Type Sample Description Fe(mol.g−1)

δ56Fe(‰)

δ57Fe(‰)

FFe

(‰/amu)Fe/Ti

Ca/Fe

Mg/Fe

Mn/Fe

T4D2#1 Basalt (Loihi, HI) 0.081±0.082 0.059±0.158 0.041±0.041Naoned#1 Eclogite (St-Philbert-

de-Bouaine, France)0.089±0.175 0.206±0.152 0.045±0.088

Payun Hmt, pseudomorphafter mgt (Argentina)

−0.039±0.080 −0.045±0.167 −0.020±0.040

−0.077±0.110 −0.140±0.074 −0.039±0.055Average Payun −0.052±0.065 −0.124±0.068 −0.026±0.032

Vigarano CV3 chondrite 0.034±0.110 −0.060±0.587 0.017±0.055Ornans CO3 chondrite −0.094±0.082 −0.224±0.212 −0.047±0.041Lance CO3 chondrite 0.030±0.157 0.191±0.135 0.015±0.079BCR-2 Basalt (Columbia

River, OR)0.001730 0.022±0.080 0.089±0.167 0.011±0.040 6.133 0.734 0.514 0.0160

IF-G BIF geostandard(Isua, Greenland)

0.469±0.087 0.763±0.211 0.235±0.044

0.648±0.087 0.997±0.177 0.324±0.0440.496±0.194 0.760±0.332 0.248±0.0970.608±0.098 0.897±0.154 0.304±0.0490.577±0.139 0.834±0.268 0.289±0.0700.622±0.083 0.934±0.125 0.311±0.0420.592±0.091 0.816±0.097 0.296±0.0460.379±0.193 0.625±0.256 0.190±0.097

Average IF-G 0.006995 0.576±0.037 0.857±0.057 0.288±0.018 10346 0.040 0.067 0.0008N Québecigneousrocks

IN05019 Massive amphibolite 0.001628 −0.024±0.06 −0.078±0.078 −0.012±0.030 15.695 1.012 1.399 0.0164

IN05047 Ultramatic near theBIFs (mgt-rich)

0.002505 −0.025±0.151 −0.044±0.243 −0.013±0.076 22.431 0.074 2.230 0.0124

IN05013 Amphibolite withinthe BIFs (plg-rich)

0.001618 0.007±0.071 0.029±0.087 0.004±0.036 18.495 0.691 1.490 0.0157

IN05045 Banded amphibolite25 m from the BIFs

0.001155 −0.148±0.06 −0.246±0.078 −0.074±0.030 8.496 0.958 0.527 0.0196

N Québec BIFs IN05048 Coarse Qtz–Px rock 0.000901 0.083±0.06 0.056±0.078 0.042±0.030 720.886 0.358 0.501 0.03460.066±0.071 0.081±0.087 0.033±0.036

Average IN05048 0.076±0.046 0.067±0.058 0.038±0.023IN05007 Relatively well

preserved BIF0.004145 0.436±0.151 0.634±0.243 0.218±0.076 3300.864 0.021 0.125 0.0126

0.507±0.151 0.742±0.243 0.254±0.076Average IN05007 0.472±0.107 0.688±0.172 0.236±0.053

IN05009 Intermediatebetween IN05007and IN05048

0.004328 0.552±0.067 0.758±0.092 0.276±0.034 153.159 0.035 0.198 0.0080

0.654±0.067 0.886±0.092 0.327±0.034Average IN05009 0.603±0.047 0.822±0.065 0.302±0.024

IN05008 Relatively wellpreserved BIF

0.003960 0.572±0.139 0.91±0.229 0.286±0.070 3153.561 0.058 0.167 0.0123

IN05010 Relatively wellpreserved BIF

0.004937 0.749±0.092 1.062±0.145 0.375±0.046 3931.905 0.033 0.135 0.0070

IN1S1 MicrosamplingIN05007 (0.17 cm)

0.492±0.107 0.822±0.222 0.246±0.054


0.556±0.139 0.809±0.232 0.278±0.070


0.436±0.088 0.659±0.182 0.218±0.044


0.588±0.113 0.867±0.160 0.294±0.056


0.616±0.165 0.893±0.264 0.308±0.083


Table 1 (continued)

Type Sample Description Fe(mol.g−1)

δ56Fe(‰)

δ57Fe(‰)

FFe

(‰/amu)Fe/Ti

Ca/Fe

Mg/Fe

Mn/Fe


0.628±0.159 1.082±0.268 0.314±0.080


0.566±0.088 0.934±0.182 0.283±0.044


0.373±0.107 0.606±0.222 0.187±0.054


0.614±0.113 0.890±0.160 0.307±0.056


0.510±0.173 0.790±0.245 0.255±0.087


0.524±0.173 0.787±0.245 0.262±0.087


0.757±0.159 1.037±0.268 0.379±0.080


0.511±0.091 0.669±0.097 0.256±0.045


0.510±0.091 0.797±0.097 0.255±0.045


0.509±0.193 0.830±0.256 0.255±0.097


0.597±0.193 0.787±0.256 0.299±0.097


0.569±0.193 0.741±0.256 0.285±0.097


0.638±0.091 0.940±0.097 0.319±0.045


0.758±0.125 1.123±0.174 0.379±0.063


0.716±0.125 1.073±0.174 0.358±0.063


0.455±0.135 0.720±0.202 0.228±0.068

IN9S4P MicrosamplingIN05007 (1.52 cm)

0.703±0.125 0.993±0.174 0.352±0.063


0.639±0.091 0.976±0.097 0.319±0.045


0.570±0.179 0.883±0.291 0.285±0.090


0.591±0.165 0.917±0.264 0.296±0.083

N Québecconglomerate

IN05004 Qtz–bt polymictconglomerate

0.001750 −0.104±0.092 −0.138±0.145 −0.052±0.046 28.008 0.057 0.836 0.0040


0.002109 −0.013±0.139 0.008±0.229 −0.007±0.070 41.861 0.193 0.779 0.0074


0.001279 −0.061±0.092 −0.17±0.145 −0.031±0.046 18.821 0.888 1.231 0.0144


0.000914 −0.008±0.092 −0.14±0.145 −0.004±0.046 29.248 – 0.881 0.0062

N Québecschists

IN05050 Bt–Qtz schist 0.002370 −0.235±0.139 −0.355±0.229 −0.118±0.070 72.800 0.010 0.870 0.0088

IN05053 Bt–Qtz schist 0.000695 −0.479±0.139 −0.655±0.229 −0.240±0.070 21.360 0.090 1.180 0.0099

Isotopic compositions are expressed relative to IRMM-014 (Eqs. (1) and (2)). The chemical compositions of geostandards are from [17] for Ti in IF-Gand [33] for others. The chemical compositions of supracrustal rocks from the NSB are from [12]. Uncertainties are 95% confidence intervals. Theycorrespond to 2–8 replicate analyses of the same solution and were calculated using student's t-distribution. For the samples that have beenmicromilled from the section of IN05007, the numbers in parentheses are the distances from the bottom (Fig. 3).



known as the χ2 per degree of freedom or reduced χ2)for the microdrill analyses of individual magnetite bandsin sample IN05007 is 2.64. This is outside the 95%confidence interval of 0.52 to 1.64 given by χ2 statistics.At the scale of the section studied (10 cm2), the Feisotopic composition appears to vary more than predictedfrom analytical uncertainty (Table 1). This may be a relictdepositional feature [19] or may indicate the presence ofinter-mineral equilibrium fractionation in BIFs at N550 °C[36] if the micro-drilled magnetite was mixed with otherphases. The average of the 25 microsample analyses(+0.284±0.012‰/amu) is identical within error to theaverage of the 2 bulk measurements performed on thesame sample (+0.236±0.054‰/amu).

The Fe/Ti ratio has proven to be useful for tracing Femobilization in altered basalts [37] and distinguishingbetween metasomatized igneous rocks and rocks ofchemical sedimentary origin such as BIFs [17]. Themafic and ultramafic rocks, quartz–biotite (polymictconglomerate?) schists, and pelitic schists from the NSBhave identical Fe/Ti ratios (18–42) as metaigneous rocksfrom the Akilia association. The only exception isIN05045, a banded amphibolite with a low Fe/Ti ratio of8.50 and low Fe isotopic composition (−0.15‰/amu).For comparison, the bulk silicate Earth (BSE) Fe/Ti ratiois 44.5. All candidate rocks of BIF origin investigated in

Fig. 4. Comparison betweenFFe (‰/amu relative to IRMM-014) and theatomic ratio of Fe to Ti in amphibolites, quartz–biotite conglomeratic (?)schists, psammitic schists, and quartz±magnetite±amphibole/pyroxenerocks from the NSB. The Fe/Ti ratio of the bulk silicate Earth is 44.53[40] and the FFe of modern (b0.1 Ga) magmatic rocks (continentalbasalts, OIBs, and MORBs) is ∼0.05‰/amu [30,54]. The hatched fieldis the convex hull enclosing measurements of BIFs and ferruginousquartz–amphibole/pyroxene rocks from the ISB and AA (SW Green-land) [17]. The heavy line is the trajectory for mixing between IF-G (BIFgeostandard from the ISB) and igneous material.

this study have extremely high Fe/Ti ratios compared tosurrounding igneous rocks (Fig. 4). For some, Ti wasbelow detection level and only a lower limit could bederived for the Fe/Ti ratio. In those cases, the Fe/Ti ratiois at least 3 orders of magnitude higher than the ratiomeasured in mafic and ultramafic units. This is similar towhat has been observed for themetamorphosed chemicalsedimentary precipitates documented from the ISB andAkilia association [6,7,17,38,39]. IN05048, a quartz–pyroxene rock with Fe isotopic composition indistin-guishable to igneous rocks has high Fe/Ti ratio (720.9).

5. Discussion

5.1. Protolith identification

Quartz–biotite schists of possible conglomeratic originhave Fe isotopic compositions that resemble mafic andultramafic rocks from the NSB, but are distinct frommodern basalts (Fig. 4). Cates and Mojzsis [12] showedthat bulk element compositions normalized to NASC(North American Shale Composite) are consistent with adetrital origin for this unit. Modern clastic sedimentsformed in a variety of environments (aerosols, loess, soils,marine sediments, suspended river loads) have Fe isotopiccompositions similar to igneous rocks [20]. Thus, Feisotopes are not useful to distinguish between differentsources of clastic sediments.

Two interpretations on the possible origin of thepelitic schist unit have been proposed in the literature.Nadeau suggested that they might derive from metaso-matic alteration of a massive amphibolite unit [13] whileStevenson and David described them as “metapelites”[25].We find that the FFe value of this unit (−0.2‰/amu)is significantly lower than that of amphibolites (∼0‰/amu) while the Fe/Ti ratio is close to that of surroundingigneous rocks. The later observation suggests that Fe wasnot extensively mobilized and that the low FFe valuesmeasured in these rocks are likely primary depositionalfeatures. Such low values have been interpreted inArchean shales and greywackes to represent the imprintof authigenic phases (carbonates, oxides, and sulfides)carrying isotopically fractionated Fe [16]. Our resultssupport a sedimentary rather than a metasomatic originfor this unit.

Iron isotopes have proven to be very useful todistinguish between rocks of true chemical sedimentaryorigin andmetasomatized igneous rocks in SWGreenland[17]. This is an important step in early Earth studiesbecause all known Eoarchean supracrustal rocks havebeen metamorphosed to at least amphibolite facies andrecognition of their protoliths is rarely straightforward.


Our results show that most banded quartz–magnetiterocks from the NSB have heavy Fe isotopic compositionsand high Fe/Ti (Fig. 4). Many samples fall in the fieldcovered by BIFs from the ISB and Fe-bearing quartz–pyroxene units from the Akilia association [17]. High Fe/Ti in these rocks probably reflects the low solubility of Tiin seawater, a feature that is transferred to the chemicalsediment at precipitation. Furthermore, the highFFe valuesof these rocks may be the imprint of partial Fe oxidation[15,17], which is known to enrich the precipitate in theheavy isotopes [21–24]. These results clearly establish thepresence of rocks of BIF parentage in the NSB. SampleIN05048 (a quartz–pyroxene rock) has an Fe isotopiccomposition (+0.038±0.023‰/amu) similar to igneousrocks. This signature does not reflect simple admixture ofigneousmaterial because the Fe/Ti ratio of this rock is high(721). Fig. 4 shows a simple two-componentmixing curvebetween IF-G, a BIF from Isua with the highest Fe/Ti ratiomeasured thus far (10343) and heavy Fe isotopiccomposition (+0.316‰/amu) [17,18], and igneous mate-rial (Fe/Ti=45 and FFe=+0.04‰/amu) [20,40]. At an Fe/Ti ratio of 721, the FFe value of the mixture would still be+0.300‰/amu. We conclude that the range of Fe isotopiccompositions measured in BIFs from the NSB (from 0 to+0.4‰/amu) is a primary feature of the rock.

5.2. Metamorphic overprint

In this section, we examine the effect of upperamphibolite/lower granulite facies metamorphism onthe Fe isotopic composition of BIFs in more detail.Comparison with the more extensively characterizedISB provides some insights on what may have been themetamorphic reactions involved. Magnetite-bearingquartzite units from the NSB contain amphiboles (andsometimes pyroxenes) that presumably formed byreaction between primary carbonates and quartz follow-ing the general reaction [41],

primary carbonates þ quartzþ water→secondary carbonatesþ amphiboles=pyroxenes þ carbon dioxide: ð3ÞThe primary carbonates may have had different

origins. They could have been deposited as primarychemical sediments together with Fe-hydroxides, as isobserved in younger, better preserved BIFs such as thoseof the Hamersley Range (Western Australia) or theTransvaal Supergroup (South Africa) [15,41]. They couldhave been introduced in the rock through carbonatemetasomatism as has been extensively documented in theISB [42–44]. It is also possible that they formed during

diagenesis by microbial dissimilatory iron reduction ofhydrous ferric oxide [19,45]. Carbonates were found andcharacterized in IN05007 (Figs. 2 and 3) but given thedegree of metamorphism of the rock, it is impossible todistinguish between sedimentary, diagenetic, or metaso-matic origins based solely on petrography or chemistry.

With amphiboles and pyroxenes derived from reactionbetween carbonates and quartz, magnetite (Fe3O4) is amajor carrier of Fe in quartzitic supracrustals (includingBIFs) from the NSB (Figs. 2 and 3). Magnetite is not anauthigenic phase but was probably formed diageneticallyshortly after deposition from the transformation of ferro–ferric hydroxide precursors [41,46]. After formation,magnetite is not involved in any metamorphic reaction[19,41]. There is little doubt that magnetite was ultimatelyderived from chemical precipitation in the water columnand its Fe isotope compositionmay help us understand themechanism that causedFe oxidation andBIF precipitationin the Eoarchean oceans. The FFe values of individualphases were presumably equilibrated at some scale andone cannot rely on single-phase measurements to infertheir compositions before metamorphism.

Two approaches can be used to estimate the initialcomposition ofmagnetite. In the element map of IN05007(Fig. 3, 10.00 cm2), the surface area of magnetite is1.38 cm2 and that of amphibole (mainly cummingtonite)is 2.15 cm2 (quartz and other phases occupy 6.34 cm2 and0.14 cm2, respectively). Assuming that the surfacedistribution is representative of the volume distribution,this analysis shows that approximately 74% of Fe is inmagnetite and 26% is in cummingtonite (using densitiesof 5.1 and 3.4 g/cm3 and Fe concentrations of 0.724 and0.246 g/g formagnetite and cummingtonite, respectively).Dideriksen et al. [47] andMarkl et al. [48] analyzed the Feisotopic compositions of natural carbonate mineralsformed in a variety of geological settings (magmatic,metamorphic, hydrothermal, and sedimentary). All sam-ples, without exception, have light Fe isotope composi-tions relative to IRMM-014, from −0.6 to 0‰/amu. Thisis similar to the range of FFe values reported forcarbonates in well-preserved BIFs from the KaapvaalCraton (South Africa) and the Biwabik formation(Minnesota), from −0.5 to 0.1‰/amu [15,19]. One canreasonably assume that the carbonates that were involvedin the formation of amphiboles had light Fe isotopiccompositions. If the rock behaved as a closed system forFe, it is straightforward to make the mass balance(0.26⁎FFe carbonate+0.74⁎FFe magnetite=FFe bulk)and calculate that the initial FFe value of Fe-oxides inIN05007 must have been in the range 0.3 to 0.5‰/amu toexplain the present value of 0.24‰/amu (forFFe values ofcarbonates of 0 and −0.5‰/amu, respectively).

Fig. 5. Binarymixing diagrams between hypothetical end-members Fe-oxide (Fe-ox) and carbonate (Carb). The filled symbols are measure-ments of quartz±magnetite±amphibole/pyroxene rocks from theNSB (Table 1) and the empty diamond corresponds to the compositionof IF-G, a BIF geostandard from the ISB [17,33]. Extrapolatingthe trends to Ca/Fe=0 or Mg/Fe=0 gives the composition of the Fe-oxide end-member, 0.3–0.4‰/amu. The equations of the best-fit linesare y=−0.8016x+0.3279 for Ca, y=−0.7285x+0.4112 for Mg, andy=−10.917x+0.4095 for Mn.


Another way to tackle this question is to recognizethat carbonates should leave an imprint on the bulkchemistry of the rock if they were present at the time offormation. The possible cations present in appreciablequantities in the carbonates from rocks of the composi-tion explored in this study are Fe, Mn, Ca, and Mg. InFig. 5, the FFe values of bulk ferruginous quartzites fromthe NSB are plotted as a function of Ca/Fe, Mg/Fe, andMn/Fe ratios. Simple binary mixing is represented asstraight lines in such diagrams. As discussed in Section5.2, the low FFe value of IN05048 (0.038±0.023‰/amu) cannot be attributed to admixture of an igneouscomponent. Instead, this sample has high Ca/Fe, Mg/Fe,and Mn/Fe and its isotopic composition may reflect thecontribution of carbonates with low FFe values (Fig. 5).Note that similar relationships are obtained for ratios oftransition metals with ionic radii similar to Fe (Ni/Fe, Cu/Fe, and Zn/Fe, see source data in [12]). If the mixinginterpretation is correct, then one can extrapolate mixinglines shown in Fig. 5 to Ca/Fe or Mg/Fe equal to 0(magnetite contains insignificant Ca or Mg) and estimatethe FFe value of an Fe-oxide end-member. The result ofthis analysis is that the initial FFe value of the primaryFe-oxide phase must have been around 0.3–0.4‰/amu,in agreement with the value inferred previously (0.3 to0.5‰/amu). The relationship between FFe and Mn/Fe isless useful because the Mn/Fe ratio of the Fe-oxide end-member is not well defined. It is interesting to note thatIF-G, a BIF geostandard from Isua [33] falls close to themixing lines defined by quartz–magnetite rocks from theNSB on all diagrams [17,30,33]. This analysis alsosuggests that the carbonate end-member had light Feisotopic composition, lower than 0‰/amu. Theoretical-ly, it should be possible to follow the same approachusing a trivalent ion that can partition into magnetite (Al,Cr, V) but not in carbonate to derive the isotopiccomposition of the carbonate end-member. However, noclear correlation was found between FFe values and Al/Fe, Cr/Fe, and V/Fe ratios, possibly because theseelements are present at trace levels and their concentra-tions in the source fluid may have been variable. If weassume that the primary carbonates had FFe valuesbetween −0.5 and 0‰/amu, then it is possible tocalculate the Ca/Fe, Mg/Fe, andMn/Fe ratios of this end-member. The composition of the primary carbonate end-member is thus estimated to be between Ca0.2Mg0.3Fe0.5and Ca0.3Mg0.4Fe0.3. It is a virtual component that mustreflect mixing between two or more carbonate phases(e.g., ankerite 0.5×Ca0.5Mg0.3Fe0.2+siderite 0.5×Mg0.3Fe0.7=Ca0.25Mg0.3Fe0.45). In a ternary diagram, thiscomposition falls on a line that joins amphiboles andcarbonates now present in IN05007, which is required by

mass balance if these phases formed from reactionbetween primary carbonates and quartz (Fig. 6). Theanalysis presented here allows us to better define whatmay have been the metamorphic reactions involved inthe formation of the mineral assemblage preserved inIN05007,

Ca0:25Mg0:32Fe0:43CO3

zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{1=2Ca0:25Mg0:35Fe0:15þ1=2Mg0:3Fe0:7

ankeriteþsiderite

þ SiO2quartz

þH2Owater

YCa0:2Mg2:7Fe4:1½Si8O22�ðOHÞ2cummingtonite

þCa1:8Mg2:7Fe2:5½Si8O22�ðOHÞ2actinolite

þCa0:95Mg0:02Fe0:03CO3calcite

þCa0:54Mg0:26Fe0:21CO3ankerite

þ CO2carbon dioxide

: ð4Þ


There are 7 free coefficients and 7 elements and thismetamorphic reaction can be balanced (Fig. 6). Thepossible presence of siderite in the primary mineralassemblage supports deposition under high PCO2

[49,50]. Whether this high PCO2 translates into highPCO2 in the Eoarchean atmosphere or reflects localconditions of precipitation is open to debate [51,52].

Because magnetite does not participate in thesemetamorphic reactions [41], its Fe isotopic compositionmay be less affected than other phases by metamor-phism. This feature was documented during contactmetamorphism in the Biwabik iron-formation (Minne-sota), where primary Fe isotope heterogeneity ofmagnetite was preserved, up to temperatures in excessof 500 °C [19]. The self-diffusion coefficient of Fe inmagnetite at 500 °C depends on the activity of O2 but isgreater than ∼10−16 cm2/s [53]. The correspondingdiffusion length for 1 My ð ffiffiffiffiffiffiffiffiffiffi

4DtÞpis ∼0.1 cm. Thus,

over timescales relevant to regional metamorphism, theFe isotope signature of magnetite in rocks of BIFparentage may have been homogenized at the scale of ahand specimen. The best-preserved quartz–magnetiterocks in the NSB have well-preserved bands of

Fig. 6. Carbonate–quartz metamorphic reaction in the Ca–Mg–Feternary diagram. The measured compositions of amphiboles (cumming-tonite and actinolite, green squares) and carbonates (ankerite and calcite,yellow squares) are from Fig. 2. The possible composition of the primarycarbonate end-member was calculated from Fig. 5 assuming an FFevalue between −0.5 and 0‰/amu [15,47] (heavy yellow line, l1). Itmust also lie along the line that connects the average compositions ofamphiboles and carbonates, the products of the metamorphic reactioncarbonate+qtz+H2O→cum+act+ank+cal+CO2 (solid line l2).The calculated composition of the primary carbonate end-member (Ca0.25Mg0.325Fe0.425) may reflect mixing between ankeriteand siderite (e.g., 1/2Ca0.5Mg0.35Fe0.15+1/2Mg0.3Fe0.7, dashed line l3).(For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

Fig. 7. Possible reaction paths for O2-mediated abiotic oxidation.During homogeneous oxidation, oxidation of Fe(II)aq followed bypolymerisation of Fe(III)aq allows formation of a solid precipitate.During heterogeneous oxidation, part of Fe(II)aq is first adsorbed onthe precipitate to then be oxidized. The second reaction path isautocatalytic in the sense that the rate of oxidation increases with theconcentration of ferrihydrite in suspension. Assuming that Fe(III)aqand Fe(II)ad have very short lifetimes, the system can be modelledusing Rayleigh transfer of Fe isotopes from Fe(II)aq to Fe(III)s (see textand Appendix for details).

magnetite, quartz, and amphibole. Although deformed,these bands most likely reflect original sedimentarystrata. Multiple measurements of IN05007 using amicromill apparatus (Fig. 3) show that magnetite hasalmost homogeneous Fe isotopic composition at thescale of the sample (small variations may be present butit is unclear whether they represent a depositionalfeature or a problem with sampling single phases, seeSection 4). Overall, it seems that in the Nuvvuagittuqsupracrustal samples we studied, precipitated Fe (hydr)oxides had almost constant Fe isotope compositions,around +0.3‰/amu.

5.3. Homogeneous/heterogeneous oxidation, source ofFe and mechanism of BIF precipitation

The aim of this section is to integrate the resultsobtained on the Fe isotopic composition of magnetite in


a feasible scenario for banded iron-formation deposi-tion. As discussed previously, various chemical path-ways may have led to the formation of magnetite[19,41]. Below, we shall make the simple assumptionthat Fe(II)aq was oxidized and precipitated as Fe(III)s.Upon diagenesis, part of Fe(III)s was reduced to formmixed valence magnetite. Beard and Johnson [54] andDauphas and Rouxel [30] developed a two-stagedistillation model to calculate Fe isotope fractionationduring oxidation–precipitation of Fe(II)aq. This modelmay be relevant to Fe(II)aq oxidation at low pH, wheresignificant Fe(III)aq can form. At circum-neutral pH,polymerization of ferric hydroxide occurs on a shorttime-scale so one would not expect large amounts of Fe(III)aq to be present at any time [55].

Walker [56] reviewed the possible conditions in theArchean ocean when BIFs formed. For PCO2 of∼0.1 atm, the pH would have been around 7 and theconcentration of Fe(II)aq around 10−3 mol/L (also see[46]). At these conditions of circum-neutral pH andhigh Fe concentrations, it is well documented that Feoxidation can proceed along two paths [57–59] (Fig. 7).For low concentrations of solids (b3 mg/L), precipita-tion proceeds predominantly through oxidation of Fe(II)aq and subsequent polymerisation of ferric hydroxideto form colloids which eventually precipitate. In thiscase (homogeneous oxidation) the rate is,

d½FeðIIÞaq�dt

¼ −k1½O2�½OH−�2½FeðIIÞaq�: ð5Þ

This formula applies to most natural waters at present[60]. When the concentration of solids in suspensionincreases to above ∼3 mg/L, Tamura et al. [57] showedthat an autocatalytic effect occurs whereby the rate ofoxidation increases with the concentration of Fe(III)hydroxide present. In this case (heterogeneous oxida-tion) the rate is,

d½FeðIIÞaq�dt

¼ −k2½O2�½FeðIIÞaq�½FeðIIIÞs�=½Hþ�: ð6Þ

The two processes can happen concurrently. Atconstant pH and PO2 we can write,

d½FeðIIÞaq�dt

¼ −k½FeðIIÞaq�−kV½FeðIIÞaq�½FeðIIIÞs�: ð7Þ

The ratio k′/k only depends on the nature of thephase precipitated and pH (∝[H+]). It is equal to∼5×103 mol−1 L at pH 7 for an amorphous ferrichydroxide formed by hydrolysis of Fe(III) perchlo-rate [57,58]. The dimensionless parameter (k is in s−1

and k′ in mol−1 s−1) that governs the evolution of thesystem is,

c ¼ k V½FeðIIÞaq�0k

ð8Þ

where [Fe(II)aq]0 is the initial concentration of dissolvedferrous iron. For an initial concentration of 10−3 molL−1, c is equal to 5. This is higher than 1 andheterogeneous oxidation could clearly be relevant toBIF precipitation. In a closed system with only Fe(II)aqpresent at the beginning, the fraction of Fe precipitatedis given by,

FeðIIIÞsFeðIIÞ0aq

¼ eð1þcÞu−1cþ eð1þcÞu ; ð9Þ

where u is the dimensionless variable kt [57,59]. Ironisotopes could be fractionated to different extents alongthe two oxidation paths. Assuming that Fe oxidationresults in Rayleigh-type transfer of isotopes (i.e., afterprecipitation there is no isotopic exchange between theprecipitate and the fluid), then the composition of theprecipitate can be calculated,

F IIIsFe ¼ F IIaq

Fe ð0Þ

þ 1þ c

1−eð1þcÞu ðDIII heIIaq −DIII ho

IIaq Þuþ DIII heIIaq ln

1þ c

cþ eð1þcÞu

� �� ;

ð10Þwhere FFe

IIaq(0) is the initial composition of ferrousiron, ΔIIaq

III he and ΔIIaqIII ho are the isotopic fractionations in

‰/amu between the instantaneous precipitate and thepool of Fe(II)aq for heterogeneous and homogeneousoxidations, respectively (see Appendix for details).

The three processes most commonly advocated forFe oxidation in an anoxic atmosphere are, (i) oxidationfrom O2 generated by oxygenic photosynthesis, (ii)oxidation during anoxygenic photosynthesis where Fe(II)aq is used as an electron donor in place of H2O, and(iii) photo-oxidation of Fe at the surface of oceans byenergetic photons penetrating through an atmospheretransparent to UV radiation. As yet, the effect of photo-oxidation (iii) on the isotopic composition of Fe hasnot been documented. The other processes (i, ii)are known to enrich Fe(III)s in the heavy isotopes ofFe relative to Fe(II)aq. Welch et al. [22] and Balci et al.[24] studied Fe isotope fractionation during oxidationat low pH (b3), which may not be relevant to BIFformation. The isotopic fractionation between Fe(II)aqand Fe(III)aq at room temperature under sterile condi-tions is +1.4‰/amu [22]. Bullen et al. [21] measured a

Fig. 8. Iron isotopic composition of Fe precipitate as a function of thefraction of Fe precipitated (from Eqs. (9) and (10), also see Appendix).The parameters used in the model calculation are for O2-mediatedhomogeneous/heterogeneous oxidation, FFe

IIaq(0) =−0.15‰/amu,c=5, ΔIIaq

III ho=0.5‰/amu, and ΔIIaqIII he=0.2‰/amu and for anoxygenic

photosynthetic oxidation, F FeIIaq(0)=−0.15‰/amu, c=0, ΔIIaq

III ho=0.7‰/amu (see text for details). The composition of the precursor of magnetitein chemical sediments from Nuvvuagittuq (∼0.35‰/amu, Fig. 5) isshown for comparison (grey band).


lower fractionation between Fe(II)aq and Fe(III)s ofaround 0.5‰/amu during precipitation of ferrihydrite atpH around 6, which may partly reflect kinetic isotopefractionation during precipitation of Fe(III)aq into Fe(III)s [54]. These values would be relevant to oxidationfrom O2 generated by oxygenic photosynthesis. Croalet al. [23] measured the Fe isotopic fractionation be-tween Fe(II)aq and Fe(III)s during anoxygenic photo-synthesis at pH∼7 and found values of around 0.7‰/amu. Again, this value may be the superposition ofeffects associated with Fe(II)aq–Fe(III)aq oxidation andFe(III)aq–Fe(III)s precipitation. Precipitation of Fe(III)aqinto hematite is associated with little equilibriumfractionation but with a significant kinetic isotope effectthat enriches the precipitate in the light isotopes [61].This fractionation increases as the rate of precipitationincreases and can reach values as low as −0.5‰/amu.The isotopic fractionation associated with heteroge-neous (O2-mediated) oxidation is unknown. Adsorptionmay play a role in this process, which can potentiallycreate isotopic fractionation, with Fe(II)ad beingenriched in the heavy isotopes of Fe relative to thepool of Fe(II)aq. Icopini et al. [62] argued that thefractionation may be as large as 1.5‰/amu but values ofaround 0.2 to 0.4‰/amu may be more reasonable[63,64]. Multiple lines of evidence suggest that theprimary source of Fe found in BIFs was hydrothermal[12,26,65]. Present hydrothermal fluids sampled atvents along ocean-ridges have negative FFe values,centred on −0.15‰/amu [66–68]. The net isotopicfractionation between Fe-oxide precursors in bandedquartz–magnetite rocks in the NSB (FFe=0.35‰/amu)and the possible hydrothermal source of Fe (FFe=−0.15‰/amu) is therefore ∼0.5‰/amu.

For the purpose of illustration, we shall adopt thefollowing set of parameters to model Fe isotope frac-tionation during BIF precipitation using Eqs. (9) and (10),(i) O2-mediated abiotic oxidation: FFe

IIaq(0)=−0.15‰/amu, c=5, ΔIIaq

III ho=0.5‰/amu, and ΔIIaqIII he=0.2‰/amu.

(ii) Anoxygenic photosynthetic oxidation: standardRayleigh distillation, equivalent to FFe

IIaq(0)=−0.15‰/amu, c=0, ΔIIaq

III ho=0.7‰/amu.In Fig. 8, we report the calculated Fe isotopic com-

position of the precipitate as a function of the fraction ofFe precipitated. As shown, the Fe isotopic compositionof the precursor of magnetite can be explained by bothscenarios (i) and (ii). The fact that precipitated Fe-(hydr)oxide in Isua and Nuvvuagittuq had almost constant Feisotopic composition, around 0.3–0.4‰/amu, suggeststhat the fraction of Fe precipitated remained constantand probably represented a small fraction of the totalinventory of dissolved Fe, presumably less than 10%. To

summarize, the heavy Fe isotopic composition found inmagnetite of quartz–magnetite rocks (banded iron-formations) from the Nuvvuagittuq supracrustal beltcan be explained if Fe was derived from a hydrother-mal source and was only partially oxidized in the watercolumn.

5.4. The Fe isotopic composition of Eoarchean maficand ultramafic magmas

Mantle-derived magmas and peridotites are the onlysamples available for estimating the Fe isotopiccomposition of the Bulk Silicate Earth (BSE). Beard etal. [20] showed that igneous rocks sampled in a varietyof geological settings have almost uniform isotopiccompositions. However, further detailed studies haverevealed the presence of large-scale Fe isotope varia-tions in mantle and magmatic rocks [34,69–72]. Forinstance, peridotites and granitoids have a large range ofFFe values, from −0.2 to +0.2‰/amu [34,71]. Mostmantle peridotites analyzed so far have light Fe isotopiccompositions, down to −0.2‰/amu [70–72]. In somecases, FFe values of bulk peridotites correlate withchemical indicators of depletion and oxidation [71]. Apossible interpretation is that during partial melting, themelt does not have the same isotopic composition as thestarting material because there is mineral/melt isotopefractionation and/or there is inter-mineral isotopefractionation and melting is non-modal. For degrees ofmelting between 0 and 40% of a spinel lherzolitecomposition, Williams et al. [71] showed that the


magmas produced could be fractionated by as much as0.07‰/amu relative to the source and yet show littlevariation from one to another over a large meltinginterval. Thus, modern continental basalts, MORBs, andOIBs could be fractionated relative to the mantle but stillshow almost constant Fe isotopic compositions, around+0.05‰/amu (Fig. 9). Lherzolites may be a better proxyof the composition of the BSE [72]. The 7 bulk rocklherzolite measurements published so far range from 0to +0.03‰/amu, with a weighted average of about+0.02 [17,71,72] (dominated by a very precise mea-surement of a single xenolith from Kilbourne Hole[71]). This is lower than the composition of modernbasalts but even the most pristine peridotites haveexperienced melt depletion/metasomatism and it iscrucial at this stage to extend the database of otherwisewell-characterized fertile lherzolites.

Although this is at the limit of resolution with presentanalytical uncertainties, mafic and ultramafic magmaticrocks from the ca. 3.83 Ga Akilia island [7] and ca.

Fig. 9. Comparison between FFe values (‰/amu relative to IRMM-014) of carbonaceous chondrites (Table 1), Vesta and Mars (HED andSNC achondrite meteorites), N3.5 Ga mafic and ultramafic magmaticrocks (Table 1), lherzolites, modern (b0.1 Ga) continental basalts,OIBs (Table 1), and MORBs (see text and [30] for an extensivereference list). The curves are kernel density estimates with automaticbandwidth selection (all curves have the same surface area) [75,76].

3.75 Ga NSB [12] localities may have light Fe isotopiccompositions (+0.004±0.016 and −0.006±0.022‰/amu respectively, Table 1, Fig. 9) compared to modernbasalts [17]. This could reflect modification of initial Feisotopic compositions after emplacement of the rocks.Indeed, pervasive carbonate and alkali metasomatismhas been documented in SW Greenland [42–44] and itseffect on Fe isotopes has not been explored. However, itis unlikely that the diverse suite of samples analyzed inthis study (Table 1) and in [17] would have experiencedthe same degree of metasomatic alteration and would allshow the same shift in FFe values. This suggests that thelight Fe isotopic compositions measured in Akilia andthe NSB may represent primary magmatic signatures.Poitrasson et al. [35] measured a 3.5 Ga komatiite fromBarberton (geostandard WITS-1) and also found a lowFFe value of 0.019±0.008‰/amu. In contrast, Weyeret al. [72] measured a 2.7 Ga komatiite from Alexo(geostandard KAL-1, Ontario, Canada) and found avalue that is indistinguishable from modern basalts(0.036±0.007‰/amu). The possible change in Feisotopic composition of magmas formed by partialmelting of the mantle (from ∼0‰/amu in the earlyArchean to ∼0.05‰/amu at present) could reflectchanges in the thermal regime of the mantle and theconditions of magma generation (e.g., fraction of partialmelting, temperature, buoyancy). Clearly, the databaseof Fe isotope analyses of Archean magmas must beextended to ascertain whether the secular variationdiscussed here is real and if it is, address its origin.

6. Conclusions

The recently discovered 3.75 Ga Nuvvuagittuqsupracrustal belt provides us with a means to investigatethe surface chemistry of the Eoarchean Earth from theperspective of a “new” terrane contemporaneous withthe Isua locality. The best-preserved banded quartz–magnetite rocks (BIFs) contain alternating bands ofmagnetite and quartz/amphibole (actinolite and cum-mingtonite) with minor sulfide (pyrite) and carbonate(calcite and ankerite). The mineral paragenesis is bestexplained by amphibolite facies metamorphism of a BIFprotolith consisting of magnetite, quartz, and carbonateas major constituents.

The heavy Fe isotopic compositions (around +0.3‰/amu relative to IRMM-014) and high Fe/Ti ratios (up to∼100× the ratio of surrounding igneous rocks)measured in some BIFs demonstrate that these rocksare true chemical sediments precipitated from seawater.These samples can therefore be used in future studies todocument the chemistry of Eoarchean oceans.


We interpret the dispersion in Fe isotope compositionsat a bulk sample scale in banded iron-formations (from 0to +0.3‰/amu) to reflect binary mixing between Fe-oxides and carbonates. By extrapolating mixing lines inCa/Fe vs FFe and Mg/Fe vs FFe space to Ca/Fe=0 andMg/Fe=0, one can estimate the Fe isotopic compositionof the pure Fe-oxide end-member, which is between +0.3to +0.4‰/amu. On the other hand, if it is assumed that theprimary carbonates had Fe isotope compositions between−0.5 and 0‰/amu, which covers the range of valuesmeasured so far in carbonates from a variety of geologicalsettings, one can derive the chemistry of the carbonateend-member. In a Fe–Mg–Ca ternary diagram, thecalculated composition falls along a line connectingcarbonates and amphiboles now present in the rock,which is consistent with derivation of these phases fromreaction between quartz and primary carbonate. Thecomposition of the primary carbonate does not correspondto a real phase but must reflect a mixture, possiblybetween ankerite and siderite. The possible presence ofsiderite supports deposition under high PCO2. Ironisotopes can thus be used as an independent means toinfer the protolith mineralogy of heavily metamorphosedrocks that may otherwise be inaccessible, especially if therocks did not behave as a perfect closed-systems.

A distillation model involving two possible oxidationpaths (homogeneous and heterogeneous) was devel-oped. It is used to demonstrate that the heavy Fe isotopiccomposition inferred for the precursor phase(s) of mag-netite can be explained by partial oxidation of Fe(II)aqderiving from a hydrothermal source through oxygenicor anoxygenic photosynthesis.

Preliminary data suggest that mafic and ultramaficmagmatic rocks from the NSB have lower Fe isotopiccompositions than modern basalts. A more extensivecharacterization of magmas generated during the earlyhistory of the Earth is required to ascertain whether thiseffect is real or not, work that is currently in progress.

As far as Fe isotopes and bulk chemistry areconcerned, rocks of likely BIF protolith from theNuvvuagittuq supracrustal belt share many similaritieswith banded iron-formations from the Isua supracrustalbelt. Based on the results of this study, we find that thesimilarity is such that the two localities may in fact begenetically related and provide us with a markedlyconsistent picture of the chemistry of the early Earth'shydrosphere.

Acknowledgments

We wish to thank M. Wadhwa, P.E. Janney, A.M.Davis, M. van Zuilen, J. Levine, and R. Yokochi, for

the fruitful discussions and help with MC-ICPMSand SEM analyses. Constructive comments by 3anonymous reviewers substantially improved themanuscript and were greatly appreciated. We aregrateful to the Pituvik Landholding Corporation andthe people of Inukjuak, Québec for the logistical helpand J.D. Adam for the field assistance. This workwas supported by the National Aeronautics andSpace Administration through grants NNG06GG75G(to ND), NAG5-13497 and NASA AstrobiologyInstitute (to SJM) and NAG5-13497. Correspondenceand requests for materials should be addressed [email protected].

Appendix A. Homogeneous/heterogeneousoxidation

At neutral pH, ferrous iron in solution can be oxidizedthrough the homogeneous and heterogeneous paths,

dFeIIaq

dt¼ − kFeIIaq

zfflfflffl}|fflfflffl{homogenous

− VkFeIIaqFeIIIszfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflffl{heterogenous

: ðA1Þ

For a system with all iron as FeIIaq at time 0, we have,

dFeIIaq

dt¼ −kFeIIaq−k VFeIIaqðFeIIaq0 −FeIIaqÞ: ðA2Þ

Using the dimensionless variables and parameters,u=kt, f=FeIIaq/Fe0

IIaq, and c=k′Fe0IIaq/k, one can write

Eq. (A2) as,

dfdu

¼ −f −cf ð1−f Þ: ðA3Þ

Integration is straightforward (pose h= f −1 and usethe method of variation of the constant),

f ¼ 1þ c

cþ eð1þcÞu : ðA4Þ

For the total amount of iron precipitated (g=FeIIIs/Fe0

IIaq =1− f ) we have,

g ¼ eð1þcÞu−1cþ eð1þcÞu : ðA5Þ

The expressions had been derived in Tamura et al.[57]. We now turn to isotopes which is a newdevelopment of this work. Let us denote a and b twoisotopes of Fe and ϕa

i→ j the flux of a from reservoir i toj. We have for the pool of ferrous iron in solution,

dFeIIaqa

dt¼ −/IIaqYIII ho

a −/IIaqYIII hea : ðA6Þ

mailto:[email protected]


We note R the ratio of the abundances of isotopes aand b, with Ri the ratio in reservoir i and Ri→j the ratioin the flux from i to j. Eq. (A6) can be written,

dFeIIaqb RIIaq

dt¼ −/IIaqYIII ho

b RIIaqYIII ho−/IIaqYIII heb RIIaqYIII he:

ðA7ÞWe impose Rayleigh conditions between FeIIaq–

FeIII ho and FeIIaq–FeIII he. Mathematically, this meansthat the ratios αIIaq

III ho =RIIaq→ III ho/RIIaq and αIIaqIII he =

RIIaq→ III he/RIIaq must remain constant.After some rearrangement of Eq. (A7), it can be

shown that,

dFeIIaq

dtþ FeIIaq

dlnRIIaq

dt¼ −aIII hoIIaq kFeIIaq−aIII heIIaq k VFeIIaqðFeIIaq0 −FeIIaqÞ: ðA8Þ

In dimensionless coordinates, this takes the form,

df

duþ f

dlnRIIaq

du¼ −aIII hoIIaq f −aIII heIIaq cf ð1−f Þ: ðA9Þ

Isotope variations are in most cases small and we usethe following notation, δi=(Ri/Rstd−1)×103. The δnotation is used here to improve readability. We alsowrite Δi

j=(αij−1)×103. Using these notations and the

fact that δ/103≈0, Eq. (A9) can be rearranged,

ddIIaq

du¼ −103

dlnfdu

−ðDIII hoIIaq þ 103Þ

−cðDIII heIIaq þ 103Þð1−f Þ:

ðA10Þ

This equation can be integrated (inject the relation-ship between dln f/du and f from Eq. (A3) into Eq.(A10)),

dIIaq ¼ dIIaq0 þ ðDIII heIIaq −DIII ho

IIaq Þuþ DIII he

IIaq ln1þ c

cþ eð1þcÞu

� �ðA11Þ

The isotopic composition of the precipitate can easilybe calculated from mass-balance considerations,

dIIaq0 ¼ f dIIaq þ ð1−f ÞdIIIs: ðA12ÞIt follows that,

dIIIs ¼ dIIaq0

þ 1þ c

1−eð1þcÞu ðDIII heIIaq −DIII ho

IIaq Þuþ DIII heIIaq ln

1þ c

cþ eð1þcÞu

� �� :

ðA13Þ

References

[1] S.A. Bowring, I.S. Williams, Priscoan (4.00–4.03 Ga) orthog-neisses from northwestern Canada, Contrib. Mineral. Petrol. 134(1999) 3–16.

[2] D.O. Froude, T.R. Ireland, P.D. Kinny, I.S. Williams, W.Compston, I.R. Williams, J.S. Myers, Ion microprobe identifi-cation of 4100–4200 Myr-old terrestrial zircons, Nature 304(1983) 616–618.

[3] S. Moorbath, R.K. O'Nions, R.J. Pankhurst, Early Archaean agefor the Isua iron formation, West Greenland, Nature 245 (1973)138–139.

[4] A.P. Nutman, V.R. McGregor, C.R.L. Friend, V.C. Bennett, P.D.Kinny, The Itsaq Gneiss Complex of southern West Greenland;the world's most extensive record of early crustal evolution(3900–3600 Ma), Precambrian Res. 78 (1996) 1–39.

[5] A.P. Nutman, S.J. Mojzsis, C.R.L. Friend, Recognition of≥3850 Ma water-lain sediments in West Greenland and theirsignificance for the early Archaean Earth, Geochim. Cosmochim.Acta 61 (1997) 2475–2484.

[6] N. Cates, S.J. Mojzsis, Chemical and isotope evidence forwidespread Eoarchean (3750 Ma) metasedimentary enclaves insouthernWest Greenland, Geochim. Cosmochim. Acta 70 (2006)4229–4275.

[7] C.E. Manning, S.J. Mojzsis, T.M. Harrison, Geology, age andorigin of supracrustal rocks at Akilia, West Greenland, Am. J.Sci. 306 (2006) 303–366.

[8] A.P. Nutman, B. Fryer, D. Bridgwater, The early ArchaeanNulliak (supracrustal) assemblage, northern Labrador, Can. J.Earth Sci. 26 (1989) 2159–2168.

[9] L. Schiøtte, W. Compston, D. Bridgwater, U–Th–Pb ages ofsingle zircons in Archaean supracrustals from the Nain Province,Labrador, Canada, Can. J. Earth Sci. 26 (1989) 2636–2644 (m).

[10] J. David, M. Parent, R. Stevenson, P. Nadeau, L. Godin, Laséquence supracrustale de Porpoise Cove, région d'Inukjuak: unexemple unique de croûte paléo-archéenne (ca. 3.8 Ga) dans laProvince du Supérieur, L'exploration minérale au Québec, notresavoir, vos découvertes, vol. 2002-10, Ministère des RessourcesNaturelles, Québec, DV, 2002, p. 17.

[11] J. David, M. Parent, R. Stevenson, P. Nadeau, L. Godin, ThePorpoise Cove supracrustal sequence, Inukjuak area: a uniqueexample of Paleoarchean crust (ca. 3.8 Ga) in the SuperiorProvince, GAC–MACAnnualMeeting,Vancouver, Canada, 2003.

[12] N.L. Cates, S.J. Mojzsis, Pre-3750 Ma supracrustal rocks fromthe Nuvvuagittuq supracrustal belt, northern Québec. EarthPlanet. Sci. Lett. (in press).

[13] P. Nadeau, Structural investigation of the Porpoise Cove area,Northeastern Superior Province, Northern Québec, M.Sc. thesis,Simon Fraser University (2003).

[14] J.L. Boak, R.F. Dymek, Metamorphism of the ca. 3800 Masupracrustal rocks at Isua, West Greenland: implications for earlyArchaean crustal evolution, Earth Planet. Sci. Lett. 59 (1982)155–176.

[15] C.M. Johnson, B.L. Beard, N.J. Beukes, C. Klein, J.M. O'Leary,Ancient geochemical cycling in the Earth as inferred from Feisotope studies of banded iron formations from the TransvaalCraton, Contrib. Mineral. Petrol. 144 (2003) 523–547.

[16] K.E. Yamaguchi, C.M. Johnson, B.L. Beard, H. Ohmoto,Biogeochemical cycling of iron in the Archean–PaleoproterozoicEarth: constraints from iron isotope variations in sedimentaryrocks from the Kaapvaal and Pilbara Cratons, Chem. Geol. 218(2005) 135–169.


[17] N. Dauphas, M. van Zuilen, M. Wadhwa, A.M. Davis, B. Marty,P.E. Janney, Clues from Fe isotope variations on the originof Early Archean BIFs from Greenland, Science 306 (2004)2077–2080.

[18] O. Rouxel, A. Bekker, K. Edwards, Iron isotope constraints onthe Archean and Paleoproterozoic ocean redox state, Science 307(2005) 1088–1091.

[19] C.D. Frost, F. vonBlanckenburg, R. Schoenberg, B.R. Frost, S.M.Swapp, Preservation of Fe isotope heterogeneities duringdiagenesis andmetamorphism of banded iron-formation, Contrib.Mineral. Petrol. 153 (2007) 211–235.

[20] B.L. Beard, C.M. Johnson, J.L. Skulan, K.H. Nealson, L. Cox, H.Sun, Application of Fe isotopes to tracing the geochemical andbiological cycling of Fe, Chem. Geol. 195 (2003) 87–117.

[21] T.D. Bullen, A.F. White, C.W. Childs, D.V. Vivit, M.S. Schulz,Demonstration of significant abiotic iron isotope fractionation,Geology 29 (2001) 699–702.

[22] S.A. Welch, B.L. Beard, C.M. Johnson, P.S. Braterman, Kineticand equilibrium Fe isotope fractionation between aqueous Fe(II)and Fe(III), Geochim. Cosmochim. Acta 67 (2003) 4231–4250.

[23] L.R. Croal, C.M. Johnson, B.L. Beard, D.K. Newman, Ironisotope fractionation by Fe(II)-oxidizing photoautotrophic bac-teria, Geochim. Cosmochim. Acta 68 (2004) 1227–1242.

[24] N. Balci, T.D. Bullen, K. Witte-Lien, W.C. Shanks, M. Motelica,K.W. Mandernack, Iron isotope fractionation during microbiallystimulated Fe(II) oxidation and Fe(III) precipitation, Geochim.Cosmochim. Acta 70 (2006) 622–639.

[25] R.K. Stevenson, J. David, Geochemistry and isotopic evolution(Nd, Hf) of the 3.825 Ga Porpoise Cove sequence, NortheasternSuperior Province, Québec, GAC–MAC Annual Meeting,Vancouver, Canada, 2003.

[26] W. Ciolkiewicz, J. Larocque, D. Francis, A habitable Earth at3.8 Ga? Annual Undergraduate Research Conference, Faculty ofScience, McGill University.

[27] L.M. Anovitz, E.J. Essene, Phase equilibria in the systemCaCO3–MgCO3–FeCO3, J. Petrol. 28 (1987) 389–414.

[28] R.F. Dymek, C. Klein, Chemistry, petrology, and origin ofbanded iron formation lithologies from the 3800 Ma IsuaSupracrustal Belt, West Greenland, Precambrian Res. 39 (1988)247–302.

[29] N. Dauphas, P.E. Janney, R.A. Mendybaev, M. Wadhwa, F.M.Richter, A.M. Davis, M. van Zuilen, R. Hines, C.N. Foley,Chromatographic separation and multicollection—ICPMS anal-ysis of iron. Investigating mass-dependent and -independentisotope effects, Anal. Chem. 76 (2004) 5855–5863.

[30] N. Dauphas, O. Rouxel, Mass spectrometry and natural varia-tions of iron isotopes, Mass Spectrom. Rev. 25 (2006) 515–550(Erratum 25:831–832).

[31] P.D.P. Taylor, R. Maeck, P. De Bièvre, Determination of theabsolute isotopic composition and atomic weight of a referencesample of natural iron, Int. J. Mass Spectrom. Ion Process. 121(1992) 111–125.

[32] V. Busigny, N. Dauphas, Tracing paleofluid circulations usingiron isotopes: A study of hematite and goethite concretions fromthe Navajo sandstone (Utah, USA). Earth Planet. Sci. Lett.(in press).

[33] K. Govindaraju, 1994 compilation of working values and sampledescription for 383 geostandard, Geostand. Newsl. 18 (1994)1–158.

[34] F. Poitrasson, R. Freydier, Heavy iron isotope composition ofgranites determined by high resolution MC-ICP-MS, Chem.Geol. 222 (2005) 132–147.

[35] F. Poitrasson, A.N. Halliday, D.-C. Lee, S. Levasseur, N.Teutsch, Iron isotope differences between Earth, Mars andVesta as possible records of contrasted accretion mechanisms,Earth Planet. Sci. Lett. 223 (2004) 253–266.

[36] V.B. Polyakov, S.D. Mineev, The use of Mössbauer spectroscopyin stable isotope geochemistry, Geochim. Cosmochim. Acta 64(2000) 849–865.

[37] O. Rouxel, N. Dobbek, J. Ludden, Y. Fouquet, Iron isotopefractionation during oceanic crust alteration, Chem. Geol. 202(2003) 155–182.

[38] R.F. Dymek, C. Klein, Chemistry, petrology, and origin ofbanded iron formation lithologies from the 3800 Ma IsuaSupracrustal Belt, West Greenland, Precambrian Res. 39 (1988)247–302.

[39] V.R. McGregor, B. Mason, Petrogenesis and geochemistry ofmetabasaltic and metasedimentary enclaves in the Amitsoqgneisses, West Greenland, Am. Mineral. 62 (1977) 887–904.

[40] W.F. McDonough, S.-S. Sun, The composition of the Earth,Chem. Geol. 120 (1995) 223–253.

[41] C. Klein, Some Precambrian banded iron-formations (BIFs) fromaround the world: their age, geologic setting, mineralogy,metamorphism, geochemistry, and origin, Am. Mineral. 90(2005) 1473–1499.

[42] M.T. Rosing, N.M. Rose, D. Bridgwater, H.S. Thomsen, Earliestpart of Earth's stratigraphic record: a reappraisal of the N3.7 GaIsua (Greenland) supracrustal sequence, Geology 24 (1996)43–46.

[43] N.M. Rose, M.T. Rosing, D. Bridgwater, The origin ofmetacarbonate rocks in the Archean Isua supracrustal belt,West Greenland, Am. J. Sci. 296 (1996) 1004–1044.

[44] M.A. van Zuilen, A. Lepland, J. Teranes, J. Finarelli, M. Wahlen,G. Arrhenius, Graphite and carbonates in the 3.8 Ga old IsuaSupracrustal Belt, southern West Greenland, Precambrian Res.126 (2003) 331–348.

[45] C.M. Johnson, E.E. Roden, S.A. Welch, B. Beard, Experimentalconstraints on Fe isotope fractionation during magnetite and Fecarbonate formation coupled to dissimilatory hydrous ferric oxidereduction, Geochim. Cosmochim. Acta 69 (2005) 963–993.

[46] W.E. Ewers, Chemical factors in the deposition and diagenesis ofbanded iron-formation, in: A.F. Trendall, R.C. Morris (Eds.),Iron-Formation: Facts and Problems, Elsevier, Amsterdam, 1983,pp. 491–512.

[47] K. Dideriksen, J.A. Baker, S.L.S. Stipp, Iron isotopes in naturalcarbonate minerals determined by MC-ICP-MS with a 58Fe–54Fedouble spike, Geochim. Cosmochim. Acta 70 (2006) 118–132.

[48] G. Markl, F. von Blanckenburg, T. Wagner, Iron isotopefractionation during hydrothermal ore deposition and alteration,Geochim. Cosmochim. Acta 70 (2006) 3011–3030.

[49] R. Rye, P.H. Kuo, H.D. Holland, Atmospheric carbon dioxideconcentrations before 2 billion years ago, Nature 378 (1995)603–605.

[50] H. Ohmoto, Y. Watanabe, K. Kumazawa, Evidence from massivesiderite beds for a CO2-rich atmosphere before∼1.8 billion yearsago, Nature 429 (2004) 395–399.

[51] R.H. Becker, R.N. Clayton, Carbon isotopic evidence for theorigin of a banded iron-formation inWestern Australia, Geochim.Cosmochim. Acta 36 (1972) 577–595.

[52] J.F. Kasting, Archaean atmosphere and climate, Comment on «H. Ohmoto, Y. Watanabe, K. Kumazawa, Evidence from massivesiderite beds for a CO2-rich atmosphere before ∼1.8 billionyears ago. Nature 429 (2004) 395–399 », Nature 432 (2004).doi:10.1038/nature03166.


[53] A. Atkinson, M.L. O'Dwyer, R.I. Taylor, 55Fe diffusion inmagnetite crystals at 500 °C and its relevance to oxidation ofiron, J. Mater. Sci. 18 (1983) 2371–2379.

[54] B.L. Beard, C.M. Johnson, Fe isotope variations in the modernand ancient Earth and other planetary bodies, Rev. Mineral. 55(2004) 319–357.

[55] A. Ninh Pham, A.L. Rose, A.J. Feitz, T.D. Waite, Kinetics of Fe(III) precipitation in aqueous solutions at pH 6.0–9.5 and 25 °C,Geochim. Cosmochim. Acta 70 (2006) 640–650.

[56] J.C.G. Walker, Possible limits on the composition of the Archeanocean, Nature 302 (1983) 518–520.

[57] H. Tamura, K. Goto, M. Nagayama, The effect of ferrichydroxide on the oxygenation of ferrous ions in neutral solutions,Corros. Sci. 16 (1976) 197–207.

[58] H. Tamura, S. Kawamura, M. Hagayama, Acceleration of theoxidation of Fe2+ ions by Fe(III)-oxyhydroxides, Corros. Sci. 20(1980) 963–971.

[59] W. Sung, J.J. Morgan, Kinetics and product of ferrous ironoxygenation in aqueous systems, Environ. Sci. Technol. 14(1980) 561–568.

[60] W. Davison, G. Seed, The kinetics of the oxidation of ferrous ironin synthetic and natural waters, Geochim. Cosmochim. Acta 47(1983) 67–79.

[61] J.L. Skulan, B.L. Beard, C.M. Johnson, Kinetic and equilibriumFe isotope fractionation between aqueous Fe(III) and hematite,Geochim. Cosmochim. Acta 66 (2002) 2995–3015.

[62] G.A. Icopini, A.D. Anbar, S.S. Ruebush, M. Tien, S.L. Brantley,Iron isotope fractionation during microbial reduction of iron: theimportance of adsorption, Geology 32 (2004) 205–208.

[63] N. Teutsch, U. von Gunten, D. Porcelli, O.A. Cirpka, A.N.Halliday, Adsorption as a cause for iron isotope fractionation inreduced groundwater, Geochim. Cosmochim. Acta 69 (2005)4175–4185.

[64] H.A. Crosby, C.M. Johnson, E.E. Roden, B.L. Beard, CoupledFe(II)–Fe(III) electron and atom exchange as a mechanism for Feisotope fractionation during dissimilatory iron oxide reduction,Environ. Sci. Technol. 39 (2005) 6698–6704.

[65] S.B. Jacobsen, M.R. Pimentel-Klose, A Nd isotopic study of theHamersley and Michipicoten banded iron formations: the sourceof REE and Fe in Archean oceans, Earth Planet. Sci. Lett. 87(1988) 29–44.

[66] M. Sharma, M. Polizzotto, A.D. Anbar, Iron isotopes in hotsprings along the Juan de Fuca Ridge, Earth Planet. Sci. Lett. 194(2001) 39–51.

[67] S. Severmann, C.M. Johnson, B.L. Beard, C.R. German, H.N.Edmonds, H. Chiba, D.R.H. Green, The effect of plumeprocesses on the Fe isotope composition of hydrothermallyderived Fe in the deep ocean as inferred from the Rainbow ventsite, Mid-Atlantic Ridge, 36°14′N, Earth Planet. Sci. Lett. 225(2004) 63–76.

[68] B.L. Beard, C.M. Johnson, K.L. von Damm, R.L. Poulson, Ironisotope constraints on Fe cycling and mass balance in oxygenatedEarth oceans, Geology 31 (2003) 629–632.

[69] X.K. Zhu, Y. Guo, R.J.P. Williams, R.K. O'Nions, A. Matthews,N.S. Belshaw, G.W. Canters, E.C. de Waal, U. Weser, B.K.Burgess, B. Salvato, Mass fractionation processes of transitionmetal isotopes, Earth Planet. Sci. Lett. 200 (2002) 47–62.

[70] B.L. Beard, C.M. Johnson, Inter-mineral Fe isotope variations inmantle-derived rocks and implications for the Fe geochemicalcycle, Geochim. Cosmochim. Acta 68 (2004) 47274743.

[71] H.M. Williams, A.H. Peslier, C. McCammon, A.N. Halliday, S.Levasseur, N. Teutsch, J.-P. Burg, Systematic iron isotopevariations in mantle rocks and minerals: the effects of partialmelting and oxygen fugacity, Earth Planet. Sci. Lett. 235 (2005)435–452.

[72] S. Weyer, A.D. Anbar, G.P. Brey, C. Münker, K. Mezger, A.B.Woodland, Iron isotope fractionation during planetary differen-tiation, Earth Planet. Sci. Lett. 240 (2005) 251–264.

[73] K.D. Card, K.H. Poulsen, Geology and mineral deposits of theSuperior Province of the Canadian Shield, in: S.B. Lucas, M.R.St-Onge (Eds.), Geology of the Precambrian Superior andGrenville Provinces and Precambrian Fossils in North America,Geology of Canada Series, vol. 7, Geological Survey of Canada,1998, pp. 13–194.

[74] M. Simard, M. Parent, J. David, K.N.M. Sharma, Géologie de larégion de la rivière Innuksuac (SNRC 34K et 34L), vol. 2002-10,Ministère des Ressources Naturelles, Québec, RG, 2003 43 pp.

[75] S.J. Sheather, M.C. Jones, A reliable data-based bandwidthselection method for kernel density estimation, J. R. Stat. Soc., B53 (1991) 683–690.

[76] A.W. Bowman, A. Azzalini, Applied Smoothing Techniques forData Analysis: the Kernal Approach With S-Plus Illustrations,Oxford University Press, Oxford, 1997.

identification of chemical sedimentary protoliths using

Documents