Available online at www.sciencedirect.

com

ScienceDirect
Geochimica et Cosmochimica Acta 133 (2014) 443–462
www.elsevier.com/locate/gca

Iron isotopes in an Archean ocean analogue

Vincent Busigny a,⇑, Noah J. Planavsky b,c,⇑, Didier Jézéquel a, Sean Crowe d,e,
Pascale Louvat a, Julien Moureau a, Eric Viollier a, Timothy W. Lyons b
a
Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Univ. Paris Diderot, Paris, France
b
University of California, Riverside, Department of Earth Sciences, Riverside, CA, USA
c
Yale University, Department of Geology and Geophysics, New Haven CT, USA
d
University of British Columbia, Department of Microbiology & Immunology, Vancouver, BC, Canada
e
University of British Columbia, Department of Earth, Ocean & Atmospheric Sciences, Vancouver, BC, Canada

Received 26 September 2013; accepted in revised form 4 March 2014; Available online 14 March 2014

Abstract

Iron isotopes have been extensively used to trace the history of microbial metabolisms and the redox evolution of the
oceans. Archean sedimentary rocks display greater variability in iron isotope ratios and more markedly negative values than
those deposited in the Proterozoic and Phanerozoic. This increased variability has been linked to changes in either water col-
umn iron cycling or the extent of benthic microbial iron reduction through time. We tested these contrasting scenarios
through a detailed study of anoxic and ferruginous Lac Pavin (France), which can serve as a modern analogue of the Archean
ocean. A depth-profile in the water column of Lac Pavin shows a remarkable increase in dissolved Fe concentration (0.1–
1200 lM) and d56Fe values (2.14& to +0.31&) across the oxic–anoxic boundary to the lake bottom. The largest Fe isotope
variability is found at the redox boundary and is related to partial oxidation of dissolved ferrous iron, leaving the residual Fe
enriched in light isotopes. The analysis of four sediment cores collected along a lateral profile (one in the oxic layer, one at the
redox boundary, one in the anoxic zone, and one at the bottom of the lake) indicates that bulk sediments, porewaters, and
reactive Fe mostly have d56Fe values near 0.0 ± 0.2&, similar to detrital iron. In contrast, pyrite d56Fe values in sub-chemo-
cline cores (60, 65, and 92 m) are highly variable and show significant deviations from the detrital iron isotope composition
(d56Fepyrite between 1.51& and +0.09&; average 0.93&). Importantly, the pyrite d56Fe values mirror the d56Fe of dis-
solved iron at the redox boundary—where near quantitative sulfate and sulfide drawdown occurs—suggesting limited iron
isotope fractionation during iron sulfide formation. This finding has important implications for the Archean environment.
Specifically, this work suggests that in a ferruginous system, most of the Fe isotope variability observed in sedimentary pyrites
can be tied to water column cycling—foremost to the oxidation of dissolved ferrous iron. This supports previous suggestions
that enhanced iron isotope variability in the Archean may record a unique stage in Earth’s history where partial ferrous iron
oxidation in upwelling water masses was a common process, probably linked to oxygenic or anoxygenic photosynthesis.
Ó 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION
⇑ Corresponding and equal contribution authors. Tel.: +33 1 83
95 74 34 (V. Busigny). Addresses: Yale University, Department of
Changes in the iron cycle through time have played a
Geology and Geophysics, New Haven CT, USA (N.J. Planavsky),
Institut de Physique du Globe de Paris, Sorbonne Paris Cité , Univ.
key role in shaping global biogeochemical evolution (Johnson
Paris Diderot, Paris, France (V. Busigny). et al., 2008b; Poulton and Canfield, 2011). Iron isotopes
E-mail addresses: [email protected] (V. Busigny), noah.planavsky have emerged in the past decade as tracers of iron cycling
@yale.edu (N.J. Planavsky). on the early Earth—paving the way to test long-standing

http://dx.doi.org/10.1016/j.gca.2014.03.004
0016-7037/Ó 2014 Elsevier Ltd. All rights reserved.
444 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

models about the chemical evolution of the ocean (Dauphas explanations for the presence of markedly light d56Fe values
and Rouxel, 2006; Anbar and Rouxel, 2007; Johnson et al., in the Archean sedimentary rock record. The highly nega-
2008b). There are several well-established trends in the glo- tive d56Fe values have been proposed to mark the onset
bal sedimentary iron isotope record across geologic time of extensive benthic dissimilatory iron reduction (Johnson
(Fig. 1). These trends are most prevalent in the pyrite re- et al., 2008b; Heimann et al., 2010). In this case, the iron
cord but are also observed from bulk-rock analyses of or- isotopes would record fundamental changes in marine het-
ganic-rich shales and carbonates (e.g., Rouxel et al., 2005; erotrophic communities through time. Alternatively, the
Johnson et al., 2008b). Archean and Paleoproterozoic pyr- markedly negative d56Fe values and their disappearance
ites, studied by analysis of isolated pyrite crystals or chem- from the sedimentary record over time might indicate a
ical sequential extraction, have d56Fe values that range shift in the chemical composition of the oceans tied to pro-
from 3.51 to +1.19 per mil (&), while modern, Phanero- cesses other than microbial iron reduction. Specifically, the
zoic, and Neoproterozoic sedimentary pyrites show a much negative d56Fe values may instead result from partial oxida-
narrower range in d56Fe values between 1.65& and tion of the dissolved iron pool, producing a dissolved resi-
+0.16& (Rouxel et al., 2005; Fehr et al., 2010). Throughout due depleted in heavy Fe isotopes, which can be transferred
Earth’s history, a larger range of d56Fe values is recorded in to the rock record as pyrite (Rouxel et al., 2005). In this
ancient sediments probed by microanalysis techniques (sec- model, the iron isotope composition of marine sediments
ondary ion mass spectrometry or laser ablation systems; would track the oxidation of the oceans. Lastly, it has also
Steinhoefel et al., 2009, 2010; Yoshiya et al., 2012; Virtasalo been proposed that in iron-rich aqueous systems, pyrite will
et al., 2013). However, these values likely track both small- be isotopically light due to large kinetic isotope fractiona-
scale diagenetic processes and fluctuating water column tions (Guilbaud et al., 2011). The sedimentary iron isotope
chemistry and therefore cannot readily be compared to composition will then record marine dissolved iron avail-
other bulk sedimentary iron isotope records. ability, with markedly negative d56Fe values and greater
The greatest amount of iron isotope variability identified isotope variability reflecting a large dissolved Fe reservoir,
in the sedimentary record (Fig. 1; Rouxel et al., 2005; Johnson dominating over reduced S compounds (Guilbaud et al.,
et al., 2008a) is found in Neoarchean rocks deposited be- 2011). This model is intriguing in that it requires us to re-
tween 2.9 and 2.5 Ga, clearly delineating an enigmatic yet think how ocean chemistry has changed through time given
unique period of iron cycling. There are a few possible that iron-rich marine conditions are now widely envisioned

Fe sulfides
Bulk shales
3
Fe oxides
GOE

BIF & hydrothermal deposits

2 Carbonates

1
δ56Fe, ‰

-1

-2

-3

-4
0 500 1000 1500 2000 2500 3000 3500 4000
Age, Ma
Fig. 1. Iron isotope composition of sedimentary minerals and bulk rocks through time. Data on Fe sulfides, oxides, and carbonates, together
with bulk shales, iron formation and hydrothermal deposits are compiled from the literature. For comparison, the horizontal light-grey zone
represents the range of Fe isotope compositions for detrital sediments and hydrothermal fluids. Earth’s earliest sedimentary record (3.8–
3.0 Ga) is characterized by near crustal or positive d56Fe values, suggesting a biosphere with limited oxidative capacity. After 3.0 Ga but prior
to the rise of atmospheric oxygen (so called Great Oxidation Event, GOE) at 2.4 Ga, there were sedimentary pyrites and shales with
anomalously negative d56Fe values. Data complied for pyrites, Fe oxides and carbonates are from Johnson et al. (2003, 2008a), Matthews
et al. (2004), Rouxel et al. (2005), Archer and Vance (2006), Severmann et al. (2006), Staubwasser et al. (2006), Busigny and Dauphas (2007),
Dauphas et al. (2007a, b), von Blanckenburg et al. (2008), Duan et al. (2010), Craddock and Dauphas (2011), Halverson et al. (2011), Czaja
et al. (2012). Carbonates from high-grade metamorphic rocks from Greenland and Canada were avoided due to their possible metasomatic
origins. Data on bulk shales, banded iron formation and hydrothermal deposits are from Matthews et al. (2004), Yamaguchi et al. (2005),
Staubwasser et al. (2006), Severmann et al. (2006), Frost et al. (2007), Dauphas et al. (2007a, b), Valaas Hyslop et al. (2008), Planavsky et al.
(2009, 2012), Steinhoefel et al. (2009), Tsikos et al. (2010), Duan et al. (2010), Halverson et al. (2011) and Czaja et al. (2012). Iron isotope data
obtained using Ion Microprobe and Laser Ablation MC-ICP-MS analyses were discarded from the present compilation because the large
heterogeneities observed on small-scale (within single crystals) that are not linked to marine biogeochemical iron cycling.
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 445

to have persisted well past the observed time interval of en- 2. SAMPLES AND METHODS
hanced iron isotope variability (e.g., Planavsky et al., 2011;
Poulton and Canfield, 2011). Constraining the origin of 2.1. Overview of Lac Pavin
negative Archean d56Fe values, therefore, will not only im-
prove our understanding of iron isotope systematics but Lac Pavin is a crater lake, within the youngest volcano
will also advance our understanding of Earth’s biogeo- in the French Massif Central. The last volcanic eruption
chemical evolution. and the ensuing lake formation occurred 6900 years ago
Debate on the origin of negative d56Fe values stems in (Juvigné and Gilot, 1986; Chapron et al., 2010). It is located
part from the lack of work in modern aqueous systems 35 km southwest of Clermont-Ferrand, 1197 m above sea
that are analogous to the early oceans. The early oceans level (45°29.740N, 2°53.280E). The lake is a maar, i.e., a
are thought to have had low sulfate concentrations (e.g., crater lake resulting from a phreatomagmatic eruption
Habicht et al., 2002) and a significant dissolved iron reser- (Glangeaud, 1916). It is characterized by a high aspect ratio
voir (Holland, 1984). These characteristics may be essen- (maximum depth/surface0.5 = 0.138), a surface area of
tial in allowing for enhanced iron isotope fractionations 0.445 km2, a diameter of 750 m, and a maximum depth
during microbial iron reduction (Johnson et al., 2008b) of 92 m. Because of this specific lake geometry and wind
or during inorganic pyrite precipitation (Guilbaud et al., protection afforded by the surrounding tuff-ring (50 m
2011). In stark contrast to the increasing number of stud- high), water column mixing is depth-limited. Lac Pavin is
ies on sulfide-rich, iron-poor modern anoxic basins (e.g. meromictic with two stable stratified layers (Fig. 2a). The
Fehr et al., 2008, 2010; Severmann et al., 2008; Staubwas- three factors common to most meromictic settings are pres-
ser et al., 2013), there has been limited work on low ent in this lake: appreciable biological production in the
sulfate, iron-rich modern aqueous systems to test these surface waters, ectogenic controls, and crenogenic inputs
models (Malinovsky et al., 2005; Teutsch et al., 2009; (deep water inflow via springs) (Assayag et al., 2008; Bon-
Song et al., 2011). With this in mind, we have conducted homme et al., 2011). The upper layer, called mixolimnion,
a detailed iron isotope study in ferruginous Lac Pavin in extends from the surface to ca. 60 m depth and is oxygen-
the French Massif Central, an aquatic system that may ated because of seasonal mixing. The deeper layer, called
serve as one of our best modern analogs for the Earth’s monimolimnion, extends from 70 to 92 m depth and never
early ferruginous oceans. mixes with upper oxygenated waters. The monimolimnion
Lac Pavin is permanently redox-stratified, with anoxic is oxygen depleted due to high organic matter load, and
and ferruginous deep waters (from 60 to 92 m depth) has been permanently anoxic over the last 100 years, at
topped by oxic shallow waters (from 0 to 60 m). It contains least (Glangeaud, 1916). An intermediate layer, called mes-
dissolved ferrous Fe (up to 1.2 mM) below the chemocline olimnion, which is characterized by sharp dissolved Fe and
at 60 meters (Michard et al., 1994; Viollier et al., 1995). salinity gradients, separates the mixolimnion and monimo-
Additionally, the lake has remarkably low sulfate concen- limnion. Over the last 40 years, the physicochemical condi-
trations (<20 lM). Lac Pavin, therefore, has conditions tions of the lake and associated sediments have been
similar to those inferred for Archean oceans—that is, high studied extensively, and geochemical cycling of many ele-
dissolved Fe concentrations, a strong vertical gradient in ments is well constrained (for example, C, Fe, P, N, and
dissolved iron concentrations, and low levels of sulfate S). On a decadal time scale, the available data show that
and sulfide (e.g., Sumner, 1997; Beukes and Gutzmer, the deep part of the lake (below 60 m depth) has reached
2008). a steady state (Michard et al., 1994, 2003; Aeschbach-
In the present work, we analyzed Fe concentrations Hertig et al., 1999; Jézéquel et al., 2011). It is thus an ideal
and isotope compositions along a depth-profile in the natural laboratory to study Fe cycling and associated
water column of Lac Pavin. To track the record of water isotope fractionations in a stratified water body.
column Fe cycling and its isotopic effect, we also collected The chemical profiles in the mixolimnion are shaped by
four sediment cores: one from the oxic zone (32 m), two the activity of photosynthetic organisms and associated het-
from below but near the chemocline under anoxic waters erotrophs. Oxygen (O2) is produced by photosynthesis and
(60, 65 m), and one from the anoxic bottom of the lake accumulates in the photic zone (Assayag et al., 2008), while
(92 m). Following Severmann et al. (2006), we determined nutrients such as nitrate (NO 3
3 ) and phosphate (PO4 ) are
iron isotope compositions for the bulk, reactive (labile Fe depleted due to biological assimilation (Fig. 2; Michard
phases such as amorphous Fe-oxides, FeS, and potentially et al., 1994). Diatoms make up the majority of the plank-
some reactive silicate Fe), pyrite, and porewater iron in tonic biomass in the photic zone (Amblard and Bourdier,
each of the cores. Based on dissolved Fe data from the 1990) and draw down dissolved silica levels within the
water column with solid phase and porewater analyses mixolimnion. Dissolved Fe and sulfate (SO2 4 ) concentra-
from sediment cores, we suggest that benthic iron recy- tions in the mixolimnion are a few tens of nM and 15–
cling and kinetic isotope fractionations during pyrite for- 20 lM, respectively (Fig. 2; Michard et al., 1994, 2003).
mation have little to no effect on the sedimentary iron The monimolimnion is devoid of oxygen but contains
isotope record. Instead, our iron isotope work in Lac very high concentrations of dissolved ionic species, as
Pavin provides evidence for iron oxidation in the water shown by the strong increase in specific conductivity with
column controlling the iron isotope composition of under- depth (Fig. 2b). These high concentrations induce high-
lying sediments in the lake and likely in many ancient sys- density anoxic deep water that counter balances the
tems, by analogy. 1 °C temperature increase observed below 60 m depth
446 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

(A) (B) T(°C) ; O2 (mg/L)

0 5 10 15 20
0 0

10 10
O2
20 20 T
30 MIXOLIMNION 30
Depth (m)

(oxic) 40
40

50 50

60 60
MESOLIMNION
70 70
C25
MONIMOLIMNION 80
80
(anoxic)
90 90

0 100 200 300 400 500 600

μS/cm)
Specific conductivity (μ

(C) Fe2+, PO43- (μM) (D) SiO2, NH4+ (μM)

0 400 800 1200 0 400 800 1200
0 0

10 10

20 20

30 30
SO42-
Mo
Depth (m)

40 40

NO3-
50 50

60 60
SiO2
70 70
Fe2+
80 PO4- 80 NH4+

90 90

100 100
0 4 8 12 16 0 1 2 3 4 5
SO4 2- (μM), Mo (nM) -
NO3 (μM)

Fig. 2. Illustration of the physico-chemical stratification in Lac Pavin. (A) Schematic profile of the Lake. (B) Temperature, dissolved oxygen
content and specific conductivity (expressed as C25) determined in July 2007. (C) and (D) Selected dissolved chemical species (data from
Michard et al., 1994; Viollier et al., 2000).

and thus contributes to maintaining water column stratifi- Most major compounds and nearly all measured trace
cation. The main charged dissolved species in the monimo- elements show strong concentration gradients, with high
limnion are bicarbonate (HCO 2+
3 ), ferrous iron (Fe ), concentrations at the sediment–water interface and low
+
ammonium (NH4 ) (with concentrations of up to 4.8, 1.1, concentrations at the oxycline (Michard et al., 1994, 2003;
and 1 mM, respectively). The main uncharged species are Viollier et al., 1995; Takayanagi and Cossa, 1997). These
carbonic acid (H2CO3), methane (CH4), and dissolved silica concentration gradients reflect upward diffusion of dis-
(H4SiO4) (with concentrations of up to 12.1, 4.3 and solved ions sourced from the sediment and consumed at
1.1 mM, respectively) (Fig. 2; Michard et al., 1994). Phos- the redox boundary. This is the case for iron, which is re-
phate is also highly concentrated and reaches up to leased from the sediment in the ferrous form (Fe2+) and
340 lM. Accumulation of the reduced compounds as well precipitates as ferric iron (Fe3+) at the redox interface (Mi-
as carbonate species mainly derives from sub-mixolimnion chard et al., 2003). Because of high stability, the monimo-
organic matter remineralization. Sulfate is reduced to sul- limnion is enriched in dissolved gases such as CH4
fide just below the oxycline (>60 m) with sulfate drawdown produced by microbial methanogenesis (Biderre-Petit
to stable low lM levels (Bura-Nakic et al., 2009). et al., 2011) and CO2 resulting from anaerobic respiration,
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 447

anaerobic CH4 oxidation, and/or a deep spring input related the lake (92 m; 45°29.743N, 2°53.279E) was collected
to volcanic activity (Aeschbach-Hertig et al., 1999; Olive and using an Uwitec gravity corer with a diameter of 90 mm.
Boulègue, 2004; Assayag et al., 2008; Lopes et al., 2011). The total length recovered from this core was 167 cm. In
Microbial density (bacteria and archaea) increases with the present study, we focused only on the first half of this
depth from 50 m to the bottom of the lake (Lehours core, from the top down to 83.5 cm.
et al., 2007)—following the increase in dissolved ion species. Water samples were collected from a sampling platform
Facultative Fe(III) reducing bacteria are likely present, but anchored near the center of the lake (45°29.745N,
well-known obligatory Fe(III) reducers (such as Geobacter) 2°53.274E) using an 1-L syringe sampler. The sampler
have not been identified (e.g., Lehours et al., 2009). was a custom-built automatic polycarbonate syringe (mod-
The sediments deposited in Lac Pavin are dominated by ified from the Model S-1000, HamiltonÒ), equipped with an
diatom silica frustules, organic matter, authigenic iron electronic depth gauge allowing a precision of ±0.2 m
minerals, and detrital components (Meybeck et al., 1975; depth. Waters were filtered to 0.2 lm using Millipore cellu-
Schettler et al., 2007; Cosmidis et al., 2014). SiO2 content lose acetate membranes and were then acidified with a few
generally varies between 20 and 70 wt%, and bulk organic drops of ultrapure HNO3 to pH 1–2. All samples were
carbon between 4 and 18 wt%. The FeBulk concentrations stored at 4 °C until chemical and isotope analyses were
in bottom sediments are elevated (2.5–25 wt%) compare conducted.
to Al (2–7 wt%)—with Fe/Al molar ratios of 0.3–36 Sediment cores were transferred into glove bags and
(Schettler et al., 2007), which are either similar to or well placed under anoxic conditions (N2 pressure) immediately
above the crustal average and the ratio observed for volca- after collection. They were processed and split into cm-scale
nic rocks in the Pavin catchment (see discussion below). Ex- fractions along the core’s vertical axis. The porewaters were
treme iron-enrichment found in Lac Pavin sediments is separated from the solid phases once extracted from the
similar to deposits of the Earth’s early oceans. This rela- core barrel using RhizonÒ samplers connected via tubing
tionship indicates that a large portion of Fe is authigenic to septum capped vials under vacuum (filtration to
and precipitates in the water column and/or from sediment 0.2 lm) and were then acidified with ultrapure HCl. During
porewaters. Authigenic Fe phases in Lac Pavin sediments the whole separation process, oxygen levels were monitored
are mostly siderite (ferrous carbonate) and vivianite (fer- with an Oxi 340i WTW oxygen meter and were always be-
rous phosphate), together with iron sulfide (iron monosul- low the detection limit of 0.1 mg/L. The solid phases were
fides, and pyrite) (Viollier et al., 1997; Schettler et al., then frozen, and the porewaters were stored at 4 °C. The so-
2007; Cosmidis et al., 2014). A detailed mineralogical study lid phase of the 90 m core was fully dehydrated by freeze
demonstrated that ferric particles were present in the water drying before analysis.
column but could not be found in the lake bottom sedi- Three volcanic rocks were collected from outcrops of the
ment, suggesting intense Fe reduction at the sediment water inner Pavin crater and are assumed to represent the detrital
interface (Cosmidis et al., 2014). These ferric particles were Fe flux to the lake sediments. These volcanics are trachy-
mainly Fe oxyhydroxides in both oxic and anoxic waters, basalts formed by eruption of the Puy de Montchal, a vol-
and some ferric phosphates, which were found only in cano overhanging the Pavin crater on its south side.
anoxic waters, at and below the chemocline. Texturally, samples RUD#1 and RUD#2 are vesicle-rich
trachybasalts, while RUD#3 is massive and mostly vesi-
2.2. Sample collection and description cle-free. Thin section observations and electron microprobe
analyses indicate that these three samples are composed of
Four sampling campaigns were carried out from 2006 to olivine, clinopyroxene (augite), and plagioclase feldspar
2009. The goal of the first one in November 2006 was to test microliths in a glass matrix, with minor titano-magnetite.
if Fe isotopes were fractionated during sample collection, The sample RUD#1 also contained additional hornblende.
with particular attention paid to filtration processes on Limited secondary phases, such as Fe oxyhydroxides and
board under atmospheric conditions (see results in Appen- clay minerals formed by partial alteration of the rocks, were
dix A). The second campaign, conducted in July 2007, fo- also observed in the thin sections. Electron microprobe
cused on the collection of water samples along a vertical analyses for all three samples show Fe/Al molar ratios of
profile in the water column from 55 to 90.5 m depth, and 0.3 for the glasses and 0.1 for clay minerals. In order
constraining sulfur cycling in the mesolimnion and monim- to evaluate if alteration processes of the trachy-basalts
olimnion (Bura-Nakic et al., 2009). Dissolved oxygen, tem- could produce Fe isotope fractionation and thus modify
perature, and specific conductivity were determined using a the isotope signature of the detrital Fe flux, the three sam-
Seabird SBE 19 Seacat profiler (Sea-Bird Electronics Inc., ples were cut into several sub-samples. RUD#1 and
Washington, DC; Fig. 2b). During the third campaign in RUD#2 were split into two sub-samples: (1) a “dark” core
July 2009, we collected three sediment cores from different lacking any visible surface alteration (RUD#1-C and
depths: 32 m (oxic zone; 45°29.912N, 2°53.257E), 60 m (re- RUD#2-C) and (2) a “lighter” border apparently affected
dox interface; 45°29.833N, 2°53.267E), and 65 m (anoxic by alteration (RUD#1-B and RUD#2-B). Sample
zone; 45°29.822N, 2°53.207E). These cores were taken RUD#3 was characterized by a clear light-grey alteration
using a Mortimer gravity corer, allowing us to sample a ver- rind on its outer rim and was divided into a core
tical profile over 10–14 cm below the water–sediment inter- (RUD#3-C), alteration rind (RUD#3-A), and an interme-
face. During the fourth and last sampling campaign in diate sub-sample between the core and the outer altered
September 2009, a sediment core from the deepest part of rind (RUD#3-M).
448 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

2.3. Chemical analyses: Fe(II)/Fe(III), Mn(II), and SO2

4 Several papers have reported various experimental con-
concentrations ditions to obtain the most precise and accurate Fe isotope
analyses (Beard and Johnson, 1999; Malinovsky et al.,
The dissolved Fe(II)/Fe(III) speciation was performed 2003; Weyer and Schwieters, 2003; Dauphas et al., 2004,
on board using filtered samples from the water column 2009; Schoenberg and von Blanckenburg, 2005). While
and spectrophotometric measurement (Merck SQ300 spec- the analysis of solid geological material has been widely ex-
trophotometer) following the method of Viollier et al. plored (for example in Beard and Johnson, 1999; Malinovsky
(2000). Briefly, spectrophotometric analyses (k = 562 nm) et al., 2003; Schoenberg and von Blanckenburg, 2005), no
of the iron-ferrozine complexes were performed in a single analytical tests for measurements of samples from anoxic
aliquot before and after a reduction step with hydroxyl- waters have been published until now. In the present work,
amine. The procedure was calibrated using Fe(III) stan- we aimed at (1) evaluate the precision and accuracy of our
dards stable under normal conditions of analysis. The protocol used at the Geochemistry and Cosmochemistry
concentration of bulk dissolved Fe was also measured by Laboratory of the IPG Paris and (2) test potential Fe iso-
MC-ICP-MS after separation by ion chromatography (see tope fractionations related to anoxic water sampling. The
description below). The spectrophotometric and MC-ICP- Appendix A presents the detailed protocol for Fe isotope
MS methods provide nearly identical Fe(II) concentrations measurement as well as precision and accuracy estimated
in the monimolimnion, demonstrating that Fe extraction from long-term analyses of international standards. Briefly,
yields by chromatography were always better than 95%. the method can be summarized as follow. All solid samples
Manganese concentrations were determined on filtered were acidified with mixtures of HF, HCl, and/or HNO3 to
and acidified samples from the water column using ICP- ensure (1) complete digestion and (2) that all Fe is in the
AES (Perkin Elmer Optima 3000) at the Laboratory of ferric state. Iron was then separated from matrix elements
Water Geochemistry, IPGP. Sulfate (SO2 4 ) concentrations on anion exchange chromatography in HCl medium (Stre-
were measured by ion chromatography (Dionex DX-600 IC low, 1980; Dauphas et al., 2004). Iron concentrations and
System) from filtered samples poisoned with zinc acetate to isotopic compositions were measured using a Neptune
avoid reoxidation of any sulfide present. We also performed ThermoFischer MC-ICP-MS (Multiple Collector Induc-
incubations to determine sulfate reduction rates using radi- tively Coupled Plasma Mass Spectrometer). We corrected
olabeled sulfate method (Fossing and Jorgensen, 1989). for instrumental mass discrimination using the conven-
tional sample-standard bracketing (SSB) approach (Bel-
2.4. Iron isotope analysis shaw et al., 2000; Beard et al., 2003; Rouxel et al., 2003;
Albarède and Beard, 2004). The 56Fe/54Fe and 57Fe/54Fe
Iron isotope compositions were determined on (1) water ratios were expressed in the usual d notation in per mil
samples collected from various depths in the lake; (2) sedi- (&) as:
ment core samples, specifically porewaters (bulk dissolved
d56 Fe ¼ ½ð56 Fe=54 FeÞsample =ð56 Fe=54 FeÞstandard  1  1000
Fe), bulk solid Fe, highly reactive Fe, and pyrite Fe; and
(3) volcanic rocks collected on outcrops of the Pavin crater. and
The extractions of reactive Fe and pyrite in sediments were d57 Fe ¼ ½ð57 Fe=54 FeÞsample =ð57 Fe=54 FeÞstandard  1  1000
performed following sequential extraction methods de-
scribed previously (Rouxel et al., 2005; Severmann et al., where the standard is IRMM-014, from the Institute for
2006; Fehr et al., 2008). We chose to follow previous meth- Reference Materials and Measurements (Taylor et al.,
ods, despite alternative, more selective extraction schemes, 1992). The analytical blank was always below 40 ng Fe
to allow us to compare our results with those from other (average 30 ng) and represented less than 0.4% of the bulk
studies. Briefly, reactive Fe was leached using cold 1 M Fe in samples. Based on replicate analyses of international
HCl for 3 h. This procedure extracted labile Fe phases such rock standards over a 5-year period (see Appendix A), the
as amorphous Fe(III)-oxides, FeS, Fe-carbonates, and external precision and accuracy were always better than
potentially some reactive silicate Fe. However, well-crystal- 0.06& for d56Fe and 0.12& for d57Fe (2SD).
lized phases such as goethite, hematite, magnetite, and pyr-
ite were not dissolved during this step. Following HCl 3. RESULTS
treatment, Fe contained in pyrite was extracted using the
sequential extraction method of Huerta-Diaz and Morse 3.1. Water column
(1990). The residual sediment was treated in 10 N HF for
16 h to dissolve any silicate Fe, and this fraction was not Results of the chemical and isotopic analyses on filtered
analyzed further. After the HF step, the sediment residue water samples are given in Table 1 and plotted in Fig. 3.
was cleaned (with HCl and water rinses). Pyrite was as- The concentration of dissolved ferrous iron in the oxic layer
sumed to be the only significant residual Fe phase after of the water column is low, with values <0.1 lM. Because
HCl and HF treatment (Severmann et al., 2006) and was of this low concentration, no isotopic measurement was at-
extracted using concentrated HNO3 overnight on a hotplate tempted for these samples. The concentration and isotopic
at 120 °C. Although crystallized Fe(III) phases were not composition of dissolved Fe in the anoxic zone both in-
analyzed with this extraction scheme, they are a minor crease with depth in the water column, with values from less
trace component of these sediments (Viollier et al., 1997; than 2 to more than 1200 lM and 2.14& to +0.31&,
Schettler et al., 2007; Cosmidis et al., 2014). respectively (Table 1; Fig. 3). The strongest gradients in
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 449

Table 1
Dissolved manganese, sulfate and ferrous iron concentrations, together with isotopic composition of the bulk dissolved iron (expressed as
d56Fe and d57Fe), in water samples from Lac Pavin, collected in July 2007.
Sample # Depth (m) Mn(II) (lM) SO2
4 (lM) Fe(II) (lM) d56Fe (&) 2SD (&) d57Fe (&) 2SD (&)
MX-15_8 55.0 0.08 0.0
MX-15_9 56.0 0.09 0.0
MX-15_10 57.0 0.42 14 0.0
MX-15_11 57.5 0.29 14 0.0
MX-15_12 58.0 0.31 14 0.1
MX-15_13 58.5 1.65 14 0.0
MX-15_14 59.0 2.39 15 0.0
MX-15_15 59.5 7.49 14 0.1
MX-15_16 60.0 14.9 14 2.1 2.14 0.04 3.16 0.06
MX-15_17 60.5 16.1 14 20.8 1.30 0.08 1.98 0.10
MX-15_18 61.0 17.4 14 106 1.34 0.02 1.97 0.08
MX-15_19 61.5 21.2 13 108 1.25 0.04 1.87 0.06
MX-15_20 62.0 19.6 9 174 1.25 0.08 1.84 0.08
MX-15_21 62.5 16.5 2 308 0.51 0.04 0.77 0.08
MX-15_22 63.0 15.7 4 378 0.40 0.04 0.58 0.06
MX-15_23 64.0 16.5 2 435 0.33 0.06 0.49 0.08
MX-15_24 65.0 16.7 1 470 0.26 0.04 0.38 0.08
MX-15_25 66.0 18.1 1 568 0.15 0.04 0.21 0.08
MX-15_26 68.0 19.0 1 688 0.23 0.02 0.33 0.10
MX-15_27 70.0 20.4 1 757 0.11 0.04 0.17 0.12
MX-15_28 72.0 21.2 2 864 0.05 0.06 0.06 0.12
MX-15_29 75.0 22.2 964 0.01 0.08 0.00 0.14
MX-15_30 80.0 23.8 1004 0.23 0.04 0.34 0.12
MX-15_31 85.0 25.3 1074 0.20 0.06 0.28 0.18
MX-15_32 87.0 25.1 1105
MX-15_33 89.0 26.4 1189
MX-15_34 90.0 26.5 1210 0.31 0.08 0.45 0.12
MX-15_35 90.5 27.1 1203 0.31 0.08 0.48 0.08
Uncertainties correspond to 2SD (standard deviation).
Depth: depth in the water column. The oxic–anoxic interface is at 60 m depth.

dissolved Fe concentration and d56Fe values are in the tracks the increase of dissolved Fe concentration in the
depth range between 60 and 62.5 m. Sulfate concentrations water column. The Fe concentration in bulk solid sediment
are mostly constant around 14 lM in the oxic layer (data also increases with depth below the sediment–water inter-
from 57 to 61 m, see also Fig. 2c) but decrease abruptly face (Table 2). In order to avoid any dilution effect due to
for depths deeper than 61.5 m. These concentration values diatoms, organic matter, and detrital minerals deposition,
are similar to those measured 20 years ago by Michard it is preferable to trace authigenic Fe enrichment by using
et al. (1994), indicating a long term stability of SO2
4 level Fe/Al ratios rather than absolute Fe concentrations (e.g.,
in the lake. The concentration of dissolved Mn(II) shows Lyons and Severmann, 2006). The Fe/Al molar ratio in
an increase with depth in the water column from 0.08 lM bulk sediment is significantly higher in the anoxic cores
at a depth of 55 m to 27.1 lM in the deepest part of the than in the oxic core, with average values of 0.21 ± 0.11,
lake. There is a peak of dissolved Mn(II) concentration lo- 0.68 ± 0.32, 0.84 ± 0.35 and 0.88 ± 0.37 for cores at 32,
cated around 61.5 m depth (Fig. 3), suggesting that sinking 60, 65 and 92 m depth, respectively (Fig. 5b). The Fe/Al ra-
Mn particles may be dissolved at that depth. tio of the core in the oxic zone is lower than the average
crustal value of 0.5 (e.g., Taylor and McLennan, 1995;
3.2. Sediments and porewaters Lyons and Severmann, 2006) but similar to the 0.3 ratio ob-
served for volcanic rocks from Pavin catchment. In con-
Iron concentrations and isotope compositions measured trast, the three sediment cores sampled under anoxic
in sediments (bulk and sequential extractions) and associ- waters at 60, 65, and 92 m are all significantly enriched in
ated porewaters are reported in Table 2 and Fig. 4. Iron authigenic Fe, in good agreement with results from a previ-
concentrations in porewater of the four sediment cores ous study (Schettler et al., 2007).
are roughly constant within each individual core. However, The three sediment cores sampled under anoxic waters
comparison among the different cores show that the aver- all show similar Fe isotope patterns among porewaters, bulk
age Fe concentration increases with depth in the lake sediments, reactive Fe (cold HCl leach fraction), and pyrites.
(Fig. 5a), with values of 30 ± 3, 180 ± 70, 510 ± 80, On average, porewaters, bulk sediment, and reactive Fe
1170 ± 260 lM for cores at 32, 60, 65, and 92 m depth, show d56Fe values near 0&. In detail, mean d56Fe values
respectively. The increase in Fe concentration with depth of porewaters display a slight increase with depth, with
450 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

(A) Fe(II), μM (B) δ56Fe, ‰

0 400 800 1200 -2 -1.5 -1 -0.5 0 0.5
55 55
O2 δ56Fe
60 60

Depth, m 65 65

70 70
Fe(II)
75 75 SO42-
80 80

85 85
Zone of sulfate
90 reduction and 90
sulfide formation Mn(II)
95 95
0 50 100 150 200 0 5 10 15 20 25 30
O2, μM Mn(II), SO42-, μM
Fig. 3. Water column chemical profiles in Lac Pavin. (A) Dissolved oxygen and ferrous iron concentrations. (B) d56Fe value of dissolved iron
together with manganese and sulfate concentrations. Below 60 m, the lake is anoxic and contains dissolved iron. There is a large shift in iron
isotope values and iron concentrations, between 60 and 64 m, linked to ferrous iron oxidation. The limited iron isotope variability below 64 m
is likely caused by ferric-ferrous interactions or limited amounts of water column iron reduction. Essentially all sulfate reduction and
particulate sulfide formation in the lake occurs in the highlighted zone near the oxic–anoxic transition.

averages of 0.08 ± 0.13& (n = 6), 0.02 ± 0.07& (n = 9), with subsequent precipitation of Fe sulfides (Severmann
and 0.15 ± 0.07& (n = 12) for cores at 60, 65 and 92 m, et al., 2006).
respectively (Table 2). Bulk sediments have d56Fe values
averaging 0.10 ± 0.11& (n = 4), 0.11 ± 0.20& (n = 12), 3.3. Volcanic rocks from Lac Pavin
and 0.02 ± 0.12& (n = 13) for cores at 60, 65 and 92 m,
respectively. Reactive Fe averages 0.00 ± 0.09& (n = 4), Iron concentrations and isotope compositions of the
0.18 ± 0.07& (n = 6), and 0.01 ± 0.11& (n = 13) for the volcanic rocks are given in Table 3. They show a very
cores at 60, 65 and 92 m, respectively. In contrast with these homogeneous distribution, with Fe concentrations ranging
rather homogeneous values for porewaters, bulk sediment, from 5.81 to 6.66 wt% and d56Fe values from 0.16& to
and reactive Fe, d56Fe values of pyrites in sub-chemocline 0.22&. Considering the analytical uncertainties, these are
cores (60, 65 and 92 m) are highly variable and show signif- mostly constant Fe concentrations and d56Fe values, with
icant deviations from the detrital iron isotope composition an overall average for all samples at 6.41 ± 0.34 wt% and
(overall d56Fepyrite between 1.51& and +0.09&; average 0.20 ± 0.02& (1r; n = 7), respectively. For each sample,
0.93&). For the individual cores, the average d56Fe values there is no difference in Fe concentration or d56Fe values
of pyrite are 0.88 ± 0.13& (n = 4), 0.61 ± 0.39& between the different sub-samples (core vs. border vs. inter-
(n = 7), and 1.20 ± 0.24 & (n = 9) for cores at 60, 65 mediate locations; Table 3). These data imply that alter-
and 92 m, respectively. The core from the oxic zone is dis- ation had no effect on bulk Fe budget in these rocks,
tinct from the sub-chemocline sediment cores, with pyrite although some redistribution within secondary mineral
d56Fe values that match the detrital iron flux (d56Fepyrite be- phases may have occurred. The d56Fe values of these volca-
tween 0.09& and +0.26&, average = 0.1 ± 0.18&, nic rocks are similar to the average bulk values measured in
n = 4). Importantly, a significant range of variation our core collected at 32 m depth in Lac Pavin below fully
(1&) for porewater d56Fe values in this core contrasts with oxic waters (average  0.16 ± 0.06& [n = 9]; Table 2) but is
the limited variability (<0.3&) observed for the sub- slightly different from the average bulk values measured for
chemocline cores, where pyrite formation occurred in Fe- subchemocline cores (0.10 ± 0.11& [n = 4], 0.11 ± 0.20&
rich porewaters. The d56Fe of porewaters in sediments sam- [n = 12], 0.02 ± 0.12& [n = 13] for cores at 60, 65, and
pled from the oxic zone tend to increase with depth in the 92 m, respectively). The close similarity between d56Fe val-
core, with values from 0.44& to +0.64& (Table 2; ues of the volcanic rocks with those of sediments from Lac
Fig. 4). This relationship is similar to that observed in sedi- Pavin suggests that dominant authigenic phases present in
ments collected off California continental margin and prob- the sediments (mostly vivianite and siderite) have d56Fe val-
ably reflects the effect of initial dissimilatory Fe reduction ues close to the detrital input.
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 451

Table 2
Fe concentrations (in wt% for solid and in mM for porewater) and Fe isotope compositions in samples of the sediment cores collected in Lac
Pavin.
Sample # Sample type Deptha (cm) d56Fe (&) 2SD (&) d57Fe (&) 2SD (&) Fe conc (wt% or mM) Fe/Al (molar)
Core from the oxic layer (32 m depth)
C3-1-1 Bulk 0 0.23 0.01 0.31 0.10 0.9 0.24
C3-2-1 Bulk 1.5 0.05 0.06 0.07 0.04 0.32
C3-3-1 Bulk 2.5 0.11 0.09 0.12 0.17 0.7 0.36
C3-4-1 Bulk 3.5 0.22 0.02 0.38 0.04 0.6 0.19
C3-5-1 Bulk 4.5 0.15 0.01 0.18 0.13 0.7 0.19
C3-6-1 Bulk 5.5 0.19 0.09 0.34 0.15 0.9 0.33
C3-7-1 Bulk 6.5 0.14 0.08 0.21 0.02 1.9 0.14
C3-8-1 Bulk 7.5 0.19 0.02 0.28 0.14 2.3 0.07
C3-9-1 Bulk 10 0.21 0.04 0.31 0.06 2.4 0.07
C3-1-1R HCl leach 0 0.10 0.02 0.13 0.16 0.1
C3-4-1R HCl leach 3.5 0.07 0.08 0.12 0.14 0.1
C3-8-1R HCl leach 7.5 0.01 0.06 0.01 0.12 0.8
C3-1-1P Pyrite 0 0.26 0.08 0.33 0.12 0.1
C3-2-1P Pyrite 1.5 0.09 0.04 0.15 0.06 0.1
C3-4-1P Pyrite 3.5 0.25 0.06 0.36 0.10 0.2
C3-8-1P Pyrite 7.5 0.01 0.09 0.01 0.11 0.1
C3-2-2 Porewater 1.5 0.26 0.04 0.41 0.10 0.04
C3-3-2 Porewater 2.5 0.44 0.06 0.65 0.08 0.03
C3-4-2 Porewater 3.5 0.25 0.08 0.41 0.06 0.04
C3-5-2 Porewater 4.5 0.06 0.09 0.11 0.04 0.03
C3-6-2 Porewater 5.5 0.13 0.04 0.18 0.16 0.03
C3-7-2 Porewater 6.5 0.20 0.04 0.18 0.10 0.04
C3-8-2 Porewater 7.5 0.64 0.02 0.90 0.15 0.03
C3-9-2 Porewater 10 0.09 0.06 0.20 0.16 0.04

Core from the oxic–anoxic boundary (60 m depth)

D-2 Bulk 2.75 0.15 0.02 0.23 0.06 0.60 0.94
F-2 Bulk 4.3 0.20 0.02 0.35 0.06 0.60 1.10
I-2 Bulk 9 0.10 0.04 0.09 0.10 0.70 0.32
L-2 Bulk 12.75 0.06 0.06 0.07 0.04 0.50 0.34
B-2R HCl leach 0.85 0.10 0.01 0.17 0.06 0.40
D-2R HCl leach 2.75 0.11 0.09 0.16 0.04 0.50
F-2R HCl leach 4.3 0.03 0.04 0.05 0.10 0.20
I-2R HCl leach 9 0.03 0.06 0.05 0.17 0.30
B-2P Pyrite 0.85 1.02 0.08 1.51 0.10 0.20
D-2P Pyrite 2.75 0.76 0.02 1.17 0.13 0.20
F-2P Pyrite 4.3 0.89 0.04 1.38 0.12 0.04
I-2P Pyrite 9 0.86 0.09 1.29 0.04 0.04
A-2 Porewater 0.85 0.14 0.06 0.25 0.08 0.05
C-2 Porewater 2.75 0.20 0.02 0.27 0.02 0.20
E-2 Porewater 4.3 0.17 0.02 0.25 0.06 0.22
G-2 Porewater 5.9 0.03 0.04 0.02 0.08 0.23
J-2 Porewater 9 0.13 0.04 0.13 0.10 0.18
K-2 Porewater 12.75 0.10 0.06 0.10 0.06 0.20

Core from the anoxic layer (65 m depth)

C2-1-1 Bulk 0 0.27 0.02 0.43 0.08 1.00 0.83
C2-2-1 Bulk 1.5 0.18 0.06 0.31 0.14 1.90 0.77
C2-3-1 Bulk 2.5 0.24 0.06 0.37 0.13 1.50 0.62
C2-4-1 Bulk 3.5 0.34 0.04 0.53 0.06 1.10 1.30
C2-5-2 Bulk 4.5 0.32 0.02 0.45 0.06 0.60
C2-6-1 Bulk 5.5 0.25 0.04 0.40 0.06 0.60 0.85
C2-7-1 Bulk 6.5 0.09 0.04 0.15 0.06 1.46
C2-8-1 Bulk 7.5 0.19 0.04 0.27 0.08 0.80 1.16
C2-9-1 Bulk 8.5 0.02 0.04 0.05 0.04 0.82
(continued on next page)
452 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

Table 2 (continued)
Sample # Sample type Deptha (cm) d56Fe (&) 2SD (&) d57Fe (&) 2SD (&) Fe conc (wt% or mM) Fe/Al (molar)
C2-10-1 Bulk 9.5 0.12 0.09 0.22 0.14 1.10 0.58
C2-11-1 Bulk 10.5 0.11 0.04 0.18 0.02 0.51
C2-12-1 Bulk 11.5 0.27 0.08 0.39 0.12 1.50 0.29
C2-1-1R HCl leach 0 0.21 0.02 0.33 0.16 0.80
C2-2-1R HCl leach 1.5 0.09 0.04 0.16 0.06 0.60
C2-3-1R HCl leach 2.5 0.13 0.09 0.20 0.04 0.60
C2-5-1R HCl leach 4.5 0.28 0.08 0.40 0.18 0.50
C2-6-1R HCl leach 5.5 0.24 0.09 0.37 0.12 0.50
C2-10-1R HCl leach 9.5 0.14 0.02 0.28 0.08 0.30
C2-1-1P Pyrite 0 0.48 0.04 0.83 0.10 0.10
C2-2-1P Pyrite 1.5 0.09 0.06 0.12 0.08 0.10
C2-3-1P Pyrite 2.5 0.43 0.02 0.69 0.02 0.02
C2-4-1P Pyrite 3.5 0.56 0.04 0.84 0.15 0.10
C2-5-1P Pyrite 4.5 1.01 0.08 1.53 0.08 0.09
C2-6-1P Pyrite 5.5 0.84 0.02 1.23 0.14 0.11
C2-10-1P Pyrite 9.5 1.03 0.02 1.50 0.12 0.15
C2-1-2 Porewater 0 0.12 0.06 0.20 0.06 0.46
C2-2-2 Porewater 1.5 0.01 0.06 0.03 0.12 0.61
C2-3-2 Porewater 2.5 0.03 0.04 0.06 0.10 0.41
C2-4-2 Porewater 3.5 0.01 0.06 0.01 0.10 0.55
C2-6-2 Porewater 5.5 0.03 0.04 0.06 0.09 0.63
C2-7-2 Porewater 6.5 0.08 0.02 0.15 0.10 0.51
C2-8-2 Porewater 7.5 0.03 0.08 0.05 0.19 0.56
C2-9-2 Porewater 8.5 0.09 0.04 0.16 0.02 0.44
C2-12-2 Porewater 11.5 0.09 0.04 0.13 0.08 0.46

Core at the bottom of Lac Pavin (92 m depth)

1-1 Bulk 3.5 0.06 0.06 0.09 0.12 1.88 0.71
3-1 Bulk 8.5 0.15 0.04 0.20 0.08 2.70 1.18
5-1 Bulk 13.5 0.10 0.08 0.17 0.06 5.30 1.03
7-1 Bulk 18.5 0.01 0.08 0.09 0.04 2.20 0.79
9-1 Bulk 23.5 0.01 0.02 0.04 0.02 3.50 0.58
11-1 Bulk 28.5 0.12 0.09 0.22 0.16 6.70 0.88
13-1 Bulk 33.5 0.12 0.04 0.20 0.17 6.50 0.48
15-1 Bulk 38.5 0.10 0.02 0.21 0.04 6.50 0.46
17-1 Bulk 46 0.15 0.06 0.23 0.14 6.90 1.62
19-1 Bulk 51 0.08 0.02 0.09 0.06 7.10 0.60
21-1 Bulk 56 0.19 0.06 0.25 0.04 6.60 1.50
23-1 Bulk 63.5 0.01 0.02 0.02 0.08 0.73
25-1 Bulk 73.5 0.15 0.06 0.26 0.14 4.40 0.83
1-1R HCl leach 3.5 0.01 0.08 0.02 0.06
3-1R HCl leach 8.5 0.10 0.04 0.16 0.14 2.40
5-1R HCl leach 13.5 0.05 0.06 0.09 0.08 4.90
7-1R HCl leach 18.5 0.03 0.08 0.06 0.10 1.60
9-1R HCl leach 23.5 0.00 0.06 0.02 0.06 3.40
11-1R HCl leach 28.5 0.06 0.02 0.09 0.06 6.00
13-1R HCl leach 33.5 0.04 0.08 0.10 0.10 3.80
15-1R HCl leach 38.5 0.13 0.08 0.20 0.15
17-1R HCl leach 46 0.11 0.06 0.20 0.04 6.50
19-1R HCl leach 51 0.19 0.06 0.24 0.14 6.30
21-1R HCl leach 56 0.14 0.05 0.19 0.14
25-1R HCl leach 73.5 0.18 0.06 0.28 0.06 0.40
27-1R HCl leach 83.5 0.06 0.08 0.11 0.08
LPC-1 Pyrite 3.5 1.08 0.07 1.64 0.12 0.14
LPC-2 Pyrite 5.5 1.51 0.06 2.25 0.12 0.02
LPC-3 Pyrite 8.5 0.78 0.02 1.19 0.04 0.08
LPC-4 Pyrite 10.5 1.14 0.02 1.75 0.06 0.55
LPC-5 Pyrite 13.5 1.14 0.04 1.70 0.06 0.33
LPC-6 Pyrite 15.5 1.37 0.02 2.00 0.04 0.39
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 453

Table 2 (continued)
LPC-11 Pyrite 28.5 1.00 0.02 1.45 0.04 0.06
LPC-15 Pyrite 38.5 1.28 0.02 1.76 0.08 0.22
LPC-23 Pyrite 63.5 1.48 0.04 2.14 0.04 0.14
LPC-1-1 Porewater 3.5 0.09 0.15 0.01 0.04 1.41
LPC-5-2 Porewater 13.5 0.21 0.02 0.35 0.18 0.84
LPC-7-2 Porewater 18.5 0.04 0.02 0.09 0.10 0.86
LPC-11-2 Porewater 28.5 0.09 0.06 0.15 0.02 0.83
LPC-13-2 Porewater 33.5 0.09 0.02 0.26 0.12 0.90
LPC-15-2 Porewater 38.5 0.24 0.02 0.40 0.12 1.38
LPC-17-2 Porewater 46 0.12 0.04 0.20 0.04 1.31
LPC-19-2 Porewater 51 0.18 0.02 0.29 0.04 1.46
LPC-21-2 Porewater 56 0.16 0.02 0.26 0.10 1.48
LPC-23-2 Porewater 63.5 0.17 0.09 0.27 0.14 1.31
LPC-25-2 Porewater 73.5 0.18 0.08 0.31 0.04 1.23
LPC-27-2 Porewater 83.5 0.24 0.02 0.46 0.10 0.99
LPC-27-2 Porewater 83.5 0.24 0.02 0.46 0.10 0.99
Bulk: bulk Fe in solid sediment; HCl leach: reactive Fe representing easily soluble chemical species; pyrite: Fe extracted from the residue using
HNO3 treatment after sequential extraction (Severmann et al., 2006).
Uncertainties correspond to 2SD (standard deviation).
a
Depth in the sediment core. The water–sediment interface is at 0 cm depth.

32 m depth
(oxic zone)

δ56Fe, ‰ OXIC
-1.0 0.0 1.0
0
60 m depth
2 (oxic-anoxic
boundary) ANOXIC
Depth, cm

4
δ56Fe, ‰ 65 m depth
6 -1.0 0.0 1.0 (anoxic zone)
0

8 2 δ56Fe, ‰
-1.0 0.0 1.0 92 m depth
4 0
Depth, cm

10 (anoxic bottom)
6 2
12 δ56Fe, ‰
8 4 -2.0 -1.0 0.0 1.0
Depth, cm

0
10 6

12 8 20
Depth, cm

bulk sediment 14 10
reactive iron 40
12
pyrite
porewater 14 60

80

Fig. 4. Iron isotope composition of porewater, bulk rock, reactive iron, and pyrite in four sediment cores collected at Lac Pavin. Error bars on
d56Fe values are smaller than dot sizes (typically <0.03&, 1r). There is limited isotope variability and deviations away from the detrital iron
flux except for the sedimentary pyrite. The pyrite is isotopically variable and has d56Fe values similar to those seen in the zone near the oxic–
anoxic transition in the water column where the majority of the iron sulfides form. For comparison, the vertical light-grey zones represent the
typical range of Fe isotope compositions for detrital sediments, volcanic rocks and hydrothermal fluids.

4. DISCUSSION (0.1–1200 lM) and d56Fe values (2.14 to +0.31&)

(Fig. 3; Table 1). These results are similar to those obtained
4.1. Iron biogeochemical cycle in Lac Pavin in two recent isotope studies focused on Fe cycling in other
anoxic Fe-rich basins (Malinovsky et al., 2005; Teutsch
4.1.1. Iron cycling in the water column et al., 2009). Malinovsky et al. (2005) examined Fe isotope
Water samples in Lac Pavin show an increase, variations in two lakes from Sweden affected by seasonally
with increasing depth, in dissolved Fe concentration anoxic conditions. Iron isotope compositions of suspended
454 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

10 matter were found to be enriched in the heavy Fe isotope

Avg Fe pw (mM)
(A) relative to dissolved Fe (1&). The authors interpreted
these results to reflect preferential partitioning of heavy iso-
1 topes during aqueous Fe(II) oxidation to Fe(III) and subse-
quent Fe(III) precipitation. Teutsch et al. (2009) measured
the Fe contents and isotope compositions of the water col-
0.1
umn of meromictic Lake Nyos (Cameroon). Iron contents
and isotope compositions showed a strong increase with
0.01 depth, with the highest gradient at the oxic–anoxic bound-
0 20 40 60 80 100 ary. The very negative d56Fe values (1.25&) observed for
Core depth (m) dissolved Fe at the oxic–anoxic boundary were tentatively
interpreted to record intense dissimilatory iron reduction
(DIR) from sinking Fe(III) particles previously formed at
1.2 (B)
Avg (Fe/Al)bulk sed

the surface of the lake. This interpretation is plausible since

DIR can enrich Fe(II) in the light isotopes by 3& relative
0.8 to initial Fe(III) particles (Crosby et al., 2005). However, an
assumption implicit in this model was that Fe oxidation
0.4
was quantitative and occurring only when oxygen levels
were >1 lM. Therefore, the iron isotope profile in Lake
Nyos could also be driven by oxidation (Teutsch et al.,
0
0 20 40 60 80 100 2009).
Core depth (m) In Lac Pavin, oxidation generates the dissolved iron iso-
tope gradient. Dissolved iron is mainly brought to the
Fig. 5. Average Fe concentrations in porewaters (A) and Fe/Al chemocline via diffusion from Fe-rich bottom waters (Mi-
molar ratio in bulk sediments (B) reported for each core as a chard et al., 1994). The oxidation of dissolved Fe(II) to
function of the depth of the sediment core (32, 60, 65 and 92 m). Fe(III) at the chemocline enriches Fe(III) in the heavy iso-
Error bars on Y-axis represent standard deviation calculated for
topes by up to 3&, independent of whether this oxidation
each core (±1r). Light grey zone in figure (B) indicates the range of
is biologically mediated or abiotic (Johnson et al., 2002;
Fe/Al ratios (0.1–0.3) of the volcanic rocks from the Pavin Crater,
and is supposed to be representative of the detrital flux values. Fe/ Welch et al., 2003; Jarzecki et al., 2004; Anbar et al.,
Al ratios higher than these detrital values result from an authigenic 2005; Balci et al., 2006). The precipitation of Fe(III)
Fe enrichment of the sediment. particles from dissolved Fe(II) produces either (1) no frac-
tionation if the precipitation is slow or (2) possibly a kinetic
isotope fractionation, enriching the solid in the light

Table 3
Iron concentrations and isotope compositions (expressed as d56Fe and d57Fe) in volcanic rocks from Lac Pavin.
Sample Sub-sample description Analysis number [Fe] (wt%) d56Fe (&) 2SD (&) d57Fe (&) 2SD (&)
RUD#1-C Core 1 6.62 0.22 0.07 0.27 0.12
2 5.64 0.21 0.07 0.31 0.11
Avg 6.13 0.21 0.29
RUD#1-B Border 1 5.81 0.21 0.04 0.35 0.08
2 5.80 0.21 0.03 0.35 0.04
Avg 5.81 0.21 0.35
RUD#2-C Core 1 6.79 0.22 0.05 0.32 0.09
2 5.90 0.16 0.01 0.28 0.06
Avg 6.34 0.19 0.30
RUD#2-B Border 1 6.46 0.21 0.02 0.30 0.05
2 6.87 0.17 0.05 0.27 0.05
Avg 6.66 0.19 0.29
RUD#3-C Core 1 7.36 0.14 0.05 0.21 0.08
2 5.90 0.18 0.04 0.28 0.11
Avg 6.63 0.16 0.25
RUD#3-M Intermediate 1 6.95 0.19 0.06 0.27 0.12
2 6.36 0.19 0.03 0.25 0.11
Avg 6.66 0.19 0.26
RUD#3-A Altered surface 1 6.67 0.25 0.04 0.33 0.05
2 6.61 0.19 0.04 0.27 0.01
Avg 6.64 0.22 0.30
Uncertainties correspond to 2SD (standard deviation).
Avg: average value of the two duplicate analyses for each sub-sample.
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 455

isotopes of Fe, if the reaction is rapid (Skulan et al., 2002). (Buffle et al., 1989; Hongve, 1997; Konhauser et al., 2007).
Overall, the net isotopic fractionation during Fe(III) parti- However, some phosphate is being removed in ferric Fe
cle formation from dissolved Fe(II) can thus be modulated phosphates (Cosmidis et al., 2014).
by the rate of precipitation but, in most cases, it produces a Ferric particles formed at the chemocline sink in the
precipitate enriched in the heavy isotopes of Fe. For in- water column down to the lake bottom. In contrast to Fe
stance, in experiments of abiotic precipitation of ferrihy- oxidation, microbial iron reduction at Lac Pavin occurs
drite from aqueous Fe(II), the precipitate is typically mostly in the sediment pile during early diagenesis rather
enriched in the heavy isotopes by 1–2& (e.g., Bullen than in the water column (Viollier et al., 1997). Porewater
et al., 2001; Beard and Johnson, 2004). It is important to iron released to the water column is transported upward
note that opposite Fe isotope fractionations have recently to the chemocline, predominantly through eddy diffusion
been found in water samples from anoxic Baltic Sea, where (Aeschbach-Hertig et al., 1999).
Fe(III) particles were depleted in heavy isotopes relative to Below 65 m, the increase with depth in the d56Fe values
the residual dissolved Fe (Staubwasser et al., 2013). This re- for dissolved iron is moderate compared to the sharp in-
sult was interpreted as due to a strong kinetic isotope frac- crease at the redox boundary (i.e., 60–64 m). This progres-
tionation during oxidation–precipitation process sive d56Fe increase may derive from Fe isotope
(Staubwasser et al., 2013). fractionation related to (1) adsorption of dissolved Fe(II)
The largest gradient of d56Fe values for dissolved Fe in on Fe(III) particles formed at the chemocline (Icopini
the water column of Lac Pavin is observed near the chemo- et al., 2004), (2) limited amount of iron reduction in the
cline, between 60 and 64 m depth, with the most negative water column of the sinking Fe(III) particles (Beard and
values at 60 m. Oxygen levels are low (<2 lM O2) at that Johnson, 2004; Crosby et al., 2005) and/or (3) a putative
depth interval, suggesting microbial microaerophillic or an- isotopic effect due to ferrous phosphate (vivianite) precipi-
oxic pathways are the dominant modes of oxidation. Inter- tation in the deeper part of the water column, as evidenced
estingly, this large gradient in d56Fe coincides with a from mineralogical data in the lake (Cosmidis et al., 2014).
significant increase in dissolved Mn concentration A contribution from a ferruginous sub-chemocline spring is
(Fig. 3), suggesting that part of the Fe(II) oxidation in this also likely, and may explain part of the Fe isotope variabil-
depth interval is coupled to Mn(IV) reduction. In detail, ity. The presence of a mineralized spring within the anoxic
Fe(II) is oxidized to Fe(III) and precipitates as Fe oxyhy- zone of Lake Pavin was proposed as early as 1975 in order
droxides (e.g., Fe(OH)3), while Mn(IV), initially present to explain observed gradients at the redox boundary
as MnO2 particles, is reduced and thus solubilized to (Meybeck et al., 1975) and was hypothesized later from
Mn(II). The overall reaction can be expressed as the follow- hydrological budgets and isotopic mass balances (Martin,
ing two half-reactions (Hongve, 1997): 1985; Michard et al., 1994; Olive and Boulègue, 2004;
Assayag et al., 2008; Jézéquel et al., 2011). Overall, observa-
2Fe2þ þ 6H2 O ! 2FeðOHÞ3 þ 6Hþ þ 2e tion of the lake level indicates a relative stability, within an
MnO2 þ 4Hþ þ 2e ! Mn2þ þ 2H2 O uncertainty of ±50 cm, over several decades (e.g. Assayag
2Fe2þ þ MnO2 þ 4H2 O ! 2FeðOHÞ3 þ 2Hþ þ Mn2þ et al., 2008). Water inputs to the lake include direct precip-
itation (18 L/S) and surface streams into the lake (20 L/
Considering the reaction stoichiometry and the relative S) (Meybeck et al., 1975; Martin, 1985; Aeschbach-Hertig
concentrations of dissolved Fe(II) and Mn(II) (Fig. 3), et al., 2002). Water outputs are dominated by evaporation
Mn(IV) reduction could be quantitatively driven by Fe(II) (7 L/s) and surface outlet flowing into the Couze Pavin
oxidation but the reverse appears impossible. There is too river (on average 50 L/s) (Meybeck et al., 1975; Martin,
much Fe being oxidized (hundred of lM/L) compare to 1985; Aeschbach-Hertig et al., 2002; Assayag et al., 2008).
Mn being reduced (tens of lM/L). Thus, another type of Thus, the difference between water inputs and outputs leads
Fe oxidation is required either by microbial microaerophillic to a deficit of about 20 L/s (Aeschbach-Hertig et al., 1999,
or anoxic oxidation. From 64 to 65 m, there is a subsequent 2002), which is assumed to be balanced by sub-surface
Mn(II) drop, representing Mn sorption onto iron oxides or springs (Glangeaud, 1916; Meybeck et al., 1975; Martin,
precipitation of other Mn-bearing mineral (Fig. 3). 1985; Viollier et al., 1997; Aeschbach-Hertig et al., 1999,
Ferric precipitates formed at the redoxcline in Pavin 2002; Olive and Boulègue, 2004; Assayag et al., 2008;
may also include Fe(III)-phosphates due to particularly Jézéquel et al., 2011). The main question concerns the dis-
high concentrations of dissolved PO3 4 in the monimolim- tribution and nature of these sub-surface springs in the
nion (Cosmidis et al., 2014). In any case, the Mn(II) con- mixolimnion and monimolimnion. High precision and high
centration profiles suggest that Fe isotope variations at frequency temperature and conductivity profiles recently
60–65 m depth is likely related to Fe oxidation rather than confirmed the presence of a sub-lacustrine, intermittent
reduction. Molybdenum also shows significant decrease in cold spring at the bottom of the mixolimnion at depth be-
this zone (Fig. 2c), which can be linked to adsorption on tween 50 and 55 m (Bonhomme et al., 2011), as earlier de-
Mn oxides and Fe oxyhydroxides, and Mo coprecipitation tected in a previous study (Aeschbach-Hertig et al., 2002).
with FeS (Viollier et al., 1995, 2014; Barling and Anbar, In the monimolimnion, a deep ferruginous spring is clearly
2004). The sharp decrease in dissolved phosphate concen- required to feed and sustain high Fe concentrations in the
trations near the oxic–anoxic boundary (Fig. 2c) is likely anoxic water layer, especially considering the high Fe re-
also a signal for significant oxidation in this zone, since moval flux through burial in the sediments deposited
phosphate can strongly adsorbed onto Fe oxide precipitates under anoxic waters. The precise depth of this purported
456 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

ferruginous spring is unknown but has been hypothesized 1.6

92 m (lake bottom)
to be located at the lake bottom (Michard et al., 1994; 65 m (anoxic core)

Fe porewater (mM)
Assayag et al., 2008) or in the upper part of the monimo- 1.2 60 m (oxic-anoxic boundary)
limnion at 65 m (Jézéquel et al., 2011). In terms of Fe iso-
topes, there is a slight shift towards negative d56Fe values at
0.8
68 m depth, which may confirm a deep spring input at
that depth (Fig. 2b).
0.4

4.1.2. Iron isotope signature of sedimentary pyrite inherited

from water column cycling 0.0
In theory, negative d56Fe values in pyrite can be ex- -1.6 -1.2 -0.8 -0.4 0.0

plained by three different scenarios: (1) Fe isotope fraction- δ56Fepyr (‰)

ation associated with DIR in the sediment, releasing
Fig. 6. Iron concentration in porewaters reported as a function of
preferentially light Fe in the porewaters and subsequent Fe isotope composition of pyrite for anoxic sediment cores.
incorporation of this Fe into diagenetic pyrite (e.g.,
Johnson et al., 2008b); (2) Fe isotope fractionation related composition of sedimentary pyrite is controlled by kinetic
to diagenetic pyrite formation from porewater Fe(II) (e.g., isotope fractionation during precipitation of pyrite in the
Guilbaud et al., 2011); and (3) pelagic pyrite formation in anoxic sediment cores, one might predict a relationship be-
depth intervals where Fe(II) is driven light as a result of tween Fe concentrations in sediment porewater and pyrite
partial oxidation (e.g., Rouxel et al., 2005). These three d56Fe values. Interestingly, this is not observed; there is
scenarios are explored in detail below with regard to the no clear trend between these two parameters in the anoxic
present Pavin data. sediment cores (Fig. 6). The range of pyrite d56Fe values
First, if the light d56Fe values of pyrite were linked to for the various sediment cores overlap (around 1&) while
DIR, porewater d56Fe values need to be significantly nega- porewater Fe concentrations show strong variability,
tive, with values similar to those of pyrite, around 0.9&. increasing with depth in the water column. We therefore
This is clearly not the case; porewaters in sub-chemocline suggest that pyrite d56Fe values in Lac Pavin sediments
cores have d56Fe values near 0& (Table 2), indicating that do not reflect a kinetic isotope fractionation. However,
DIR is not the process driving to negative d56Fe values in our conclusion would only apply to ferruginous systems
pyrite. It must be noted that very high Fe concentrations and it is not in agreement with previous results obtained
in porewaters attest to extensive DIR (average Fe concen- on pyrites from sulfate- and sulfide rich sediments in Cali-
trations of 180 ± 70, 510 ± 80, 1170 ± 260 lM for cores fornia, which suggest that the apparent Fe isotope fraction-
at 60, 65, and 92 m, respectively; Table 2 and Fig. 5a). ation between aqueous Fe2+ and pyrite was >1.5&
The iron isotope fractionation typically associated with (Severmann et al., 2006). The discrepancy between the var-
DIR is likely not expressed in these porewaters because ious data published so far (either calculated, determined
Fe reduction is complete, similar to what is expected in con- experimentally or from natural systems) may reflect strong
tinental margin marine sediments where limited iron flux is pathway dependence of the isotopic fractionations involved
associated with high organic matter content. in pyrite formation.
The second scenario implies that negative d56Fe re- At Lac Pavin, negative and highly variable pyrite d56Fe
corded in pyrite results from Fe isotope fractionation asso- values in sub-chemocline cores may reflect that pyrite Fe
ciated with its precipitation either in the water column or in was sourced from an Fe(II) pool with variable and depleted
the sediment. Experimental studies show that a depletion of d56Fe values. In Lac Pavin, such variability and depletion
2.2& in pyrite can be produced during formation from only occurs in the water column within the vicinity of the
aqueous Fe2+ (Guilbaud et al., 2011). However, estimates oxycline. We propose that Fe-sulfides form in the water col-
of the isotopic effects of this process range from kinetically umn at the oxic–anoxic boundary (60 m depth) and trans-
controlled fractionations that enrich pyrite in the light Fe fer their Fe isotope signature to sedimentary pyrite with
isotopes (e.g., Severmann et al., 2006; Guilbaud et al., limited post formation iron isotope fractionation. Water
2011), to equilibrium fractionations, which yield pyrite column profiles show that sulfate reduction occurs within
extensively enriched in heavy isotopes (e.g., Polyakov the redox boundary (61–65 m depth), where the Fe isotope
et al., 2007). One of the main pyrite precursor, mackinawite gradient is the strongest (Table 1, Fig. 3b). Below these
(FeS), is also affected by kinetic Fe isotope fractionation depths sulfide concentrations are controlled by FeS solubil-
when precipitated at low pH (4; Butler et al., 2005; ity and sulfate levels are constant, presumably at concentra-
Guilbaud et al., 2010) but can reach equilibrium, with an tions below what microbial sulfate reducers can utilize.
enrichment in heavy Fe isotopes, at near-neutral pH condi- Consistent with this view, we found sulfate reduction rates
tions (Wu et al., 2012). At Lac Pavin, the presence of pyrite of <103 lmol L1 d1 at 55 m (oxic waters) and 70 m
with near crustal values in the oxic core (32 m), where pyrite depth, but notably higher rates (2  103 lmol L1 d1)
formation occurs in porewaters with moderate dissolved at 61 m, confirming that bacterial sulfate reduction occurs
ferrous iron concentrations (>30 lM Fe(II)) and low extent mainly at the redox boundary and is restricted there by
of pyritization, is not consistent with presence of large ki- the availability of sulfate. This scenario is supported by
netic fractionations in low temperature ferruginous aque- the results from a previous study on reduced sulfur species
ous systems. Additionally, assuming that Fe isotope distribution in the anoxic water column of Lac Pavin
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 457

(Bura-Nakic et al., 2009). From a comparison between tra- 4.2. Implications for Archean environment and biosphere
ditional and voltammetry S data, this study demonstrated
that free H2S/HS concentrations are extremely low and Work at Lac Pavin can help to improve our foundation
about 80% of the total sulfide occurs as iron sulfide colloids. for interpreting the sedimentary iron isotope record over
Therefore, constraints on water column S cycling indicate geologic time (Fig. 1). There is not strong evidence at Lac
that particulate iron sulfide formation occurs near the Pavin for large kinetic iron isotope fractionations during
chemocline. The idea that pyrite in the anoxic cores tracks pyrite formation under iron-rich conditions. We suggest
the dissolved iron isotope values of the zone of particulate that our observations at Pavin can be broadly applied to
iron sulfide formation near the chemocline is corroborated other ferruginous systems, modern and ancient. Iron iso-
by the similarity between the Fe isotope compositions tope variability at Pavin is most easily linked to iron redox
found in sediment pyrites (Fig. 4) and dissolved Fe(II) at cycling within the water column. Specifically, oxidation of a
the oxic–anoxic interface (Fig. 3b). large dissolved iron pool is the main process that induces
Down-core variability in pyrite d56Fe values over short iron isotope variability within the water column, and some
distances can be explained by expansion and contraction of this variability is likely transferred to the sediments via
of the thickness of the zone of iron oxidation in the water iron sulfide formation in the water column. Since there is
column and the relative depth of water-column sulfate a discrete redoxcline at Pavin, with fully oxic surface waters
reduction and FeS formation with respect to the d56Fe gra- above, there is ultimately quantitative oxidation and limited
dient. For instance, when oxic shallow waters are affected isotope variability in the bulk sediment.
by seasonal mixing down to 60 m depth, the iron gradient Using Lac Pavin as a guide, the oxidative side of the iron
at the chemocline is sharpened both in concentration and cycle likely imparted iron isotope variability to the Archean
isotope composition. This condition affects the d56Fe values sedimentary record. Work at Pavin supports the model that
of dissolved iron in the zone where the majority of sulfides enhanced sedimentary iron isotope variability in the Neo-
form (Bura-Nakic et al., 2009). As noted above, the large archean is linked to an extensive oxidative iron cycle but
variability in sediment pyrite values are difficult to explain one in which partial oxidation prevailed in broad swaths
if pyrite Fe isotope composition is inherited within the sed- of the shallow ocean. In this model, non-quantitative
iment, given mostly constant porewater Fe concentration behavior is the key to inducing large sediment iron isotope
and isotope profiles. We cannot, however, completely rule variability (Rouxel et al., 2005). Before the permanent rise
out with current constraints that below the chemocline of atmospheric oxygen [2.5–2.3 Ga (Bekker et al., 2004;
there is porewater pyrite formation from FeS precursors Konhauser et al., 2011)], iron oxidation in an upwelling
and that this process involves an isotope fractionation water mass could have been a protracted, incomplete pro-
(e.g., Severmann et al., 2006). cess—allowing for burial of iron oxides with positive
From X-ray diffraction (XRD) and Scanning Electron d56Fe values and leaving behind an isotopically depleted
Microscopy (SEM) analyses of Lac Pavin sediments, dissolved iron reservoir that could be transferred to the sed-
Viollier et al. (1997) identified spherical aggregates imentary record with or without associated fractionations
(20–30 lm) of bipyramidal pyrite crystals (0.5–1 lm) (as pyrite, carbonate or iron oxides). From this perspective,
termed framboids. These authors suggested that pyrite the decrease in iron isotope variability in the sedimentary
could form after diagenetic dissolution of proto-vivianite record around 2.45 billion years ago (Rouxel et al., 2005;
(ferrous phosphate) in the sediment because they were not Johnson et al., 2008b) marks the first full-scale and persis-
able to find any phosphate particles in their sediment sam- tent oxidation of the upper ocean. We further propose that
ple. This hypothesis is clearly not supported by latter stud- the appearance of pronounced negative iron isotope values
ies, showing that vivianite is the main authigenic mineral in in the sedimentary record around 3.0 billion years ago,
Lac Pavin sediments, with concentrations representing up compared to heavy or near crustal values that seem to char-
to 70 wt% of the total sediment (Schettler et al., 2007; Cos- acterize the earlier sedimentary iron isotope record (Fig. 1)
midis et al., 2014). In a recent and more detailed mineralog- reflect the first onset of large-scale partial oxidation of iron
ical study of Lac Pavin, pyrite was not readily identified by in response to an important evolution of photosynthetic
SEM or Transmission Electron Microscopy (TEM) tech- machinery. Photosynthesis, either oxygenic or anoxygenic
niques in either particulate matter from the water column (Cloud, 1973; Kappler et al., 2005), are likely to provide
or surface sediments (Cosmidis et al., 2014). This is likely means to oxidize the extensive amount of iron (>50% of dis-
due to the very low abundance of pyrite in these samples solved iron pool) needed to cause large iron isotope gradi-
(<0.2 wt%; see also Table 2) relative to diatoms, ferric ents (Rouxel et al., 2005). Locally, dissimilatory Fe
and ferrous phosphate and organic matter (dilution effect). reduction in sediment porewater and associated diagenetic
Consistent with this view, the authors noted that EXAFS reactions may have further enhanced Fe isotope variability
(Extended X-ray Absorption Fine Structure) spectra ob- in marine sediments. For instance, earlier studies based on
tained from synchrotron-based analyses were better fitted C and Fe isotope compositions suggested that Fe cycling re-
by addition of pyrite, particularly in a sediment trap placed corded in Archean carbonates largely reflect dissimilatory
at 67 m in the water column (just below the sulfate reduc- iron reduction (Heimann et al., 2010; Craddock and Dau-
tion zone). Better constraints on the extents porewater phas, 2011). Using Lac Paving as a guide, however, it seems
and water column pyrite formation at Lac Pavin are, there- an oxidation model can explain a large part of the iron iso-
fore, required to incontrovertibly link sediment pyrite val- tope variability observed in the Archean sediments. Banded
ues with water column iron cycling. Iron Formations have a large range of d56Fe values with
458 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

markedly positive and negative d56Fe values (Fig. 1). This ferrous iron oxidation in the water column and capture of
range can reflect partial oxidation and distillation of the remaining isotopically light reservoir (Rouxel et al., 2005).
dissolved Fe reservoir due to progressive removal of Fe oxi- Before the large-scale rise of atmospheric oxygen around
des to the sediment (e.g. Rouxel et al., 2005; Tsikos et al., the Great Oxidation Event at 2.4 Ga iron oxidation in
2010). Pyrites measured so far in Archean rocks are gener- upwelling water masses could have been a protracted pro-
ally contained in organic-rich sediments. In these settings, cess—allowing for burial of iron oxides with positive
one can expect extensive sediment microbial Fe reduction d56Fe values and leaving behind an isotopically depleted
with limited Fe isotope fractionation, similar to what we dissolved iron reservoir (Rouxel et al., 2005). With the
observed in Lac Pavin. establishment of a discrete redoxcline in the oceans quanti-
If we are correct, then the Archean data would finger- tative iron oxidation would have prevailed. Thus oxidation
print a fundamental shift in the oxidative capacity of the of the surface oceans would have prevented efficient separa-
ocean. This capacity would be regulated by photosynthesis. tion of ferric oxides with positive d56Fe values from the dis-
Therefore, we propose that the sedimentary iron isotope re- solved negative residue. In this light, the iron isotope record
cord pinpoints the timing of one of the most significant bio- provides support for the presence of only restricted ‘oxygen
logical evolutionary events in Earth’s history—the ability of oasis’ in the Archean rather than a fully oxic surface layer.
organisms to utilize sunlight to mediate electron transfer— Partial oxidation would be most likely in water masses with
at 3.0 Ga. Recently, novel phylogenomic approaches high rates of both upwelling and lateral transport, which is
looking at the collective evolutionary record of gene fami- consistent with the presence of the lightest sedimentary iron
lies and protein domain structures have suggested a period isotope values in highly organic rich systems that would be
of rapid evolutionary innovation during a genetic expan- fed by a strong nutrient fluxes (Rouxel et al., 2005). Linking
sion around 3.0 Ga (David and Alm, 2011; Wang et al., the iron isotope record to redox evolution provides a simple
2011). More focused phylogenomic reconstructions suggest explanation for why sulfidic, organic matter-rich shales, a
the emergence of cyanobacteria in the Mesoarchean sedimentary rock type relatively common throughout the
(Schirrmeister et al., 2013), contemporaneous with the first Earth’s history, would have unusually negative Fe isotope
appearance of appreciable atmospheric oxygen (Crowe values prior to the rise of atmospheric oxygen around 2.4
et al., 2013). Our work hints at a convergence in molecular billion years ago.
and geochemical estimates for the timing of the first major
diversification of life on Earth and the ability of organisms ACKNOWLEDGMENTS
to use sunlight to fuel electron transfer.
This work was partly funded by the ANR-EC2CO (META-
5. SUMMARY NOX project) and by the BQR program of IPGP. We are thankful
to Gil Michard, Magali Ader, Oanez Lebeau, Patrick Alberic, Oliv-
ier Rouxel, Anne-Catherine Lehours, Corinne Biderre-Petit, Julie
The present study shows strong dissolved iron isotope Cosmidis, Karim Benzerara, Donald Canfield, and Guillaume
variability in the water column of anoxic and ferruginous Morin for constructive discussions. Yves Gamblin and Antonio
Lac Pavin, with d56Fe values ranging from 2.1& to Vieira from the IPGP mechanics workshop are thanked for pro-
+0.3&. The low d56Fe values and strong Fe isotope gradi- ducing the 1 L syringe-based water sampler. The University Blaise
ent observed close to the oxic–anoxic boundary reflect the Pascal of Clermont-Ferrand allowed the use of Besse biological sta-
effect of Fe oxidation and precipitation, leaving residual fer- tion nearby the lake and provided boats and platform. The US Na-
rous iron depleted in heavy isotopes. tional Science Foundation (EAR-PF) and NASA Exobiology and
Compared with water column, there is limited Fe iso- Astrobiology programs supported the work of Planavsky and
tope variability in sediment porewaters, bulk sediments, Lyons.
and reactive Fe, with d56Fe values essentially near 0&.
The largest range of d56Fe in sediments is found in pyrites,
with values between 1.51& and +0.09&; average APPENDIX A. SUPPLEMENTARY DATA
0.93&), similar to those measured in dissolved iron at
the oxic–anoxic boundary of the water column. We suggest Supplementary data associated with this article can be
that iron sulfides precipitate at the redox boundary where found, in the online version, at http://dx.doi.org/10.1016/
sulfate is reduced to sulfide and that these sulfides record j.gca.2014.03.004.
the Fe isotope variability of the chemocline. However, we
cannot completely rule out porewater pyrite formation, REFERENCES
which could have had an accompanying kinetic iron isotope
Aeschbach-Hertig W., Hofer M., Kipfer R., Imboden D. I. and
fractionation. The limited iron isotope variability in pore-
Wieler R. (1999) Accumulation of mantle gases in a perma-
waters, bulk sediments, and reactive Fe is linked to near
nently stratified volcanic lake (Lac Pavin, France). Geochim.
quantitative oxidation at chemocline and quantitative Cosmochim. Acta 63, 3357–3372.
microbial iron reduction in porewaters. In Lac Pavin, there Aeschbach-Hertig W., Hofer M., Schmid M., Kipfer R. and
is not clear Fe isotope signal for microbial iron reduction. Imboden D. M. (2002) The physical structure and dynamics of
Using Lac Pavin as a guide for interpreting Archean re- a deep, meromictic crater lake (Lac Pavin, France). Hydrobi-
cords of ferruginous oceans, these findings support that ologia 487, 111–136.
variable and markedly negative iron isotopes values in sed- Albarède F. and Beard B. L. (2004) Analytical methods for non-
imentary pyrites and shales in the Archean reflect partial traditional isotopes. In Geochemistry of Non-Traditional Stable
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 459

Isotopes (eds. C. M. Johnson, B. L. Beard and F. Albarède). Bura-Nakic E., Viollier E., Jézéquel D., Thiam A. and Ciglenecki I.
Mineralogical Society of America and Geochemical Society, (2009) Reduced sulfur and iron species in anoxic water column
Washington, DC, pp. 113–152. of meromictic crater Lake Pavin (Massif Central, France).
Amblard C. and Bourdier G. (1990) The spring bloom of the Chem. Geol. 266, 320–326.
diatom Melosira italica subsp. subarctica in Lake Pavin: Butler I. B., Archer C., Vance D., Oldroyd A. and Rickard D.
biochemical, energetic and metabolic aspects during sedimen- (2005) Fe isotope fractionation on FeS formation in ambient
tation. J. Plank. Res. 12, 645–651. aqueous solution. Earth Planet. Sci. Lett. 236, 430–442.
Anbar A. D., Jarzecki A. A. and Spiro T. G. (2005) Theoretical Busigny V. and Dauphas N. (2007) Tracing paleofluid circulations
investigation of iron isotope fractionation between Fe(H2O)3+ 6 using iron isotopes: a study of hematite and goethite concre-
and Fe(H2O)2+ 6 : implications for iron stable isotope geochem- tions from the Navajo Sandstone (Utah, USA). Earth Planet.
istry. Geochim. Cosmochim. Acta 69, 825–837. Sci. Lett. 254, 272–287.
Anbar A. and Rouxel O. (2007) Metal stable isotopes in paleoce- Chapron E., Albéric P., Jézéquel D., Versteeg W., Bourdier J.-L.
anography. Annu. Rev. Earth Planet. Sci. 35, 717–746. and Sitbon J. (2010) Multidisciplinary characterisation of
Archer C. and Vance D. (2006) Coupled Fe and S isotope evidence sedimentary processes in a recent maar lake (Lake Pavin,
for Archean microbial Fe(III) and sulfate reduction. Geology French Massif Central) and implication for natural hazards.
34, 153–156. Nat. Hazards Earth Syst. Sci. 10, 1815–1827.
Assayag N., Jézéquel D., Ader M., Viollier E., Michard G., Prévot Cloud P. (1973) Paleoecological significance of banded iron-
F. and Agrinier P. (2008) Hydrological budget, carbon sources formation. Econ. Geol. 68, 1135–1143.
and biogeochemical processes in Lac Pavin (France): con- Cosmidis J., Benzerara K., Morin G., Busigny V., Lebeau O.,
straints from (d18O of water and (d13C of dissolved inotganic Jézéquel D., Noël V., Dublet G. and Othmane G. (2014)
carbon. Appl. Geochem. 23, 2800–2816. Biomineralization of mixed valence iron-phosphates in the
Balci N., Bullen T. D., Witte-Lien K., Shanks W. C., Motelica M. anoxic water column of Lake Pavin (Massif Central, France).
and Mandernack K. W. (2006) Iron isotope fractionation Geochim. Cosmochim. Acta 126, 78–96.
during microbially stimulated Fe(II) oxidation and Fe(III) Craddock P. R. and Dauphas N. (2011) Iron and carbon isotope
precipitation. Geochim. Cosmochim. Acta 70, 622–639. evidence for microbial iron respiration throughout the Archean.
Barling J. and Anbar A. D. (2004) Molybdenum isotope fraction- Earth Planet. Sci. Lett. 303, 121–132.
ation during adsorption by manganese oxides. Earth Planet. Crosby H. A., Johnson C. M., Roden E. E. and Beard B. L. (2005)
Sci. Lett. 217, 315–329. Coupled Fe(II)-Fe(III) electron and atom exchange as a
Beard B. L. and Johnson C. M. (1999) High precision iron isotope mechanism for Fe isotope fractionation during dissimilatory
measurements of terrestrial and lunar materials. Geochim. iron oxide reduction. Environ. Sci. Technol. 39, 6698–6704.
Cosmochim. Acta 63, 1653–1660. Crowe S. A., Døssing L. N., Beukes N. J., Bau M., Kruger S. J.,
Beard B.L. and Johnson C.M. (2004) Fe isotope variations in the Frei R. and Canfield D. E. (2013) Atmospheric Oxygen 3.0
modern and ancient Earth and other planetary bodies. In: C.M. billion years ago. Nature 501, 535–538.
Johnson, B.L. Beard and F. Albarède (Editors), Geochemistry Czaja A. D., Johnson C. M., Roden E. E., Beard B. L., Voegelin A.
of non-traditional stable isotopes. Reviews in Mineralogy and R., Nägler T. F., Beukes N. J. and Wille M. (2012) Evidence for
Geochemistry. Mineralogical Society of America and Geo- free oxygen in the Neoarchean ocean based on coupled iron-
chemical Society, Washington DC., pp. 319–357. molybdenum isotope fractionation. Geochim. Cosmochim. Acta
Beard B. L., Johnson C. M., Skulan J. L., Nealson K. H., Cox L. 86, 118–137.
and Sun H. (2003) Application of Fe isotopes to tracing the Dauphas N., Janney P. E., Mendybaev R. A., Wadhwa M., Richter
geochemical and biological cycling of Fe. Chem. Geol. 195, 87– F., Davis A. M., van Zuilen M., Hines R. and Foley C. N.
117. (2004) Chromatographic separation and multicollection-
Bekker A., Holland H. D., Wang P. L., Rumble D., Stein H. J., ICPMS analysis of iron – investigating mass-dependent and -
Hannah J. L., Coetzee L. L. and Beukes N. J. (2004) Dating the independent isotope effects. Anal. Chem. 76, 5855–5863.
rise of atmospheric oxygen. Nature 427, 117–120. Dauphas N. and Rouxel O. J. (2006) Mass spectrometry and
Belshaw N. S., Zhu X. K. and O’Nions R. K. (2000) High precision natural variations of iron isotopes. Mass. Spectrom. Rev. 25,
measurement of iron isotopes by plasma source mass spec- 515–550.
trometry. Int. J. Mass Spectrom. 197, 191–195. Dauphas N., Cates N. L., Mojzsis S. J. and Busigny V. (2007a)
Beukes N. J. and Gutzmer J. (2008) Origin and Paleoenvironmen- Identification of chemical sedimentary protoliths using iron
tal significance of major iron formations at the Archean- isotopes in the >3750 Ma Nuvvuagittuq supracrustal belt,
Paleoproterozoic boundary. Soc. Econ. Geol. Rev. 15, 5–47. Canada. Earth Planet. Sci. Lett. 254, 358–376.
Biderre-Petit C., Jézéquel D., Dugat-Bony E., Lopes F., Kuever J., Dauphas N., van Zuilen M., Busigny V., Lepland A., Wadhwa M.
Borrel G., Viollier E., Fonty G. and Peyret P. (2011) Identi- and Janney P. E. (2007b) Iron isotope, major and trace element
fication of microbial communities involved in the methane cycle characterization of early Archean supracrustal rocks from SW
of a freshwater meromictic lake. FEMS Microbiol. Ecol. 77, Greenland: Protolith identification and metamorphic overprint.
533–545. Geochim. Cosmochim. Acta 71, 4745–4770.
Bonhomme C., Poulin M., Vincßon-Leite B., Saad M., Groleau A., Dauphas N., Pourmand A. and Teng F.-Z. (2009) Routine isotopic
Jézéquel D. and Tassin B. (2011) Maintaining meromixis in analysis of iron by HR-MC-ICPMS: how precise and how
Lake Pavin (Auvergne, France): the key role of a sublacustrine accurate? Chem. Geol. 267, 175–184.
spring. C. R. Geosci. 343, 749–759. David L. A. and Alm E. J. (2011) Rapid evolutionary innovation
Buffle J., De Vitre R. R., Perret D. and Leppard G. G. (1989) during an Archaean genetic expansion. Nature 469, 93–96.
Physico-chemical characteristic of a colloidal iron phosphate Duan Y., Severmann S., Anbar A., Lyons T. W., Gordon G. W.
species formed at the oxic-anoxic interface of a eutrophic lake. and Sageman B. B. (2010) Isotopic evidence for Fe cycling and
Geochim. Cosmochim. Acta 53, 399–408. repartitioning in ancient oxygen-deficient settings: examples
Bullen T. D., White A. F., Childs C. W., Vivit D. V. and Schulz M. from black shales of the mid-to-late Devonian Appalachian
S. (2001) Demonstration of significant abiotic iron isotope basin. Earth Planet. Sci. Lett. 290, 244–253.
fractionation in nature. Geology 29, 699–702.
460 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

Fehr M. A., Anderson P. S., Halenius U. and Morth C.-M. (2008) Johnson C. M., Beard B. L. and Roden E. E. (2008b) The iron isotope
Iron isotope variations in Holocene sediments of the Gotland fingerprints of redox and biogeochemical cycling in modern and
Deep, Baltic Sea. Geochim. Cosmochim. Acta 72, 807–826. ancient Earth. Annu. Rev. Earth Planet. Sci. 36, 457–493.
Fehr M. A., Anderson P. S., Halenius U. and Mörth C.-M. (2010) Juvigné E. and Gilot E. (1986) Age et zones de dispersion de téphra
Iron enrichments and Fe isotopic composition of surface émises par les volcans du Montcineyre et du lac Pavin (Massif
sediments from the Gotland Deep, Baltic Sea. Chem. Geol. Central, France). Z. Dtsch. Geol. Ges. 137, 613–623.
277, 310–322. Kappler A., Pasquero C., Konhauser K. O. and Newman D. K.
Fossing H. and Jorgensen B. B. (1989) Measurement of bacterial (2005) Deposition of banded iron formations by anoxygenic
sulfate reduction in sediments: evaluation of a single-step phototrophic Fe(II)-oxidizing bacteria. Geology 33, 865–868.
chromium reduction method. Biogeochemistry 8, 205–222. Konhauser K. O., Lalonde S. V., Amskold L. and Holland H. D.
Frost C. D., von Blanckenburg F., Schoenberg R., Frost B. R. and (2007) Was there really an Archean phosphate crisis? Science
Swapp S. M. (2007) Preservation of Fe isotope heterogeneities 315, 1234.
during diagenesis and metamorphism of banded iron forma- Konhauser K. O., Lalonde S. V., Planavsky N. J., Pecoits E.,
tion. Contrib. Mineral. Petrol. 153, 211–235. Lyons T. W., Mojzsis S. J., Rouxel O. J., Barley M. E., Rosı̀ere
Glangeaud P. (1916) Le cratère-lac Pavin et le volcan de Monchalm C., Fralick P. W., Kump L. R. and Bekker A. (2011) Aerobic
(Puy-de-Dôme). C. R. Acad. Sci. 162, 428–430. bacterial pyrite oxidation and acid rock drainage during the
Guilbaud R., Butler I. B., Ellam R. M. and Rickard D. (2010) Fe Great Oxidation Event. Nature 478, 369–373.
isotope exchange between Fe(II)aq and nanoparticulate mack- Lehours A.-C., Evans P., Bardot C., Joblin K. and Fonty G. (2007)
inawite (FeSm) during nanoparticule growth. Earth Planet. Sci. Phylogenetic diversity of Archea and bacteria in the anoxic
Lett. 300, 174–183. zone of a meromictic lake (Lake Pavin, France). Appl. Environ.
Guilbaud R., Butler I. B. and Ellam R. M. (2011) Abiotic pyrite Microb. 73, 2016–2019.
formation produces a large Fe isotope fractionation. Science Lehours A.-C., Batisson I., Guedon A., Mailhot G. and Fonty G.
332, 1548–1551. (2009) Diversity of culturable bacteria, from the anaerobic zone
Habicht K. S., Gade M., Thamdrup B., Berg P. and Canfield D. E. of the meromictic Lake Pavin, able to perform dissimilatory-
(2002) Calibration of sulfate levels in the Archean Ocean. iron reduction in different in vitro conditions. Geomicrobiol. J.
Science 298, 2372–2374. 26, 212–223.
Halverson G. P., Poitrasson F., Hoffman P. F., Nédélec A., Montel Lopes F., Viollier E., Thiam A., Michard G., Abril G., Groleau A.,
J.-M. and Kirby J. (2011) Fe isotope and trace element Prévot F., Carrias J.-F. and Jézéquel D. (2011) Biogeochemical
geochemistry of the Neoproterozoic syn-glacial Rapitan iron modeling of anaerobic vs. aerobic methane oxidation in a
formation. Earth Planet. Sci. Lett. 309, 100–112. meromictic crater lake (Lake Pavin, France). Appl. Geochem.
Heimann A., Johnson C. M., Beard B. L., Valley J. W., Roden E. 26, 1919–1932.
E., Spicuzza M. J. and Beukes N. J. (2010) Fe, C, and O isotope Lyons T. W. and Severmann S. (2006) A critical look at iron
compositions of banded iron formation carbonates demon- paleoredox proxies: new insights from modern euxinic marine
strate a major role for dissimilatory iron reduction in 2.5 Ga basins. Geochim. Cosmochim. Acta 70, 5698–5722.
marine environments. Earth Planet. Sci. Lett. 294, 8–18. Malinovsky D. N., Stenberg A., Rodushkin I., Andren H., Ingri J.,
Holland H. D. (1984) The Chemical Evolution of the Atmosphere Ohlander B. and Baxter D. C. (2003) Performance of high
and Oceans. Princeton Series in Geochemistry. Princeton Uni- resolution MC-ICP-MS for Fe isotope ratio measurements in
versity Press, Princeton, NJ, 598 pp. sedimentary geological materials. J. Anal. At. Spectrom. 18,
Hongve D. (1997) Cycling of iron, manganese, and phosphate in a 687–695.
Meromictic Lake. Limnol. Oceanogr. 42, 635–647. Malinovsky D. N., Rodyushkin I. V., Scherbakova E. P., Ponter
Huerta-Diaz M. A. and Morse J. W. (1990) A quantitative method C., Ohlander B. and Ingri J. (2005) Fractionation of Fe isotopes
for determination of trace metal concentrations in sedimentary as a result of redox processes in a basin. Geochem. Int. 43, 797–
pyrite. Mar. Chem. 29, 119–144. 803.
Icopini G. A., Anbar A. D., Ruebush S. S., Tien M. and Brantley Martin J.-M. (1985) The Pavin Crater Lake. In: W. Stumm
S. L. (2004) Iron isotope fractionation during microbial (Editor), Chemical processes in lakes. Reviews in Mineralogy
reduction of iron: the importance of adsorption. Geology 32, and Geochemistry. John Wiley & Sons, New York, pp. 169-188.
205–208. Matthews A., Morgans-Bell H. S., Emmanuel S., Jenkyns H. C.,
Jarzecki A. A., Anbar A. D. and Spiro T. G. (2004) DFT analysis of Erel Y. and Halicz L. (2004) Controls on iron-isotope
Fe(H2O)3+ 6 and Fe(H2O)2+ 6 structure and vibrations: implica- fractionation in organic-rich sediments (Kimmeridge Clay,
tions for isotope fractionation. J. Phys. Chem. A 108, 2726–2732. Upper Jurassic, southern England). Geochim. Cosmochim. Acta
Jézéquel D., Sarazin G., Prévot F., Viollier E., Groleau A., 68, 3107–3123.
Agrinier P., Albéric P., Binet S., Bergonzini L. and Michard G. Meybeck M., Martin J. M. and Olive P. (1975) Géochimie des eaux
(2011) Bilan hydrique du lac Pavin – water balance of the Lake et des sédiments de quelques lacs volcaniques du Massif Central
Pavin. Rev. Sci. Nat. d’Auvergne 74, 75–96. francßais. Verh. Internat. Verein. Limnol. 19, 1150–1164.
Johnson C. M., Skulan J. L., Beard B. L., Sun H., Nealson K. H. Michard G., Viollier E., Jézéquel D. and Sarazin G. (1994)
and Braterman P. S. (2002) Isotopic fractionation between Geochemical study of a crater lake: Lac Pavin, France –
Fe(III) and Fe(II) in aqueous solutions. Earth Planet. Sci. Lett. identification, location and quantification of the chemical
195, 141–153. reactions in the lake. Chem. Geol. 115, 103–115.
Johnson C. M., Beard B. L., Beukes N. J., Klein C. and O’Leary J. Michard G., Jézéquel D. and Viollier E. (2003) Vitesses des
M. (2003) Ancient geochemical cycling in the Earth as inferred réactions de dissolution et précipitation au voisinage de
from Fe isotope studies of banded iron formations from the l’interface oxydo-réducteur dans un lac méromictique: Le lac
Transvaal Craton. Contrib. Mineral. Petrol. 144, 523–547. Pavin (Puy-de-Dôme, France). Revue des Sciences de l’Eau 16,
Johnson C. M., Beard B. L., Klein C., Beukes N. J. and Roden E. 199–218.
E. (2008a) Iron isotopes constrain biologic and abiologic Olive P. and Boulègue J. (2004) Biogeochemical study of a
processes in banded iron formation genesis. Geochim. Cosmo- meromictic lake: Pavin lake, France. Geomorphologie 4, 305–
chim. Acta 72, 151–169. 316.
 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462 461

Planavsky N., Rouxel O., Bekker A., Shapiro R., Fralick P. and Steinhoefel G., von Blanckenburg F., Horn I., Konhauser K. O.,
Knudsen A. (2009) Iron-oxidizing microbial ecosystems thrived Beukes N. J. and Gutzmer J. (2010) Deciphering formation
in late Paleoproterozoic redox-stratified oceans. Earth Planet. processes of banded iron formations from the Transvaal and
Sci. Lett. 286, 230–242. the Hamersley successions by combined Si and Fe isotope
Planavsky N., McGoldrick P., Scott C. T., Li C., Reinhard C. T., analysis using UV femtosecond laser ablation. Geochim. Cos-
Kelly A. E., Chu X., Bekker A., Love G. D. and Lyons T. W. mochim. Acta 74, 2677–2696.
(2011) Widespread iron-rich conditions in the mid-Proterozoic Strelow F. W. E. (1980) Improved separation of iron from copper
ocean. Nature 477, 448–451. and other elements by anion-exchange chromatography on a
Planavsky N., Rouxel O., Bekker A., Hofmann A., Little C. T. S. 4% cross-linked resin with high concentrations of hydrochloric
and Lyons T. (2012) Iron isotope composition of some Archean acid. Talanta 27, 727–732.
and Proteroizoic iron formations. Geochim. Cosmochim. Acta Sumner D. Y. (1997) Carbonate precipitation and oxygen strati-
180, 158–169. fication in late Archean seawater as deduced from facies and
Polyakov V. B., Clayton R. N., Horita J. and Mineev S. D. (2007) stratigraphy of the Gamohaan and Frisco formations, Trans-
Equilibrium iron isotope fractionation factors of minerals: vaal Supergroup, South Africa. Am. J. Sci. 297, 455–487.
reevaluation from the data of nuclear inelastic resonant X-ray Takayanagi K. and Cossa D. (1997) Vertical distribution of Sb(III)
scattering and Mössbauer spectroscopy. Geochim. Cosmochim. and Sb(V) in Pavin Lake, France. Water Res. 31, 671–674.
Acta 71, 3833–3846. Taylor S. R. and McLennan S. M. (1985) The continental crust: its
Poulton S. W. and Canfield D. E. (2011) Ferruginous conditions: a composition and evolution. Blackwell Scientific Publications,
dominant feature of the ocean through Earth’s history. Oxford, 312 p.
Elements 7, 107–112. Taylor P. D. P., Maeck R. and De Bievre P. (1992) Determination
Rouxel O. J., Dobbek N., Ludden J. and Fouquet Y. (2003) Iron of the absolute isotopic composition and atomic weight of a
isotope fractionation during oceanic crust alteration. Chem. reference sample of natural iron. Int. J. Mass Spectrom. 121,
Geol. 202, 155–182. 111–125.
Rouxel O. J., Bekker A. and Edwards K. J. (2005) Iron isotope Teutsch N., Schmid M., Müller B., Halliday A. N., Bürgmann H.
constraints on the Archean and Paleoproterozoic ocean redox and Wehrli B. (2009) Large iron isotope fractionation at the
state. Science 307, 1088–1091. oxic-anoxic boundary in Lake Nyos. Earth Planet. Sci. Lett.
Schettler G., Schwab M. J. and Stebich M. (2007) A 700 years 285, 52–60.
record of climate change based on geochemical and palynolog- Tsikos H., Matthews A., Erel Y. and Moore J. M. (2010) Iron
ical data from varved sediments (Lac Pavin, France). Chem. isotopes constrain biogeochemical redox cycling of iron and
Geol. 240, 11–35. manganese in a Paleoproterozoic stratified basin. Earth Planet.
Schirrmeister B. E., de Vos J. M., Antonelli A. and Bagheri H. C. Sci. Lett. 298, 125–134.
(2013) Evolution of multicellularity coincided with increased Valaas Hyslop E., Valley J. W., Johnson C. M. and Beard B. L.
diversification of cyanobacteria and the Great Oxidation Event. (2008) The effects of metamorphism on O and Fe isotope
Proc. Natl. Acad. Sci. USA 110, 1791–1796. compositions in the Biwabik Iron Formation, northern Min-
Schoenberg R. and von Blanckenburg F. (2005) An assessment of nesota. Contrib. Mineral. Petrol. 155, 313–328.
the accuracy of stable Fe isotope ratio measurements on Viollier E., Jézéquel D., Michard G., Pèpe M., Sarazin G. and
samples with organic and inorganic matrices by high-resolution Albéric P. (1995) Geochemical study of a crater lake (Pavin
multicollector ICP-MS. Int. J. Mass Spectrom. 242, 257–275. Lake, France): trace-element behaviour in the monimolimnion.
Severmann S., Johnson C. M., Beard B. L. and McManus J. (2006) Chem. Geol. 125, 61–72.
The effect of early diagenesis on the Fe isotope compositions of Viollier E., Michard G., Jézéquel D., Pèpe M. and Sarazin G.
porewaters and authigenic minerals in continental margin (1997) Geochemical study of a crater lake: Lake Pavin, Puy de
sediments. Geochim. Cosmochim. Acta 70, 2006–2022. Dôme, France. Constraints afforded by the particulate matter
Severmann S., Lyons T. W., Anbar A. D., McManus J. and distribution in the element cycling within the lake. Chem. Geol.
Gordon G. (2008) Modern iron isotope perspective on the 142, 225–241.
benthic iron shuttle and the redox evolution of ancient oceans. Viollier E., Inglett P. W., Hunter K., Roychoudhury A. N. and
Geology 36, 487–490. Van Cappellen P. (2000) The ferrozine method revisited: Fe(II)/
Skulan J. L., Beard B. L. and Johnson C. M. (2002) Kinetic and Fe(III) determination in natural waters. Appl. Geochem. 15,
equilibrium Fe isotope fractionation between aqueous Fe(III) 785–790.
and hematite. Geochim. Cosmochim. Acta 66, 2995–3015. Viollier E., Thiam A., Darmoul Y., Neubert N., Nägler T., Lopes
Song L., Liu C.-Q., Wang Z.-L., Zhu X., Teng Y., Liang L., Tang F., Michard G., Jézéquel D. (2014) Molybdenum isotopes
S. and Li J. (2011) Iron isotope fractionation during biogeo- fractionation in a crater lake: understanding the signature of
chemical cycle: information from suspended particulate matter biogeochemical processes interplay. Submitted.
(SPM) in Aha Lake and its tributaries, Ghizhou, China. Chem. Virtasalo J., Whitehouse M. J. and Kotilainen A. T. (2013) Iron
Geol. 280, 170–179. isotope heterogeneity in pyrite fillings of Holocene worm
Staubwasser M., von Blanckenburg F. and Schoenberg R. (2006) burrows. Geology 41, 39–42.
Iron isotopes in the early marine diagenetic iron cycle. Geology von Blanckenburg F., Mamberti M., Schoenberg R., Kamber B. S.
34, 629–632. and Webb G. E. (2008) The iron isotope composition of
Staubwasser M., Schoenberg R., von Blankenburg F., Krüger S. microbial carbonate. Chem. Geol. 249, 113–128.
and Pohl C. (2013) Isotope fractionation between dissolved and Wang M., Jiang Y. Y., Kim K. M., Qu G., Ji H. F., Mittenthal J.
suspended particulate Fe in the oxic and anoxic water column E., Zhang H. Y. and Caetano-Anollés G. (2011) A universal
of the Baltic Sea. Biogeosciences 10, 233–245. molecular clock of protein folds and its power in tracing the
Steinhoefel G., Horn I. and von Blanckenburg F. (2009) Micro- early history of aerobic metabolism and planet oxygenation.
scale tracing of Fe and Si isotope signatures in banded iron Mol. Biol. Evol. 28, 567–582.
formation using femtosecond laser ablation. Geochim. Cosmo- Welch S. A., Beard B. L., Johnson C. M. and Braterman P. S.
chim. Acta 73, 5343–5360. (2003) Kinetic and equilibrium Fe isotope fractionation
462 V. Busigny et al. / Geochimica et Cosmochimica Acta 133 (2014) 443–462

between aqueous Fe(II) and Fe(III). Geochim. Cosmochim. Acta proterozoic Earth: constraints from iron isotope variations in
67, 4231–4250. sedimentary rocks from the Kaapvaal and Pilbara Cratons.
Weyer S. and Schwieters J. B. (2003) High precision Fe isotope Chem. Geol. 218, 135–169.
measurements with high mass resolution MC-ICPMS. Int. J. Yoshiya K., Nishizawa M., Sawaki Y., Ueno Y., Komiya T.,
Mass Spectrom. 226, 355–368. Yamada K., Yoshida N., Hirata T., Wada H. and Maruyama
Wu L., Druschel G., Findlay A., Beard B. L. and Johnson C. M. S. (2012) In situ iron isotope analyses of pyrite and organic
(2012) Experimental determination of iron isotope fractiona- carbon isotope ratios in the Fortescue Group: metabolic
tions among Fe2+ aq -FeSaq-Mackinawite at low temperatures: variations of a Late Archean ecosystem. Precamb. Res. 212,
implications for the rock record. Geochim. Cosmochim. Acta 89, 169–193.
46–61.
Yamaguchi K. E., Johnson C. M., Beard B. L. and Ohmoto H.
Associate editor: Nicolas Dauphas
(2005) Biogeochemical cycling of iron in the Archean-Paleo-
APPENDIX A. Procedure for the analysis of accurate Fe isotope composition

Iron separation by ion chromatography

Before isotopic analysis, Fe must be separated from other elements contained in the

sample matrix that can affect the mass bias in the mass spectrometer and/or form complex

compounds creating interference on Fe masses. Major isobaric interferences may derive from

(i) neighbor elements such as Cr+ and Ni+ on the masses 54 and 58 respectively, or (ii) sample

matrix elements such as Ca that lead to molecular interferences such as 40Ca16O+ and 40Ca17O+

interfering on the masses 56 and 57, respectively. Doubly charged species such as Pd2+, Cd2+,

Sn2+ can also induce interferences at masses 54, 56 and 57. The most commonly used method

to separate Fe from other elements is based on anion exchange chromatography in HCl

medium (Strelow, 1980). Over the last decade, several laboratories have developed similar

techniques for Fe separation by chromatography (Beard and Johnson, 1999; Malinovsky et

al., 2003; Rouxel et al., 2003; Poitrasson et al., 2004; Dauphas et al., 2004; Schoenberg and

von Blanckenburg, 2005). We follow essentially the chemical protocol laid out by Dauphas et

al. (2004).

Sample chemistry was implemented in a clean laboratory at the Laboratory of

Geochemistry and Cosmochemistry, IPGP. Water samples were loaded in Teflon beakers and

evaporated to dryness. They were twice acidified with mixtures of HCl-HNO3 and

evaporated, to ensure that all Fe is in the ferric state before loading on the resin. Bulk

sediment and volcanic rock samples analyzed over the course of this study were digested in a

sequence of acid mixtures including HF, HCl and HNO3. All samples were finally dissolved

in 6 M HCl before ion exchange chromatography. The following ion chromatography

procedure was performed twice for each sample to ensure separation of Fe from matrix

  1  
elements. Bio-Rad Poly-Prep columns were filled with 1 mL anion exchange resin (AG1-X8

200-400mesh chloride form). The resin was cleaned three times with 10 ml H2O and 5 ml 1 M

HNO3. It was then preconditioned in HCl medium by running 10 ml H2O, 10 ml 0.4 M HCl, 5

ml H2O and 2 ml 6 M HCl. Half of the sample solution (250 µl) was loaded on the column in

6 M HCl (containing between 10 and 500 µg of Fe). Matrix elements were eluted with 8 ml 6

M HCl, whereas Fe(III) is strongly adsorbed on the resin and is quantitatively retained. Fe

was subsequently eluted in 10 ml 0.4 M HCl, with a procedural yield of >94%. The Fe blank

level of the present procedure has been evaluated by systematic analyses of one blank in each

sample series, prepared as described above but without any sample powder or solution. The

blank was always below 40 ng Fe (average ~30 ng), thus representing less than 0.4 % of the

bulk Fe.

Mass spectrometry

Iron concentrations and isotopic compositions were measured using a Neptune

ThermoFischer MC-ICP-MS (Multiple Collector Inductively Coupled Plasma Mass

Spectrometer) at the Laboratory of Geochemistry and Cosmochemistry. The Neptune is

equipped with eight adjustable Faraday cups and one fixed center cup. Iron isotopes were

measured simultaneously at masses 54, 56, 57 and 58, while the contributions of Cr and Ni on

masses 54 and 58 were monitored and corrected for by using ion intensities measured at

masses 53 and 60, respectively. Masses 53, 54, 56, 57, 58 and 60 were collected in the

Faraday cups Low 3, Low 2, C, High 1, High 2 and High 4, respectively. A drawback in the

use of the MC-ICP-MS is that polyatomic argide interferences are produced from Ar plasma

(Weyer and Schwieters, 2003; Anbar, 2004; Dauphas and Rouxel, 2006; Dauphas et al., 2009)

. These are mainly 40Ar14N+, 40Ar16O+, 40Ar16OH+ and 40Ar18O+ interfering at m/z 54, 56, 57 and

  2  
 58, respectively. Iron isotopes were fully resolved from argide interferences using the high-

resolution mode of the Neptune (for details see Weyer and Schwieters, 2003). This high-

resolution mode allows us to achieve a resolving power with m/Δm of ~12,500. Sample

solutions were nebulized and introduced using two different systems: the ThermoFinnigan

stable introduction system (SIS) and the ESI Apex-HF desolvating apparatus. We found that

the signal intensity was more stable with the SIS system but the Apex-HF apparatus gave

better sensitivity (~3 times more signal) and better precision on the Fe isotope measurement,

in agreement with previous results from another study (Schoenberg and von Blanckenburg,

2005). Thus all measurements presented herein were carried out with the Apex-HF apparatus.

The samples were analyzed in 0.3 M HNO3 at a concentration of ~0.6 ppm Fe.

Nebuliser uptake rate was between 40 to 60   µl/min. Signal intensity on the mass 56 was

between 10 to 15 volts with the high-mass resolution mode. Fe isotope measurements were

performed only when a flat Fe shoulder-peak plateau with a width > 200 ppm was achieved.

The analytical routine included baseline correction measured for 120 s before each

acquisition. Each sample measurement consisted of 1 block of 20 cycles with 7.2 seconds

duration. The take up time before analysis and wash time after analysis were fixed at 70 and

120 seconds respectively.  

Instrumental mass discrimination was corrected for using the conventional sample-

standard bracketing (SSB) approach (Belshaw et al., 2000; Beard et al., 2003; Rouxel et al.,

2003; Albarède and Beard, 2004). The SSB method was shown to give results as accurate and

precise as the Cu-dopping method for correction of instrumental mass bias during Fe isotope

measurement (Schoenberg and von Blanckenburg, 2005; Dauphas et al., 2009). The 56Fe/54Fe

and 57Fe/54Fe ratios were expressed in the usual δ notation in per mil (‰) as,

$ '
δ 56Fe = &( 56
Fe/ 54 Fe ) (
/ 56
Fe/ 54 Fe
) − 1) × 1000  
% sample standard (

€
  3  
 $ '
δ 57Fe = &( 57
Fe/ 54 Fe ) (
/ 57
Fe/ 54 Fe
) − 1) × 1000  
% sample standard (

where the standard is IRMM-014, a pure synthetic Fe metal from the Institute for Reference

€ Materials and Measurements (Taylor et al., 1992). A comparison between two solutions of

IRMM-014 prepared respectively in IPG Paris and at the Origins Laboratory of the University

of Chicago showed homogeneous Fe isotopic composition, with δ56Fe of 0.021± 0.032 ‰ and

δ57Fe of -0.017 ± 0.064 ‰ (2σ; n = 27).

Accuracy and analytical precision

Three water samples collected in Lac Pavin were selected to test if they could be

affected by any matrix effect during column chemistry and/or Fe isotope measurement. In

order to gauge matrix effects, we replaced Fe contained in the water samples with Fe that has

a known isotope composition. After preparation of the selected samples in 6 M HCl medium

as described in the experimental section, they were passed through a first column. The matrix

was recovered (6 M HCl fraction) and passed through a second column to get rid of any

potential residual Fe. The Fe-free matrix was doped with 50 µg of IRMM-014 (i.e. our Fe

standard from the Institute for Reference Materials and Measurements). These solutions of

Fe-free matrix+IRMM-014 were processed twice through column chemistry and analyzed on

the Neptune as described above. The results are presented in Table A.1. The three solutions of

Fe-free matrix+IRMM-014 show Fe isotope compositions undistinguishable, within error,

from pure IRMM-014 standard (i.e. δ56Fe ~ 0‰). Iron quantification by MC-ICP-MS also

indicates that the final solution contained > 47 µg Fe thus yielding more than 94% of the bulk

Fe initially introduced. These data illustrate that the present procedure is not affected by any

matrix effect and that the yield is near 100 %.

  4  
 We assessed the accuracy and analytical precision through the analysis of multiple

preparations of individual geostandards. Such replicate analyses take into account not only the

external precision of the technique, but various parameters such as the nature of the sample

and variations due to chemical separation and handling. Because no standards have been

characterized for Fe isotope composition of lakes or river waters, we analyzed international

rock geostandards (Dauphas and Rouxel, 2006; Craddock and Dauphas, 2010). Specifically,

we used the geostandards (i) IF-G: a 3.8 Ga old BIF (Banded Iron Formation) from Isua

(Greenland), (ii) BCR-2: a basalt from Columbia River (Portland, Oregon), (iii) AC-E: a

granite from Ailsa Craig Island (SW Scotland). These 3 geostandards represent various

lithologies, thus allowing us to test further if any matrix effect may occur. Interestingly they

cover a large range of Fe contents from high (39.10 wt%) to low (1.77 wt%) concentration,

providing an opportunity to test the sensitivity of the technique. Iron isotope compositions of

IF-G, BCR-2 and AC-E geostandards measured over five years in our laboratory are shown in

Figure A.1 and their weighted averages are given in Table A.1. Mean δ56Fe values of IF-G

(102 analyses of 23 solutions), BCR-2 (59 analyses of 10 solutions) and AC-E (29 analyses of

5 solutions) were 0.637 ± 0.049, 0.082 ± 0.060, and 0.337 ± 0.048 ‰ respectively (2SD, i.e.

Standard Deviation), in good agreement with available data for these geostandards (e.g. 0.647

± 0.025 ‰, 0.091 ± 0.025 ‰ and 0.322 ± 0.025 ‰ in Dauphas et al., 2009; see also Fe

standards compilation in Dauphas and Rouxel, 2006; Craddock and Dauphas, 2010). These

results demonstrate the accuracy of our measurements. The long-term reproducibility on the

mass spectrometer was also evaluated from the analyses of a pure Fe solution that was

measured regularly over a 5-years period. This "in house" solution, named ALD, is a solution

from ALDRICH (catalog number 35,630-1), which was diluted to 0.6 ppm in 0.3 M HNO3, in

order to match IRMM-014 Fe concentration. Thirty-six measurements of ALD gave δ56Fe and

δ57Fe values of 0.373 ± 0.054 and 0.546 ± 0.120 ‰ (2SD), respectively (Fig. A.1). The

  5  
reproducibility on δ56Fe and δ57Fe for ALD is similar to the reproducibility obtained on

several solutions of IF-G, BCR-2 and AC-E, which represent different Fe chemical separates

after chromatography. This indicates that the chemistry does not introduce any variability in

the Fe isotope composition. The accuracy seems to be as good as the reproducibility, i.e.

better than 0.06 ‰ for δ56Fe and 0.12 ‰ for δ57Fe (2SD). The larger uncertainty observed on
57
the Fe/54Fe ratio relative to the 56
Fe/54Fe ratio is likely a consequence of the greater

difference in the isotope abundances. A three-isotope plot for the standards together with all

Lac Pavin samples (n=141) analyzed in the course of this study is presented in Figure A.2. All

data points plot on a single mass fractionation line, with a slope of 1.480 and an intercept

value (0.005) not significantly different from 0. This again illustrates the reliability of our Fe

isotope measurements.

Collection of anoxic water samples for Fe isotope analyses

The collection of water samples is devoted to the analysis of either dissolved species

or particulate matter. The separation of dissolved species from particles is usually performed

through a filtration process where the particles are retained on a filter. When anoxic water

samples are filtered under atmospheric conditions, we must make sure that redox sensitive

elements initially dissolved in water do not precipitate on the filter. For instance, Fe can be

highly concentrated as dissolved Fe(II) in anoxic water. When samples are taken up to the

surface, they can be contaminated by atmospheric oxygen and Fe(III) precipitate may form.

Partial precipitation of Fe(III)-oxyhydroxyde from dissolved Fe(II) leads to a strong isotope

fractionation up to ~3‰ in 56Fe/54Fe ratio (Johnson et al., 2002; Welch et al., 2003; Jarzecki

et al., 2004; Anbar et al., 2005). In the present work, Fe-rich water samples from the anoxic

zone (75 m depth) of Lac Pavin have been collected in November 2006 using a lab-made 50

  6  
ml PE syringe sampler (Viollier et al., 1995) and filtered directly in situ (0.2 µm Millipore

filter with Luer connexion) from the syringe. Two kinds of filtration were performed either (i)

at depth during water collection (i.e. during introduction of water into the syringe) or (ii)

under atmospheric condition just after taken up the syringe to the surface (Table A.2). Both

methods lead to identical Fe concentrations and isotope compositions for the dissolved Fe

(samples LP2006-7 and LP2006-8 were filtered at depth, while samples LP2006-5 and

LP2006-6 were filtered in surface just after retrieving the sampler). Accordingly, filtration at

the surface does not modify Fe concentrations or isotope compositions when sampling anoxic

waters.

REFERENCES

Albarède F. and Beard B.L. (2004) Analytical methods for non-traditional isotopes. In: C.M.
Johnson, B.L. Beard and F. Albarède (Editors), Geochemistry of non-traditional stable
isotopes. Mineralogical Society of America and Geochemical Society, Washington
DC., pp. 113-152.
Anbar A.D. (2004) Iron stable isotopes: beyond biosignatures. Earth Planet. Sci. Lett. 217,
223-236.
Anbar A.D., Jarzecki A.A. and Spiro T.G. (2005) Theoretical investigation of iron isotope
fractionation between Fe(H2O)63+ and Fe(H2O)62+: Implications for iron stable isotope
geochemistry. Geochim. Cosmochim. Acta 69, 825-837.
Beard B.L. and Johnson C.M. (1999) High precision iron isotope measurements of terrestrial
and lunar materials. Geochim. Cosmochim. Acta 63, 1653-1660.
Beard B.L., Johnson C.M., Skulan J.L., Nealson K.H., Cox L. and Sun H. (2003) Application
of Fe isotopes to tracing the geochemical and biological cycling of Fe. Chem. Geol.
195, 87-117.
Belshaw N.S., Zhu X.K. and O'Nions R.K. (2000) High precision measurement of iron
isotopes by plasma source mass spectrometry. Int. J. Mass Spectrom. 197, 191-195.
Craddock P. and Dauphas N. (2010) Iron isotopic compositions of geological reference
material and chondrites. Geostandards and Geoanalytical Research 35, 101-123.
Dauphas N., Janney P.E., Mendybaev R.A., Wadhwa M., Richter F., 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., Pourmand A. and Teng F.-Z. (2009) Routine isotopic analysis of iron by HR-
MC-ICPMS: How precise and how accurate? Chem. Geol. 267, 175-184.
Dauphas N. and Rouxel O.J. (2006) Mass spectrometry and natural variations of iron
isotopes. Mass. Spectrom. Rev. 25, 515-550.

  7  
Jarzecki A.A., Anbar A.D. and Spiro T.G. (2004) DFT analysis of Fe(H2O)63+ and Fe(H2O)62+
structure and vibrations: implications for isotope fractionation. J. Phys. Chem. A. 108,
2726-2732.
Johnson C.M., Skulan J.L., Beard B.L., Sun H., Nealson K.H. and Braterman P.S. (2002)
Isotopic fractionation between Fe(III) and Fe(II) in aqueous solutions. Earth Planet.
Sci. Lett. 195, 141-153.
Malinovsky D.N., Stenberg A., Rodushkin I., Andren H., Ingri J., Ohlander B. and Baxter
D.C. (2003) Performance of high resolution MC-ICP-MS for Fe isotope ratio
measurements in sedimentary geological materials. J. Anal. At. Spectrom. 18, 687-
695.
Poitrasson F., Halliday A.N., Lee D.-C., Levasseur S. and Teutsch N. (2004) Iron isotope
differences between Earth, Moon, Mars and Vesta as possible records of contrasted
accretion mechanisms. Earth Planet. Sci. Lett. 223, 253-266.
Rouxel O., Dobbek N., Ludden J. and Fouquet Y. (2003) Iron isotope fractionation during
oceanic crust alteration. Chem. Geol. 202, 155-182.
Schoenberg R. and von Blanckenburg F. (2005) An assessment of the accuracy of stable Fe
isotope ratio measurements on samples with organic and inorganic matrices by high-
resolution multicollector ICP-MS. Int. J. Mass Spectrom. 242, 257-275.
Strelow F.W.E. (1980) Improved separation of iron from copper and other elements by anion-
exchange chromatography on a 4% cross-linked resin with high concentrations of
hydrochloric acid. Talanta 27, 727-732.
Taylor P.D.P., Maeck R. and De Bievre P. (1992) Determination of the absolute isotopic
composition and atomic weight of a reference sample of natural iron. Int. J. Mass
Spectrom. 121, 111-125.
Viollier E., Jézéquel D., Michard G., Pèpe M., Sarazin G. and Albéric P. (1995) Geochemical
study of a crater lake (Pavin Lake, France): Trace-element behaviour in the
monimolimnion. Chem. Geol. 125, 61-72.
Welch S.A., Beard B.L., Johnson C.M. and Braterman P.S. (2003) Kinetic and equilibrium Fe
isotope fractionation between aqueous Fe(II) and Fe(III). Geochim. Cosmochim. Acta
67, 4231-4250.
Weyer S. and Schwieters J.B. (2003) High precision Fe isotope measurements with high mass
resolution MC-ICPMS. Int. J. Mass Spectrom. 226, 355-368.

  8  
Table A.1
Average Fe isotope composition of geostandards and analytical tests performed on water samples.
Type Sample Description Fe δ56Fe δ57Fe
(wt%) (‰) (‰)

0.637 ± 0.049 0.943 ± 0.100

Banded Iron Formation IF-G SARM Geostandard 39.10
(n=102) (n=102)
0.082 ± 0.060 0.125 ± 0.098
Basalt BCR-2 USGS Geostandard 9.66
(n=59) (n=59)
0.337 ± 0.048 0.487 ± 0.122
Granite AC-E USGS Geostandard 1.77
(n=29) (n=29)
0.373 ± 0.054 0.546 ± 0.120
"Home" standard ALD Aldrich Fe solution -
(n=36) (n=36)
IRMM/LP2006- IRMM-014 in LP2006-
Reference - 0.043 ± 0.060 0.062 ± 0.132
1D 1 matrix
IRMM/LP2006- IRMM-014 in LP2006-
Reference - 0.004 ± 0.052 0.031 ± 0.102
2D 2 matrix
IRMM/LP2006- IRMM-014 in LP2006-
Reference - 0.032 ± 0.070 0.054 ± 0.086
3D 3 matrix
Uncertainties correspond to 2SD (standard deviation).
IF-G: distributed by the SARM (Service d'Analyse des Roches et des Minéraux) of the CRPG (Centre de
Recherche Pétrographique et Géochimique) in Nancy, France
BCR-2 and AC-E: distributed by the USGS (United States Geological Survey).
ALD: pure Fe solution (ALDRICH: catalog No 35,630-1; 9833mg/ml of Fe in 5% HNO3).
 
 
 
 
Table A.2
Depth, Fe concentration and Fe isotope composition of water samples from Lac Pavin, collected in November
2006.
Sample Filtration type Depth Fe δ56Fe δ57Fe
(m) (µM) (‰) (‰)

LP2006-1 atmospheric conditions 61.5 77 -0.80 ± 0.08 -1.20 ± 0.14

LP2006-2 atmospheric conditions 62 153 -0.64 ± 0.04 -0.95 ± 0.06
LP2006-2-duplicate atmospheric conditions 62 160 -0.71 ± 0.08 -1.00 ± 0.16
LP2006-3 atmospheric conditions 62.5 264 -0.46 ± 0.06 -0.70 ± 0.10
LP2006-4 atmospheric conditions 70 727 0.00 ± 0.02 0.05 ± 0.08
LP2006-5 atmospheric conditions 75 874 0.06 ± 0.04 0.12 ± 0.12
LP2006-6 atmospheric conditions 75 855 0.07 ± 0.04 0.10 ± 0.08
LP2006-7 At depth (75m) 75 841 0.06 ± 0.06 0.08 ± 0.06
LP2006-8 At depth (75m) 75 853 0.05 ± 0.04 0.06 ± 0.10

Uncertainties correspond to 2SD (standard deviation).

 
 

  9  
  
Fig. A.1. Iron isotope composition of international (IF-G, AC-E, BCR-2) and "home" (ALD) standards.
Analyzes over a 5-years period. Mean δ56Fe values are reported with 2SD uncertainty and are in excellent
agreement with previous data for these standards.
 

 
Fig. A.2. Three-isotope plot representing δ57Fe versus δ56Fe. Data measured for international (IF-G, AC-E, BCR-
2) and "home" (ALD) standards, together with Lac Pavin samples analyzed in the course of this study (n=141).
All data plot on a single mass fractionation line, with a slope close to the theoretical value (~1.5).
 
 
 

  10