silicon isotope fractionation during microbial reduction of … · 2017-01-08 · beard et al.,...

$: Silicon isotope fractionation during microbial reduction of … · 2017-01-08 · Beard et al., 2010; Wu et al., 2010; Frierdich et al., 2014a, 2014b; Reddy et al., 2015). Stable$
Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 190 (2016) 85–99

Silicon isotope fractionation during microbial reductionof Fe(III)–Si gels under Archean seawater conditions

and implications for iron formation genesis

Thiruchelvi R. Reddy ⇑, Xin-Yuan Zheng, Eric E. Roden, Brian L. Beard,Clark M. Johnson

Department of Geoscience, University of Wisconsin-Madison, WI 53706, United States

NASA Astrobiology Institute, United States

Received 30 November 2015; accepted in revised form 26 June 2016; Available online 4 July 2016

Abstract

Microbial dissimilatory iron reduction (DIR) is a deeply rooted metabolism in the Bacteria and Archaea. In the Archeanand Proterozoic, the most likely electron acceptor for DIR in marine environments was Fe(III)–Si gels. It has been recentlysuggested that the Fe and Si cycles were coupled through sorption of aqueous Si to iron oxides/hydroxides, and throughrelease of Si during DIR. Evidence for the close association of the Fe and Si cycles comes from banded iron formations(BIFs), which consist of alternating bands of Fe-bearing minerals and quartz (chert). Although there has been extensive studyof the stable Fe isotope fractionations produced by DIR of Fe(III)–Si gels, as well as studies of stable Fe isotope fractiona-tions in analogous abiologic systems, no studies to date have investigated stable Si isotope fractionations produced by DIR.

In this study, the stable Si isotope fractionations produced by microbial reduction of Fe(III)–Si gels were investigated insimulated artificial Archean seawater (AAS), using the marine iron-reducing bacterium Desulfuromonas acetoxidans. Micro-bial reduction produced very large 30Si/28Si isotope fractionations between the solid and aqueous phase at �23 �C, whereD30Sisolid–aqueous isotope fractionations of �3.35 ± 0.16‰ and �3.46 ± 0.09‰ were produced in two replicate experimentsat 32% Fe(III) reduction (solid-phase Fe(II)/FeTotal = 0.32). This isotopic fractionation was substantially greater than thatobserved in two abiologic controls that had solid-phase Fe(II)/FeTotal = 0.02–0.03, which produced D30Sisolid–aqueous isotopefractionations of �2.83 ± 0.24‰ and �2.65 ± 0.28‰. In a companion study, the equilibrium D30Sisolid–aqueous isotope frac-tionation was determined to be �2.3‰ for solid-phase Fe(II)/FeTotal = 0. Collectively, these results highlight the importanceof Fe(II) in Fe–Si gels in producing large changes in Si isotope fractionations. These results suggest that DIR should producehighly negative d30Si values in quartz that is the product of diagenetic reactions associated with Fe–Si gels. Such Si isotopecompositions would be expected to be associated with Fe-bearing minerals that contain Fe(II), indicative of reduction, such asmagnetite. Support for this model comes from recent in situ Si isotope studies of oxide-facies BIFs, where quartz in magnetite-rich samples have significantly more negative d30Si values than quartz in hematite-rich samples.� 2016 Elsevier Ltd. All rights reserved.

Keywords: Si isotopes; Fe–Si gels; Microbial reduction; Fractionation; BIFs

http://dx.doi.org/10.1016/j.gca.2016.06.035

0016-7037/� 2016 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Department of Geoscience, Univer-sity of Wisconsin-Madison, WI 53706, United States.

E-mail address: [email protected] (T.R. Reddy).

1. INTRODUCTION

Iron-rich cherts are common in the Archean andPaleoproterozoic rock record and include lithologies suchas jasper and banded iron formations (BIFs). Such deposits


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2016.06.035&domain=pdf

86 T.R. Reddy et al. /Geochimica et Cosmochimica Acta 190 (2016) 85–99

are commonly used to infer a Precambrian marine systemthat had high dissolved Fe(II) and Si (e.g., Holland, 1984;Siever, 1992; Klein, 2005), and it is now recognized that oxi-dation of hydrothermally sourced Fe(II) in the presence ofdissolved Si may produce extensive primary Fe(III)–Si pre-cipitates in Precambrian marine environments (e.g., Fischerand Knoll, 2009; Rasmussen et al., 2015). The mechanismfor Fe precipitation in BIFs has been attributed tochemolithoautotrophic iron oxidation, anoxygenic photo-synthesis (photoferrotrophy), and oxidation by ambientoxygen produced by oxygenic photosynthesis (e.g.,Kappler et al., 2005; Klein, 2005). Stable Fe isotope studiesof jaspers and BIFs have shown that the extent of oxidationof hydrothermal Fe(II) has increased with decreasing age inthe Archean (e.g., Dauphas et al., 2004; Czaja et al., 2013;Li et al., 2013; Satkoski et al., 2015), and recently, com-bined Fe–Nd isotopes have documented microbial andhydrothermal Fe sources in 2.5 Ga BIFs (Li et al., 2015).The wide range in isotopic compositions of Si in BIFs hasbeen generally interpreted to be a mixture of hydrothermaland continental sources (e.g., Andre et al., 2006; Robertand Chaussidon, 2006; van den Boorn et al., 2007, 2010;Heck et al., 2011; Chakrabarti et al., 2012), although recentstudies have increasingly highlighted the potential role of Sicycling through diagenetic processes (e.g., Stefurak et al.,2015).

Numerous laboratory experiments have constrainedstable Fe isotope fractionations associated with interactionsbetween aqueous Fe(II) and various iron oxides/hydroxidesin abiologic systems under a variety of pH conditions (e.g.,Beard et al., 2010; Wu et al., 2010; Frierdich et al., 2014a,2014b; Reddy et al., 2015). Stable Fe isotope fractionationsare significantly affected by the presence of Si, either as asorbed species, or in the solid phase as Fe–Si gels, includinga major influence by Fe:Si ratios relative to equivalent Si-free systems (Wu et al., 2010, 2011, 2012). In addition,the stable Fe isotope fractionation between aqueous Fe(II) and iron oxides/hydroxides during microbial dissimila-tory iron reduction (DIR) has been extensively studied inSi-free (Beard et al., 1999, 2003; Crosby et al., 2005,2007) and Si-bearing (Wu et al., 2009; Percak-Dennettet al., 2011) systems. The potential importance of DIR inFe–Si cycling in the Archean lies in the deeply rooted nat-ure of microbial Fe(III) reduction in both the Bacteriaand the Archaea (e.g., Vargas et al., 1998). In contrast,there have been few stable Si isotope studies of Fe–Si sys-tems, where, until recently, only the effect of sorption ofaqueous Si to iron oxides/hydroxides has been studied(Delstanche et al., 2009), although there has been extensivestudy of pure Si systems. There have been no studies ofstable Si isotope fractionation associated with DIR.

In this contribution, we describe the results of experi-ments that investigated stable Si isotope fractionation dur-ing microbial reduction of Fe(III)–Si gels in artificialArchean seawater. The use of Fe(III)–Si gels in the experi-ments reflects the recognition that such gels were likelyimportant primary precipitates in the Archean ocean duringoxidation of aqueous Fe(II) in the presence of high dis-solved Si (Percak-Dennett et al., 2011), and is aligned withthe importance of nanometer-scale Fe–silicate phases in

iron formation cherts (Rasmussen et al., 2015), which sug-gest that Fe and Si co-precipitated from anoxic seawaterenriched in dissolved Fe and Si. Control experiments areused to determine Si isotope fractionation in the absenceof biological production of aqueous Fe(II). The microbiallymediated reductive dissolution of the Fe(III)–Si gels resultsin both solid-phase and aqueous Fe(II), the former of whichhas a significant effect on Si isotope fractionation, produc-ing very large isotopic fractionations relative to pure silicasystems. Our results are compared to those from a compan-ion study (Zheng et al., 2016), where Si isotope exchangekinetics between the same Fe–Si gel and AAS in the pres-ence or absence of aqueous Fe(II) was studied through aseries of abiologic 29Si-enriched tracer experiments, andequilibrium Si isotope fractionations in abiologic Fe–Sigel systems were determined through extrapolation to100% isotope exchange using the three-isotope method.We use the Si isotope exchange kinetics determined byZheng et al. (2016) to evaluate the likelihood that the bio-logical experiments were close to Si isotope equilibrium.The comparison of the two studies permits us to understandthe similarities and differences between Si isotope fraction-ation imparted by Fe(II) produced as a result of microbialreduction of Fe–Si gels and abiotic introduction of aqueousFe(II). Our results bear on long-standing puzzles in the Siisotope compositions of BIFs, which tend to have lowerd30Si values for cherts in BIFs, particularly those that havemajor portions of Fe(II)-bearing minerals such asmagnetite.

2. EXPERIMENTAL DESIGN AND ANALYTICAL

METHODS

DIR of Fe–Si gel was conducted in an anoxic media ofseawater-like matrix (termed ‘‘artificial Archean seawater”,or AAS). Two sets of experiments (cell and control) wereconducted in duplicate, for a duration of 30 d. Experimentsthat were inoculated with Desulfuromonas acetoxidans and20 mM acetate are termed cell 1 and cell 2 (ExperimentsA and B). Two uninoculated control experiments, control1 and control 2 (Experiments C and D) were run in parallelto the cell experiments. All reactors were maintained understrictly anaerobic conditions. The experiments were con-ducted and maintained at room temperature (�23 �C).The reactor vessels were agitated at the beginning of theexperiments, and caution was taken to ensure homogenoussampling and extractions (see Section 2.4).

2.1. Synthesis of Fe–Si gel (electron acceptor)

The Fe–Si gel was synthesized using the procedure fromPercak-Dennett et al. (2011), where a solution containing100 mM NaHCO3, 100 mM Na2SiO3�9H20, and 50 mMFeCl2�4H2O was produced and allowed to oxidize underambient lab conditions for �18 d with continuous shaking.The extent of Fe(II) oxidation in the solid was monitoredregularly during the synthesis process by Ferrozine mea-surements (Stookey, 1970) on small aliquots of the gel thatwas dissolved in 0.5 M HCl. The final Fe–Si gel was cen-trifuged and washed with distilled water, and contained

T.R. Reddy et al. /Geochimica et Cosmochimica Acta 190 (2016) 85–99 87

�97–98% of Fe(III) (i.e., a Fe(II)/FeTotal ratio of�0.02–0.03).

This synthesis method produces a homogeneous, amor-phous Fe–Si gel, as documented by XRD, SEM, TEM, andchemical assays in Percak-Dennett et al. (2011). The Fe–Sigel produced by this method consisted of <50 nm solid par-ticles (based on TEM with selected area electron diffraction(SAED) analysis, see Percak-Dennett et al., 2011) with astoichiometric composition of approximately FeSi2(OH)11as a homogenous solid and not individual particles of Fe-and Si-bearing material. This was confirmed in the currentstudy using XRD, SEM, and chemical assays, the latter ofwhich analyzed a completely digested gel in HCl andNAOH for Fe and Si concentrations, respectively, whichproduced a molar ratio of Fe:Si in the gel of 0.475:1 (Sup-plementary Information, Table S1).

2.2. Artificial Archean seawater (AAS)

Because the goal of the experiment was to provide ananalogy to Archean iron-chert formations, including jas-pers and BIFs, key components of Archean marine condi-tions were incorporated into the experimental design.Archean seawater is generally considered to have beenanoxic, high in dissolved Si (Siever, 1992; Maliva et al.,2005), and low in sulfate contents, 6200 lM (Habichtet al., 2002). The AAS prepared for this study followedthe recipe of Percak-Dennett et al. (2011) (Table S2), wherethe phosphate concentration was 0.1 mM to minimize for-mation of Fe-phosphates, with the exception that aqueousSi was excluded to determine if DIR dissolved the Fe–Sigel in stoichiometric proportions (Fe:Si molar ratio of0.475:1; see above), as well as maximize sensitivity to anyisotopic effects during DIR and mobilization of Si. Theexclusion of Si was also to minimize Si sorption to the gelsurfaces, which would inhibit release of Si into solution.Although it is estimated that Archean seawater may havecontained �0.1–1 mM aqueous Fe(II) (e.g., Holland,1984), initial aqueous Fe(II) contents were zero in the celland control experiments, to monitor the extent of reduc-tion, as well as to determine if dissolution during DIRwas stoichiometric, and follows standard practice in exper-imental studies of DIR. The Fe–Si gel was added to theAAS to make up a final concentration of �21 mM Fe asgel. The medium was then rendered anoxic by bubblingwith an 80:20 mixture of O2-free N2:CO2 gas passedthrough a reduced Cu column to remove any traces ofO2. This method resulted in a final pH of �6.9. The slightlylower pH relative to modern seawater is a realistic condi-tion for Archean seawater at equilibrium with elevatedatmospheric CO2 concentrations (Rye et al., 1995;Ohmoto et al., 2004, 2006; Sheldon, 2006), and still lieswithin the physiological limits of modern marine dissimila-tory iron-reducing bacteria (6.5–8.5; Lovley et al., 2004).The near-neutral pH also minimizes Si polymerization(Iler, 1979; Svensson et al., 1986; Davis et al., 2001;Hiemstra et al., 2007), and prevents the inhibition of DIRthat occurs at more alkaline pH due to Si polymerizationon the surfaces of Fe(III) oxides (Wu et al., 2009). The

medium was not autoclaved to avoid re-crystallizing theFe–Si gel and inducing phase transitions.

2.3. DIR culture

D. acetoxidans was obtained from the DSMZ-GermanCollection of Microorganisms and Cell Cultures, Germany.The cells were grown to a stationary phase and collected bycentrifugation and washing with AAS, as previouslydescribed in Roden and Lovley (1993), where fumarateserved as the electron acceptor and sodium acetate servedas the carbon and energy source. A portion of washed cellswas added to AAS medium containing Fe(III)–Si gel toobtain a final cell concentration of ca. 108 cells/mL.

2.4. Sampling and extraction procedures

Subsamples of 2 mL volume were collected at each timepoint from shaken reactor vessels to ensure homogenoussampling of the solid suspension. All removals from the ves-sels were done using syringes that had been flushed with O2-free N2:CO2 gas to maintain anoxic conditions. The sub-samples were separated into different aliquots for Fe andSi measurements by colorimetry (see Section 2.6) and Si iso-tope analysis. All the extractions were conducted in ananaerobic chamber (Coy Products, Grass Lake, MI,USA). Lab apparatus and chemical reagents were allowedto equilibrate in the anoxic chamber for at least 24 h priorto use.

Aqueous and solid phases were sampled at 1, 2, 5, 9, 16,23, and 30 d (Table 1), where the collected suspension wascentrifuged for �5 min at �6000g, followed by removal ofthe supernatant as the aqueous phase. The remaining solidwas dissolved in 1 mL 0.5 M HCl for �24 h. Prior to Feand Si concentration measurements, the aqueous phasewas centrifuged and only the upper part of the solutionwas sampled for analysis to avoid contamination of anynon-visible precipitate in the aqueous component.

2.5. X-ray diffraction (XRD) and variable pressure scanning

electron microscopy (VPSEM) analysis

XRD analyses of the initial Fe(III)–Si gel after synthesisconfirmed the amorphous nature of the gel, which was con-sistent with our previous studies (Percak-Dennett et al.,2011; Wu et al., 2012). In addition, solids collected aftermicrobial reduction (Experiments A and B) and from thecontrol experiments (Experiments C and D) were also ana-lyzed by XRD and VPSEM. Samples were dried under astream of O2-free N2 gas prior to XRD and VPSEM.XRD data were obtained using a Rigaku Rapid II XRDsystem with a two-dimensional image plate detector (MoKa radiation) in the Department of Geoscience, Universityof Wisconsin-Madison. Using Rigaku’s 2DP software, theimages were integrated to produce the final XRD pattern(Fig. S1).

VPSEM analyses were carried out on the solids of micro-bial reduced samples (Experiments A and B) and of thecontrol experiments (Experiments C and D) at the Depart-ment of Geoscience, University of Wisconsin-Madison.

Table 1Si isotope composition of dissolved and solid experimental run products of both cell (Experiments A and B) and control (Experiments C andD) experiments and deduced isotopic fractionation between the two components. The average values refer only to samples t = 16–30 d.

Time (d) Aqueous Solid D30SiFe–Si gel-AAS

d30Si (‰) 2 SDa d29Si (‰) 2 SDa n d30Si (‰) 2 SDa d29Si (‰) 2 SDa n

Exp A cell 1 (Fe(II) bearing)

1 1.16 0.09 0.57 0.06 2 �0.13 0.16 �0.03 0.02 2 �1.292 2.07 0.05 1.09 0.10 2 �0.09 0.08 �0.08 0.05 3 �2.165 2.56 0.18 1.35 0.15 4 �0.17 0.11 �0.08 0.10 2 �2.739 2.89 0.18 1.51 0.19 4 �0.13 0.18 �0.06 0.12 2 �3.0216 3.19 0.16 1.62 0.05 3 �0.24 0.15 �0.11 0.07 3 �3.4323 3.23 0.12 1.66 0.19 3 �0.04 NAb 0.01 NAb 1 �3.2730 3.25 0.14 1.67 0.18 2 �0.11 0.15 �0.06 0.12 2 �3.36

Average �3.35 ± 0.16

Exp B cell 2 (Fe(II) bearing)

1 1.22 0 0.67 0.15 2 �0.09 0.11 �0.03 0.06 5 �1.312 2.08 0.12 1.09 0.09 3 �0.32 0.09 �0.13 0.02 3 �2.405 2.44 0.07 1.25 0.07 3 �0.23 0.12 �0.08 0.04 2 �2.679 2.97 0.13 1.55 0.21 3 �0.13 0.08 �0.03 0.08 3 �3.1016 3.09 0.19 1.56 0.06 3 �0.32 0.07 �0.20 0.08 2 �3.4123 3.36 0.23 1.69 0 2 �0.10 0.19 �0.08 0.14 4 �3.4630 3.40 0.19 1.69 0.07 3 �0.10 0.13 �0.05 0.03 3 �3.50

Average �3.46 ± 0.09

Exp C control 1 (non-Fe(II) bearing)

1 1.22 0.09 0.63 0.15 2 �0.08 0.09 �0.07 0.05 5 �1.302 1.95 0.13 0.99 0.18 3 �0.19 0.07 �0.08 0.10 6 �2.145 2.28 0.06 1.16 0.08 3 �0.24 0.02 �0.13 0.09 2 �2.529 2.32 0.08 1.23 0.09 3 �0.24 0.11 �0.08 0.10 3 �2.5616 2.62 0.06 1.30 0.12 4 �0.14 0.06 �0.06 0.12 3 �2.7623 2.60 0.13 1.33 0.10 3 �0.16 0.13 �0.07 0.14 2 �2.7630 2.74 NAb 1.47 NAb 1 �0.23 0.14 �0.06 0.09 3 �2.97

Average �2.83 ± 0.24

Exp D control 2 (non-Fe(II) bearing)

1 1.06 0.14 0.59 0.04 2 �0.13 0.12 �0.07 0.03 3 �1.342 2.20 0.08 1.12 0.14 3 �0.07 0.20 �0.06 0.15 3 �2.275 2.29 0.17 1.19 0.18 3 �0.11 0.06 �0.06 0.20 2 �2.409 2.53 0.12 1.31 0.17 3 �0.19 0.15 �0.08 0.04 3 �2.7216 2.39 0.06 1.22 0.07 3 �0.18 0.05 �0.08 0.06 3 �2.5723 2.66 0.05 1.37 0.15 2 �0.15 0.14 �0.08 0.03 2 �2.8130 2.49 0.05 1.29 0.05 2 �0.09 0.07 �0.03 0.14 3 �2.61

Average �2.65 ± 0.28

All isotopic data from wet plasma except isotopic data from Experiment A t = 5, 16 d (solid), Experiment B t = 5 d (aqueous), t = 2–16 d(solid), Experiment C t = 5–9 d (solid), and Experiment D t = 1, 16, 30 d (aqueous), t = 1, 9–30 d (solid) were from dry plasma.a 2 SD = 2*standard deviation, where standard deviation is based on the number of replicate isotopic analysis for a particular sample.b NA = Not applicable.


The solids were imaged (Fig. S3) with a backscattered elec-tron (BSE) detector of the VPSEM operated at 15 kV, sui-ted for biological specimens (Newbury, 2002). The energydispersive X-ray spectrometric analyses were done to con-firm the 0.475:1 Fe:Si ratio in the reduced material on themicrometer scale.

2.6. Fe and Si concentration measurements

Iron contents in the aqueous and solid component (dis-solved by 0.5 M HCl) were quantified using a small aliquotof these solutions and the Fe(II)-selective reagent Ferrozine(Stookey, 1970). The total Fe concentration was deter-mined using hydroxylamine hydrochloride to reduce all

Fe to Fe(II). The Fe(III) abundance is calculated as the dif-ference between total Fe and Fe(II). Light absorbance ofthese samples was measured on a spectrophotometer at awavelength of 562 nm and converted to concentrationsbased on calibrations to Fe standards of known concentra-tions. The uncertainty associated with this method was�5% (1 SD) based on replicate measurements of samplesand the detection limit was �5 lM. The concentration ofsilica was determined colorimetrically using the heteropolyblue method (Clesceri et al., 1989) that detects molybdate-reactive Si species, which are mostly present asmonomers and minor dimers and trimmers (Iler, 1979;Tanakaa and Takahashib, 2001). For Si concentrationmeasurements, solutions were allowed to react with the


ammonium molybdate for �5 min before being reduced by1-amino-2-naphthol-4-sulfonic acid that resulted in blue-colored Si complexes. Light absorbance of these sampleswas measured on a spectrophotometer at wavelength of815 nm and converted to concentrations based on calibra-tions to Si standards of known concentrations. The uncer-tainty associated with this method was �5% (1 SD) basedon replicate measurements of samples and the detectionlimit was �5 lM.

2.7. Si isotope analysis

Silicon isotope compositions are reported in standarddelta (d) notation relative to the international standardNBS-28:

d29Si ¼ 29Si=28Si� �

sample= 29Si=28Si� �

NBS-28� 1

h i� 1000 ð1Þ

and

d30Si ¼ 30Si=28Si� �

sample= 30Si=28Si� �

NBS-28� 1

h i� 1000 ð2Þ

Isotopic fractionation between two phases, A and B, isexpressed as:

a30SiA�B ¼ 30Si=28Si� �

A= 30Si=28Si� �

Bð3Þ

following standard notation, and this can be approximatedby:

103 lna30SiA�B � D30SiA� B ¼ d30SiA � d30SiB ð4ÞPrior to sample purification by cation-exchange chro-

matography for Si, the aqueous samples were acidified withdouble-distilled 7 M HCl to pH � 2–3 (verified with pHpapers). Because Si precipitated as insoluble gel due to highSi contents and the resulting fast Si polymerization in thesolution when Fe–Si gel was dissolved in �1 mL of 0.5 MHCl, it is essential to separate the supernatant and thegel. Although the supernatant contained primarily Fe, itcan also contain some soluble, monomeric Si. Insoluble Sigel was dissolved in NaOH, and the dissolved solutionwas then diluted to low Si concentrations (�20 ppm) sothat acidification of the solution did not result in re-precipitation of Si. The supernatant and the dissolved gelwere finally recombined to ensure accurate total Simeasurement.

The pre-treated samples were then purified following thechromatographic approach of Georg et al. (2006). BecauseSi existed in the samples largely as uncharged monosilicicacid (H4SiO4), along with some minor negatively chargedspecies (H3SiO4

�), at pH � 2–8, Si is not retained by thecation exchange resin (Bio-Rad AG 50W-12X, H form,200–400 mesh). The resin was filled up to the 1.8 mL markon BioRad columns and pre-cleaned by 8 M HCl, 4 M HCl,and ultrapure water sequentially. Multiple passes of HCland water were used to ensure clean resin. Prior to sampleloading, the eluted water from the columns was checked forneutral pH to verify complete removal of the HCl. Siliconwas collected after the sample was loaded on the column,and further eluted with �4.5 mL of H2O.

Silicon isotope analysis was done using a Nu Instru-ments Nu Plasma II for 30Si/28Si and 29Si/28Si ratios.

Interferences, particularly on 30Si by 14N16O+, wereresolved by narrowing the width of the source defining slitand the two alpha slits to operate in pseudo-high-mass-resolution mode. Silicon-bearing sample solutions wereinlet into the mass spectrometer using a self-aspirating neb-ulizer with an uptake rate of �90–110 lL/min and a Peltiercooled (7 �C) Glass Expansion Twister spray chamber (wetplasma) or using an Aridus desolvating unit (dry plasma).Silicon concentrations were set at �3 ppm for the wetplasma measurements and �1 ppm for the dry plasma mea-surements, yielding a total Si ion intensity of �5–10 volts(1011 ohm resistors). Silicon backgrounds from reagentsand instrument were negligible (<0.5%) relative to Si pro-cessed through column chemistry, and were corrected byan on-peak-zero routine for each analysis. Instrumentalmass-bias corrections were made using a standard-sample-standard bracketing protocol, where the international stan-dard for Si isotopes, NBS-28, was used as the bracketingstandard.

Accuracy and precision of Si isotope measurements wereassessed through analysis of standards of known isotopiccomposition, Diatomite and BHVO-2, which were pro-cessed through the entire chemical separation procedurealong with samples, and measured in each analytical ses-sion. All samples yielded mass-dependent d30Si and d29Sivalues, d30Si � 1.95 � d29Si, similar to d30Si � 1.93 � d29Sireported by Young et al. (2002). The long-term analysesusing wet plasma for Diatomite resulted in d30Si = 1.28± 0.20‰ and d29Si = 0.66 ± 0.12‰ (2 SD, n = 43) andBHVO-2 yielded d30Si = �0.26 ± 0.19‰ andd29Si = �0.13 ± 0.15‰ (2 SD, n = 68). Analyses using dryplasma yielded d30Si = 1.21 ± 0.17‰ and d29Si = 0.63± 0.12‰ (2 SD, n = 18) for Diatomite and d30Si = �0.29± 0.13‰ and d29Si = �0.13 ± 0.10‰ (2 SD, n = 17) forBHVO-2. These values are identical within error to pub-lished values of d30Si = 1.26‰ and d29Si = 0.64‰ for Dia-tomite and d30Si = �0.28‰ and d29Si = �0.15‰ forBHVO-2 (Reynolds et al., 2007; Savage et al., 2014). More-over, repeated measurements for the same samples underwet and dry plasma modes were reproducible within uncer-tainty (�0.20‰) (Fig. S2), indicating that the datasets areequivalent.

In addition to measurements of Diatomite and BHVO-2, Si isotope analysis was evaluated using anion and cationdoping tests. The use of cation exchange columns does notallow for removal of anions such as Cl�, which was a majoranion in AAS. Therefore, NBS-28 that had been passedthrough the ion-exchange column separation was dopedwith HCl at concentrations between 0.02 and 0.5 M, brack-eting the range in our samples, and this was measuredagainst NBS-28 that was not doped with HCl. This testshowed that the measured Si isotope compositions ofNBS-28 doped with a range of Cl� were identical withinanalytical uncertainty of the measurements (�0.20‰), sug-gesting a negligible matrix effect associated with anions.Cation doping tests were also conducted to evaluate theeffectiveness of the ion-exchange separation. The Big BatchSi standard was doped with FeCl3 to produce Fe:Si molarratios of 1:1, 1:2.5, and 1:5 and processed through ion-exchange columns prior to isotopic analysis. This produced

Fig. 1. Temporal variation in Fe and Si concentrations presentedin terms of aqueous Fe concentrations (A), % total reduction (B),which takes into account both the aqueous and solid phase Fe, andaqueous Si concentrations (C).


Si isotope compositions that were indistinguishable fromthose measured for pure Big Batch, indicating effectiveremoval of Fe through the ion-exchange column separationprocedure.

3. RESULTS

3.1. Fe–Si gel structure

X-ray diffraction and VPSEM analysis indicated that theinitial Fe(III)–Si gel and the microbially reduced gel wereamorphous. The XRD patterns contained no significantpeaks to be correlated with any minerals (Fig. S1). EDSanalysis in VPSEM produced Fe:Si ranging between0.5–0.6 over areas of 20 � 20, 50 � 50, and 200 � 200 lm,indicating that there was no significant change in Fe:Siratios within the uncertainties of operating in VP mode.Although the measured aqueous Si concentrations in thecell and control experiments (Experiments A–D; �1 mM;see below) were significantly above saturation levels forquartz: �0.3 mM at 20–30 �C (Gunnarsson andArnorsson, 2000), there was no evidence for formation ofmicro-crystalline quartz in XRD or SEM measurements.The Si concentrations were below amorphous silica satura-tion, �2 mM (Gunnarsson and Arnorsson, 2000). For thetime-series samples, there was no detectable change in theamorphous (nanoparticulate) nature of the solids.

3.2. Temporal changes in Fe and Si concentrations

The total Fe concentrations reported (Fig. 1A; Table S3)for the aqueous fraction are all Fe(II) because anaerobicconditions were used and the electron acceptor was solid-phase Fe(III). The lack of significant aqueous Fe in the con-trol experiments (Experiments C and D), beyond the smallamount carried over from preparation of the initial Fe–Sigel (see above), demonstrates that abiologic dissolution ofFe in the gel did not occur under the experimental condi-tions used. The aqueous Fe concentrations in the cell exper-iments increased with time (Experiments A and B),although all measurements were below 0.25 mM. In thecontrol experiments (Experiments C and D), very low levelsof aqueous Fe were measured reflecting the small amountsof residual Fe(II) (Fe(II)/FeTotal = 0.02–0.03) thatremained after preparation of the initial Fe(III)–Si gel thatwas produced by �97–98% oxidation (see above).

The majority of Fe(II) produced during microbialreduction was retained in the solid phase, consistent withprevious studies, which showed that the majority of Fe(II) is probably incorporated into the gel or bound tothe surfaces of the gel (see discussion in Percak-Dennettet al., 2011). Solid-phase Fe(II) concentrations increasedto an average of �5.6 mM in the cell experiments after30 d. The Fe(II) measured in the control experimentsreflects a Fe(II)/FeTotal ratio of 0.02–0.03 that remainedafter preparation of the initial Fe(III)–Si gel, and the Fe(II)/FeTotal ratios and percent reduction for the cell exper-iments are calculated after subtracting this backgroundlevel of Fe(II) (Fig. 1B). The correction for backgroundFe(II) is equivalent to �2–3% reduction, and therefore

would be negligible relative to the extent of reduction overthe total time of the experiment even if no correction wasmade. For each sample, the total solid phase Fe concen-tration was measured and was variable due to variationsin suspension density sampled at each time point. TheFe(II)/FeTotal ratios and percent reduction were thereforecalculated using the total solid phase Fe concentrationof individual time points for the experiments, and notthe initial amount delivered to the reactors.

The initial AAS had no aqueous Si in the beginning ofthe experiments (t = 0), as Si was excluded during thepreparation AAS. The aqueous Si concentrations quicklyrose to �1 mM within the first day of the experiments forboth cell and control experiments (Fig. 1D; Table S3). Mostof the Si resided in the solid phase (�97%) and the rise inaqueous Si concentration is equivalent to �3% dissolutionof the solid Fe–Si gel (Table S4). In contrast, aqueous Fe


did not increase beyond that expected from initialbackground levels (above) indicating that this early dissolu-tion of the Fe–Si gel was non-stoichiometric. This extent ofdissolution is equivalent to changing the Fe:Si molar ratioof the bulk Fe:Si gel from 0.475:1 to 0.475:0.97, whichwould not be detected in the chemical analyses. Att = 9 d, cell and control experiments reached 1.1–1.2 mMaqueous Si, which may reflect saturation or slightly super-saturated conditions for the Fe–Si gel. The aqueous Si con-centrations thereafter remains constant within uncertainty(±5%, 1 SD), although the cell experiments seem to main-tain slightly higher aqueous Si concentrations for later timepoints than the controls. One anomalously high aqueous Simeasurement of 1.25 mM for Experiment A at t = 16 d isinterpreted to reflect minor contamination by the solidphase during separation of the supernatant, as this is notseen in the same time point for Experiment B.

3.3. Reduction of Fe–Si gel

The extent of reduction was calculated using the sum ofaqueous and solid Fe(II) contents relative to total solidphase Fe at each sampling. Following previous work(Percak-Dennett et al., 2011), 20 mM acetate was used inthe cell experiments (Experiments A and B) as the electrondonor. An average of 32% reduction was achieved over 30 din the cell experiments (Fig. 1; Table S3). The significantlyslower reduction rates in the current study, as compared tothat of Percak-Dennett et al. (2011) (79%), are probablyassociated with the lower intrinsic activity of the cells inthe pure culture used in the current study, as compared tothe natural enrichment culture used by Percak-Dennettet al. (2011).

3.4. Si isotope variations

The first sampled solution (t = 1 d), where aqueous Sirapidly increased from zero to �1 mM (Fig. 1), corre-sponded with a d30Si value for aqueous Si (d30Siaqueous) of�1‰ for both cell experiments (Experiments A and B)

Fig. 2. Temporal variation in the d30Si values for the aqueous (AAS) andof AAS, therefore any Si in the aqueous phase was derived from dissoindicated on the figure is d30Si = �0.07 ± 0.14‰, d29Si = 0.04 ± 0.09‰ (2on the graph.

and control experiments (Experiments C and D) (Fig. 2).This is significantly higher than the starting d30Si valuefor the Fe–Si gel of �0.07‰. For the cell experiments,d30Siaqueous values continued to increase between 1 and9 d, and then stabilized at d30Siaqueous � 3.2–3.3‰ towardthe end of the experiment between 16 and 30 d (Fig. 2A;Table 1). There was a significant difference (i.e., �0.7‰)in the d30Siaqueous values reached at the end of the experi-ment for the cell and control experiments, where the controlexperiments plateaued at d30Siaqueous � 2.5–2.7‰ at the endof the experiment (Fig. 2). An important distinctionbetween the cell experiments and controls is that the formerproduced significant quantities of solid-phase Fe(II) duringreduction, whereas no significant Fe(II) was produced inthe controls. The Fe(II)/FeTotal of the controls remainedconstant within error at 0.02–0.03. In contrast to the Si iso-tope changes in the aqueous phase, the d30Si values for thesolids remained constant within analytical error of the ini-tial d30Si value for the Fe–Si gel. This reflects the fact that97% of the Si (molar ratio) in the experiments was con-tained in the gel (Table S4).

Using the d30Si values determined individually on thetime series of solid-aqueous pairs, the Si isotope fractiona-tions at each time were calculated, and an average D30Siso-lid–aqueous isotope fractionation was estimated using the lastthree time points (Table 1), representing an approximationto steady-state conditions with regard to Si (Fig. 2). Com-mensurate with the changes in d30Siaqueous values, the Si iso-tope fractionations between Fe–Si gel and aqueous Sibecome larger over time and reach a relatively constant iso-topic fractionation by the second half of the experiments,both in cell and control experiments (Fig. 3; Table 1).The D30Sisolid–aqueous isotope fractionation in the cell exper-iments reached a value ��3.4‰ to �3.5‰, and in the con-trols, the D30Sisolid–aqueous isotope fractionation reached avalue of ��2.7‰ to �2.8‰. The larger D30Sisolid–aqueousisotope fractionation in the cell experiments wasobtained after �32% reduction of the Fe–Si gel, whereFe(II)/FeTotal � 0.32, whereas the smaller D30Sisolid–aqueousisotope fractionation in the controls was not associated

solid (Fe–Si gel) phases. Si was excluded from the starting materiallution of the gel. The isotope composition of the initial Fe–Si gelSD, n = 12). The error bars (Table 1) are smaller than the symbols

Fig. 3. Temporal variation in the 30Si/28Si isotope fractionationsbetween the solid (Fe–Si gel) and aqueous (AAS) phases for theindividual time-series sampling. Although the starting compositionof the AAS did not contain Si, over time the composition of theAAS evolved as Si increased in the aqueous phase due todissolution. The cell experiments (Experiments A and B) record alarger fractionation of ��3.4‰ to �3.5‰ as compared to that ofthe control experiments (Experiments C and D) of ��2.7‰ to�2.8‰. The error bars (�0.28‰ for D30Sisolid–aqueous, 2 SD) aresmaller than the symbols on the graph.


with reduction of the Fe–Si gel, where Fe(II)/FeTotalremained constant at 0.02–0.03.

4. DISCUSSION

4.1. Aqueous Si concentrations and isotopic exchange

mechanism

The temporal changes in aqueous Si concentrations mayreflect the interplay between net dissolution and net sorp-tion, which may be illustrated through comparison of rateconstants if a first-order rate law is assumed. Rapid disso-lution of the solid occurred because the aqueous phasewas initially under-saturated with respect to Si. Att = 1 d, aqueous Si concentrations rapidly rose, and att = 2 d, aqueous Si concentrations reached a maximum(Fig. 1D; Table S3). With continued time, the Si concentra-tions remained constant over t = 9–30 d except for Experi-ment A at t = 16 d. Initial dissolution of the solid phase wasonly �3% of the total solid mass (Section 3.2 above), andthe rapid increase in aqueous Si in the first day producesa first-order rate constant of �2.29 d�1. The rate constantswere calculated based on the fraction of Si that was presentin the aqueous phase over time. Since dissolution was rapid(reaching saturation levels within two days), only the firsttwo time points were used to derive the dissolution rateconstant since the later time points (5–30 d) may representmore complex sorption–desorption processes. This is sup-ported by comparing the rates of Si sorption to ferrihydritebased on the work of Delstanche et al. (2009), which pro-duces a first-order rate constant of �0.073 d�1, suggestingthat the rate of sorption of Si to the mixed Fe(III)–Fe(II)–Si gel used in our study was likely much slower thanthe rate of initial dissolution (Fig. S5). We therefore inter-pret the peak in aqueous Si concentrations between 2 and

30 d to reflect Si saturation in the absence of significantsorption, whereas the slight decline in aqueous Si concen-trations with longer time periods likely reflects net sorption.It does not seem likely that the temporal changes reflectchanging aqueous Si polymerization at the circum-neutralpH and low Si concentrations of our experiments (Stummet al., 1967; Iler, 1979). The aqueous Si contents in both celland control experiments (�1 mM) are the same throughoutthe individual sampling times (t = 2 d onwards), most likelyreflecting minimal differences in Si solubility among theFe–Si gels. This appears to indicate that the Fe:Si ratio ofgel is the major control of the gel solubility, as observedby Zheng et al. (2016), rather than the Fe(III):Fe(II) ratioof gel, although further experiments would be required toconfirm this.

Extensive previous characterization of the Fe–Si gelused in the current study (Percak-Dennett et al., 2011;Wu et al., 2012), and Fe–Si gels in general (Doelsch et al.,2000, 2001, 2003), indicates that such gels are composedof Fe–Si–O networks rather than distinct Fe- and Si-richphases. We envision that our Fe–Si gel is probably similarin size to the Fe silicate nanoparticles in the HamersleyBIFs, <10–600 nm long and �1–50 nm wide (Rasmussenet al., 2015). For mobilization of Si to occur during isotopicexchange, breakage of Fe–Si–O bonds is required. The sim-ilar solid-aqueous Si isotope fractionations for the cell andcontrol experiments in the first few days (Fig. 3) suggestssimilar extents of Si isotope exchange when the solid phasesare most similar between the cell and control experiments(i.e., only very low levels of Fe(II) in the solid phase).The isotopic fractionations, however, begin to deviate by�10 d (Fig. 3) between the cell and control experiments.This could reflect greater extents of Si isotope exchange inthe cell experiments as Fe(II) is produced by in situ reduc-tion, drawing upon the observation of Zheng et al. (2016)that the presence of Fe(II) increases Si isotope exchangerates. In addition, where Fe(II) is present, previous stableFe isotope studies have shown that Fe atom exchange iscatalyzed by electron transfer reactions that may producelarge extents of Fe isotope exchange in Fe oxides/hydrox-ides (e.g., Handler et al., 2009; Beard et al., 2010;Frierdich et al., 2014a, 2014b), as well as Fe–Si gels (Wuet al., 2011, 2012). If Fe–Si–O bonds are broken duringFe atom exchange, we would expect this to also enhanceSi atom exchange during electron transfer reactions. Inaddition, the presence of Fe(II) in the solid phase of theexperiments, particularly the cell experiments that producedhigh levels of Fe(II), may impart an intrinsically different Siisotope fractionation for mixed Fe(III)–Fe(II)–Si gels ascompared to Fe(III)–Si gels (Zheng et al., 2016), and thisis discussed in the next section.

4.2. Role of Fe(II) and biological reduction

The results of Zheng et al. (2016), which studied Si iso-tope exchange under abiologic conditions using a nearlyidentical Fe–Si gel, document that the presence of Fe(II)in the aqueous and solid phase leads to more extensive Siisotope exchange between Fe–Si gel and AAS, and alsoleads to a larger Si isotope fractionation relative to Fe


(II)-free systems, reflecting bonding changes that occurupon incorporation of Fe(II) into the solid. We comparethese results with those of the current study in Fig. 4. Animportant distinction between our results and those ofZheng et al. (2016), however, is that solid phase Fe(II) con-centrations in the current study were much higher in the cellexperiments due to microbial Fe(III) reduction. Dissimila-tory Fe(III)-reducing organisms such as D. acetoxidans uti-lize specific biomolecular mechanisms (i.e., outer membraneproteins) to catalyze DIR (Liu et al., 2014), which raises thepossibility that there were enzymatic or ‘‘vital” effects onthe Si isotope fractionation produced by D. acetoxidans.It seems more likely, however, that the ‘‘biological” effectin producing larger D30Sisolid–aqueous isotope fractionationthan in the abiological studies lies in the production ofhigher proportions of Fe(II) (Fig. 4; Table 1) and the con-comitant bonding changes that may produce more extremeSi isotope fractionations. Zheng et al. (2016) determined anequilibrium D30Sisolid–aqueous isotope fractionation of��3.2‰ in the presence of aqueous Fe(II) and a solidFe–Si gel that had an Fe(II)/FeTotal ratio of 0.09, whereequilibrium conditions were estimated based on extrapolatingexperiments that exchanged �80–90% to 100% exchange.Zheng et al. (2016) determined a much smaller equilibriumisotope fractionation of D30Sisolid–aqueous � �2.3‰ for anFe(II)-free system, obtained by extrapolating experimentsthat exchanged �70–80% to 100% exchange. In contrast,the D30Sisolid–aqueous isotope fractionation measured in thecell experiments, where Fe(II)/FeTotal ratios reached�0.32, ranged from �3.4‰ to �3.5‰ (Fig. 4). It remainsunknown if the cell experiments attained complete Siisotope exchange because an enriched Si isotope tracer

Fig. 4. Changes in the D30Sisolid–aqueous isotope fractionation as afunction of the Fe(II)/FeTotal ratio of the solid (Fe–Si gel). For thecell experiments, the Fe(II)/FeTotal ratios increased with time,reflecting biological reduction. For the controls, the Fe(II)/FeTotalratio was 0.02–0.03, reflecting small carryover of Fe(II) from theinitial preparation of the gel. For comparison, the abiologicexperiments of Zheng et al. (2016) are also plotted, whichdetermined the equilibrium D30Sisolid–aqueous isotope fractionationfor Fe(III)–Si gel (Fe(II)/FeTotal = 0) and a mixed Fe(III)–Fe(II)–Sigel (Fe(II)/FeTotal = 0.09). Together, these two studies clearlyindicate that the D30Sisolid–aqueous isotope fractionation increaseswith increasing Fe(II) abundance in the solid. The error bars(�0.28‰ for D30Sisolid–aqueous, 2 SD) are smaller than the symbolson the graph.

was not used, as in the experiments of Zheng et al.(2016). It seems likely, therefore, that the equilibriumD30Sisolid–aqueous isotope fractionation at high Fe(II)/FeTotalratios are larger than the measured values. A conservativeestimate, therefore, is that for Fe(II)/FeTotal ratios greaterthan 0.3, the equilibrium D30Sisolid–aqueous isotope fraction-ation is <�3.5‰. It is unknown if the D30Sisolid–aqueous iso-tope fractionations scale linearly with increasing Fe(II)/FeTotal ratios, although, despite uncertainties in the extentof exchange, the relations in Fig. 4 suggest that they do not.

Comparison of the Si isotope fractionations measured inthe controls in this study with the Fe(II)-free exchangeexperiments of Zheng et al. (2016) suggest differences thatprobably reflect the distinct experimental conditions ofthe two studies. The controls in the current study used aFe–Si gel that had Fe(II)/FeTotal ratios of 0.02 and 0.03,reflecting minor carryover of Fe(II) during initial prepara-tion of the gel, and this small amount of Fe(II) is a possibleexplanation for the slightly larger D30Sisolid–aqueous isotopefractionation of ��2.7‰ to �2.8‰ compared to Zhenget al.’s (2016) Fe(II)-free experiments that produced anequilibrium D30Sisolid–aqueous isotope fractionation of��2.3‰. The larger D30Sisolid–aqueous isotope fractionationmeasured in our controls, at a Fe(II)/FeTotal ratio of�0.02–0.03, lies between those measured by Zheng et al.(2016) at Fe(II)/FeTotal ratios of 0 and 0.09 (Fig. 4).

An additional difference between the studies is thatZheng et al. (2016) set the initial aqueous Si concentrationsnear those estimated to reflect Si solubility for the Fe–Si gel,whereas no initial aqueous Si was used in our control or cellexperiments, and this produced distinct temporal changesin Si isotope fractionations (Fig. 5). In experiments 2aand 3a of Zheng et al. (2016), which had a Fe(II)/FeTotalratio of 0, �70% exchange occurred after 30 d, and in theirexperiments 2b and 3b, which had a Fe(II)/FeTotal ratio of�0.09, �80% exchange occurred after 30 d. In our controls,the isotopic fractionations become larger in the first 10 d,probably reflecting the lack of buffering that occurred inthe Zheng et al. (2016) experiments that contained initialaqueous Si, but, over time, the isotopic fractionations inour controls moved toward those measured for the Fe(II)-bearing experiments of Zheng et al. (2016) (Fig. 5).Although we cannot rule out an initial kinetic effect inour controls, such effects, if present, were not large enoughto produce dramatically different temporal trends relativeto the Fe(II)-bearing results of Zheng et al. (2016). In factit is possible that our control experiments reached Si iso-tope equilibrium faster than those of Zheng et al. (2016)because the presence of initial aqueous Si in their experi-ments would mute isotopic changes with time.

4.3. Implications for the Precambrian rock record

Two important conclusions can be drawn from the cur-rent study and the companion work of Zheng et al. (2016)that bear on interpretation of Archean jaspers and BIFs.The first is the role of microbial iron reduction which resultsin Fe(II) that creates the conditions for the largest Si iso-tope fractionations yet measured, which offers a possibleexplanation for the wide range in Si isotope compositions

Fig. 5. The measured D30Sisolid–aqueous isotope fractionationbetween the solid (Fe–Si gel) and aqueous (AAS) phases for theindividual time-series sampling of the control experiments (Exper-iments C and D) in the current study, compared with the resultsfrom Zheng et al. (2016), who studied Si isotope exchange usingFe–Si gel that had an Fe(II)/FeTotal ratio of 0 (Experiments 2a and3a) and 0.09 (Experiments 2b and 3b). The controls in our studyhad an average Fe(II)/FeTotal ratio of 0.02–0.03. Using a 29Sitracer, Zheng et al. (2016) showed that after 30 d, Experiments 2aand 3a exchanged �70% and Experiments 2b and 3b exchanged�80%. Based on extrapolation to 100% exchange, the estimatedequilibrium D30Sisolid-aqueous isotope fractionation for Experiments2a and 3a is �2.3‰, and �3.2‰ for Experiments 2b and 3b. It isunknown if our control experiments approached equilibrium atfaster or slower rates than the experiments of Zheng et al. (2016),but the apparent D30Sisolid–aqueous isotope fractionation measuredhere for the controls of ��2.7‰ and �2.8‰ lies between theinferred equilibrium results from Zheng et al. (2016), in agreementwith the intermediate Fe(II)/FeTotal ratios of the controls. The errorbars (�0.28‰ for D30Sisolid–aqueous, 2 SD) are smaller than thesymbols on the graph.


of jasper- and BIF-related cherts. The second is the diffi-culty in attaining sufficient Fe(II) contents in Fe–Si gelsto produce the precursors of magnetite + quartz assem-blages under abiologic reactions between Fe(III)–Si gelsand aqueous Fe(II), raising the possibility that the presenceof magnetite may essentially require a role for DIR whenabundant magnetite is found in low-clastic, quartz-richmarine sediments.

The exceptionally large Si isotope fractionations mea-sured here during microbial reduction of Fe(III)–Si gelsprovides a potential explanation for the highly negatived30Si values measured in Archean and Proterozoic BIFsthat contain Fe(II)-bearing oxides such as magnetite. InFig. 6, we illustrate the interactions that may occur betweenaqueous Si and Fe(III)–Si gel, its reduced counterpart, andthe resulting authigenic mineral products. It is not theintent of Fig. 6 to fully describe the precise Si isotope com-positions of quartz in all BIFs; we recognize that factorssuch as the proportions of hydrothermal and continentalsources will influence the range of isotopic compositionsof cherts in BIFs. Our goal here is to illustrate potential dia-genetic pathways for hematite- and magnetite-bearingcherts in jaspers and BIFs that may produce distinct Si iso-tope compositions based on Si isotope fractionations docu-mented here through microbial iron reduction.

Building on the equations and discussion in Percak-Dennett et al. (2011) and Posth et al. (2013), we initially

assume one-third reduction of Fe(III)–Si gel by DIR, whichcan be described by

3FeSi2ðOHÞ11 þ 0:25CH2O

! 3FeðIIÞ1=3FeðIIIÞ2=3Si2ðOHÞ10 2=3 þ 0:25HCO3�

þ 0:25Hþ þ 0:5H2O ð5Þwhich in turn may produce a magnetite–quartz assemblagevia the reaction

3FeðIIÞ1=3FeðIIIÞ2=3Si2ðOHÞ10 2=3 þ 0:25Hþ þ 0:5H2O

! Fe3O4 þ 6SiO2 þ 0:25Hþ þ 16:5H2O ð6ÞIn contrast, in the absence of reduction, burial diagene-

sis of Fe(III)–Si gel would produce a hematite–quartzassemblage via the reaction

2FeðIIIÞSi2ðOHÞ11 ! 4SiO2 þ Fe2O3 þ 11H2O ð7ÞWe use the following isotope fractionations between

aqueous Si and (1) Fe(III)–Si gel, D30Sisolid–aqueous� �2.3‰ (Zheng et al., 2016); (2) microbially reducedFe–Si gel, D30Sisolid–aqueous � �3.5‰ (this study); and (3)quartz, D30Sisolid–aqueous � 1–2‰ (Douthitt, 1982; Dupuiset al., 2015; Pollington et al., 2016), to illustrate the differ-ences between closed- and open-system isotopic composi-tions, focusing on late-stage diagenetic Fe reduction(Fig. 6).

The Fe(II)-free experiments of Zheng et al. (2016)provide insights into the d30Si values expected for hema-tite–quartz assemblages in BIFs. If Si isotope exchangeoccurred between aqueous Si and Fe(III)–Si gel during ini-tial precipitation, and the aqueous:gel molar Si ratio wasvery high, such as would be the case for direct precipitationfrom a large water mass, the d30Si values of the Fe(III)–Sigel would be expected to be ��2‰, assuming a near-zerod30Si value for aqueous Si (Fig. 6A). Dewatering and diage-netic interactions between gel and pore fluids may impartadditional isotopic fractionations, but in our illustration,we assume the isotopic compositions are determined inthe initial gel precipitate because this is most closely analo-gous to the experimental conditions. In our simple model,quartz would inherit the Si isotope compositions of the ini-tial Fe(III)–Si gel if produced via burial diagenesis in aclosed system or when there is incomplete equilibrationwith seawater. If, however, quartz maintained Si isotopeequilibrium with aqueous Si (open system), its d30Si valueswould be significantly higher, depending upon the quartz-aqueous Si isotope fractionation, D30Sisolid–aqueous � 2‰at �23 �C (Pollington et al., 2016) and the extent of isotopicexchange during transformation of Fe(III)–Si gel to quartzand hematite.

In contrast, the very large Si isotope fractionations pro-duced during reduction of Fe(III)–Si gel indicates thatquartz, in the presence of Fe(II)-bearing oxides such asmagnetite, may have highly negative d30Si values(Fig. 6B). Assuming Si isotope exchange between the par-tially reduced Fe–Si gel and aqueous Si, in the presenceof a large molar proportion of aqueous Si, as expected ina marine setting, d30Si values down to <�3.5‰may be pro-duced, using the relations shown in Fig. 4. Reaction tomagnetite and quartz during burial diagenesis in a closed

Fig. 6. Interactions of aqueous Si with Fe–Si gel of different Fe(II) and Fe(III) proportions and their end products during later diageneticprocesses along with the predicted Si isotope compositions of the resulting chert layers in hematite- (Fig. 6A) and magnetite-rich (Fig. 6B)BIFs. The interactions of aqueous Si with the unreduced (Fig. 6A) and reduced (Fig. 6B) gels are depicted in Eqs. 5–7.


system would produce quartz that inherited the Si isotopecomposition of the reduced Fe–Si gel. If, however, quartzformed in an open system, where moderate aqueous-quartz isotopic fractionations controlled the Si isotopecomposition of quartz, higher d30Si values would beexpected for quartz (Pollington et al., 2016).

In situ Si isotope studies of quartz in BIFs that have dif-ferent Fe oxide mineralogy (hematite or magnetite) supportour model in Fig. 6. Work by Steinhoefel et al. (2010) com-pared BIF cores from the Kuruman Iron Formation (sam-ples 3/79 and 3/59), Dales Gorge Member of the BrockmanIron Formation (sample DGM-36), and Penge Iron Forma-tion (sample TBT). Samples 3/59 and DGM-36 aremagnetite-rich BIFs, whereas samples 3/79 and TBT arehematite-rich BIFs. The quartz layers of the magnetite-rich BIFs have consistently lower d30Si values as comparedto the quartz layers in the hematite-rich BIFs, which havesignificantly higher d30Si values (Fig. 7). This contrast fol-lows the experimental results obtained in this study, wherethe most negative d30Si values for quartz would be expectedto be associated with mixed Fe(II)–Fe(III) (correspondingto magnetite + quartz assemblages), whereas relativelyhigher d30Si values for quartz would be expected in theabsence of Fe(II) (corresponding to hematite + quartzassemblages) (Fig. 6). This interpretation contrasts withthat offered by Steinhoefel et al. (2010) and Heck et al.(2011). Steinhoefel et al. (2010) concluded that very lowd30Si values represent hydrothermal sources, while Hecket al. (2011) in their study of similar BIF samples suggestedthat Si, mostly hydrothermally sourced were fractionatedby non-equilibrium processes when Si precipitated byadsorption on Fe oxides in ocean waters. Our results, how-ever, offer an alternative explanation for quartz in BIFs thathas very low d30Si values, and the occurrence of such Si

isotope compositions in magnetite-rich layers can beexplained through microbial reduction of primary Fe(III)–Si precipitates, followed by dewatering and burial dia-genesis to quartz–magnetite assemblages.

We conclude by addressing the possibility that mag-netite–quartz BIFs, particularly if accompanied by highlynegative d30Si values, may essentially require an indirectrole for biology in producing Fe(II) and creating the con-ditions required for the observed isotopic fractionations.Although magnetite paragenesis can occur at many stagesin BIFs (e.g., Bekker et al., 2010), from early diagenesis toore formation, very early magnetite is commonly inter-preted to have formed either through direct precipitationof mixed Fe(II)–Fe(III) precursors or through reactionof Fe(III) hydroxides with aqueous Fe(II) (e.g., Klein,2005; Beukes and Gutzmer, 2008). Virtually all discus-sions on magnetite formation pathways to date, be theybased on the rock record or experiment (e.g., Klein,2005; Posth et al., 2013), have focused on pure ferrihy-drite starting materials, and rarely have considered Si,particularly Fe–Si gels. The results of Zheng et al.(2016), who used aqueous Fe(II) concentrations of�1 mM, an extreme upper limit estimated for the Archeanocean, only produced a mixed Fe(III)–Fe(II)–Si gel thathad an Fe(II)/FeTotal ratio of 0.09, much lower than thestoichiometry of magnetite (Fe(II)Fe(III)2O4). Althoughmagnetite is easily produced through interaction of aque-ous Fe(II) and ferrihydrite in pure Fe systems (e.g.,Frierdich et al., 2014b), the presence of Si, either as asorbed species to ferrihydrite, or an Fe(III)–Si gel,strongly inhibits conversion to more stable Fe(III) oxi-des/hydroxides, or production of Fe(II)-bearing oxides(Wu et al., 2011), and the results of Zheng et al. (2016)support this observation.

Fig. 7. Histograms of d30Si values of quartz layers from various BIF core sections as studied via in situ analysis (Steinhoefel et al., 2010),grouped by Fe-bearing mineralogy. Gray bars are magnetite rich samples (sample 3/59 from the Kuruman Iron Formation, and sampleDGM-36 from the Brockman Iron Formation) while white dotted bars are hematite rich samples (sample 3/79 from the Kuruman IronFormation, and sample TBT from the Penge Iron Formation). Samples that are magnetite rich have highly negative d30Si values, whereassamples that are hematite rich have more moderate d30Si values, which can be explained by the diagenetic reactions in Fig. 6, whereproduction of Fe(II) during DIR of Fe(III)–Si gel produces more extreme Si isotope fractionations than Fe(II)-free Fe–Si gel.


Although additional experimental studies are clearlyneeded, the results to date using Fe–Si gels suggests thatit is very difficult to produce magnetite through interactionswith aqueous Fe(II). In particular, because estimates foraqueous Fe(II) contents in the Archean oceans are gener-ally lower than the 1 mM used by Zheng et al. (2016), per-haps closer to 100 lM (see discussion in Czaja et al., 2012),it seems unlikely that abiologic production of solid-phaseFe(II)/FeTotal ratios sufficient to reach magnetite stoichiom-etry could have easily occurred in Si-bearing Archean mar-ine environments. In addition, if sorption of Si inhibits Feisotope exchange over long time periods (Wu et al., 2012),this might inhibit abiotic Fe(II) incorporation and develop-ment of high enough Fe(II)/FeTotal ratios to reach mag-netite stoichiometry. In contrast, the highly reduciblenature of Fe(III)–Si gels provides a compelling mechanismfor producing magnetite–quartz assemblages that havehighly negative d30Si values. Although it is difficult to cur-rently prove such assemblages could not be produced byabiologic means, on balance, the evidence at hand suggestan indirect role for biology in producing Fe(II) as a viableexplanation for the observed isotopic fractionation. Whencoupled with highly negative d30Si values for quartz inmagnetite-rich BIFs, the argument for a microbial reduc-tion pathway seems highly compelling.

5. CONCLUSIONS

Microbial Fe(III) reduction of Fe–Si gels produces thelargest yet measured Si isotope fractionation betweensolid-phase and aqueous Si, where the apparent isotopicfractionation increases with increasing extent of reduction.Collectively this study, and that of Zheng et al. (2016), indi-cates that the D30Sisolid–aqueous isotope fractionationbetween Fe–Si gel and aqueous Si moves from ��2.3‰for an Fe(II)/FeTotal ratio of zero for the gel, to ��2.7‰to �2.8‰ for an Fe(II)/FeTotal ratio of 0.02–0.03 for the

gel (controls in the current study). The largestD30Sisolid–aqueous isotope fractionations of ��3.4‰ to�3.5‰ were obtained at an Fe(II)/FeTotal ratio of �0.32where 32% of the gel was reduced. Following observationsby Zheng et al. (2016) that the presence of Fe(II) increasesSi isotope exchange rates, it is possible that the large Si iso-tope fractionations measured here approached those ofequilibrium conditions.

Microbial reduction of Fe(III)–Si gel readily producesmixed Fe(III)–Fe(II)–Si gels that, upon dewatering andburial diagenesis, could produce magnetite–quartz assem-blages that are characterized by highly negative d30Si val-ues. Recent in situ Si isotope studies of BIFs (Steinhoefelet al., 2010; Heck et al., 2011) have identified highly neg-ative d30Si values in quartz from magnetite-rich layers, ascompared to quartz from hematite-rich layers, providingcompelling evidence that the magnetite–quartz assem-blages reflect DIR. Although it is possible that the min-eralogical and Si isotope characteristics of such BIFsamples may be produced through abiologic interactionsbetween aqueous Fe(II) and Fe–Si gels, Fe exchangemay be inhibited through sorption of Si, and it is becom-ing increasingly apparent that Archean seawater aqueousFe(II) contents may not have been sufficient to attainsolid-phase Fe(II) contents required to produce mag-netite. In contrast, microbial reduction provides an effi-cient means for producing magnetite and highlynegative d30Si values in quartz from magnetite-rich layersin BIFs.

ACKNOWLEDGEMENTS

The authors thank Elizabeth Percak-Dennett for her insights inthe Fe–Si gel synthesis and microbial culture growth proceduresand Rie Fredrickson and Phillip Gopon for guidance on XRDand SEM analysis respectively. Comments made by Michael Tat-zel, two anonymous reviewere, and AE Edwin Schauble haveimproved the manuscript. This work was supported by the NASA


Astrobiology Institute under grant NNA13AA94A, and NationalScience Foundation grant 1122855.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.gca.2016.06.035.

REFERENCES

Andre L., Cardinal D., Alleman L. Y. and Moorbath S. (2006)Silicon isotopes in �3.8 Ga West Greenland rocks as clues tothe Eoarchaean supracrustal Si cycle. Earth Planet. Sci. Lett.

245(1–2), 162–173.Beard B. L., Johnson C. M., Cox L., Sun H., Nealson K. H. and

Aguilar C. (1999) Iron isotope biosignatures. Science 285(5435),1889–1892.

Beard B. L., Johnson C. M., Skulan J. L., Nealson K. H., Cox L.and Sun H. (2003) Application of Fe isotopes to tracing thegeochemical and biological cycling of Fe. Chem. Geol. 195(1–4),87–117.

Beard B. L., Handler R. M., Scherer M. M., Wu L., Czaja A. D.,Heimann A. and Johnson C. M. (2010) Iron isotope fraction-ation between aqueous ferrous iron and goethite. Earth Planet.

Sci. Lett. 295(1–2), 241–250.Bekker A., Slack J. F., Planavsky N., Krapez B., Hofmann A.,

Konhauser K. O. and Rouxel O. J. (2010) Iron formation: Thesedimentary product of a complex interplay among mantle,tectonic, oceanic, and biospheric processes. Econ. Geol. 103,467–508.

Beukes N. J. and Gutzmer J. (2008) Origin and paleoenvironmentalsignificance of major iron formations at the Archean-Paleo-proterozoic Boundary. Econ. Geol. 15, 5–47.

Chakrabarti R., Knoll A. H., Jacobsen S. B. and Fischer W. W.(2012) Si isotope variability in Proterozoic cherts. Geochim.

Cosmochim. Acta 91, 187–201.Clesceri L., Greenberg A. and Eaton A. (1989) Standard Methods

for the Examination of Water and Wastewater. American PublicHealth Association, Washington.

Crosby H. A., Johnson C. M., Roden E. E. and Beard B. L. (2005)Coupled Fe(II)–Fe(III) electron and atom exchange as amechanism for Fe isotope fractionation during dissimilatoryiron oxide reduction. Environ. Sci. Technol. 39(17), 6698–6704.

Crosby H. A., Roden E. E., Johnson C. M. and Beard B. L. (2007)The mechanisms of iron isotope fractionation produced duringdissimilatory Fe(III) reduction by Shewanella putrefaciens andGeobacter sulfurreducens. Geobiology 5(2), 169–189.

Czaja A. D., Johnson C. M., Roden E. E., Beard B. L., Voegelin A.R., Nagler T. F., Beukes N. J. and Wille M. (2012) Evidence forfree oxygen in the Neoarchean ocean based on coupled iron–molybdenum isotope fractionation. Geochim. Cosmochim. Acta

86, 118–137.Czaja A. D., Johnson C. M., Beard B. L., Roden E. E., Li W. and

Moorbath S. (2013) Biological Fe oxidation controlled depo-sition of banded iron formation in the ca. 3770 Ma IsuaSupracrustal Belt (West Greenland). Earth Planet. Sci. Lett.

363, 192–203.Dauphas N., Zuilen M. V., Wadhwa M., Davis A. M., Marty B.

and Janney P. E. (2004) Clues from Fe isotope variations on theorigin of early Archean BIFs from Greenland. Science 306

(5704), 2077–2080.Davis C. C., Knocke W. R. and Edwards M. (2001) Implications of

aqueous silica sorption to iron hydroxide: mobilization of iron

colloids and interference with sorption of arsenate and humicsubstances. Environ. Sci. Technol. 35(15), 3158–3162.

Delstanche S., Opfergelt S., Cardinal D., Elsass F., Andre L. andDelvaux B. (2009) Silicon isotopic fractionation during adsorp-tion of aqueous monosilicic acid onto iron oxide. Geochim.

Cosmochim. Acta 73(4), 923–934.Doelsch E., Rose J., Masion A., Bottero J. Y., Nahon D. and

Bertsch P. M. (2000) Speciation and crystal chemistry of iron(III) chloride hydrolyzed in the presence of SiO4 ligands. 1. AnFe K-Edge EXAFS study. Langmuir 16(10), 4726–4731.

Doelsch E., Stone W. E. E., Petit S., Masion A., Rose J., BotteroJ.-Y. and Nahon D. (2001) Speciation and crystal chemistry ofFe(III) chloride hydrolyzed in the presence of SiO4 ligands. 2.Characterization of Si–Fe aggregates by FTIR and 29Si solid-state NMR. Langmuir 17(5), 1399–1405.

Doelsch E., Masion A., Rose J., Stone W. E. E., Bottero J. Y. andBertsch P. M. (2003) Chemistry and structure of colloidsobtained by hydrolysis of Fe(III) in the presence of SiO4

ligands. Colloids Surf., A 217(1–3), 121–128.Douthitt C. B. (1982) The geochemistry of the stable isotopes of

silicon. Geochim. Cosmochim. Acta 46(8), 1449–1458.Dupuis R., Benoit M., Nardin E. and Meheut M. (2015)

Fractionation of silicon isotopes in liquids: the importance ofconfigurational disorder. Chem. Geol. 396, 239–254.

Fischer W. W. and Knoll A. H. (2009) An iron shuttle fordeepwater silica in late Archean and early Proterozoic ironformation. Geol. Soc. Am. Bull. 121(1–2), 222–235.

Frierdich A. J., Beard B. L., Reddy T. R., Scherer M. M. andJohnson C. M. (2014a) Iron isotope fractionation betweenaqueous Fe(II) and goethite revisited: new insights based on amulti-direction approach to equilibrium and isotopic exchangerate modification. Geochim. Cosmochim. Acta 139, 383–398.

Frierdich A. J., Beard B. L., Scherer M. M. and Johnson C. M.(2014b) Determination of the Fe(II)aq–magnetite equilibriumiron isotope fractionation factor using the three-isotope methodand a multi-direction approach to equilibrium. Earth Planet.

Sci. Lett. 391, 77–86.Georg R. B., Reynolds B. C., Frank M. and Halliday A. N. (2006)

New sample preparation techniques for the determination of Siisotopic compositions using MC-ICPMS. Chem. Geol. 235(1–2), 95–104.

Gunnarsson I. and Arnorsson S. (2000) Amorphous silica solubil-ity and the thermodynamic properties of H4SiO4 in the range of0� to 350 �C at Psat. Geochim. Cosmochim. Acta 64(13), 2295–2307.

Habicht K. S., Gade M., Thamdrup B., Berg P. and Canfield D. E.(2002) Calibration of sulfate levels in the Archean ocean.Science 298(5602), 2372–2374.

Handler R. M., Beard B. L., Johnson C. M. and Scherer M. M.(2009) Atom exchange between aqueous Fe(II) and goethite:an Fe isotope tracer study. Environ. Sci. Technol. 43(4), 1102–1107.

Heck P. R., Huberty J. M., Kita N. T., Ushikubo T., Kozdon R.and Valley J. W. (2011) SIMS analyses of silicon and oxygenisotope ratios for quartz from Archean and Paleoproterozoicbanded iron formations. Geochim. Cosmochim. Acta 75(20),5879–5891.

Hiemstra T., Barnett M. O. and van Riemsdijk W. H. (2007)Interaction of silicic acid with goethite. J. Colloid Interface Sci.

310(1), 8–17.Holland H. D. (1984) The Chemical Evolution of the Atmosphere

and Oceans. Princeton University Press, p. 582.Iler R. K. (1979) The Chemistry of Silica Solubility, Polymerization,

Colloid and Surface Properties, and Biogeochemistry.Kappler A., Pasquero C., Konhauser K. O. and Newman D. K.

(2005) Deposition of banded iron formations by anoxygenic



http://refhub.elsevier.com/S0016-7037(16)30365-9/h0005


































































































































phototrophic Fe(II)-oxidizing bacteria. Geology 33(11), 865–868.

Klein C. (2005) Some precambrian banded iron formations (BIFs)from around the world: their age, geologic setting, mineralogy,metamorphism, geochemistry, and origin. Am. Mineral. 90,1473–1499.

Li W., Huberty J. M., Beard B. L., Kita N. T., Valley J. W. andJohnson C. M. (2013) Contrasting behavior of oxygen and ironisotopes in banded iron formations revealed by in situ isotopicanalysis. Earth Planet. Sci. Lett. 384, 132–143.

Li W., Beard B. L. and Johnson C. M. (2015) Biologically recycledcontinental iron is a major component in banded iron forma-tions. Proc. Natl. Acad. Sci. U.S.A. 112(27), 8193–8198.

Liu Y., Wang Z., Liu J., Levar C., Edwards M. J., Babauta J. T.,Kennedy D. W., Shi Z., Beyenal H., Bond D. R., Clarke T. A.,Butt J. N., Richardson D. J., Rosso K. M., Zachara J. M.,Fredrickson J. K. and Shi L. (2014) A trans-outer membraneporin-cytochrome protein complex for extracellular electrontransfer by Geobacter sulfurreducens PCA. Environ. Microbiol.

Rep. 6(6), 776–785.Lovley D. R., Holmes D. E. and Nevin K. P. (2004) Dissimilatory

Fe(III) and Mn(IV) reduction. Adv. Microbiol. Physiol. 49, 219–286.

Maliva R. G., Knoll Andrew H. and Simonson Bruce M. (2005)Secular change in the Precambrian silica cycle: insights fromchert petrology. Geol. Soc. Am. Bull. 117(7-8), 835–845.

Newbury D. E. (2002) X-ray microanalysis in the variable pressure(environmental) scanning electron microscope. J. Res. Nat. Inst.

Stand. Technol. 107(6), 567.Ohmoto H., Watanabe Y. and Kumazawa K. (2004) Evidence

from massive siderite beds for a CO2-rich atmosphere before�1.8 billion years ago. Nature 429(6990), 395–399.

Ohmoto H., Watanabe Y., Yamaguchi K., Naraoka H., HarunaM., Kakegawa T., Hayashi K. and Kato Y. (2006) Chemicaland biological evolution of early Earth: constraints frombanded iron formations. Geol. Soc. Am. Mem. 198, 291.

Percak-Dennett E. M., Beard B. L., Xu H., Konishi H., Johnson C.M. and Roden E. E. (2011) Iron isotope fractionation duringmicrobial dissimilatory iron oxide reduction in simulatedArchaean seawater. Geobiology 9(3), 205–220.

Pollington A. D., Kozdon R., Anovitz L. M., Georg R. B.,Spicuzza M. J. and Valley J. W. (2016) Experimental calibra-tion of silicon and oxygen isotope fractionations betweenquartz and water at 250 �C by in situ microanalysis ofexperimental products and application to natural samples.Chem. Geol. 42, 127–142.

Posth N. R., Kohler I. D., Swanner E., Schroder C., Wellmann E.,Binder B., Konhauser K. O., Neumann U., Berthold C.,Nowak M. and Kappler A. (2013) Simulating Precambrianbanded iron formation diagenesis. Chem. Geol. 362, 66–73.

Rasmussen B., Krapez B., Muhling J. R. and Suvorova A. (2015)Precipitation of iron silicate nanoparticles in early Precambrianoceans marks Earth’s first iron age. Geology 43(4), 303–306.

Reddy T. R., Frierdich A. J., Beard B. L. and Johnson C. M.(2015) The effect of pH on stable iron isotope exchange andfractionation between aqueous Fe(II) and goethite. Chem. Geol.

397, 118–127.Reynolds B. C., Aggarwal J., Andre L., Baxter D., Beucher C.,

Brzezinski M. A., Engstrom E., Georg R. B., Land M., LengM. J., Opfergelt S., Rodushkin I., Sloane H. J., van den BoornS. H. J. M., Vroon P. Z. and Cardinal D. (2007) An inter-laboratory comparison of Si isotope reference materials. J.

Anal. At. Spectrom. 22(5), 561–568.Robert F. and Chaussidon M. (2006) A palaeotemperature curve

for the Precambrian oceans based on silicon isotopes in cherts.Nature 443, 969–972.

Roden E. E. and Lovley Derek. R. (1993) Dissimilatory Fe(III)reduction by the marine microorganism Desulfuromonas ace-

toxidans. Appl. Environ. Microbiol. 59(3), 734–742.Rye R., Kuo P. H. and Holland H. D. (1995) Atmospheric carbon-

dioxide concentrations before 2.2-billion years ago. Nature 378,603–605.

Satkoski A. M., Beukes N. J., Li W., Beard B. L. and Johnson C.M. (2015) A redox-stratified ocean 3.2 billion years ago. EarthPlanet. Sci. Lett. 430, 43–53.

Savage P. S., Armytage R. M. G., Georg R. B. and Halliday A. N.(2014) High temperature silicon isotope geochemistry. Lithos190–191, 500–519.

Sheldon N. D. (2006) Precambrian paleosols and atmospheric CO2

levels. Precambrian Res. 147, 148–155.Siever R. (1992) The silica cycle in the Precambrian. Geochim.

Cosmochim. Acta 56(8), 3265–3272.Stefurak E. J. T., Fischer W. W. and Lowe D. R. (2015) Texture-

specific Si isotope variations in Barberton Greenstone Beltcherts record low temperature fractionations in early Archeanseawater. Geochim. Cosmochim. Acta 150, 26–52.

Steinhoefel G., von Blanckenburg F., Horn I., Konhauser K. O.,Beukes N. J. and Gutzmer J. (2010) Deciphering formationprocesses of banded iron formations from the Transvaal andthe Hamersley successions by combined Si and Fe isotopeanalysis using UV femtosecond laser ablation. Geochim. Cos-

mochim. Acta 74(9), 2677–2696.Stookey L. L. (1970) Ferrozine-a new spectrophotometric reagent

for iron. Anal. Chem. 42(7), 779–781.Stumm W., Huper H. and Champlin R. L. (1967) Formulation of

polysilicates as determined by coagulation effects. Environ. Sci.Technol. 1(3), 221–227.

Svensson I. L., Sjoberg S. and Ohman L.-O. (1986) Polysilicateequilibria in concentrated sodium silicate solutions. J. Chem.

Soc., Faraday Trans. 82(12), 3635–3646.Tanakaa M. and Takahashib K. (2001) Silicate species in high pH

solution molybdate, whose silica concentration is determined bycolorimetry. Anal. Chim. Acta 429(1), 117–123.

van den Boorn S. H. J. M., van Bergen Manfred. J., NijmanWouter. and Vroon Pieter. Z. (2007) Dual role of seawater andhydrothermal fluids in early Archean chert formation: evidencefrom Si isotopes. Geology 35(10), 939–942.

van den Boorn S. H. J. M., van Bergen M. J., Vroon P. Z., de VriesS. T. and Nijman W. (2010) Silicon isotope and trace elementconstraints on the origin of �3.5 Ga cherts: implications forEarly Archaean marine environments. Geochim. Cosmochim.

Acta 74(3), 1077–1103.Vargas M., Kashefi K., Blunt-Harris E. L. and Lovley D. R. (1998)

Microbiological evidence for Fe(III) reduction on early Earth.Nature 395(6697), 65–67.

Wu L., Beard B. L., Roden E. E. and Johnson C. M. (2009)Influence of pH and dissolved Si on Fe isotope fractionationduring dissimilatory microbial reduction of hematite. Geochim.

Cosmochim. Acta 73(19), 5584–5599.Wu L., Beard B. L., Roden E. E., Kennedy C. B. and Johnson C.

M. (2010) Stable Fe isotope fractionations produced byaqueous Fe(II)–hematite surface interactions. Geochim. Cos-

mochim. Acta 74(15), 4249–4265.Wu L., Beard B. L., Roden E. E. and Johnson C. M. (2011) Stable

iron isotope fractionation between aqueous Fe(II) and hydrousferric oxide. Environ. Sci. Technol. 45(5), 1847–1852.

Wu L., Percak-Dennett E. M., Beard B. L., Roden E. E. andJohnson C. M. (2012) Stable iron isotope fractionation betweenaqueous Fe(II) and model Archean ocean Fe–Si coprecipitatesand implications for iron isotope variations in the ancient rockrecord. Geochim. Cosmochim. Acta 84, 14–28.









































































































































Young E. D., Galy A. and Nagahara H. (2002) Kinetic andequilibrium mass-dependent isotope fractionation laws innature and their geochemical and cosmochemical significance.Geochim. Cosmochim. Acta 66(6), 1095–1104.

Zheng X.-Y., Beard B. L., Reddy T. R., Roden E. E. and JohnsonC. M. (2016) Abiologic silicon isotope fractionation between

aqueous Si and Fe(III)–Si gel in simulated Archean seawater:implications for Si isotope records in Precambrian sedimentaryrocks. Geochim. Cosmochim. Acta 187, 102–122.

Associate editor: Edwin Schauble










silicon isotope fractionation during microbial reduction of … · 2017-01-08 · beard et al.,...

Documents