Abiologic silicon isotope fractionation between aqueous Si ... · Kinetic processes, such as Si precipitation, clearly aﬀect Si isotope fractiona-tion (e.g., Geilert et al., 2015),

$Page 1: Abiologic silicon isotope fractionation between aqueous Si ... · Kinetic processes, such as Si precipitation, clearly aﬀect Si isotope fractiona-tion (e.g., Geilert et al., 2015),$
Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 187 (2016) 102–122

Abiologic silicon isotope fractionation between aqueous Si andFe(III)–Si gel in simulated Archean seawater: Implications for

Si isotope records in Precambrian sedimentary rocks

Xin-Yuan Zheng ⇑, Brian L. Beard, Thiruchelvi R. Reddy, Eric E. Roden,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 5 May 2016; available online 14 May 2016

Abstract

Precambrian Si-rich sedimentary rocks, including cherts and banded iron formations (BIFs), record a >7‰ spread in30Si/28Si ratios (d30Si values), yet interpretation of this large variability has been hindered by the paucity of data on Si isotopeexchange kinetics and equilibrium fractionation factors in systems that are pertinent to Precambrian marine conditions. Usingthe three-isotope method and an enriched 29Si tracer, a series of experiments were conducted to constrain Si isotope exchangekinetics and fractionation factors between amorphous Fe(III)–Si gel, a likely precursor to Precambrian jaspers and BIFs, andaqueous Si in artificial Archean seawater under anoxic conditions. Experiments were conducted at room temperature, and inthe presence and absence of aqueous Fe(II) (Fe(II)aq).

Results of this study demonstrate that Si solubility is significantly lower for Fe–Si gel than that of amorphous Si, indicatingthat seawater Si concentrations in the Precambrianmay have been lower than previous estimates. The experiments reached�70–90% Si isotope exchange after a period of 53–126 days, and the highest extents of exchange were obtained where Fe(II)aq waspresent, suggesting that Fe(II)–Fe(III) electron-transfer and atom-exchange reactions catalyze Si isotope exchange throughbreakage of Fe–Si bonds. All experiments except one showed little change in the instantaneous solid–aqueous Si isotope frac-tionation factor with time, allowing extraction of equilibrium Si isotope fractionation factors through extrapolation to 100%isotope exchange. The equilibrium 30Si/28Si fractionation between Fe(III)–Si gel and aqueous Si (D30Sigel–aqueous) is �2.30± 0.25‰ (2r) in the absence of Fe(II)aq. In the case where Fe(II)aq was present, which resulted in addition of �10% Fe(II) inthe final solid, creating a mixed Fe(II)–Fe(III) Si gel, the equilibrium fractionation between Fe(II)–Fe(III)–Si gel and aqueousSi (D30Sigel–aqueous) is �3.23 ± 0.37‰ (2r). Equilibrium Si isotope fractionation for Fe–Si gel systems is significantly larger inmagnitude than estimates of a near-zero solid–aqueous fractionation factor between pure Si gel and aqueous Si, indicating amajor influence of Fe atoms on Si–O bonds, and hence the isotopic properties, of Fe–Si gel. Larger Si isotope fractionationin the Fe(II)-bearing systems may be caused by incorporation of Fe(II) into the solid structure, which may further weakenFe–Si bonds and thus change the Si isotope fractionation factor. The relatively large Si isotope fractionation for Fe–Si gel,relative to pure Si gel, provides a new explanation for the observed contrast in d30Si values in the Precambrian BIFs and cherts,as well as an explanation for the relatively negative d30Si values in BIFs, in contrast to previous proposals that the more negatived30Si values in BIFs reflect hydrothermal sources of Si or sorption to Fe oxides/hydroxides.� 2016 Elsevier Ltd. All rights reserved.

Keywords: Si isotopes; Fe–Si system; Fractionation factor; Precambrian; BIFs; Chert

http://dx.doi.org/10.1016/j.gca.2016.05.012

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

⇑ Corresponding author at: Department of Geoscience, University of Wisconsin–Madison, WI 53706, United States.E-mail address: [email protected] (X.-Y. Zheng).


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2016.05.012&domain=pdf

X.-Y. Zheng et al. /Geochimica et Cosmochimica Acta 187 (2016) 102–122 103

1. INTRODUCTION

The silica cycle in the Precambrian oceans was con-trolled by silica saturation due to the absence of Si secretingorganisms, in marked contrast to the modern marine silicacycle that is dominated by biological utilization (Treguerand De La Rocha, 2013). Silica concentrations in Precam-brian seawater, if determined by solubility of amorphoussilica, as well as sorption to silicate minerals, are estimatedto have been between �1 and �2 mM (Siever, 1992; Malivaet al., 2005; Konhauser et al., 2007). High silica concentra-tions in the Precambrian oceans, sustained by input fromcontinental and hydrothermal sources (Siever, 1992), haveproduced abundant Si-rich sedimentary rocks, includingcherts and banded iron formations (BIFs) (Maliva et al.,2005; Perry and Lefticariu, 2013). Formation of Si-rich pre-cipitates would also impact cycling of other elements in thePrecambrian oceans, including iron (Fe) and phosphorus(P) (e.g., Konhauser et al., 2007; Fischer and Knoll, 2009;Jones et al., 2015), and therefore study of secular changein the silica cycle bears on our general understanding ofgeochemical cycles in Precambrian marine systems.

Stable Si isotopes are potential tracers for the ancient sil-ica cycle. High-temperature fractionation is limited basedon the narrow range (<�0.5‰) of d30Si values for igneousrocks (see review by Savage et al., 2014). Surface processes,however, including continental weathering (e.g., Basile-Doelsch et al., 2005; Ziegler et al., 2005; Georg et al.,2007) and terrestrial and marine biological uptake (e.g.,De La Rocha et al., 1998; Ding et al., 2005; Opfergeltet al., 2006), can fractionate Si isotopes up to several permil (e.g., see review by Basile-Doelsch, 2006).

Published d30Si values for Precambrian cherts and BIFsspan a range of over 7‰ (Andre et al., 2006; Robert andChaussidon, 2006; van den Boorn et al., 2007, 2010;Steinhoefel et al., 2009, 2010; Abraham et al., 2011; Hecket al., 2011; Chakrabarti et al., 2012; Delvigne et al.,2012; Marin-Carbonne et al., 2014; Geilert et al., 2014a;Stefurak et al., 2015). Based on a simple Si mass-balancemodel, d30Si values in Precambrian cherts have been inter-preted to reflect seawater temperature (Robert andChaussidon, 2006; Marin-Carbonne et al., 2014). Thisinterpretation, however, was not based on an explicit deter-mination of Si isotope fractionation as a function of tem-perature, and was challenged by subsequent studies thatattributed changes in d30Si values in cherts to different silicasources (e.g., van den Boorn et al., 2007; Steinhoefel et al.,2009). It has been increasingly recognized that both inter-pretations are likely oversimplified (Chakrabarti et al.,2012; Stefurak et al., 2015), and some first-order observa-tions remain unexplained. For example, d30Si values in Pre-cambrian BIFs are often lower than those recorded inbroadly coeval cherts (Chakrabarti et al., 2012; Marin-Carbonne et al., 2014; Stefurak et al., 2015), and the causeof this is still debated. Recent discussions on the Si isotoperecord in the early Precambrian have stressed understand-ing the processes of Si cycling rather than focusing primar-ily on Si sources (e.g., Stefurak et al., 2015), and such anapproach requires detailed knowledge of stable Si isotopefractionation factors.

Experimental studies of Si isotope fractionations underconditions relevant to the Precambrian oceans have beenrelatively few. Adsorption experiments suggest that Si iso-tope fractionation factors (D30Sisorbed–fluid) range between��3‰ and ��1‰ during adsorption of Si onto Fe andAl oxides (Delstanche et al., 2009; Oelze et al., 2014). Batchand flow-through experiments yield solid–aqueous Si iso-tope fractionation factors that vary between ��3‰ and0‰ for 30Si/28Si during precipitation from Si-saturatedsolutions (Li et al., 1995; Geilert et al., 2014b; Oelzeet al., 2015; Roerdink et al., 2015), consistent with fieldstudies that document low d30Si values for siliceous cementsand secondary clay minerals (Basile-Doelsch et al., 2005;Ziegler et al., 2005; Georg et al., 2007). Kinetic processes,such as Si precipitation, clearly affect Si isotope fractiona-tion (e.g., Geilert et al., 2015), but all low-temperatureexperimental work to date has been conducted under con-ditions associated with strong Si precipitation and/or disso-lution, making differentiation between equilibrium andkinetic fractionation challenging. In addition, none of thelow-temperature experimental work to date has assessedSi isotope exchange rates.

Here we present the results of Si isotope exchange exper-iments that involved aqueous Si and amorphous Fe(III)–Sigel, using the ‘‘three-isotope method” to rigorously con-strain the extent and kinetics of isotopic exchange (e.g.,Matthews et al., 1983a,b; Young et al., 2015). In contrastto most previous experiments that explored silica precipita-tion in simple silica solutions (e.g., Li et al., 1995; Geilertet al., 2014b), our experiments were conducted in a syn-thetic seawater matrix intended to be analogous to Archeanseawater chemistry in key aspects such as high aqueous Fe(II) (Fe(II)aq) but low sulfate, building on earlier work thatdemonstrated significant effects on Fe isotope fractionationand exchange kinetics relative to simple pure systems (Wuet al., 2012). In addition, Fe(III)–Si gel, rather than pureFe(III) oxides/hydroxides, was used in the current workto represent a more appropriate marine precipitate thatwas a likely precursor to iron formations and jaspers thatformed through oxidation of Fe(II) in a Si- and Fe-richArchean ocean, which aligns our experiments with modelsfor a coupled Fe–Si shuttle in the Precambrian (Fischerand Knoll, 2009), as well as recent work that suggests animportant role for Fe–Si nanoparticles in the genesis of ironformations (Rasmussen et al., 2015). Our results demon-strate that the magnitude of equilibrium Si isotope fraction-ation as well as the isotopic exchange kinetics aresignificantly controlled by the presence or absence of Fe(II)aq, and that the unique Si isotope properties of Fe–Sigels can well explain previously puzzling aspects of the Siisotope compositions of certain Precambrian BIFs andcherts.

2. NOMENCLATURE AND EXPERIMENTAL DESIGN

2.1. Notation

Silicon isotope compositions are reported relative to theinternational standard NBS-28, and are expressed usingstandard d notation:

104 X.-Y. Zheng et al. /Geochimica et Cosmochimica Acta 187 (2016) 102–122

d29Si ¼29Si=28Sisample

29Si=28SiNBS-28

� 1

� �� 1000 ð1Þ

d30Si ¼30Si=28Sisample

30Si=28SiNBS-28

� 1

� �� 1000 ð2Þ

The Si isotope fractionation factor (a) between phase Aand B is defined as:

a29=28A–B ¼ ð29Si=28SiÞA

ð29Si=28SiÞBð3Þ

a30=28A–B ¼ ð30Si=28SiÞA

ð30Si=28SiÞBð4Þ

More conveniently, Si isotope fractionation betweenphase A and B can be expressed by DA–B, which directlylinks the Si isotope compositions of phase A and B to thefractionation factor a through the following relations:

D29SiA–B ¼ d29SiA � d29SiB � 103 ln a29=28A–B ð5ÞD30SiA–B ¼ d30SiA � d30SiB � 103 ln a30=28A–B ð6Þ

2.2. Application of the three-isotope method to Si isotopes

The three-isotope method was initially developed tostudy isotopic exchange kinetics and to allow estimatesfor equilibrium fractionation factors for O isotopes(Matsuhisa et al., 1978; Matthews et al., 1983a,b), and,more recently, has been extensively applied to stable Feand Mg isotopes (e.g., Shahar et al., 2008; Beard et al.,2010; Li et al., 2011; Frierdich et al., 2014a). This methodwas previously applied to experiments on Si isotope frac-tionation between silicate melt and Si-bearing Fe metalunder high-temperature conditions (Shahar et al., 2009).To the best of our knowledge, the experiments reportedhere represent the first application of this method to deter-mine Si isotope exchange kinetics and fractionation factorsin low-temperature experiments.

Fig. 1. A schematic diagram showing the principle of the three-isotope method, as applied to stable Si isotope exchange. Phase Acontains a 29Si-enriched tracer, whereas phase B contains isotopi-cally ‘‘normal” Si.

The three-isotope method involves isotope exchangebetween two components (Fig. 1), one of ‘‘normal” isotopiccomposition that initially plots on the primary mass-dependent fractionation line in a three-isotope (i.e. d–d) dia-gram, and the other that is enriched in one isotope (29Si inthis study) and hence initially plots on a distinct mass-dependent fractionation line. Isotopic compositions of thetwo components gradually evolve during isotopic exchange,and, at 100% exchange, will fall on a secondary mass-dependent fractionation line. For Si isotopes, d30Si andd29Si follow a mass-dependent relation ofd30Si � 1.93 � d29Si (Young et al., 2002).

Progressive isotopic exchange is represented on a d–ddiagram by rotation of the tie-line between the two compo-nents towards the secondary fractionation line, the locationof which is determined by the isotopic mass balance of thesystem and the slope of the mass-dependent fractionationline (Fig. 1). Because the three isotope method permitsextrapolation to 100% exchange, it provides one of the mostrobust methods of estimating equilibrium isotope fraction-ation factors, although it is important to note that 100%exchange may not be the same as attainment of isotopicequilibrium if the mechanism of exchange occurs undernon-equilibrium conditions (e.g., Li et al., 2011; Frierdichet al., 2014b).

2.3. Preparation of 29Si-enriched Archean Artificial Seawater

(AAS)

Archean Artificial Seawater (AAS) was prepared as theaqueous Si phase in our experiments, and followed therecipe from previous studies (Percak-Dennett et al., 2011;Wu et al., 2012) to mimic some key features of Archean sea-water chemistry, including low sulfate and high Si contents.AAS was adjusted and buffered by 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) to a pH of �7.4,consistent with pH estimates for the Archean oceans(Grotzinger and Kasting, 1993). It is important to note thatthere is no significant change in aqueous Si speciation at cir-cumneutral pH (�6–8) (Alexander et al., 1954; Iler, 1979),although pH may control Si solubility of Fe–Si gel, asdemonstrated recently in Fe(II)–Si systems (Tosca et al.,2015). Different batches of AAS with varied aqueous Siconcentrations were prepared with or without Fe(II)aq(details in Section 2.5). The goal of the exchange experi-ments was to operate as close to solubility as possible tominimize net precipitation or dissolution, and hence to pro-vide true isotopic exchange information. AAS that con-tained Fe(II)aq was prepared in an anaerobic chamber toprevent oxidation of Fe(II), and this was monitoredthrough controls.

The 29Si spike added to AAS was prepared from a 29Sistock solution from 99.9% enrichment 29SiO2 powders,fused by sodium hydroxide (NaOH, 99.99% purity) in a sil-ver crucible at �750 �C for �13 min, and followed by com-plete dissolution in ultrapure (18.2 MX-cm) water (Georget al., 2006). The concentrated 29Si spike solution wasmixed with isotopically ‘‘normal” sodium silicate (Na2SiO3)solution to produce AAS that had a d29Si value of �99‰,and a d30Si value of �0‰.


2.4. Preparation of Fe(III)–Si gel

Isotopically ‘‘normal” amorphous Fe(III)–Si gel was syn-thesized following the methods described in Percak-Dennettet al. (2011) and Wu et al. (2012). A solution containing100 mM Na2SiO3, 50 mM FeCl2 and 100 mM NaHCO3

was allowed to oxidize under ambient laboratory condi-tions for �18 days with continuous shaking. Iron and Sico-precipitated during gradual oxidation of Fe(II) andassociated Fe hydrolysis, forming an amorphous Fe(III)–Si gel. The Fe(III)–Si gel was centrifuged and washed withultrapure water to remove excessive salts before use inexchange experiments. Iron speciation measurements bythe Ferrozine method (details in Section 3.1) showed that>98% of Fe present in the solid was Fe(III). Two batchesof Fe(III)–Si gel (termed ‘‘A” and ‘‘B”) were produced dur-ing the course of this study. Analysis of Fe and Si contentsfor both batches of gels yielded a consistent Si:Fe molarratio of 1:0.475, which, for convenience, we will refer toas 2:1. The Fe(III)–Si gel, therefore, may be stoichiometri-cally described as Fe(III)Si2(OH)11 based on charge bal-ance. X-ray Diffraction (XRD) analysis of the Fe–Si gelused in the current study indicates an amorphous nature(Supplementary Information, Fig. S1). Previous spectro-scopic studies indicate that co-precipitation of Fe and Sifrom a mixed solution tends to form an intimately bondedFe–Si phase, rather than separate Fe and Si phases(Doelsch et al., 2000, 2001, 2003). This is consistent withtransmission electron microscopy (TEM) analysis on Fe(III)–Si gel produced with the same synthesis procedure inour previous companion study, which showed a homoge-nous distribution of Fe and Si atoms at the nanometer scalein the gel (Wu et al., 2012).

2.5. Experimental details

All experiments were conducted in an anaerobic cham-ber (Coy Laboratory, MI, USA), filled with �4% (v/v)H2 and �96% Ar mixed gas. Because O2 was continuouslyremoved by the palladium-catalyzed reaction with H2 insidethe anaerobic chamber, the O2 level in the chamber wasalways maintained below the detection limit of the O2 mon-itor (<1 ppm). Reagents used in the experiments, includingAAS, were sparged using oxygen-free N2 gas before beingtransferred into the anaerobic chamber. Both reagentsand experimental apparatus were left to equilibrate withthe atmosphere in the anaerobic chamber for at least�24 h prior to use. The lack of oxidation of Fe(II) duringthe experiments was confirmed by constant Fe(II)aq concen-trations measured in control reactors that contained Fe(II)aq-bearing AAS but no Fe–Si gel.

Five exchange experiments, differing mainly in initial Siconcentrations and the presence or absence of initial Fe(II)aqin AAS, were conducted at room temperature (�23 �C)(Table 1). The pH of starting AAS was not buffered forExperiment 1 (pH � 7.8), but was buffered to �7.4 withHEPES for all other experiments. Each experiment com-prised a series of parallel batch reactors, each in 15 ml plasticcentrifuge tubes which contained identical amounts of Fe(III)–Si gel and 10 ml of AAS. Reactors that contained Fe

(II)aq were wrapped in aluminum-foil to avoid possiblephoto-oxidation of Fe(II). To avoid cross-contamination,one parallel reactor for each experiment, treated in the sameway as sample reactors, was dedicated to monitor pH of thesolution. Control reactors paralleled each experiment, andonly contained AAS with or without Fe(II)aq; the controlreactors allowed for monitoring of changes in Si polymeriza-tion and, for Fe(II)aq-bearing controls, oxidation of Fe(II).All reactors were continuously rolled on an end-over-endrotator throughout the experiments.

Experiment 1 was designed to have an initial aqueous Siconcentration (2.33 mM) that slightly exceeds the solubilityof amorphous Si (1.86 mM at 23 �C; Gunnarsson andArnorsson, 2000), and did not contain Fe(II)aq. Experi-ments 2a and 2b used AAS that had a starting aqueous Siconcentration of 0.71 mM, a level chosen based on theresults of Experiment 1, as well as previous work thatdemonstrated a lower solubility of Fe–Si gel (i.e., �0.45–1.5 mM Si) relative to that of pure amorphous Si (Wuet al., 2011, 2012). Experiment 2a used Fe(II)aq-free AAS,but Experiment 2b used AAS that contained an initial Fe(II)aq concentration of 0.92 mM; this concentration waslower than that used in our earlier work (1.8 mM; Wuet al., 2012), but lies closer to the range estimated forArchean seawater (Czaja et al., 2012). Experiments 1, 2a,and 2b all used the same batch of Fe(III)–Si gel, denotedas ‘‘Fe–Si gel (A)” (Table 1). Experiments 3a and 3b reacteda second batch of Fe(III)–Si gel, denoted as ‘‘Fe–Si gel (B)”,with AAS that had an initial aqueous Si concentration of0.99 mM, a concentration level slightly higher than thatin Experiment 2. Experiment 3a used Fe(II)aq-free AAS,and Experiment 3b used AAS that initially contained0.96 mM Fe(II)aq. All experiments were started when amor-phous Fe(III)–Si gel was added into AAS.

Fe(III)–Si gel and AAS in parallel reactors were left toreact for varied durations before sampling, and parallelreactors were sampled as a time series over 53–126 days.The pH of the aqueous phase was checked in the dedicatedreactor in each set of experiments with a pH meter at eachsampling time point. For sampling, the solid phase was sep-arated by centrifuging at 10,000 rpm (6238g RCF) for�5 min, and the aqueous phase was filtered through 0.22-lm filters, followed by centrifuging at 10,000 rpm foranother �10 min. The control reactors, which did not con-tain Fe(III)–Si gel, were typically sampled at the end of theexperiments. Although no visible precipitate was observedin control reactors for all experiments, AAS collected fromcontrol reactors was centrifuged at 10,000 rpm for �10 minprior to collection of only the upper portion of the solutionfor Si and Fe concentration measurements. Sampling,together with subsequent sample dissolution and measure-ments of Fe contents, was conducted in the anaerobicchamber.

3. ANALYTICAL PROCEDURES

3.1. Analysis of Si and Fe concentrations

Silica concentrations were measured colorimetricallyusing the heteropoly blue method (Clesceri et al., 1999).

Table 1A summary of experimental details.

ExperimentNo.

Description AAS Fe–Si gel

[Fe(II)aq](mM)

[Si](mM)

Total Si mass(lmole)

d30Si(‰)

d29Si(‰)

Percentexchange

Fe:Si moleratio

Total Si mass(lmole)

d30Si(‰)

d29Si(‰)

Percentexchange

Exp-1 Fe(II)aq-free AAS + Fe–Si gel(A)Initial conditions – 2.33 23.29 0.00 99.21 0% 1:2 40.36 �0.07 �0.04 0%Final conditions (126 days) – 0.88 8.80 2.24 49.54 80% 54.85 �0.47 28.68 80%

Exp-2a Fe(II)aq-free AAS + Fe–Si gel(A)Initial conditions – 0.71 7.10 0.07 98.63 0% 1:2 40.36 �0.07 �0.04 0%Final conditions (80 days) – 0.90 9.00 1.50 26.16 88% 38.46 �0.36 11.06 76%

Exp-2b Fe(II)aq-containing AAS+ Fe–Si gel (A)Initial conditions 0.92 0.71 7.10 0.09 98.29 0% 1:2 40.36 �0.07 �0.04 0%Final conditions (80 days) 0.10 0.83 8.30 3.06 16.65 99% 39.16 �0.56 13.17 91%

Exp-3a Fe(II)aq-free AAS + Fe–Si gel(B)Initial conditions – 0.99 9.89 �0.05 97.41 0% 1:2 50.36 �0.07 �0.04 0%Final conditions (53 days) – 0.94 9.40 1.61 36.14 76% 50.85 �0.29 13.35 85%

Exp-3b Fe(II)aq-containing AAS +Fe–Si gel (B)Initial conditions 0.96 0.99 9.89 �0.03 98.26 0% 1:2 50.36 �0.07 �0.04 0%Final conditions (53 days) 0.18 0.79 7.90 2.51 29.92 84% 52.35 �0.34 14.73 93%

106X.-Y

.Zhenget

al./Geochim

icaet

Cosm

ochim

icaActa

187(2016)

102–122


This method measures molybdate-reactive Si, which com-prises monomeric silica, and possibly some low-weight sil-ica oligomers (e.g., dimers, trimers) (Iler, 1979; Tanakaaand Takahashib, 2001). Briefly, Si-bearing solutions werereacted with ammonium molybdate, and were then reducedby 1-amino-2-naphthol-4-sulfonic acid to produce blue-colored Si complexes. Light absorbance of samples wasmeasured on a spectrophotometer at the wavelength of815 nm. Conversion of absorbance values to Si concentra-tions was calibrated against gravimetrically prepared Sistandards with known concentrations at the beginning ofeach analysis session. The precision of the method is esti-mated to be �5% (1r).

Ferrous Fe concentrations were measured colorimetri-cally on a spectrophotometer at the wavelength of562 nm, using the Ferrozine method (Stookey, 1970). TotalFe concentrations were measured by the same method afteradding a reducing reagent, hydroxylamine hydrochloride.Ferric Fe contents are calculated from the differencebetween measured Fe(II) and total Fe contents. The instru-ment was calibrated against gravimetrically prepared Fesolutions with known Fe concentrations before each analyt-ical session. The precision of the method is �5% (1r).

3.2. Sample pre-treatment

Aqueous samples from the experiments were acidifiedwith Teflon–distilled 7 M HCl to pH �2–3 (checked bypH papers), followed by chromatographic purification forSi. Solid samples of Fe–Si gel were first dissolved in0.5 M HCl for at least 12 h, and then centrifuged to sepa-rate the supernatant (primarily Fe fraction) and insolubleSi gel. The resulting fractions of white Si gel were then dis-solved in 0.5 M NaOH, prepared from NaOH beads of99.99% purity, for another 12 h at least until complete dis-solution was achieved, followed by dilution with ultrapurewater to a Si concentration of �30 ppm. The diluted Sifractions were acidified with 7 M HCl, and subsequentlycombined with the Fe fractions. Final sample solutionshad a Si concentration of �25 ppm and a pH between 2and 3.

3.3. Si isotope analysis

Silicon was purified following the chromatographicapproach of Georg et al. (2006), in which sample solutionswere passed through Bio-Rad AG 50W–12X (H form, 200–400 mesh) cation-exchange resin. Under the ion-exchangeconditions used, Si existed predominantly as unchargedmonosilicic acid (H4SiO4) in equilibrium with minor anio-nic species (H3SiO4

�), and therefore Si passes through theresin directly. Ultrapure water was used to elute the remain-der of Si from the resin to achieve an average yield of�100%.

Isotopic compositions for Si were measured on an MC–ICP–MS (Nu plasma II) in the Department of Geoscience,University of Wisconsin–Madison. Polyatomic interfer-ences on Si isotopes, notably 14N16O+ on 30Si+, wereresolved in pseudo-high-resolution mode, achieved by nar-rowing the width of the source defining slit and two alpha

slits. Sample solutions were introduced into the instrumentvia a self-aspirating nebulizer (�40–100 ll/min uptake rate)and a peltier-cooled cyclonic spray chamber (i.e., ‘‘wet plas-ma”). Typical Si concentration for the isotopic measure-ments was �3 ppm, yielding an ion intensity of �5–6volts of 28Si on Faraday collectors that were equipped with1011 X resistors. Backgrounds from the instrument andultrapure water used during sample pre-treatment andchromatographic separation were corrected using an on-peak zero routine, where ion intensities of Si isotopes weredetermined in the ultrapure water before each analysis andsubtracted from their respective beam intensities measuredin the subsequent analyte. On-peak zero intensities,accounting for <1% (usually �0.2%) of Si signals in ana-lytes, were negligible.

Silicon isotope ratios were measured using a sample–standard bracketing protocol to correct for instrumentalmass bias. The international standard for Si isotopes,NBS-28, was used as the bracketing standard, and all Si iso-tope data reported in this study are relative to NBS-28. Ref-erence materials, including Diatomite and BHVO-2, wereroutinely processed through the entire ion-exchange col-umn separation along with samples, and measured in eachanalytical session to check accuracy and precision of the Siisotope analyses. Each sample was typically measured threetimes during an analytical session. Measured d29Si andd30Si values for reference materials and samples that werenot spiked with the 29Si tracer fell on a mass-dependentfractionation line. Repeated analyses of Diatomite yieldedd29Si = 0.66 ± 0.12‰ and d30Si = 1.28 ± 0.20‰ (2r,n=43), and repeated analyses of BHVO-2 yieldedd29Si = �0.11±0.12‰ and d30Si = �0.21 ± 0.19‰ (2r,n = 64). Our results are in good agreement with publishedconsensus values for these materials (Diatomite:d29Si = 0.64‰, d30Si = 1.26‰; BHVO-2: d29Si = �0.15‰,d30Si = �0.28‰) (Reynolds et al., 2007; Savage et al.,2014). Long-term reproducibility (2r), based on repeatedmeasurements of the reference materials, is estimated tobe ±0.15‰ for d29Si and ±0.20‰ for d30Si.

The accuracy of the Si isotope measurements wasassessed using cation and anion doping tests. Differentamounts of FeCl3 solution was added into solution of a ref-erence material, Big Batch, to yield Si:Fe molar ratios of1:1, 1:2.5 and 1:5. The Fe-doped solutions were processedthrough the ion-exchange column separation and then mea-sured for their Si isotope compositions. Solutions dopedwith varied amount of Fe yielded Si isotope compositionsthat were indistinguishable from each other, as well asresults measured for pure Big Batch, suggesting that effec-tive removal of Fe was achieved during chromatographicseparation. For Big Batch, the pooled d30Si value is�10.71 ± 0.22‰ (2r, n = 8), which agrees within error ofthe inter-laboratory calibration result of �10.48 ± 0.54‰(2r) (Reynolds et al., 2007). Because solid samples of Fe–Si gel from the experiments only had a Si:Fe molar ratioof �1:0.5, the Fe-doping tests indicate that a matrix effectassociated with cations during analysis is unlikely.

Anions cannot be separated from Si during the ion-exchange separation due to the use of cation exchangeresin. Because Cl� was the predominant anion in AAS

(A) Exp-1 – no Fe(II)aq

amorphous Si solubility1.86 mM (23°C)

(C) Exp-2b and 3b – with Fe(II)aq

(B) Exp-2a and 3a – no Fe(II)aq

Fig. 2. Silica concentrations of Archean Artificial Seawater (AAS)measured at each sampling time point. (A) Results of Experiment 1that used Fe(II)aq-free AAS and initial Si concentrations thatexceeded that of amorphous Si gel (horizontal dashed line indicatesamorphous Si solubility in H2O at 23 �C; Gunnarsson andArnorsson, 2000). (B) Results of Experiments 2a (red) and 3a(blue) that used Fe(II)aq-free AAS. (C) Results of Experiments 2b(red) and 3b (blue) that used AAS which contained �1 mM Fe(II)aq. The vertical gray bar between Day 3 and 5 in (B) and (C)indicates an axis-break. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web versionof this article.)


(�0.6 M), variable amounts of HCl were doped into anNBS-28 solution that had been processed after column sep-aration to produce Cl� concentrations ranging from 0.02 Mup to 0.5 M; these solutions were then directly measured onthe MC-ICP-MS against the same NBS-28 solution that didnot contain added HCl. No matrix effect was detected withthe increased Cl� concentration. Potential matrix effectsassociated with anions were further evaluated through com-parison of the Si isotope compositions measured for puresodium silicate that was dissolved in ultrapure water, withthose measured for AAS that was prepared using the samesodium silicate powers. The measured d30Si value of 0.13‰for sodium silicate dissolved in ultrapure water agreeswithin analytical uncertainty (±0.2‰) with the averaged30Si value of 0.02‰ measured for different batches ofAAS that had a more complex anionic matrix. This agree-ment is expected, because, without a matrix effect, additionof the 29Si spike at the level of enrichment used in ourexperiments will only shift d30Si values of AAS by<0.05‰. These tests, therefore, demonstrate that matrixeffects are entirely negligible, and the accuracy of the Si iso-tope measurement is similar to the precision.

4. RESULTS

4.1. Temporal changes in aqueous chemistry and solid phase

The pH of the aqueous phase for Experiment 1, whichwas not pH-buffered, showed a minor drift from 7.8 to8.0 over 126 days. For Experiments 2a and 3a, which didnot contain Fe(II)aq, but were pH-buffered, the pH of aque-ous phase remained at 7.4 throughout the experiments (80and 53 days, respectively). In contrast, there was a rapid�0.2-unit pH drop after addition of Fe(III)–Si gel in Exper-iments 2b and 3b, which contained Fe(II)aq, but after thisinitial drop, pH remained constant at �7.2 for the durationof these experiments (80 and 53 days, respectively).

Aqueous Si concentrations in Experiment 1 decreasedcontinuously from the initial concentration of2.33 to 0.88 mM over the 126 day experiment (Fig. 2), indi-cating a loss of 62% Si from solution. Aqueous Si concen-trations in the control for Experiment 1, which containedAAS but no Fe(III)–Si gel, decreased from 2.33 mM to1.89 mM (Appendix Table A.1). Fe(III)–Si gel dissolvedshortly after initiation of Experiments 2a and 2b, as demon-strated by an increase in Si concentration of AAS at thebeginning of both experiments (Fig. 2). For Experiment2a, aqueous Si concentrations increased rapidly from theinitial value of 0.71 mM to �0.9 mM within �4 h, corre-sponding to dissolution of �5% of total Si from the solid,and then gradually increased to �1 mM within a weekbefore returning to a relatively constant value of�0.92 mM during the rest of the experiment. In Experiment2b, which contained Fe(II)aq, aqueous Si concentrationsincreased from the initial value of 0.71 mM to �0.9 mMduring the first 5 days, corresponding to dissolution of�4% of total Si from the solid, and then returned to a con-stant level of �0.85 mM during the rest of the experiment.For Experiment 3a, aqueous Si concentrations remained ata relatively constant value of �0.95 mM for the duration of

the experiment, essentially the same as the initial Si concen-tration of 0.99 mM within the analytical uncertainty(Fig. 2). For Experiment 3b, however, which contained


Fe(II)aq, there was net precipitation of aqueous Si, asdemonstrated by a decrease in Si concentration of AASfrom the initial 0.99 mM to �0.80 mM (Fig. 2). Controlsfor Experiments 2 and 3 showed no change in aqueous Siconcentrations, indicating negligible Si polymerizationand precipitation (Appendix Table A.1).

Although Si concentrations in initial AAS were differentin Experiments 2a and 3a (0.71 and 0.99 mM, respectively),they reached a similar value of �0.95 mM within a few days(Fig. 2). Similarly, Si concentrations of AAS in the Fe(II)aq-bearing Experiment 2b and 3b, starting at different initiallevels, evolved towards a similar final value of �0.82 mM.The Si solubility of Fe(III)–Si gel in AAS containing�1 mM Fe(II)aq (Experiments 2b and 3b) is slightly lowerthan that in AAS without Fe(II)aq (Experiments 2a and3a). The final equilibrium aqueous Si concentrations forboth sets of experiments were significantly lower than thesolubility of amorphous Si (1.86 mM) in H2O at room tem-perature (23 �C) (Gunnarsson and Arnorsson, 2000), but fitwell between the solubility determined for Fe(II)–Si precip-itates at pH 7.0 and 7.5 (Tosca et al., 2015).

Aqueous Fe(II)aq contents decreased with time for all Fe(II)aq-bearing experiments (Experiments 2b and 3b)(Fig. 3), similar to what was previously observed in Fe iso-tope exchange experiments with the same Fe(III)–Si gel inAAS (Wu et al., 2012). The lack of changes in Fe(II)aq con-tents in control reactors that contained Fe(II)aq-bearingAAS but no Fe(III)–Si gel demonstrates that the Fe(II)aqdecrease in our experiments is not due to oxidation (Appen-dix Table A.1). Transfer of Fe(II)aq to Fe–Si gel was alsoconfirmed in Experiment 3b where the measured Fe(II)addition in the solid phase matched the Fe(II) loss fromthe aqueous phase within the analytical uncertainty at eachsampling time point. Our new results may be directly com-pared with those of Wu et al. (2012) (Fig. 3), where thesame amount of Fe(III)–Si gel was used and the reactor vol-umes were the same, although the initial Fe(II)aq contentsof the experiments of Wu et al. (2012) were higher

Fig. 3. Temporal changes in absolute amount of Fe(II)aq inExperiments 2b and 3b of this study, compared with analogousexperiments reported in Wu et al. (2012). The decrease in moles ofFe(II)aq reflects a combination of sorption and incorporation of Fe(II) into the Fe–Si gel, producing a change in stoichiometry fromFe(III)Si2(OH)11, to Fe(II)0.1Fe(III)Si2(OH)11.2. See text fordiscussion.

(�1.8 mM) relative to the current study (�1 mM). Despitethe different initial Fe(II)aq contents, the absolute amount(i.e., mole) of Fe(II) loss from solution were very similarbetween the two studies, at �8 � 10�6 moles. Wu et al.(2012) did 5 mM HCl and 0.5 M HCl extractions, whichprovides an assessment of loosely bound Fe(II) sorbed onFe–Si gel and Fe(II) incorporated into the structure ofFe–Si gel, respectively. They found that a total of�5 � 10�6 moles of Fe(II) was sorbed, and �3 � 10�6

moles of Fe(II) were contained in the solid phase, equiva-lent to a change in stoichiometry from that of the initialmaterial, Fe(III)Si2(OH)11, to Fe(II)0.1Fe(III)Si2(OH)11.2at the end of their experiments. We did not perform HClextractions in the current study because it was unknownif this would affect Si isotope compositions of the solid,but if we assume the similar Fe(II)aq losses that wereobserved in the current study as compared to Wu et al.(2012) are apportioned the same between sorbed and solidcomponents, we would estimate the same solid phase stoi-chiometry Fe(II)0.1Fe(III)Si2(OH)11.2 at the end of the Fe(II)aq-bearing experiments in the current study. Suchchanges are consistent with visible color changes in the solidphase for Experiments 2b and 3b, where the initial browncolor of the Fe(III)–Si gel changed to green, indicatingthe presence of Fe(II) in the solid.

XRD analysis of the Fe(III)–Si and Fe(II)–Fe(III)–Sigels indicates that the solid remained X-ray amorphousthroughout all experiments in this study (SupplementaryInformation, Fig. S1). This is consistent with the XRDand TEM results of our previous companion Fe isotopestudy that has shown the amorphous nature of the gel dur-ing the course of experiments in a highly similar AAS andFe–Si gel system (Wu et al., 2012). Our observation is, how-ever, different from a recent study that found nm-size Fe(II)–Si nanoparticles in Fe(II)–Si gel formed in similarexperimental settings (Tosca et al., 2015), probably reflect-ing inhibition of crystallization due to higher Si contents inour Fe–Si gel (Si:Fe = �2:1) relative to those (Si:Fe = �1:1) produced in Tosca et al. (2015).

4.2. Evolution of Si isotope compositions

Extensive Si isotope exchange was evident from an ini-tial rapid and then gradual convergence in d29Si values ofaqueous and solid phases with time in all experiments(Fig. 4). For Experiments 2a and 2b, which involved mod-est extents of solid dissolution, the aqueous d29Si valuesdecreased more than those of the aqueous component forExperiments 3a and 3b, which involved negligible, or smallamounts of, net precipitation. Overall, the Fe(II)aq-bearingexperiments had larger changes in d29Si values than the Fe(II)aq-free experiments (Fig. 4). Accompanying isotopicexchange, as monitored using the enriched 29Si tracer, therewas a divergence in d30Si values between the two phases,where d30Si values of the aqueous phase increased(Fig. 4). The magnitude of d30Si changes was larger in theaqueous Si than the Fe(III)–Si and Fe(II)–Fe(III)–Si gels,because most of the Si mass resided in the gels for all exper-iments (Table 1). For Experiment 1, d30Si values of aqueousSi increased from �0‰ to 2.24‰ over 126 days, and d30Si

Exp–1 (no Fe(II) )aq Exp–2a and 3a (no Fe(II) )aq Exp–2b and 3b (with Fe(II) )aq(A) (B) (C)

Fig. 4. Plots of d29Si and d30Si values of Fe–Si gel and AAS as a function of time for all experiments conducted in this study. Error bars (2r)shown where they exceed the size of the symbol.


values in the gel decreased from �0‰ to ��0.5‰. For Fe(II)aq-free Experiments 2a and 3a, which involved smallamounts of net Si dissolution or precipitation, respectively,both experiments show closely similar d30Si results; d30Si ofaqueous Si gradually increased from �0‰ to �1.5‰, andd30Si values of the gel decreased from �0‰ to ��0.3‰.Consistent d30Si results were also obtained for Fe(II)aq-bearing Experiments 2b and 3b; aqueous d30Si values grad-ually increased from �0‰ to �2.5‰, and d30Si values ofthe Fe(II)–Fe(III)–Si gel decreased from �0‰ to��0.5‰. The magnitude of the changes in d30Si valuesfor the Fe(II)aq-bearing experiments were larger than thoseobserved in Fe(II)aq-free experiments (Fig. 4).

4.3. Calculation of fraction of Si isotope exchange

Following the approach of previous studies (e.g., Coleand Chakraborty, 2001; Johnson et al., 2002; Li et al.,2011; Wu et al., 2012), the fraction of Si isotope exchange(F) can be calculated using the following equation:

F ¼ dt � dide � di

ð7Þ

where dt indicates the Si isotope composition of the sampleat any given time t; di and de represent initial and equilib-rium Si isotope compositions, respectively. The parameterF quantifies the extent to which the system has moved tocomplete isotopic exchange, although, in detail, Eq. (7)does not directly quantify the number of atoms in the solidor solution that have been mixed (Frierdich et al., 2015).The fraction of exchange is useful to investigate the kineticsof isotope exchange (e.g., Cole and Chakraborty, 2001), aswell as to evaluate changes in the instantaneous fractiona-

tion factors that might indicate kinetic isotope effectsduring isotopic exchange (Frierdich et al., 2014a,b).

The de values for phase A (de(A)) and B (de(B)) in a closedsystem can be expressed by the mass-balance equation atequilibrium:

ðNA þ NBÞ � dmean ¼ NA � deðAÞ þ NB � deðBÞ ð8Þwhere NA and Nb indicate moles of Si in phase A and B,and dmean represents the mean Si isotope composition ofthe system. dmean can be constrained by a similar mass-balance equation based on initial materials:

ðNA þ NBÞ � dmean ¼ NA � diðAÞ þ NB � diðBÞ ð9Þwhere di(A) and di(B) represent initial Si isotopic composi-tions of phase A and B.

If there is no natural, mass-dependent, Si isotopefractionation between the two phases at equilibrium (i.e.DA–B = 0), de(A) equates to de(B), and equates to dmean. Con-sequently, F can be calculated based on Eqs. (7)–(9) withoutuncertainty. In the case where DA–B – 0, de(A) and de(B)depend on DA–B, which can be expressed by rearrangingEq. (8) as:

deðAÞ ¼ dmean þ NB

NA þ NB

� �� DA–B ð10Þ

deðBÞ ¼ dmean � NA

NA þ NB

� �� DA–B ð11Þ

Substituting Eqs. (9)–(11) into Eq. (7), calculation of F,which can be based on either phase A (FA) or B (FB), alsodepends on DA–B:

F A ¼ dtðAÞ � diðAÞf � ½ðdiðBÞ � diðAÞÞ þ DA–B� ð12Þ


F B ¼ dtðBÞ � diðBÞðf � 1Þ � ½ðdiðBÞ � diðAÞÞ þ DA–B� ð13Þ

where

f ¼ NB

ðNA þ NBÞ ð14Þ

In principle, either d29Si or d30Si can be used to calculateF if DA–B is accurately known. Because, however, DA–B isthe parameter that our experiments are intended to deter-mine, it is not accurately known. It can be inferred fromEqs. (12) and (13) that the larger the initial isotopic contrastbetween starting materials (i.e. di(B) – di(A)) relative to DA–B,the less influence of DA–B would impose on the calculationof F, and hence use of an enriched 29Si tracer minimizes theeffect of DA–B on the calculation of F using d29Si data.Results of our experiments show that apparent fractiona-tion factors of d30Si between Fe(III)–Si (or Fe(II)–Fe(III)–Si) gel and aqueous Si, based on the last d30Si data inthe time series of each experiment, range between ��2‰and ��4‰ (Fig. 4). Because the mass-dependent fraction-ation law predicts D30Si � 1.93 � D29Si (Young et al.,2002), D29Sigel–aqueous is likely to be within the range of�1‰ and �2‰. A value of �1.5‰ for D29Sigel–aqueouswas applied to Eqs. (12) and (13) for all experiments to cal-culate F (Appendix Table A.1); if, instead, a D29Sigel–aqueousvalue of zero were assumed, the calculated F values wouldonly differ by <2%.

Although our goal was to conduct experiments veryclose to Si solubility to minimize the possibility of kineticisotopic effects caused by net precipitation or dissolution,net Si mass transfer occurred in most of our experiments(Fig. 2), and this produces slightly different F values calcu-lated from aqueous or solid components. For Experiments2a and 2b, small amounts of net dissolution of the solidduring these experiments preferentially transferred isotopi-cally ‘‘normal” Si to the solution, producing slightly higherextents of apparent exchange (Appendix Table A.1); in suchcases, the fraction of true exchange is best estimated using Fvalues calculated based on the solid phase results, althoughit should be recognized that solid-phase F values are lessprecise in the current study given the smaller overall shiftin 29Si/28Si ratios due to mass balance. For Experiments1, 3a and 3b, where net Si precipitation occurred, the frac-tion of true isotopic exchange is best estimated using F val-ues calculated from the aqueous phase.

Fig. 5. Changes in d30Si values of Fe(III)–Si gel (green squares)and aqueous Si (red diamonds) as a function of the fraction ofisotope exchange (F; calculated based on aqueous phase) forExperiment 1. This experiment involved large changes in aqueousSi concentration, reflecting rapid net precipitation. The non-lineartrajectory of d30Si values for aqueous Si is interpreted to reflectchanges in the instantaneous isotope fractionation factor, record-ing a variable kinetic isotope effect. This precludes extrapolation to100% isotope exchange to infer an equilibrium isotope fractiona-tion factor. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of thisarticle.)

4.4. Silicon isotope fractionation factors

Inference of Si isotope fractionation factors at 100%exchange is possible via extrapolation of partial exchangeresults using the three-isotope method. Extrapolation ofdata is not possible for Experiment 1 because of a non-linear trajectory of d30Si towards 100% exchange (Fig. 5),suggesting variable kinetic isotope effects (e.g., Frierdichet al., 2014b); a fractionation factor, therefore, was notdetermined for this experiment. Extrapolation of data forExperiments 2a and 3a to 100% exchange yielded consistentfractionation factors between Fe(III)–Si gel and aqueous Si(D30Sisolid–AAS) of �2.28 ± 0.17 (2r) ‰ and �2.31 ± 0.19

(2r) ‰, respectively (Fig. 6). Extrapolation of data forExperiments 2b and 3b also yielded consistent fractionationfactors between Fe(II)–Fe(III)–Si gel and aqueous Si,D30Sisolid–AAS, of �3.23 ± 0.33 (2r) ‰ and �3.23 ± 0.16(2r) ‰, respectively (Fig. 6). Importantly, the fractionationbetween Fe–Si gel and aqueous Si for Fe(II)aq-bearingexperiments that involved changes in the stoichiometry ofthe gel is �1‰ larger than that obtained from the Fe(II)aq-free experiments that maintained the same stoichiometry ofFe(III)–Si gel.

5. DISCUSSION

In following sections, we first discuss the Si solubility ofFe–Si gel and its implications for estimation of seawater Siconcentrations in the Precambrian oceans, and then discussSi isotope exchange kinetics observed in our experiments,followed by a comparison with Fe isotope exchange kineticsobserved in a companion study (Wu et al., 2012). In partic-ular, the role of Fe(II) in controlling Si isotope exchangekinetics and fractionation is highlighted. The discussion isthen followed by a comparison between the equilibriumSi fractionation factors for the Fe-bearing Si gels deter-mined in this study and those estimated for pure Si gel inthe literature, and concluded with the implications of thiscomparison for interpretation of Si isotope records in thePrecambrian BIFs and cherts.

5.1. Silica concentrations in the Precambrian oceans

Estimation of Si concentrations in the Precambrianoceans is required to understand the Precambrian silica

Fig. 6. Calculation of the 30Si/28Si fractionation factor between Fe(III)–Si gel and aqueous Si in AAS (D30Sisolid–aqueous), determined byextrapolation to 100% isotope exchange, for Fe(II)aq-free experiments (A and B) and Fe(II)aq-bearing experiments (C and D). (A) Results ofExperiment 2a. The fraction of isotopic exchange (F) is calculated based on the solid phase due to small amounts of dissolution of Fe(III)–Sigel. (B) Results of Experiment 3a. F is calculated based on the aqueous phase due to small extents of precipitation of aqueous Si. (C) Resultsof Experiment 2b. F is calculated based on the solid phase due to minor dissolution of Fe–Si gel. (D) Results of Experiment 3b. F is calculatedbased on the aqueous phase due to small amounts of precipitation of aqueous Si. Note that interaction between Fe(II)aq and the Fe(III)–Si gelin Experiments 2b and 3b produced a change in stoichiometry from Fe(III)Si2(OH)11, to Fe(II)0.1Fe(III)Si2(OH)11.2. In all figures, grayhorizontal dashed lines represent the d30Si value (�0‰) of initial Fe(III)–Si gel and AAS used for the experiments, and black solid linesindicate the linear regression lines, and black dashed lines indicate the uncertainty of regression resulted from analytical uncertainty (95%significance level).


Amorphous Si in H O (Gunnarsson and Arnórsson, 2000)2

Fe-Si gel in Fe(II) Wu et al., 2012)aq-bearing AAS (Fe-Si gel in Fe(II) -free AAS (This study)aq

Fe-Si gel in Fe(II) -bearing AAS (This study)aq

Fe-Si gel in Fe(II) solution (Wu et al., 2011)

Amorphous Si in AAS (Modeled using PHREEQC)

Fig. 7. Solubility of Fe–Si gel as a function of Fe/Si ratio of gel atroom temperature (�23 �C). Solubility of amorphous Si (Fe/Si = 0) in H2O is based on Gunnarsson and Arnorsson (2000).Solubility of amorphous Si in AAS is modeled using thePHREEQC software and Si equilibrium constants at ionic strengthsimilar to seawater (Felmy et al., 2001). Solubility of Fe–Si gel witha Fe/Si ratio of �0.3 was inferred from aqueous Si concentrationmeasured at the end of a 28-day experiment that reacted Fe–Si gelwith AAS containing �1.6 mM Fe(II)aq (Wu et al., 2012).Solubility of Fe–Si gel with a Fe/Si ratio of �0.5 was inferredfrom equilibrium aqueous Si concentrations measured at the end ofExperiments 2 and 3 in this study. Solubility of Fe–Si gel with a Fe/Si ratio of 1 was inferred from aqueous Si concentration measuredat the end of a 15-day experiment that reacted Fe–Si gel with Fe(II)aq solution (Wu et al., 2011). Note that our results suggest aslightly lower Si solubility in the presence of Fe(II)aq, whichpresumably reflects the different stoichiometry of the solid, wherethe Fe(II)aq–free experiments used Si gel with a formula of Fe(III)Si2(OH)11, and the Fe(II)aq-bearing experiments produced a changein Si gel stoichiometry to Fe(II)0.1Fe(III)Si2(OH)11.2.


cycle. There is general consensus that solubility of differentSi-bearing phases played an essential role in controlling sea-water silica concentrations before the appearance of marinesilica-secreting organisms (Siever, 1992), and it has beenproposed that silica concentrations in seawater may havereached saturation of amorphous silica during the Precam-brian (Maliva et al., 2005; Konhauser et al., 2007). Recentstudies, however, have highlighted the importance of Fe–Sigels in the Precambrian Si cycle (Fischer and Knoll, 2009;Rasmussen et al., 2015), and this potentially changes esti-mates of Precambrian seawater silica contents.

Our experiments, together with previous experimentalstudies using Fe(III)–Si gel (Wu et al., 2011, 2012), demon-strate that Si solubility for Fe(III)–Si gels varies between1.5 and 0.5 mM at room temperature as a function of Fe/Siratio of gel (Fig. 7), significantly lower than that of pureamorphous silica (1.86 mM) in water (Gunnarsson andArnorsson, 2000). The difference in the solubility betweenFe–Si gel and amorphous silica cannot be explained by a salteffect due to the presence of an AAS matrix, because ouraqueous modeling results, using the PHREEQC software(Parkhurst and Appelo, 2013) and Si equilibrium constantsat ionic strength similar to seawater (Felmy et al., 2001), sug-gest that the presence of a AAS matrix with or without Fe(II)aq would lead to a slightly lower solubility of 1.54 mMSi for amorphous silica, but not as low as that observed forFe(III)–Si gel, indicating that Fe(III)–Si gel has an intrinsi-cally lower solubility than amorphous silica. This is consis-tent with a recent study that also observed lower Sisolubility of �0.5–1.2 mM, depending on aqueous pH, forFe(II)–Si precipitates in simulated Precambrian seawater(Tosca et al., 2015). Ferrous iron in Precambrian seawatermay have been oxidized by anoxygenic photosynthesis(e.g., Walker, 1987; Kappler et al., 2005), or by photo-oxidation (e.g., Braterman et al., 1983), or by free O2 (e.g.,Morris, 1993), leading to formation of Fe(III)–Si gels. Also,fluctuations in seawater Fe and Si concentrations associatedwith secular changes in the Fe and Si cycles in Precambrianoceans may have led to formation of Fe–Si gels with differentFe/Si ratios. Because the solubility of Fe(III)–Si gel varieswith the Fe/Si ratio (Fig. 7), any temporal changes in aque-ous Fe contents in the Precambrian oceans, as might occurdue to changes in oxygen contents, may have had an impor-tant impact on Si concentrations in Precambrian seawater.

5.2. Silica speciation in AAS and Fe–Si gel

An understanding of mechanisms for the observed Siisotope exchange kinetics and fractionations requiresknowledge on Si species involved during the isotopicexchange process. Aqueous modeling, using thePHREEQC software (Parkhurst and Appelo, 2013), indi-cates that uncharged monomeric silicic acid (H4SiO4) isthe dominant species (>97%) in AAS between pH 7 and 8at 23 �C. Dissociated H4SiO4 (e.g., H3SiO4

�) species areminor because of the high pKa value of H4SiO4 (�9.5)(Exley and Sjoberg, 2014). Monomeric silicic acid can beexpressed as Si(OH)4 to reflect its tetrahedral structure withthe Si atom at the center and hydroxyl functional groups atfour corners. Polymeric Si species are expected to be minor

in AAS, because monomeric silicic acid can remain stablefor long periods at low concentrations (<�2 mM) at neutralpH and room temperature (Iler, 1979). This is supported byconstant aqueous Si concentrations in control reactors thatonly contained AAS (with or without Fe(II)aq) but no Fe(III)–Si gel over the course of the experiments, because Sipolymerization or complexation would result in formationof Si species that cannot be detected by the heteropoly bluemethod used in our study (Appendix Table A.1).

One exception is Experiment 1, where aqueous Si con-centrations in control AAS decreased by �20% at the endof the 126-day experiment (Appendix Table A.1), and thisis interpreted to be due to polymerization of Si inducedby higher initial aqueous Si concentration (2.3 mM) for thisexperiment. An abrupt �20% decrease in aqueous Si con-centration was observed in the control reactor at the endof Experiment 3b (Appendix Table A.1). This decrease,however, may only reflect a perturbation in this particularcontrol reactor or an analytical artifact, because it was


not accompanied by a similar decrease in aqueous Si con-centration, or an anomalous change in Si isotope composi-tion, in the parallel sample reactor at the same time (Fig. 2).

Silanol bonds (�Si–OH) are expected to exist at the sur-face of Fe–Si gel, but Si atoms should be also present inFe–O–Si and Si–O–Si bonds in the Si–Fe network of bulkFe–Si gel. Given high initial aqueous Si (100 mM) and Fe(50 mM) concentrations used for Fe–Si gel synthesis, theproduction of Fe(III)–Si gel must involve both Si polymer-ization and Fe hydrolysis processes. Despite the tendency ofmaximizing Si–O–Si bonds during Si polymerization(Belton et al., 2012), Si has a strong affinity to Fe. Hydrol-ysis of Fe in the presence of Si tends to form Si–O–Febonds, which has been proven to significantly inhibit thegrowth of crystalline Fe oxides or transformation of lesscrystalline Fe phases to well-crystallized Fe minerals inboth natural and controlled laboratory conditions (e.g.,Schwertmann and Thalmann, 1976; Carlson andSchwertmann, 1981; Doelsch et al., 2003; Pokrovski et al.,2003; Chatellier et al., 2004; Jones et al., 2009). Using asuite of spectroscopic techniques, previous studies havedemonstrated the presence of both Si–O–Si and Si–O–Febonds in amorphous Fe–Si gels that were synthesized bya similar approach involving Fe hydrolysis in the presenceof Si in solution under a range of Si/Fe ratios (0.25–4)and pH (3 and 10) at room temperature (Doelsch et al.,2000, 2001, 2003). These studies also indicate that Fe–Sigel is present as a single phase with intimately bonded Feand Si atoms, rather than separate Si and Fe phases, whenproduced under conditions similar to those of our synthesisprotocol (e.g., a Si/Fe ratio of 2, pH > 3). This agrees wellwith our previous TEM analysis that showed a uniform dis-tribution of Fe and Si atoms at the nm-scale without Fe- orSi-enriched domains in Fe–Si gels produced using the sameprotocol as used in the current study (Wu et al., 2012).However, an understanding of the exact ways in which Sitetrahedra are linked to Fe remains limited, because thespectroscopic approaches that have been commonly usedto detect the locations of Fe atoms in amorphous materials,such as X-ray absorption fine structure spectroscopy, arenot able to precisely detect the locations of Si atoms inthe vicinity of Fe atoms (e.g., Doelsch et al., 2000;Pokrovski et al., 2003). It is likely that silicon tetrahedramay be linked to Fe in different ways in Fe–Si gels, giventheir amorphous nature.

5.3. Kinetics of Si isotope exchange and the catalytic effect of

Fe(II)aq

The experiments in this study reached �70–90% Si iso-tope exchange after periods of �50–120 days (AppendixTable A.1), in marked contrast to the assumed low Si iso-tope exchange rate between silicates and Si-bearing fluids(Andre et al., 2006). The fast exchange of Si isotopes inour experiments is most likely caused by the amorphousnature of Fe–Si gel relative to crystalline silicate minerals.Isotopic exchange rates may be described by a general rateequation (Johnson et al., 2002):

�dð1� F Þdt

¼ knð1� F Þn ð15Þ

where F is the fraction of isotopic exchange; kn is the rateconstant; t is reaction time, and n represents the order ofthe reaction. A simple first- (n = 1) or second- (n = 2) orderrate law, derived from Eq. (15), can adequately account formany isotopic exchange reactions (e.g., Criss et al., 1987;Johnson et al., 2002; Li et al., 2011; Wang et al., 2015),but appears to be inadequate in fitting our experimentalresults (Supplementary Information, Fig. S2). This maybe explained by exchange of aqueous H4SiO4 with differentSi bonds (�Si–OH, Si–O–Si and Si–O–Fe) in the Fe–Si gel,each of which may produce different exchange rates.

Silicon isotope exchange was more extensive in the Fe(II)aq-bearing experiments (2b and 3b) relative to the Fe(II)aq-free experiments (1, 2a and 3a). This is illustratedthrough a comparison of the fraction of exchange (F) basedon fitting F data and time (t) to an empirical power relationin a form of

F ¼ a� tb ð16Þwhere a and b are two constants obtained by curve fit. Thecomparison clearly demonstrates a consistently higher frac-tion of Si isotope exchange in Fe(II)aq-bearing Experiments2b and 3b than that of their counterpart Fe(II)aq-freeExperiments 2a and 3a, respectively, at any given time(Fig. 8).

Our experiments document the first evidence for Fe(II)-catalyzed Si isotope exchange. Fe(II)-catalyzed isotopeexchange has been previously observed for Fe and O iso-topes during interaction between aqueous Fe(II) solutionand Fe minerals (Frierdich et al., 2015). Poulson et al.(2005) documented limited Fe isotope exchange betweenFe(III)aq and nm-size ferrihydrite, where exchange was lim-ited to surface Fe atoms. In contrast, using the same Fe(III)–Si gel as in this study, Wu et al. (2011, 2012) docu-mented extensive Fe isotope exchange between Fe(II)aqand Fe(III)–Si gel.

The catalytic effect of Fe(II)aq on the extent of Si isotopeexchange can be explained by electron-transfer atom-exchange (ETAE) reactions between Fe(II)aq and Fe(III)(hydro-)oxides. Ample evidence has demonstrated that elec-trons are transferred from Fe(II) to Fe(III) substrates whenFe(II)aq is adsorbed onto the Fe(III) (hydro-)oxide surfaces(e.g., Jeon et al., 2003; Williams and Scherer, 2004; Yaninaand Rosso, 2008; Rosso et al., 2010), which leads to reduc-tive release of Fe from the solid, thereby facilitating Featom exchange (e.g., Crosby et al., 2005; Handler et al.,2009, 2014; Gorski et al., 2012). Fe(II)–Fe(III) ETAE reac-tions should have occurred in our Experiment 2b and 3bduring net sorption of Fe(II)aq onto Fe(III)–Si gel. We sug-gest that Fe(II)–Fe(III) ETAE reactions may cause break-age of Si–O–Fe bonds, and formation of new Si–O–Febonds, in Fe(III)–Si gel, and this process is envisioned tobe the mechanism behind the larger extent of Si isotopeexchange in the Fe(II)aq-bearing experiments.

5.4. A comparison of Si and Fe isotope exchange kinetics

Wu et al. (2012) conducted Fe isotope exchange experi-ments using a similar Fe(III)–Si gel and AAS that con-tained Fe(II)aq, offering an opportunity to compare Fe

(A) (B)

Fig. 8. (A) Change of fraction of Si isotope exchange (F) with time for Experiments 2a, 2b, 3a and 3b, together with curve fits by a power lawequation F ¼ a� tb. Curve fits for all experiments produced R2 > 0.98. (B) Comparison of fraction of Si isotope exchange for Fe(II)aq–bearingexperiments (2b and 3b) and Fe(II)aq–free experiments (2a and 3a), as fitted by the power law. That the fraction of exchange for comparison ofExperiment 2a and 2b, and 3a and 3b, consistently lie at higher rates of exchange for the Fe(II)aq–bearing experiments (plots above the 1:1line), indicates that Fe(II) had a catalytic effect on Si isotope exchange, due to electron transfer reactions between Fe(II)aq and Fe(III) in theFe–Si gel.


and Si isotope exchange. An identical preparation methodfor the Fe(III)–Si gel was used in the current study and thatof Wu et al. (2012), and an identical Fe:Si ratio of 1:2 wasobtained for Fe(III)–Si gel used in both studies, althoughthere are likely subtle differences in the gels between thetwo studies. As described above, data for the fraction ofFe isotope exchange with time from Wu et al. (2012) werefit by a power law (Eq. (16)), allowing comparison of theextent of exchange in both studies. The results show thatthe extent of isotope exchange for Si is subtly higher thanthat for Fe in the AAS–Fe–Si gel system (Fig. 9). Thismight reflect differences in the properties of the Fe(III)–Sigel used in the two experiments, despite identical prepara-tion. If any difference, however, is expected, the higher Fe(II)aq contents in the experiments of Wu et al. (2012) wouldbe expected to produce higher rates of Fe isotope exchange,yet the opposite is observed. In a comparison of Fe isotopeexchange in the presence or absence of aqueous Si, Wuet al. (2011) observed that Fe isotope exchange was inhib-ited in experiments that contained aqueous Si, which wasinterpreted to reflect blockage of surface Fe sites by sorbedSi. This may be an alternative explanation for the slightlyhigher rates of Si isotope exchange, relative to Fe isotopeexchange (Fig. 9).

5.5. Si isotope fractionations factors in low-temperature

precipitates

Because the Fe–Si gel is amorphous, and a system com-prising this amorphous material and AAS is unlikely toreach thermodynamic chemical equilibrium in our experi-ments, a fundamental question is whether Si isotope equi-librium can be achieved. Formation of Fe–Si gels innature is ubiquitous, not only limited to the Precambrianoceans, due to the strong chemical affinity between Fe

and Si in solutions, and the incorporation of Si into Fehydroxides has been demonstrated to strongly inhibit trans-formation of the amorphous phase into a thermodynami-cally stable crystalline phase (e.g., Schwertmann andThalmann, 1976; Carlson and Schwertmann, 1981;Chatellier et al., 2004; Jones et al., 2009). The Fe–Si gelclearly did not undergo any significant phase transforma-tion in our experiments spanning over a period of a coupleof months, and hence we consider it to be a metastablephase. Reaching relatively stable aqueous Si concentrations(except for Experiment 1) in AAS and stable Fe(II)aq con-centrations in AAS in Fe(II)-bearing experiments (Figs. 2and 3) after �10 days is indicative of attainment of a meta-stable chemical equilibrium state in our experiments. It hasbeen long known in high temperature and pressure oxygenisotope exchange experiments that isotope equilibrium canbe achieved in systems that contained metastable assem-blages (e.g., Clayton et al., 1989), which reflects the fact thatisotope exchange kinetics may be faster than chemicalkinetics required for phase transformation. It remainsunclear if there is a change in Si isotope fractionation dur-ing transformation of metastable Fe–Si gels into crystallineFe–Si minerals, and this may be dependent on the pathwaysof phase transformation (e.g., solid state versus dissolution–precipitation conversion). It should be noted, however, thatthe fractionation factors experimentally determined foramorphous materials do not necessarily differ from thosedetermined for more crystalline phases. For example, oxy-gen isotope fractionation factors determined between amor-phous Si gel and water at low temperatures are close tothose extrapolated from quartz–water fractionation factorsdetermined at high temperatures (Kita et al., 1985).

The relatively linear trajectories for the Si isotopeexchange in Experiments 2a, 2b, 3a, and 3b (Figs. 6)strongly suggest that the inferred fractionation extrapolated

(A) (B)

Fig. 9. (A) Change of fraction of Si isotope exchange (F) with time for Fe(II)aq–bearing Experiments 2b and 3b in this study, and change offraction of Fe isotope exchange with time for experiments using the same Fe(III)–Si gel (Wu et al., 2012), together with curve fits by a powerlaw equation, as in Fig. 9 (R2 > 0.98 for all curve fits). ‘‘CP1a” and ‘‘CP1b” refer to two experiments using different batches of Fe(III)–Si gelby Wu et al. (2012), both of which had the same Fe/Si molar ratio of 0.5 as used in the current study. (B) Comparison of Si isotope exchange(this study; Fe(II)aq–bearing Experiments 2b and 3b) with Fe isotope exchange, as fitted by the power law. That all combinations ofcomparison between the studies produce consistently higher extents of exchange for Si isotopes suggests a small, but significant inhibition ofFe isotope exchange due to sorption of Si and blockage of surface Fe sites. See text for discussion.

Fig. 10. Comparison of solid–aqueous Si isotope fractionationfactors (D30Sisolid–aqueous) of the current study for Fe–Si gels withprevious work in pure Si systems (Li et al., 1995; Geilert et al.,2014; Oelze et al., 2015; Roerdink et al., 2015), as well as betweenaqueous Si and Si sorbed by Fe oxides (Delstanche et al., 2009).The Si isotope fractionation factors in the current study areinterpreted to reflect equilibrium isotope fractionations, whereasprevious work in pure Si systems may not have attained isotopicequilibrium. Note that the temperature dependence of the Si gel–aqueous Si experiments of Geilert et al. (2014) is not interpreted toreflect an equilibrium effect but instead precipitation rates andsaturation state.


to 100% exchange reflects equilibrium or near-equilibriumfractionation. Rapid net dissolution of Fe(III)–Si gel atthe beginning of Experiments 2a and 2b drove d30Si valuesof aqueous Si at early time points towards the initial d30Sivalue of Fe(III)–Si gel (�0‰), causing small deviations ofthese first data points from the linear regression definedby all data points (Figs. 6). Overall, the consistent fraction-ation factors between Fe(III)Si2(OH)11 and aqueous Si(D30Sisolid–aqueous = ��2.3‰) obtained for Experiments 2aand 3a, as well as consistent fractionation factors betweenFe(II)0.1Fe(III)Si2(OH)11.2 and aqueous Si (with Fe(II)aq)(D30Sisolid–aqueous = ��3.2‰) for Experiments 2b and 3b,support our interpretation that these reflect equilibrium Siisotope fractionation factors. Our results contrast with pre-cipitation experiments where large changes were measuredfor the instantaneous Si isotope fractionation factorbetween aqueous and solid phases (Li et al., 1995; Oelzeet al., 2015; Roerdink et al., 2015); such changes are analo-gous to the large changes seen in Experiment 1, whichinvolved extensive net precipitation (Fig. 5).

The isotopic fractionations obtained in the Fe–Si gel sys-tems studied here are different from those inferred for pureSi systems based on theoretical calculations and experi-ments. Early studies that calculated equilibrium Si isotopefractionation factors predicted that heavy Si isotopes wouldpreferentially partition in higher polymerized Si phases(Grant, 1954), suggesting that in a pure Si system, thesolid–aqueous Si isotope fractionation factor should bepositive. Recent low-temperature experimental studies havesuggested little Si isotope fractionation between Si gel andaqueous Si (D30Sigel–aqueous � 0‰) under equilibrium condi-tions (Fig. 10; Oelze et al., 2015; Roerdink et al., 2015),although equilibrium was not rigorously established inthese experiments, nor were the isotope exchange kineticsdetermined using an enriched tracer. A near-zero

D30Sigel–aqueous fractionation for pure Si gel is consistentwith the suggestion of Douthitt (1982), based on theobservation of Si isotope compositions of natural quartz,chalcedony, and dissolved Si. In contrast, batch or flow-through experiments observed slightly larger fractionation,D30Sigel–aqueous of ��1‰, between Si gel and aqueous Siduring precipitation from oversaturated solutions


(Fig. 10; Li et al., 1995; Geilert et al., 2014b), although theseexperiments may have not reached isotopic equilibrium.Regardless of the uncertainty in equilibrium versus kineticeffects in experimental studies of pure Si systems, it seemsclear that Si isotope fractionation between Fe-free andFe-bearing Si gel systems are very different (Fig. 10), whichrequires an assessment of the different Si bonding in thesesystems.

The first-order interpretation of the influence of Fe onthe Si isotope fractionation factors can be derived fromfirst-principles calculations of the equilibrium Si isotopefractionation factors, although it is important to note thatdetailed knowledge on Fe–Si linkages in Fe–Si gels is notcurrently possible. Broadly, theoretical calculations on theSi isotope fractionation between different silicates or aque-ous Si species suggest that the magnitude of Si isotope frac-tionation should correlate to mean Si–O distances, wherethe Si isotope fractionation factor becomes larger withincreasing mean Si–O distances (i.e., weaker bonds;Meheut and Schauble, 2014; Dupuis et al., 2015). The pres-ence of cations (M), and variations in their electronegativ-ity, in the Si tetrahedron structure can affect meandistances of Si–O bonds, following an electron donationtheory (Meheut and Schauble, 2014), where cations thathave electronegativity values lower than Si tend to shortenM–Si–O bond. Shortening of the M–Si–O bond is accom-panied by lengthening of the other three Si–O bonds,thereby leading to longer mean Si–O distances for the tetra-hedron. Because Fe has an electronegativity value of 1.83(Pauling scale), which is lower than a value of 1.90 for Si,electron donation theory would predict that overall Si–Obond strength in the presence of Fe in Fe–Si gel is weakerthan Si–O bonds of the aqueous Si tetrahedron. As heavyisotopes tend to be partitioned into a stronger bond duringequilibrium fractionation (Schauble, 2004), 30Si is preferen-tially partitioned in aqueous Si in our experiments. In addi-tion, the presence of Fe would cause weaker Si–O bonds inFe–Si gel as compared to those in pure Si gel, and this mayexplain the larger Si isotope fractionation in our experi-ments that used Fe–Si gel, relative to previous experimentsthat used pure Si gel (Li et al., 1995; Geilert et al., 2014b;Oelze et al., 2015; Roerdink et al., 2015), despite the uncer-tainty about equilibrium in the pure Si systems. The largermagnitude D30Sigel–aqueous fractionation we measured in theFe(II)-bearing experiments may reflect the lower elec-tronegativity of Fe(II) relative to Fe(III), which shouldhave an even stronger effect on weakening Si–O bonds thanFe(III), resulting in weaker Si–O–Fe(II) bonds as comparedto Si–O–Fe(III) bonds.

Finally, it is important to note that our experiments aredistinctly different from the case of sorption of aqueous Sito pure iron oxides/hydroxides as studied by Delstancheet al. (2009). Sorption of Si to Si-free iron oxides/hydrox-ides may only involve attachment of aqueous Si specieson mineral surfaces, and does not involve formation ofSi–O–Fe bonds throughout the mineral. This likely explainsthe larger D30Sigel–aqueous fractionations of�2.3‰ � �3.2‰ observed in our experiments, relative tothat observed during adsorption of aqueous Si to ironoxide/hydroxides, where D30Siadsorbed–initial fractionations

of �1.1‰ � �1.6‰ were measured (Fig. 10; Delstancheet al., 2009).

5.6. Implications for Si isotope records in Precambrian

sedimentary rocks

Silicon isotope variations in Precambrian cherts andBIFs have been interpreted to reflect changes in seawatertemperature (e.g., Robert and Chaussidon, 2006; Marin-Carbonne et al., 2014) or distinct sources of Si, includingcontinental or hydrothermal (e.g., van den Boorn et al.,2007; Steinhoefel et al., 2009; Heck et al., 2011). Theseinterpretations, however, are limited by a lack of knowledgeon Si isotope exchange kinetics and equilibrium fractiona-tion factors in systems analogous to the Precambrianocean. One puzzling feature of the Si isotope record for Pre-cambrian cherts and BIFs is the fact that d30Si values forBIFs tend to be lower than cherts (Fig. 11; e.g., Van denBoorn et al., 2010; Steinhoefel et al., 2010; Heck et al.,2011; Chakrabarti et al., 2012; Marin-Carbonne et al.,2014; Stefurak et al., 2015). Common explanations for thiscontrast invoke the negative sorbed–aqueous Si isotopefractionation factor that is observed for Si sorption to Feoxides/hydroxides (Delstanche et al., 2009), or infer ahigher portion of hydrothermal component in BIFs relativeto cherts (Chakrabarti et al., 2012). Assuming an igneousSi source (d30Si � 0‰) and a fractionation factor(D30Sisorbed–initial) of �1.6‰ (Delstanche et al., 2009), sorp-tion to Fe oxides, however, can at most produce d30Si val-ues as negative as �1.6‰, which cannot explain the verylow d30Si values for BIF cherts down to �4‰ or �5‰.Our work on Fe–Si gels provides a new means for interpret-ing very negative d30Si values in BIFs, as well as the con-trast in d30Si values in Precambrian BIFs and cherts.

The effect of distinct Si isotope fractionation factors forpure Si and Fe–Si systems, as applied to formation of Pre-cambrian cherts and BIFs, can be illustrated in a simplemass-balance model (Fig. 11). Although some workers haveargued that d30Si values in many Precambrian cherts andBIFs may be dominated by kinetic effects (e.g., Roerdinket al., 2015), the relatively fast Si isotope exchange betweenFe–Si gel and aqueous Si demonstrated in our experimentssuggests that attainment of Si isotope equilibrium is likely,at least for Fe-bearing systems. Our model assumes that Siprecipitates in sediments on the seafloor maintain Si isotopeequilibrium with seawater in the overlying water column,and the Si precipitates–seawater system has a mean d30Sivalue of 0‰, similar to that for the bulk earth (Savageet al., 2014); thus the Si isotope compositions of seawaterand Si precipitates are dictated by the equilibrium fraction-ation factor and the relative Si mass balance (Supplemen-tary Information, Fig. S3), which is determined by Simass in the water column relative to that in the underlyingSi-bearing precipitates (Siwater-column/Siprecipitates). Our cal-culation assumed a density of 3 g/cm3, close to that of BIFsand cherts, for a dense rock equivalent for Fe–Si gel (Fe(II)0.1Fe(III)Si2(OH)11.2) or pure Si gel (SiO2), and a thick-ness of 1 mm of this dense rock in Si isotope equilibriumwith overlying seawater. Consequently, the Siwater-column/Siprecipitates ratio can be equivalent to the ratio of

Fig. 11. A simple equilibrium mass-balance model that predicts changes in d30Si values in chert and BIF precipitates as a function of thewater-depth to gel-thickness ratio, Si isotope fractionation factor, and seawater Si concentration. In the model, we assumed a density of 3 g/cm3 for a dense rock equivalent for Fe–Si gel or Si gel, and a thickness of 1 mm of this rock in isotope equilibrium with overlying seawater.The water-depth to gel-thickness ratio used on the x-axis closely approximates to the Siwater-column/Siprecipitates ratio, but allows for illustrationof the impact of changing seawater Si concentrations on d30Si values in Si-bearing precipitates in the figure. Results calculated based on twoinitial d30Si values of 0‰ (red star) and 1‰ (green star) are illustrated. The vertical gray arrow indicates that the initial d30Si value of systemmay vary in response to changes in the balance of continental and hydrothermal Si sources. Two different D30Sigel–aqueous fractionation factorsof 0‰ and �1‰, encompassing estimates from recent experimental studies (e.g., Roerdink et al., 2015; Oelze et al., 2015), are shown for apure Si gel system. The D30Sigel–aqueous fractionation factor of �3.2‰ is from the current study for an Fe(II)aq–bearing Fe–Si gel system. Thelight yellow circle indicates the mean d30Si value measured in the 3.0 Ga Isua BIFs, corresponding to a higher Siwater-column/Siprecipitates ratio,and the darker yellow circle indicates the mean d30Si value measured in 2.5 Ga BIFs from the Hamersley–Transvaal (H–T) basin,corresponding to a lower Siwater-column/Siprecipitates ratio (Heck et al., 2011). For comparison, the distribution of d30Si values for Precambriancherts and BIFs, based on compiled literature data, is shown as Kernel density functions on the right (after Stefurak et al., 2015 and refer toreferences therein for details on the d30Si data compilation). (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)


water-column depth to the thickness of condensedSi-bearing gels (Fig. 11). Variations in actual density orthickness of the rock would proportionately change theSiwater-column/Siprecipitates ratio.

The relations in Fig. 11 demonstrate that the different Siisotope fractionation factors associated with Fe–Si gel andpure Si gel will have a major impact on the d30Si values ofthese precipitates. In the case of Fe–Si gel, we illustrate thecase for a D30Sigel–aqueous fractionation factor of �3.2‰, themaximum fractionation obtained from our study. For pureSi gel, we illustrate two possible D30Sigel–aqueous fractiona-tion factors of 0‰ and �1‰, encompassing estimates fromrecent experimental studies (e.g., Roerdink et al., 2015;Oelze et al., 2015). For comparison, the compiled distribu-tion of d30Si values of Precambrian cherts and BIFs areshown in Fig. 11 (Stefurak et al., 2015), which highlightsthe lower d30Si values for BIFs relative to cherts. Our modelshows that the more negative fractionation factor for Fe–Sigel, relative to Si gel, leads to more negative d30Si values ofFe–Si precipitates, which agrees qualitatively with theobserved contrast in Si isotope compositions in Precam-brian BIFs and cherts.

The initial d30Si value of the system considered in ourmodel may vary, depending on changes in the balancebetween continental and hydrothermal sources of Si inthe Precambrian ocean. Variation in the initial d30Si valueof our model can explain some of the range of d30Si values

observed in Precambrian BIFs and cherts. For example, inaddition to an initial d30Si value of 0‰, we also illustratedresults for an initial d30Si value of +1‰ in Fig. 11, presum-ably reflecting the influence of Si input from continents.The resulting d30Si values for Si gels can explain slightlypositive d30Si values measured in white cherts from 3.5 to3.2 Ga Barberton Greenstone Belt in the South Africa(Stefurak et al., 2015). A change in the Si isotope composi-tion of seawater alone, however, may not be sufficient inexplaining very negative d30Si values for Fe-bearing chertsand BIFs.

Production of BIFs that have highly negative d30Si val-ues, in a simple equilibrium mass balance model, requires arelatively high Siwater-column/Siprecipitates ratio. That the aver-age d30Si values for 3.8 Ga BIFs from Isua are generallylower than those of the 2.5 Ga BIFs from the Hamersley–Transvaal basin (Heck et al., 2011) is consistent with thedifferences in BIF thickness. The Isua BIFs are relativelythin and small deposits, with an Fe footprint of �4 � 1011

mol km�2 (Czaja et al., 2013), and would be expected tohave formed at relatively high Siwater-column/Siprecipitatesratios, which would produce relatively low d30Si values(Fig. 11). The relatively small volume of the Isua BIFs alsocorrelates with modestly positive d56Fe values for oxideminerals, which is consistent with partial oxidation of Fe(II)aq (Czaja et al., 2013), and hence supports potential for-mation of Fe(II)–Fe(III)–Si gels. In contrast, the very large


Hamersley–Transvaal BIFs are estimated to have an Fefootprint that is two orders of magnitude higher than theIsua BIFs (Czaja et al., 2013), and this should be reflectedin lower Siwater-column/Siprecipitates ratios in our model, whichin turn would produce, on average, less negative d30Si val-ues. The less negative d30Si values for the Hamersley–Transvaal BIFs corresponds with the lower average d56Fevalues for oxide minerals that reflect more extensive oxida-tion of Fe(II)aq (Czaja et al., 2013). It is certainly true that avariety of d30Si values in individual BIF layers would beexpected depending upon Si sources and pathways, but inthe broadest context, the average d30Si values for BIFs ofdifferent ages may reflect fundamental mass balance rela-tions that correlate with the size of BIF deposits.

Finally, we note that variations in seawater Si concen-trations can also affect d30Si values of Si-bearing precipi-tates, because it shifts the Si mass-balance between thesolid and solution. For example, a change in seawater Siconcentration from �2 mM, estimated based on amor-phous Si solubility, to �0.4 mM, estimated based on Fe–Si gel solubility for gels that have high Fe/Si ratios, maycause up to >1‰ increase in d30Si values of Fe–Si gel.Depending on the fractionation factor assumed for pureSi gel and aqueous Si, the same decrease in seawater Si con-centrations would cause either no change (Dgel–aqueous = 0-‰) or an up to � 0.5‰ increase (D

gel–aqueous= �1‰), in

d30Si values of Si gel (Fig. 11). The long-term trend ofincreasing d30Si values in cherts was interpreted to reflectsecular seawater cooling (Robert and Chaussidon, 2006),or reduced hydrothermal input (Chakrabarti et al., 2012),but this trend also involves major changes in dissolved Siand Fe abundances. Our data and model suggest that thechanges in d30Si values in cherts and BIFs, in response tosecular changes in seawater Si and Fe concentrations,should be considered when interpreting the long-termd30Si increase in Precambrian cherts and BIFs.

When considering the origin of Si isotope variations inPrecambrian cherts and BIFs, diagenetic reactions areclearly important, although our mass balance model doesnot explicitly include such a process. Marin-Carbonneet al. (2014) proposed a diagenetic model for cherts, whichinvolves two processes: (1) dissolution of opal A and subse-quent precipitation of opal CT from the dissolved opal Afollowing a Rayleigh process; (2) formation of microquartzfrom a Si pool that comprises aqueous Si from dissolutionof opal CT and additional diagenetic fluids. It should benoted, however, that this model does not consider a poten-tially important role of Fe–Si gel, and the associated largerSi isotope fractionation, in the Precambrian oceans.Instead, the model of Marin-Carbonne et al. (2014) focuseson additional sources of dissolved Si with negative d30Sivalues, produced from successive dissolution–precipitationprocesses, to explain very negative d30Si values (��3‰)in cherts. Importantly, this model does not consider Si iso-tope exchange rates between Si-bearing precipitates and flu-ids. Because our experiments indicate that Si exchange ratesare relatively fast, at least for Fe–Si gel, Si-bearing precip-itates may undergo isotopic exchange with Si-containingdiagenetic fluids in sediments. Depending on the d30Si valueof diagenetic fluid and Si mass balance between precipitates

and fluid, the initial Si isotope compositions of Si-bearingprecipitates can be altered towards a new precipitate–fluidequilibrium.

6. CONCLUSIONS

The results of a series of Si isotope exchange experi-ments using the three-isotope method demonstrate that Siisotope exchange kinetics are sufficiently rapid at room tem-perature between amorphous Fe(III)–Si gel and aqueous Sithat equilibrium isotope fractionations can be confidentlydetermined. The experiments were designed to simulateanoxic Archean seawater conditions and were conductedat different degrees of Si saturation, and in the presenceand absence of Fe(II)aq. An important finding of theseexperiments is that Fe(III)–Si gel has a significantly lowersolubility than that of amorphous Si, implying that Precam-brian seawater Si concentrations may have been consider-ably lower than previous estimates based on amorphousSi solubility. About 70–90% Si isotope exchange wasachieved for all experiments after 53–126 days, but experi-ments that used Fe(II)aq-bearing solutions showed moreextensive Si isotope exchange than equivalent experimentsthat used Fe(II)aq-free solutions. This suggests that Fe(II)catalyzes Si isotope exchange through Fe(II)–Fe(III)electron-transfer and atom-exchange reactions that pro-mote breakage of Si bonds in Fe–Si gel.

Possible kinetic isotope effects were evaluated based onthe linearity of changes in the d30Si values of aqueous Siover time. Only in one experiment, which initially containedaqueous Si concentrations that were twice that of solubilityfor Fe–Si gel, were non-linear changes observed in d30Si val-ues that would indicate changes in the instantaneous iso-tope fractionation factor. The majority of experimentswere operated close to equilibrium solubility, and undersuch conditions, no apparent changes in instantaneous Siisotope fractionation were observed, suggesting the absenceof kinetic isotope effects during exchange; in such cases,equilibrium Si isotope fractionation factors could be confi-dently obtained by extrapolation to 100% exchange. Exper-iments that used Fe(II)aq-free Si solution yielded aconsistent 30Si/28Si fractionation factor between Fe(III)–Sigel and aqueous Si (D30Sigel–aqueous) of �2.3‰. In contrast,a larger D30Sigel–aqueous fractionation factor of �3.2‰ wasobtained for experiments that used Fe(II)aq-bearing Si solu-tion; in this case, exposure to Fe(II)aq resulted in net addi-tion of Fe(II) to the solid, which changed the Fe–Si gelstoichiometry from Fe(III)Si2(OH)11 to Fe(II)0.1Fe(III)Si2(OH)11.2. In all experiments, the solid–aqueous Si isotopefractionation factors were significantly larger in magnitudethan previous work on pure Si gel systems, indicating thatFe has a major effect on Si isotope fractionation. This dis-crepancy is interpreted to reflect weaker Si–O bonds in Fe(III)–Si gel as compared to pure Si gel. The larger Si isotopefractionation between Fe(II)–Fe(III)–Si gel and aqueous Siis probably caused by further weakening of Si–O bonds dueto incorporation of Fe(II) into gel structures.

The large Si isotope fractionation factor between Fe–Sigel and aqueous Si may explain the observation that d30Sivalues of Precambrian BIFs are broadly lower than those


in Fe-free cherts. In addition, changing Fe and Si balancesin the Archean and Proterozoic oceans, reflecting changesin Si solubility and aqueous Fe(II) levels, may help explainthe secular trend in Si isotope records in cherts and BIFsduring the Precambrian. Finally, depositional settings canaffect the mass-balance of amorphous Si-bearing precipi-tates and seawater Si, and hence control d30Si values incherts and BIFs. Theses results indicate that Si isotopes inPrecambrian cherts and BIFs are not only affected by Sisources but also by a variety of other parameters, hencethey can be potentially useful for tracing a wide range ofprocesses that controlled the Precambrian Si cycle, and itscoupling to Fe.

ACKNOWLEDGEMENTS

We thank Dr. Rosalind Armytage for sharing her expertise dur-ing the process of setting up protocols for Si isotope measurementat UW-Madison, Prof. Mark Brzezinski for sharing the Si isotopereference materials (Big Batch, Diatomite and Big Batch), Ms. Jan-ice Jones for shipping the reference materials, and Dr. ElizabethPercak-Dennett for advice on Fe–Si gel synthesis. Comments madeby Dr. L. Trower, two anonymous reviewers and AE E. Schaublehave improved the manuscript. This work was supported by theNASA Astrobiology Institute under grant NNA13AA94A, andNational Science 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.2016.05.012.

REFERENCES

Abraham K., Hofmann A., Foley S. F., Cardinal D., Harris C.,Barth M. G. and Andre L. (2011) Coupled silicon-oxygenisotope fractionation traces Archaean silicification. Earth

Planet Sci. Lett. 301, 222–230.Alexander G. B., Heston W. M. and Iler R. K. (1954) The

solubility of amorphous silica in water. J. Phys. Chem. 58, 453–455.

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, 162–173.Basile-Doelsch I. (2006) Si stable isotopes in the Earth’s surface: a

review. J. Geochem. Explor. 88, 252–256.Basile-Doelsch I., Meunier J. D. and Parron C. (2005) Another

continental pool in the terrestrial silicon cycle. Nature 433, 399–402.

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, 241–250.Belton D. J., Deschaume O. and Perry C. C. (2012) An overview of

the fundamentals of the chemistry of silica with relevance tobiosilicification and technological advances. FEBS J. 279,1710–1720.

Braterman P. S., Cairns-Smith A. G. and Sloper R. W. (1983)Photo-oxidation of hydrated Fe2+–significance for banded ironformations. Nature 303, 163–164.

Carlson L. and Schwertmann U. (1981) Natural ferrihydrites insurface deposits from Finland and their association with silica.Geochim. Cosmochim. Acta 45, 421–429.

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.Chatellier X., West M. M., Rose J., Fortin D., Leppard G. G. and

Ferris F. G. (2004) Characterization of iron-oxides formed byoxidation of ferrous ions in the presence of various bacterialspecies and inorganic ligands. Geomicrobiol. J. 21, 99–112.

Clayton R. N., Goldsmith J. R. and Mayeda T. K. (1989) Oxygenisotope fractionation in quartz, albite, anorthite and calcite.Geochim. Cosmochim. Acta 53, 725–733.

Clesceri L. S., Greenberg A. E. and Eaton A. D. (1999) StandardMethods for the Examination of Water and Wastewater, 20th ed.American Public Health Association, USA.

Cole D. R. and Chakraborty S. (2001) Rates and mechanisms ofisotopic exchange. Rev. Mineral. Geochem. 43, 83–223.

Criss R. E., Gregory R. T. and Taylor, Jr, H. P. (1987) Kinetictheory of oxygen isotopic exchange between minerals andwater. Geochim. Cosmochim. Acta 51, 1099–1108.

Crosby H. A., Johnson C. M., Roden E. E. and Beard B. L.(2005) Coupled Fe(II)�Fe(III) electron and atom exchange asa mechanism for Fe isotope fractionation during dissimila-tory iron oxide reduction. Environ. Sci. Technol. 39,6698–6704.

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 couplediron–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.

De La Rocha C. L., Brzezinski M. A., DeNiro M. J. and ShemeshA. (1998) Silicon-isotope composition of diatoms as an indica-tor of past oceanic change. Nature 395, 680–683.

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, 923–934.Delvigne C., Cardinal D., Hofmann A. and Andre L. (2012)

Stratigraphic changes of Ge/Si, REE+Y and silicon isotopes asinsights into the deposition of a Mesoarchaean banded ironformation. Earth Planet Sci. Lett. 355–356, 109–118.

Ding T. P., Ma G. R., Shui M. X., Wan D. F. and Li R. H. (2005)Silicon isotope study on rice plants from the Zhejiang province,China. Chem. Geol. 218, 41–50.

Doelsch E., Rose J., Masion A., Bottero J. Y., Nahon D. andBertsch 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, 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, 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, 121–128.Douthitt C. B. (1982) The geochemistry of the stable isotopes of

silicon. Geochim. Cosmochim. Acta 46, 1449–1458.



http://refhub.elsevier.com/S0016-7037(16)30245-9/h0005






































































































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.

Exley C. and Sjoberg S. (2014) Silicon species in seawater.Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 117, 820–821.

Felmy A. R., Cho H., Rustad J. R. and Mason M. J. (2001) Anaqueous thermodynamic model for polymerized silica species tohigh ionic strength. J. Solution Chem. 30, 509–525.

Fischer W. W. and Knoll A. H. (2009) An iron shuttle fordeepwater silica in Late Archean and early Paleoproterozoiciron formation. Geol. Soc. Am. Bull. 121, 222–235.

Frierdich A. J., Beard B. L., Scherer M. M. and Johnson C. M.(2014a) 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.Frierdich A. J., Beard B. L., Reddy T. R., Scherer M. M. and

Johnson C. M. (2014b) 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., Rosso K. M., Scherer M. M., SpicuzzaM. J., Valley J. W. and Johnson C. M. (2015) Low temperature,non-stoichiometric oxygen-isotope exchange coupled to Fe(II)–goethite interactions. Geochim. Cosmochim. Acta 160, 38–54.

Geilert S., Vroon P. Z. and van Bergen M. J. (2014a) Siliconisotopes and trace elements in chert record early Archean basinevolution. Chem. Geol. 386, 133–142.

Geilert S., Vroon P. Z., Roerdink D. L., Van Cappellen P. and vanBergen M. J. (2014b) Silicon isotope fractionation duringabiotic silica precipitation at low temperatures: inferences fromflow-through experiments. Geochim. Cosmochim. Acta 142, 95–114.

Geilert S., Vroon P. Z., Keller N. S., Gudbrandsson S., StefanssonA. and van Bergen M. J. (2015) Silicon isotope fractionationduring silica precipitation from hot-spring waters: evidencefrom the Geysir geothermal field, Iceland. Geochim. Cos-

mochim. Acta 164, 403–427.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, 95–104.

Georg R. B., Reynolds B. C., West A. J., Burton K. W. andHalliday A. N. (2007) Silicon isotope variations accompanyingbasalt weathering in Iceland. Earth Planet Sci. Lett. 261, 476–490.

Gorski C. A., Handler R. M., Beard B. L., Pasakarnis T., JohnsonC. M. and Scherer M. M. (2012) Fe atom exchange betweenaqueous Fe2+ and magnetite. Environ. Sci. Technol. 46, 12399–12407.

Grant F. S. (1954) The geological significance of variations in theabundances of the isotopes of silicon in rocks. Geochim.

Cosmochim. Acta 5, 225–242.Grotzinger J. P. and Kasting J. F. (1993) New constraints on

Precambrian ocean composition. J. Geol. 101, 235–243.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, 2295–2307.

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

Handler R. M., Frierdich A. J., Johnson C. M., Rosso K. M.,Beard B. L., Wang C., Latta D. E., Neumann A., PasakarnisT., Premaratne W. A. P. J. and Scherer M. M. (2014) Fe(II)-catalyzed recrystallization of goethite revisited. Environ. Sci.

Technol. 48, 11302–11311.

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, 5879–5891.

Iler R. K. (1979) The Chemistry of Silica. Wiley, New York.Jeon B.-H., Dempsey B. A. and Burgos W. D. (2003) Kinetics and

mechanisms for reactions of Fe(II) with iron(III) oxides.Environ. Sci. Technol. 37, 3309–3315.

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.Jones A. M., Collins R. N., Rose J. and Waite T. D. (2009) The

effect of silica and natural organic matter on the Fe(II)-catalysed transformation and reactivity of Fe(III) minerals.Geochim. Cosmochim. Acta 73, 4409–4422.

Jones C., Nomosatryo S., Crowe S. A., Bjerrum C. J. and CanfieldD. E. (2015) Iron oxides, divalent cations, silica, and the earlyearth phosphorus crisis. Geology 43, 135–138.

Kappler A., Pasquero C., Konhauser K. O. and Newman D.K. (2005) Deposition of banded iron formations byanoxygenic phototrophic Fe(II)-oxidizing bacteria. Geology

33, 865–868.Kita I., Taguchi S. and Matsubaya O. (1985) Oxygen isotope

fractionation between amorphous silica and water at 34–93�C.Nature 314, 83–84.

Konhauser K. O., Lalonde S. V., Amskold L. and Holland H. D.(2007) Was there really an Archean phosphate crisis? Science

315, 1234.Li Y., Ding T. and Wan D. (1995) Experimental study of silicon

isotope dynamic fractionation and its application in geology.Chin. J. Geochem. 14, 212–219.

Li W., Beard B. L. and Johnson C. M. (2011) Exchange andfractionation of Mg isotopes between epsomite and saturatedMgSO4 solution. Geochim. Cosmochim. Acta 75, 1814–1828.

Maliva R. G., Knoll A. H. and Simonson B. M. (2005) Secularchange in the Precambrian silica cycle: insights from chertpetrology. Geol. Soc. Am. Bull. 117, 835–845.

Marin-Carbonne J., Robert F. and Chaussidon M. (2014) Thesilicon and oxygen isotope compositions of Precambrian cherts:a record of oceanic paleo-temperatures? Precambr. Res. 247,223–234.

Matsuhisa Y., Goldsmith J. R. and Clayton R. N. (1978)Mechanisms of hydrothermal crystallization of quartz at 250 �C and 15 kbar. Geochim. Cosmochim. Acta 42, 173–182.

Matthews A., Goldsmith J. R. and Clayton R. N. (1983a) Oxygenisotope fractionations involving pyroxenes: the calibration ofmineral-pair geothermometers. Geochim. Cosmochim. Acta 47,631–644.

Matthews A., Goldsmith J. R. and Clayton R. N. (1983b) On themechanisms and kinetics of oxygen isotope exchange in quartzand feldspars at elevated temperatures and pressures. Geol. Soc.Am. Bull. 94, 396–412.

Meheut M. and Schauble E. A. (2014) Silicon isotope fractionationin silicate minerals: insights from first-principles models ofphyllosilicates, albite and pyrope. Geochim. Cosmochim. Acta

134, 137–154.Morris R. C. (1993) Genetic modelling for banded iron-formation

of the Hamersley Group, Pilbara Craton, Western Australia.Precambr. Res. 60, 243–286.

Oelze M., von Blanckenburg F., Hoellen D., Dietzel M. andBouchez J. (2014) Si stable isotope fractionation duringadsorption and the competition between kinetic and equilib-rium isotope fractionation: implications for weathering systems.Chem. Geol. 380, 161–171.












































































































































Oelze M., von Blanckenburg F., Bouchez J., Hoellen D. andDietzel M. (2015) The effect of Al on Si isotope fractionationinvestigated by silica precipitation experiments. Chem. Geol.

397, 94–105.Opfergelt S., Cardinal D., Henriet C., Andre L. and Delvaux B.

(2006) Silicon isotope fractionation between plant parts inbanana: In situ vs. in vitro. J. Geochem. Explor. 88, 224–227.

Parkhurst D. L. and Appelo C. (2013) Description of Input and

Examples for PHREEQC Version 3: A Computer Program for

Speciation, Batch-reaction, One-dimensional Transport, and

Inverse Geochemical Calculations. US Geological Survey,2328-7055.

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, 205–220.

Perry E. C. and Lefticariu L. (2013) Formation and geochemistryof Precambrian cherts. In Treatise on Geochemistry (eds. H.Holland and K. Turekian), second ed, pp. 113–139. Treatise onGeochemistry. Elsevier.

Pokrovski G. S., Schott J., Farges F. and Hazemann J.-L. (2003)Iron (III)-silica interactions in aqueous solution: insights fromX-ray absorption fine structure spectroscopy. Geochim. Cos-

mochim. Acta 67, 3559–3573.Poulson R. L., Johnson C. M. and Beard B. L. (2005) Iron isotope

exchange kinetics at the nanoparticulate ferrihydrite surface.Am. Mineral. 90, 758–763.

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, 303–306.

Reynolds B. C., Aggarwal J., Andre L., Baxter D., Beucher C. andBrzezinski M. A., et al. (2007) An inter-laboratory comparisonof Si isotope reference materials. J. Anal. Atom. Spectrom. 22,561–568.

Robert F. and Chaussidon M. (2006) A palaeotemperature curvefor the Precambrian oceans based on silicon isotopes in cherts.Nature 443, 969–972.

Roerdink D. L., van den Boorn S. H. J. M., Geilert S., Vroon P. Z.and van Bergen M. J. (2015) Experimental constraints onkinetic and equilibrium silicon isotope fractionation during theformation of non-biogenic chert deposits. Chem. Geol. 402, 40–51.

Rosso K. M., Yanina S. V., Gorski C. A., Larese-Casanova P. andScherer M. M. (2010) Connecting observations of hematite (a-Fe2O3) growth catalyzed by Fe(II). Environ. Sci. Technol. 44,61–67.

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

Schauble E. A. (2004) Applying stable isotope fractionation theoryto new systems. Rev. Mineral. Geochem. 55, 65–111.

Schwertmann U. and Thalmann H. (1976) The influence of [Fe(II)],[Si], and pH on the formation of lepidocrocite andferrihydrite during oxidation of aqueous FeC12 solutions. ClayMiner. 11, 189–200.

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

268, 330–338.Shahar A., Ziegler K., Young E. D., Ricolleau A., Schauble E. A.

and Fei Y. (2009) Experimentally determined Si isotopefractionation between silicate and Fe metal and implicationsfor Earth’s core formation. Earth Planet Sci. Lett. 288, 228–234.

Siever R. (1992) The silica cycle in the Precambrian. Geochim.

Cosmochim. Acta 56, 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., Horn I. and von Blanckenburg F. (2009) Micro-scale tracing of Fe and Si isotope signatures in banded ironformation using femtosecond laser ablation. Geochim. Cos-

mochim. Acta 73, 5343–5360.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, 2677–2696.Stookey L. L. (1970) Ferrozine – a new spectrophotometric reagent

for iron. Anal. Chem. 42, 779–781.Tanakaa M. and Takahashib K. (2001) Silicate species in high pH

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

Tosca N. J., Guggenheim S. and Pufahl P. K. (2015) An authigenicorigin for Precambrian greenalite: Implications for iron forma-tion and the chemistry of ancient seawater. Geol. Soc. Am. Bull.

Treguer P. J. and De La Rocha C. L. (2013) The world ocean silicacycle. Ann. Rev. Mar. Sci. 5, 477–501.

van den Boorn S. H. J. M., van Bergen M. J., Nijman W. andVroon P. Z. (2007) Dual role of seawater and hydrothermalfluids in early Archean chert formation: evidence from siliconisotopes. Geology 35, 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 similar to 3.5 Ga cherts: implica-tions for early Archaean marine environments. Geochim.

Cosmochim. Acta 74, 1077–1103.Walker J. C. G. (1987) Was the Archaean biosphere upside down?

Nature 329, 710–712.Wang X., Johnson T. M. and Ellis A. S. (2015) Equilibrium

isotopic fractionation and isotopic exchange kinetics betweenCr(III) and Cr(VI). Geochim. Cosmochim. Acta 153, 72–90.

Williams A. G. B. and Scherer M. M. (2004) Spectroscopicevidence for Fe(II)�Fe(III) electron transfer at the ironoxide�water interface. Environ. Sci. Technol. 38, 4782–4790.

Wu L., Beard B. L., Roden E. E. and Johnson C. M. (2011) Stableiron isotope fractionation between aqueous Fe(II) and hydrousferric oxide. Environ. Sci. Technol. 45, 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.

Yanina S. V. and Rosso K. M. (2008) Linked reactivity at mineral-water interfaces through bulk crystal conduction. Science 320,218–222.

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, 1095–1104.

Young E. D., Manning C. E., Schauble E. A., Shahar A., Macris C.A., Lazar C. and Jordan M. (2015) High-temperature equilib-rium isotope fractionation of non-traditional stable isotopes:experiments, theory, and applications.Chem.Geol. 395, 176–195.

Ziegler K., Chadwick O. A., Brzezinski M. A. and Kelly E. F.(2005) Natural variations of d30Si ratios during progressivebasalt weathering, Hawaiian Islands. Geochim. Cosmochim.

Acta 69, 4597–4610.

Associate editor: Edwin Schauble







































































































































Documents

Abiologic silicon isotope fractionation between aqueous Si ... · Kinetic processes, such as Si precipitation, clearly aﬀect Si isotope fractiona-tion (e.g., Geilert et al., 2015),