Fe2+ catalyzed iron atom exchange and re-crystallization in a tropical soil

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 148 (2015) 191–202

Fe2+ catalyzed iron atom exchange and re-crystallizationin a tropical soil

Viktor Tishchenko a, Christof Meile b, Michelle M. Scherer c, Timothy S. Pasakarnis c,Aaron Thompson a,⇑

a University of Georgia (Crop and Soil Science), United Statesb University of Georgia (Marine Science), United States

c The University of Iowa (Civil and Environmental Engineering), United States

Received 22 April 2014; accepted in revised form 8 September 2014; available online 28 September 2014

Abstract

Aqueous ferrous iron (Fe2+(aq)) is known to transfer electrons and exchange structural positions with solid-phase ferric (FeIII)

atoms in many Fe minerals. However, this process has not been demonstrated in soils or sediments. In a 28-day sterileexperiment, we reacted 57Fe-enriched Fe2+

(aq) (57/54Fe = 5.884 ± 0.003) with a tropical soil (natural abundance57/54Fe = 0.363 ± 0.004) under anoxic conditions and tracked 57/54Fe in the aqueous phase and in sequential 0.5 M and7 M HCl extractions targeting surface-adsorbed and bulk-soil Fe, respectively; we also analyzed the reacted soil with 57FeMossbauer spectroscopy. In 28 days, the aqueous and bulk pools both moved �7% toward the isotopic equilibrium(57/54Fe = 1.33). Using a kinetic model, we calculate final adsorption-corrected 57/54Fe ratios of 5.56 ± 0.05 and0.43 ± 0.03 in the aqueous and bulk pools, respectively. The aqueous and surface/labile Fe initially exchanged atoms rapidly(10–80 mmol kg�1 d�1) decreasing to a near constant rate of 1 mmol kg�1 d�1 that was close to the 0.74 mmol kg�1 d�1

exchange-rate between the surface and bulk pools. Thus, after 28 days we calculate aqueous Fe has exchanged with20.1 mmol kg�1 of bulk Fe atoms (1.9% of total Fe) in addition to the 17.0 mmol kg�1 of surface/labile Fe atoms (1.6% oftotal Fe), which have likely turned over several times during our experiment. Extrapolating these rates, we calculate a hypo-thetical whole-soil turnover time of �3.6 yrs. Furthermore, Mossbauer spectroscopy indicates the soil-incorporated 57Fe labelre-crystallized as short-range-ordered (SRO) FeIII-oxyhydroxides: our model suggests this pool could turnover in less thanseven months via Fe2+-catalyzed recrystallization. Thus, we conclude Fe atom exchange can occur in soils at rates fast enoughto impact ecological processes reliant on Fe minerals, but sufficiently slow that complete Fe mineral turnover is unlikely,except perhaps in permanently anoxic environments.� 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Iron (Fe) minerals strongly affect the behavior of nutri-ents (Peretyazhko and Sposito, 2005) and pollutants(Wielinga et al., 2001; Charlet et al., 2002) in many ecosys-tems, with the extent of this influence depending on mineral

http://dx.doi.org/10.1016/j.gca.2014.09.018

0016-7037/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: [email protected] (A. Thompson).

composition and size. The smallest or most disorderedFe phases, called short-range-ordered (SRO) Fe, are themost reactive phases for a range of processes includingchemical sorption (Stumm, 1987; Deng and Stumm, 1994;Kogel-Knabner et al., 2008), particle adhesion, and reduc-tive dissolution (Sulzberger et al., 1989; Lovley et al.,2004). Quantifying these processes is challenging in partbecause Fe mineral composition is not fixed, but changesin response to environmental conditions that drive


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2014.09.018&domain=pdf

192 V. Tishchenko et al. / Geochimica et Cosmochimica Acta 148 (2015) 191–202

precipitation and reductive dissolution of Fe phases. Fur-thermore, adsorption of Fe2+ ions to FeIII-bearing mineralscan lead to FeIII mineral transformation (Gorski andScherer, 2011) through conversion of an adsorbed FeII intoa FeIII atom (electron transfer), possible electron conduc-tion through the mineral and subsequent dissolution of adifferent atom as Fe2+

(aq). This process—termed Fe atomexchange and recrystallization—exchanges reduced aqueousFe atoms for (originally oxidized) solid Fe atoms and thislikely will affect the behavior of nutrients and pollutants thatassociate with Fe minerals (Frierdich and Catalano, 2012).

Fe atom exchange and recrystallization has been docu-mented for several Fe-(oxyhydr)oxide minerals (Pedersenet al., 2005; Handler et al., 2009; Mikutta et al., 2009), butit does not impact all Fe minerals equally. For instance,within weeks substantial isotopic equilibration can occurbetween aqueous Fe2+

(aq) and goethite (Handler et al., 2009),ferrihydrite and lepidocrocite (Pedersen et al., 2005), andsome with magnetite (Gorski et al., 2012), whereas withhematite, no atom exchange appears to occur even over timescales of months (Pedersen et al., 2005). Furthermore, Featom exchange can be inhibited when Al or other elementssubstitute in the Fe mineral structure or if Si ions or organicmatter adsorb to the mineral surface (Jones et al., 2009;Gorski and Scherer, 2011; Masue-Slowey et al., 2011;Frierdich et al., 2012; Latta et al., 2012). This raises the ques-tion whether Fe atom exchange takes place in natural soilsand sediments, which contain Fe mineral phases that areoften highly substituted by Al and contain a range of inor-ganic ions and organic compounds on their surfaces. Addi-tionally, since substantial Fe atom exchange can proceedwithout any evidence of secondary mineralization in someminerals [e.g., goethite and magnetite (Handler et al., 2009;Gorski et al., 2012)], while causing dramatic recrystallizationin others [e.g., the SRO mineral ferrihydrite (Pedersen et al.,2005; Yang et al., 2010)] the implications of atom exchangeon nutrient or pollutant dynamics may depend on a soil’sspecific mineral composition.

This study is organized around two questions: (1) Doesiron atom exchange occur in soils and if so, how fast, and(2) which soil iron phases are involved in atom exchange?We approached these questions by selecting a soil with ahistory of active Fe redox cycling that might naturally pro-mote Fe atom exchange. We exposed this soil to isotopi-cally labeled Fe2+

(aq) and monitored the migration oflabeled Fe atoms into and out of the soil solid-phase for28 days. To answer our first question, we then interpretedthis observational data using a two-step kinetic model thatisolated Fe2+

(aq) adsorption from Fe atom exchange. Toanswer our second question we compared the 57Fe Moss-bauer spectra of the unreacted and reacted soil to isolatethe signature of accumulated 57Fe atoms from our spike.

2. MATERIALS AND METHODS

2.1. Field site and soil collection

We collected soil for this experiment from the lower val-ley portion of Bisley Research Watershed site of the Luqu-illo Experimental Forest (LEF) in Puerto Rico. These soils

are in the Luquillo series (Typic Eutrudepts) formed fromvolcanic parent material with a native pH of �5(Peretyazhko and Sposito, 2005). They experience periodicO2 fluctuations that generate intermittent anoxic conditions(Silver et al., 1999) that occur on the scale of days to weeks(Heartsill-Scalley et al., 2007; Liptzin et al., 2011; Hallet al., 2013) and their solid phase Fe is composed predom-inately of short-range-ordered Fe phases described inPeretyazhko and Sposito (2005) and expected to participatein Fe atom exchange based on pure mineral experiments.We collected the upper 10 cm of the soil from a floodplainlocation described elsewhere (Peretyazhko and Sposito,2005; Ginn et al., 2014). Soils were placed in polypropyleneziplock bags, transported at field temperature to the Uni-versity of Georgia and air-dried at 30 �C for 12 h prior todry sieving (<2 mm) and homogenizing.

2.2. Soil characterization

The total soil elemental concentrations were measuredby ICP-MS following Li-metaborate fusion (Hossner,1996) (Acme Labs, Vancouver, BC Canada) with totalorganic carbon measured via a Carlo Erba Elemental Ana-lyzer (Ginn et al., 2014). The total Fe content of the soil wasmeasured separately for greater accuracy as follows: Weadded 10 mg of finely ground soil to a Teflon digestionbomb with �3 mL of H2O2 and heated the mixture at80 �C for 30 min. After cooling, 1.25 mL HNO3, 3.75 mLHCl and 1 mL HF were added and the mixture was sealedand heated at 70 �C for 16 h, and then at 200–220 �C for anadditional 2 h. The sample was then evaporated to drynessand reconstituted in 5 mL of HNO3; evaporated again tonear dryness to oxidize all of the Fe to FeIII and analyzedvia a modified ferrozine protocol given in Thompsonet al. (2006) on a Shimadzu-1700 spectrophotometer atk = 562 nm. All acids used were ultrapure (twice distilled)purchased from Seastar, Inc. To identify the major miner-als, X-ray Diffraction (XRD) was conducted on the bulksoil and clay fraction, which was separated by centrifuga-tion, followed by vacuum filtration and filter-clay-transferto glass slides according to the method of Moore and Rey-nolds (1997). X-ray Diffraction (XRD) was carried out on aBruker Advance diffractometer with a Co-Ka source acrossa 2-theta interval of 2–70� at a scan rate of 2�/min (0.02�increments) using a 0.6 mm slit for glass-slide clay mounts.Data analysis was performed using the EVA software pack-age (Bruker). The short-range-ordered Fe, Al and Si min-eral abundance was measured via a citrate-ascorbateextraction (Ginn et al., 2014); organically-associated Fe,Al and Si were measured with using a sodium pyrophos-phate extraction (Sparks, 1996). These extract solutionswere analyzed on an ICP-MS (Perkin Elmer, model ElanDRC II). The Fe mineral composition of the unreacted soil(and reacted soil, see below) was assessed using 57Fe Moss-bauer spectroscopy (MB) collected at 13, 77, 140 and295 K. The relative abundance of each Fe site population(e.g., mineral phase) was extracted from the spectral fittingas a fraction of the total Fe spectral area. Quantifying Fephase abundance in this manner assumes equal Mossbauerrecoilless fractions of all detected phases; this assumption is

V. Tishchenko et al. / Geochimica et Cosmochimica Acta 148 (2015) 191–202 193

expected to be valid at cryogenic temperatures, and also tobe a good approximation at least up to 295 K with dry sam-ples (Lalonde et al., 1998; Rancourt, 1998). Detailed expla-nation of the MB spectroscopy methods and fitting routinesare given in the electronic annex (EA-Section 2).

2.3. Isotope tracer experiments

Approximately 0.75 g of soil was added to a 125 mLflask in an anoxic glovebox (Coylabs, Grasslake, MI) witha 95%:5% N2:H2 atmosphere and mixed with 75 mL ofincubation buffer solution with or without 6.9 lM of HgCl2to create “sterile” and “live” treatment sets to prevent FeII

oxidation and suppress Fe reduction in our natural soil(Ginn et al., 2014). The incubation solution was preparedto contain 25 mM KCl and 25 mM of 2-(N-morpho-lino)ethanesulfonic acid (MES) buffer and �2.5 mM Fe2+

(aq).The 57Fe-enriched Fe2+

(aq) was introduced to the incubationsolution by adding an aliquot of acidified 57FeCl2 stock(113 mM) and an aliquot of reagent-grade FeCl2 with nor-mal isotopic composition (Sigma–Aldrich) to obtain a final57/54Fe ratio of 5.884 ± 0.003, which is sufficiently distinctfrom the native soil 57/54Fe ratio of 0.363 ± 0.004. Afteraddition of the Fe2+

(aq) spike (comprising 26% of the nativesoil Fe mass) the incubation solution pH was adjusted to5.5 with KOH, allowed to equilibrate for 30 min, and fil-tered (0.2 lm) to remove potential Fe(oxyhydr)oxide pre-cipitates resulting from the pH increase. The reaction pHof 5.5 was chosen to approximate the native soil pH (�5),while retaining sufficient MES buffer capacity (pKa 6.1 at25 �C). All solutions were prepared from ultra-pure water(>18 MO-cm) that was deoxygenated by sparging with N2

gas for 2 h and exposed to our glovebox atmosphere for48 h. The soil slurry was then subsampled into 2 mL poly-propylene centrifuge tubes for sacrificial sampling androtated on an end-over-end shaker (7 rpm). Triplicate sam-ples were prepared at each time point for the sterile treat-ment and single sample for each timepoint was preparedfor the “live” control to insure the Fe2+ spike did not altertypical Fe reduction activity. Controls without soil and/orwithout 57Fe spikes were included to monitor the experi-mental integrity.

2.4. Sampling and analysis

We followed an approach similar to prior work withpure Fe oxides (Williams and Scherer, 2004; Larese-Casanova and Scherer, 2007; Cwiertny et al., 2008;Handler et al., 2009; Beard et al., 2010; Gorski andScherer, 2011; Schaefer et al., 2011), reacting the soil withenriched 57Fe2+

(aq) and tracking the movement of 57Fe intothe solid phase. We used a two-step selective extraction pro-cedure to separate any 57FeII that adsorbed to the soil sur-face from the bulk soil Fe. The first extraction step (2 h,0.5 M HCl) desorbs the majority of adsorbed FeII and willalso dissolve a small portion of the surface Fe-oxyhydrox-ide layers (Heron et al., 1994; Kostka and Luther, 1994;Lovley et al., 2004; Shelobolina et al., 2004). The second

extraction step (14 d, 7 M HCl) dissolves all the Fe (oxy-hydr)oxide phases, including any remaining adsorbed FeII,and should leave the aluminosilicate phases largely intact(Kostka and Luther, 1994; Hossner, 1996). We chose theseextractions over organic-acid extractions (i.e., acid-ammo-nium oxalate) because HCl has been shown to not intro-duce any isotopic fractionation of Fe duringsolubilization (Skulan et al., 2002; Wiederhold et al.,2006). At 0.02, 1, 3, 7, 14, 21, and 28 days of reaction sac-rificial samples were centrifuged at 21,169 relative centrifu-gal force (rcf) for 30 min on an Eppendorf 5430 centrifugewith a F45-30-11 rotor. The supernatant was removed,acidified and stored for analysis. To remaining pellet weadded � 0.9 mL of 0.5 M HCl (Optima grade, Fisher Scien-tific), reacted the resulting slurry for 2 h on an end-over-endshaker (7 rpm) then centrifuged the sample at 21,169 rcf for30 min and collected the supernatant for analysis. To theremaining 0.5 M HCl pellet, we added �0.9 mL of 7 MHCl and reacted the resulting slurry for 14 d on an end-over-end shaker before centrifuging and collecting thesupernatant for analysis. Samples were always openedand manipulated within our anoxic glovebox. This proce-dure produced subsamples from the aqueous phase, thelabile Fe layer including any surface adsorbed FeII (0.5 MHCl extraction), and the bulk solid iron oxides plus a minorportion of the Fe in silicate phases susceptible to proton-promoted dissolution (7 M HCl extraction). Because theseextractions were performed sequentially, we adjusted theFeII, total Fe and isotope measurements (see below) inthe 0.5 M HCl and 7 M HCl extracts to account for thesmall amount of entrained Fe remaining in the soil pelletfrom the previous extract solution. Following extractionall samples were analyzed for FeII and total Fe with the fer-rozine method.

2.5. Isotopic analysis

Fe isotopes were measured in all subsamples by Induc-tively Coupled Plasma Mass Spectrometry (ICP-MS, Per-kin Elmer, Elan 9000). The instrument was tuned prior toanalysis on a 10 lg L�1 solution of Ba, Be, Ce, Co, In,Mg, Pb, Rh, U (Standard ELAN & DRC-e, ISO 9001).Operating conditions were as follows: RF power of1300 W; sample uptake rate of 24 rpm; cooling gas flow rateof 13 L min�1; nebulizer gas flow rate between 0.95 and0.99 L min�1; and auxiliary gas flow rate of 0.70 L min�1.65 sweeps per reading, dwell time 50 ms, 20 replicates wereused. All results were referenced to 54Fe with correction ofchromium concentrations by monitoring 52Cr. Mass bias ofthe measured isotope ratio was corrected using externalstandards containing a 57Fe/54Fe isotopic ratio of 0.3625,which were analyzed after every 10 samples to monitor itstemporal variation. The standard statistical error of 57/

54Fe ratio based on calibration curve was �0.023 (r) (Fig-ure EA1), which is sufficiently precise to discern differencesacross our experimental variation with 57/54Fe ratios rang-ing from �0.3 to 5.8. All statistical representations aregiven as one standard deviation of replicate measures in

Fig. 1. Conceptual model partitioning adsorption from the aque-ous to solid phase into mass transfer (uni-directional, superscriptm) and atom exchange components (equal bi-directional, denotedby *). While no net mass transfer occurs during atom exchange, thefluxes of individual isotopes making up atom exchange are notnecessarily in balance.


concise notation with the parenthetical number indicatingthe error in the last digit [e.g., 7.14(12) is equivalent to7.14 ± 0.12].

2.6. Mossbauer spectroscopy of the reacted soils

Scherer and co-workers (Williams and Scherer, 2004;Larese-Casanova and Scherer, 2007; Cwiertny et al., 2008;Handler et al., 2009; Beard et al., 2010; Gorski andScherer, 2011; Schaefer et al., 2011) have described themolecular structure of recently sorbed/exchanged Fe byreacting isotopically pure 56Fe solid phases with 57Fe2+

(aq)

and then examining the solid phases using Mossbauer(MB) spectroscopy. MB spectroscopy only detects 57Featoms, so the resulting 57Fe-MB spectra isolate the struc-tural composition of the sorbed/exchanged atoms, whichoriginated as aqueous Fe2+

(aq). Since our soil contains a2.1% natural abundance of the 57Fe isotope, we cannotavoid a strong MB signal from the native soil Fe atoms.However, the addition of the 57Fe spike will bias thepost-reaction MB signal towards the recently sorbed/exchanged 57Fe atoms relative to the unreacted MB spectraby a degree that can be calculated from simultaneouslymeasured Fe isotope composition data. We collected MBspectra on the sterile soil reacted with 57Fe2+

(aq) for 28 days,after rinsing it three times with an Fe-free KCl-MES buffersolution to remove any Fe2+

(aq). The spectra were collected at20 and 77 K using two different spectral fitting methods toevaluate the Fe site populations and MB spectral character-istics. Detailed explanation of these methods and fittingroutine are given in section two of the electronic annex.

2.7. Kinetic model

We developed a kinetic model to differentiate atomexchange—the balanced, bi-directional movement of atomsbetween the aqueous and solid phase—from net mass trans-fer between the aqueous and solid phases associated withsorption/desorption or dissolution/precipitation reactions(Figure 1). Our conceptual model involves sorption/desorp-tion of Fe2+

(aq) on/from the surface/labile pool. Surface/labileiron can then be incorporated into the bulk iron phase, orbe transported back to the aqueous phase, i.e.aqueous Fe2þ $ surface=labile$ bulkFe. The surface/labile and bulk iron concentrations are operationallydefined by the 0.5 and 7 M HCl extractable fractions,respectively.

Adsorption of aqueous iron onto the surface wasdescribed as

aqueous! surface : iRads ¼ kads�iFeaq

�Sopen ð1Þ

where Feaq is the iron concentration in solution, Sopen is thenumber of open mineral surface adsorption sites, and i

denotes the isotope (i = 54, 57, 56&58). Desorption isdefined as:

surface! aqueous : iRdes ¼ kdes�iFesfc

�ðSopen þ CÞ ð2Þ

where Fesfc is the concentration of iron in the 0.5 M HClextractable surface/labile pool. This representation linksdesorption with the size of the surface pool (kdes * C * iFesfc,

where C is a constant reflecting that there is always someintrinsic desorption, even in the absence of net mass trans-fer), and to capture the rapid desorption coinciding withadsorption in the beginning of the experiment we includeda similar formulation for simplicity (kdes * iFesfc * Sopen).The exchange between surface and bulk pool was modeledsimilarly, with

surface! bulk : iRinc ¼ kinc�iFesfc

�Bincorp ð3Þbulk ! surface : iRrelease ¼ krelease

�iFebulk ð4Þ

where Febulk is the Fe concentration in the 7 M HCl extract-able bulk phase and Bincorp represents the potential amountof uni-directional Fe incorporation into the bulk pool forthis specific experiment.

These processes resulted in the following system of cou-pled ordinary differential equations for 54Fe, 57Fe, and(56+58)Fe, respectively:

dFeaq

dt¼ �Rads þ Rdes ð5Þ

dFesfc

dt¼ Rads � Rdes � Rinc þ Rrelease ð6Þ

dFebulk

dt¼ Rinc � Rrelease ð7Þ

which were implemented in MATLAB and solved for 54Fe,57Fe, and (56+58)Fe in aqueous, surface and bulk pools,expressed in mmol kgsoil

�1 .Reaction parameters were adjusted to match measured

iron concentrations and isotope ratios, where the surfaceand bulk pools were operationally identified by the 0.5 and7 M HCl extractable fractions, respectively. Observationaldata from the experiment include total iron (

PFe) in each

of the three pools and the 57/54Fe isotopic ratio. The molarabundance of 54Fe, 57Fe, and (56+58)Fe was estimated from

il


these measurements by assuming that the 54Fe:56Fe:58Feratio remains at the native abundance throughout the exper-iment so that 54Fe/(56Fe + 58Fe) = 0.063508. Each isotopecan then be expressed from the measured total iron(P

Fe) and the 57/54Fe ratio [e.g., 54Fe =P

Fe/(57/54Fe + 16.746); see electronic annex]. The fit for the 57/54Feratios of aqueous and bulk pools was weighted more stronglyover the 57/54Fe ratios in the surface pool, because the latterwas the smallest and least well defined of the three pools. Thefitted parameter values used here are: kads = 0.064 d�1

(mmol kgsoil� 1)�1; kdes = 0.1125 d�1 (mmol kgsoil

� 1)�1;kinc = 0.027 d�1 (mmol kgsoil

� 1)�1; krelease = 0.00075 d�1;C = 0.08 mmol kgsoil

� 1; Stotal = Sopen + Fesfc = 56 mmol kgso-�1; Btotal = Bincorp + Febulk = 982 mmol kgsoil

� 1. The sensitiv-ity of atom exchange towards changes in the parameters isquantified by the rate of desorption at 28 days (see below)and described further in Section 3 of the Electronic Annex.Qualitatively, Stotal and Btotal are well constrained by themeasured values, while the relative magnitude of rate con-stants (k-values) was adjusted to mimic the temporal evolu-tion of the 57/54Fe ratios.

Atom exchange has previously been estimated by thedegree of approach to equilibrium (A) expressed as:

A ¼ ðf ðtÞ � f initialÞ=ðf equilibrium � f initialÞ ð8Þ

where f(t), fequilibrium and finitial are the 57Fe/54Fe of an Fepool at time t, after complete mixing of the enriched 57FeII

spike and solid phase Fe, and at t = 0, respectively. How-ever, this calculated approach to equilibrium is influencednot only by atom exchange, but also by any net mass trans-fers (sorption and incorporation) between the aqueous andsolid phases. To correct for this, we separate total adsorp-tion (ads) into adsorption associated with mass transfer(adsm) and a component (ads*) that is balanced by desorp-tion (ads* = des, i.e. atom exchange; Figure 1).

Due to the addition of iron to the aqueous solution, thetotal rate of iron adsorption, Rads, exceeds that of desorp-tion, and the atom exchange component Rads

* = Rdes =57Rdes + 54Rdes + 56&58Rdes. The rates of sorption anddesorption for individual isotopes do not necessarily match(i.e., iRdes – iRads

*), leading to a change in isotopic signa-tures, and the change in 57Fe and 54Fe with time due toatom exchange between the aqueous and surface poolscan be written as:

ðd57Fe=dtÞ�aq ¼57Rdes � Rdes=Rads� 57Rads ð9Þ

ðd54Fe=dtÞ�aq ¼ 54Rdes � Rdes=Rads� 54Rads ð10Þ

We use the same approach to quantify atom exchangewith the bulk Fe pool, partitioning the incorporation ofFe atoms to the bulk pool into uni-directional mass transfer(incm) and an atom exchange flux (inc*) that is balanced byrelease from the bulk to the surface/labile pool balanced bydesorption (i.e., inc* = release), such that:

ðd57Fe=dtÞ�bulk ¼ 57Rinc � Rrelease=Rinc� 57Rinc ð11Þ

ðd54Fe=dtÞ�bulk ¼ 54Rinc � Rrelease=Rinc� 54Rinc ð12Þ

Finally, the 57/54Fe ratios associated only with atomexchange over a time interval t can be recalculated asfollows:

f ’ ¼57Fejðt ¼ 0Þ þ

R t0 ðd

57Fedt Þ

�j

dt

54Fejðt ¼ 0Þ þR t

0 ðd54Fedt Þ

�j

dt; ð13Þ

where j denotes the pool, and can be computed for theaqueous and the bulk pools by substituting Eqs.(9)–(12),respectively (see section 3 of the electronic annex for a com-parison of f and f0).

3. RESULTS AND DISCUSSION

3.1. Soil characterization

The first step initiating Fe atom exchange is the adsorp-tion of Fe2+

(aq) to the surface of FeIII-bearing solids such asiron-bearing minerals under redox disequilibrium. Theamount and speciation of solid phase Fe should influencea soil’s susceptibility to Fe atom exchange. The redox-dynamic soils (Silver et al., 1999) from the Bisley Valley siteat the Luquillo Critical Zone Observatory contained1039 ± 40 mmol kg�1 total Fe (Table 1). XRD revealedquartz (SiO2) as the major soil mineral, and a kaolinitegroup phase, a 2:1 layered aluminosilicate (likely chlorite),and goethite present in the soil clay fraction (<2 mm) (Fig-ure EA2; Table 1). Goethite is the only Fe-rich phase iden-tified by XRD, but Fe can substitute for some of the Alatoms in the identified aluminosilicates, or in less crystallinephases with short-range-ordering (SRO) that XRD doesnot resolve well. Irrespective of crystal ordering, the 57FeMossbauer spectra of this soil reveals Fe is distributedacross four broad coordination environments (see EA-sec-tion 2 for full details). The largest coordination environ-ment, containing 80% of the soil Fe, is FeIII nuclei inFeIII-oxyhydroxides, which includes the goethite identifiedby XRD (Figure EA3; Table 1). MB spectroscopy suggests13% and 3% of the soil Fe are present as FeIII and FeII

nuclei, respectively, substituted within aluminosilicatephases (likely kaolinite or chlorite based on MBS parame-ters, see EA-section 2.3.2 for details) or potentially as sur-face adsorbed or organically complexed ions; and 4% ofthe soil Fe is present as FeII in a paramagnetic phase similarto ilmenite (FeIITiO3) (Figure EA4; Table 1).

Ferric oxyhydroxides have been demonstrated to partic-ipate in Fe atom exchange (Gorski and Scherer, 2011). Thismineral group can have substantial variation in crystalordering, from micron-sized, well-crystalized goethite toSRO nano-goethite and ferrihydrite phases that have crys-tal domains less that 5 nm. The quantity of SRO Fe in soilsis commonly quantified by extraction with acid-ammoniumoxalate or citrate-ascorbate (C/A) (Reyes and Torrent,1997) in combination with sodium pyrophosphate fororganically bound iron determination, which for this soilwe found previously to be 144 mmol Fe kg�1 (14% of totalFe) and 53 mmol Fe kg�1 (5.1% of total Fe) (Table 1). Thecrystallinity of FeIII-oxyhydroxide phases can be quantifiedby the temperature at which those phases magneticallyorder (form a sextet) in a Mossbauer spectrum. The mostcrystalline goethite forms a sextet above 295 K(Schwertmann et al., 1985), whereas laboratory-preparedSRO phases such as ferrihydrite or microcrystalline goe-thite require cooling to at least 77 K (Murad and

Table 1Soil chemical characteristics.

Total Elemental Composition XRD Mineral identificationc

mmol kg�1 % total massb Bulk soil Clay Fraction

Si 9781(58)a 59(1)% Quartz (1)kaoliniteAl 2965(6) 15(0)% (2)goethiteFe 1039(40) 8(0)% (3)chloriteC 2125(8) 6(0)%Ti 107(1) 1(0)%

Fe in selective extractionsd Mossbauer Fe assignments

mmol Fe kg�1 % total Fe mmol kg�1 % total Fe

0.5 M HCl 17(4) 2(0)% FeIII-oxyhydroxide

Citrate/Ascorbate 144(6) 14(1)% >295 K Below detection7 M HCl 952(16) 92(2)% 295–140 K 468(10) 45.0(10)%Pyrophosphate 53(6) 5(1)% 140–77 K 98(20) 9.4(19)%

Al and Si in selective extractionse 77 K-13 K 114(19) 11.0(18)%

mmol Al kg�1 mmol Si kg�1 <13 K 156(10) 15.0(10)%

Citrate/Ascorbate 36(4) 6(1) FeIII-silicate/organic 137(4) 13.2(4)%Pyrophosphate 35(4) 2(1) FeII-silicate 27(3) 2.6(3)%

FeII-site#2f 37(4) 3.6(4)%

a Errors are 1 standard deviation (1r) based on triplicate analysis for total elemental composition and selective iron extractions and 1standard error (s.e.) calculated from the Recoil Software for the Mossbauer data. Uncertainty is presented as the error in the last digit inconcise notation (e.g., 5.0(15) = 5.0 ± 1.5).

b Assuming metals in their oxide forms (SiO2, Al2O3, TiO2, and Fe2O3) and C as CH2O.c See Figure EA2.d Note these extractions are separate (not sequential).e These are the same solutions measured for Fe above.f This site is likely illmenite, but we are less confident about this than the other site assignments.


Schwertmann, 1980), and the most disordered SRO phaseswith high degrees of substitution (Masue-Slowey et al.,2011) and very low crystallinity may require cooling below13 K (Murad and Schwertmann, 1983; Murad andCashion, 2004; Thompson et al., 2011). For our soil, wefind none of the FeIII-oxyhydroxides order at 295 K, 54%of the FeIII-oxyhydroxides order by 77 K, and one-fifth ofthe FeIII-oxyhydroxides (or 15% of the total soil Fe) remainunordered at 13 K (Table 1). The ensemble average FeIII-oxyhydroxide in this soil lies between that of nano-goethiteand ferrihydrite (Figure EA5). Since both Al substitution(based on Al in C/A extraction SRO-Fe phases could beAl substituted up to 20%, Table 1) and small particle sizewill lower the Mossbauer ordering temperature—and alterthe rate of Fe dissolution in C/A solutions—we cannot pre-cisely quantify the soil Fe susceptible to atom exchange, butwe suggest the best candidate phases are those orderingbelow 77 K (26% of total Fe, see Table 1) and those Fephases that are extractable by C/A (14% of total Fe, seeTable 1).

3.2. Iron atom exchange in soil

At the beginning of the experiment the first extractionstep (2 h, 0.5 M HCl) removed 17.0 mmol kg�1 or 2% ofthe total soil Fe (Table 1). The second extraction step(7 M HCl) extracted 951.6 mmol kg�1 or 92% of total Fe,12% greater than the size of the Fe-oxyhydroxide pool iden-tified by Mossbauer spectroscopy; this suggests some of thealuminosilicate/organically-bound Fe was also extracted(see Table 1).

The FeII and FeIII concentrations changed systemati-cally in the aqueous and 0.5 M HCl extractable pools ofour triplicate sterile soils over the 28 days of reaction.Adsorption decreased the Fe2+

(aq) concentrations 15 ± 6%within 1 h and 20 ± 4% after 1 d of reaction (Table 2). Mostof this sorbed Fe (82 ± 2%) was recovered as an increase inFe extracted by 0.5 M HCl, but interestingly this increasewas split nearly equally between FeII

(0.5M HCl) and FeIII(0.5M

HCl) (Figure 2 and Table 2). We suspect this reflects rapidelectron transfer reactions at FeIII-mineral surfaceswhereby Fe2+ adsorbed from the aqueous phase transfersan electron to the bulk FeIII-mineral and is oxidized to FeIII

in a labile surface layer that is extractable by 0.5 M HCl(Williams and Scherer, 2004).

Following this initial adsorption step, net mass transferof FeII between aqueous to solid phase was minimal in thesterilized slurries. The mass of Fe extracted was more vari-able in the 7 M HCl (Fe(7M HCl)) pools than in the aqueousand 0.5 M HCl pools, but remained similar within errorthroughout the experiment (Figure 2). The rapid Fe2+

(aq)

sorption in the first hour of the experiment (Figure 2) coin-cides with a decrease in the 57/54Fe(aq) ratio from5.884 ± 0.028 to 5.766 ± 0.012 and a correspondingincrease in the 57/54Fe(0.5M HCl) and 57/54Fe(7M HCl) ratiosfrom their initial natural abundance values of0.363 ± 0.004 to 3.401 ± 0.004 and 0.386 ± 0.001, respec-tively (Figure 3). Part of the increase in 57/54Fe(0.5M HCl)

and 57/54Fe(7M HCl) ratios reflects adsorption of the enriched57Fe2+

(aq) to the solid phase and not atom exchange. How-ever, atom-exchange also occurred during the first day ofexperiment as evidenced by a large decrease in the 57/54Fe

Fig. 2. Total Fe (see Table 2 for the fraction FeII) in the aqueous,surface layer (0.5 M HCl extraction), and bulk (7 M HCl extrac-tion) pools in the atom exchange experiment. Error bars repre-sent ± 1r (n = 3) for the sterile treatments. Only one set of livecontrol samples was used.

Table 2Mass balance and % recovery data for the sterile treatment.a

Treatment Day Aqueous Fe (mmol kg�1 soil) 0.5 M HCl Fe (mmol kg�1 soil) 7 M HCl Fe (mmol kg�1 soil) % Recovery

Total Fe 57Fe/54Fe bf Fe2+ Total Fe 57Fe/54Fe f Fe2+ Total Fe 57Fe/54Fe f Fe2+

Sterile treatment 0 269.9(97) 5.884(28) 0.981(21) 17.0(46) 0.363(4) 0.26(2) 952(16) 0.363(4) b.d. -0.02 232.6(61) 5.766(12) 0.973(39) 55.7(1) 3.401(4) 0.52(6) 937(15) 0.386(1) b.d. 94.7(37)1 218.0(3) 5.675(5) 0.975(8) 59.4(6) 3.970(66) 0.46(2) 977(52) 0.437(3) b.d. 89.0(3)3 213.4(28) 5.671(30) 0.995(5) 58.3(11) 4.135(68) 0.45(0) 917(39) 0.475(17) b.d. 88.6(15)7 206(10) 5.582(82) 0.980(17) 56.0(30) 4.140(99) 0.49(3) 1011(28) 0.498(14) b.d. 85.1(25)14 211.5(97) 5.587(22) 0.975(24) 57.6(46) 4.109(28) 0.65(3) 970(31) 0.504(10) b.d. 90.4(12)21 201.5(63) 5.535(51) 0.956(23) 60.0(48) 4.073(77) 0.64(2) 1000(26) 0.536(34) b.d. 85.9(35)28 212.5(80) 5.530(54) 0.970(22) 61.3(28) 3.883(76) 0.69(3) 956(11) 0.525(34) b.d. 92.2(20)

Standard deviations were calculated from triplicate reactors. All statistical representations are given as one standard deviation of replicatemeasures in concise notation with the parenthetical number indicating the error in the last digit [e.g., 7.14(12) is equivalent to 7.14 ± 0.12].

a Data for the Live Control is contained in the electronic annex (Table EA1).b f Fe2+ is a fraction of Fe2+ relative to total Fe.


ratio of the aqueous phase. This decrease cannot beexplained through isotopic fractionation associated withadsorption alone as that process would not change the57/54Fe(aq) value more than a few per mil (<0.01 in the57/54Fe ratio) (Mikutta et al., 2009).

3.3. Rates of Fe atom exchange in soil

We can isolate changes in 57/54Fe ratios due to atomexchange from the computed concentrations of individualFe isotopes (see section 2.7). The adsorption event is mod-eled well by defining a fixed amount of adsorption sites(Sopen) in the surface/labile pool (Figure 2). The rapid initialincrease in the bulk 57/54Fe(7M HCl) ratios (Figure 3) is notmatched by a corresponding decrease in 57/54Fe(0.5M HCl)

and thus suggests rapid incorporation of labeled 57Fe intobulk pool, which certainly includes SRO Fe phases. Wemodel rapid incorporation by including a fixed quantityof incorporation sites (Bincorp) analogous to the adsorptionsites in the surface/labile pool (Figure 1; Eq. (3)). Weascribe changes in the isotopic ratios over longer time scalesto atom exchange—expressed by balanced desorption/adsorption reactions between the aqueous and surfacepools and by balanced release/incorporation reactionsbetween the surface and bulk pools (see section 2.7). Thisrepresentation leads to an excellent fit of the 57/54Fe(aq)

and 57/54Fe(bulk) ratios, with a mismatch only early in theexperiment in the much smaller and more poorly con-strained surface/labile pool 57/54Fe(surface).

Using our model, we can isolate processes with zero netmass transfer (Eqs.(9)–(13)) and calculate that by the end ofour experiment the aqueous and bulk pools have moved7.1% and 7.2% toward isotopic equilibrium due to atomexchange, respectively (Eq. (13), also see Section 3 of theelectronic annex, Figure EA8). This compares with anunadjusted calculation of the approach to equilibrium valueof 7.8 ± 1.2% and 16.7 ± 3.0% for the aqueous and bulkpools, respectively. The corrected and uncorrected‘approach to equilibrium’ values are equal within errorfor the aqueous pool, but substantially different for the bulkpool. In prior work, the adsorption-bias on the solid phase‘approach to equilibrium’ value has been overcome by

efficient removal of sorbed and surface Fe atoms usingweak-acid extractions in systems where near-complete equi-librium is achieved (Handler et al., 2009) or by reportingthe ‘approach to equilibrium’ for only the aqueous phasein systems with variable extents of exchange (Gorski andScherer, 2011; Gorski et al., 2012; Latta et al., 2012). Thelater approach works well if all Fe isotopes are measuredindividually and separate ‘approach to equilibrium’ calcula-

Fig. 3. 57/54Fe ratios in the aqueous, surface layer and bulk pools, and the corresponding with dashed lines are the model fit. Sorptioncorrected values were not calculated for the 0.5 M HCl extraction. The isotopic ratio upon complete equilibration between solution, surfaceand bulk oxide pools is 57/54Fe � 1.3344.


tions are performed for 57Fe and 54Fe to confirm the frac-tion of 57Fe lost from the aqueous phase is matched by afractional gain of 54Fe from the solid phase. Our kineticmodeling approach provides a means of constraining theextent of atom exchange when only isotope ratios (e.g.,57/54Fe) and total iron can be measured. In addition, it facil-itates description of atom exchange rates.

The rate of atom exchange can be calculated from thekinetic model by summing the movement of all Feisotopes between two pools as a function of reaction time(Figure 4, see Section 2.7). These rates are fastest(10–80 mmol kg�1 d�1) between the aqueous and surface/labile pools in the first day of the experiment—coinciding

with the rapid adsorption event—and decreased to anear constant rate of �1 mmol kg�1 d�1 within five days(Figure 4). Conversely, the exchange rate between thesurface and bulk pool was lower (�0.74 mmol kg�1 d�1)and less variable throughout the experiment. Thus, thesurface/labile layer modulates exchange between aqueousand bulk soil Fe, effectively shielding the bulk Fe frominteraction with the labeled Fe2+

(aq) with the exception ofthe initial incorporation event. This is consistent withprevious observations that atom exchange is most prevalentbetween aqueous Fe2+ and surface atoms (Mikutta et al.,2009), although in our case 0.5 M HCl likely extracts morethan a surface monolayer.

0 5 10 15 20 25 1000 2000

1

10

100

Rat

e of

Fe

atom

Exc

hang

e

mm

ol k

g-1 d

-1

Days

Aqueous Bulk

0 400 800 1200 1600 20001E-3

0.01

0.1

1

10

1333 days

Bulk Fe phase

Frac

tion

of p

ool t

urno

ver

Days

Aqueous Fe

164 days

Aqueous Fe

Bulk Fe phase

Fig. 4. Calculated rate of atom exchange (all Fe isotopes) correctedfor sorption-effects in log scale for the aqueous and bulk Fe poolsas a function of reaction time. Inset shows the modeled fraction ofpool turnover in log scale (moles of Fe exchanged per mole of totalFe in the pool) as a function of reaction time. The dashed lineindicates one complete pool turnover.


Implicit in our calculation of different atom exchangerates for the aqueous and bulk pools is that the fractionof atoms exchanged will be unique to each pool and depen-dent on both pool size and the rate. Thus, the fraction ofaqueous Fe2+ pool undergoing atom exchange will be lar-ger than the fraction of the bulk Fe pool (Figure 4 inset)and both will exchange smaller fractions than the surface/labile Fe pool. Integrating over 28 days we calculate a totalof 65 mmol kg�1 of atom exchange between the aqueousand surface/labile pools and 20.1 mmol kg�1 (1.9% of totalsoil Fe) of atom exchange between the surface and bulkpools. Projecting our simulation beyond the 28 d experi-ment, we calculate the aqueous pool would turn over com-pletely in �164 days, while �3.6 yrs of constant Fe2+

(aq)

exposure would be required for the bulk Fe to completelyturn over (Figure 4 inset). Of course, mineral transforma-tion and other processes would likely alter atom exchangerates over such timescales. Regardless, the turnover timeestimated for our soil is considerably longer than for a puregoethite system, but may be consistent with atom-exchangefor Al-substituted goethite or with goethite with significantSi or organic matter coverage of the mineral surface (Joneset al., 2009; Gorski and Scherer, 2011; Masue-Slowey et al.,2011; Frierdich et al., 2012; Latta et al., 2012). Indeed, Al ispresent in the SRO mineral pool at a ratio of 1:4 Al:Fe,consistent with some Al-substitution of at least the cit-rate/ascorbate extractable Fe phases.

3.4. Fe phases involved in atom exchange

The Bisley soils contain a 2.1% natural abundance of the57Fe isotope or 22 mmol kg�1 soil. We estimate the soilaccumulated an additional 16.4 ± 4 mmol kg�1 soil of57Fe (see Section 4 of the electronic annex) after reactionwith the isotopically labeled 57Fe2+

(aq) and removal of theentrained solution and weakly sorbed Fe (15.1 mmol Fekg�1 soil). Thus, the MB spectra of our 57Fe2+ reacted soil(Figure 6 and Figure EA4 of the electronic annex) reflect aroughly 3:2 mixture of native Fe and 57Fe labeled addition.The MB spectrum at the lowest collection temperature,20 K, shows a �5% increase in FeIII-oxyhydroxides relativeto the unreacted soil Fe phases collected at 13 K (Figure 5and Figure EA6 of the electronic annex) consistent with93% of the 57Fe label accumulating as FeIII-oxyhydroxides.The slight difference in MBS collection temperaturesbetween the reacted (20 K) and unreacted (13 K) samplesis not likely to produce significant differences based onour prior work with these soils, because nearly all FeIII-(oxyhydr)oxide phases in these soils order above 20 K.The potential error would be that by collecting MBS at20 K, instead of 13 K, we would be slightly underestimatingthe abundance of FeIII-oxyhydroxides and slightly overesti-mating the abundance of FeIII-silicate/organic populationrelative to the unreacted sample. Thus, the MB spectra con-firm the aqueous 57Fe2+ underwent electron transfer reac-tions (Williams and Scherer, 2004) with solid-phase FeIII

and re-crystallized overwhelmingly as Fe-oxyhydroxidephases, with potential minor re-crystallization within clayminerals or organic complexes (Figure 5).

Examining the MB spectra at 77 K—where a significantportion of the FeIII-oxyhydroxides are within proximity oftheir Neel temperatures—reveals the 57Fe label accumulatespreferentially in lower crystallinity portions of the FeIII-oxyhydroxide population. The FeIII-oxyhydroxide sextetsin the 57Fe2+-reacted soil exhibit broader line widths (13.9vs. 13.2 T) and lower average hyperfine field strengths(38.5 T vs. 41.6 T) than FeIII-oxyhydroxide sextets in theunreacted soil (Figure 6). This broadening of line widthsand lowering of the average hyperfine field strengths at77 K is consistent with a greater proportion the FeIII-oxy-hydroxide population exhibiting Neel temperatures at orbelow 77 K, rather than above 77 K (i.e., exhibiting mag-netic ordering at lower temperatures, see the electronicannex of Thompson et al., 2011 for more details). A lowerNeel point for FeIII-oxyhydroxide phases is consistent withan increase in phase disorder or decrease in particle size(i.e., for both, indicating a lower crystallinity). However,note that the peak FeIII-oxyhydroxide hyperfine fieldstrength for the 57Fe2+-reacted soil (48.3 T) remained simi-lar to that of the unreacted soil (48.4 T). This suggests themost crystalline FeIII-oxyhydroxide sites were unaffected byreaction with 57Fe2+

(aq) (Figure 6) as the those phases areexpected to exhibit the highest hyperfine field strengths(i.e., most ordered) within the parameter range for FeIII-oxyhydroxides (Thompson et al., 2011).

3.5. Implications of Fe atom exchange in soils

Re-crystallization of the 57Fe label as SRO Fe-oxyhy-droxides is consistent with observations of rapid Fe atomexchange with pure ferrihydrite and nano-scale goethitephases, but no atom exchange for the highly crystallinephase hematite (Pedersen et al., 2005; Handler et al.,2009; Gorski and Scherer, 2011). We estimate37 mmol kg�1 of atom exchange has occurred during our

0%

10%

20%

80%

90%

100%

2.2(5)%3.5(4)%2.2(4)%

2.6(3)%

10.4(7)%

13.2(4)%

85(3)%

57F

e at

om

dis

trib

uti

on

FeII-ilmenite FeII-silicate FeIII-silicate/organic FeIII-(oxyhydr)oxides

80(2)%

Unreacted 57Fe-reacted

Fig. 5. Comparison of the spectral area of Fe solid phasecomponents from Mossbauer spectra at 13 K for the unreactedsoil and 20 K for the 57Fe2+-reacted soil following a two washes toremove entrained and surface adsorbed 57Fe2+. Nearly all FeIII-oxyhydroxide phases are magnetically ordered below 20 K (seeSection 2 of the electronic annex for further detail). Because thereacted soil was measured at 20 K, instead of 13 K, the FeIII-oxyhydroxide population of the reacted soil may be slightlyunderestimated and FeIII-silicate/organic population may beslightly overestimated relative to the unreacted soil (see text formore details). Standard errors on the last digit are given in conciseform.

Unreacted

57Fe-reacted

A

B

Co

un

t In

ten

sity

(ar

bit

rary

un

its)

-10 -8 -6 -4 -2 0 2 4 6 8 10

Velocity (mm s-1)

Fig. 6. Mossbauer spectra at 77 K for the unreacted soil (A) and the 57Fe2

adsorbed 57Fe2+ (B). In each spectrum, the black solid line is the total cacomponents and assignments are: (1) Q-FeIII-1, the deep central doublet (FeII-1, the narrow ferrous doublet corresponding to ilmenite (red line); (3(green line); (4) HFD-OxHy, the dominant sextet (purple line) correspohyperfine field strengths for the HFD-OxHy sextets (i.e., FeIII-oxyhydroxline) soils at 77 K. Component designations (i.e., Q-FeIII-1) are describereferences to colour in this figure legend, the reader is referred to the we


experiment by summing the atom exchange between thesurface/labile and bulk pools (20.1 mmol kg�1; 1.9% oftotal Fe) and the original surface/labile pool(17.0 mmol kg�1; 1.6% of total Fe), which likely turnedover several times during our experiment. This is substan-tially smaller than the total amount of Fe extractable by cit-rate-ascorbate (144 ± 6 mmol Fe kg�1 soil; n = 3), whichtargets short-range-ordered (SRO) FeIII-oxyhydroxides.According to our model, completely exchanging this poolwould require �7 months of exposure to 2.5 mM Fe2+

(aq).Soils from the Luquillo CZO valley site do not remain

anoxic beyond a month (Liptzin et al., 2011) and field soilporewater Fe2+

(aq) concentrations would rarely reach the mil-limolar concentrations used in this experiment (Hall et al.,2013). However, even intermittent anoxia, generating sub-millimolar Fe2+

(aq) concentrations, could result in turnoverof the SRO Fe pool within a decade. On the century or mil-lennial timescales of soil genesis and development, atomexchange certainly plays a role in shaping iron mineralcomposition and their ecological roles. Atom exchangeand re-crystallization is known to expel trace-metals fromco-precipitated iron phases (Pedersen et al., 2005;Frierdich et al., 2012) and would likely impact the availabil-ity of nutrients to plants and microbes over the course ofpedogenesis.

30 35 40 45 50 55 60

0.00

0.05

0.10

0.15

0.20

0.25

Pro

bab

ility

Dis

trib

uti

on

Absolute Value of BHF (T)

Unreacted57Fe-reacted

C

+-reacted soil following two washes to remove entrained and surfacelculated fit, through the discrete data points. The resolved spectralblue line) corresponding to FeIII in aluminosilicate minerals; (2) Q-) Q-FeII-2 the wider ferrous doublet corresponding to FeII in claysnding to FeIII-oxyhydroxides. (C) The probability distribution ofides) in the unreacted (black line) and 57Fe2+-reacted (Blue, dashedd in more detail in the electronic annex. (For interpretation of theb version of this article.)


Anoxic conditions and the generation of Fe2+(aq)—prereq-

uisites of atom exchange—facilitate iron-reducing organ-isms presence in natural systems. Although un-replicated,our non-sterile control soil (prepared without HgCl2) exhib-ited an increase in Fe2+

(aq) and FeII(0.5M HCl) after 14 and 7 d of

reaction, respectively that likely reflects microbially-cata-lyzed iron reduction (Figure 2, Tables 2 and EA2 of the elec-tronic annex). Coincident with this increase in FeII, the57/54Fe(aq) ratio begins to decrease after 7 d (Figure 3), presum-ably reflecting reductive dissolution of FeIII-oxyhydroxideswith a lower 57/54Fe ratio than the initial Fe2+(aq) spikesolution. More investigations of microbial Fe reductionin the presence of isotopically labeled Fe are clearlywarranted.

4. CONCLUSION

This study documents atom exchange between aqueousFe2+ and solid phase Fe in a natural soil. The most rapidatom exchange took place concurrently with Fe2+ adsorp-tion, and continued slowly thereafter. Mossbauer analysissuggests Fe2+-facilitated atom exchange occurred predomi-nately with the short-range-ordered (SRO) portion ofsolid-phase Fe, which typically has high surface areas andhence high affinity for nutrient adsorption (e.g., P) and highsusceptibility to reductive dissolution. Soils and sedimentsexposed to Fe reducing conditions are thus likely to undergoatom exchange and exhibit Fe-atom fluidity within at leastthe SRO Fe phases. While it is not clear if all soil Fe will par-ticipate in these reactions, our data suggests atom exchangereactions could be substantial over pedogenic timescales,and might need to be considered when interpreting paleotra-cers associated with Fe minerals. A key need emerging fromthis work is to quantify how much Fe atom exchange altersnutrient, carbon and pollutant mobility in naturalenvironments.

5. CONFLICT OF INTEREST

The authors declare no competing financial interest.

ACKNOWLEDGEMENTS

We thank Whendee Silver for providing access to the Bisley sitefor soil collection, and Drew Latta for Mossbauer measurement ofthe unreacted soil. We also thank Jared Wilmoth for XRD analy-sis, Laiming Huang for total digestion of the soil and Gene Weeksand Sayed Hasan for their assistance in setting up the ICP-MS forFe isotope ratio analysis. We also thank three anonymous review-ers and associate editor Jeff Catalano for excellent comments thatimproved this manuscript. This work was funded by USDA -NIFA Soil Processes Program AFRI- NIFA Grant no. 2009-65107-05830 to AT, CM and MS and the NSF Critical ZoneObservatory Program grant no. EAR-1331841 to AT.

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.2014.09.018.

REFERENCES

Beard B. L., Handler R. M., Scherer M. M., Wu L. 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.Charlet L., Bosbach D. and Peretyashko T. (2002) Natural

attenuation of TCE, As, Hg linked to the heterogeneousoxidation of Fe(II): an AFM study. Chem. Geol. 190, 303–319.

Cwiertny D. M., Handler R. M., Schaefer M. V., Grassian V. H.and Scherer M. M. (2008) Interpreting nanoscale size-effects inaggregated Fe-oxide suspensions: reaction of Fe(II) withgoethite. Geochim. Cosmochim. Acta 72, 1365–1380.

Deng Y. and Stumm W. (1994) Reactivity of aquatic iron(III)oxyhydroxides–implications for redox cycling of iron in naturalwaters. Appl. Geochem. 9, 23–36.

Frierdich A. J. and Catalano J. G. (2012) Controls on Fe(II)-activated trace element release from goethite and hematite.Environ. Sci. Technol 46, 1519–1526.

Frierdich A. J., Scherer M. M., Bachman J. E., Engelhard M. H.,Rapponotti B. W. and Catalano J. G. (2012) Inhibition of traceelement release during Fe(II)-activated recrystallization of Al-,Cr-, and Sn-substituted goethite and hematite. Environ. Sci.

Technol. 46, 10031–10039.Ginn B. R., Habteselassie M. Y., Meile C. and Thompson A.

(2014) Effects of sample storage on microbial Fe-reduction intropical rainforest soils. Soil Biol. Biochem. 68, 44–51.

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, 2399–12407.

Gorski C. A. and Scherer M. M. (2011) Fe2+ Sorption at the FeOxide-Water Interface: A Revised Conceptual Framework. InAquatic Redox Chemistry, American Chemical Society, pp.315–343.

Hall S., McDowell W. and Silver W. (2013) When wet gets wetter:decoupling of moisture, redox biogeochemistry, and greenhousegas fluxes in a humid tropical forest soil. Ecosystems 16, 576–589.

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.

Heartsill-Scalley T., Scatena F. N., Estrada C., McDowell W. H.and Lugo A. E. (2007) Disturbance and long-term patterns ofrainfall and throughfall nutrient fluxes in a subtropical wetforest in Puerto Rico. J. Hydrol. 333, 472–485.

Heron G., Crouzet C., Bourg A. C. M. and Christensen T. H.(1994) Speciation of Fe(II) and Fe(III) in contaminated aquifersediments using chemical-extraction techniques. Environ. Sci.

Technol. 28, 1698–1705.Hossner L. R. (1996) Dissolution for total elemental analysis. In

Methods of Soil Analysis: Part 3-Chemical Methods (ed. D. L.Sparks). Soil Science Society of America, Madison, WI, pp. 49–64.

Jones A. M., Collins R. N., Rose J. and Waite T. D. (2009) Theeffect of silica and natural organic matter on the Fe(II)-catalysed transformation and reactivity of Fe(III) minerals.Geochim. Cosmochim. Acta 73, 4409–4422.

Kogel-Knabner I., Guggenberger G., Kleber M., Kandeler E.,Kalbitz K., Scheu S., Eusterhues K. and Leinweber P. (2008)Organo-mineral associations in temperate soils: integratingbiology, mineralogy, and organic matter chemistry. J. Plant

Nutr. Soil Sci. 171, 61–82.Kostka J. E. and Luther G. W. (1994) Partitioning and speciation

of solid-phase iron in salt-marsh sediments. Geochim. Cosmo-

chim. Acta 58, 1701–1710.



http://refhub.elsevier.com/S0016-7037(14)00564-X/h0005






























































Lalonde A. E., Rancourt D. G. and Ping J. Y. (1998) Accuracy offerric/ferrous determinations in micas: a comparison ofMossbauer spectroscopy and the Pratt and Wilson wet-chem-ical methods. Hyperfine Inter. 117, 175–204.

Larese-Casanova P. and Scherer M. M. (2007) Fe(II) sorption onhematite: new insights based on spectroscopic measurements.Environ. Sci. Technol. 41, 471–477.

Latta D. E., Bachman J. E. and Scherer M. M. (2012) Fe electrontransfer and atom exchange in goethite: influence of Al-substitution and anion sorption. Environ. Sci. Technol. 46,10614–10623.

Liptzin D., Silver W. and Detto M. (2011) Temporal dynamics insoil oxygen and greenhouse gases in two humid tropical forests.Ecosystems 14, 171–182.

Lovley D. R., Holmes D. E. and Nevin K. P. (2004) DissimilatoryFe(III) and Mn(IV) reduction. Adv. Microb. Physiol. 49, 219–286.

Masue-Slowey Y., Loeppert R. H. and Fendorf S. (2011) Alter-ation of ferrihydrite reductive dissolution and transformationby adsorbed As and structural Al: implications for As retention.Geochim. Cosmochim. Acta 75, 870–886.

Mikutta C., Wiederhold J. G., Cirpka O. A., Hofstetter T. B.,Bourdon B. and Von Gunten U. (2009) Iron isotope fraction-ation and atom exchange during sorption of ferrous iron tomineral surfaces. Geochim. Cosmochim. Acta 73, 1795–1812.

Murad E. and Cashion J. (2004) Mossbauer Spectroscopy of

Environmental Materials and their Industrial Utilization. KluwerAcademic Publishers Group, Norwell, Massachusetts.

Murad E. and Schwertmann U. (1980) The Mossbauer spectrum offerrihydrite and its relations to those of other iron-oxides. Am.

Mineral. 65, 1044–1049.Murad E. and Schwertmann U. (1983) The influence of aluminum

substitution and crystallinity on the Mossbauer-spectra ofgoethite. Clay Miner. 18, 301–312.

Pedersen H. D., Postma D., Jakobsen R. and Larsen O. (2005) Fasttransformation of iron oxyhydroxides by the catalytic action ofaqueous Fe(II). Geochim. Cosmochim. Acta 69, 3967–3977.

Peretyazhko T. and Sposito G. (2005) Iron(III) reduction andphosphorous solubilization in humid tropical forest soils.Geochim. Cosmochim. Acta 69, 3643–3652.

Rancourt D. G. (1998) Mossbauer spectroscopy in clay science.Hyperfine Interact. 117, 3–38.

Reyes I. and Torrent J. (1997) Citrate-ascorbate as a highlyselective extractant for poorly crystalline iron oxides. Soil Sci.

Soc. Am. J. 61, 1647–1654.Schaefer M. V., Gorski C. A. and Scherer M. M. (2011)

Spectroscopic evidence for interfacial Fe(II)–Fe(III) electrontransfer in a clay mineral. Environ. Sci. Technol. 45, 540–545.

Schwertmann U., Cambier P. and Murad E. (1985) Properties ofgoethites of varying crystallinity. Clays Clay Miner. 33, 369–378.

Shelobolina E. S., Anderson R. T., Vodyanitskii Y. N., Sivtsov A.V., Yuretich R. and Lovley D. R. (2004) Importance of clay sizeminerals for Fe(III) respiration in a petroleum-contaminatedaquifer. Geobiology 2, 67–76.

Silver W. L., Lugo A. E. and Keller M. (1999) Soil oxygenavailability and biogeochemistry along rainfall and topographicgradients in upland wet tropical forest soils. Biogeochemistry

44, 301–328.Skulan J., Beard B. and Johnson C. (2002) Kinetic and equilibrium

Fe isotope fractionation between aqueous Fe(III) and hematite.Geochim. Cosmochim. Acta 66, 2995–3015.

Sparks D. L. (1996) Methods of soil analysis. Part 3. Soil ScienceSociety of America: American Society of Agronomy, Madison,Wis.

Stumm W. (1987) Aquatic Surface Chemistry: Chemical Processes

at the Particle-Water Interface. Wiley, New York.Sulzberger B., Suter D., Siffert C., Banwart S. and Stumm W.

(1989) Dissolution of Fe(III)(hydr)oxides in natural waters;laboratory assessment on the kinetics controlled by surfacecoordination. Mar. Chem. 28, 127–144.

Thompson A., Chadwick O. A., Rancourt D. G. and Chorover J.(2006) Iron-oxide crystallinity increases during soil redoxoscillations. Geochim. Cosmochim. Acta 70, 1710–1727.

Thompson A., Rancourt D. G., Chadwick O. A. and Chorover J.(2011) Iron solid-phase differentiation along a redox gradient inbasaltic soils. Geochim. Cosmochim. Acta 75, 119–133.

Wiederhold J. G., Kraemer S. M., Teutsch N., Borer P. M.,Halliday A. N. and Kretzschmar R. (2006) Iron isotopefractionation during proton-promoted, ligand-controlled, andreductive dissolution of goethite. Environ. Sci. Technol. 40,3787–3793.

Wielinga B., Mizuba M. M., Hansel C. M. and Fendorf S. (2001)Iron promoted reduction of chromate by dissimilatory iron-reducing bacteria. Environ. Sci. Technol. 35, 522–527.

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

Yang L., Steefel C. I., Marcus M. A. and Bargar J. R. (2010)Kinetics of Fe(II)-catalyzed transformation of 6-line ferrihy-drite under anaerobic flow conditions. Environ. Sci. Technol. 44,5469–5475.

Associate editor: Jeffrey G Catalano


























































































1

Electronic Annex for

“Fe2+ catalyzed Fe atom exchange and re-crystallization in a tropical soil”

Viktor Tishchenko1, Christof Meile2, Michelle Scherer3, Tim Pasakarnis3, and Aaron Thompson1*

1University of Georgia (Crop and Soil Sci.); 2University of Georgia (Marine Sci.); 3University of Iowa (Civil and Environ. Engineering)

*Corresponding author E-mail: [email protected] Prepared: 6 September 2014 Contains four sections with eight Figures and three Tables.

2

Section 1. Additional Tables and Figures

Figure EA1. Precision of 57/54Fe ratio measurements made by single collector ICP-MS (ELAN 6000). Error bars are 2σ based on 30 measurements. The inset illustrates that samples can be statistically distinguished at the 2σ level when they are separated by a 57/54Fe ratio of ~0.015

3

Figure EA2. X-ray diffraction of the soil clay fraction using a Co-Kα source. Values adjacent to the peaks are d-spacings in Å. Mineral identifications include Ch (Chlorite); Ka (Kaolinite); Gt (Goethite); Qz-L (Quartz).

Degrees&2θ&

4

Table EA1. Mass balance and % recovery data for the live control soil. Tr

eatm

ent

Day

Aqueous Fe (mmol kg-1 soil)

0.5M extracted Fe (mmol kg-1 soil)

7M extracted Fe (mmol kg-1 soil)

% re

cove

ry

Total Fe 57Fe/54Fe 1 ƒFe2+ Total Fe 57Fe/54Fe ƒFe2+ Total Fe 57Fe/54Fe ƒFe2+

Live c

ontro

l

0 269.9 5.884 0.981 17.03 3.917 0.259 950.1 0.363 0.0000 -

0.02 234.3 5.740 0.993 34.60 4.289 0.980 882.8 0.392 0.0000 99.0

1 214.4 5.589 0.988 57.82 4.469 0.484 971.1 0.426 0.0000 89.1

3 217.0 5.568 0.984 55.43 4.301 0.464 911.4 0.452 0.0000 88.9

7 214.4 4.992 0.960 66.61 4.206 0.622 1034 0.475 0.0000 91.8

14 241.3 5.302 0.953 52.73 4.223 0.600 950.7 0.473 0.0000 97.2

21 247.3 5.200 0.919 52.93 3.587 0.691 1006 0.501 0.0000 98.1

28 249.1 5.112 0.965 50.01 3.917 0.653 926.9 0.471 0.0000 101.4

1 ƒFe2+ is a fraction of Fe2+ relative to total Fe Standard deviations were calculated from triplicate reactors. All statistical representations are given as one standard deviation of replicate measures in concise notation with the parenthetical number indicating the error in the last digit [e.g., 7.14(12) is equivalent to 7.14 ± 0.12].

5

Section 2. 57Fe Mössbauer spectroscopy of the unreacted and 57Fe2+ exchanged samples

EA-2.1 Introduction We employ 57Fe Mössbauer spectroscopy to identify the major Fe site populations in the Bisley soil. EA-2.2 Methods EA-2.2.1 MB collection and analysis methods Transmission 57Fe Mössbauer spectroscopy was performed with a variable temperature He-cooled system with a 1024 channel detector. A 57Co source (~50 mCi) embedded in a Rh matrix was used at room temperature. Freeze-dried powder samples were mounted between two pieces of 0.127 mm thickness Kapton tape in an anoxic glovebox, and transferred immediately to the spectrometer cryostat to avoid sample oxidation prior to analysis. Velocity (i.e., gamma-ray energy) was calibrated using α-Fe foil at 298 K and all center shifts (CSs) and peak positions are reported with respect to this standard. The transducer was operated in constant acceleration mode and folding was performed to achieve a flat background. EA-2.2.2 MB spectral analysis approach

Mössbauer spectral fitting of all spectra was performed using the Voigt-based fitting (VBF) method of Rancourt and Ping(1991) for quadrupole splitting distributions (QSDs) and combined hyperfine field distributions (HFDs), as implemented in the RecoilTM software, ISA Inc. (http://www.isapps.ca/ recoil). All VBF Mössbauer parameter definitions and a description of the relevant notation are given by Rancourt and Ping(1991). All errors in Mössbauer fitting parameters are two-standard deviation (2σ) errors, as calculated by RecoilTM. In reporting quantitative phase abundances or site populations it is assumed that the Mössbauer recoilless fractions of all detected phases or Fe-bearing components are equal, such that subspectral areas (expressed as fractions of total spectral area) are equal to the amounts of Fe (expressed as fractions of total Fe) in the corresponding phases or components. This assumption is expected to be valid at cryogenic temperatures, and also to be a good approximation at RT with dry samples (Lalonde et al., 1998; Rancourt, 1998). EA-2.2.3 MB spectral fitting approach

In Mössbauer spectroscopy, each spectral component corresponds to one Fe-bearing solid phase or to a group of unresolved Fe-bearing solid phases. These components take the form of a doublet, sextet, octet (none resolved here) or a collapsed sextet—indicating a solid-phase near its magnetic ordering temperature (TN). Solid-phases well above (doublet) or below (sextet) their TN will not exhibit any vertical (i.e. count axis) distance between the peak troughs and the baseline. When Fe solid-phases are near their TN, they exhibit an intermediate shape between a doublet and full sextet, which fills the area between the upper baseline and the inverse troughs of the peaks. We approximate this either by using a separate collapsed sextet component (i.e., a sextet with exceedingly large line widths and Bhf = 0 T) or by mixing low field strength and broad line width elements into the fitting of a dominant broadened sextet (e.g., the 77K spectra in Figure 6). At 13K this soil has very few Fe-oxyhydroxide phases near their TN and there is little advantage to splitting out a separate collapsed sextet component. However, at 77K, 140K, and 295K,

6

significant portions of the Fe population are near their TN and it is more appropriate to fit these Fe populations as a separate component. For comparison of the unreacted and 57Fe2+-exchanged soil at 77K (Figure 6; Table EA3) we combine this collapsed sextet with the broad Fe-OxHy sextet; while in our variable temperature characterization of the unreacted soil we fit the 13K spectra with a separate collapsed sextet component to facilitate comparison of the collapsed sextet spectral area at 77K, 140K and 295K. Instrument difficulties prevented us from collecting a spectrum for the 57Fe2+-reacted soil at 13K, but we were able to collect an excellent spectrum at 20K, which is sufficiently close for comparison between the unreacted and reacted spectra (Table EA2).

Assuming equal Mössbauer recoilless fractions (i.e., that the measurable gamma rays emitted by each phase are proportional to the phase abundance), the total spectral area of a given component is proportional to the amount of Fe in the corresponding solid phase(s). Furthermore, it can then be assumed that each phase occupies the same spectral area at all temperatures, although not the same spectral shape. For instance, nano-scale FeIII-(oxyhydr)oxide minerals form a doublet at 295K and magnetically order to form a sextet only when cooled sufficiently to prevent thermal disordering of the macroscopic magnetic ordering within the material (Rancourt, 2001).

We left all fitting parameters normally used to fine-tune the fit at their most general, conservative values. These include: constraining all Lorentzian half widths at twice the Heisenberg values; disallowing parameter coupling between the center shift (CS), quadrapole splitting (QS or ε, for sextets) and average hyperfine field (Bhf); and constraining the area ratios between the doublet lines to 1:1 and area ratios between sextet lines 2 and 3 to 2:1 and lines 1 and 3 to 3:1 (see Tables EA2 and EA3). An explanation of the four key MB spectral parameters used in our analysis (center shift, quadrapole splitting, hyperfine field strength and the line width) is summarized in the electronic annex of Thompson et al (2011). EA-2.2.4 Mössbauer spectral components

Across the four collection temperatures (295K, 140K, 77K and 13K) we resolved five distinct spectral components, except in the 295K spectra where we did not resolve a magnetically-ordered FeIII-sextet. Spectral components include: (1) an FeIII quadrupole doublet (labelled Q-FeIII-1) corresponding to FeIII in silicates, surface-complexed to solids, and in all FeIII-(oxy)hydroxides that are superparamagnetic (SP) at the collection temperature; (2) a narrow FeII quadrupole doublet (labelled Q-FeII-1) that we attribute to paramagnetic ferrous-bearing phase best approximated by ilmenite; (3) a wide FeII quadrupole doublet (labelled Q-FeII-2) that we attribute to paramagnetic ferrous in silicate phases; (4) a broadened FeIII sextet (labelled HFD-OxHy) that corresponds to magnetically-ordered FeIII oxyhydroxides; and (5) a partially collapsed FeIII ‘sextet’ (labelled HFD-(b)Oxhy) due to a FeIII-(oxy)hydroxide having its SP blocking temperature near the collection temperature. In the 13K spectra, if we assume all the FeIII-oxyhydroxides have magnetically ordered we can refine our component descriptions of the the FeIII quadrupole doublet (Q-FeIII-1) to correspond to FeIII complexed to solids such as organic matter or in non-Fe-rich silicates; the FeIII sextet (HFD-OxHy) would represents all of the Fe-OxHy phases except the most disordered end-member (HFD-(b)OxHy) that is represented by the partially collapsed FeIII ‘sextet’ attributed to FeIII-(oxyhydr)oxides having their SP blocking temperatures near 13K. These components and their parameter values at each temperature are summarized in Tables EA2.

7

In the 77K spectra used to compare the 57Fe-reacted sample with the unreacted sample (Figure 6; Table EA3), we did used a four component fit that was identical to the above described five component fit except we do not include the separate collapsed sextet component (HFD-(b)OxHy), but instead add an additional low field strength element with broad line widths into the dominant sextet (HFD-OxHy). This approach presents some advantages in that the probability distribution of the hyperfine field strength of the whole FeIII-oxyhydroxide population can be conveniently extracted compare between samples (see Figure 5 of the main text). SI-2.2.5 Calculation of phase abundances We calculated the abundance of each Fe-bearing phase from the spectral area of each respective component in the 13K spectra. FeIII-clay, FeII-Illimenite, and FeII-clay abundances are calculated from spectra area of the Q-FeIII-1, Q-FeII-2, and Q-FeII-1 component spectra areas of the 13K spectra, respectively. The FeIII-oxyhyroxide abundance is calculated by summing the HFD-OxHy and HFD-(b)OxHy components in the five component fitting (Figure EA3). EA-2.3. Mössbauer phase analysis Below we provide an analysis of the four Fe populations identified in the unreacted sample. EA-2.3.1 FeIII-oxyhydroxide populations Collection of Mössbauer spectra at 295K, 140K, 77K, and 13K allows us to characterize the crystallinity-continuum of the FeIII-oxyhydroxide solid phases in these soils. All resolved FeIII-(oxyhydr)oxides in this soil at 295K are superparamagnetic or near their blocking temperature. As temperature is lowered portions of the 57FeIII-oxyhydroxide populations are resolved as they magnetically order into a sextet (Figure EA3), with the more crystalline portions of the population ordering at higher temperatures. The spectral area assigned to the full FeIII-oxyhydroxide sextet increases from 0%, 45%, 54% to 64% at 295K, 140K, 77K and 13K, respectively. The spectral parameters of this sextet (QS = -0.122(2); Bhf peak = 49.4 T @ 13K) approach those reported for nano-goethite standards at 4.2K as temperature is lowered (Figure S5). FeIII-oxyhydroxides that are blocked or near their blocking temperature at 13K comprise 15% of the spectral area (Table EA2 and Figure EA3). We have fit this Fe population using a collapsed sextet (HFD-(b)OxHy) (3Q, 2H fit in Table EA2). EA-2.3.2 FeIII-silicate/organic-bound Fe populations Fe atoms in silicates and monomeric organic complexes are too distant from each other to order magnetically, and thus both site populations produce a doublet at 13K with CS (0.4 mm s-1) and QS (0.7 mm s-1) values consistent with octahedrally coordinated, trivalent Fe. FeIII substituted kaolinite (Murad, 1998) typically yields a lower QS value of 0.52 mm s-1, however, the hypothesized presence of Fe2+ in these kaolinites (see below) would likely increase structural distortion and lead to higher Fe3+ quadropole spliting values (Murad and Cashion, 2004). Some of this FeIII may be associated with the 2:1:1 mineral chlorite as XRD analysis suggests this mineral is present in the clay size fraction (Peretyazhko and Sposito, 2005). FeIII in octahedral positions of Fe-poor 2:1 minerals, or those with significant FeII substitution, also yield larger FeIII quadropole splitting values consistent with our observations (Murad and Wagner, 1994).

8

EA-2.3.3 Paramagnetic ferrous populations We observe two paramagnetic ferrous contributions in this sample consistent with (1) ilmenite and (2) sheet silicates or sorbed Fe2+. No octets consistent with the presence of magnetite are observed. The position of the high-energy line at 1.4 mm s-1 is diagnostic for ilmenite. The high-energy lines (HL) at 2.6 mm s-1 are consistent with those commonly reported for aluminosilicate clays(Murad and Cashion, 2004), including chlorite, which was identified previously in these soils via XRD. The HL for ferrous substitution in kaolinite averages 2.4 ± 0.1 mm s-1 based on a compilation of eight studies reporting FeII in kaolinites (Murad and Cashion, 2004) and may contribute to our observed peak position. In addition, adsorbed Fe2+ may also contribute to this peak, as Rancourt et al. (2005) have shown Fe2+ sorbed to bacterial cells can yield HL lines even above 2.8 mm yr-1. EA-2.4 References Lalonde, A. E., Rancourt, D. G., and Ping, J. Y., 1998. Accuracy of ferric/ferrous determinations in micas:

A comparison of Mössbauer spectroscopy and the Pratt and Wilson wet-‐chemical methods. Hyperfine Interact. 117, 175-‐204.

Murad, E., 1998. Clays and clay minerals: What can Mössbauer spectroscopy do to help understand them? Hyperfine Interactions 117, 39-‐70.

Murad, E. and Cashion, J., 2004. Mössbauer Spectroscopy of Environmental Materials and their Industrial Utilization. Kluwer Academic Publishers Group, Norwell, Massachusetts.

Murad, E. and Wagner, U., 1994. Mössbauer study of pure illite and its firing products. Hyperfine Interact. 91, 685-‐688.

Peretyazhko, T. and Sposito, G., 2005. Iron(III) reduction and phosphorous solubilization in humid tropical forest soils. Geochim. Cosmochim. Acta 69, 3643.

Rancourt, D. G., 1998. Mössbauer spectroscopy in clay science. Hyperfine Interact. 117, 3-‐38. Rancourt, D. G. and Ping, J. Y., 1991. Voigt-‐Based Methods For Arbitrary-‐Shape Static Hyperfine

Parameter Distributions In Mössbauer-‐Spectroscopy. Nucl. Instrum. Methods Phys. Res., Sect. B 58, 85-‐97.

Rancourt, D. G., Thibault, P.-‐J., Mavrocordatos, D., and Lamarche, G., 2005. Hydrous ferric oxide precipitation in the presence of nonmetabolizing bacteria: Constraints on the mechanism of a biotic effect. Geochim. Cosmochim. Acta 69, 553-‐577.

Thompson, A., Rancourt, D. G., Chadwick, O. A., and Chorover, J., 2011. Iron solid-‐phase differentiation along a redox gradient in basaltic soils. Geochim. Cosmochim. Acta 75, 119-‐133.

9

Figure EA3. 57Fe Mössbauer spectra of the unreacted soil at 295K, 140K, 77K, and 13K. In each spectrum, the black solid line is the total calculated fit, through the discrete data points. The resolved spectral components and assignments are: (1) Q-FeIII-1, the deep central doublet (blue line) corresponding to FeIII in aluminosilicate minerals; (2) Q-FeII-1, the narrow ferrous doublet corresponding to ilmenite (red line); (3) Q-FeII-2 the wider ferrous doublet corresponding to FeII in clays (green line); (4) HFD-OxHy, the dominant sextet (purple line) corresponding to FeIII-oxyhydroxides that are magnetically ordered; and (5) HFD-(b)OxHy the collapsed ‘sextet’ (yellow line) corresponding to FeIII-oxyhyroxides near their blocking temperature. No HFD-OxHy component is present in the 295K spectrum.

-10 -8 -6 -4 -2 0 2 4 6 8 107.0

7.1

7.2

In

tens

ity (M

c ch

-1)

-10 -8 -6 -4 -2 0 2 4 6 8 10

5.4

5.5

In

tens

ity (M

c ch

-1)

-10 -8 -6 -4 -2 0 2 4 6 8 10

5.0

5.1

In

tens

ity (M

c ch

-1)

-10 -8 -6 -4 -2 0 2 4 6 8 104.5

4.6

4.7

4.8

In

tens

ity (M

c ch

-1)

Velocity (mm s-1)

13 K

77 K

140 K

295 K

10

Figure EA4. Expanded low velocity, doublet region of the unreacted Bisley sample at 13K. The three doublets described here are the same as those in Figure S3: (1) Q-FeIII-1, the deep central doublet (blue line) corresponding to FeIII in aluminosilicate minerals; (2) Q-FeII-1, the narrow ferrous doublet corresponding to ilmenite (red line); (3) Q-FeII-2 the wider ferrous doublet corresponding to FeII in clays (green line). CS=center shift and QS=quadrapole splitting. The FeII doublets are assigned based on the position of the high-energy lines (HL) to FeII in clays and ilmenite.

-2 0 2 47.0

7.1

7.2

In

tens

ity (M

c ch

-1)

Velocity (mm s-1)

Fe(II) – Clay CS = 1.1 QS = 3.0 HL = 2.6

2.6 1.4

Fe(II) – ilmenite CS = 0.89 QS = 1.1 HL = 1.4

Fe(III) – silicate/organic CS = 0.4 QS = 0.7

-0.20 -0.15 -0.10 -0.05 0.00 0.05

45

48

51

54

Hpe

ak (T

)

<e> (mm s-1)

140K

77K

13K

nGt Fh

Lp

Hm

11

Figure EA5. Most probable hyperfine field, Hpeak, vs. average quadrupole shift, e, for the HFD-OxHy sextets (purple lines in Figures S3 and S4) at 13K, 77K and 140K (red star symbols) and sextets from reference oxides and (oxy)hydroxides collected at 4.2K from Van der Zee et al. (2003) (Fh, ferrihydrites; Gt, goethites; Hm, hematites; Lp, lepidocrosites).

-2 0 2 47.0

7.1

7.2

In

tens

ity (M

c ch

-1)

Velocity (mm s-1)

Fe(II) – Clay CS = 1.1 QS = 3.0 HL = 2.6

2.6 1.4

Fe(II) – ilmenite CS = 0.89 QS = 1.1 HL = 1.4

Fe(III) – silicate/organic CS = 0.4 QS = 0.7

-0.20 -0.15 -0.10 -0.05 0.00 0.05

45

48

51

54

Hpe

ak (T

)

<e> (mm s-1)

140K

77K

13K

nGt Fh

Lp

Hm

12

Figure EA6. 57Fe Mössbauer spectra of the 57Fe-reacted soil at 20K. The black solid line is the total calculated fit, through the discrete data points. The resolved spectral components and assignments are: (1) Q-FeIII-1, the deep central doublet (blue line) corresponding to FeIII in aluminosilicate minerals; (2) Q-FeII-1, the narrow ferrous doublet corresponding to ilmenite (red line); (3) Q-FeII-2 the wider ferrous doublet corresponding to FeII in clays (green line); (4) HFD-OxHy, the dominant sextet (purple line) corresponding to FeIII-oxyhydroxides that are magnetically ordered; and (5) HFD-(b)OxHy the collapsed ‘sextet’ (yellow line) corresponding to FeIII-oxyhyroxides near their blocking temperature.

-10 -8 -6 -4 -2 0 2 4 6 8 10

Inte

nsity

(Mc

ch-1)

57-Reacted soil at 20K

Velocity (mm s-1)

20 K

13

Table-EA2: Temperature profile Mossbauer parameters (see Figure EA3)

Sample Phase Spectral Area δ0 ε0 P Δ or H σ Red-Χ2 <CS> < ε > <QS> or <H> s.d. MC*mm/s % mm/s mm/s % mm/s or T mm/s or T mm/s mm/s mm/s or T mm/s or T

Untreated soil 295K (3Q, 2H) (BG = 4.762 MC/ch)

Q-FeIII-1 0.176(3) 56(1) 0.358(3) n/a 100* 0.601(5) 0.253(7) 1.37 0.358 n/a 0.602 0.25 Q-FeII-1 0.011(2) 3.6(6) 0.71(2) n/a 100* 1.11(4) 0.09(7) 0.71 n/a 1.11 0.09 Q-FeII-2 0.012(2) 4.0(7) 0.99(3) n/a 100* 2.85(7) 0.28(8) 0.99 n/a 2.85 0.28

HFD-OxHy 0 0 n/a n/a n/a n/a n/a n/a n/a n/a n/a

HFD-(b)OxHy 0.113(6) 36(1) 0.48(7) 0* 100* 0* 23(2) 0.48 0 18.28 13.8

Untreated soil 140K (3Q, 2H) (BG =5.1471 MC/ch)

Q-FeIII-1 0.141(4) 20.6(7) 0.424(7) n/a 100* 0.701(9) 0.39(2) 1.08 0.424 n/a 0.713 0.37 Q-FeII-1 0.017(3) 2.5(4) 0.81(2) n/a 100* 1.12(4) 0.14(5) 0.81 n/a 1.12 0.14 Q-FeII-2 0.017(2) 2.4(3) 1.16(2) n/a 100* 2.95(4) 0.21(5) 1.16 n/a 2.95 0.21

HFD-OxHy 0.31(1) 45(1) 0.463(6) -0.112(6) 44.23* 46.44(9) 1.8(2) 0.463 -0.112 43.97 4.05 55.8 42.0(7) 4.3(4) (Peak H 46.17 T)

HFD-(b)OxHy 0.20(1) 29(1) 0.56(7) 100* 0* 31.9(27) 0.56 0 25.4 19.2



HFD-OxHy 0.406(7) 54.4(9) 0.478(3) -0.125(3) 57.9* 48.51(4) 1.36(8) 0.478 -0.125 47.14 2.796 42.1 45.3(6) 3.1(3) (Peak H 48.4 T)

HFD-(b)OxHy 0.177(10) 23.8(11) 0.53(9) 0* 100* 0* 34.5(30) 0.533 0 27.54 20.81



HFD-OxHy 0.660(6) 65.4(9) 0.481(2) -0.122(2) 58.3* 49.42(2) 1.09(4) 0.481 -0.122 48.64 2.22 41.7 47.6(2) 2.9(1) (Peak H 49.4 T)

HFD-(b)OxHy 0.15(1) 15.0(10) 0.41(18) 0* 100* 0* 43(5) 0.412 0 33.97 25.66

57Fe reacted soil 20K (3Q, 1H)

(BG =18.719 MC/ch)


HFD-OxHy 0.573(11) 64.4(16) 0.484(4) -0.116(4) 49.3* 49.40(4) 1.19(8) 0.484 -0.116 47.98 2.92 50.7 46.6(3) 3.4(2) (Peak H 49.3 T)

HFD-(b)OxHy 0.185(19) 20.8(17) 0.41* 0* 100* 0* 42(7) 0.41 0 34.1 25.8

14

Notes

All fits performed using the Voigt-based fitting method of Rancourt and Ping (1991) with the Recoil™ software. All fitting and calculated parameters are as defined in by Rancourt and Ping (1991). All delta-1 couplings between CS and z (or DELTA) were taken to be zero. All line-1 to line-2 area ratios in all (distributed) elemental doublets were taken to be 1. All line-2/line-3 and line-1/line-3 area ratios in all (distributed and symmetric) elemental sextets were taken to be 2 and 3, respectively. All epsilon-1 couplings between epsilon and z (in a HFD) are taken to be 0. BG = background level, in mega-counts per channel (MC/ch). phase = assigned spectral component, as described in the text. DELTA = the center (or position) of a Gaussian component in the quadrupole splitting distribution (QSD) of a given doublet spectral component. z = the center (or position) of a Gaussian component in the hyperfine field distribution (HFD) of a given 'sextet' spectral component. sigma = the Gaussian standard deviation width of a given Gaussian component of a given QSD or HFD. p (in %) is the weight factor for a given Gaussian component in a given QSD or HFD. <A> signifies the average of the absolute value of A. <z> is the average magnitude of the hyperfine field (expressed as an excited state Zeeman splitting, in mm/s) in a given HFD of a given sextet spectral component. <QS> is the average magnitude of the quadrupole splitting in a given QSD of a given doublet spectral component. <eps> is the average magnitude of the slave distribution of quadrupole shifts (epsilons) associated to a given HFD of a given sextet spectral component. s.d. is the standard deviation width of a given distribution (QSD, HFD, or slave distributions) chi-2-red is the reduced chi-squared value for the fit: chi-squared divided by the number of degrees of freedom. It has an ideal value of 1 for a correct model. All center shifts (CSs, delta-0, <CS>) are given with respect to the CS of metallic Fe at 295K.

15

Table-EA3: Fitting and calculated Mossbauer parameters for 77 K spectra.

Sample Phase Spectral Area δ0 ε0 P Δ or H σ Red-Χ2

<CS> < ε > <QS> or <H> s.d.

MC*mm/s % mm/s mm/s % mm/s or T mm/s or T mm/s mm/s mm/s or T mm/s or T


Q-FeIII-1 0.134(4) 18.0(5) 0.428(7) n/a 100* 0.72(1) 0.43(2) 2.33 0.428 n/a 0.733 0.40 Q-FeII-1 0.020(1) 2.7(3) 0.86(1) n/a 100* 1.19(3) 0.13(4) 0.86 n/a 1.19 0.126 Q-FeII-2 0.019(2) 2.6(3) 1.16(2) n/a 100* 3.04(3) 0.16(4) 1.16 n/a 3.04 0.159 HFD-OxHy

0.571(7) 76.7(6) 0.477(3) -0.124(3) 39.0* 48.58(4) 1.3(1) 0.477 -0.124 41.56 13.19

29.3 45.5(7) 2.8(3) (Peak H 48.44 T) 31.7 28(2) 20(2)

57Fe reacted soil 77K (3Q, 1H) (BG =3.8373 MC/ch)

Q-FeIII-1 0.34(1) 19.6(6) 0.449(7) n/a 100* 0.732(9) 0.37(2) 1.87 0.449 n/a 0.739 0.359 Q-FeII-1 0.033(7) 1.9(4) 0.82(3) n/a 100* 1.09(6) 0.15(7) 0.82 n/a 1.09 0.153 Q-FeII-2 0.040(5) 2.3(3) 1.17(2) n/a 100* 2.99(4) 0.16(5) 1.17 n/a 2.99 0.155 HFD-OxHy 1.33(2) 76.2(7) 0.476(5) -0.123(5) 29.5* 48.45(6) 1.5(1) 0.476 -0.123 38.52 13.91

31.2 44.2(7) 3.8(5) (Peak H 48.29 T) 39.3 26(2) 17(1)

See Table S2 for notes

16

Section 3. Kinetic modelling annex Parameter sensitivity was assessed by introducing a 1% perturbation to each of the following parameters and examining changes in the rate of desorption (Rdes), which is the essential measurement for calculating the rate of atom exchange. We found that the rate of desorption (Rdes) increases directly when kdes, C and Stotal are increased (Eq. 2), with kdes and Stotal having a larger relative influence than C, because changing kdes and Stotal also alters the calculation of Sopen. An increase in kads and—most significantly—a decrease in Ttotal also results in a lower Sopen value, which in our model leads to a smaller rate of desorption (Rdes , e.g., Eq. 2).

Figure EA7: Impact of a 1% positive (+) and negative (-) parameter perturbation on the rate of desorption

(Rdes) at day 28, defined as (Rdes(p+Δpi) – Rdes(p))/Rdes(p) * pi/Δpi, where pi is the parameter of interest.

-‐2

-‐1

0

1

2

3

4

5

6

ΔR d

es/R

des * p/Δp

+ -‐

17

Figure EA8. Model-derived approach to equilibrium for the isolated Fe atom exchange processes (f’).

0 5 10 15 20 25 300%

2%

4%

6%

8%

10%

%

app

roac

h to

equ

ilibr

ium

Days

Aqueous Bulk

18

Section 4. Calculation of 57Fe in the reacted soil sample We collected MB spectra on the sterile soil reacted with 57Fe2+ for 28 days, after rinsing it three times with an Fe-free KCl-MES buffer solution to remove any Fe2+

(aq). We quantified the 57Fe accumulated by soil sample by following steps:

57Fe accumulated before KCl-MES rinsing

Surface/labile and bulk pools accumulated 11.08 and 9.26 mM kg-1 of 57Fe respectively (see below for 57Fe calculation). Fe accumulated by soil sample before two washes in both pools was calculated as 57Fes = 57Fef – 57Fei EA-4.1 57Fes(surface) = 11.44 mM kg-1 - 0.36 mM kg-1 = 11.08 mM kg-1 57Fes(bulk) = 20.13 mM kg-1 – 29.39 mM kg-1 = 9.26 mM kg-1 where 57Fes is the iron-57 accumulated, 57Fei is the iron in the surface/labile and bulk pools before the spike was added, and 57Fef is the average amount of iron surface/labile and bulk pools after 28 days of reaction.

57Fe removed by KCl-MES rinsing

Rinsing the soil with KCl-MES buffer removed a significant portion of the weakly sorbed Fe in the surface/labile layer. We quantified this using a separate experiment with un-labeled Fe2+(aq) under conditions identical to our atom exchange experiments (see main text methods). In this experiment we calculated that out of 47.6 mM kg-1 of sorbed Fe2+, 19.4 and 4.88 mM kg-1 of Fe2+ desorbed after the first and the second washes respectively, accounting for 24.3% ± 6.3 (±σ, n=6) of the sorbed Fe2+ removed by the washes. This allowed us to calculate the quantity of the desorbed 57Fe in our main experiment as follows: The 57Fe-reacted soil sorbed 62.18 mM kg-1 calculated as Fes =Fei.aq. – Fef.aq. EA-4.2 Fes = 269.9 mM kg-1 - 207.8 mM kg-1 = 62.18 mM kg-1 where Fes is the iron sorbed, Fei.aq. is the aqueous iron before soil sample added, and Fef.aq. is the average amount of aqueous Fe remaining after 28 days. The KCl-MES buffer rinsing removed 24.3% or 15.1 mM kg-1 of Fe2+. Assuming that desorbed Fe had a similar isotopic composition to the surface/labile Fe after sorption/desorption equilibrium the removed iron contained 3.93 mM kg-1

of 57Fe.

57Fe accumulated after KCl-MES rinsing

The final amount of 57Fe accumulated in the reacted soil following two KCl-MES washes was calculated as

19

57Fes.f. = 57Fes.(surface) + 57Fes.(bulk)– 57Fer. EA-4.3 57Fes.f. = 11.08 mM kg-1 + 9.26 mM kg-1 - 3.93 mM kg-1 = 16.40 mM kg-1 where 57Fes.f. is the final 57Fe accumulated after KCl-MES rinsing, 57Fes is the 57Fe sorbed by soil before KCl-MES rinsing, and

57Fer is the 57Fe removed by KCl-MES rinsing.

Calculation of the individual 57Fe, 54Fe and 56+58Fe content

We measured the total Fe concentration and the 57Fe/54Fe isotopic ratio on our ICP-MS, however we need to calculate approximate molar abundances of 57Fe, 54Fe, and 56Fe + 58Fe for use in our model (Section 3) and to approximate the amount of 57Fe accumulated in the reacted soil (this Section). We do this by assuming the ratios of the other three Fe isotopes (54Fe:56Fe:58Fe) remain at their native abundance ratios throughout the experiment. This is not strictly valid since the 54Fe:56Fe:58Fe in the spike solution likely differs from that in the soil. However, our spike solution was a mixture of natural isotopic abundance Fe and an aliquot of 97.82% pure 57Fe, so this assumption is valid within the analytical limits of this experiment. Using this assumption we calculate the concentration of the isotopes as follows: Since, ∑Fe =54Fe+57Fe+56Fe+58Fe EA-4.4 and we assume a constant 54Fe:56Fe:58Fe ratio, we can solve for the amount of 54Fe as: 𝐹𝑒!" = !"

!"!"!"!"!!".!"#$

EA-4.5

Subsequently 57Fe can be solved knowing 57Fe/54Fe ratio and 54Fe 54Fe = 57Fe/54Fe * 54Fe EA-4.6 The other isotopes can be calculated by substituting for 54Fe in the above equations.

Fe2+ catalyzed iron atom exchange and re-crystallization in a tropical soil

Documents

Transcript of Fe2+ catalyzed iron atom exchange and re-crystallization in a tropical soil