Products of abiotic U(VI) reduction by biogenic magnetite and vivianite

17 Pages·2011·1.37 MB·English

by  Harish Veeramani

Available online at www.sciencedirect.com
GeochimicaetCosmochimicaActa75(2011)2512–2528
www.elsevier.com/locate/gca
Products of abiotic U(VI) reduction by biogenic magnetite
and vivianite
Harish Veeramania,1,⇑, Daniel S. Alessia, Elena I. Suvorovaa,
Juan S. Lezama-Pachecob, Joanne E. Stubbsb, Jonathan O. Sharpa,2, Urs Dipponc,
Andreas Kapplerc, John R. Bargarb, Rizlan Bernier-Latmania
aEnvironmentalMicrobiologyLaboratory,EcolePolytechniqueFe´de´raledeLausanne(EPFL),Switzerland
bStanfordSynchrotronRadiationLightsource(SSRL),USA
cGeomicrobiology,CenterforAppliedGeosciences(ZAG),Universita¨tTu¨bingen,Germany
Received20September2010;acceptedinrevisedform14February2011;availableonline21February2011
Abstract
Reductiveimmobilizationofuraniumbythestimulationofdissimilatorymetal-reducingbacteria(DMRB)hasbeeninves-
tigatedasaremediationstrategyforsubsurfaceU(VI)contamination.Inthoseenvironments,DMRBmayutilizeavarietyof
electron acceptors, such as ferric iron which can lead to the formation of reactive biogenic Fe(II) phases. These biogenic
phases could potentially mediate abiotic U(VI) reduction. In this work, the DMRB Shewanella putrefaciens strain CN32
was used to synthesize two biogenic Fe(II)-bearing minerals: magnetite (a mixed Fe(II)–Fe(III) oxide) and vivianite (an
Fe(II)-phosphate). Analysis of abiotic redox interactions between these biogenic minerals and U(VI) showed that both
biogenicmineralsreducedU(VI)completely.XASanalysisindicatessignificantdifferencesinspeciationofthereducedura-
niumafterreactionwiththetwobiogenicFe(II)-bearingminerals.Whilebiogenicmagnetitefavoredtheformationofstruc-
turally ordered, crystalline UO , biogenic vivianite led to the formation of a monomeric U(IV) species lacking U–U
2
associations in the corresponding EXAFS spectrum. To investigate the role of phosphate in the formation of monomeric
U(IV)suchassorbedU(IV)speciescomplexedbymineralsurfaces,versusaU(IV)mineral,uraniumwasreducedbybiogenic
magnetitethatwaspre-sorbedwithphosphate.XASanalysisofthissamplealsorevealedtheformationofmonomericU(IV)
speciessuggestingthatthepresenceofphosphatehindersformationofUO .ThisworkshowsthatU(VI)reductionproducts
2
formedduringinsitubiostimulationcanbeinfluencedbythemineralogicalandgeochemicalcompositionofthesurrounding
environment, aswell asbythe interfacial solute–solid chemistry of the solid-phasereductant.
(cid:2)2011Elsevier Ltd. Allrights reserved.
1. INTRODUCTION contaminated sites, in situ reductive bioremediation has
received appreciable attention due to its perceived cost-
Uranium mining and processing for nuclear weapons effectiveness when compared to pump-and-treat methods,
production has led to extensive uranium contamination of andbecauseitobviatestheneedforoff-sitehandlingofhaz-
soil and groundwater at US Department of Energy ardousmaterials(PalmisanoandHazen,2003).Hexavalent
(DOE) sites. Among options for remediating uranium- uranium[U(VI)],thevalencestateofcontaminanturanium
at most sites, is stable in oxic environments and typically
occursasaqueouscarbonatecomplexesinoxicgroundwa-
⇑ Correspondingauthor.Tel.:[phone].
ter at circumneutral pH. In contrast, tetravalent uranium
E-mailaddress:[email protected](H.Veeramani).
1 Current address: Department of Geosciences, Virginia Poly- [U(IV)], produced by biological or abiotic processes, is
technicInstituteandStateUniversity,USA. stable in anoxic environments and often occurs as the
2 Currentaddress:EnvironmentalScience&Engineering,Colo- sparingly soluble mineral, uraninite (UO2) (Langmuir,
radoSchoolofMines,USA. 1978).
0016-7037/$-seefrontmatter(cid:2)2011ElsevierLtd.Allrightsreserved.
doi:10.1016/j.gca.2011.02.024
AbioticU(VI)reductionbybiogenicmagnetiteandvivianite 2513
Subsurfaceuraniumcanbereducedbyanumberofabi- (BehrendsandVanCappellen,2005).Previousstudiesthat
otic (Behrends and Van Cappellen, 2005) and microbially- focused on chemogenic analogs may not have accounted
mediated processes (Abdelouas et al., 1998; Elias et al., forimportantpropertiescharacteristicofbiogenicminerals
2003) including reductive immobilization of uranium by such as their nano-size and associated enhanced reactivity
dissimilatory metal-reducing bacteria (DMRB) (Lovley, (O’Loughlin et al., 2003; Regenspurg et al., 2009). Two
1993).ThesebacteriacatalyzeU(VI)reductionusingorgan- such minerals are biogenic magnetite and vivianite both
ic acids, alcohols or H as electron donors and utilize of which have shown to be produced as an end product
2
Fe(III) asgrowth-supporting electronacceptors. of microbial Fe(III) reduction and environmentally perti-
Fe-oxides and iron-bearing clay minerals, which are nent under Fe(III) reducing conditions (Fredrickson
widely distributed in soils and sediments, represent a large et al.,1998; Kostkaet al.,2002).
reserveofFe(III)forDMRB(Kostkaetal.,2002;Zachara InteractionsbetweenU(VI)andmagnetitehavereceived
etal.,2002;KapplerandStraub,2005).Thisincludescon- appreciable attention because magnetite is a ubiquitous,
taminated US-DOE sites where cleanup efforts are under- environmentally relevant ferrous-bearing oxide, a meta-
way to immobilize uranium as uraninite (N’Guessan bolic byproduct of bacterial respiration, and a corrosion
etal.,2008).Evidencefrombothfieldandlaboratorystud- productofsteelwithramificationsfornuclearwasterepos-
iesalsosuggestanexusbetweenironredoxcyclingandura- itories(Ishikawaetal.,1998;Dodgeetal.,2002;Iltonetal.,
niumredoxprocesses(Galloway,1978;Posey-Dowtyetal., 2010).Microbialreductionofamorphousferricoxyhydrox-
1987).ThebiostimulationofDMRBwilllikelyleadtobio- ide(Fe(OH) )hasbeenreportedtoinducetheformationof
3
logical Fe(III) reduction (Wielinga et al., 2000; Finneran magnetite(Belletal.,1987;Lovleyetal.,1987;Moskowitz
et al., 2002; Anderson et al., 2003; Elias et al., 2004;) and etal.,1989;Zhangetal.,1997;Konhauser,1998).Similarly,
production of sorbed Fe(II) or Fe(II)-bearing minerals as magnetiteformationfromthereductionofaqueousFe(III)
metabolic products. The Fe(II)-bearing phases found in- precursors catalyzed by sulfate-reducing microorganisms
clude magnetite, siderite, vivianite, ferruginous smectite, such as Desulfovibrio spp. has been reported (Sakaguchi
and green rust (Bell et al., 1987; Roden and Zachara, et al., 1993; Sakaguchi et al., 2002). Magnetite formation
1996; Fredrickson et al., 1998; Zachara et al., 1998; Dong has also been reported during biooxidation of Fe(II) cou-
et al., 2000; Roh et al., 2003; O’Loughlin et al., 2007; pled to denitrification (Chaudhuri et al., 2001). A number
Komlosetal.,2008;O’Loughlinetal.,2010).SorbedFe(II) of studies have investigated the role of magnetite in ura-
andtheFe(II)-bearingbiogenicphasescanprovideareser- nium reduction and the findings varied greatly ranging
voir of reducing capacity where reduction of U(VI) may from noobservablereduction(Dodgeetal.,2002)toclear
occur due to abiotic interactions (O’Loughlin et al., evidence of reduction (Scott et al., 2005; Aamrani et al.,
2010). This process may compete with direct enzymatic 2007; O’Loughlin et al., 2010) to the formation of a
microbial reduction of U(VI)(Fredricksonet al.,2000). mixed-valence U(IV)–U(VI) phase (Missana et al., 2003;
AlthoughreductionofU(VI)byaqueousFe(II)isther- Aamranietal.,2007;Regenspurgetal.,2009)ortheforma-
modynamicallyfavorable,itcanbekineticallylimited,often tionofU(V)(Iltonetal.,2010).Thevariationinfindingsis
necessitating an appropriate adsorbent to react with aque- presumably linked to variability in morphology, specific
ousFe(II)andcatalyzethereaction.Researchthusfarhas surfaceareaandphasestoichiometry(GorskiandScherer,
demonstratedU(VI)reductionbyFe(II)sorbedontoavari- 2009, Gorski et al., 2010) of the magnetite used as well as
etyofironoxides/oxyhydroxides(Charletetal.,1998;Liger differences in experimental conditions.
et al., 1999; Fredrickson et al., 2000; Jeon et al., 2004), In phosphate-rich reducing environments, vivianite
Fe(II)-containing natural sediments (Behrends and Van (Fe (PO ) (cid:2)8H O)isanimportantsinkfordissolvedFe(II)
3 42 2
Cappellen, 2005; Jeon et al., 2005), Fe(II)-containing and is considered a stable mineral due to its low solubility
carboxyl-functionalized microspheres (Boyanov et al., at neutral pH (Nriagu and Dell, 1974; Buffle et al., 1989;
2007), Fe(II) sorbed on corundum (Regenspurg et al., Manningetal.,1991;Al-BornoandTomson,1994;Viollier
2009) and Fe(II) sorbed on montmorillonite (Chakraborty et al., 1997; Sapota etal., 2006). Underanoxic conditions,
et al., 2010). These studies primarily consider surface vivianite is very stable (K =10(cid:3)36; (Nriagu, 1972)) and
sp
catalyzed processes that involved either concomitant or can exert significant control over the geochemical cycles
sequential adsorption of aqueousFe(II) and U(VI) species of Fe and P. Vivianite has also been reported as an end
ontoasolidphaseadsorbentormineraltomediateabiotic product of bacterial Fe(III) reduction (Fredrickson et al.,
U(VI)reduction. 1998; Zachara et al., 1998; Roh et al., 2007; Peretyazhko
Likewise,U(VI)canadsorbdirectlyontoFe(II)-bearing etal.,2010).Toourknowledge,theroleofbiogenicvivia-
minerals and undergo reduction by structurally bound nite in abioticuranium reduction andsubsequentimmobi-
Fe(II). For instance, chemogenic green rust and silicates lization hasnotbeen investigated.
including various micas as well as ferrous-bearing sulfide Redoxprocesseslinkingbiogenicmagnetiteorvivianite
mineralssuchasgalenaandpyritehavebeenshowntoad- and uranium were systematically investigated and the fac-
sorb and reduce U(VI) (Wersin et al., 1994; O’Loughlin tors controlling the product of U(VI) reduction probed in
et al., 2003; Ilton et al., 2004; Ilton et al., 2005; Ilton the present study. Biogenic magnetite and vivianite were
et al.,2006; BruggemanandMaes,2010). produced by Shewanella putrefaciens CN32 and character-
Biogenic Fe(II)-bearing minerals are of interest in the izedbyscanning electronmicroscopy (SEM),transmission
context of uranium redox cycling and bioremediation electron microscopy (TEM), X-ray powder diffraction
because they are formed under Fe-reducing conditions (XRD) and Mo¨ssbauer spectroscopy. Their propensity to
2514 H.Veeramanietal./GeochimicaetCosmochimicaActa75(2011)2512–2528
reduceaqueousuranylspecieswasinvestigatedbyacombi- diffraction (XRD) did not show significant diffraction
nation of wet chemistry analyses and X-ray absorption peaks. The HFO suspension was stored at 4(cid:3)C and used
spectroscopy(XAS).Understandingdifferencesinuranium within 4 days to prevent phase transformation. The HFO
speciation in the presence of various environmentally rele- suspension was not sterilized due to the risk of crystalliza-
vant biogenic Fe(II)-bearing minerals is critical due to its tion and/or phase transformation at elevated temperature
potential implication for long-termU(IV) reactivity. and pressure.
2. MATERIALSAND METHODS 2.1.4. Ferriccitratestock solution
About 0.5M ferric citrate (Sigma–Aldrich) was dis-
2.1.Cultures, media andsolutions solved in boiling water on a magnetic stirrer, cooled to
room temperature, adjusted to pH 7 with NaOH, and
2.1.1.Culture andmedia autoclaved.
Microbial culturing was carried out according to stan-
dard microbiological procedures under aseptic conditions. 2.1.5. Uranylacetate solution
A frozen stock culture (50% glycerol) of S. putrefaciens Uranyl acetate powder was dissolved in MilliQ water
CN32wasstreakedonsterilePetridishescontainingLuria resultingina20mMstocksolutionthatwasfilter-sterilized
Bertani(LB)agar(AppliChemGmbH,Germany).Follow- (0.2lm polyethersulfone (PES)), and stored in an amber
inga24hincubationat30(cid:3)C,asinglecolonywaspicked, colored bottle withinthe anoxic chamberatmosphere.
transferredto10mlofsterileLBbrothandincubatedover-
night(ca.12h)at30(cid:3)C.Analiquotofthebrothwasinoc- 2.2. Biological Fe(III)reduction
ulated(1%)into100mlofLBbrothinabaffledflask,which
was incubated at 30(cid:3)C in a shaker (Excella E32, New M4medium(1L)inaduplicatesetof2Lanoxicbottles
BrunswickScientific,NewJersey,USA)for6–8htoobtain was amended either with HFO or Fe(III)-citrate to a con-
cellsatthemid-logarithmicphasethatwasusedforexper- centration of approximately 50±5mM Fe(III). Sodium
iments. Iron reduction experiments were carried out in a dihydrogen phosphate (NaH PO ) to a concentration of
2 4
minimalmediumM4(allmediaingredientspurchasedfrom 5mM was added to medium containing Fe(III)-citrate to
Acron Organics), the composition of which is tabulatedin favor the formation of vivianite. A 1% inoculum of LB-
Table SC-1 of the supplementary content (SC). A liter of grown mid-logarithmic phase culture of S. putrefaciens
themediumwastransferredtoeachoffour2-lanoxicbot- CN32 or G. sulfurreducens was added to each Fe(III) con-
tles and the headspace was purged with sterile nitrogen, taining anoxic bottle mentioned above. The cultures were
sealedunderslightpositivepressureandautoclaved.Geob- incubatedonarotaryshaker(140rpm)at30(cid:3)C(InforsHT
acter sulfurreducens (DSM 12127) was purchased from Multitron Standard). At timed intervals, 0.5ml of sample
DSMZ (Deutsche Sammlung von Mikroorganismen und from anoxic bottles above was withdrawn using a sterile
ZellkulturenGmbH)asalivecultureandculturedaccord- syringeandneedle(pre-purgedwithsterileN )andacidified
2
ing to theirrecommendations. using0.5mlof1MHCl(incubatedfor1week).Fe(II)was
measured using the ferrozine colorimetric assay as de-
2.1.2.Solutions scribed by Stookey (1970) on a UV–vis spectrophometer
Unless indicated otherwise, sample preparation, experi- (Shimadzu UV-2501PC).
mental setup and procedures were conducted under strict
anoxic conditions – either in serum bottles equipped with 2.3. Characterization of Fe(II)phases
a butyl rubber septum and an aluminum crimp or inside
an anoxic chamber with an atmosphere of 2.2% H and After no further observable Fe(III) reduction, the bio-
2
97.8% N . All chemicals used were of ultrapure analytical genicmagnetiteandvivianitesettledtothebottomofeach
2
grade. Stock solutions were boiled and purged for several anoxic bottle leaving bacterial cells mostly in suspension.
hours with N before use. Glassware was soaked in 10% Inside an anoxic chamber (COY Laboratory Products,
2
HClovernight(ca.14h)andwashedfivetimeswithdeion- Inc., Grass Lake, MI) the supernatant from each bottle
izedwaterand MilliQ water, respectively priorto use. was decanted and the remaining solid phase was resus-
pended in 100ml of anoxic MilliQ water that was loaded
2.1.3.Hydrousferric oxide(HFO)synthesis into gas-tight centrifuge bottles equipped with an O-ring.
A1litersolutionof0.5Mferricchloride(Sigma-Aldrich The bottles were centrifuged at 10,000g for 10min. This
GmbH,Germany)wasplacedonamagneticstirrerandti- washing procedure wasrepeated fivetimes.
trated drop-wise with 1M NaOH until a pH of 6.5 was
reached. The resulting precipitate was allowed to settle, 2.3.1. X-raypowderdiffraction (XRD)
resuspended in MilliQ water and centrifuged (Beckman Insidetheanoxicchamber,washedsampleswereapplied
Coulter Avanti J-26XP) at 10,000g. Water washing and tosilicon“zerobackground”sampleholders(PANalytical),
centrifugationwererepeatedseveraltimestoremoveresid- whichweresealedingas-tightjarsfortransporttotheXRD
ual chloride. The resulting pellet was re-suspended in facility. Diffraction measurements were carried out using
100mlofpure MilliQ waterandsonicated(BransonSoni- Cu Ka radiation (X‘Pert Pro MD – PANalytical) in
fier 150, Alys Technologies, Lausanne) briefly to break up Bragg-Brentano geometry with a start and end angle of
large aggregates. An aliquot of HFO analyzed by X-ray 5(cid:3)/2h and 90(cid:3)/2h, respectively. A step size of 0.016, a time
AbioticU(VI)reductionbybiogenicmagnetiteandvivianite 2515
perstepof3sandascanspeedof0.083(cid:3)/swereused.Qual- agglomerated and non-agglomerated particles observed
itative analysis was done by using the X’Pert HighScore withvariousmasksforbetterviewingtheedgesofparticles.
Plussoftwarethatenabledautomaticbackgroundsubtrac-
tion of thespectra. 2.3.4. Mo¨ssbauer spectroscopy
Sampleswerepreparedinsideananoxicchamberbyfil-
2.3.2.BET tering the mineral suspensions through 0.45lm pore-size
Inside an anoxic chamber 2ml of each washed mineral filters (MF-Millipore Membrane, mixed cellulose esters,
suspension was centrifuged (Eppendorf Mini-spin) in Hydrophilic,13mm).Filterscontainingtheretainedsolids
microfuge tubes and the resulting pellets were dried at were sealed using Kapton tape and transported under an-
30(cid:3)Cusingaheatingblock.Thedriedmineralsweretrans- oxic conditions to the spectrometer (Wissenschaftliche
ferredintospecificsamplecellsthatweresealedwithrubber Elektronik GmbH, Germany). Samples were mounted in
caps to maintain anoxic conditions during transportation a closed-cycle helium cryostat (Janis Research Company,
andinstrumentmanipulation.Thesurfaceareasofthemin- Inc., USA) that allowed cooling of the sample to 5.5K.
erals were analyzed by a multi-point BET analyzer Mo¨ssbauer spectra were collected in transmission mode
(Micromeritics Gemini 2360) using nitrogen as the probe using a 57Co source embedded within a Rhodium matrix.
gas. The estimated error in the measurement according to Thesourcewasoperatedatroomtemperatureusingacon-
the manufacturerwas ca. 5%(relative). stant acceleration drive system set to a velocity range of
±12mm/s with movement error of <1%. The absorption
2.3.3.Electron microscopy (EM) was recorded using a proportional counter and a 1024-
Samples for scanning electron microscopy (SEM) were multichannel analyzer.The spectrawerecalibrated against
prepared inside the anoxic chamber by loading dried, a room temperature spectrum of alpha-Fe metal foil.
crushed and homogenized samples on a carbon film that Folding and fitting of the spectra was performed using
wasmountedonastandardaluminumstubandloadedinto Recoil software suite (University of Ottawa, Canada). All
thescanningelectronmicroscope(FEIXLF-30-FEG).The models except vivianite (5.5K) were modeled using Voigt
energy was set to 10–20keV and all images were captured based spectral lines.
inscanningmode.Thespecimensfortransmissionelectron
microscopy(TEM)examinationwerepreparedoncarbon- 2.4. Abiotic uranium reduction
coatedcoppergrids(QuantifoilMicroTools,GmbHJena)
and allowed to dry. The composition, structure and mor- ThemolarFe:Uratiowasvariedsystematicallybymod-
phology of particles were examined by high-resolution ifying the volume of magnetite suspension of known con-
transmission electron microscopy (HRTEM), scanning centration transferred to tubes containing 40ml of anoxic
transmission electron microscopy (STEM) imaging and MilliQ water. This ratio was varied to determine its effect
X-ray energy dispersive spectrometry (EDS) chemical on uranium reduction and speciation. The nomenclature
microanalysis (INCA, Oxford) in a FEI CM300UT FEG of samples in these experiments is indicated as BioMagX
transmission electron microscope (300kV field emission or BioVivX, where X is the concentration ratio of iron to
gun,0.65mmsphericalaberration,and0.17-nmresolution uranium. For example, BioMag50 refers to an experiment
atScherzerdefocus).TheimageswererecordedonaGatan involving 50mM biogenic magnetite that was incubated
797 slow scan CCD camera with a 1024(cid:4)1024pixels/14 with1000lMU(VI).TheratioofFe:Uincaseofbiogenic
bit detector and processed with the Gatan S5 Digital vivianite was set to 50:1 due to limitations in producing
Micrograph 3.11.0 software including Fourier filtering. substantial quantities of the mineral. All the resulting sus-
Low dose illumination conditions were used to record the pensions were amended with 20mM (final concentration)
images in order to prevent sintering of particles under the anoxic PIPES buffer set to pH 7 (AppliChem GmbH,
electron beam. The phase composition was determined by Germany). Parallel controlmixtures withpasteurized min-
analyzing selected area electron diffraction (SAED) pat- erals (80(cid:3)C for 20min) were carried out in the presence
terns and fast Fourier transforms of HRTEM images on and absence of bicarbonate (1mM) in a similarly buffered
the micrometer and nanometer scale, respectively. The matrix. Uranium reduction was initiated by amending the
interpretationofHRTEMimages,SAEDpatternsanddif- suspensionswith1mMU(VI)fromasterilestocksolution
fractograms were performed according to the method de- of 20mM uranyl acetate (Sigma–Aldrich GmbH). To
scribed by Veeramani et al. (2009) using the known investigate the effects of inorganic phosphate on uranium
electron-optical parameters and crystallographic data for reduction and speciation, 463mg of biogenic magnetite
several phases containing iron or uranium (ICSD 2003). (50mM total Fe) was equilibrated in 40ml of 100mM
Briefly, the interpretation of electron diffraction patterns dihydrogen sodium phosphate overnight. The concentra-
was based on the spacing of the diffraction rings. Electron tion of phosphate in the supernatant was analyzed by
diffraction patterns were indexed by the standard ratio inductively coupled plasma optical emission spectroscopy
method, in which sample patterns were compared to pat- (ICP-OES)beforeandafterequilibration.Followingasin-
terns for specific crystal structure using the JEMS (EMS gleanoxicMilliQwaterwashofapelletofmagnetiteonto
Java version) software package (Stadelmann, 1987). The which phosphate was adsorbed, uranium reduction was
usualaccuracy of thecalibration wasca. 1–2%. performed as mentioned above. All experiments were car-
The size of the biogenic Fe(II) phase particles was ried out in duplicates at 25±1(cid:3)C under strictly anoxic
estimated using Fourier filtered HRTEM images of conditions in the dark except for short-term exposure to
2516 H.Veeramanietal./GeochimicaetCosmochimicaActa75(2011)2512–2528
light during sampling. Two samples (0.5ml each) were
withdrawn at every time point from each anoxic bottle.
One sample set was filtered through a syringe filter
(0.2lm PES), amended with oxic 1% HNO and analyzed
3
for aqueous U(VI) using a Kinetic phosphorescence ana-
lyzer (KPA) (Chemchek Instruments, Inc.). This measure-
ment targeted the disappearance of hexavalent uranium
from solution. The second sample set was treated with an
anoxicsolutionof0.5Mbicarbonate(finalconcentration),
stored at 25(cid:3)C overnight, filtered anoxically through a
0.2lm pore size filter and analyzed using the KPA as
above. This sample treatment procedure enabled preferen-
tial de-sorption of U(VI) from the mineral surface (due to
formationofuranylcarbonatecomplexes)andtheanalysis
ofthebicarbonateextractrevealedtheamountofadsorbed
uranylspecies.TheamountofU(VI)reducedcouldbecal- Fig. 1. (h) HFO reduction by S. putrefaciens, (O) Fe(III)-citrate
culatedbysubtractingtheamountofU(VI)recoveredfrom reduction by S. putrefaciens, (D) Fe(III)-citrate reduction by G.
the totalamount ofU associated withthe solid phase. sulfurreducens.
2.5.Biological U(VI) reduction
citrateandG.sulfurreducensculturesamendedwithFe(III)
citrate. However, the rates of Fe(III) reduction differed
Batch enzymatic U(VI) reduction was carried out by
(Fig. 1). Whilethe reduction of Fe(III)-citrate by S.putre-
S. putrefaciens CN32 and G. sulfurreducens under non-
faciens reached a plateau in 24h, HFO reduction by the
growth conditions. These experiments were performed in
samebacteriumoccurredoverthecourseof9days.G.sul-
asimplechemicalmatrixaccordingtothemethoddescribed
furreducens reduced Fe(III) citrate in about 50h. Fe(III)
bySharp etal. (2009).
reductionbyallcultureswasassumedtobecompletewhen
steady-stateFe(II)concentrationasmeasuredbytheferro-
2.6.X-ray absorptionspectroscopy(XAS)
zine assay was reached. HFO reduction by S. putrefaciens
CN32 resulted in a black precipitate while reduction of
Samples were prepared and shipped via express over-
Fe(III)-citratebybothculturesinthepresenceofphosphate
nightcourierservicetoSSRL(StanfordSynchrotronRadi-
produced a grayish whiteprecipitate.
ation Light-source) for XAS analysis in septum bottles to
maintainahermeticseal.Priortoshipping,septum-bottles
containing the samples were placed inside of a gas-tight 3.2. Characterization of biogenic Fe(II)phases
stainlesssteelanoxicjar(SchuttbiotechGmbH,Go¨ttingen,
Germany) inside the anoxic chamber. XAS analysis in- 3.2.1. BET
cluded X-ray absorption near edge structure (XANES) Surface area analysis indicated that the specific surface
and extended X-ray absorption near edge structure (EX- area (SSA) of biogenic magnetite (53.63±2.68m2/g) was
AFS). All sample manipulation at SSRL was performed approximately eight times greater than that of biogenic
under an anoxic atmosphere (2-5% hydrogen, balance vivianite(7.66±0.38m2/g)(bothoriginatingfromS.putre-
nitrogen).CentrifugedwetsampleswereloadedinAlsam- faciens).TheSSAofbiogenicvivianiteproducedbyG.sul-
ple holders with Kapton windows. Anoxic samples were furreducens was not determined due to difficulties in
stored wet until analysis. U L -edge transmission spectra producing asufficient amountof the mineral foranalysis.
III
at liquid nitrogen temperature (77K) were collected at
SSRL beamlines 4-1, using a Si (220) double-crystal 3.2.2. XRD
monochromators. Beam line resolution was controlled by Qualitative background-subtracted XRD analysis
vertical beam dimensions to insure that the energy resolu- (Fig. 2) confirmed the resultant crystalline biogenic
tion was smaller than the intrinsic U L -edge line width. Fe(II)-bearing phases (originating from S. putrefaciens) to
III
EXAFS spectra were background subtracted, splined and be magnetite and vivianite. Similarly the resultant Fe(II)
analyzed using the Athena program (Ravel and Newville, phase produced by G. sulfurreducens was confirmed to be
2005). Backscattering phase and amplitude functions re- vivianite. XRD analyses did not show indications of sec-
quired for fitting of spectra were obtained from FEFF8 ondary phases.
(Rehretal., 1992).
3.2.3. Chemicalcharacterization
3.RESULTS An aliquot of HFO was digested with concentrated ni-
tric acid prior to reduction and analyzed for total iron.
3.1.Fe(III) reduction Comparing the amount of total iron to the steady-state
Fe(II) concentration following HFO reduction yielded an
Iron reduction was observed in S. putrefaciens cultures Fe(II):Fe ratio of 0.3±0.02 for magnetite, consistent
Total
amended with hydrous ferric oxide (HFO) and Fe(III) with thefindings of Ilton etal. (2010).
AbioticU(VI)reductionbybiogenicmagnetiteandvivianite 2517
oxidized quickly upon brief exposure to air. In addition,
electronmicroscopydidnotrevealthepresenceofbacteria
in the biogenic magnetite and biogenic vivianite samples,
suggesting efficient separation of bacteria from the bio-
genic minerals during the washing procedure.
3.2.5. Mo¨ssbauer spectroscopy
Mo¨ssbaueranalysisofbiogenicmagnetiteat245,77and
5K confirmed that the most abundant iron phase in the
sample was magnetite (about 48% of the total iron)
(Fig.4).However,thepeaksofthespectrawerebroadened
compared to highly crystalline synthetic magnetite. The
peak broadening indicates small particle sizes in the sub-
micron range. The Fe(II):Fe ratio of the magnetite
Total
phase was determined as 0.28 (±1.5%) consistent with the
wet chemistry analysis as described in Section 3.2.3 above
(Table SC-2).
Besidesmagnetite,apoorlycrystalline,unidentifiediron
phaseexhibitingastronglybroadenedmagneticsextetwith
a very low hyperfine field (H) of 34.8T was detected at
245K. For this phase, it was not possible to resolve the
two sub models for iron at tetrahedral and octahedral lat-
ticesitestypicalformagnetite.Similarresultshavebeenre-
ported by Goya et al. (2003) and Corr et al. (2004) for
synthetic, nanoparticulate (<10nm) magnetite. However,
thesmallHisnotuniquetonano-magnetiteandcouldalso
correspondtoothersmallparticlesizesand/orphaseswith
lowcrystallinity,consistentwithnano-goethitephasesasre-
ported by Van der Zee et al. (2003) and Thompson et al.
(2006)and.Sideritewasdetectedasthethirdmineralphase
in the sample(ca. 10%).
Fig.2. XRD–BiogenicMagnetiteandVivianite.Thedifferencein
background between two vivianite samples could be due to the Measurements at 77K also indicated an unidentified
extreme reactivity of the Shewanella produced vivianite. All nanoparticulate iron species that could not be resolved.
sampleswereanalyzedunderidenticalconditions. The hyperfine field corresponding to the iron phase in-
creases to 39.T, again consistent with nanoparticulate
phases. The magnetite phase could be confirmed at 77K.
3.2.4.Electron microscopy The iron(II) phase in the 77K spectrum could be inter-
Electron microscopy revealed that biogenic magnetite preted as siderite, consistent with the analysis at 245K.
was agglomerated into micron and submicron-sized The spectra at 77K spectra also showed broader peaks
clumps comprised of individual particles of about 10nm compared to those of highly crystalline samples, typical
in diameter (Fig. 3). Selected area electron diffraction for small particle sizes that have previously been referred
(SAED) of biogenic magnetite was found to be consistent toas‘collapsedsextet’(VanderZeeetal.,2003;Thompson
with the structure of magnetite. Electron diffraction pat- et al.,2006).
terns taken from agglomerates showed traces of goethite. At5K,theHfieldofthenanoparticulateFe(III)phase
ElectronmicroscopyofbiogenicvivianitefromS.putrefac- increasestovaluesintherangeofmagnetite(36–52T)and
iens CN32 (Fig. 3) revealed an average particle size in the cannot be resolved due to overlapping features. All peaks
sub-micron to micron range. The Fe/P ratio was close to arebroad,whichistypicalforsmallparticles.Sincesiderite
1.5 in all EDS spectra suggesting a stoichiometric formula splitsintoeightpeaksat5K,thepeakintensitybecametoo
of Fe (PO ) . Although rings in the electron diffraction small to bemodeled.
3 42
pattern were broad, the diffuse ring close to the center is Weconcludethatthebiogenic magnetitephaseisarel-
consistentwithvivianite.Thebiogenicvivianitewasshown atively complex mixture dominated by magnetite ((cid:5)50%)
to possess two different morphologies suggestive of phos- but including significant contributions from other nano
phoferrite co-existing with vivianite. Phosphoferrite crystalline iron phases (such as goethite as indicated by
[Fe++ (PO ) (cid:2)3(H O)] is chemically similar to vivianite electron microscopy) and some contribution ((cid:5)10%) from
3 42 2
[Fe++ (PO ) (cid:2)8(H O)] but with a different hydration state. siderite.
3 42 2
Both, biogenic vivianite and phosphoferrite exhibit similar Mo¨ssbaueranalysesofbiogenicvivianite(Fig.4)at245,
symmetry and diffraction patterns. This was consistent and 77 and 5K spectra were found to match well with
with X-ray powder diffraction data (Fig. 2) that indicated Fe(II)-phosphate references. Measurements carried out at
peaks corresponding to vivianite without secondary 245and77KindicatedasinglephasefeaturingtwoFe(II)
phases. The biogenic vivianite was extremely reactive and lattice sites, suggesting homogeneity in the sample.
2518 H.Veeramanietal./GeochimicaetCosmochimicaActa75(2011)2512–2528
Fig. 3. (A–D) S. putrefaciens-derived biogenic magnetite: secondary electron SEM, dark-field STEM, HRTEM and SAED, respectively;
(E–H)S.putrefaciens-derivedbiogenicvivianite:secondaryelectronSEM,dark-fieldSTEM,darkfieldSTEMandSAED,respectively;(I–L)
G.sulfurreducens-derivedbiogenicvivianite:secondaryelectronSEM,darkfieldSTEM,dark-fieldSTEMandSAED,respectively.
Fig. 4. Mossbauer spectra at various temperatures showing biogenic magnetite and vivianite, both derived from S. putrefaciens. (Data
representedbyopencirclesandmodelrepresentedbygreysolidlines).
Mo¨ssbauer analysis of the biogenic vivianite sample sug- Thespectradonotshowthetypicalasymmetryofvivianite
gested the material to be at least partially dehydrated. spectra(Forsythetal.,1970;MattievichandDanon,1977;
AbioticU(VI)reductionbybiogenicmagnetiteandvivianite 2519
McCammon and Burns, 1980). The isomer shift (CS) and
quadrupole splitting values provided satisfactory fits for
phosphoferrite, [Fe (PO ) (cid:2)3H O] (Mattievich and Danon,
3 42 2
1977). The large quadrupole splitting of >3mm/s of one
of the doublets, however, is inferred to be a strongindica-
tion of vivianite(TableSC-3).
The peaks in the 5K spectrum matched vivianite refer-
ences well in general, although peak intensities were
variable and the entire spectrum included a strong back-
ground(Fig.4).Thisissuggestiveofthepresenceofasingle
nanoparticulate phase. Due to the incomplete splitting,
the spectrum obtained at 5K could not be modeled
quantitatively.
3.3.Uranium reduction
U(VI)amendedtosuspensionsofbiogenicmagnetiteor Fig. 5. Surface area normalized reduction of U(VI) by ( )
vivianitedisappearedfromsolutionrapidly(1–4h),suggest- BioMag100+1mM bicarbonate, ( ) BioMag100 without bicar-
ing a rapid sorption process but not necessarily reduction bonate,(j)BioMag50+1mMbicarbonate,( )BioMag50with-
(Fig. SC-1). To probe the extent of mineral-associated U out bicarbonate ( ) Pasteurized BioMag50+1mM bicarbonate,
( ) Pasteurized BioMag50 without bicarbonate, ( ) BioMag20
reduction, samples from the suspension were treated with
without bicarbonate ( ) BioMag10+1mM bicarbonate, ( )
bicarbonate which recovered U(VI) but not U(IV) from
PasteurizedBioMag10+1mMbicarbonate.
thesurface.Thiswasconfirmedbykineticphosphorescence
analysis(KPA)thatisspecificforhexavalenturanium.The
rate of uranium reduction increased with increasing Fe:U 5and6).AdsorbedU(VI)decreasedovertimeasthereduc-
ratiointhemagnetitesystem(Table1,Fig.SC-2).Inaddi- tionofsorbedU(VI)tosurface-associatedU(IV)proceeded.
tion,reductionwasobservedinbothpasteurizedandunpas- Therateofreductionfollowedafirst-orderreactionmodel
teurized suspensions (Figs. 5 and 6), suggesting an abiotic (Figs. SC-2 and SC-3), which was used to determine rate
process. Graphs evidencing U(VI) reduction were plotted constantsandhalf-livesofthereaction(Table1).Formag-
asmicromolesofsorbedU(VI)perm2ofsurfacearea(Figs. netite at low Fe:U ratios (10:1 and 20:1), reduction rates
Table1
Summary of experimental conditions, sample descriptions and rates of abiotic uranium reduction by biogenic magnetite and vivianite.
BioMag100,BioMag50,BioMag20,BioMag10andBioMag2refertobiogenicmagnetiteincubatedwithU(VI)ataFe:Uratioof100:1,50:1,
20:1,10:1and2:1,respectively.BioViv50referstobiogenicvivianiteincubatedwithU(VI)ataFe:Uratioof50:1.Samplemarkedwithan
asteriskindicatethepresenceof1mMbicarbonate.+Extractedwith100mMbicarbonate.ND-Notdetermined.pHwasverifiedbeforeand
aftertheexperiment(12h)andwasfoundtodriftinsignificantly(±0.02).
Sample Solids Molarratio Bicarbonate Effectivesurface K Halflife r2
obs
(mg) (Fe:U) conc(M) area(m2) (h(cid:3)1) (h)
Biogenicmagnetite
Unpasteurized
BioMag100* 926 100:1 0.001 49.29 2.52 0.27 0.99
BioMag100 926 100:1 0 49.29 3.28 0.21 0.99
*
BioMag50 463 50:1 0.001 24.64 0.98 0.70 0.99
BioMag50 463 50:1 0 24.64 0.87 0.79 1.00
BioMag20 185.23 20:1 0 9.85 0.69 0.99 0.98
BioMag10 92.61 10:1 0 4.92 0.33 2.05 0.91
*
BioMag2 18.52 2:1 0.001 0.98 0.14 4.95 0.96
Pasteurized
BioMag50 463 50:1 0 24.64 0.58 1.17 0.99
*
BioMag50 463 50:1 0.001 24.64 0.77 0.89 0.99
*
BioMag10 92.61 10:1 0.001 4.92 0.81 0.85 0.96
BioMag50+P* 436 50:1 0.001 24.64 0.38f 1.81 0.98
Biogenicvivianite
BioViv50(ShewanellaPasteurized) 250 50:1 0.001 1.90 0.19f 3.64 0.98
BioViv50(Shewanella) 250 50:1 0.001 1.90 0.11f 6.30 0.95
BioViv50(Geobacter) 250 50:1 0.001 ND ND ND ND
ChemogenicFe(II)
Phosphate
ChemoViv50 250 50:1 0.001 ND 2.02 0.34 0.99
2520 H.Veeramanietal./GeochimicaetCosmochimicaActa75(2011)2512–2528
Finally,magnetitetowhichphosphatewaspre-sorbedat
a surface coverage of (cid:5)9lmol/m2, also reduced U(VI)
(Fig.6)buttherateofreductionwasintermediatebetween
thatofuntreatedmagnetiteandthatofvivianite(Table1).
Similar to biogenic vivianite, extractions were carried out
using 0.1M bicarbonate because treatment with 0.5M
bicarbonate lead to extraction of total uranium (data not
shown).
Consistentwithourpriorwork(Sharpetal.,2009)iron-
freebatchcellsuspensionsofS.putrefaciensandG.sulfur-
reducens completely reduced U(VI) to UO in a matter of
2
hours (Figs.SC-4).
3.4. X-rayabsorption spectroscopy(XAS)
Fig. 6. Surface area normalized reduction of U(VI) by (d)
XANES analyses and linear combination fitting using
PasteurizedBioViv50(S.putrefaciens),(U)BioViv50(S.putrefac-
iens)and(N)BioMag50+phosphate.Allreductionsperformedin U(VI) and U(IV) standards indicated the oxidation state
of uranium to be tetravalent in magnetite samples with a
thepresenceof1mMbicarbonate.
Fe :U ratio>10:1 (Fig. SC-5). The pasteurized sample
total
Mag10-Pas+Bicarb (Fe :U ratio 10:1) displayed a lag
total
wereconsiderablylowerthanforhigherFe:Uratios(Table priortouraniumreduction,buteventuallyunderwentnear
1), and reduction was initiated after a 3-day lag (Figure complete uranium reduction. This is consistent with lower
ure5). In experiments containing a 2:1 molar ratio of Fe:U ratios having lower reduction capacities and leading
Fe:U, reduction did not take place (data not shown), sug- toincomplete(10:1)orminimal(2:1)U(VI)reduction.Ura-
gesting insufficient reducing capacity to mediate the niumL -edgeEXAFSwasusedtocharacterizetheshort-
III
reduction. range(<4A˚)andintermediate-range(4(cid:3)10A˚)structureof
Similarly, uranium adsorption onto vivianite (originat- thereduceduraniumprecipitates(Fig.7).Allspectrashow
ingfromS.putrefaciensorG.sulfurreducens)wasveryrapid generalsalientfeaturesexpectedforuraninite,mostnotably
(Fig. SC-1). The complete reduction of sorbed U(VI) oc- an U–O shell at (cid:5)2.34A˚ (Table 2) and a U–U shell at
curredoveracourseof8daysasevidencedbythedecrease (cid:5)3.83A˚ (corresponding to the Fourier Transform (FT)
in sorbed U(VI) via conversion to U(IV) over time (Fig. 6 peaks at 1.8 and 3.8A˚, R+dR). In addition to these fea-
and SC-3). Uranium sorption and subsequent reduction tures, FT peaks are present in the sample data for
was observed in both pasteurized andunpasteurized vivia- R>4A˚ (terminating at about A˚), indicating the presence
nitesuspensions,suggestingadominantabioticredoxpro- ofsignificantintermediate-rangestructuralorder(Schofield
cess. Unlike biogenic magnetite, treatment of biogenic etal.,2008).Areductionintheexpectedcoordinationnum-
vivianite with 0.5M bicarbonate led to extraction of total ber (CN) for the U–U shell at 3.8A˚ was observed
uranium (data not shown). Hence all biogenic vivianite (CN=5.7–6.4 atoms) as compared to the bulk value of
samples were treated with 0.1M bicarbonate which selec- 12. A typical structurally ordered stoichiometric uraninite
tively removed U(VI) (confirmed by kinetic phosphores- crystal structure has a U–U coordination number (CN) of
cence analysis). 12 at approximately 3.8A˚. For small uraninite particles,
Table2
Shell-by-shell EXAFS fits of all biogenic magnetite samples containing UO and biogenic UO produced by S. putrefaciens and G.
2 2
sulfurreducens.Valueswithinparenthesisindicateuncertainty.Valuesinparenthesesarefitderived1sigmauncertaintiesinthelastreported
digit.AllthesamplesabovealsoincludedtwoadditionalU–Ushellsatdistancesfitto6.66(3)and7.84(6)A˚;coordinationnumbersfixedat24
and 12 and the Debye–Waller factor fixed at 0.008 and 0.004, respectively. These fixed values were derived from a model developed by
Schofieldetal.(2008)andallowtheestimationoftheorderedparticlecorediameteratabout1.5nm.
U–O U–U U–U
Samples N R(A˚) r2A˚2 N R(A˚) r2A˚2 N R(A˚) r2A˚2
Unpasteurized
Mag100+bicarb 8(1) 2.342(9) 0.010(2) 5(1) 3.832(7) 0.004(1) 6 5.72(3) 0.006
Mag100 7(1) 2.34(1) 0.009(2) 6(1) 3.835(7) 0.004(1) 6 5.75(3) 0.006
Mag50+bicarb 6.4(9) 2.335(8) 0.007(1) 6(1) 3.830(7) 0.005(1) 6 5.38(3) 0.006
Mag20+bicarb 8(1) 2.343(8) 0.010(2) 5.1(9) 3.831(6) 0.0033(8) 6 5.73(3) 0.006
BioUO 8(1) 2.339(8) 0.011(1) 4(1) 3.833(7) 0.005(1) 6 5.46(2) 0.006
2
Pasteurized
Mag10-Pas+bicarb 7(1) 2.337(9) 0.009(1) 6(1) 3.829(6) 0.0041(9) 6 5.72(3) 0.006
Mag50-Pas+bicarb 7(1) 2.338(8) 0.009(2) 5(1) 3.831(6) 0.0033(9) 6 5.72(3) 0.006
Mag50-Pas 7(1) 2.339(8) 0.009(1) 5.2(9) 3.831(6) 0.0034(9) 6 5.72(3) 0.006
AbioticU(VI)reductionbybiogenicmagnetiteandvivianite 2521
thenumberofUatomsboundatthesurfaceoftheparticle detectionlimitbeingaround10%UO (2)completedesorp-
2
becomescomparabletothenumberofUatomswithinthe tionofU(IV)byahighconcentrationsolutionofbicarbon-
interioroftheparticle,resultinginadecreaseintheaverage ate(datanotshown)and(3)noevidenceofaprecipitateby
U–UCN(4.1–6.1),whichisconsistentwithEXAFSinthe electronmicroscopy.Wecanthusconcludethatthemono-
present study. This under-coordination is consistent with mericU(IV)mostlylikelyoccursassurfacecomplexedspe-
thenanoparticulatesizeandconsequentialhighproportion cies on vivianitesurfaces.
ofsurface-exposedatoms(O’Loughlinetal.,2003;Schofield Phosphate is a structural component of vivianite and
etal.,2008;Veeramanietal.;2009;Sharpetal.,2009). shell-by-shell fits indicate the coordination of tetravalent
ThesameEXAFSresultswereobservedforbothunpas- uranium with two phosphorus shells at about (cid:5)3.1A˚ and
teurized and pasteurized magnetite samples, confirming another P shell appears to be present at (cid:5)3.8A˚ (Table 3).
that an abiotic process leads to uraninite formation. EX- Attempts at fitting the data using Fe atoms (at 2.98 and
AFS analysis of the magnetite-containing samples in 3.16A˚)didnotconverge(orconvergedonnegativecoordi-
Fig.7–withtheexceptionofthelowFe :U(2:1)sample– nationnumbers).Thus,U(IV)appearstobecoordinatedto
total
indicate the reduced uranium to be dominantly uraninite phosphateratherthantoFe(II).AU(IV)-carbonatecoordi-
with intermediate range structure at >4A˚. Similarly, sys- nationwasruledoutduetotheabsenceofthecharacteristic
tematicvariationoftheFe:Uratiodidnotaffecttheprod- multiple scattering feature of the distal oxygen atom
uct as characterized by EXAFS (Fig. 7). Electron (Hennigetal.,2010).Attemptsatcarryingouthigh-resolu-
microscopy of the magnetite after uranium reduction also tion electron microscopy and SAED on biogenic vivianite
revealedtheformationofuraniniteasconfirmedbyselected containing reduced uranium were unsuccessful since the
areaelectrondiffraction (SAED) (Fig. SC-6). uranium species was destroyed quickly when exposed to
Incontrasttothemagnetiteexperiments,theproductof the electron beam(Fig. SC-6).
U(VI) reduction by vivianite was a non-uraninite, mono- Toinvestigatetheroleofphosphateintheformationof
meric uranium species. While XANES analysis confirmed monomeric U(IV) species, uranium reduction was per-
complete reduction of uranium (Fig. SC-7), EXAFS did formed with biogenic magnetite that was pre-sorbed with
notdisplaytheU–UFouriertransformpeakat3.86A˚ typ- phosphate. The rate of U(VI) reduction was slower for
icalofuraninite(Fig.8).HadaU–Upaircorrelationbeen phosphate-treated magnetite when compared to untreated
presentinthissample,thispeakwouldhaveappearedinthe magnetite,presumablyduetosurface-associatedphosphate
FTdespitetheshorterdatarangeascomparedtothemag- blocking transfer of electrons from Fe(II) to U(VI). The
netite samples. The absence of a U–U peak indicates the phosphate-treated magnetite also produced monomeric
formation of a monomeric form of U(IV), which is likely U(IV) as evidencedbyXAS analysis (Fig. 8).The product
to be a molecular species complexed by surface ligands. wasalmostidenticaltothatobtainedinthebiogenicvivia-
The formation of a precipitate is highly unlikely based on nite experiments: it lacked the typical spectral signature of
(1)theEXAFSspectranotshowingtheU–UFouriertrans- uraninite and was best fit with one or two U-P shells at
form peak at 3.86A˚ typical of uraninite and the EXAFS (cid:5)3.1A˚ and (cid:5)3.8A˚, respectively (Table 3).
Fig.7. (A)UL EdgeEXAFSspectraforUO producedbypasteurizedandunpasteurizedmagnetitesamples,S.putrefaciens(S-BioUO)
III 2 2
andG.sulfurreducens(G-BioUO)inabsenceofFeB)CorrespondingFouriertransforms.Dataareindicatedbysolidlinesandfits(shownin
2
Table 2) by dashed lines. Samples marked with an asterisk indicate the presence of 1mM bicarbonate. The small FT peak at R(cid:5)1.5A˚
(R+dR)isknowntoresultfromthepresenceofamulti-electronexcitationatk=10.2A˚(cid:3)1inallspectra.