Reduction of Hg(II) to Hg(0) byMagnetiteH E A T H E R A . W I A T R O W S K I , † , |

S O U M Y A D A S , ‡ , ⊥ R A V I K U K K A D A P U , §

E U G E N E S . I L T O N , § T A M A R B A R K A Y , †

A N D N A T H A N Y E E * , ‡

Department of Biochemistry and Microbiology, RutgersUniversity, New Brunswick, New Jersey, Department ofEnvironmental Sciences, Rutgers University, New Brunswick,New Jersey, and Pacific Northwest National Laboratory,Richland, Washington

Received February 4, 2009. Revised manuscript receivedMay 8, 2009. Accepted May 20, 2009.

Mercury (Hg) is a highly toxic element, and its contaminationof groundwater presents a significant threat to terrestrialecosystems. Understanding the geochemical processes thatmediate mercury transformations in the subsurface is necessaryto predict its fate and transport. In this study, we investigatedthe redox transformation of mercuric Hg (Hg[II]) in thepresence of the Fe(II)/Fe(III) mixed valence iron oxide mineralmagnetite. Kinetic and spectroscopic experiments wereperformed to elucidate reaction rates and mechanisms. Theexperimental data demonstrated that reaction of Hg(II)with magnetite resulted in the loss of Hg(II) and the formationof volatile elemental Hg (Hg[0]). Kinetic experiments showedthat Hg(II) reduction occurred within minutes, with reaction ratesincreasing with increasing magnetite surface area (0.5 to 2 m2/L) and solution pH (4.8 to 6.7), and decreasing with increasingchloride concentration (10-6 to 10-2 mol/L). Mossbauerspectroscopic analysis of reacted magnetite samples revealeda decrease in Fe(II) content, corresponding to the oxidationof Fe(II) to Fe(III) in the magnetite structure. X-ray photoelectronspectroscopy detected the presence of Hg(II) on magnetitesurfaces, implying that adsorption is involved in the electrontransfer process. These results suggest that Hg(II) reaction withsolid-phase Fe(II) is a kinetically favorable pathway for Hg(II)reduction in magnetite-bearing environmental systems.

IntroductionIn the United States, mercury (Hg) associated with mixedwaste generated by nuclear weapons manufacturing hascontaminated vast areas of soil and groundwater (1, 2).Mercury released from spills and waste disposal typicallyenters the subsurface as inorganic mercury. In anoxicsediments, mercuric Hg (Hg[II]) can be subsequently con-verted into the neurotoxic substance methylmercury (MeHg)by anaerobic bacteria (3-5). Elevated levels of MeHg havebeen shown to accumulate in fish inhabiting surface waters

receiving hydrologic inputs from mercury-contaminated sites(6-8). Critical to understanding the formation of methylm-ercury is an accurate knowledge of the chemical reactionsthat proceed as inorganic mercury moves from contaminantsources to anoxic methylation zones.

The fate of Hg(II) in soil and groundwater is stronglyinfluenced by complexation and redox reactions (9). Hg(II)complexation with organic matter and mineral surfaces canretard its subsurface migration (10, 11). Oxide minerals inparticular have been found to be efficient sorbents of Hg(II)(12-15). The adsorption of Hg(II) onto iron and aluminumoxide surfaces has been studied extensively, with the extentof adsorption varying as a function of mineral surface area,pH, and chloride concentrations. Iron oxides such asferrihydrite and goethite are known to adsorb Hg(II) ionsabove pH 4 (16-18). For example, in ferrihydrite suspensions,Hg(II) adsorption increases from near zero at pH 3 to greaterthan 90% at pH 5 (17). The mechanism for this process isattributed to surface complexation of Hg(II) ions with surfacehydroxyl functional groups at the mineral-water interface(19). X-ray absorption spectroscopy has shown that thedominant mode of Hg(II) adsorption to goethite occurs bythe formation of monodentate and bidentate inner spheresurface complexes (18). At high pH, the extent of adsorptiononto iron oxide surfaces decreases due to pH dependenceof Hg(II) hydrolysis (16). The presence of chloride also inhibitsadsorption due to the formation of nonsorbing mercurychloride complexes (13, 17, 20).

The redox transformation of Hg(II) to Hg(0) significantlyalters the fate of mercury in soil and groundwater. Due toits low solubility in water and high volatility, Hg(0) readilypartitions to the gas phase in the vadose zone (21, 22). Lossof gaseous Hg(0) to the atmosphere decreases the amountof mercury remaining for groundwater transport and limitsthe concentration of Hg(II) available for methylation. Ingroundwater aquifers where gas exchange is restricted, Hg(0)may become supersaturated and mobilized to drinking watersources (23). Important chemical reductants of Hg(II) includedissolved organic carbon and sorbed/precipitated ferrousiron. Alberts et al. (24) and Allard and Arsenie (25) demon-strated that natural organic matter such as humic and fulvicacids can reduce Hg(II). Bacteria can promote mercuryreduction by catalyzing electron transfer from an electrondonor to Hg(II) (26-28). Mineral-associated ferrous iron hasbeen identified as another possible reductant for Hg(II).Charlet et al. (29) reported the reduction of Hg(II) to Hg(0)by Fe(II) adsorbed onto phlogopite surfaces, and O’Loughlinet al. (30) observed reduction of Hg(II) by Fe(II)-containingmineral hydroxysulfate green rust.

In anoxic groundwater, ferrous iron often accumulates insoils and sediments as the iron oxide mineral magnetite(Fe3O4). Magnetite is a mixed-valence iron oxide that containsboth Fe(II) and Fe(III) ions in an inverse spinel structurewith oxygen atoms in a cubic closest packing array. Whiteand Peterson showed that the Fe(II) in magnetite can reducea wide range of metal ions, including ferric iron, copper(II),vanadate, and chromate (31). Previous experimental studieshave found that magnetite can also reduce Np(V) to Np(IV)(32), Pu(V) to Pu(IV) (33), U(VI) to U(IV) (34), and Se(IV) toSe(0) (35). However, despite its affinity to reduce metalcontaminants (31-35), and its common occurrence in soilsand sediments (36), the extent to which magnetite can actas a chemical reductant for the reduction of Hg(II) is currentlyunknown.

In this study, we conducted laboratory experiments toinvestigate the interaction of Hg(II) with magnetite in

* Corresponding author e-mail: nyee@envsci.rutgers.edu.† Department of Biochemistry and Microbiology, Rutgers Univer-

sity.‡ Department of Environmental Sciences, Rutgers University.§ Pacific Northwest National Laboratory.| Department of Biology, Clark University, Worcester, Mas-

sachusetts (present address).⊥ Department of Geological Sciences, University of Saskatch-

ewan, Canada (present address).

Environ. Sci. Technol. 2009, 43, 5307–5313

10.1021/es9003608 CCC: $40.75 2009 American Chemical Society VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5307

Published on Web 06/12/2009

deoxygenated water. We present macroscopic and spectro-scopic evidence that the ferrous iron in magnetite reducesHg(II) to Hg(0). Hg(II) reduction experiments were conductedas a function of magnetite surface area, pH, and chlorideconcentration to quantify the rates of reaction. Reactedmagnetite samples were analyzed using Mossbauer and X-rayphotoelectron spectroscopy to examine solid-phase changesin Fe speciation, and to identify the Hg charge state on themagnetite surface. The results of this work suggest thatsurface-catalyzed Hg(II) reduction is a kinetically favorablepathway for the formation of Hg(0) in magnetite-bearingsoils and sediments.

Experimental SectionSynthesis and Characterization of Iron Oxides. Magnetite,goethite, and ferrihydrite were synthesized according to themethods described by Cornell and Schwertmann (36). Briefly,magnetite was prepared in an anaerobic glovebox (CoyLaboratories, Grass Lake, MI) by titrating a FeSO4 ·7H2Osolution with KOH and KNO3 at 90 °C. Ferrihydrite wassynthesized by titrating a FeCl3 solution with NaOH to pH7. Goethite was prepared by reacting ferrihydrite in a KOHsolution at 70 °C for 60 h. All precipitates were washed withdeoxygenated distilled deionized water until the supernatantexhibited a constant pH approximately equal to the pHzpc ofthe iron oxide mineral. Aliquots of the suspension werefiltered, dried under N2 atmosphere, and characterized usingX-ray powder diffraction (XRD) (Philips X’Pert diffractometer).The identities of the minerals were confirmed by comparingthe X-ray diffraction patterns to standards in the JointCommittee on Powder Diffraction Standards database.Surface area was determined with an 11-pt BET-Nitrogenisotherm (Micromeritics Gemini 2375). BET measurementswere conducted on samples outgassed at 80 °C for 24 h.

Trapping of Hg(0). Magnetite (0.2 g/L) suspended in 20mL of deoxygenated water was reacted with 48.4 ( 2.0 nMHgCl2, corresponding to a total of 0.63 ( 0.04 µg of Hg in thereactor. Experiments were conducted in foil-wrapped sealedserum bottles, and Hg(0) gas was collected continuously bypurging N2 through the reactor for 1 h into midget bubblers(Ace Glass, Vineland, NJ, catalog 75320-20) containing anHg(0) trapping solution (0.6% potassium permanganate, 2.5%sulfuric acid, 2.5% nitric acid). Samples were collected fromthe reaction vessel and trapping solution at the beginningand at the end of the experiment. Additionally, at the end ofthe experiment, the walls of the bubblers and serum bottleswere washed with concentrated acid to remove any mercurysorbed onto the glassware. Mercury was digested using 1:1concentrated sulfuric and nitric acids (trace metal grade)and 10 MΩ Milli-Q water, according to a variation of EPAmethod 245.1 (37). These samples were heated at 65 °C for2 h, incubated with 250 µL of 5% potassium permanganateat room temperature overnight, and reacted with 100 µL of12% hydroxylamine hydrochloride. Finally, samples werediluted with 2% HCl and analyzed by cold vapor atomicabsorbance spectroscopy (CVAAS) using a Leeman Labo-ratories Hydra AA (Hudson, NH). The detection limit of ourinstrument, defined as 3 times the standard deviation of 10blank samples, was 0.4 nM.

Hg(II) Reduction Kinetic Experiments. For the kineticexperiments, Hg was provided as HgCl2, with 203HgCl2 as aradioactive tracer (provided by Prof. D. Barfuss). Reactionswere performed in sealed serum bottles containing deoxy-genated water. Experiments with magnetite were conductedas a function of magnetite surface area (0.5-2 m2/L), pH(4.8-6.7), and chloride concentration (10-6 to 10-2 mol/L).Samples (0.5 mL) were removed from the sealed serum bottlesusing a needle and syringe, every 35 s for approximately 20min. The radioactivity of the unfiltered mineral suspensionwas measured to determine the amount of total Hg remaining

(e.g., Hg in solution and adsorbed to the mineral particles).203Hg analysis was performed by liquid scintillation countingusing a Beckman LS-6500 Counter (Beckman Instruments,Fullerton, CA), with EcoLume Scintillation Cocktail (ICNRadiochemicals, Irvine, CA). Counts per minute (CPM) weredetermined for 3 min using a wide channel. Mercuryreduction experiments were also performed with goethiteand ferrihydrite suspensions (0.2 g/L) as Fe(II)-free controls.Additional control experiments were conducted with deoxy-genated water in absence of iron oxide minerals. As initialCPM readings in mineral suspensions were indistinguishablefrom those from the water controls, it was assumed thatquench due to minerals in suspension was not complicatingdata collection.

57Fe Mossbauer Spectroscopy. 57Fe Mossbauer spectros-copy was performed on magnetite samples reacted with 1mM of HgCl2 for 14 days. Samples from a control experimentconducted with magnetite suspended in deoxygenated waterwithout Hg(II) for 14 days were also analyzed. All sampleswere dried in an anoxic chamber and care was taken toprevent magnetite oxidation by sample handling (38). Thesample holder was filled with the magnetite sample andsealed with transparent tape and an oxygen-impermeablepolymer film (aluminized Mylar stable to 4 K). The tape andpolymer were snapped into the holder with carbonizedpolyethyletherketone (PEEK) polymer rings to ensure tight-ness. Mossbauer spectra were collected according to theprocedure given in Kukkadapu et al. (38). A closed-cyclecryostat was used for the 125 K measurements. The Moss-bauer data were modeled with the Recoil software using aVoight-based spectral fitting routine (39). Additional detailsof the Mossbauer analysis are included in the SupportingInformation.

X-ray Photoelectron Spectroscopy. XPS analysis wasperformed on magnetite samples reacted with 0.1 mM and1 mM HgCl2. After reaction for 14 days, magnetite suspensionswere pipetted from the glass serum bottle in a glovebox at<0.1 ppm O2, and centrifuged/filtered at 4500 rpm using 30Kmolecular weight cut off Whatman VectaSpin centrifugefilters. The resulting pastes were smeared onto tantalumcoupons with a stainless steel spatula and allowed to dry.Samples were then placed in a dry seal desiccator andtransferred to a glovebag (∼35 ppm O2) attached to the XPSentry port. Exposure to atmosphere was minimized.

The XPS measurements were performed using a PhysicalElectronics Quantum 2000 Scanning ESCA Microprobe.Details of the XPS analysis are included in the SupportingInformation. Scans of the Hg4f and Cl2p regions wererecorded and the energy scale was referenced to adventitiouscarbon 1s at 285.0 eV. Because the Hg4f region is stronglyoverlapped with Fe3s, the Fe3s region was characterized forunreacted magnetite and used in the fit for the experimentalspectra. Silicon (Si) was detected and the Si2p line was alsoincluded in the fit. The Hg4f lines are characterized by twosimple spin-orbit split peaks, where the Hg4f5/2 peak is clearlyvisible but the Hg7/2 peak is buried under the Fe3s and Si2ppeaks. The Hg7/2 peak was generated using known values forthe spin orbit splitting and relative intensities of the twopeaks. The spectra were best fit by nonlinear least-squaresusing the CasaXPS curve resolution software. A spin-orbitsplitting of 4.0 eV was used. The Hg4f5/2:Hg4f7/2 intensityratio was set at 0.75, which is the ideal branching ratio. Eachspin-orbit peak was modeled using only one component,with variable but equal fwhm (full width at half-maximum).This is consistent with the closed shell electronic structureof Hg(II) and resulting lack of multiplet structure. Elementalratios were semiquantified using Scofield photoionizationcross sections.

5308 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

Results and DiscussionReduction of Hg(II) to Hg(0). The reduction of Hg(II) bymagnetite was investigated by reacting 0.2 g/L of magnetitewith 100 nM Hg(II) in deoxygenated water, using 203Hg as atracer (Table 1). After 2 h, radioactivity measurementsindicated that 79.5 ( 0.8% of the total Hg was lost from themagnetite suspension, compared to 1.0 ( 1.8% Hg loss inthe reaction vessel containing only deoxygenated water. Thereaction with magnetite was nearly complete in 2 h, asreaction for 24 h resulted in only a marginal increase in Hgloss (82.5( 0.2%). In the control experiments with the Fe(III)oxide minerals goethite and ferrihydrite, the amounts of Hglost from the suspensions after 2 h were 4.2 ( 1.5% and 9.0( 1.2%, respectively. These measurements indicate thatreaction of Hg(II) with magnetite results in loss of Hg thatdoes not occur with goethite or ferrihydrite.

A mercury trapping experiment was conducted to de-termine if the loss of mercury in the magnetite suspensionwas due to formation of volatile Hg gas. After 1 h of reactiontime, 30.8 ( 5.2% of the Hg remained in suspension with themagnetite, and 70.9 ( 16.3% was recovered in a potassiumpermanganate trap (Figure 1). For the control experimentperformed in deoxygenated water without magnetite, 110.2( 2.8% of the Hg remained in the reaction vessel and 2.5 (1.5% was recovered in the trap. These results indicate thatgaseous Hg is a product of Hg(II) reaction with magnetite.We interpret this gaseous Hg to be strong evidence for theformation of Hg(0).

Kinetic Experiments. Experiments were conducted tomeasure Hg(II) reduction rates, and to determine the effectof magnetite surface area, pH, and chloride concentrationon the rates of reaction. The data show that the rate ofreduction increases with increasing magnetite surface area

(0.5 to 2 m2/L) (Figure 2A) and pH (4.8 to 6.7) (Figure 2B),and decreases with increasing chloride concentration (10-6

to 10-2 mol/L) (Figure 2C). At a magnetite surface areaconcentration of 2 m2/L, over 80% of the Hg(II) loss occurredin less than 15 min of reaction (Figure 2A). This Hg(II)reduction rate is approximately 10 times faster than Hg(II)reduction by Fe(II) sorbed onto phlogopite (29).

As the magnetite surface site density is in large excesscompared to the Hg(II) concentration, we can assume thatthe magnetite concentration remains constant during theexperiment. Accordingly, the Hg(II) reduction kinetics canbe described using a pseudo first-order kinetic model:

where krxn is pseudo first-order rate constant, and [Hg(II)] isthe concentration of total Hg(II) remaining in the system.The reaction rate constants determined for each experimentalsystem are given in Table 2. The reaction rates predicted byeach pseudo first-order rate constant are plotted in Figure

TABLE 1. Loss of Hg from Iron Oxide Mineral Suspensions

percent Hg lost from suspension

2 h 24 h surface area (m2/g)

magnetitea 79.5 ( 0.8b 82.5 ( 0.2 10.4ferrihydrite 9.0 ( 1.2 16.2 ( 1.2 46.4goethite 4.2 ( 1.5 9.1 ( 4.2 13.6water 1.0 ( 1.8 5.7 ( 1.8 NAc

a All iron oxide minerals were suspended at aconcentration of 0.2 g/L in deoxygenated water. b Valuesrepresent means of triplicate experiments, and errorsrepresent the standard deviation of the mean. c Notapplicable.

FIGURE 1. Formation of gaseous Hg(0) during Hg(II) reactionwith magnetite. Grey bars represent mercury remaining in thereaction vessel, and white bars represent mercury in thetrapping solution. Reactions were performed under anoxicconditions in sealed 100 mL serum bottles with 70 mL of wateror magnetite suspension (0.2 g/L). The initial concentration ofHg(II) added to the reaction vessel was 48.3 ( 2.0 nM.Reactions were performed in triplicate, and error barsrepresent standard deviation.

FIGURE 2. Kinetics of Hg(II) reduction by magnetite.Experiments were conducted in deoxygenated water with 100nM of HgCl2, using 203Hg as a tracer. Symbols indicateexperimentally determined data, and lines indicate the pseudofirst-order kinetic model fit. Percent Hg(II) remaining in solutionis plotted as a function of (A) magnetite surface area (0.5-2 m2/L), (B) pH (4.8-6.7), and (C) chloride concentration (10-6 to 10-2

mol/L).

d[Hg(II)]dt

) -krxn[Hg(II)] (1)

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5309

2. The rate constants describe the overall reaction rates, whichinclude the rates of adsorption, electron transfer, andvolatilization.

Comparison between the experimental measurementsand model fits indicate that the pseudo first-order kineticmodel provides an excellent description of the Hg(II)reduction data. The magnetite, [H+], and [Cl-] reaction orderterms are reported in the Supporting Information (FigureS1).

Spectroscopic Analysis. To examine the solid-phasechanges in Fe speciation by 57Fe-specific Mossbauer spec-troscopy and the Hg surface state by surface-sensitive X-rayphotoelectron spectroscopy, magnetite samples were treatedwith higher concentrations of Hg(II) (100 µM to 1.4 mM).Mossbauer spectra were collected at room temperature and125 K to examine the purity of the product and to followoxidation of Fe(II) magnetite by Hg(II). Figure 3 shows 125K spectra of a sample that was suspended in deoxygenatedwater (Figure 3A) and a sample reacted with Hg(II) (Figure3B). The spectrum of the water-treated sample exhibited twosextet peaks with relative areas and Mossbauer parametersthat are similar to stoichiometric magnetite [Fe(II)/Fe-totalof 0.33] (40). The spectral features are also in agreement withthe absence of any impurity phases, e.g., ferrihydrite,hematite, or goethite. The outer sextet (36% area) is due toFe(III) in tetrahedral sites of the inverse spinel structure. Theinner sextet (64% area), on the other hand, is due to an averageof Fe(II) and Fe(III) or “Fe2.5+” in the octahedral sublattice,which is the result of fast electron hopping between the Fe(II)and Fe(III) sites, at temperatures above 120 K (40). Becausehalf of the Fe in the octahedral sublattice is occupied byFe(II), the fit-derived Fe(II)oct/[Fe(II,III)oct + Fe(III)tet] of 0.36is approximately a third of the total Fe, in agreement withFe charge state distribution in stoichiometric magnetite.

The redox reaction with Hg(II) resulted in significantchanges in the 57Fe Mossbauer spectrum of the magnetitesample. This is evident from different relative areas of thesextets in the water-treated and Hg(II)-treated spectra (Figure3A and B). Increase in the outer sextet area to 46% from 36%in the water-treated sample implied partial oxidation of theoctahedral Fe(II). Oxidized Fe(II) exhibits parameters similarto the tetrahedral Fe(III), hence their peaks are unresolvedfrom each other. Based on the spectral area of the Fe2.5+contribution (54%), the Fe(II) content of the Hg(II)-treatedsample was estimated to be 27% or half of the inner sextetcontribution. This change represents oxidation of ∼18% ofthe Fe(II) in the magnetite sample by ionic mercury. Featurescharacteristic of ferrihydrite or goethite were absent in theoxidized sample indicating that secondary Fe(III) oxidephases did not form.

Results of the XPS analysis are summarized in the Table3 (also see Figure S2 in Supporting Information). AdsorbedHg was detected on all samples that were reacted with HgCl2,where Hg4f binding energies (BE) are consistent with Hg(II).As expected, the Hg/Fe ratio was about 10-fold smaller for

the 0.1 mM compared to the 1 mM HgCl2 magnetiteexperiments, indicating that the amount of adsorbed Hg(II)was proportional to the amount of initial HgCl2(aq). Incontrast, the same decrease in HgCl2(aq) only achieved a2-fold decrease in surface Cl/Fe ratio. The Hg/Cl ratios forthe magnetite experiments are all far less than 0.5, the ratiofor HgCl2. This is consistent with preferential loss of total Hgfrom the system.

Hg(II) Interaction with Magnetite Surfaces. Electrontransfer at the magnetite-water interface involves directinteractions between the Hg(II) and structural ferrous ironas free electrons are not readily transferred into aqueoussolution. Adsorption sites are one location where reductionof metal ions can occur (41). The XPS data indicate that Hg(II)ions can adsorb onto magnetite surfaces. Furthermore, theMossbauer results suggest that the adsorbed Hg(II) interactswith magnetite surfaces by accepting electrons from Fe(II)in the magnetite structure. On the magnetite surface, thehalf cell potential for the solid-state oxidation reaction Fe(II)f Fe(III) + e- is approximately -0.34 to -0.65 V (31). Formercury reduction, the standard potential for the half reactionHg(II) + 2e- f Hg(0) is +0.85 V. The sum of the half cellpotentials yields a positive value, indicating that electronscan spontaneously flow from Fe(II) on magnetite surfaces toadsorbed Hg(II) ions. The reactivity of solid phase Fe(II) onmagnetite surface can be attributed to the shift in electrondensity from surface hydroxyl groups, which increases thereducing power of the Fe(II) ion (42). In comparison, thereduction of Hg(II) by aqueous Fe(II) (-0.77 V) is energeticallyless favorable, and kinetically inhibited (29).

Our experimental data demonstrate that the kinetics ofHg(II) reduction by magnetite systematically varies as afunction of magnetite concentration, pH, and chlorideconcentration. We propose that the rate of reaction iscontrolled by the chemical speciation of Hg(II) ions andmagnetite surfaces. First, the effect of magnetite concentra-tion on overall reaction rates is attributed to the increase ofmineral surface area available for interaction with Hg(II).The surface area of the magnetite particles used in ourexperiments was 10.4 m2/g (Table 1), with a surface sitedensity of approximately 3.62 × 10-5 mol/m2 (43). At theconcentration range of 0.5 to 2 m2/L, the concentration ofsurface sites increased from 1.88 × 10-5 to 7.52 × 10-5 mol/L.Second, the effect of pH on Hg(II) reduction rates can beexplained by the pH-dependent adsorption of Hg(II) ontomagnetite. Hg(II) complexation with surface hydroxyl groupsonto iron oxide surfaces occurs in the pH range of 4-7, withadsorption increasing at pH > 4 and decreasing at pH > 7(16, 17). This adsorption behavior is controlled by thedeprotonation of surface hydroxyl groups above pH 4, andthe formation of neutral Hg(OH)2 aqueous complexes abovepH 7. Between pH 4 and 7, deprotonated surface hydroxylgroups generate negative surface charge and electrostaticallyattract Hg(II) cations to adsorption sites at magnetite-water interface. Finally, the effect of chloride concentrationon Hg(II) reduction is attributed to mercury-chlorideaqueous complexation. At high chloride concentrations,Hg(II) predominately exists as mercuric chloride complexes(13, 19). The formation of stable nonsorbing aqueous mercurycomplexes limits direct contact of mercuric Hg with magnetitesurfaces, thereby hindering the electron transfer reaction.

Environmental Significance. The results presented in thisstudy show that magnetite can rapidly reduce Hg(II). Theevidence for this reaction includes the loss of Hg(II) frommagnetite suspensions, the formation of gaseous Hg(0), andthe oxidation of solid-state Fe(II). These experimentalobservations may help us understand the natural redoxtransformations of Hg in magnetite-bearing soils and sedi-ments. In oxic surface water, biologic and photoreductionmay dominate. However, because magnetite is common in

TABLE 2. Pseudo First-Order Reaction Rate Constants: Hg(II)Reduction by Magnetite at Varying pH, Magnetite SurfaceArea, and Chloride Concentrations

pH [Fe3O4] (m2/L) [Cl-] (mol/L) rate constant ((2.0 × 10-4) (s-1)

6.73 ( 0.08 2.08 ( 0.11 - 1.6 × 10-3

6.73 ( 0.08 1.04 ( 0.05 - 9.0 × 10-4

6.73 ( 0.08 0.52 ( 0.04 - 4.0 × 10-4

4.77 ( 0.23 2.08 ( 0.11 - 3.0 × 10-4

6.05 ( 0.15 2.08 ( 0.11 - 9.0 × 10-4

6.63 ( 0.11 2.08 ( 0.11 1 × 10-2 1.0 × 10-4

6.63 ( 0.11 2.08 ( 0.11 1 × 10-4 5.0 × 10-4

6.63 ( 0.11 2.08 ( 0.11 1 × 10-6 9.0 × 10-4

5310 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

iron-reducing sediments and waterlogged soils, the reductionof Hg(II) by solid-phase Fe(II) may occur in anoxic ground-water. This process may limit the discharge of MeHg fromhyporheic zones to surface waters (44) by controlling the

concentration of Hg(II), the substrate for mercury meth-ylation. Conversely, Hg(II) reduction in saturated sedimentsmay promote the mobilization of Hg as Hg(0). The formationof mobile Hg(0) is of particular concern as groundwater canbe a source of Hg to surface water (45) and water distributionsystems where Hg(0) is the major form of Hg (23, 46).Intriguingly, elevated Hg concentrations in groundwater havebeen correlated with high levels of Fe(II) and low redox (47).

Based on our laboratory experiments, Hg(II) reductionby magnetite in soils and sediments is expected to be facile,however the reaction rates would be strongly dependent ongroundwater composition and reactive surface area. For thereaction to proceed, the pH of the groundwater must be highenough for Hg(II) adsorption on the magnetite surface tooccur. In the presence of elevated chloride concentrations,Hg(II) adsorption is inhibited and surface-mediated reductionby magnetite is expected to be slow. Furthermore, Hg(II)reduction by magnetite is likely to be most favorable underanoxic conditions, or in soil environments where oxygendiffusion is limited. White and Peterson showed that naturalmagnetite sands weathered under reducing conditions are

FIGURE 3. 125 K 57Fe Mossbauer spectra of magnetite. (A) magnetite samples suspended in deoxygenated water for 14 days; and (B)magnetite samples reacted with 1.37 ( 0.07 mM of Hg(II) for 14 days.

TABLE 3. Elemental Ratios and Binding Energies from X-rayPhotoelectron Spectroscopy

sample Hg/Fe Cl/Fe Hg/Clbinding energiesa

Hg4f7/2 Hg4f5/2

H2O+mag. Hg NDb Hg ND Hg ND1 mM Hg+mag 0.0095 Cl NAc Cl NA 102.0 106.0

area 21 mM Hg+mag 0.012 0.25 0.049 102.0 106.0

area 40.1 mM Hg+mag 0.0008 0.13 0.0066 102.0 106.0

area 50.1 mM Hg mag 0.0015 0.11 0.014 102.1 106.1

area 6a (0.1 eV. b not detected. c not analyzed.

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5311

electrochemically active, where as magnetite particles weath-ered under oxidizing vadose conditions show minimalreactivity toward metal reduction (31). Oxidative weatheringof magnetite results in the formation of surface oxidationproducts that may impede electron transfer. Therefore, theoxidation of magnetite surfaces by oxygen is expected topacify its reactivity toward Hg(II). Other factors such asorganic matter and sulfide complexation with Hg(II) mayalso inhibit mercury reduction by magnetite through theformation of nonsorbing Hg chemical species. Additionalfield and experimental studies are required to furtherelucidate the competing pathways of Hg(II) transformationin contaminated subsurface environments.

AcknowledgmentsThis research was supported by the Office of Science (BER),U.S. Department of Energy Grant DE-FG02-08ER64544. XPSand Mossbauer experiments were performed at EMSL, anational scientific user facility sponsored by the Departmentof Energy’s Office of Biological and Environmental Researchlocated at Pacific Northwest National Laboratory.

Supporting Information AvailableThis material is available free of charge via the Internet athttp://pubs.acs.org.

Literature Cited(1) Riley, R. G.; Zachara, J. M.; Wobber, F. J. Chemical contaminants

on DOE lands and selection of contaminant mixtures forsubsurface science research; DOE/ER-0547; U.S. DOE, 1992.

(2) Campbell, K. R.; Ford, C. J.; Levine, D. A. Mercury distributionin Poplar Creek, Oak Ridge, Tennessee, USA. Environ. Toxicol.Chem. 1998, 17 (7), 1191–1198.

(3) Compeau, G. C.; Bartha, R. Sulfate-reducing bacteria: principlemethylators of mercury in anoxic estuarine sediment. Appl.Environ. Microbiol. 1985, 50, 498–502.

(4) King, J. K.; Kostka, J. E.; Frischer, M. E.; Saunders, F. M. Sufate-reducing bacteria methylate mercury at variable rates in pureculture and in marine sediments. Appl. Environ. Microbiol. 2000,66 (6), 2430–2437.

(5) Kerin, E. J.; Gilmour, C. C.; Roden, E.; Suzuki, M. T.; Coates,J. D.; Mason, R. P. Mercury methylation by dissimilatory iron-reducing bacteria. Appl. Environ. Microbiol. 2006, 72 (12), 7919–21.

(6) Southworth, G. R.; Turner, R. R.; Peterson, M. J. Form of mercuryin stream fish exposed to high-concentrations of dissolvedinorganic mercury. Chemosphere 1995, 30 (4), 779–787.

(7) Southworth, G. R.; Peterson, M. J.; Ryon, M. G. Long-termincreased bioaccumulation of mercury in largemouth bassfollows reduction of waterborne selenium. Chemosphere 2000,41 (7), 1101–5.

(8) Santoro, E. D.; Koepp, S. J. Mercury levels in organisms inproximity to an old chemical site (Berry’s Creek, HackensackMeadowlands, New Jersey, USA). Mar. Pollut. Bull. 1986, 17(5), 219–224.

(9) Gabriel, M. C.; Williamson, D. G. Principal biogeochemicalfactors affecting the speciation and transport of mercury throughthe terrestrial environment. Environ. Geochem. Health 2004, 26(4), 421–34.

(10) Melamed, R.; Boas, R. C. V. Phosphate-background electrolyteinteraction affecting the transport of mercury through a BrazilianOxisol. Sci. Total Environ. 1998, 213 (1-3), 151–156.

(11) Miretzky, P.; Bisonti, M. C.; Jardim, W. F. Sorption of mercury(II)in Amazon soils from column studies. Chemosphere 2005, 60(11), 1583–1589.

(12) Lockwood, R. A.; Chen, K. Y. Adsorption of Hg (II) by ferrichydroxide. Environ. Lett. 1974, (6), 151–166.

(13) Mac Naughton, M. G.; James, R. O. Adsorption of aqueousmercury (II) complexes on the oxide/water interface. J. ColloidInterface Sci. 1974, 46 (2), 431–440.

(14) Gunneriusson, L.; Sjorberg, S. Surface Complexation in theH+ - Goethite (a-FeOOH)-Hg(II)-Chloride System. J. ColloidInterface Sci. 1993, 156 (1), 121–128.

(15) Sarkar, D.; Essington, M. E.; Misra, K. C. Adsorption of Mercury(II)by Variable Charge Surfaces of Quartz and Gibbsite. Soil Sci.Soc. Am. J. 1999, 63 (6), 1626–1636.

(16) Barrow, N. J.; Cox, V. C. The effects of pH and chlorideconcentration on mercury sorption. I. By goethite. Eur. J. SoilSci. 1992, 43 (2), 295–304.

(17) Kinniburgh, D. G.; Jackson, M. L. Adsorption of Mercury(II) byIron Hydrous Oxide Gel. Soil Sci. Soc. Am. J. 1978, 42 (1), 45–47.

(18) Kim, C. S.; Rytuba, J. J.; Brown, G. E. EXAFS study of mercury(II)sorption to Fe- and Al-(hydr)oxides: I. Effects of pH. J. ColloidInterface Sci. 2004, 271 (1), 1–15.

(19) Tiffreau, C.; Lutzenkirchen, J.; Behra, P. Modeling the adsoprtionof mercury(II) on (hydr)oxides 0.1. amorphous iron-oxide andalpha-quartz. J. Colloid Interface Sci. 1995, 172 (1), 82–93.

(20) Kim, C. S.; Rytuba, J. J.; Brown, G. E. EXAFS study of mercury(II)sorption to Fe- and Al-(hydr)oxides: II. Effects of chloride andsulfate. J. Colloid Interface Sci. 2004, 270 (1), 9–20.

(21) Schluter, K. Review: evaporation of mercury from soils. Anintegration and synthesis of current knowledge. Environ. Geol.2000, 39 (3), 249–271.

(22) Zhang, H.; Lindberg, S. Processes influencing the emission ofmercury from soils: a conceptual model. J. Geophys. Res. 1999,104 (D17), 21889–21896.

(23) Murphy, E. A.; Dooley, J.; Windom, H. L.; Smith, R. G. Mercuryspecies in potable ground water in southern New Jersey. WaterAir Soil Pollut. 1994, 78, 61–72.

(24) Alberts, J. J.; Schindler, J. E.; Miller, R. W.; Nutter, D. E. Elementalmercury evolution mediated by humic acid. Science 1974, 184,895–897.

(25) Allard, B.; Arsenie, I. Abiotic reduction of mercury by humicsubstances in aquatic system s an important process for themercury cycle. Water Air Soil Pollut. 1991, 56 (1), 457–464.

(26) Barkay, T.; Miller, S. M.; Summers, A. O. Bacterial mercuryresistance from atoms to ecosystems. FEMS Microbiol. Rev. 2003,27, 355–384.

(27) Takeuchi, F.; Iwahori, K.; Kamimura, K.; Negishi, A.; Maeda, T.;Sugio, T. Volatilization of Mercury under Acidic Conditions fromMercury-polluted soil by a Mercury-resistant Acidothiobacillusferooxidans SUG2-2. Biosci. Biotechnol. Biochem. 2001, 65 (9),1981–1986.

(28) Wiatrowski, H. A.; Ward, P. M.; Barkay, T. Novel reduction ofmercury(II) by mercury-sensitive dissimilatory metal reducingbacteria. Environ. Sci. Technol. 2006, 40 (21), 6690–6696.

(29) Charlet, L.; Bosbach, D.; Peretyashko, T. Natural attenuation ofTCE, As, Hg linked to the heterogeneous oxidation of Fe(II): anAFM study. Chem. Geol. 2002, 190 (1-4), 303–319.

(30) O’Loughlin, E. J.; Kelly, S. D.; Kemner, K. M.; Csencsits, R.; Cook,R. E. Reduction of Ag(I), Au(III), Cu(II), and Hg(II) by Fe(II)/Fe(III) hydroxysulfate green rust. Chemosphere 2003, 53 (5),437–46.

(31) White, A. F.; Peterson, M. L. Reduction of aqueous transitionmetal species on the surfaces of Fe(II) - containing oxides.Geochim. Cosmochim. Acta 1996, 60 (20), 3799–3814.

(32) Nakata, K.; Nagasaki, S.; Tanaka, S.; Sakamoto, Y.; Tanaka, T.;Ogawa, H. Sorption and reduction of neptunium(V) on thesurface of iron oxides. Radiochim. Acta 2002, 90 (9-11), 665–669.

(33) Powell, B. A.; Fjeld, R. A.; Kaplan, D. I.; Coates, J. T.; Serkiz, S. M.Pu(V)O2+ Adsorption and Reduction by Synthetic Magnetite(Fe3O4). Environ. Sci. Technol. 2004, 38 (22), 6016–6024.

(34) Scott, T. B.; Allen, G. C.; Heard, P. J.; Randell, M. G. Reductionof U(VI) to U(IV) on the surface of magnetite. Geochim.Cosmochim. Acta 2005, 69 (24), 5639–5646.

(35) Scheinost, A. C.; Charlet, L. Selenite Reduction by Mackinawite,Magnetite and Siderite: XAS Characterization of NanosizedRedox Products. Environ. Sci. Technol. 2008, 42 (6), 1984–1989.

(36) Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure,Properties, Reactions, Occurences, and Uses; Wiley-VCH: NewYork, 2003.

(37) O’Dell, J. W.; Potter, B. B.; Lobring, L. B.; Martin, T. D.Determination of mercury in water by cold vapor atomicabsorption spectroscopy; Method 245.1; U.S. EnvironmentalProtection Agency, 1994.

(38) Kukkadapu, R. K.; Zachara, J. M.; Fredrickson, J. K. Biotrans-formation of two-line silica-ferrihydrite by a dissimilatory Fe(III)-reducing bacterium: Formation of a carbonate green rust in thepresence of phosphate. Geochim. Cosmochim. Acta 2004, 68(13), 2799–2814.

(39) Rancourt, D. J.; Ping, J. Y. Voigt-based methods for arbitrary-shape static hyperfine parameter distribution in Mossbauerspectroscopy. Nucl. Instrum. Meth. Phys. Res., B 1991, 58 (1),85–97.

(40) Cashion, J.; Murad, E. Mossbauer Spectroscopy of EnvironmentalMaterials and their Industrial Utilization; Kluwer AcademicPublishers: Boston, MA, 2004.

5312 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

(41) Stumm, W.; Hering, J. G. Oxidative and reductive Dissolutionof Minerals. In Mineral-Water Interface Geochemistry; Hochella,M. F., Jr., White, A. F., Eds.; Mineralogical Society of America:Chantilly, VA, 1990; pp 427-465.

(42) Luther, G. W. The frontier-molecular-orbital theory approachin geochemical processes. In Aquatic Chemical Kinetics; Stumm,W., Ed.; Interscience: New York, 1990; pp 173-198.

(43) Wesolowski, D. J.; Machesky, M. L.; Palmer, D. A. Magnetitesurface change studies from 190 degrees C from in situ pHtitrations. Chem. Geol. 2000, 167 (1-2), 193–229.

(44) Stoor, R. W.; Hurley, J. P.; Babiarz, C. L.; Armstrong, D. E.Subsurface sources of methyl mercury to Lake Superior froma wetland-forested watershed. Sci. Total Environ. 2006, 368,99–110.

(45) Bone, S. E.; Charette, M. A.; Lamborg, C. H.; Gonneea, M. E. HasSubmarine Groundwater Discharge Been Overlooked as a Sourceof Mercury to Coastal Waters? Environ. Sci. Technol. 2007, 41(9), 3090–3095.

(46) Barringer, J. L.; Szabo, Z. Overview of investigations into mercuryin ground water, soils, and septage, New Jersey Coastal Plain.Water Air Soil Pollut. 2006, (175), 193–221.

(47) Barringer, J. L.; Szabo, Z.; Kauffman, L. J.; Barringer, T. H.;Stackelberg, P. E.; Ivahnenko, T.; Rajagopalan, S.; Krabbenhoft,D. P. Mercury concentrations in water from an unconfinedaquifer system, New Jersey coastal plain. Sci. Total Environ.2005, 346 (1-3), 169–83.

ES9003608

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5313