Positive Impact of Microorganismson the Performance ofLaboratory-Scale PermeableReactive Iron BarriersT H O M A S V A N N O O T E N , † , ‡

D I R K S P R I N G A E L , ‡ A N DL E E N B A S T I A E N S * , †

Environmental and Process Technology, Flemish Institute forTechnological Research (VITO), Boeretang 200, 2400 Mol,Belgium, and Division of Soil and Water Management,Katholieke Universiteit Leuven, Kasteelpark Arenberg 20, 3001Heverlee, Belgium

Received July 17, 2007. Revised manuscript receivedDecember 3, 2007. Accepted December 7, 2007.

Degradation efficiencies of zerovalent iron (Fe0) containingdifferent bacterial inocula, i.e., an iron(III)-reducing Geobactersulfurreducens strain and/or a bacterial consortium, werecompared to degradation efficiencies of noninoculated Fe0 ina laboratory-scale column experiment. Contaminant removalefficienciesandhydrogenproductionrates indicatedanincreasingreactivity in time for all inoculated iron columns, whilereactivity of the noninoculated columns remained the same.The main mineral precipitates, including carbonate green rust,ferrous hydroxy carbonate, aragonite, and to a lesser extentgoethite, were observed under all imposed conditions. The higherreactivity of the inoculated column material is explicable bythe reduction of ferric iron species by iron(III)-reducing bacteria,resulting in the observed higher amounts of highly reactivecarbonate green rust. However, contributions of other bacteriacould not be excluded. Although different groups of hydrogen-consuming bacteria were detected in the columns, noindication was found that hydrogen consumption was sufficientlyhigh to affect reactivity or permeability of the iron matrix, asthe abiotic generation of H2 was substantially exceedingits potential consumption.

Introduction

The long-term efficiency of granular zerovalent iron (Fe0)permeable reactive barriers (PRBs) to treat groundwatercontaminated with chlorinated aliphatic hydrocarbons (CAHs)is still a point of concern as corrosion of iron grains and thesubsequent precipitation of secondary minerals can lead tosubstantial decreases in reactivity (1) and permeability (2)of the iron barrier matrix. Only little information is availableabout the microbial community composition in and aroundiron barriers and its effect on barrier performance (3, 4).Cathodic hydrogen generated by the anaerobic corrosion ofiron is a potential electron donor for various types ofmicroorganisms whose biochemical activity can contributein a direct way to contaminant removal in the barrier. For

example, several halorespiring bacteria use hydrogen as thesole electron donor to use and dechlorinate CAHs asrespiratory electron acceptors (5). In addition, various otherfacultative hydrogen-consuming microorganisms, includingmethanogenic (6), acetogenic, sulfate-reducing, denitrifying(7), and iron(III)-reducing bacteria (IRB) (8) are reported todegrade CAHs either by cometabolical means or by usingthe contaminants as electron donor. On the other hand,bacteria can indirectly affect barrier efficiency. The con-sumption of entrapped hydrogen gas might alleviate porevolume reduction and benefit the hydraulic flow in the ironbarriers (3). However, removal of the passivating hydrogenlayer by hydrogen-consuming bacteria might accelerate ironcorrosion and shorten the life span of iron barriers (9). IRBcan solubilize precipitate layers by reducing ferric ironcorrosion products to ferrous iron compounds, resulting intoa reactivation of passivated iron surfaces (10). The activityof sulfate-reducing bacteria (SRB) results into precipitationof ferrous iron sulfides which are significantly more reactiveper unit surface area than iron metal in transformation oftrichloroethylene (TCE) and other CAHs (11, 12). Denitrifyingbacteria can increase iron barrier efficiency by convertingNO3

-, which is reported to affect Fe0 reactivity by competingfor reactive sites (13), to more innocuous products as N2Oand N2, instead of the aesthetically less favorable ammoniumwhich is formed by abiotic reduction of nitrate (14). Biologi-cally produced N2 gas, however, has been reported to causeheterogeneity in column-scale iron barriers (2).

It was investigated in this study whether microbialactivities can affect the degradation efficiency of column-scale Fe0 PRBs toward TCE. Special attention was given toIRB which may increase iron reactivity by affecting ironprecipitate formation. Therefore, we compared the perfor-mance of poisoned noninoculated iron columns with ironcolumns seeded with different microbial inocula, i.e., aniron(III)-reducing Geobacter sulfurreducens (G. sulfurre-ducens) strain and/or a soil bacterial consortium containingIRB. The columns were operated under conditions whichmimic in situ conditions. Bacterial community structureswere studied via DNA-based PCR-DGGE analysis usingvarious group-specific primer sets, targeting the eubacterialand archaeal community, and different functional groups ofbacteria which might affect iron barrier performance,including IRB, SRB, methanogens, denitrifying bacteria, andhalorespiring bacteria. The impact of the microbial inoculumon barrier performance was investigated by measuringcontaminant degradation capacities and gas production ratesduring the experiment and by identifying iron corrosionproducts and other mineral precipitates after columndismantling with optical microscopy, scanning electronmicroscopy (SEM) combined with energy dispersive X-ray(EDX), and X-ray diffraction (XRD) analysis.

Experimental SectionDescription of Column Setup, Feed Solution, and Inocula.An overview of the column setup is given in Figure S1 of theSupporting Information. Glass columns (height, 10 cm; insidediameter, 2.4 cm) were dry-filled with granular iron particlesto a height of 8 cm and sealed with a viton stopper perforatedwith a needle to form the outlet port. The iron packingmaterial (Gotthart Maier Metallpulver) had an initial porosityof 0.47 (12). The dry-packed columns were autoclaved inclosed recipients to avoid corrosion of the iron by contactwith water. Before autoclaving, the columns were flushedwith CO2 to facilitate saturation with the feed solution at thestart of the experiment.

* Corresponding author phone: +3214335179; fax: +3214580523;e-mail: leen.bastiaens@vito.be

† VITO.‡ Katholieke Universiteit Leuven.

Environ. Sci. Technol. 2008, 42, 1680–1686

1680 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008 10.1021/es071760d CCC: $40.75 2008 American Chemical SocietyPublished on Web 02/05/2008

Artificial groundwater was prepared by autoclaving two5 L bottles of MilliQ water. After autoclaving, when solutionshad a temperature of ∼50 °C, sterile stock solutions of theappropriate salts were aseptically added to the water, resultingin final concentrations of 1 mM CaCl2 ·2H2O, 1 mMMgCl2 ·6H2O, 1 mM Na2SO4, 0.075 mM KNO3, 0.001 mMNa2HPO4, and 1 mL L-1 of a trace element solution (DSMZmedium 141). While cooling, the medium was deoxygenatedby flushing with N2 gas for at least 1 h. From a sterile stocksolution, 2 mM amounts of NaHCO3 and KHCO3 were thenaseptically added, and pH was adjusted to 7.47 ( 0.47 with30% HCl. After flushing with N2 gas for another 15 min, thebottles were connected to an aluminum foil balloon filledwith nitrogen. The feed solution was pumped into a first 12mL mixing vial where a concentrated aqueous solution ofTCE was added (11.8 ( 2.2 mg L-1) via a syringe pump. Thisvial was overflowing in a second mixing vial, receiving 13.8( 7.2 mg L-1 acetate. The second vial was overflowing intoa third mixing vial which received 6.4 g L-1 formaldehyde.The medium was continuously pumped from the mixingvials in an upward flow through the columns with a flow rateof 3.2 ( 0.3 mL h-1, corresponding to an initial pore watervelocity of 0.36 ( 0.04 m day-1.

Two different inocula were used, i.e., a single-strain G.sulfurreducens culture (DSMZ No. 12127) and a bacterialconsortium. G. sulfurreducens was precultured anaerobicallyfor 3 weeks at 29 °C in Geobacter medium 826 (DSMZ),resulting in a homogeneous culture containing (2.85 ( 0.45)× 108 cells mL-1 as counted in a microscopic countingchamber (HELBER). The consortium was recovered from ananaerobic soil sample by shaking 8 g of soil for 1 h in 72 mLof anaerobic water at 12 °C. After 30 min settling time, thetop liquid phase was taken as inoculum. Characteristics ofthe soil sample are presented in Table S1 of the SupportingInformation. Cell numbers in the extract were not determined.

Five different column conditions (Table S2) were set up,each performed in triplicate. Columns 1-3 (GEO-A) wereinoculated with 5 mL of the G. sulfurreducens culture andfed with medium originating from the first mixing vial,containing only TCE. Columns 4-12 were fed with mediumoriginating from the second mixing vial, containing TCEtogether with acetate as carbon source to stimulate bacterialactivity. Columns 4-6 (GEO) were inoculated with 5 mL ofthe G. sulfurreducens culture, columns 7-9 (GEO + CON)with a mixture of 5 mL of G. sulfurreducens culture and 5 mLof the bacterial consortium, and columns 10-12 (CON) with5 mL of consortium. The latter condition was included toinvestigate if iron(III)-reducing bacteria present in theconsortium can increase iron reactivity, thereby alleviatingthe need for bioaugmentation with a pure G. sulfurreducensculture. The condition seeded with the mixed inoculum wasincluded to evaluate possible out-competition of G. sul-furreducens by consortium bacteria. Columns 13-15 (blank)were noninoculated control columns and were fed withmedium originating from the third mixing vial, containingTCE and acetate. The control columns were poisoned startingon day 45 by the addition of formaldehyde to the feed.Columns were inoculated at day 21 of operation by slowlyinjecting the bacterial suspension through the inlet port ofthe columns. After inoculation, flow was stopped for 1 h toallow cell attachment. To ensure the presence of Geobacterspecies, columns operated under conditions 1 and 2 werereinoculated after 121 days with 9 mL of a G. sulfurreducensculture containing (2.46 ( 0.60) × 108 cells mL-1.

All column parts were sterilized by autoclaving, and sterileluer lock filters (0.22 µm pore size) were inserted betweencolumn inlet ports and upstream flow lines, and after thesyringes in the syringe pump, in order to avoid accidentalmicrobial contamination of the columns. In case of bacterial

contamination, all upstream column parts, including mixingvials, tubing, and bacterial filters were replaced by sterilecolumn parts. Gases were collected by passing the columneffluent through a sealed 20 mL glass tube (Figure S1). Columneffluent ports and the overflow of the third mixing vial werecontinuously sterilized by UV radiation during the operationperiod, to prevent bacterial contamination. The wholecolumn system was operated at room temperature.

Sampling Procedures and Analysis. The composition ofthe gas escaping the columns was regularly analyzed for theparameters described in Table S3. Therefore, gas was sampledfrom gas collection tubes connected to the outlet of thecolumns. The tubes were weighed before and after a gascollection period in order to determine the volume ofcaptured gas and to calculate gas generation rates. Periodi-cally, liquid samples (∼10 mL) collected from the columninlet and outlet were analyzed for TCE, TCE dechlorinationproducts, and acetate. Samples were taken by connectingsterile 10 mL glass syringes to sterilized sampling ports andallowing the liquid to flow into the syringes. Flow rates werecalculated by weighing the sampling syringes before and aftersampling. Acetate samples were prepared for analysis byadding 0.5 mL of sample to 2 mL of 50% H2SO4. Fatty acidswere extracted with diethyl ether and analyzed on a TraceGC-FID (Thermoquest). TCE concentrations, Eh, pH, anddissolved oxygen (DO) concentrations were determined asdescribed in Table S3.

After 233 days of operation, ∼400 mL samples were takenfrom the influent and effluent solutions of columns 1 (GEO-A), 4 (GEO), 7 (GEO + CON), 10 (CON), and 13 (blank) tomeasure alkalinity, metal concentrations, and concentrationsof sulfate, sulfide, nitrate, nitrite, and ammonia as describedin Table S1. The effluent samples were collected during sixdays of operation in nitrogen-flushed glass bottles connectedto an empty aluminum foil balloon that could expand, therebyavoiding overpressure.

Column Dismantling and Precipitate Identification.After 258 days of operation (corresponding to 1095 ( 103pore volumes, taking into account the initial porosity of theiron packing), columns were dismantled in an anaerobicglovebox (Don Whitley Scientific Ltd.) and the iron packingmaterial was divided into three sections of equal length. ForXRD analysis, 20 g subsamples were taken from selectedsections. To improve XRD detection, the precipitates weredetached from Fe0 filings by sonification for 30 min in 40 mLof acetone (15). The fine precipitate fraction was recoveredby filtration of the acetone solution and stored anaerobicallyin capped glass vials for a maximum of 1 week before analysis.XRD analysis was performed using a PANanalytical X-raydiffractometer (X’pert Pro) and Cu KRX-radiation. The samplewas scanned from 2 to 120° 2θ (40 kV, 40 mA; step size, 0.04°;step time, 1 s/step). For microscopic analysis, approximately1 g subsamples were taken from selected sections and driedin the anaerobic glovebox. Dry Fe0 filings were embedded inepoxy and polished after hardening for 12 h. Polished sectionswere studied using an AXIOPLAN Imaging reflected-lightmicroscope (Zeiss), and selected sections were platina coatedfor examination with an Environmental scanning electronmicroscope (JEOL JSM-6340F), fitted with an Omega detector(PGT SPIRIT) for energy dispersive X-ray analysis (EDX).

Microbial Characterization. After 85 and 182 days ofoperation, effluent samples (∼425 mL) were taken for DNA-based PCR-DGGE-analysis. During 6 days, effluent sampleswere collected while being cooled at 4 °C to minimize bacterialgrowth. Sample solutions were filtered through a 0.45 µmfilter (Millipore) and filters were transferred immediately toTris-glycerol (10 mM, 15%) buffer for DNA extraction. Afterdismantling the columns, 2 g solid samples were taken from

VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1681

the entrance section of each column. DNA was extractedfrom both liquid and solid samples as described by Hendrickxet al. (16).

PCR was performed on undiluted DNA samples withdifferent group specific primer sets, targeting Eubacteria (17),Archaea (18), IRB (Geobacteraceae and Geothrix) (19, 20), SRB(21), denitrifying bacteria (22), methanogens (23), Dehalo-coccoides (24), and dehalogenase genes encoding PCE-reductase (pceA), TCE-reductase (tceA) (25), and VC-reductasegenes (bvcA, vcrAB) (26, 27). The used primer sets aresummarized in Table S4, while PCR-DGGE conditions aredescribed elsewhere (12).

ResultsColumn Operation. Contaminant Degradation. To study theevolution of iron reactivity, TCE removal rates (mg h-1) werecalculated by multiplying the difference between the influentand effluent TCE concentrations with the flow rate. TCE wasinitially completely removed, due to low influent concentra-tions (<5 mg L-1) and possible sorption losses. Therefore,TCE removal rates are reported starting from day 50 (Figure1A), i.e., when the TCE influent concentrations were 11.8 (2.2 mg L-1. Similar degradation rates (0.024 ( 0.003 mg h-1)were observed for all columns at day 50. Although TCEdegradation rates were fluctuating, a similar gradual increasewas observed for all inoculated columns. Degradation ratesrecorded after 258 days of operation were ∼1.6 times higherthan those recorded after 50 days of operation. The noni-noculated control columns showed an increasing reactivityduring the first 120 days, similar to the inoculated columns.However, from day 120, degradation rates decreased to thevalues recorded at day 50. cis-Dichloroethylene (cis-DCE)

was recorded as the main TCE degradation product withgradually decreasing effluent concentrations (from 0.48 (0.07 to 0.20 ( 0.06 mg L-1) in time for all columns. Vinylchloride (VC) was not detected.

Influent and Effluent Characterization and Acetate Con-sumption. Chemical parameters of the influent and effluentsolutions recovered from the different columns are reportedin Table 1. The pH and Eh of the influent feed solution wererespectively 7.67 ( 0.57 and 67 ( 58 mV for all columns,except for the poisoned control columns where the additionof formaldehyde after 45 days of operation seems to causeinterference with the Eh measurement, resulting in lowervalues (-43 ( 37 mV) than those recorded before addition.However, during column transit, Eh decreased similarly to-230 ( 53 mV for all columns. The pH of the feed solutionslightly increased to 8.02 ( 0.48. Although the DO concen-tration of the feed solution in the bottles was kept below 0.9mg L-1, transport through the column tubing and the additionof aqueous TCE in the first mixing vial apparently caused anincrease in DO concentration in the influent solution of thenon-acetate columns with an average of 4.2 mg L-1. Theaverage influent DO concentration of the other inoculatedcolumns was lowered to 2.3 mg L-1 as the addition of acetatein the second mixing vial caused a partial oxygen consump-tion due to growth of a bacterial contamination as indicatedby an increasing turbidity. The addition of an aerobicformaldehyde solution to the third mixing vial again increasedthe average DO concentration up to 3.6 mg L-1 for theinfluents of the control columns (blank). The acetateconsumption rate, slightly fluctuating in time but similar forall inoculated columns, was 0.011 ( 0.006 mg h-1, while noacetate was consumed in the poisoned control columns.

FIGURE 1. TCE removal rates (mg h-1; A) and hydrogen production rates (mL day-1; B) recorded for the columns operated under thefive different conditions. Data are shown as mean ( standard deviation of triplicate columns, except for the hydrogen production ofthe non-acetate-fed columns (GEO-A) whose values were plotted separately due to low reproducibility.

TABLE 1. Summary of Chemical Parameters of the Column Influent and Effluent Solution for Each Conditiona

concentration, mg L-1

pH Eh, mV DO HCO3-/CO3

2- Ca2+ Fe Mg2+ K+ SO42- N (NO3

-) N (NH4+)

influent 7.67 ( 0.57 67 ( 58 (-43 ( 37)b variable 195.6 37.4 <0.025 23.5 69.2 98 1.09 <0.1effluentGEO-A 8.21 ( 0.47 -262 ( 51 0.52 (4.23)c 109.8 14 0.066 21.7 65.3 96 <0.23 1.4GEO 7.85 ( 0.49 -204 ( 38 0.47 (2.35) 148.2 25.2 0.047 21.8 66 98 <0.23 0.5GEO + CON 7.90 ( 0.46 -208 ( 51 0.47 (2.29) 147.0 24.2 0.143 21.7 65.8 98 <0.23 0.65CON 8.12 ( 0.44 -235 ( 61 0.42 (2.21) 154.2 24.9 0.361 21.4 66.6 97 <0.23 0.42blank 8.05 ( 0.47 -236 ( 46 0.35 (3.61) 118.8 34.5 0.087 21.5 65.7 96 0.75 d

a pH, Eh, and DO were regularly measured and are given as mean values of triplicate columns, while other parameterswere only determined after 238 days of operation for one column of each condition (columns 1, 4, 7, 10, and 13). b Theaddition of formaldehyde was causing interference with the Eh measurement, resulting into lower values. c Influent DOvalues are placed between parentheses. Although zero DO concentrations were expected in the effluent due to theoxidation of iron, the measured oxygen can be explained by slight air contamination during sampling. d Disturbedmeasurement, due to the presence of formaldehyde.

1682 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008

Alkalinity (HCO3- and CO3

2-), calcium, and to a lesserextent other cations (Mg2+, K+), were removed from theinfluent solutions by secondary mineral precipitation. Al-kalinity especially decreased in the non-acetate columns(43.9%) and the control columns (39.3%), and to a lesserextent in the inoculated acetate-fed columns (23.4 ( 2.0%).Calcium was partly removed from solution in the non-acetatecolumns (62.6%) and in the inoculated acetate-fed columns(33.8 ( 1.4%), while its removal was substantially lower inthe control columns (7.8%). Although the ratio between theremoval percentages of carbonates and calcium was similarfor all inoculated columns (4:1), it was substantially higherfor the control columns (27:1). Nitrate was removed to valuesbelow the detection limit in all inoculated columns, whileonly 31% removal was observed in the control columns.Ammonium concentrations in the effluent were highest forthe non-acetate columns (1.4 mg L-1 N) and similar for theinoculated acetate-fed columns (0.6 ( 0.1 mg L-1 N). Almostno sulfate was consumed (Table 1).

Gas Production. Gas produced in the columns wascomposed mainly of hydrogen (40-70%) and nitrogen, withtrace amounts of oxygen. Hydrogen production results fromthe reduction of water by iron under reducing conditions.Nitrogen, present in the gas phase, originated from the N2-flushed feed solution and possibly as a product of dissimila-tory denitrification. The contribution of dissimilatory den-itrification was relatively low, as nitrate was present inrelatively low concentrations in the feed solution, and ispreferentially reduced to ammonia by Fe0 (28). In case allnitrate was converted by dissimilatory denitrification, theproduced nitrogen could only have accounted for a maximum18% of the total nitrogen concentration. Recorded traceamounts of oxygen can be explained by air contaminationduring sampling.

Hydrogen gas production rates (Figure 1B) were similarfor all inoculated acetate-fed columns and gradually in-creased from 0.13 ( 0.05 mL day-1 at day 13 to 1.20 ( 0.08mL day-1 at day 251. The hydrogen production rates of thenon-acetate-fed columns demonstrated a low reproducibility.The maximum production rates were substantially higherfor column 1 and column 3 (2.5-3 mL day-1), compared tocolumn 2 which exhibited production rates similar to theinoculated acetate-fed columns. At the end of the experimenthydrogen production rates decreased to 1.6 mL day-1 forcolumn 1 and 1.0 mL day-1 for column 3. Lowest rates wereobserved for the poisoned control columns. Rates increasedslightly during the first 30 days of operation but remainedconstant (0.34 ( 0.09 mL day-1) for the rest of the operationperiod.

Microbial Characterization. All columns, except for thepoisoned control columns, were inoculated with either asingle-strain G. sulfurreducens culture, a bacterial consortium,or a mixture of both. The consortium contained SRB and IRBbelonging to the Geobacteraceae family as determined byPCR, while numbers of Archaea, methanogens, and deni-trifying bacteria were apparently below the detection limit(Table S4). In addition, 16S rDNA of Dehalococcoides andPCE- and TCE-reductase genes were detected in the con-sortium, but no VC-reductase genes.

Eubacteria were detected in solid and effluent samplesoriginating from all columns, including the noninoculatedcontrol columns. At the 85th day of operation, Geobacteraceaewere detected in almost all inoculated acetate-fed columns(GEO, GEO + CON, and CON), with strong PCR signalsrecorded for the columns inoculated exclusively with G.sulfurreducens (GEO). The latter columns were reinoculatedafter 121 days and showed again strong signals at the secondeffluent sampling after 182 days of operation, while noGeobacteraceae could be detected in the effluents of the othercolumns (except for column 7). Despite repetitive inoculation,

no Geobacteraceae could be detected in the effluents of thecolumns without acetate (GEO-A), probably due to a lack ofa suitable carbon source. After dismantling the columns,however, strong PCR signals were observed for the solidsamples, originating from all columns inoculated with G.sulfurreducens, but not for the columns inoculated exclusivelywith the consortium. Geobacteraceae could not be detectedin the control columns.

SRB and nirS-containing denitrifying bacteria were de-tected in both liquid (at days 85 and 182) and solid samples(at day 258). In contrast, Archaea, methanogens, IRB belong-ing to the Geothrix family, and nirK-containing denitrifyingbacteria were not detected in any of the samples. AlthoughDehalococcoides was present in the original inoculum, theywere not detected in the solid samples. tceA could be detectedin solid samples of columns 2 and 3 (GEO-A), of column 11(CON), and of the poisoned control column 14 (blank).

DGGE profiles of the eubacterial 16S rDNA PCR ampli-fication products are presented in Figure S2 for the solidsamples. Replicate columns were apparently dominated bydifferent populations. The columns inoculated exclusivelywith G. sulfurreducens (GEO-A and GEO) showed a weakband similar to the dominant band in the original pureculture. However, other stronger bands were visible, indicat-ing that the majority of the amplification products originatedfrom other bacteria, not present in the original inoculum.The presence of G. sulfurreducens in columns 1–9 (GEO-A,GEO, and GEO + CON), however, was clearly confirmed bya Geobacteraceae-specific PCR. Moreover, DGGE profiles ofthe corresponding amplification products were identical tothe banding pattern of G. sulfurreducens (Figure S3).

Precipitate Identification at Column Shutdown. Ac-cumulation of precipitates, occupying pore spaces andinterconnecting iron grains, could be visually observed inthe columns during the experiment. The hardness of thepacking material that was noticed during removal of the irongrains indicated a high degree of cementation over the wholecolumn length. Iron oxyhydroxides were observed withpolarized reflected-light microscopy as up to 150 µm thickreddish precipitate layers on iron grains taken directly afterthe column inlet, as a consequence of the relatively highinfluent DO concentrations. More downstream, precipitatelayers were generally thinner and varied in thicknessdepending on the imposed conditions. Iron grains originatingfrom the nonacetate-fed columns (GEO-A) showed thehighest degree of corrosion and were covered with up to 130µm thick precipitate layers. In all other columns, thicknessof precipitate layers ranged from 10 to 60 µm, except forcolumns 4 and 6 (GEO), where a significant portion of theiron grains was only covered with a thin precipitate layer ofless than 10 µm.

Samples for XRD analysis were taken from differentsections of appropriate column setups. An overview ispresented in Figure 2 in which for each sample the amountof the detected phases is indicated by the height of thediffraction peaks (counts s-1). This is a semiquantitativeoverview that enables to compare the concentration of aspecific mineral phase between different columns andcolumn sections, but that does not allow one to compareconcentrations of different phases in the same sample, asthe height of the diffraction peaks depends on the absorptioncoefficient which is characteristic for each phase. Moreover,the height of diffraction peaks depends not only on theconcentration but also on the degree of crystallinity and onthe orientation of the grains (29).

Four different mineral phases were detected. Carbonategreen rust (CGR), ferrous hydroxy carbonate (FHC), andaragonite were found in all columns. Goethite was detectedin most columns but only in relatively low amounts, asprincipal diffraction peaks were generally lower than 100

VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1683

counts s-1. CGR was almost exclusively detected in theentrance section of the columns. Samples taken from thepoisoned control columns (blank) clearly showed the leastintense CGR diffraction peaks. FHC and aragonite diffractionpeaks were most intense in the middle and top section of thecolumns, while only low diffraction peaks were observed atthe entrance. Samples taken from the poisoned controlcolumns showed the highest FHC peaks and the lowest peaksfor aragonite. FHC was present as needlelike crystals with amaximal length of ∼10 µm (Figure S4A). Aragonite wasidentified as clusters of elongated hexagonal prismatic crystals(up to 100 µm) (Figure S4B).

DiscussionLaboratory-scale iron columns were inoculated with aniron(III)-reducing G. sulfurreducens strain and/or a soilbacterial consortium to evaluate the effect of specific bacterialactivities on Fe0 reactivity. However, the detection of Eu-bacteria in the noninoculated control columns, and thepresence of other species in columns inoculated exclusivelywith G. sulfurreducens, indicated that bacterial contaminationoccurred during operation, contributing to the low repro-ducibility of microbial community structures in the replicatecolumns. The mixing vials were frequently contaminated bymicrobes, as indicated by an increasing turbidity and adecreasing oxygen content of the fluid, and may havecontaminated the columns by improper functioning of thebacterial filters in upstream flow lines. Bacteria couldproliferate in the control columns during the first 45 days ofoperation, although subsequent poisoning with formalde-hyde inhibited further bacterial growth.

Bacterial activity, indicated by acetate removal, appearedto be beneficial for column performance as TCE removalrates and hydrogen production rates of the inoculatedcolumns were significantly higher in comparison with thepoisoned control columns where acetate was not removed.Although biodegradation of contaminants in Fe0 PRBs hasbeen suggested by Lampron et al. (30), it likely was not amain removal process in our iron columns. Biodegradationwould not result in the observed elevated hydrogen produc-tion rates which indicate the increased iron reactivity in theinoculated columns. Moreover, Dehalococcoides and metha-nogens were not detected and tceA was only detected in afew columns, which did not show higher TCE removal ratesthan those recorded in other columns. Although biodegra-dation of TCE by other species cannot be completelyexcluded, we suggest that bacteria have mainly contributedto column reactivity in an indirect way by influencingprecipitate formation. As green rusts have a significantlyhigher specific surface area than Fe0 (31) and have beenreported to efficiently reduce chlorinated organics (32),especially the precipitation of CGR is believed to havecontributed to the increasing iron reactivity. Indeed, theinoculated columns showed clearly higher amounts of CGRthan the poisoned control columns. Although CGR phasescan also result from abiotic reactions, the higher amountsof CGR in the inoculated columns might be explained byreduction of amorphous hydrous ferric oxide by IRB (33).Indeed, Ona-Nguema et al. (34) reported that green rust wasformed as a result of bioreduction of well-crystallizedlepidocrocite by an iron(III)-reducing Shewanella putrefa-ciens culture. The green rust transformed into more stable

FIGURE 2. Semiquantitative representation of the XRD results. The concentration of the detected phases is indicated by the height ofthe diffraction peaks (counts s-1). Samples were taken from different sections (entrance, middle, and top) of the columns, operatedunder the five different conditions (GEO-A (1), GEO (2), GEO + CON (3), CON (4), and blank (5)). The column numbers were indicatedby C1-C15 in the X-axis.

1684 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008

forms (magnetite and siderite) after microbial activity wasstopped. Shin et al. (35) reported that the reduction of ironoxides by Shewanella alga BrY in batch experiments resultedinto increased TCE removal rates. The fact that DGGE profilesof the eubacterial communities did not always point towardthe presence of Geobacter in the columns can be explainedby other populations being numerically dominant over G.sulfurreducens in those columns. However, Geobacteraceae-specific PCR-DGGE analysis clearly demonstrated the pres-ence of G. sulfurreducens in the columns inoculated withthis strain. Contributions of other bacteria could not beexcluded and other iron(III)-reducing species, such asShewanella, might have performed bioreduction in columnsinoculated with the consortium, although they were nottargeted by PCR. Interestingly, Geobacter seem to be especiallyassociated with the solid matrix of the columns as the mostintense PCR signals were obtained with DNA from the solidsamples. This can be explained by the fact that Geobacteraceaerequire direct contact with iron(III) oxides to reduce them(36), resulting in attached Biofilm bacteria.

In contrast with CGR, the amount of FHC was apparentlyhigher in the control columns than in the inoculated columns.Although not much is known about its in situ redox reactivity,it has been suggested that FHC is a redox-active phase thatcan reduce TCE and conduct electrons between the ironcore and the contaminants (37). On the other hand, thethicker FHC precipitate layers might become insulating dueto an excessive distance for efficient electron transfer (37).The high carbonate removal due to FHC precipitation in thecontrol columns resulted into the lowest calcium retention(Table 1) and aragonite precipitation (Figure 2). Precipitationof aragonite and calcite did not significantly decrease ironreactivity in column-scale (37) and pilot-scale (38) reactiveiron barriers, although other researchers have reportedreactivity losses due to calcium carbonate precipitation(39, 40). In this study, aragonite precipitation was not believedto have significantly decreased iron reactivity as aragonitecrystals were only covering a small fraction of the overalliron surface.

Elevated concentrations of corrosive species such asoxygen in the influent of the non-acetate-fed columnsprobably accelerated iron corrosion and the subsequentmineral precipitation. In case of higher oxygen concentra-tions, the oxidation of Fe2+ to Fe3+ will be more competitivewith the reduction by the underlying Fe0, resulting into theformation of a passive oxide layer (41). Corrosion rates of thenon-acetate-fed columns 1 and 3 decreased at later stagesof the experiment, as indicated by the decreasing hydrogenproduction. However, no significantly higher amounts ofgoethite or other crystalline (oxyhydr)oxides could beobserved, indicating that the decreasing corrosion rates wereprobably caused by an increasing thickness of (amorphous)precipitate layers.

Although SRB were detected, column conditions appearedto be unfavorable to their growth, as sulfate consumptionwas very low and almost no iron sulfide could be detected.Insignificant SRB activity was also reflected by the poor SRBdetection at the end of the experiment. Therefore, SRB arenot believed to have significantly affected column perfor-mance. High nitrate removal activities and relatively lowammonium levels in the effluent indicate a positive con-tribution of denitrifying bacteria in the inoculated acetate-fed columns. The higher ammonium level in the non-acetatecolumns might be explained by slower denitrification rates,due to the lack of acetate as a suitable electron donor.Different groups of hydrogen-consuming bacteria weredetected in the inoculated columns, but hydrogen productionwas significantly lower in the poisoned control columns. Thisindicates that microbial hydrogen consumption did notsignificantly affect column reactivity or permeability as the

abiotic generation of H2 was substantially exceeding itspotential consumption.

In conclusion, the laboratory-scale experiment predictsthat in situ Fe0 PRBs should not be considered as purephysical-chemical remediation systems, but rather assystems in which a specific microbial community can developand contribute to barrier reactivity in an indirect way byaffecting precipitate formation. Efforts were made to conductthe column experiment under conditions that closely mimicin situ conditions. Flow rate, pH, and geochemical composi-tion of the synthetic feed solution were based on field-scalegeochemical parameters. The concentration of corrosivespecies such as nitrate and carbonate were based on fielddata (4, 38, 42, 43) and the used carbon source, acetate, isone of the main products in the anaerobic degradation oforganic matter, and a highly relevant source of electrons (5).Although oxygen concentrations were high compared tomany field sites, similar concentrations were reported inexisting barrier situations (42). On the other hand, bacterialcontributions are expected to be less explicit in field-scalePRBs as groundwater temperatures decrease bacterial activ-ity, compared to room temperature that was maintained inthis experiment.

AcknowledgmentsThis work was funded by the Institute for the Promotion ofInnovation through Science and Technology in Flanders(IWT-Vlaanderen) and the Flemish Institute for TechnologicalResearch (VITO). We thank T. Vanhentenrijck, M. Maesen,A. Bossus, M. Schoeters, R. Kemps, and M. Mertens fortechnical support to this study.

Supporting Information AvailableSchematic overview of the column setup, tables with PCRresults, soil sample characteristics, column conditions, andmethods used, DGGE-gel pictures, and photomicrographsof secondary precipitates. This material is available free ofcharge via the Internet at http://pubs.acs.org.

Literature Cited(1) Klausen, J.; Ranke, J.; Schwarzenbach, R. P. Influence of solution

composition and column aging on the reduction of nitroaro-matic compounds by zero-valent iron. Chemosphere 2001, 44,511–517.

(2) Kamolpornwijit, W.; Liang, L.; West, O. R.; Moline, G. R.; Sullivan,A. B. Preferential flow path development and its influence onlong-term PRB performance: column study. J. Contam. Hydrol.2003, 66, 161–178.

(3) Gu, B.; Watson, D. B.; Wu, L.; Phillips, D. H.; White, D. C.; Zhou,J. Microbial characteristics in a zero-valent iron reactive barrier.Environ. Monit. Assess. 2002, 77, 293–309.

(4) Wilkin, R. T.; Puls, R. W.; Sewell, G. W. Long-term performanceof permeable reactive barriers using zero-valent iron: Geochem-ical and microbiological effects. Ground Water 2002, 41, 493–503.

(5) Smidt, H.; De Vos, W. M. Anaerobic microbial dehalogenation.Annu. Rev. Microbiol. 2004, 58, 43–73.

(6) Weathers, L. J.; Parkin, G. F.; Alvarez, P. J. J. Utilization of cathodichydrogen as electron donor for chloroform cometabolism bya mixed methanogenic culture. Environ. Sci. Technol. 1997, 31,880–885.

(7) Field, J. A.; Sierra-Alvarez, R. Biodegradability of chlorinatedsolvents and related chlorinated aliphatic compounds. Rev.Environ. Sci. Biotechnol. 2004, 3, 185–254.

(8) Sung, Y.; Fletcher, K. E.; Ritalahti, K. M.; Apkarian, R. P.; Ramos-Hernandez, N.; Sanford, R. A.; Mesbah, N. M.; Löffler, F. E.Geobacter lovleyi sp. nov. strain SZ, a novel metal-reducing andtetrachloroethene-dechlorinating bacterium. Appl. Environ.Microbiol. 2006, 72, 2775–2782.

(9) Gregory, K. B.; Mason, M. G.; Picken, H. D.; Weathers, L. J.;Parkin, G. F. Bioaugmentation of Fe(0) for the remediation ofchlorinated aliphatic hydrocarbons. Environ. Eng. Sci. 2000,17, 169–180.

VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1685

(10) Gerlach, R.; Cunningham, A. B.; Caccavo, F., Jr. Dissimilatoryiron-reducing bacteria can influence the reduction of carbontetrachloride by iron metal. Environ. Sci. Technol. 2000, 34,2461–2464.

(11) Butler, E. C.; Hayes, K. F. Kinetics of the transformation ofhalogenated aliphatic compounds by iron sulfide. Environ. Sci.Technol. 2000, 34, 422–429.

(12) Van Nooten, T.; Lieben, F.; Dries, J.; Pirard, E.; Springael, D.;Bastiaens, L. Impact of microbial activities on the mineralogyand performance of column-scale permeable reactive ironbarriers operated under two different redox conditions. Environ.Sci. Technol. 2007, 41, 5724–5730.

(13) Schlicker, O.; Ebert, M.; Fruth, M.; Weidner, M.; Wüst, W.;Dahmke, A. Degradation of TCE with iron: The role of competingchromate and nitrate reduction. Ground Water 2000, 38, 403–409.

(14) Till, B. A.; Weathers, L. J.; Alvarez, P. J. J. Fe(0)-supportedautotrophic denitrification. Environ. Sci. Technol. 1998, 32, 634–639.

(15) Liang, L.; Moline, G. R.; Kamolpornwijit, W.; West, O. R. Influenceof hydrogeochemical processes on zero-valent iron reactivebarrier performance: A field investigation. J. Contam. Hydrol.2005, 80, 71–91.

(16) Hendrickx, B.; Dejonghe, W.; Boënne, W.; Brennerova, M.;Cernik, M.; Lederer, T.; Bucheli-Witschel, M.; Bastiaens, L.;Verstraete, W.; Top, E. M.; Diels, L.; Springael, D. Dynamics ofan oligotrophic bacterial aquifer community during contact witha groundwater plume contaminated with benzene, toluene,ethylbenzene, and xylenes: An in situ mesocosm study. Appl.Environ. Microbiol. 2005, 71, 3815–3825.

(17) Marchesi, J. R.; Sato, T.; Weightman, A. J.; Martin, T. A.; Fry,J. C.; Hiom, S. J.; Wade, W. G. Design and evaluation of usefulbacterium-specific PCR primers that amplify genes coding forbacterial 16S rRNA. Appl. Environ. Microbiol. 1998, 64, 795–799.

(18) Casamayor, E. O.; Massana, R.; Benlloch, S.; Øvreås, L.; Díez,B.; Goddard, V. J.; Gasol, J. M.; Joint, I.; Rodríguez-Valera, F.;Pedrós-Alió, C. Changes in archaeal, bacterial and eukaryalassemblages along a salinity gradient by comparison of geneticfingerprinting methods in a multipond solar saltern. Environ.Microbiol. 2002, 4, 338–348.

(19) Holmes, D. E.; Finneran, K. T.; O’Neil, R. A.; Lovley, D. R.Enrichment of members of the family Geobacteraceae associatedwith stimulation of dissimilatory metal reduction in uranium-contaminated aquifer sediments. Appl. Environ. Microbiol. 2002,68, 2300–2306.

(20) Snoeyenbos-West, O. L.; Nevin, K. P.; Anderson, R. T.; Lovley,D. R. Enrichment of Geobacter species in response to stimulationof Fe(III) reduction in sandy aquifer sediments. Microb. Ecol.2000, 39, 153–167.

(21) Geets, J.; Vanbroekhoven, K.; Borremans, B.; Vangronsveld, J.;van der Lelie, D.; Diels, L. Molecular monitoring of SRBcommunity structure and dynamics in batch experiments toexamine the applicability of in situ precipitation of heavy metalsfor groundwater remediation. J. Soils Sediments 2005, 5, 149–163.

(22) Braker, G.; Zhou, J.; Wu, L.; Devol, A. H.; Tiedje, J. M. Nitritereductase genes (nirK and nirS) as functional markers toinvestigate diversity of denitrifying bacteria in pacific northwestmarine sediment communities. Appl. Environ. Microbiol. 2000,66, 2096–2104.

(23) Luton, P. E.; Wayne, J. M.; Sharp, R. J.; Riley, P. W. The mcrAgene as an alternative to 16S rRNA in the phylogenetic analysisof methanogen populations in landfill. Microbiology 2002, 148,3521–3530.

(24) Löffler, F. E.; Sun, Q.; Li, J.; Tiedje, J. M. 16S rRNA gene-baseddetection of tetrachloroethene-dechlorinating Desulfuromonasand Dehalococcoides species. Appl. Environ. Microbiol. 2000,66, 1369–1374.

(25) Regeard, C.; Maillard, J.; Holliger, C. Development of degenerateand specific PCR primers for the detection and isolation ofknown and putative chloroethene reductive dehalogenase genes.J. Microbiol. Methods 2004, 56, 107–118.

(26) Krajmalnik-Brown, R.; Hölscher, T.; Thomson, I. N.; Saunders,F. M.; Ritalahti, K. M.; Löffler, F. E. Genetic identification of aputative vinyl chloride reductase in Dehalococcoides sp. strainBAV1. Appl. Environ. Microbiol. 2004, 70, 6347–6351.

(27) Müller, J. A.; Rosner, B. M.; Von Abendroth, G.; Meshulam-Simon, G.; McCarty, P. L.; Spormann, A. M. Molecular iden-tification of the catabolic vinyl chloride reductase from Deh-alococcoides sp. strain VS and its environmental distribution.Appl. Environ. Microbiol. 2004, 70, 4880–4888.

(28) Huang, C.-P.; Wang, H.-W.; Chiu, P.-C. Nitrate reduction bymetallic iron. Water Res. 1998, 32, 2257–2264.

(29) Cullity, B. D. Elements of X-Ray Diffraction, 2nd ed.; Addison-Wesley: London, 1978.

(30) Lampron, K. J.; Chiu, P. C.; Cha, D. K. Reductive dehalogenationof chlorinated ethenes with elemental iron: the role of micro-organisms. Water Res. 2001, 35, 3077–3084.

(31) Williams, A. G. B.; Scherer, M. M. Kinetics of Cr(VI) reductionby carbonate green rust. Environ. Sci. Technol. 2001, 35, 3488–3494.

(32) Erbs, M.; Hansen, H. C. B.; Olsen, C. E. Reductive dechlorinationof carbon tetrachloride using iron(II) iron(III) hydroxide sulfate(green rust). Environ. Sci. Technol. 1999, 33, 307–311.

(33) Fredrickson, J. K.; Zachara, J. M.; Kennedy, D. W.; Dong, H.;Onstott, T. C.; Hinman, N. W.; Li, S.-M. Biogenic iron miner-alization accompanying the dissimilatory reduction of hydrousferric oxide by a groundwater bacterium. Geochim. Cosmochim.Acta 1998, 62, 3239–3257.

(34) Ona-Nguema, G.; Abdelmoula, M.; Jorand, F.; Benali, O.; Géhin,A.; Block, J.-C.; Génin, J.-M. R. Iron(II,III) hydroxycarbonategreen rust formation and stabilization from lepidocrocitebioreduction. Environ. Sci. Technol. 2002, 36, 16–20.

(35) Shin, H.-Y.; Singal, N.; Park, J.-W. Regeneration of iron fortrichloroethylene reduction by Shewanella alga BrY. Chemo-sphere 2007, 68, 1129–1134.

(36) Holmes, D. E.; Finneran, K. T.; O’Neil, R. A.; Lovley, D. R.Enrichment of members of the family Geobacteraceae associatedwith stimulation of dissimilatory metal reduction in uranium-contaminated aquifer sediments. Appl. Environ. Microbiol. 2002,68, 2300–2306.

(37) Kohn, T.; Livi, K. J. T.; Roberts, A. L.; Vikesland, P. J. Longevityof granular iron in groundwater treatment processes: Corrosionproduct development. Environ. Sci. Technol. 2005, 39, 2867–2879.

(38) Vogan, J. L.; Focht, R. M.; Clark, D. K.; Graham, S. L. Performanceevaluation of a permeable reactive barrier for remediation ofdissolved chlorinated solvents in groundwater. J. Hazard. Mater.1999, 68, 97–108.

(39) Zhang, Y.; Gillham, R. W. Effects of gas generation andprecipitates on performance of Fe0 PRBs. Ground Water 2005,43, 113–121.

(40) Jeen, S.-W.; Gillham, R. W.; Blowes, D. W. Effects of carbonateprecipitates on long-term performance of granular iron forreductive dechlorination of TCE. Environ. Sci. Technol. 2006,40, 6432–6437.

(41) Agrawal, A.; Ferguson, W. J.; Gardner, B. O.; Christ, J. A.; Bandstra,J. Z.; Tratnyek, P. G. Effects of carbonate species on the kineticsof dechlorination of 1,1,1-trichloroethane by zero-valent iron.Environ. Sci. Technol. 2002, 36, 4326–4333.

(42) Liang, L.; Korte, N.; Gu, B.; Puls, R.; Reeter, C. Geochemical andmicrobial reactions affecting the long-term performance of insitu ‘iron barriers’. Adv. Environ. Res. 2000, 4, 273–286.

(43) Phillips, D. H.; Watson, D. B.; Roh, Y.; Gu, B. Mineralogicalcharacteristics and transformation during long-term operationof a zerovalent iron reactive barrier. J. Environ. Qual. 2003, 32,2033–2045.

ES071760D

1686 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008