Journal of Contaminant Hydrology 116 (2010) 47–57

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology

j ourna l homepage: www.e lsev ie r.com/ locate / jconhyd

Iron hydroxy carbonate formation in zerovalent iron permeable reactivebarriers: Characterization and evaluation of phase stability

Tony R. Lee, Richard T. Wilkin⁎U.S. Environmental Protection Agency, National Risk Management Research Laboratory, Ground Water and Ecosystems Restoration Division, 919 Kerr Research Drive,Ada, OK 74820, United States

a r t i c l e i n f o

⁎ Corresponding author. Tel.: +1 580 436 8874; faxE-mail address: wilkin.rick@epa.gov (R.T. Wilkin).

0169-7722/$ – see front matter. Published by Elseviedoi:10.1016/j.jconhyd.2010.05.009

a b s t r a c t

Article history:Received 11 January 2010Received in revised form 19 May 2010Accepted 25 May 2010Available online 31 May 2010

Predicting the long-term potential of permeable reactive barriers for treating contaminatedgroundwater relies onunderstanding the endpoints of biogeochemical reactions between influentgroundwater and the reactive medium. Iron hydroxy carbonate (chukanovite) is frequentlyobserved as a secondary mineral precipitate in granular iron PRBs. Mineralogical characterizationwas carried out using X-ray diffraction, scanning electron microscopy, thermogravimetricanalysis, and X-ray absorption spectroscopy onmaterials collected from three field-based PRBs intheUS (EastHelena,MT; Elizabeth City, NC;Denver Federal Center, CO). These PRBswere installedto treat a range of contaminants, including chlorinated organics, hexavalent chromium, andarsenic. Results obtained indicate that chukanovite is a prevalent secondary precipitate in thePRBs. Laboratory experiments on high-purity chukanovite separateswere carried out to constrainthe room-temperature solubility for this mineral. An estimated Gibbs energy of formation (ΔfG°)for chukanovite is−1174.4±6 kJ/mol. Amineral stability diagram is consistentwith observationsfrom the field. Water chemistry from the three reactive barriers falls inside the predicted stabilityfield for chukanovite, at inorganic carbon concentrations intermediate to the stability fields ofsiderite and ferrous hydroxide. These new data will aid in developing better predictive models ofmineral accumulation in zerovalent iron PRBs.

Published by Elsevier B.V.

Keywords:Permeable reactive barrierZerovalent ironChukanoviteIron hydroxy carbonate

1. Introduction

A critical factor governing the long-term effectiveness ofsubsurface installations of granular iron is the inevitablebuildup of mineral precipitates on the surfaces of the irongrains. Consequently, an active area of research relating totechnology verification of PRBs focuses on the description andprediction of mineral precipitate accumulation that encom-passes variable groundwater chemistries and hydrologicregimes. Several studies have reported on mineralizationprocesses in controlled laboratory batch and column tests(e.g., Gu et al., 1999; Köber et al., 2002; Kamolpornwijit et al.,2004; Zhang and Gillham, 2005; Kohn et al., 2005; Huang andZhang, 2006; Nooten et al., 2007; Jeen et al., 2007; Parbs et al.,2007). Reactive transport simulations have also been con-

: +1 580 436 8703.

r B.V.

ducted in order to assess the impact of mineral fouling on thehydraulic performance of iron PRBs (e.g., Li et al., 2005, 2006;Mayer et al., 2001). However, comparatively few case studiesare available that report on the long-term behavior of field-based systems, particularly with respect to trends in thebuildup of mineral precipitates and efficiency of contaminantremoval (e.g., Phillips et al., 2000; Furukawa et al., 2002).

Besides impacting contaminant species, zerovalent ironalso affects the biogeochemical behavior of the moreconcentrated assortment of groundwater solutes. Reactionprocesses that involve the major anionic (e.g., Cl−, SO4

2−, andHCO3

−) and cationic groundwater components (e.g., Ca2+,Mg2+, and Na+) govern the kinetics and pathways of ironcorrosion, mineral precipitation, microbial activity, and gasproduction within and around the reactive medium (e.g.,Wilkin and Puls, 2003; Reardon, 2005). The cumulative effectof these processes through time can lead to changes in thereactivity, porosity, and hydraulic permeability of a PRB.

48 T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

An additional complicating factor is that mineral precip-itation and adsorption processes can result in either disad-vantageous or beneficial effects. For example, excessiveoxidation can lead to rapid pore clogging, bypassing, andreductions in flow rate (Liang et al., 2000). Adsorption of silicaat the iron–water interface can substantially reduce thereduction rate of certain organohalides (e.g., Kohn et al.,2003). On the other hand, accumulation of secondary iron-bearing phases including hydroxides and sulfides can provideadditional sorption capacity and/or reactivity in PRBs (Butlerand Hayes, 2001; Furukawa et al., 2002; Wilkin et al., 2005;Van Nooten et al., 2007). Better predictive models ofcontaminant treatment by PRBs require an improved under-standing of mineral accumulation processes as they relate tothe sustainability of hydraulic performance, contaminantdegradation, and contaminant sequestration.

Ferrous hydroxy carbonate (FHC, Fe2(OH)2CO3) has provento be a commonmineral precipitate identified in lab-based andfield-based applications of granular iron (e.g., Wilkin and Puls,2003; Kamolpornwijit et al., 2004; Kohn et al., 2005; Jeen et al.,2006). The first description of this phase is from a study ofcorrosion deposits that formed on steel between pH 9–11 at180 °C (Erdös and Altofer, 1976). More recently, Kukkadapu etal. (2005) showed that FHC formed by slow reaction ofmicrobially produced carbonate with Fe(II)-excess magnetite.FHC has recently been given a mineral name, chukanovite(Pekov et al., 2007). The mineral description of chukanovite isbased on occurrences in cavities of weathered fragments of aniron meteorite which was discovered about 350 km southeastofMoscow, Russia (Pekov et al., 2007). Chukanovite is commonin zerovalent iron PRBs and Kukkadapu et al. (2005) suggestthat the phase may be commonly overlooked in anoxicsediments. One limitation of geochemical modeling effortsthat attempt to predict mineral formation and accumulationin PRBs through time is the lack of thermodynamic datapertaining to this phase.

In this study, we examine trends in iron carbonateformation in three field-scale PRBs. The occurrence of chuka-novite is examined using X-ray diffraction (XRD), scanningelectron microscopy (SEM), thermal methods (TGA–MS), andX-ray absorption spectroscopy (XAS). Concentrated specimenswere subjected to solubility studies in order to estimate theGibbs energy of formation for chukanovite.

2. Methods

2.1. Core samples

Core samples were collected from granular iron PRBs at 3different field sites in the US: East Helena, Montana (EH),Elizabeth City, North Carolina (EC), and the Denver FederalCenter, Colorado (DFC). Background information on thesePRBs has been presented in previous publications (e.g.,Wilkin and Puls, 2003; Paul et al., 2003; Wilkin et al., 2008).Angle coring and vertical coringmethods utilizing a Geoprobewere used to collect samples of the reactive medium from thethree sites (Beck et al., 2002). Immediately after collection,the cores were frozen and shipped back to the laboratory ondry ice. This preservation method is effective for maintainingthe redox-sensitive mineralogy in granular iron cores(Wilkin, 2006). Frozen cores were thawed and sampled in a

Coy anaerobic glove box with a maintained H2–N2 atmo-sphere. All sub-samples were retained in airtight vials underan anaerobic atmosphere to prevent any air oxidation ofredox-sensitive constituents.

2.2. X-ray diffraction

Powder X-ray diffraction analysis on core samplescollected from the EH, EC, and DFC sites were conducted todetermine the mineralogy of precipitates formed in thezerovalent iron treatment zones. Materials for analysis wereprepared by sonicating iron core samples in methanol for10 min, followed by filtration of the released particulatesthrough 47-mm diameter, 0.2-micron filter paper (polycar-bonate). The separated particles were mounted on a zero-background quartz plate and scanned with Fe (X-ray)radiation over a range of 10° to 90° 2-theta (0.1°/min. and0.01° step intervals) using a RigakuMiniflex II Diffractometer.All sample manipulations, including the X-ray scans, werecarried outwithin a Coy anaerobic glove box. Using these scanconditions the estimated detection limit for crystalline solidscontained in the samples is about 5 wt.%.

2.3. Scanning electron microscopy

Samples were sent to Spectrum Petrographics, Inc. (Van-couver, WA) for the preparation of standard (27 mm×46mm)polished thin-sections. The samples were shipped in sealedglass bottles. To the extent possible, air exposure was mini-mized during the sample mounting and polishing procedures.A gold coating was applied to the thin-sections with a Kurt J.Lesker Company coating device. A JEOL JSM-6360 scanningelectron microscope (SEM) was used to evaluate the morphol-ogy and spatial relationships among mineral precipitates onthe surfaces of zerovalent iron particles from the different sites.The thickness of precipitate layers on the zerovalent iron wasmeasured and energy dispersive spectroscopy (EDS) was usedto determine on a semi-quantitative basis the composition ofsurface precipitates. The SEM system is equipped with a high-resolution Ge detector and imaging was conducted in thebackscatter electron composite mode.

2.4. Thermal analysis

Gradient and isothermal thermogravimetric analyseswere performed using a Netzsch STA409 TGA/DSC coupledto a Pfeiffer QMS 300 mass spectrometer using a methodbased on Kamolpornwijit et al. (2004). Samples were placedinto alumina crucibles and heated to 900 °C (10 °C/min)under Ar gas flow (50 mL/min). At 900 ° C a 2% H2 gas in Arwas passed through the furnace causing transformation ofiron oxides to Fe0 (Kamolpornwijit et al., 2004). Off-gaseswere monitored for m/z 18 (H2O) and 44 (CO2) using themass spectrometer.

2.5. X-ray absorption spectroscopy

X-ray absorption spectroscopy (XAS) measurements weremade at the high-photon-flux insertion device beamline (10-ID), MR-CAT, at the Advanced Photon Source (ArgonneNational Laboratory). The Fe K-edge spectra were collected

49T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

using a silicon (111) double-crystal monochromator at room-temperature in transmission mode with gas-filled ionchambers. Samples and reference compounds were anoxi-cally packed into 1 mm thick plastic holders and sealed withKapton tape. Energy calibration was accomplished duringeach scan by setting the first inflection of an iron foilspectrum to 7112.0 eV. Multiple scans of each sample werecollected and each scan was aligned using the iron foilreference. Raw data from multiple scans were merged andprocessed using the Athena software package (Ravel andNewville, 2005). The merged and averaged spectra werebackground corrected and step height normalized. ExtendedX-ray absorption fine structure (EXAFS) spectra were fit withtheoretical phase-shift and amplitude functions calculatedusing the ab initio FEFF6 program (Rehr et al., 1992; Rehr,1993) from the crystal structure refinement for chukanovitereported in Pekov et al. (2007). E0 was set at the absorptionedge inflection point for all samples. Radial distributionfunctions around the Fe absorber were determined by Fouriertransformation of the k3χ(k) EXAFS functions using atruncated k-range of 2.0–12 Å−1. The EXAFS data were fitwith the coordination number (N), interatomic distance (R),Debye–Waller (σ2), and the energy offset (ΔE0) parameters.

Fig. 1. a) X-ray diffraction scans for selected mineral separates from the East Helena,iron hydroxy carbonate (PDF #33-0650; Erdös and Altofer, 1976) is also plotted for cPeak positions for magnetite (M) and aragonite (A) are also shown. b) expandedchukanovite.

2.6. Solution analyses

Solubility studies were conducted on selected mineralseparates from the DFC PRB. Measurements of pH were madewith a combination electrode calibrated against NIST-trace-able buffer solutions. Solution measurements were made forNa, Fe, and Ca using inductively coupled plasma–opticalemission spectrometry (ICP–OES, Perkin-Elmer Optima3300DV) or inductively coupled plasma–mass spectrometry(ICP–MS, PQExcell, Thermo Elemental). Concentrations ofdissolved inorganic carbon were measured with a DohrmannDC-80 Carbon Analyzer. For all solution measurements,quality assurance tests involved duplicate samples, blanks,sample matrix spikes, calibration check standards, andsecond-source quality control samples.

3. Results and discussion

3.1. Solid-phase characterization

X-ray diffraction results show that chukanovite is theprimary mineral constituent in all three of the granular ironPRBs examined in this study (Fig. 1). In addition to

Denver Federal Center, and Elizabeth City PRBs. The powder diffraction file foromparison. Lines for angles b55° 2-theta are extrapolated onto the data plot.view of the 2-theta region from 30 to 50° shows the excellent match with

50 T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

chukanovite, other identified phases are magnetite andaragonite. The occurrence of chukanovite in these PRBs,along with other reports from the literature (e.g., Gavaskar etal., 2002; Wilkin and Puls, 2003; Kamolpornwijit et al., 2004;Jeen et al., 2007; Kohn et al., 2005), indicate that this phase isindeed a major corrosion product in zerovalent reactivebarriers. Interestingly, siderite is not identified as the primaryferrous carbonate and although siderite is commonly as-sumed to be present in zerovalent iron PRBs, we are notaware of studies that show demonstrative XRD evidenceconfirming the formation of siderite. Phase relationshipsamong siderite, chukanovite, and ferrous hydroxide areexamined in a subsequent section, however, the increasedpH typically observed in granular iron systems, that resultsfrom iron corrosion, is likely responsible for forcing theprecipitation of chukanovite over siderite. Chukanoviteclearly appears to be a product of iron metal corrosion inthe presence of bicarbonate; however, Kukkadapu et al.

Fig. 2. Scanning electron microscope images of polished thin-sections from the Elileading edge and mid-barrier section of the PRB. Images b and f show incorporatio

(2005) show that another formation pathway is by reactionof biogenic magnetite with bicarbonate. Kohn et al. (2005)also suggested that iron hydroxy carbonate converts tomagnetite and maghemite through time in PRBs. Sincemagnetite is a commonly observed mineral in zerovalentiron PRBs, there may be multiple formation and transforma-tion pathways for chukanovite.

Expanded view of an X-ray diffraction scan of fine-grainedprecipitates from the Denver Federal Center over the 2-thetaregion from 30 to 50° shows an excellent match with the XRDreference data reported in Erdös and Altofer (1976), PDF 00-033-0650; Fig. 1b). The X-ray diffraction pattern for chuka-novite as reported by Pekov et al. (2007) closely matches thepattern originally presented by Erdös and Altofer (1976),from which the powder diffraction file is assigned.

Scanning electron micrographs of mineral precipitatesformed on surfaces of iron particles from the Elizabeth CityPRB are shown in Fig. 2. Particle morphologies range from

zabeth City PRB. Images a, c, d, and e show typical surface coatings near then of aquifer material (quartz grains) in the precipitation layers.

51T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

radiating tabular or needle-like crystals to amorphousmassesand are comparable to the mineral accumulations on granulariron observed in other field-based and laboratory columnstudies (e.g., Kamolpornwijit et al., 2004; Kohn et al., 2005;Jeen et al., 2007). The thickness of precipitate crusts rangesfrom b5 µm to about 200 µm. Generally the greatest amountof accumulation is observed near the leading edge of the PRB.Relatively thin or non-existent surface coverage of mineralprecipitates is observed near the downgradient edge of thePRB. In some cases, grains of aquifer sediment (quartzparticles) are encrusted within neoformed mineral masseson the iron particles (Fig. 2b and f). These mineral massesclearly block reactive surface area of the granular ironmedium and this micro-scale observation suggests thatreactions between contaminants in the aqueous phase andiron surfaces become less efficient as the iron surface iscoated with fresh mineral precipitates.

Micro-scale compositional data were collected using EDSat points represented by the red circles in Fig. 2. Analyticaltransects were established perpendicular to iron grainsurfaces to evaluate whether element concentration gradi-ents are present in the mineral coatings. The abundances ofiron and oxygen for all points analyzed are plotted in Fig. 3,along with the idealized compositions of wüstite, magnetite,goethite, ferrihydrite, chukanovite, and quartz for reference.The Fe–O compositions span a range from Fe(0) to quartz,representative of the transition from granular iron particles tothe native aquifer. Platy to needle-like particles, for example,representative of mineral precipitates shown in Fig. 2a and e,generally cluster around the ideal composition of chukano-vite. X-ray maps for selected grains clearly show the relativeenrichment in O and depletion of Fe in themineral precipitatehalos surrounding granular iron particles (Fig. 4). Enrichmentof sulfur is found in areas exterior to iron grains in Fig. 4a. X-ray maps clearly reveal the distribution of aquifer quartzgrains in Fig. 4b.

TGA–MS was performed on a high-purity chukanovitesample collected from the Denver Federal Center PRB (Fig. 5).Weight loss patterns reveal inflections at about 220, 370, and

Fig. 3. EDX composition of surface precipitates with ideal Fe and O contentsindicated for wüstite, magnetite, goethite, chukanovite, ferrihydrite, andquartz.

500 °C. The lower temperature thermal events are associatedwith H2O(g) and CO2(g) loss. The 500 °C event is largelyassociated with the loss of CO2(g). Differential ScanningCalorimetry (DSC) shows that the sample underwent a broadexothermic reactionwhich corresponds to the loss of H2O andCO2. During the isothermal reduction period at 900 °C, a slightweight loss also accompanied by H2O(g) production wasobserved. The final weight loss was about 30% of the initialweight, which is equivalent to the amount of CO2+H2O in Fe2(OH)2CO3. These thermogravimetric features are broadlysimilar to results reported in Erdös and Altofer (1976) whoshowed a loss of H2O and CO2 from ferrous hydroxy carbonateat temperatures of 250 and 500 °C. For example, theyindicated that Fe2(OH)2CO3 breaks down under an inertatmosphere at 500 °C following the decomposition reaction(Kamolpornwijit et al., 2004):

3Fe2ðOHÞ2CO3⇒2Fe3O4 þ H2OðgÞ þ 2H2ðgÞ þ 3CO2ðgÞ ð1ÞNormalized X-ray absorption near edge structure (XANES)

and first-derivative spectra for chukanovite, siderite, andmagnetite are shown in Fig. 6a and b. The edge and post-edgeXANES structure of chukanovite differs from that of siderite anddisplays a dampened pre-edge peak compared to magnetite.The distinct pre-edge feature in magnetite reflects iron in var-iable oxidation states situated in both tetrahedral and octahe-dral sites (Wilke et al., 2001). Chukanovite shows primary peakinflections at 7121.0 and 7124.5 eV which are similar to mag-netite inflections and corresponds to iron primarily in octa-hedral coordination (O'Day et al., 2004). Pekov et al. (2007)describe the chukanovite crystal structure as edge-sharingoctahedra that form ribbons, interlinked by corner-sharingstructural units to form corrugated octahedral layers. Quanti-tative analysis of the iron EXAFS data for three fine-grainedisolates from the East Helena and Denver Federal Center PRBsshows iron in coordinationwith oxygen. Fig. 5c and d show thek3-weighted EXAFS spectra. The fits of the theoretical EXAFSspectra to the experimental data are also shown (solid curves inFig. 6c and d). Interatomic distances derived from fits of theEXAFS data (Table 1) indicate one predominant feature seenin the Fourier-transformed spectra that corresponds to two setsof Fe–O backscatterers in the first shell. Second-neighborbackscattering past the first shell of octahedral O atoms is notapparent in the chukanovite spectra. The interatomic Fe–Odistances in the PRB separates are similar to those in chuka-novite as reported by Pekov et al. (2007) based on Rietveldrefinement of X-ray diffraction data, with Fe–O distances of2.03–2.08 Å and 2.44–2.46 Å. Slight mismatch in the fits atlow k possibly indicates local octahedral distortion or resultsfrom impurities in the samples (O'Day et al., 2004).

3.2. Laboratory solubility studies

Mineral separates from the Denver Federal center werenoted to be relatively pure in chukanovite. Mineral powdersfrom this site were selected for dissolution studies toconstrain the solubility of chukanovite. Approximately100 mg of chukanovite was added to 50 mL of 3 to 40 mMNaHCO3 solutions. In some cases pH was adjusted by addingaliquots 1 mM HCl; experimental pH values ranged from 8.3to 9.1. All experiments were conducted in an anaerobic

Fig. 4. X-ray maps for selected iron grains from the Elizabeth City PRB: a) Fe, O, and S showing precipitate overgrowths and b) Fe, O, and Si showing quartzencrustations. The upper left image in each panel is a backscattered electron image. X-ray maps were produced by rastering the electron beam through 1 µm-spaced lateral scans.

52 T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

Fig. 5. TGA/DSC–MS results for weight loss, heat transfer, and off-gas composition for a chukanovite-rich sample from the Denver Federal Center PRB.

53T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

glovebox and all solutionswere purgedwith N2 gas to removedissolved oxygen. The temperature in the glovebox averaged22.3±1.9 °C. Dissolution experiments were conducted for 2to 12 days. At the end of an experiment, the final pH wasmeasured and the solution was filtered (0.2 µm) andanalyzed for dissolved Na, Fe, Ca, and inorganic carbon. Thesolution composition was speciated using the Geochemist'sWorkbench software (v.8, Rockware, Inc.; thermo.dat data-base; Delany and Lundeen, 1990). Iron species consideredincluded Fe2+, FeCO3

0, FeHCO3+, FeCl+, FeOH+, Fe(OH)20, and

Fe(OH)3−. Activity coefficients were calculated using the

extended B-dot equation (Helgeson and Kirkham, 1974).The solution data were evaluated following the reactiondescribing the dissolution of chukanovite:

Fe2ðOHÞ2CO3 þ 3Hþ⇔2Fe

2þ þ HCO−3 þ 2H2O ð2Þ

The equilibrium constant for this reaction is given by:

logKsp ¼ 2logfFe2þg þ logfHCO−3 g þ 3pH ð3Þ

Experimental results for chukanovite solubility are shownin Table 2. The average log Ksp value for reaction 2

Fig. 6. a) Normalized iron K-edge XANES spectra of chukanovite separated from the Denver Federal Center PRB and reference samples magnetite and siderite, b)corresponding first-derivative spectra, c) normalized background-subtracted k3-weighted EXAFS spectra for three PRB samples. The open circles represent theexperimental data and the solid lines represent nonlinear least-squares best fits (numerical fit results are shown in Table 1), d) Fourier-transforms of the k3-weighted spectra shown in panel c (uncorrected for backscatterer phase-shift).

54 T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

determined from the set of experiments is 11.9±0.5. Itshould be noted that this value is a preliminary estimatebecause it was only approached from undersaturated condi-tions. Experiments were not attempted from oversaturatedconditions because of uncertainties about whether chukano-vite seed crystals would grow from aqueous solution. Log Ksp

Table 1EXAFS fit results for chukanovite from field-based PRBs.

Sample Bond pair R (Å) σ2 (Å−1) ΔE0 R XRD (Å)

East Helena Fe–O 2.03 0.006 7.7 2.073Fe–O 2.46 0.008 2.411Fe–OH 2.30 0.002 2.260

East Helena Fe–O 2.08 0.013 9.1 2.073Fe–O 2.46 0.028 2.411Fe–OH 2.37 0.018 2.260

Denver Federal Center Fe–O 2.03 0.012 5.1 2.073Fe–O 2.44 0.006 2.411Fe–OH 2.25 0.002 2.260

R=interatomic distance; σ2=Debye–Waller parameter; ΔE0=energy shifin the least-squares fit; R XRD is the interatomic distance from the Rietveldrefinement of Pekov et al. (2007).

Table 2Experimental solubility data for chukanovite separates from the DFC PRB.

Experiment Time,d

pH ΣIC,ppm

ΣFe(II),ppm

I LogaHCO3

−LogaFe2+

LogKsp

A1 5 8.31 156 2.15 0.013 −1.95 −5.22 12.54B1 5 8.26 41 1.24 0.003 −2.56 −5.00 12.24A12 8 8.62 70 0.51 0.006 −2.27 −5.71 12.17B12 8 8.36 21 0.30 0.002 −2.80 −5.53 11.22A13 2 8.79 99 0.49 0.009 −2.14 −5.93 12.37B13 2 8.87 94 0.31 0.008 −2.17 −6.16 12.12A14 12 9.00 495 0.02 0.046 −1.51 −7.15 11.20B14 12 9.05 458 0.12 0.043 −1.55 −7.22 11.17

Notes: ΣIC is the dissolved inorganic carbon concentration. ΣFe(II) is the totaldissolved ferrous iron concentration. I is the calculated ionic strength.Bicarbonate and Fe2+ activities were calculated using the Geochemist's

t

values appear to be largely time-independent over the timeperiod of 2 to 5 days, which provides some evidence that asteady-state solubility was achieved from undersaturation.Also consistent values of log Ksp were determined over arange of pH and bicarbonate concentrations (Table 2). X-raydiffraction experiments at the end of the experiments

Workbench software (React module).

55T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

showed that the mineral composition did not change throughthe dissolution experiment.

The solubility reaction for chukanovite, expressed withcarbonate as the reaction product is:

Fe2ðOHÞ2CO3 þ 2Hþ⇔2Fe

2þ þ CO2−3 þ 2H2O ð4Þ

with log Ksp=1.56. The Gibbs energy of formation ofchukanovite can be computed from this value along withΔfG° for Fe2+ (−90.53 kJ/mol; Parker and Kodakovskii,1995), CO3

2− (−527.9 kJ/mol; Wagman et al., 1982), andH2O (−237.18 kJ/mol; Wagman et al., 1982). The derivedΔfG° value for chukanovite is −1174.4±6 kJ/mol.

Based on this new log Ksp value for chukanovite, a mineralstability diagram was constructed for the system Fe(II)–CO2–

H2O (Fig. 7), i.e., the Fe(II)/Fe(III) redox pair is decoupled inFig. 7. The wüstite (FeO) field was suppressed as this phase isunexpected as a low-temperature corrosion product andfurther it has not been identified in X-ray diffraction scans ofprecipitates developed on granular iron. Stability fields forsiderite, chukanovite, and ferrous hydroxide are consistentwith chukanovite (FeCO3·Fe(OH)2) occupying compositionalspace intermediate to siderite and ferrous hydroxide. Bénézethet al. (2009) recently reviewed the available laboratory-basedsolubility studies on siderite and provided a new set of tem-perature-dependent solubility measurements and derivedthermodynamic constants. At 25 °C a range of log Ksp values,from −10.25 to −11.05, have been reported for the sideritesolubility reaction:

FeCO3⇔Fe2þ þ CO

2−3 ð5Þ

Fig. 7. Mineral stability diagram for the system Fe(II)–CO2–H2O. The wüstitefield was suppressed. The speciation of dissolved iron is shown for thecondition ΣFe=10−6. Groundwater compositions for the East Helena(triangles), Elizabeth City (diamonds), and Denver Federal Center (circles)PRBs are shown. Upgradient groundwater is shown with filled symbols; in-wall compositions are indicated with open symbols.

The median value, −10.65, matches the value in thethermo.dat dataset, which was used to construct Fig. 7. Thesolubility expression for ferrous hydroxide is:

FeðOHÞ2 þ 2Hþ⇔Fe

2þ þ 2H2O ð6ÞThe log Ksp=14 value for this reaction was taken from

Feitknecht and Schindler (1963).Representative upgradient and in-wall groundwater com-

positions from the three PRB sites are plotted on Fig. 7.Upgradient groundwater from all of the sites is undersatu-rated in iron carbonates and hydroxides. Within the PRBs,groundwater is moderately alkaline and the data points fallmainly within the stability field of chukanovite or along thechukanovite/siderite boundary, with one point from the EastHelena PRB falling in the siderite stability field. This result isconsistent with the prevalence of chukanovite in granulariron PRBs and suggests that the derived thermodynamic datafor chukanovite are reasonable. The stability diagram indi-cates that ferrous hydroxide is only expected to precipitatefrom water that is highly depleted in bicarbonate/carbonateand that siderite formation requires elevated inorganiccarbon concentrations.

Agrawal et al. (2002) conducted batch experiments toexamine the reactivity of granular iron with 1,1,1-trichlor-oethane in groundwater containing dissolved carbonatespecies. In their study, siderite was reported as the primarymineral precipitate. The final pH and carbonate speciesconcentration after reaction with granular iron is notreported in this study, however, the initial pH and dissolvedcarbonate concentration was ∼5.5 and 38 mM, respectively.Inspection of Fig. 7 suggests that pH-bicarbonate trajectoriesin the batch experiments of Agrawal et al. (2002) wouldreasonably end up in the siderite stability field because of thehigh initial inorganic carbon concentration used in their batchexperiments. Thus, high groundwater concentrations ofdissolved inorganic carbon are expected to result in theformation of siderite over chukanovite.

4. Conclusions

The iron-based permeable reactive barrier (PRB) technol-ogy has gained acceptance as an alternative in-situ remedy topump-and-treat for cleaning up groundwater contaminatedwith chlorinated organic compounds, radionuclides, metals,and metalloids (Higgins and Olson, 2009). PRB technologyemploying zerovalent iron combines subsurface fluid-flowmanagement with contaminant treatment by biogeochemicalprocesses, such as reductive transformation, reductive pre-cipitation, and adsorption. Long-term success of PRBsdepends on their ability to effectively maintain theseprocesses over time. Biogeochemical processes that controliron corrosion and mineral accumulation in PRBs are keyfactors that control PRB longevity. Results reported hereindicate that chukanovite is a major precipitate that forms infield-based PRBs and the new solubility data allow for a moreaccurate prediction of mineral formation in these systems.Chukanovite is the result of reaction between groundwatercontaining dissolved carbonate species and granular iron. Theprecipitation of this phase blocks reactive sites at the iron–water interface and is expected to impede contaminant

56 T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

removal through time. Preliminary thermodynamic for thisphase are consistent with its occurrence in moderatelyalkaline PRB systems. Additional study is needed to betterunderstand whether chukanovite plays any role in contam-inant degradation processes.

Acknowledgements

We thank B. Scroggins and P. Clark for field soil coreassistance. The U.S. Environmental Protection Agencythrough its Office of Research and Development funded theresearch described here. It has not been subjected to agencyreview and therefore does not necessarily reflect the views ofthe agency, and no official endorsement should be inferred.Use of the Advanced Photon Source was supported by theU. S. Department of Energy, Office of Science, Office of BasicEnergy Sciences, under Contract No. DE-AC02-06CH11357.MR-CAT operations are supported by the Department ofEnergy and the MR-CAT member institutions. Additionallythe authors would like to thank Soma Chattopadhyay andTomohiro Shibata for their assistance with the XAS datacollection and an anonymous reviewer for providing con-structive comments. Mention of trade names or commercialproducts does not constitute endorsement or recommenda-tion for use.

References

Agrawal, A., Ferguson, W.J., Gardner, B.O., Christ, J.A., Bandstra, J.Z., Tratnyek,P.G., 2002. Effects of carbonate species on the kinetics of dechlorinationof 1, 1, 1-trichloroethane by zero-valent iron. Environmental Science andTechnology 36, 4326–4333.

Beck, F.P., Clark, P.J., Puls, R.W., 2002. Direct push methods for locating andcollecting cores of aquifer sediment and zero-valent iron from apermeable reactive barrier. Ground Water Monitoring and Remediation22, 165–168.

Bénézeth, P., Dandurand, J.L., Harrichoury, J.C., 2009. Solubility product ofsiderite (FeCO3) as a function of temperature (25–250 °C). ChemicalGeology 265, 3–12.

Butler, E.C., Hayes, K.F., 2001. Factors influencing rates and products in thetransformation of trichloroethylene by iron sulfide and iron metal.Environmental Science and Technology 35, 3884–3891.

Delany, J.M., Lundeen, S.R., 1990. The LLNL thermochemical database.Lawrence Livermore National Laboratory Report, UCRL-21658. LawrenceLivermore National Laboratory.

Erdös, V.E., Altofer, H., 1976. Ein dem Malachit ähnliches basischesEisenkarbonat als Korrosionsprodukt von Stahl. Werkstoffe und Korro-sion 27, 304–312.

Feitknecht, W., Schindler, P., 1963. Solubility constants of metal oxides, metalhydroxides and metal hydroxide salts in aqueous solution. Pure andApplied Chemistry 6, 130–199.

Furukawa, Y., Kim, J., Watkins, J., Wilkin, R.T., 2002. Formation of ferrihydriteand associated iron corrosion products in Permeable Reactive Barriers ofzero-valent iron. Environmental Science and Technology 36, 5469–5475.

Gavaskar, A., Sass, B., Gupta, N., Drescher, E., Yoon, W., Sminchak, J., Hicks, J.,Condit, W., 2002. Evaluating the longevity and hydraulic performance ofpermeable reactive barriers at Department of Defense sites. Prepared forNaval Facilities Engineering Service Center, Port Hueneme, CA.

Gu, B., Phelps, T.J., Liang, L., Dickey, M.J., Roh, Y., Kinsall, B.L., Palumbo, A.V.,Jacobs, G.K., 1999. Biogeochemical dynamics in zero-valent ironcolumns: implications for permeable reactive barriers. EnvironmentalScience and Technology 33, 2170–2177.

Helgeson, H.C., Kirkham, D.H., 1974. Theoretical prediction of the thermo-dynamic behavior of aqueous electrolytes at high pressures andtemperatures II. Debye–Hückel parameters for activity coefficients andrelative partial molal properties. American Journal of Science 274,1199–1261.

Higgins, M.R., Olson, T.M., 2009. Life-cycle case study comparison ofpermeable reactive barrier versus pump-and-treat remediation. Envi-ronmental Science and Technology 43, 9432–9438.

Huang, Y.H., Zhang, T.C., 2006. Nitrite reduction and formation of corrosioncoatings. Chemosphere 64, 937–943.

Jeen, S.W., Gillham, R.W., Blowes, D.W., 2006. Effects of carbonateprecipitates on long-term performance of granular iron for reductivedechlorination of TCE. Environmental Science and Technology 40,6432–6437.

Jeen, S.-W., Jambor, J.L., Blowes, D.W., Gillham, R.W., 2007. Precipitates ongranular iron in solutions containing calcium carbonate with trichlor-oethene and hexavalent chromium. Environmental Science and Tech-nology 41, 1989–1994.

Kamolpornwijit, W., Liang, L., Moline, G.R., Hart, T., West, O.R., 2004.Identification and quantification of mineral precipitation in Fe(0) filingsfrom a column study. Environmental Science and Technology 38,5757–5765.

Köber, R., Schlicker, O., Ebert, M., Dahmke, A., 2002. Degradation ofchlorinated ethylenes by Fe0: inhibition processes and mineral precip-itation. Environmental Geology 41, 644–652.

Kohn, T., Kane, S.R., Fairbrother, D.H., Roberts, A.L., 2003. Investigation of theinhibitory effect of silica on the degradation of 1, 1, 1-trichloroethane bygranular iron. Environmental Science and Technology 37, 5806–5812.

Kohn, T., Livi, K.J.T., Roberts, A.L., Vikesland, P.J., 2005. Longevity of granulariron in groundwater treatment processes: corrosion product develop-ment. Environmental Science and Technology 39, 2867–2879.

Kukkadapu, R.K., Zachara, J.M., Fredrickson, J.K., Kennedy, D.W., Dohnalkova,A.C., McCready, D.E., 2005. Ferrous hydroxy carbonate is a stabletransformation product of biogenic magnetite. American Mineralogist90, 510–515.

Li, L., Benson, C.H., Lawson, E.M., 2005. Impact of mineral fouling on hydraulicbehavior of permeable reactive barriers. Ground Water 43, 582–596.

Li, L., Benson, Lawson, E.M., 2006. Modeling porosity reductions caused bymineral fouling in continuous-wall permeable reactive barriers. Journalof Contaminant Hydrology 83, 89–121.

Liang, L., Korte, N., Gu, B., Puls, R., Reeter, C., 2000. Geochemical andmicrobiological reactions affecting the long-term performance of in situbarriers. Advances in Environmental Research 4, 273–286.

Mayer, K.U., Blowes, D.W., Frind, E.O., 2001. Reactive transport modeling ofan in situ reactive barrier for the treatment of hexavalent chromium andtrichloroethylene. Water Resources Research 37, 3091–3103.

Nooten, T.V., Lieben, F., Dries, J., Pirard, E., Spingael, D., Bastiaens, L., 2007.Impact of microbial activities on the mineralogy and performance ofcolumn-scale permeable reactive iron barriers operated under twodifferent redox conditions. Environmental Science and Technology 41,5724–5730.

O'Day, P.A., Rivera, N., Root, R., Carroll, S.A., 2004. X-ray absorptionspectroscopic study of Fe reference compounds for the analysis ofnatural sediments. American Mineralogist 89, 572–585.

Parbs, A., Ebert, M., Dahmke, A., 2007. Long-term effects of dissolvedcarbonate species on the degradation of trichloroethylene by zerovalentiron. Environmental Science and Technology 41, 291–296.

Parker, V.B., Kodakovskii, I.L., 1995. Thermodynamic data of the aqueous ions(2+ and 3+) of iron and the key compounds of iron. Journal of Physicaland Chemical Reference Data 24, 1699–1745.

Paul, C.J., McNeil, M.S., Beck Jr., F.P., Clark, P.J., Wilkin, R.T., Puls, R.W., 2003.Capstone report on the application, monitoring, and performance ofpermeable reactive barriers for ground-water remediation. Long-termmonitoring of PRBs: Soil and Ground Water Sampling. EPA/600/R-03/045b. Office of Research and Development, Cincinnati, OH. 133 pp.

Pekov, I.V., Perchiazzi, N., Merlino, S., Kalachev, V.N., Merlini, M., Zadov, A.E.,2007. Chukanovite, Fe2(CO3)(OH)2, a new mineral from the weatherediron meteorite Dronino. European Journal of Mineralogy 19, 891–898.

Phillips, D.H., Gu, B., Watson, D.B., Roh, Y., Liang, L., Lee, S.Y., 2000.Performance evaluation of a zerovalent iron reactive barrier: mineral-ogical characteristics. Environmental Science and Technology 34,4169–4176.

Ravel, B., Newville, M., 2005. Athena, Artemis, Hephaestus: data analysis forX-ray absorption spectroscopy using IFEFFIT. Journal of SynchrotronRadiation 12, 537–541.

Reardon, E.J., 2005. Zerovalent irons: styles of corrosion and inorganiccontrol on hydrogen pressure buildup. Environmental Science andTechnology 39, 7311–7317.

Rehr, J.J., 1993. Recent developments in multiple-scattering calculations ofXAFS and XANES. Japan Journal of Applied Physics 32, 8–12.

Rehr, J.J., Albers, R.C., Zabinksky, S.I., 1992. High-order multiple-scatteringcalculations of X-ray absorption fine structure. Physical Review Letters69, 3397–3400.

Wagman, D.D., Evans, W.H., Parker, V.B., Schumm, R.H., Halow, I., Bailey, S.M.,Churney, K.L., Nuttal, R.L., 1982. The NBS tables of chemical thermody-namic properties. Selected values for inorganic C1 and C2 organicsubstances in SI units. Journal of Physical and Chemical Reference Data11, 1–392.

57T.R. Lee, R.T. Wilkin / Journal of Contaminant Hydrology 116 (2010) 47–57

Wilke, M., Farges, F., Petit, P.-E., Brown, G.E., Martin, F., 2001. Oxidation stateand coordination of Fe in minerals: an Fe K-XANES spectroscopic study.American Mineralogist 86, 714–730.

Wilkin, R.T., 2006. Mineralogical Preservation of Solid Samples Collectedfrom Anoxic Subsurface Environments. EPA/600/R-06/112. Office ofResearch and Development, Cincinnati, OH. 7 pp.

Wilkin, R.T., Puls, R.W., 2003. Capstone Report on the Application, Monitoringand Performance of Permeable Reactive Barriers for Ground WaterRemediation. Performance Evaluations at Two Sites. EPA/600/R-03/045a.Office of Research and Development, Cincinnati, OH. 140 pp.

Wilkin, R.T., Su, C., Ford, R.G., Paul, C.J., 2005. Chromium-removal processesduring groundwater remediation by a zerovalent iron permeablereactive barrier. Environmental Science and Technology 39, 4599–4605.

Wilkin, R.T., Acree, S.D., Ross, R.R., Beak, D.G., Lee, T.R., Paul, C.J., 2008. FieldApplication of a Permeable Reactive Barrier for Treatment of Arsenic inGround Water. EPA/600/R-08/093. Office of Research and Development,Cincinnati, OH. 88 pp.

Zhang, Y., Gillham, R.W., 2005. Effects of gas generation and precipitates onperformance of Fe0 PRBs. Ground Water 43, 113–121.