Fe(III) Carbonate Green

Immobilization of chromate in hyperalkaline waste streams by green rusts and zero-valent iron. Christine M. Rogers , Ian T. Burke , Imad A.M. Ahmed , ...
19 downloads 0 Views 205KB Size
Download PDF
Environ. Sci. Technol. 2004, 38, 4587-4595

Reduction of Aqueous Chromate by Fe(II)/Fe(III) Carbonate Green Rust: Kinetic and Mechanistic Studies LUDOVIC LEGRAND,* ALAAEDDINE EL FIGUIGUI, FLORENCE MERCIER, AND ANNIE CHAUSSE Laboratoire Analyse et Environnement, UMR 8587 CNRSs Universite´ d’Evry Val d’Essonne-CEA, Universite´ d’Evry Val d’Essonne, Baˆtiment des Sciences, 1 Rue du Pe`re Jarland, 91025 Evry Cedex, France

toxic and potentially carcinogens; they occur mainly as CrO42anions, which are mobile in the environment. In contrast, CrIII species has a limited hydroxide solubility and forms strong complexes with minerals. Previous research about the CrVI reduction by Fe metal has resulted in understanding the surface reactions, kinetics, and mechanisms (5-9). But Fe corrodes in aqueous solution by forming oxidized solid products that can alter the long time performances of PRBs. Moreover, Fe(II)-bearing minerals and dissolved Fe(II) species are present in the environment and can interfere in the CrVI reduction. So, studies have been extended to Fe(II) or Fe(II)-Fe(III) products in order to evaluate their contribution in the reduction of redox active contaminants (10-20). Reduction of CrVI in the presence of both aqueous (10, 11) and solid Fe(II) species (12-14) can be written as

Cr(VI) + 3Fe(II) f Cr(III) + 3Fe(III)

This work describes the heterogeneous reaction between FeII in carbonate green rust and aqueous chromate, in NaHCO3 solutions at 25 °C, and at pH values of 9.3-9.6. Evidence for reduction of CrVI to CrIII and concomitant solidstate oxidation of lattice FeII to FeIII was found from FeII titration and from structural analysis of the solids using FTIR, XRD, SEM, and XPS methods. Results indicate the formation of ferric oxyhydroxycarbonate and the concomitant precipitation of CrIII monolayers at the surface of the iron compound that induce passivation effects and progressive rate limitations. The number of CrIII monolayers formed at the completion of the reaction depends on {FeII}t)0, the molar concentration of FeIIsolid at t ) 0; on {no}t)0, the molar concentration of reaction sites present at the surface of the solid phase at t ) 0; and on [CrVI]t)0, the molar concentration of CrVI at t ) 0. Kinetic data were modeled using a model based on the formation of successive CrIII monolayers, -(d[CrVI]/dt) ) Σ1jki{S}[CrVI]({ni-1} - {ni}) with ki{S} (in s-1 L mol-1), the rate coefficient of formation of CrIII monolayer i, and {ni} and {ni-1}, the molar concentration of CrIII precipitated in monolayer i and monolayer i - 1, respectively. Good matching curves were obtained with kinetic coefficients: k{S} ) 5-8 × 10-4, 1 {S} {S} -5 k2 ) 0.5-3 × 10 , and k3 about 1.7 × 10-6 s-1 m-2 L. The CrVI removal efficiency progressively decreases along with the accumulation of CrIII monolayers at the surface of carbonate green rust particles. In the case of thick green rust particles resulting from the corrosion of iron in permeable reactive barriers, the quantity of FeII readily accessible for efficient CrVI removal should be rather low.

Introduction Permeable reactive barriers (PRBs) containing zerovalent iron have been identified as an effective means of in-situ remediation of contaminated groundwaters with chlorinated organic solvents and redox active metals (1-4). Chromium is one of the most common contaminant metals in industrial regions due to its use as a metal corrosion inhibitor or leather tanning. Chemical reduction of CrVI to CrIII via PRBs offers an attractive alternative for reducing the mobility and toxicity of chromium species. Indeed, CrVI species are known to be * Corresponding author phone: +33 (0)1 69 47 77 05; fax: +33(0)1 69 47 76 55; e-mail: [email protected]. 10.1021/es035447x CCC: $27.50 Published on Web 07/28/2004

 2004 American Chemical Society

(1)

CrVI trapping by green rusts was also investigated (1518). Green rusts have been identified in anoxic soils and ferrous-rich groundwaters; they are now recognized as an important intermediate species in the corrosion of iron. Among them, carbonate green rust or GR(CO32-), a potential source of FeII, is reported as having the general formula [FeII4FeIII2(OH)12]‚[CO3‚mH2O] with m likely equal to 2 and a crystal structure resembling that of pyroaurite (21-23). Works on the redox reaction have focused on characterizing end products and thermodynamic parameters and studying the reaction kinetics for different conditions. Williams and Scherer (17) and Bond and Fendorf (18) have reported that the rate of CrVI reduction was proportional to the green rust surface area concentration and was pseudo-first-order with respect to CrVI concentration. At high CrVI concentrations, a decrease of the reaction rate was observed by Williams and Scherer; this rate limitation was attributed to the surface passivation by accumulation of precipitates on the green rust surface or to depletion of available FeII within the green rust structure (19). Analyses of the resulting Cr(III)-bearing solids are not well-achieved but their nature is of utmost importance because they will govern the subsequent behavior of chromium in the environment. On one hand, LoyauxLawniczak et al. (15) proposed the formation of a poorly ordered chromium(III)-iron(III) oxyhydroxide similar to the “2-line ferrihydrite”. But on the other hand, Kendelewicz et al. (20) proposed the incorporation of CrIII in an overlayer when CrVI is reduced by magnetite according to a heterogeneous redox process; the thickness of this overlayer depends on the reaction time. The build-up of chromium compounds on the iron surface that acts as a diffusion barrier for CrVI reduction was also proposed from electrochemical data relative to iron in contact with aqueous solutions containing CrVI (8). The present study investigates the hypothesis that decreasing reaction rates between CrVI and carbonate green rust are due to a surface passivation. This objective was accomplished by combining chemical titration with surface analysis method and characterization methods of solids: X-ray diffraction (XRD), scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), and X-ray photoelectron spectroscopy (XPS).

Materials and Methods Synthesis of Carbonate Green Rust. Suspensions of carbonate green rust were obtained from the aerial oxidation of Fe(II) suspensions at 25 °C according to the procedure already described in ref 24. Fifty milliliters of 0.4 M NaHCO3 solution VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4587

TABLE 1. Initial Quantities of Carbonate Green Rust (GR(CO32-)) and Chromate (or H2O2) Involved in Experiments 1-8a

expt 1 2 1 3 4 5 1 6 7

Figure 2 Figure 3 Figure 3 Figure 3 Figure 4 Figure 4 Figure 4 Figure 4 Figure 4

expt 8

Figure 5b

n Fe in GR(CO32-) (µmol)

n CrO42(µmol)

R ratio, {FeII}/[CrVI]

n CrO42consumed at t ) 0.5 min (µmol)

500 500 500 500 1000 750 500 250 0

111 78 111 151 111 111 111 111 111

3.00b 4.27 3.00b 2.21 6.00 4.50 3.00b 1.50 0

∼40 ∼40 ∼40 ∼40 ∼80 ∼55 ∼40 ∼28 ∼0 (control)

n Fe in GR(CO32-) (µmol)

n H2O2 (µmol)

ratio {FeII}/[H2O2]

n H2O2 consumed at t ) 0.5 min (µmol)

500

167

2.00b

∼135

a

Quantity of chromate (or H2O2) consumed during first 0.5 min. The volume of solution is 50 mL. Carbonate green rust contains 2/3FeII. b Stoichiometric ratio with respect to reactions: 3Fe(II) + Cr(VI) f 3Fe(III) + Cr(III) or 2Fe(II) + H2O2 f 2Fe(III) + H2O.

(prepared with NaHCO3 from Fluka, purity 99.5% and 18 MΩ‚cm nano-pure water) was introduced in a cylindrical glass cell (about 50 mm diameter) and stirred (300 rpm). Prior to Fe(II) addition, the solution was deaerated with argon and 10 M NaOH was added dropwise to get a pH value of 9.5. A total of 0.5 mL (10-2M) of a 1 M FeCl2 solution was then introduced into the solution. The cell was opened to air, and the oxidation reaction was monitored by recording pH and potential of the solution. No significant change of pH was recorded during the preparation of carbonate green rust suspension due to the low Fe(II)/NaHCO3 ratio. During the oxidation of Fe(II) into carbonate green rust suspension, the potential values remained rather stable, about -0.4 VSHE. The completion of the oxidation was revealed by the sharp increase of the potential. As E reached -0.25 VSHE, the cell was closed and argon was bubbling in the solution. The oxidation time required to get the carbonate green rust suspension was 25 ( 2 min. Moreover, EDTA (ethylenediaminetetraacetic acid) complexing titration of the solution after filtration of the carbonate green rust suspension did not give any detection of iron species (detection limit less than 2 × 10-4 M), indicating that all Fe atoms were present within the carbonate green rust structure (500 µmol). The following procedure was used to determine the FeII content in carbonate green rust. Once the potential reached -0.25 VSHE, the cell was closed and argon was bubbling through the solution. The green rust suspension was then dissolved by acidification with 10 M H2SO4 solution down to a pH around 0.5. Titration of Fe2+ ions was finally done with a 10-2 M KMnO4 solution. CrVI/GR(CO32-) Redox Reaction. Experiments were initiated by adding to the green rust suspensions a volume of 0.111 M Na2CrO4 solution (prepared with Na2CrO4 from Fisher, purity >99.5% and 18 MΩ ‚m nano-pure water) as indicated in Table 1 under argon atmosphere. The temperature of the cell was maintained at 25 °C during the CrVI/ GR(CO32-) reaction. The pH was remaining in the 9.3-9.6 range during the oxidation experiments. Kinetics of the CrVI reduction were studied by monitoring the time-dependent concentration of this species. At the chosen times, solid products were quickly separated from the solution by collecting on 0.22 µm PVDF filters under argon atmosphere. 4588

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004

Acidification of the solution until pH 0.5 was then done by addition of a 10 M H2SO4 solution. CrVI concentration was finally determined by titration with a 2 × 10-2 M FeCl2 solution. Potentiometric titration was monitored using the same electrodes and the same material as for the green rust preparation. The equivalence was revealed by a potential drop. Oxidation of GR(CO32-) by H2O2 or by Dissolved Oxygen. The oxidation of carbonate green rust by H2O2 was performed by adding 1.67 mL of a 0.1 M H2O2 solution to the GR(CO32-) suspension under argon. At various oxidation times, the solid phase was quickly separated from the solution by filtration, rinsed with deaerated 18 MΩ‚cm nano-pure water, and then titration of the remaining FeII by 0.01 M KMnO4 was done after digestion of the solid by 1 M H2SO4. The oxidation of carbonate green rust by dissolved oxygen was investigated by determining the remaining FeII present in the suspension at various aeration times. After the synthesis of GR(CO32-) suspension, the cell was maintained open to air. At chosen times, the oxidation was stopped by closing the cell and purging with argon. Then, the solution was acidified with 10 M H2SO4 down to a pH around 0.5 to induce the dissolution of the solid phase. Titration of Fe2+ ions was done with a 0.01 M KMnO4 solution. Characterization of Solids. After filtration, solid products were rinsed with deaerated 18 MΩ‚cm nano-pure water and then dried under argon flow before characterization. XRD measurements were carried out by using a Bruker D8 diffractometer with a Cu KR radiation (1.5406 Å). The carbonate green rust sample was protected from air oxidation by admixing glycerol according to Hansen’s procedure (22). SEM examinations were obtained using a Philips XL 30 microscope. FTIR data were recorded on a Bruker IFS 28 FTIR spectrometer. Solids were pressed as pellets with KBr and analyzed by direct transmission mode. Twenty scans were done for the spectrum acquisition, and the time duration was about 30s. XPS experiments were carried out by using a VG Escalab MKII spectrometer with an unmonochromated Al KR X-ray source (1486.6 eV). A source power of 20 kV and of 10 mA was used. The pressure in the analysis chamber was below 10-9 Pa. Calibration in binding energy (BE) was achieved using the Ag 3d3/2 and Ag 3d5/2 transitions (368.3 and 374.3 eV respectively). The BE resolution of the spectrometer was estimated to 0.2 eV. Solids as powders were deposited on an ultrahigh vacuum double-tape scotch until its complete coverage. The binding energy values were corrected using the C1s signal of adventitious carbon at 285 eV.

Kinetic Background Homogeneous Reaction. The reduction of aqueous CrVI by aqueous FeII can be described by the overall reaction:

CrVI(aq) + 3FeII(aq) f CrIII + 3FeIII

(2)

The reaction should depend on R, which is defined as the ratio between the molar concentration of aqueous FeII at t ) 0 and the molar concentration of CrVI at t ) 0; R ) ([FeII]t)0/ [CrVI]t)0). This reaction was studied by Eary and Rai (10) in solutions with pH values ranging from about 3 to 9. A mixed (chromium, iron) hydroxide was obtained from the simultaneous precipitation of CrIII and FeIII. Heterogeneous Reaction with Precipitation of CrIII on Reaction Sites. The reduction of aqueous CrVI by an FeIIcontaining solid phase can be described by the overall reaction CrVI(aq) + 3FeII(s) + s f CrIII(precipitated on reaction site s) + 3FeIII(s), with s designating reaction site. Two parameters have to be used: R ) ({FeII}t)0/[CrVI]t)0) and R1, which is defined as the ratio between the molar concentration of CrVI at t ) 0 and the molar concentration of reaction sites present

(iii) R1 > 1. The reaction stops as the molar concentration of available reaction sites reaches zero. The remaining CrVI ions are not allowed to react with lattice FeII since it would lead to the precipitation of a second CrIII monolayer on the first one. On the other hand, the kinetic constant related to the formation of the second monolayer is zero. Heterogeneous Reaction with Precipitation of Successive CrIII Monolayers. A general model for the carbonate green rust/chromate reaction should account for the accumulation of successive monolayers of CrIII precipitate at the surface of green rust particles along with the decrease of chromate concentration in solution: j

v)

∑

VI k{S} i [Cr ]({ni-1}

- {ni}) ) -

d[CrVI]

1

(5)

dt

The percentage of surface coverage (θi) is expressed as θ1 ) ({n1}/{no}t)0), θ2 ) ({n2}/{n1}), ..., θj ) ({nj}/{nj-1}). Equation 5 can be rewritten in a more appropriate form for numerical integration:

d{n1} VI ) k{n} 1 [Cr ]({no}t)0 - {n1}) dt d{n2} VI ) k{n} 2 [Cr ]({n1} - {n2}) dt ... FIGURE 1. (a) Theoretical normalized CrVI decays as calculated -1 L mol-1, R ) 3, and R ranging from from eq 3 with k{n} 1 1 ) 0.02 s 0 to 1.3. (b) Theoretical normalized CrVI decays as calculated from {n} {n} eq 5 with k1 ) 0.1 s-1 L mol-1, k2 ) 0.01 s-1 L mol-1, k{n} 3 ) 0.002 s-1 L mol-1, R ) 3 (stoichiometric ratio), R1 ) 3, and j ) 3. at the surface of the solid phase at t ) 0, R1 ) ([CrVI]t)0/ {no}t)0). By assuming that (i) the reaction is first order with respect to chromate concentration, (ii) it is also first order with respect to the number of available sites, (iii) one precipitated CrIII occupies one reaction site (this assumption can reasonably be admitted since the ionic radius of Cr3+ (0.063 nm) is very close to that of Fe3+ (0.064 nm), the second order reaction rate can be schematically written as {n} VI VI v ) k{n} 1 [Cr ]{no} ) k1 [Cr ]({no}t)0 - {n1}) ) VI k{n} 1 [Cr ]{no}t)0(1 - θ1) ) -

d[CrVI] (3) dt

where (1 - θ1) is the percentage of free surface sites. θ1 ) ({n1}/{no}t)0) ) ([CrVI]t)0 - [CrVI]t/{no}t)0). {no}t)0 is related to the surface area concentration {S} and to the mass concentration {m} through the two following equations, {no}t)0 ) ({S}/AsN) and {no}t)0 ) (A{m}/AsN), respectively. Figure 1a gives the theoretical normalized [CrVI] decays obtained from eq 3 with k{n} ) 0.02 s-1 L mol-1, R ) 3 1 (stoichiometric ratio), and R1 ranging from 0.1 to 1.3. (i) R1 f 0: (1 - θ1) f 1. The reaction rate can be expressed as VI VI v ) k{n} 1 [Cr ]{no}t)0 ) kobs[Cr ] ) -

d[CrVI] dt

(4)

corresponding to a pseudo-first-order reaction. (ii) R1 < 1. The molar concentration of reaction sites at t ) 0, {no}t)0, is greater than the molar concentration of CrVI at t ) 0, [CrVI]t)0. At t ) ∞, the percentage of sites coverage is equal to R1. The time corresponding to the completion of the reaction increases with R1.

d{nj} VI ) k{n} j [Cr ]({nj-1} - {nj}) dt 1 {n1} + {n2} + ... + {nj} ) ([CrVI]t)0 - [CrVI]) V

(6)

Figure 1b gives the theoretical normalized [CrVI] decays obtained from eq 5 with k{n} ) 0.1 s-1 L mol-1, k{n} ) 0.01 1 2 {n} -1 -1 -1 s L mol , k3 ) 0.002 s L mol-1, R ) 3 (stoichiometric ratio), R1 )3, and j ) 3. The insert in Figure 1b gives the evolutions of θi values as a function of time.

Results Characterization of Carbonate Green Rust. Carbonate green rust was analyzed through XRD, FTIR, SEM, and Fe(II) titration. XRD pattern (data not shown) exhibits only lines of this compound. Unit-cell parameters, a ) 3.18 Å and c ) 22.71 Å, are consistent with those already reported in the literature (21, 25-27). The average FeII content in our carbonate green rust was deduced from the titration of four samples, 66% ( 2% of Fe(II), a value consistent with the Fe(II)/Fe(III) ratio of 2 reported in the literature (21, 22). The FTIR spectrum of our carbonate green rust (see spectrum a in Figure 6) is also consistent with those already reported in the literature (25, 27). Following bands are observed: OH stretching modes at 3460 and 3350 cm-1; H-bounded OH at about 3090 cm-1; H2O bending vibration at 1630 cm-1; OH deformation at 1550 cm-1; CO32- stretching modes at 1350; and Fe-OH lattice modes at 845, 768, 510, and 480 cm-1. SEM image (data not shown) reveals hexagonal platy crystals of about 100-300 nm diameter. This shape is typical of green rust particles (27-29). The average thickness was too low to be accurately determined but it is less than 40 nm. Kinetics of Cr(VI) Reduction by Carbonate Green Rust. Table 1 summarizes the initial moles of carbonate green rust and chromate used in the experiments performed in this work. Figure 2 gives the CrVI moles remaining in solution, as a function of the interaction time with carbonate green rust; stoichiometric conditions were first chosen, referred to as expt 1 in Table 1. The complete reduction of CrVI lasts for VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4589

FIGURE 2. Evolution of CrVI mole number (open circles) as a function of the interaction time in expt 1: 500 µmol of Fe in GR(CO32-) and 111 µmol of CrVI. Theoretical curves calculated according to the pseudo-first-order model (full lines) given in eq 4 with k{m} ) 3.34 1 × 10-2 (from ref 18), 2.2 × 10-2 (from ref 17), 10-3, and 3 × 10-4 s-1 L g-1. Theoretical curves calculated according to the second-order model (dashed lines) given in eq 3 with k{m} ) 2 × 10-3 and 6 × 1 10-4 s-1 L g-1 and with R1 ) 1. about 1000 min. A sharp drop corresponding to the reduction of about 40 µmol of CrVI is first observed from 0 to about 0.5 min. Then, the rate significantly decreases. The further removal of 40 µmol of CrVI extends within about 40 min (from 70 to 30 µmol). Finally, the consumption of the remaining 30 µmol requires an important time duration, ∼1000 min. Results from the titration of remaining FeII in solid samples and CrVI in solution were consistent with the electron balance given in eq 1. The experimental data reported in Figure 2 were tentatively modeled by using the second order (dashed lines) and pseudo-first-order (full lines) models given in eqs 3 and 4. All attempts were unsuccessful, and the rate coefficient values were chosen in order to clearly show that it was impossible to reach a satisfactory matching. It is to be noted that the pseudo-first-order curves calculated from the kinetic coefficients reported by Bond and Fendorf (18) and by Williams and Scherer (17) should give an approximate matching only for time less than about 1 min. In our experimental conditions, pseudo-first-order and secondorder models can unequivocally be discarded. We have next focused our studies on the first hour of the interaction in order to clarify the influence of R ratio on the redox reaction rate. Figure 3 reports the CrVI removal as a function of the interaction time in experiments with 500 µmol of Fe in carbonate green rust and different CrVI moles (referred as expts 1-3 in Table 1). A similar evolution is displayed whatever the initial CrVI mole number introduced into the carbonate green rust suspension. An initial rapid drop, corresponding to the same removal of CrVI, ∼40 µmol (see Table 1), is obtained. The rate of the CrVI reduction then decreases and the time where the complete CrVI reduction or FeII removal should be attained depends on R. Curve (d) will be commented later. To test whether varying the initial quantity of carbonate green rust would affect the kinetics, experiments were carried out with 111 µmol of CrVI and various initial quantities of carbonate green rust. Figure 4 reports the CrVI decays obtained with 500 (referred to expt 1 in Table 1) and 1000 µmol (referred to as expt 4 in Table 1) of Fe in green rust. The quantity of CrVI removed within the first 0.5 min is doubled as the initial mole number of Fe is made 2-fold higher. Moreover, the complete removal of CrVI proceeds within about 20 min for R ) 6 against about 1000 min for R ) 3. Extra experiments 4590

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004

FIGURE 3. Evolution of CrVI mole number as a function of the interaction time in expts 1-3. The initial quantity of Fe in GR(CO32-) is 500 µmol: (a) R ) 3, (b) R ) 2.21, and (c) R ) 4.27 (see Table 1 for details). (d) CrVI decay obtained with partially oxidized carbonate green rust by dissolved oxygen (see text for details).

FIGURE 4. Evolution of CrVI mole number as a function of the interaction time in expts 1 and 4. Initial CrVI quantity: 111 µmol and various R ratios (see Table 1 for details). Insert: Chromate removal within the first 0.5 min as a function of the initial quantity of (GR(CO32-). Initial CrVI quantity ) 111 µmol and various R ratios (see Table 1 for details). were done (referred to as expts 5-7 in Table 1), and we plotted the quantity of chromate remaining at t ) 0.5 min as a function of the initial mole number of Fe in carbonate green rust. A linear relationship is observed (see insert in Figure 4). The slope is 0.082 ( 0.002 mol of CrVI with respect to 1 mol of Fe in carbonate green rust. Assuming that the specific surface area of the carbonate green rust is the same in all experiments, this result implies that the reduction of CrVI into CrIII strongly depends on the surface area displayed by carbonate green rust particles. Comparison with Other Strong Oxidizing Species. Experiments were also done with oxidizing species that have a standard potential higher than 1 V: dissolved oxygen O2 (E° O2/H2O ) 1.23 VSHE) and hydrogen peroxide H2O2 (E° H2O2/H2O ) 1.78 VSHE). Both species are reduced into H2O, a soluble reduction product. Figure 5 reports the data obtained for the oxidation of 500 µmol of Fe in carbonate green rust by chromate (referred to as expt 1 in Table 1, stoichiometric conditions), by hydrogen peroxide (referred

FIGURE 5. Disappearance of FeII in carbonate green rust as a function of the interaction time with various strong oxidizing species. Initial quantity of Fe in GR(CO32-) equal to 500 µmol, i.e., 333 µmol of FeII. Oxidizing species: (a) chromate, expt 1; (b) H2O2, expt 8; and (c) dissolved oxygen. to as expt 8 in Table 1, stoichiometric conditions), and by dissolved oxygen (progressive supply of oxygen by stirring of suspension in contact with air). As already mentioned, the completion of the CrVI/GR(CO32-) reaction is obtained after about 1000 min. With H2O2, the reaction is much faster since less than 10 min is sufficient to get a complete FeII removal. With dissolved O2, the reaction lasts for 120 min. The linear decrease of FeII mole number with time indicates that the reaction rate remains constant. The oxidation might depend on the transfer of oxygen from air to solution. Effect of GR(CO32-) Preoxidation on the Reaction Rate. A total of 500 µmol FeII was converted to GR(CO32-) suspension following the procedure described above. After the synthesis, the cell was maintained open to air for 38 min to obtain a partial solid-state oxidation of GR(CO32-) suspension. This time corresponds to the oxidation of 120 µmol of FeII in GR(CO32-) (see Figure 5c); the latter value is the quantity of FeII that should have been oxidized if the reaction was done with 40 µmol of CrVI. Then, the cell was placed under argon and 71 µmol of CrVI was added to the solution. The CrVI decay as a function of time is reported in Figure 3d. The chromate removals within 1.5 min were calculated from Figure 3a (from 0.5 to 2 min), Figure 3d (from 0 to 1.5 min), and Figure 3c (from 0 to 1.5 min). We obtained 6, 17, and 49 µmol, respectively. The efficiency of the removal decreases in the following manner: GR(CO32-) > GR(CO32-) partially oxidized by dissolved O2 > GR(CO32-) partially oxidized by chromate ions. Characterization of Solid Ferric Product Resulting from Oxidation of Carbonate Green Rust. The solid phases sampled at different interaction times in expt 1 were analyzed through FTIR and XRD measurements. Figure 6 reports the FTIR spectra of solids sampled at 0, 7, 180, and 930 min. Increasing the interaction time induces a decrease of the bands relative to the carbonate green rust (see above). At the same time, new bands corresponding to a ferric product that we already identified in previous studies (24, 30) as a ferric oxyhydroxycarbonate compound increase. The bands at 470 and 655 cm-1 are assigned to the lattice Fe-O mode. The presence of carbonate is revealed by the adsorption bands at 690, 840, 1080, 1360, and 1490 cm-1. Modifications in the 3000-3500 cm-1 region corresponding to OH stretching are also observed. The FTIR spectra of the solid ferric products resulting from the oxidation of carbonate green rust by H2O2

FIGURE 6. FTIR spectra of the solid products sampled at various times in expt 1: (a) 0, (b) 7, (c) 180, and (d) 930 min.

FIGURE 7. XRD pattern of the solid product (ferric oxihydroxicarbonate) sampled at t ) 930 min in expt 1, corresponding to the completion of the reaction. Step size 0.05° 2θ; step time 30 s. or by dissolved O2 were also recorded (spectra not shown). The same ferric product is formed (i.e., ferric oxyhydroxycarbonate). The XRD pattern of the solid-phase sampled after the complete oxidation of the carbonate green rust by CrVI (Figure 7) displays main lines at 7.36 and 2.53 Å, consistently with the formation of ferric oxihydroxicarbonate (24, 25, 30, 31). XPS Analysis of the Solid Product. Cr-containing species could not be identified in solid products by using FTIR and XRD, mainly due to their low amount in the solid samples. XPS measurements were therefore carried out since this technique is very convenient to analyze surface species and to discriminate CrVI from CrIII. Solid samples corresponding to four interaction times in expt 1 were analyzed, respectively 0, 0.5, 45, and 1000 min, corresponding to the reduction of 0, 40, 80, and 111 µmol of CrVI. After reaction, the solids were very carefully rinsed, dried under argon flow, and finally left for several hours in contact with air; the latter treatment provoked the complete oxidation of the remaining lattice VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4591

FIGURE 8. C 1s, Cr 2p, Fe 2p, and Fe 3p regions displayed by the XPS spectra of the solid product sampled at t ) 0.5 min in expt 1. Fe2+ ions into Fe3+ ions via a solid-state reaction (24-31). Figure 8 gives the C 1s, Cr 2p, Fe 2p, and Fe 3p regions of the XPS spectra relative to the solid sampled at t ) 0.5 min. The XPS peak positions of the other solid samples are the same as those of the solid sampled at 0.5 min (note that chromium peaks are not observed for the solid sampled at t ) 0 min, as expected). Assuming 285 eV as a reference binding energy for C 1s from adventitious carbon, the Cr 2p3/2 binding energy is 577.1 ( 0.2 eV. The binding energy of the Cr 2p3/2 line is reported at about 580 eV in Cr(VI) salts (32). It extends from 576.2 to 577.5 eV for Cr(III) compounds such as Cr2O3, CrOOH, or Cr(OH)3 (33). One can state that chromium is present as a solid Cr(III)-bearing product. This statement is consistent with the spin-orbit splitting between 2 p1/2 and 2 p3/2 that is 9.9 eV, a value typical of Cr(III) compounds (34). The Fe 2p3/2 (711.7 ( 0.2 eV) and Fe 3p (56.0 ( 0.2 eV) peaks are characteristic of iron(III) oxyhydroxide compounds (35). The remaining of carbonate anions inside the lattice of the ferric product resulting from the oxidation of carbonate green rust is revealed by the C 1s peak at about 289.5 eV observed in all of the spectra (32). Figure 9 reports the normalized intensities of Cr 2p3/2 and Fe 2p3/2 peaks worked out from XPS spectra, as a function of the Cr(VI) removal. A quasi-linear increase of the Cr 2p intensities is observed. This result mainly indicates that the whole quantity of Cr(III) resulting from the Cr(VI) reduction by carbonate green rust is present in the solid phase; the concentration of Cr(III) as a soluble species is therefore very negligible, consistently with the results from Williams and Scherer (17). The Fe 2p3/2 intensities decrease with the interaction time (i.e., with the increase of the Cr(III) amount in the solid phase). The attenuation of the Fe peaks might arise from a progressive coverage of the carbonate green rust surface by a Cr(III) precipitate, as a result of the heterogeneous green rust/chromate reaction. The detection of Fe peaks in all of the spectra suggests that the thickness of the Cr(III) layer remains thinner than the depth probed by XPS.

Discussion End Products of Reaction. Carbonate green rust oxidizes into ferric oxyhydroxycarbonate, as shown by FTIR and XRD measurements. Oxidation proceeds via a solid-state reaction 4592

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004

FIGURE 9. Evolutions of the intensities of XPS peaks as a function of the quantity of chromate consumed for solid products obtained from expt 1. The sampling times are 0, 0.5, 45, and 1000 min, corresponding to the consumption of 0, 40, 80, and 111 µmol of chromate, respectively. combining electron transfer (lattice Fe2+ into Fe3+ transformation) and deprotonation of the water molecules in the interlayers and of some of the hydroxyl groups in the Fe(O,H) octaedra sheets (24, 30). The following formula may be proposed for ferric oxihydroxicarbonate: FeIII6O(2+n)(OH)(12-2n)(H2O)nCO3 with n between 0 and 2 (24). Use of XPS allows surface analysis of carbonate green rust after its reaction with chromate ions. Results suggest that the chromate removal from solution occurs through a CrVI into CrIII reduction process leading to the precipitation of a Cr(III) compound. An accurate formula for the Cr(III) precipitate cannot be established from XPS results, only a general formula is proposed, CrIII(O)x(OH)y(H2O)z with 2x + y ) 3. Finally, the schematic global reaction can be proposed:

3[FeII4FeIII2(OH)12]‚[CO3‚2H2O] + 4CrO42- f 3FeIII6O(2+n)(OH)(12-2n)(H2O)nCO3 + 4CrIII(O)x(OH)y(H2O)z (7) Passivation of Carbonate Green Rust by Cr(III). Several authors have postulated the formation of a poorly crystallized mixed chromium(III)-iron(III) oxyhydroxide as the solid product resulting from the reaction of chromate ions with soluble Fe(II) species (10-12), sulfate green rust, and chloride green rust (15). Whereas the formation of such mixed oxyhydroxide from a redox reaction between two soluble species is easily understandable, it is more difficult to consider it in the case of a heterogeneous reaction between aqueous chromate ions and solid particles. As an example, results given by Kendelewicz et al. (20) relative to the reaction of aqueous CrVI with magnetite (Fe3O4) can be cited. These authors proposed the solid-state oxidation of magnetite into maghemite (γ-Fe2O3) and the concomitant precipitation of chromium(III) oxyhydroxide as an overlayer at the surface of the particles. The oxidation of carbonate green rust by CrVI, as well as by H2O2 or dissolved oxygen (at pH near 9.5), involves a solid-state reaction (i.e., a reaction occurring without a dissolution step of carbonate green rust prior to

oxidation) (24). The CrVI reduction at the carbonate green rust/solution interface is associated to the precipitation of chromium(III) oxyhydroxide at the surface of particles. The initial rapid reduction within the first 0.5 min occurs with readily accessible FeII at the carbonate green rust/ solution interface. The subsequent reduction of CrVI by FeII requires the electron transfer across the chromium(III) and ferric oxyhydroxycarbonate overlayers, from bulk FeII to chromate ions. The large diminution of the reaction rates after the first 0.5 min suggests that these overlayers generate passivation effects and/or that they significantly hinder the sorption of chromate ions. When H2O2 or dissolved oxygen are used as oxidizing species, the passivation effects are much lower since the reduction product is H2O. Surface Sites on Green Rust Particles. The unit cell parameters of carbonate green rust are a ) b ) 3.18 Å, c ) 22.71 Å, R ) β ) 90°, and γ ) 120°; the following formula can be used [FeII4FeIII2(OH)12][CO3‚2H2O], with the corresponding molar mass, M ) 635 g mol-1 (24, 31). Taking into account these data, the volume of the rhombohedral unit cell, the surface site area, and the density of GR(CO32-) can be computed as 199 Å3, 8.76 Å2, and 2.95 g cm-3, respectively. Figure 4 shows a linear relationship between the CrVI moles consumed during the first 0.5 min and the quantity of carbonate green rust initially present in the solution. From the slope, one can calculate a removal of 775 ( 19 µmol of chromate with respect to 1 g of carbonate green rust during the first 0.5 min. The rapid drop of the chromate mole number is associated to a saturation of the surface sites of GR(CO32-) by the resulting CrIII species. Assuming that one CrIII occupies one surface site, the specific area of carbonate green rust can be determined, 40.9 ( 0.9 m2 g-1. This value agrees well with that reported by Williams and Scherer (17) who found 47 ( 7 m2 g-1 by using BET. Depth of Oxidation. The precipitation of one CrIII on a surface site requires the transfer of three e- from three Fe2+ cations of the carbonate green rust to one aqueous chromate ion. The height of the rhombohedral unit cell is c ) 22.71 Å. The unit cell contains 3.33 Fe atoms (2.22 Fe2+ and 1.11 Fe3+). Thus, the precipitation of one monolayer of CrIII species (775 ( 19 µmol of Cr(III) with respect to 1 g of carbonate green rust) implies the oxidation of lattice Fe2+ ions within a depth of 3.07 nm. The electron balance indicates that 1 g of carbonate green rust can reduce 2.1 µmol of CrVI (stoichiometric conditions, R ) 3). Both the complete removal of CrVI and the oxidation of carbonate green rust are therefore associated to the precipitation of 2.7 CrIII monolayers (R1 ) 2.7) and to the oxidation of lattice Fe2+ ions into Fe3+ ions within a depth of 8.3 nm. The average thickness of the carbonate green rust particles can be assessed as twice the oxidation depth, about 16.6 nm. Schematic Representation of Green Rust/CrVI Reaction. In Figure 10, we present a schematic representation of the processes occurring during the CrVI removal by carbonate green rust in the conditions of expt 1 (stoichiometric conditions). Following processes are considered, CrVI reduction, precipitation of CrIII species at the surface of solid GR(CO32-) particles, solid-state oxidation of GR(CO32-), and formation of ferric oxihydroxicarbonate. The initial step (Figure 10b) is modeled as the formation of a CrIII monolayer at the green rust surface with the removal of 775 ( 19 µmol of chromate with respect to 1 g of carbonate green rust. The further removal of CrVI occurs with significantly lower rates. Finally, 2.7 monolayers of Cr(III) precipitate are formed at the surface of ferric oxyhydroxycarbonate, as the result of the complete CrVI removal and oxidation of green rust (Figure 10c). Kinetics of GR(CO32-)/CrVI Reaction. Williams and Scherer (17) and Bond and Fendorf (18) claimed that the kinetics of the carbonate green rust/chromate reaction could be

FIGURE 10. Simplified cross-sectional sketches of the chromatereacted green rust surface modeling expt 1.

FIGURE 11. Theoretical curve drawn according to the model given in eq 5, as compared with the experimental data from expt 1. The insert reports the values of θi as a function of time. modeled by applying a pseudo-first-order reaction rate with respect to CrVI concentration (see eq 4). But, the theoretical curves drawn according to this model match quite well the experimental data only when the carbonate green rust particles are introduced in large excess compared to CrVI. In fact, the heterogeneous reaction between carbonate green rust and aqueous chromate strongly depends on the availability of surface sites for CrIII species. In the case where only one CrIII monolayer with a final coverage ranging from about 20 to 100% (∼0.2 < R1 < 1) forms at completion, the secondorder model must be used (see eq 3). The experimental data were fitted by using the general kinetic model given in eq 5. Figure 11 reports the theoretical curve drawn with j ) 3, as compared with the experimental data of expt 1 where the surface area concentration of carbonate green rust, {S}, is {S} 43.3 m2 L-1. The best matching is obtained with k{S} 1 , k2 , and VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4593

TABLE 2. Parameters Used for Fitting Experimental Data with Kinetic Model Given in Equation 5 expt

R

A (m2 g-1)

R1

j

k{S} 1 (s-1 m2 L-1)

k{S} 2 (s-1 m2 L-1)

1 2 3 4

3 4.27 2.21 6

40.9 43.8 40.9 43.8

2.70 1.77 3.67 1.26

3 2 3 2

5.2 × 10-4 8.3 × 10-4 7.7 × 10-4 4.8 × 10-4

1.4 × 10-5 2.9 × 10-5 0.5 × 10-5 3.0 × 10-5

TABLE 3. Percentage of Fe(II) Involved in Formation of the First CrIII Monolayer as a Function of GR(CO32-) Particles Sizes

thickness (nm) surface area (m2 g-1) FeII involved in formation of first CrIII monolayer (%)

GR(CO32-) particles optimized for Cr(VI) removal

this work

GR(CO32-) as iron corrosion product

110.4 100

16.6 40.9 37

∼100 ∼6.8 ∼6

k{S} 3

values of 5.2 × 10-4, 1.4 × 10-5, and 1.7 × 10-6 s-1 m-2 L, respectively. The insert in Figure 11 reports the evolutions of θi values as a function of time. The data of expts 2-4 were modeled from the parameters values in Table 2. The fit curves are represented as full lines in Figures 3 and 4. Our present k{S} 1 values are close to the k value that can be worked out from the slope of the linear kobs - {S} relationship reported by Williams and Scherer (17, insert in Figure 6), 5.6 10-4 s-1 m-2 L. The k{S} values are about 40-fold higher than k{S} 1 2 values, which are about 10-fold higher than k{S} 3 values. This result is consistent with the progressive passivation of solid particles by successive CrIII monolayers. Implications for Iron Permeable Reactive Barriers. The pH range of the present study (9.3-9.6) is close to values that are measured in iron PRBs (36). The formation of carbonate green rust as a corrosion product can therefore be considered; thus, several studies report the identification of this compound along with other iron oxidation products (3638). In previous studies (24, 27, 30), we showed that the size of carbonate green rust particles formed from iron oxidation was significantly larger than that observed for green rust suspensions obtained from aerial oxidation of Fe(II) solutions, 1-3 µm diameter and about 100-200 nm thickness against 0.1-0.3 µm diameter and 16.6 nm average thickness (present study). The most efficient process for the removal of Cr(VI) corresponds to the formation of up to one monolayer of Cr(III) precipitate at the surface of green rust. This process should be related to an oxidation depth of 3.07 nm (see above). The thickness of carbonate green rust optimized for CrVI removal should therefore be less than 6.1 nm, which corresponds to a specific surface area of 110.4 m2 g-1. In this case, all Fe2+ ions present in green rust should strictly serve to the building of the first CrIII monolayer. This efficiency (Table 3) is compared to those obtained from carbonate green rust particles with thickness of 16.6 nm (present study) and 100 nm (green rust as an iron corrosion product in PRBs). In the latter case, only 6% of Fe(II) should be readily accessible.

Acknowledgments We thank G. Renou (Universite´ d’Evry Val d’Essonne) for the XRD measurements. XPS experiments were carried out at the Laboratoire des Lasers et d’Etudes des Surfaces of the Direction de l’Energie Nucle´aire (CEA, Saclay, France). 4594

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004

k{S} 3 (s-1 m2 L-1) 1.7 × 10-6 not determined

Symbols A

specific surface area (m2 g-1)

N

Avogadro number (mol-1)

{S}

surface area concentration (m2 L-1)

{m}

mass concentration (g L-1)

{no}

molar concentration of reaction sites present at surface of solid phase (mol L-1)

As

reaction site surface area (m2)

{FeII}

molar concentration of FeIIsolid (mol L-1)

[FeII]

molar concentration of aqueous FeII (mol L-1)

[CrVI]

molar concentration of aqueous chromate ions (mol L-1)

{ni}

molar concentration of CrIII precipitated in monolayer i (mol L-1)

k{S} 3

rate coefficient of formation of CrIII monolayer i (s-1 L mol-1)

k{S} i

rate coefficient of formation of CrIII monolayer i (s-1 L m-2)

k{m} i

rate coefficient of formation of CrIII monolayer i (s-1 L g-1)

k{S} 3 {no}t)0 ) k{S} i {S} ) k{m} {m} i

interconversion equations for rate coefficients

Literature Cited (1) Yabusaki, S.; Cantrell, K.; Sass, B.; Steefel, C. Environ. Sci. Technol. 2001, 35, 1493-1503. (2) Liang, L.; Korte, N.; Gu, B.; Puls, R.; Reeter, C. Adv. Environ. Res. 2000, 4, 273-286. (3) Blowes, D. W.; Ptacek, C.; Jambor, J. L. Environ. Sci. Technol. 1997, 31, 3348-3357. (4) Puls, R. W.; Paul, C. J.; Powell R. M. Appl. Geochem. 1999, 14, 989-1000. (5) Powell, R. M.; Puls, R. W.; Hightower, S. K.; Sabatini, D. A. Environ. Sci. Technol. 1995, 29, 1913-1922. (6) Pratt, A. R.; Blowes, D. W.; Ptacek, C. J. Environ. Sci. Technol. 1997, 31, 2492-2498. (7) Astrup, T.; Stipp, S. L. S.; Christensen, T. H. Environ. Sci. Technol. 2000, 34, 4163-4168. (8) Melitas, N.; Chuffe-Moscoso, O.; Farrell, J. Environ. Sci. Technol. 2001, 35, 3948-3953. (9) Alowitz, M. J.; Scherer, M. M. Environ. Sci. Technol. 2002, 36, 299-306. (10) Eary, L. E.; Rai, D. Environ. Sci. Technol. 1988, 22, 972-977. (11) Buerge, I. J.; Hug, S. J. Environ. Sci. Technol. 1997, 31, 14261432. (12) Buerge, I. J.; Hug, S. J. Environ. Sci. Technol. 1999, 33, 42854291. (13) Peterson, M. L.; White, A. F.; Brown, G. E., Jr.; Parks, G. A. Environ. Sci. Technol. 1997, 31, 1573-1576. (14) Anderson, L. D.; Kent, D. B.; Davis, J. A. Environ. Sci. Technol. 1994, 28, 178-185. (15) Loyaux-Lawniczak, S.; Refait, Ph.; Ehrhardt, J. J.; Lecomte, P.; Ge´nin, J.-M. R. Environ. Sci. Technol. 2000, 34, 438-443.

(16) Loyaux-Lawniczak, S.; Lecomte, P.; Ehrhardt, J. J. Environ. Sci. Technol. 2001, 35, 1350-1357. (17) Williams, A. G. B.; Scherer, M. M. Environ. Sci. Technol. 2001, 35, 3488-3494. (18) Bond, D. L.; Fendorf, S. Environ. Sci. Technol. 2003, 37, 27502757. (19) Zepp, R. G.; Wolfe, N. L. Aquatic Surface Chemistry: Chemical Processes at the Particle-Water Interface; Stumm, W., Ed.; Wiley: New York, 1987; pp 423-455. (20) Kendelewicz, T.; Liu, P.; Doyle, C. S.; Brown, G. E., Jr. Surf. Sci. 2000, 469, 144-163. (21) Drissi, S. H.; Refait, Ph.; Abdelmoula, M.; Ge´nin, J. M.-R. Corros. Sci. 1995, 37, 2025-2041. (22) Hansen, H. C. B. Clay Miner. 1989, 24, 663-669. (23) Allmann, R. Acta Crystallogr. 1968, B24, 972-977. (24) Legrand, L.; Mazerolles, L.; Chausse´, A. Geochim. Cosmochim. Acta (in press). (25) Taylor, R. M. Clay Miner. 1980, 15, 369-382. (26) McGill, J. R.; McEnaney, B.; Smith, D. C. Nature 1976, 259, 200201. (27) Legrand, L.; Abdelmoula, M.; Ge´hin, A.; Chausse´, A.; Ge´nin, J. M.-R. Electrochim. Acta 2001, 46, 1815-1822. (28) Stampfl, P. P. Corros. Sci. 1969, 9, 185-187. (29) Ona-Nguema, G.; Abdelmoula, M.; Jorand, F.; Benali, O.; Ge´hin, A.; Block, J. C.; Ge´nin, J. M.-R. Environ. Sci. Technol. 2002, 36, 16-20.

(30) Legrand, L.; Maksoub, R.; Sagon, G.; Lecomte S.; Dallas, J. P.; Chausse´, A. J. Electrochem. Soc. 2003, 150, B45-B51. (31) Adbelmoula, M.; Refait, Ph.; Drissi, S. H.; Mihe, J. P.; Ge´nin, J. M.-R. Corros. Sci. 1996, 38, 623-633. (32) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F.; Muilenberg, G. E. Handbook of X-ray Photoelectron Spectroscopy; Perkin-Elmer: Eden Prairie, MN, 1979. (33) Stypula B.; Stoch J. Corros. Sci. 1994, 36, 2159-2167. (34) Ikemoto, I.; Ishii, K.; Kinoshita, S.; Kuroda, H.; Franco, M, A. A.; Thomas, J. M. J. Solid State Chem. 1976, 17, 425-430. (35) Descostes, M.; Mercier, F.; Thromat, N.; Beaucaire, C.; GautierSoyer, M. Appl. Surf. Sci. 2000, 165, 288-302. (36) Furukawa, Y.; Kim, J. W.; Watkins, J.; Wilkin, R. T. Environ. Sci. Technol. 2002, 36, 5469-5475. (37) Bonin, P. M. L.; Odziemkowski, M. S.; Reardon, E. J.; Gillham, R. W. J. Solution Chem. 2000, 29, 1061-1074. (38) Ritter, K.; Odziemkowski, M. S.; Gillham, R. W. J. Contam. Hydrol. 2002, 55, 87-111.

Received for review December 23, 2003. Revised manuscript received May 11, 2004. Accepted June 8, 2004. ES035447X

VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4595