Environ. Sci. Techno/. 1995, 29, 2396-2404
Reduction of Substituted Nitrobenzenes by Fe(ll) in Aaueous Mineral Susiensions 4
m
JORG KLAUSEN, SERGE P. TROBER, STEFAN B. HADERLEIN, AND R E N E P. SCHWARZENBACH* Swiss Federal institute f o r Environmental Science and Technology (EAWAG) and Swiss Federal institute of Teclznology (ETHZ), CH-8600 Dubendorf: Switzerland
The kinetics of the reduction of 10 monosubstituted nitrobenzenes (NBs) by Fe(ll) has been investigated under various experimental conditions in aqueous suspensions of minerals commonly present in soils and sediments. Aqueous solutions of Fe(l1) were unreactive. In suspensions of Fe(ll1)-containing minerals (magnetite, goethite, and lepidocrocite), Fe(11) readily reduced the NBs to the corresponding anilines in a strongly pH-dependent reaction. Our results suggest that on other mineral surfaces (y-aluminum oxide, amorphous silica, titanium dioxide, and kaolinite) iron (hydr)oxide coatings are indispensable to promote the reduction of NBs by adsorbed Fe(ll). Apparent pseudo-first-order rate constants, kobs, were used to describe the initial kinetics of the NB reduction, covering several half-lives of the compounds. The distinct effect of substituents on kobs and the observed pronounced competition between different NBs indicate that precursor complex formation as well as the (re)generation of reactive surface sites are ratedetermining steps in the overall reduction of the NBs. The results of this study demonstrate that Fe(l1) adsorbed on iron (hydr)oxide surfaces or surface coatings may play an important role in the reductive transformation of organic pollutants in subsurface environments. Our findings may also contribute to a better understanding of the various redox processes involved in groundwater remediation techniques based on Fe(0) as the bulk reductant.
Introduction Reductive dehalogenation and reduction of nitro and azo functional groups of organic pollutants in subsurface environments are of considerable interest for two reasons. These processes are significant ecotoxicologicallybecause they can form products that are of greater toxicity andlor mobility than their parent compounds (1, 2). Also, from an engineering point of view, reductive transformations * Corresponding author e-mail address:
[email protected]; telephone: 41- 1-8235109; fax: 41-1-8235471.
2396 ENVIRONMENTAL SCIENCE &TECHNOLOGY / VOL. 29, NO. 9,1995
may be important with respect to the treatment of hazardous wastes or contaminated soils (3-5). Laboratory and field studies in anoxic sediments, soils, and aquifers have shown that the reduction of organic pollutants may occur by abiotic chemical reactions (6, 7 ) . In most of these studies, the investigators did not attempt to identify the types of natural reductants responsible for the observed reactions. However, given the abundance and the range of reduction potentials of Fe(I1) species that may exist in anoxic environments (8-12), it seems likely that, particularly under iron-reducing conditions, such species play a pivotal role as electron donors or electron transfer mediators in redox transformations of organic compounds. Figure 1summarizes the reduction potentials in aqueous solution, Eho(w),of some key redox couples of various Fe(I1) and Fe(II1) species as well as of organic compounds (2). Under standard environmental conditions, polyhalogenated alkanes, nitroaromatic compounds (NACs),and even aromatic azo compounds can in principle be reduced by a variety of Fe(I1) species. For example, it has been demonstrated that dissolved Fe(I1)porphyrins (FePs)readily reduce polyhalogenated hydrocarbons (13-16) as well as NACs ( 1 7 ) . In addition, it was found that Fe(I1) adsorbed to lepidocrocite was able to reduce nitrite (NO*-) in a relatively fast reaction (18). Finally, structural Fe(I1)in clays and oxides was also found to reduce organic substances such as CCh (19,201. However, even though the reduction potentials of Fe(I1) in these solids are much lower than those of FePs, the rates observed were smaller. Obviously, the reactivity of the various Fe(I1) species cannot be predicted from their Eho(W)values alone; processes other than the electron transfer (such as sorption, desorption, or dissolution reactions) may be rate-limiting and must be considered. We have previously found that various NACs are rapidly reduced both in batch systems containing microbially formed magnetite and Fe(I1) (21)and in laboratory columns containing aquifer material from a river water/groundwater infiltration site (22). For the column systems, we have postulated that the NAC reduction occurred by oxidation of surface-bound Fe(I1) that was continuously produced by dissimilatory iron-reducing microorganisms and that the (relgeneration of these reactive Fe(I1) surface species (and not the actual electron transfer to the NACs) is rate limiting (Figure 2). In this paper, we report the effects of a variety of minerals on the reduction kinetics of a series of NACs in de-aerated aqueous suspensions in the presence of dissolved Fe(I1). Magnetite was chosen as the primary mineral because it was previously found to be a common mineral phase under iron-reducing conditions (12, 23) and because a certain reactivity of its structural Fe(I1)was suspected. NACs were chosen not only because of their great environmental concern but also because they are very suitable model compounds for studying redox reactions potentiallyrelevant in the environment (2). Under reducing conditions, NACs are commonly transformed into well-defined, easily detected products, namely, the corresponding nitrosobenzenes (II), hydroxylamines (III), and the anilines (IV)(eq
0013-936X/95/0929-2396$09.00/0
0 1995 American Chemical Society
Ek'W (volt)
ox
ox
Red
CC1~CC1~
Red
!k'(W)
c12c=ccI~
(volt)
1 .o
1.o
factors such as pH; the concentrations of NACs, Fe(II),and surfaces;and the types of surfaces on the reduction of NACs, and (c)to get hints on the rate-limiting steps that determine overall reduction rates in heterogeneous systems.
Experimental Section 0.5
0.5
0.0
0.0
-0.5
-0.5
FIGURE 1. Representative Fe(ll)/Fe(lll)redox couples and a selection of organic redox couples showing that electron transfer by Fe(ll)/ Fe(lll) systems can occur over the entire stability range of water (h0(w) = -0.5 to $1.1 V from ref 2. The values given represent reduction potentials at pH = 7 at molar concentrations of the redox partners but at environmental concentrations of the major anions involved: [HCOJ-] = [CI-] = M; [Br-] = M; phen = phenanthroline; sal = salicylate; porph = porphyrin.
Wp) substrate
=Fe( III) ''OW
lattice-bound
FIGURE2. Reaction scheme postulatingthe concurrence of microbial and abiotic reactions involved in the reduction of NACs in laboratory aquifer columns (22).
11, thereby allowing mass and electron balances to be established. Ar-NO?
7 Ze-, 2H+
APNO
2e', 2Hf
Ar-"OH
(1)
H20 I
H20 I1
111
Ar =
aR ;
IV
R = H, C1, Me, COCH3
Furthermore, the one-electron reduction potentials, Eh1'(ArN02)(eq 21, of NACs can be determined relatively easily and are available in the literature for a variety of NACs in aqueous solutions at environmentally relevant pH (17).
As demonstrated previously (17, 24) and also used in this work, an evaluation of the relationship between reduction rates and Eh1'(ArN02)values of a series of structuallyrelated NACs, such as monosubstituted nitrobenzenes (NBs),may be very helpful for identifylng rate-determining steps and reaction mechanisms for a redox process. The major objectives of this investigation were (a) to assess the potential and significance of Fe(I1) species in aquatic systems with respect to the reduction of organic pollutants, (b) to evaluate the influence of environmental
Unless otherwise stated, all aqueous solutions were prepared with ion-exchanged high-purity water (Barnstead NANOpure) that had been filtered through 0.2-pm pore diameter membrane filters (Nuclepore; Costar Co., Cambridge, MA). All commercially available chemicals and mineral phases were used as received. Organic Chemicals. Nitrobenzene (H-NB),2-chloro-, 3-chloro-, and 4-chloronitrobenzene (Cl-NB);2-methyl-, 3-methyl-, and 4-methylnitrobenzene (Me-NB); 2-nitro-, 3-nitro-, and 4-nitroacetophenone (Ac-NB);and 4-chloroaniline (4-C1-An)were analytical grade (FlukaAG, Buchs, Switzerland). HPLC/UV analysis revealed only trace amounts of detectable impurities of other isomers or of H-NB. Methanolic stock solutions (0.1 M) of these NBs were stable at room temperature for several weeks. Stock solutions of 2-Ac-NB were unstable and were therefore freshly prepared for each set of experiments. 3-(2-F'yridyl)-5,6-bis(4-phenylsulfonicacid)- 1,2,4-triazine monosodium salt monohydrate (FerroZine reagent, Fluka) was used for the determination of dissolved ferrous iron. A 0.46 mM aqueous solution was prepared and stored at room temperature in the dark. Organic Buffers. 2-(N-Morpholino)ethanesulfonicacid monohydrate (MES, pKa 6.15; for pH 6.00 and 6.50 solutions), piperazine-N,W-bis(2-ethanesulfonic acid) (PIPES, pKa 6.80; for pH 6.75 solutions), 3-(N-morpholino)propanesulfonic acid) (MOPS, pKa 7.20; for pH 7.00, 7.25, and 7.50 solutions), and N-(2-hydroxyethyl)piperazine-N'-3propanesulfonic acid (HEPPS, pKa 8.00; for pH 7.75 and 8.00 solutions) were purchased from Fluka BioChemika (Buchs, Switzerland). These buffers were selected because of their weak complexation of metal ions, such as Fe(II), Ca(II), and Mn(1I) (25). Aqueous stock solutions (1.0 M) were titrated to the desired pH using NaOH. Inorganic Chemicals. Fe(I1)stocksolutions (1.0M) were prepared byfiltering 1OmLofasolution of 1.1 M FeCl2.4H20 (Fluka) at pH 5 through 0.02-pm pore diameter inorganic membrane filters (Anotop 25; Whatman Scientific, Kent, U.K.), or 0.22-pm pore diameter membrane filters (MillexGV Millipore, Waters, Milford, MA) directly into 1 mL of 1 M HC1. These solutions were kept under argon in the dark for a maximum of 3 d. Mineral Phases. Several batches of magnetite (Fes04) were synthesized at 25 "C by oxidation of an anaerobic solution of 30 mM FeC12*4H20 (Fluka) and 3 mM Fe(N03)3.9H20(Fluka)with air at a constant flow of 10 mL min-l (26'). The pH was kept constant at 7.20 f0.01 during the reaction by adding aliquots of 2 M de-aerated NaOH as required. After several hours, 50-100% of the stoichiometric amount of base had been consumed, and the reaction rate, determined as the amount of base added per time, approached zero. The greenish-black precipitates formed were transferred into butyl-rubber stoppered, argon-flushed glass bottles. Within 12 h, finely dispersed, black, magnetic precipitates formed. The bottles were kept upside down in distilled water in order to minimize the diffusion of oxygen into the suspensions. Excess dissolved Fe(I1)was removed by carefully exchanging the supernatant several times with 45 mM oxygen-free NaCl solution at pH VOL. 29, NO. 9, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY a 2397
TABLE 1
Characteristics of Mineral Particles mineral
magnetite lepidocrocite goethite aluminum oxide silicon dioxide titanium dioxide kaolinite
formula
PHW
6.4-6.85' 6.9d-7.3e 7.9 7.4s
FesO4 y-FeOOH a-FeOOH Y-Al203
SiOz, a m anatase (60%)' rutile (40%)
BET surface area (m2 g - l )
*
[zXOHIt' (nm-2)
2h
56 2 70h3 14' 100 rt 159 (50)
(9.8)c 9.6d 13d 5.3d
6'
46 i 2'
3
7.5k
13.2k
n/ai
5d
*Total number of surface hydroxyl groups per nm2. Ref 33. Estimated from Fe(l1) adsorption using the data shown i n Figure 4; ['XOHl,= x NL/BET. Ref 43. e Ref 32. 'Ref 44. Ref 45. Ref 8. Ref 46. Ref 47. Ref 48. n/a, not available. J
5. A freeze-dried portion of the material was identified as magnetite by X-ray powder diffraction analysis. Lepidocrocite (y-FeO0H)-produced by precipitation of Fe(I1) in the presence of (CHd6N4(urotropine) and subsequent oxidation with NaNOZat 60 "C (27)-was kindly provided by H.-U. Laubscher (EAWAG). Goethite (aFeOOH; Bayferrox 910 Standard 86, Bayer AG, Germany), y-aluminum oxide (y-Al~O3; Aluminiumoxid C, Degussa AG, Germany), amorphous silicon dioxide (Si02 am.; Aerosil 0x50, Degussa), and titanium dioxide (TiOz;P25, Degussa) were used as received. Kaolinite (China Clay Supreme; English Clays Lovering Pochrin & Co. Ltd., St. Austelll Cornwall, U.K.) was washed first with 0.2 M HC1 and then with 0.2 M NaCl or CsCI, respectively, until the pH of the supernatant was above 5.5. Excess electrolytewas removed by successively washing the material with H20 (28). Some characteristicsof the minerals used are summarized in Table 1. Experimental Procedures. Experiments were conducted in 30-mLserum flasks equippedwith aTeflon-coated magnetic stir bar. An aliquot of the appropriate sulfonic acid-based pH buffer was added; the flasks were sealed with black butyl-rubber stoppers (Maagtechnic,Dubendorf, Switzerland),evacuated to 20 mbar, refilled with argon five times, and finally equilibrated in a 25 "C water bath. An aliquot of mineral stock suspension and Fe(I1)stock solution were then added using a syringe. The total volume of the final suspension was 14 mL. It typically contained between 10 and 20 m2L-' mineral particles, 36 mM pH buffer, 0.12.3 mM Fe(II), and predetermined amounts of NaOH to compensate for the acidity introduced with the mineral suspension and the ferrous iron solution. A kinetic experiment was performed by adding an aliquot (typically 7 p L ) of 0.1 M methanolic NB stock solution to the reaction medium and by subsequent sampling at regular time intervals. A sample was withdrawn with a syringe and immediately filtered into an Eppendorf tube through an 0.2-pm pore diameter membrane filter (Nuclepore polycarbonate; Costar Co.). An 0.5-mLaliquot of the iiltered solution was added to either 15 ,uL of concentrated perchloric acid (70%;Merck AG, Darmstadt, Germany) or 50 pL of 2.16 M hydroxylamine hydrochloride (pH 6; Fluka AG). Under both conditions, NB concentrations remained unchanged for more than 48 h. The intermediate hydroxylamine (111,eq 1)was only stable when frozen in liquid NZ immediately after preparing the sample. The time elapsed between spiking the assay with NB stock solution and finishing the filtration of a sample was used as the reaction time for evaluating the reaction kinetics. t3S8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL.
29, NO. 9, 1995
'
cSat
Analytical Procedures. Both the NBs and their transformation products were determined by HPLC analysis on an RP-18 or RP-8 reversed-phase column (125 x 4 mm) equipped with a precolumn (4 x 4 mm; both LiChroCart stainless steel cartridge, 5 pm spheres; Merck) connected to a pumping system Uasco 880-PU; Jasco Spectroscopic Co., Ltd., Tokyo,Japan) supplemented with an autosampler (Gina 50; Gynkotek, Germering b. M., Germany) and a variable-wavelength UVlvis detector (Spectroflow 773, Kratos Analytical Instruments, Ramsey, NJ, or Jasco 875vv). The mobile phase was 10 mM, pH 6, NH20H.HC1 buffer in CH30H/H20,312 vlv. The flow rate was typically set to 1 mL min-', and the injected volume was 20 pL. The detector wavelength was set to 254, 265, or 280 nm depending on the compound analyzed (cf. Table 2 for A,, values). 4-Chlorophenylhydroxylamineconcentrations were calculated from a mass balance obtained during the reduction of 4-Cl-NB, assuming that the hydroxylamine compound was the only intermediate (24). The corresponding peak areas were used to calculate a response factor for the calibration of the HPLC system. Dissolved Fe(I1) was determined photometrically after complexation with FerroZine (29) using a UVlvis spectrophotometer (Uvikon 860; Kontron AG, Zurich, Switzerland) at 562 nm. Total Fe(I1) was determined as the sum of dissolved and adsorbed Fe(1I). The latter was calculated from a previously determined adsorption isotherm (data not shown). Validation of the Experimental System. HPLC analyses were run in duplicate and were reproducible within 1-296. The detector response was linear in the concentration range of 0.1-200 pM, and the detection limit was 0.1-1 pM depending on the compound. A possible loss of NBs into butyl-rubber stoppers was examined for 4-Cl-NB in the absence of mineral particles and was found to be negligible (