Environ. Sci. Technol. 1995, 29, 775-783
Reduction of Nitroaromatic Compounds Coupled to Microbial Iron Reduction in Laboratory Aquifer Columns CORNELIS G. HEIJMAN,+ ERWIN G R I E D E R , +
C H R I S T O F H O L L I G E R , + AND RENE P. SCHWARZENRACH*,+ Swiss Federal Institute for Environmental Science and Technology (EAWAG] and Swiss Federal Institute of Technology (ETH), CH-8600 Dubendoe Switzerland, and Limnological Research Center, EAWAG, CH-6047 Kastanienbaum, Switzerland
Using 10 monosubstituted nitrobenzenes as model compounds, the interdependence between the reduction of organic pollutants and microbial iron reduction in anaerobic aquifers has been studied in laboratory column systems. All nitroaromatic compounds (NACs) investigated were stoichiometrically reduced to the corresponding amino compounds. It is proposed that NAC reduction occurred primarily by a reaction with surface-bound iron species, which served as mediators for the transfer of electrons originating from microbial oxidation of organic material by iron-reducing bacteria. Although the different NACs studied exhibited very different one-electron reduction potentials, they were reduced at very similar rates under all conditions investigated, indicating that the regeneration of reactive sites and not the electron transfer to the NAC was the rate-limiting process. It is also proposed that the presence of reducible organic pollutants such as NACs may significantly enhance the activity of iron-reducing bacteria in aquifers, in that reduction of such compounds continuously regenerates easily available Fe(lll) species.
Introduction There is a growing interest in reduction reactions of organic pollutants, mainlyfor two reasons. On the one hand, it has been recognized that in anaerobic environments (e.g., sediments, soils, and aquifers) reductive transformations of anthropogenic organic chemicals can lead to products that may be of considerable (eco)toxicologicalconcern (I). On the other hand, such processes, particularly reductive dehalogenations, have been investigated extensively because of their potential application in the biological and chemical treatment of hazardous wastes or contaminated sites (2). To date, part of the published work addressing kinetic aspects of the reduction of organic pollutants in the aquatic environment has focused on studies using (bio)chemical model reductants (3-9) or liquid cultures of anaerobic microorganisms (10-14). The results of such studies have provided important information on mechanisms and ratedetermining steps of reactions of organic chemicals with potentially important environmental electron donors. Some work has also been conducted to evaluate the reduction kinetics of organic pollutants in systems that simulate the natural environment such as sediment slurries (15-18) and aquifer columns(l9-21). In most of these studies, it was however not attempted to identify the type of natural reductants primarilyresponsible for the observed reaction. In addition, it was not possible to assess unequivocally whether the reaction occurred strictly abiotically, whether it was mediated by microorganisms, or whether both chemical and biological processes were determining the overall reaction rate. In earlier work, we have investigated the reduction kinetics of nitroaromatic compounds (NACs),particularly of substituted nitrobenzenes, in a variety of homogeneous and heterogeneous batch model systems (22-26). We have chosen to investigate NACs for two different reasons. First, NACs have been found to be ubiquitous pollutants in the aquatic environment, because they are widely used as pesticides, explosives, chemical intermediates, and dyes (27). Furthermore, there is also evidence that certain NACs are formed in significant quantities by photochemical processes in the atmosphere (28, 29). Under anoxic conditions, NACs(1) may be reduced to the corresponding hydroxylamines(II1) and, ultimately, to the amines(W):
Note that& stands for any aromatic structure (e.g.,benzene ring, polycyclic aromatic ring system). Hydroxylamines and amines are both of considerable toxicological concern. The second reason for using NACs is that they are particularly well-suited probe compounds for studying redox reactions in anaerobic environments. Under reducing conditions, many NACs are transformed to clearly defined, easily detectable products (i.e., hydroxylamines+
Limnological Research Center.
* EAWAGIETH.
0013-936Xl9510929-0775$09.00/0
0 1995 American Chemical Society
VOL. 29, NO. 3, 1995 I ENVIRONMENTAL SCIENCE & TECHNOLOGY
775
TABLE 1
Overview of the Various Column Systems
column type I column type II column type 111
a
geometry of columns
aauifer material
flow parameters
length diameter total volume no. of (cm) (cm) (cm3) sampling ports'
size fraction Fetd porosity (mg g-l) bm)
flow velocity hydraulic residence time (h) (em h-l)
100 24 22
5.0 5.0 3.2
2.0 x 4.7 x 1.8 x
lo3 lo2 lo2
14 7 2
-0.4 -0.4 -0.4 -0.4 -0.4 -0.4
6.7 6.5 6.0 6.6 6.2 6.5
3.0 0.51 0.69 0.69 0.69 0.69
32 47 32 32 32 32
Including inlet and outlet.
(1111, amines(W),thus allowing the establishment of mass and electron balances. Furthermore, the transfer of the first electron to a given NAC is reversible (eq 2). Therefore,
the corresponding one-electron reduction potential, Ehl' (ArN02),can be relatively easily determined and is available for a variety of NACs in aqueous solution at ambient pH (22).
As we have shown previously, plots of reduction rate data versus Eh1'(ArN02)values of a series of structurally related NACs (e.g.,monosubstituted nitrobenzenes) are very useful for evaluating whether in a given system the actual transfer of the electron is rate limiting or whether another reaction step (e.g., formation of the precursor complex) is determining the overall reduction rate (22). In addition, the comparison of such plots derived for defined model systems with plots obtained for less defined environmental systems may even provide hints on the type of reductant@) primarily responsible for the observed NAC reduction in the environmental systems (23). In the work presented in this paper, we have extended our investigations of reductive transformations of NACs to laboratory column systems containing aquifer materials collected from a river water/groundwater infiltration site. The aquifer materials and experimental conditions were chosen in a way to mimic iron-reducing subsurface environments. In several recent studies, it has been clearly demonstrated that the importance of transformation reactions of organic pollutants under iron-reducing conditions has been largely underestimated in the past (30-32). Furthermore, in previous work (261,we have shown that in liquid enrichment cultures of iron-reducing bacteria NACs are reduced very efficientlyby an abiotic reaction involving ferrous iron and magnetite. The major purpose of this study was to evaluate the interdependence between microbial iron reduction and NAC reduction in anaerobic aquifers.
Experimental Section Chemicals. Nitrobenzene (NB); 2-chloro-, 3-chloro-, and 4-chloronitrobenzene (Cl-NB);2-methyl-, 3-methyl-, and 4-methylnitrobenzene (Me-NB); 3-(N-morpholino)propanesulfonic acid (MOPS);N- (2-hydroxyethyllpiperazineN-2-ethanesulfonic acid (HEPES); and ferrozine were purchased from Fluka AG (Buchs, Switzerland). 2-Nitro-, 3-nitro-, 4-nitroacetophenone (Ac-NB) and 2-amino-, 3-amino-,4-aminoacetophenone (Ac-An)were purchased from Merck (Darmstadt,Germany). Gas was obtained from Pan Gas (Kriens, Switzerland). All chemicals were of analytical grade and were used without further purification. 776
250-500 125-500 125-250 250-500 250-500 125-500
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 29, NO. 3, 1995
Column Design. The experiments were conductedwith three different types of columns. Some important characteristics of the columns are summarized in Table 1. Column types I and I1 were constructed of transparent Plexiglas and were fitted with a series of sampling ports equally spaced over the length of the column. In the case of column type 111, which was made of borosilicate glass, only the influent and the eMuent could be sampled. A detailed description of column types I and I1 is given by Kuhn et al. (33). Preparation and Characterization ofAquifer Material. All columns were wet-packed with natural aquifer material that was obtained from the sediment layer (5-20 cm) of a river water/groundwater infiltration site (Chriesbach, Diibendorf, Switzerland). The material was washed under aerobic conditions and wet-sieved to pass a mesh of 125250 and 250-500 pm, respectively. Some characteristic data of the material are given in Table 1. The porosities of the packed columns were determined with C1- as conservative tracer. Breakthrough curves obtained for 4-chloroaniline showed no significant retardation of this compound as compared to C1-. For determination of total particulate manganese and iron, 50 mg of dried (24 h at 110 "C) aquifer material was extracted with 1 mL of hydrogen peroxide (35%)and 4 mL of nitric acid (65%) using a microwave extractor (Model M U 1200,Henry Sarasin AG, Basel, Switzerland), followed by atomic absorption spectroscopy (Model 2100, Perlin Elmer AG, Ueberlingen, Germany). The total particulate iron contents are summarized in Table 1. Total reduced particulate sulfurwas measured by the procedure described by Canfield ~t al. (34). The concentrations obtained for both manganese and reduced sulfur components were smaller than 0.5 m g g l . Particulate organic carbon (POC) was determined by the method of Baccini et al. (35). POC values were between 1(aquifermaterial that had been stored for 1-2 years at 4 "C) and 2 mg of C gs-I (freshly collected material). Experimental Setup. Column type I was operated in the saturated, upflow mode using an HPLC pump (Kontron, Rotkreuz, Switzerland) to obtain constant flow rates. A syringe pump (Braun Perfusor, Melsungen, Germany) delivered an accurate dosage of a concentrated aqueous NAC solution (1 mM) into a stainless steel T-joint connection to the influent of the column (for details, see Kuhn etal. (33)).The influent into the column consisted of a 10 mM NaHC03 solution in double-distilled and autoclaved water (20 min at 120 "C). Prior to the addition of the NACs, this solution was passed through a 2.0-m silicone tube positioned in a gas exchange chamber flushed with N z /
COZ(90110%). As described in detail by Zeyer et al. (367, in this chamber, oxygen was efficiently removed from the solution, and the pH was adjusted to 7.1. In contrast to column type I, column types I1 and I11 were operated in the down flow mode. Flow control was accomplished by a peristaltic pump (Model IPN, Ismatec, Zurich, Switzerland)connected to the outlet of the column. The inlet of the columns were connected by stainless steel tubing (1 mm i.d.1 to 1-10-L glass bottles that served as reservoirs. The reservoirs containing the NAC solutions were prepared as follows. The glass bottles were filledwith double-distilled and autoclaved water. Subsequently, the waterwas strippedwith NZand Nz/COZ(90/10%)to remove oxygen. Then, aliquots of sterile-filtered or autoclaved oxygen-free aqueous stock solutions of NaHC03 (0.8 M), NACs (1 mM), and, where indicated, acetate (0.1 M) were added to obtain the desired concentrations of the various species (10 mM NaHC03, 5-300 pM NAC, 20-200 pM acetate). The head space in the reservoirs consisted of NL/ COZ(90110%). The pH was constant (7.10 i 0.05) in the reservoirs and throughout the columns. In general, the columns were operated at a constant flow rate (Table 1) in a thermostated room at 25 k 0.5 "C in the dark. AU experiments were conducted at least in duplicates; most experiments were, however, performed in 3-6 replicates. Prior to the addition of NACs, all columns were operated first with sterile-filtered anoxic NaHC03containing water for 2 weeks. For experiments at higher temperatures, the columns (type 111) were totally wrapped with PVC tubes (5 mm i.d.1 through which water of a given temperature was pumped continuously from a water bath. Two type I11 columns were y-irradiated for 3.5 h at 7 kGy h-l, resulting in a total effective dose of about 25 kGy. The irradiation was carried out by M. Steinemann, Paul Scherrer Institute (PSI),Villigen, Switzerland. Analytical Methods. For the determination of the concentration of the NACs and of their transformation products (Le., hydroxylamines and anilines), 0.2 mL of a water sample (inlet, outlet, sampling ports) was extracted with 0.5 mL of ethyl acetate as described by Schwarzenbach et al. (22). The analyses of the ethyl acetate extracts were performed by HPLC on a RP18 reversed-phase column (stainless steel cartouche Lichrocart 125 x 4mm, 5-pm spheres; Merck, Darmstadt, Germany) connected to a Perkin-Elmer pump system (Perkin-Elmer AG, Series 4 liquid chromatography, Ueberlingen, Germany) supplementedwith an auto sampler (Jasco,Model 851-AS,Tokyo, Japan),a Rheodyne 7125 injector with a 6-pLinjectionloop, a Uvikon 430 variable wavelength W detector (Kontron AG, Zurich, Switzerland), and a Perkin-Elmer LCI-100 computing integrator (Perkin-Elmer AG, Ueberlingen, Germany). The mobile phase consisted of a mixture of methanollwater (typically65%/35%),both containing 10% 0.1 M hydroxylammonium chloride buffer (pH 6.0). The flow rate was set at 1.0 mL min-I, and detection occurred generally at 254 nm. Total iron in aqueous samples was measured spectrophotometrically with ferrozine as described by Lovley and Phillips (37)with the modification of using a 0.5 M instead Of a0.05 M HEPES pH buffer (pH 7.0). Fe(I1)was determined by the same method in the absence of hydroxylamine. Acetate was determined by ion chromatography (Metrohm, Model 690, Herisau, Switzerland) using a Hamilton RPX300 column. The eluent was 0.5 mM sulfuric acid at a flow rate of 1.0 mL min-l. Methane in aqueous samples was
250
4-CI-NB
+
4-CI-An
n
E5. v
c
150
4-CI-An
0
5
10
15
20
25
40
50
Column length (cm) I " " I ' " ' 1 ' ' " 1 ' " ' 1 " " 1
0
10
20
30
Residence time (h) FIGURE 1. Type of concentration gradients typically found for a given NAC (here CCI-NB, 0 )and the corresponding aniline (here 4-CI-An, 0 )in an undisturbed column (type 11). Also shown is the mass balance (A)over the length of the column.
measured by head space analysis. A 5-mL water sample was transferred to a slightly evacuated 9-mL serum bottle. After equilibration at 25 "C, an aliquot (200 pL) of the headspace was analyzed on a gas chromatograph (Carlo Erba, Model 5160, Brechbtihler, Schlieren, Switzerland) equipped with a GS-Q column (30 m x 0.53 mm i.d., J &W Scientific, Folsom, CA)and an FID. The oven temperature was maintained 40 "C, and the flow rate of the carrier gas (H2) was 4 mL min-I.
Results and Discussion Phenomenological Observations. Figure 1 shows an example of the type of concentration gradients that were observed in columns of types I and I1 when supplying a constant input of a given NAC. As is illustrated by the mass balance for 4-chloronitrobenzene (4-Cl-NB1,the NACs were reduced stoichiometrically to the corresponding anilines. Furthermore, in a given column, a nearly constant slope of the concentration gradient was found over the entire column length and over the entire concentration range investigated (i.e., 5-3OOpM). Thus, for comparison of the reactivities of the various model NACs and of the different column systems, it is feasible and useful to define an apparent zero-order rate constant, /cobs, for the reduction of a given NAC. Due to the low organic carbon content of the aquifer materialused fJo, 5 0.0021, hydrophobic sorption of the NACs can be neglected. In addition, because the influent contained 10 mM Na+, it can also be assumed that formation of electron donor-acceptor complexes of the NACs to clay mineral surfaces was not significant (38). Therefore, the average residence times of the NACs in the columns were assumed to be similar to the residence time (tw) of the water. For a given column or column segment, VOL. 29, NO. 3, 1995 /ENVIRONMENTAL SCIENCE & TECHNOLOGY
777
' 9 undisturbed column (type 111)
-
r
1.0-
0.0 -
0 0
. 50
1 100
4 150
1
200
250
Time (days) FIGURE 2. Plot of 16br determined for 4-CI-NB as a function of time for 15 different column systems: (4 column type I containing fresh material (July 1988; 250-500 p m size fraction); (0)column type II containing fresh material (March 1993; 125-500 p m size fraction); (W) column type 111 containing fresh material (August 1992; 250-500 p m size fraction; average of five columns with standard deviation