Environ. Sci. Technol. 2001, 35, 196-203
In Situ Stimulation of Groundwater Denitrification with Formate To Remediate Nitrate Contamination RICHARD L. SMITH,* DANIEL N. MILLER,† AND MYRON H. BROOKS‡ U.S. Geological Survey, 3215 Marine Street, Boulder, Colorado 80303 MARK A. WIDDOWSON AND MARC W. KILLINGSTAD§ Department of Civil and Environmental Engineering, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
In situ stimulation of denitrification has been proposed as a mechanism to remediate groundwater nitrate contamination. In this study, sodium formate was added to a sand and gravel aquifer on Cape Cod, MA, to test whether formate could serve as a potential electron donor for subsurface denitrification. During 16- and 10-day trials, groundwater from an anoxic nitrate-containing zone (0.51.5 mM) was continuously withdrawn, amended with formate and bromide, and pumped back into the aquifer. Concentrations of groundwater constituents were monitored in multilevel samplers after up to 15 m of transport by natural gradient flow. Nitrate and formate concentrations were decreased 80-100% and 60-70%, respectively, with time and subsequent travel distance, while nitrite concentrations inversely increased. The field experiment breakthrough curves were simulated with a two-dimensional site-specific model that included transport, denitrification, and microbial growth. Initial values for model parameters were obtained from laboratory incubations with aquifer core material and then refined to fit field breakthrough curves. The model and the lab results indicated that formateenhanced nitrite reduction was nearly 4-fold slower than nitrate reduction, but in the lab, nitrite was completely consumed with sufficient exposure time. Results of this study suggest that a long-term injection of formate is necessary to test the remediation potential of this approach for nitrate contamination and that adaptation to nitrite accumulation will be a key determinative factor.
Introduction Nitrate is the most prevalent groundwater contaminant in freshwater aquifers. In widespread areas of the U.S., more than 25% of all wells sampled contain nitrate in excess of 0.2 * Corresponding author phone: (303)541-3032; fax: (303)447-2505; e-mail:
[email protected]. † Current address: U.S. Department of Agriculture, P.O. Box 166, Clay Center, NE 68933. ‡ Current address: U.S. Geological Survey, 2617 E. Lincoln Way, Suite B, Cheyenne, WY 82001. § Current address: ARCADIS Geraghty & Miller, Inc., 1131 Benfield Blvd., Millersville, MD 21108. 196
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
mM (3 mg N L-1), while long-term trends indicate increasing nitrate concentrations in groundwater in all parts of the country (1), particularly in certain more susceptible hydrologic settings (2). Nitrate contamination originates from agricultural, sewage-disposal, and industrial practices and from both point and nonpoint sources. Though not exclusive to the subsurface, nitrate contamination is much more pervasive in groundwater because the compound has a relatively long residence time in that environment. Thus, when groundwater is being used for a drinking water supply, effective treatment processes are needed to deal with excessive nitrate concentrations. Hydrogen-enhanced denitrification is one such potential treatment process (3-5). A select group of denitrifying bacteria can autotrophically reduce nitrate, using hydrogen as an electron donor, and subsequently produce water and nitrogen gas, two innocuous end products (5). Addition of hydrogen to nitrate-laden water can rather specifically stimulate these organisms; it avoids the need to add an organic electron donor, and upon depletion of the unwanted nitrate, excess hydrogen can be easily removed by air stripping. The process has been successfully utilized in pump and treat applications at pilot-scale industrial plants in Belgium (6) and Germany (7). In rural areas, in situ treatment via denitrification may have more utility than a pump and treat approach. In situ treatment could potentially generate a nitrate-free zone around a single-household water-supply well. This approach would exploit the natural filtration capacity of an aquifer, utilize the indigenous populations of denitrifying bacteria, and obviate the need for pump and treat technology. In at least one case, the abundance of hydrogen-oxidizing denitrifiers (HOD) in a nitrate-contaminated sand and gravel aquifer was significant relative to the abundance of heterotrophic denitrifiers (5), to the extent that in situ stimulation of the HOD population appeared to be feasible. However, little is known about the adaptability of the process to in situ stimulation or its capability to selectively remove nitrate. Moreover, due to its low solubility in water, the addition of hydrogen to the subsurface to stimulate these organisms is not straightforward. With an observed stoichiometry of ∼6:1 (H2:NO3-) for hydrogen oxidation by denitrification (5), if water contained 0.7 mM nitrate (10 mg N L-1; the U.S. drinking water limit), only about 20% of the nitrate could be degraded via hydrogen oxidation, even if the water was fully saturated with hydrogen. Thus, an effective approach is needed to achieve adequate substrate delivery and to test the potential for in situ stimulation of the hydrogen-oxidizing denitrifiers. In this study, we examined the potential to use formate as a water-soluble hydrogen analogue for in situ enhancement of groundwater denitrification. Formate can be readily hydrolyzed to hydrogen and carbon dioxide by many types of microorganisms (8, 9), and it induces autotrophic carbon dioxide fixation in some denitrifiers (10). Formate is an important intermediate in anaerobic ecosystems, such as aquatic sediments (11, 12) and ruminant mammals (13), and it has been speculated that hydrogen and formate are interconvertible electron donors in such systems (14, 15). However, little is known about formate in denitrifying ecosystems, even though formate utilization by denitrifying bacteria has been reported in a variety of metabolism studies (10, 16-18). The aims of this study were to (i) test the utility and the specificity of using formate in lieu of hydrogen to stimulate denitrification and (ii) to discern and simulate the general response of groundwater denitrification to the in situ stimulation. We report here the results of carefully 10.1021/es001360p CCC: $20.00
2001 American Chemical Society Published on Web 11/29/2000
FIGURE 2. Diagram of the field apparatus used to introduce formate into a zone of nitrate contaminated groundwater. A sterile, anoxic, concentrated solution of sodium formate was metered into continuously pumped groundwater that was being withdrawn and injected into wells screened in the same vertical interval.
TABLE 1. Characteristics of Injected Groundwater for In Situ Formate Stimulation Experimentsa
FIGURE 1. Map of well site F473 near the Massachusetts Military Reservation on Cape Cod, Massachusetts. Multilevel samplers (MLS) are designated by row and by number (e.g. 2-12, 4-12, etc.). controlled, sequential, injection experiments into a nitratecontaminated sand and gravel aquifer designed to incorporate the physical attributes of the aquifer (vertical geochemical gradients, physical flow, and hydrodynamic dispersion) into the response of the indigenous, denitrifying, microbial community.
Experimental Section Study Site. This study was conducted in a freshwater sand and gravel aquifer located on Cape Cod near Falmouth, Massachusetts. At this location, disposal of dilute, treated sewage has resulted in a large plume of contaminated groundwater (4 km long), of which nitrate is a significant component, with concentrations often exceeding 1 mM (19, 20). Dissolved organic carbon in the plume is 2-4 mg L-1 and is composed of mostly refractory compounds (21, 22). A large portion of the contaminant plume is suboxic (10200 µM O2) or anoxic. Thus, denitrification is a significant terminal electron-accepting process in portions of the plume and has been the subject of a long-term field study at this site (20, 23-25). In general, within zones of nitrate-containing groundwater, denitrification is electron-donor limited within this aquifer (23), while dissimilatory reduction of nitrate to ammonium is not occurring to any significant extent (20, 23). This study was conducted at site F473 (Figure 1), which is located about 2 years groundwater travel time from the contaminant source. In Situ Tests. Sodium formate was added to a nitratecontaminated zone of the aquifer to stimulate in situ denitrification (Figure 2). Two 19 mm diameter, 0.9 m long, stainless steel, Vylon-screened drive points (Solinst Canada, Ltd., Glen Williams, Ontario) were installed (at the same altitude vertically and ∼6 m apart horizontally) using a percussion hammer. For the injection experiment, groundwater was continuously pumped from the withdrawal well
parameter
1993 test
1994 test
groundwater pumping rate (L h-1) injection time (days) nitrate (mM) nitrite (µM) bromide (mM) formate (mM)
6.1 (0.2)
7.9 (0.5)
16 1.37 (0.12) 8.9 (1.4) 0.5 (0.05) 12.4 (1.0)
10 0.51 (0.07) 5.3 (0.5) 0-1.2b 5.4 (2.3)
a Values are the mean of samples collected during injection duration. Brackets enclose standard deviation. b Bromide added independently of formate in 1994.
into the injection well using a peristaltic pump. The lines connecting the wells were 6.4 mm diameter copper tubing (with a 0.3 m length of Norprene tubing in the pump head) configured with shutoff valves on each side of the pump. Also situated on the downstream side of the peristaltic pump were (A) a tee through which the formate solution was added; (B) an inline pressure transducer (Chem-Tech Equipment Co., Deerfield Beach, FL) connected to a Campbell Scientific (Logan, UT) CR10 data logger to record flow; and (C) a threeway ball valve to divert flow for collection of injectate samples and to manually calibrate the flow rates. Care was taken at all times to keep the water lines primed and to avoid pumping air into the injection well. The pumping rates for each experiment are listed in Table 1. O2-free (N2-sparged), filter-sterilized (0.2 µm) solutions (19 L using DI water) of NaBr and NaCOOH were prepared in glass carboys and added to the groundwater flow line with a metering pump (50-55 mL h-1). The metering lines were 1.6 mm diameter stainless steel and were connected to a variable back-pressure control valve downstream of the metering pump to prevent backflow of groundwater into the sterile stock solution. Isopropyl alcohol (100 mL) was pumped through the metering system to sterilize it prior to connection with the carboy. The volume of liquid removed from the carboy during pumping was replaced with filter-sterilized (0.2 µm) N2 at ∼3 psi. The tracer cloud that resulted from the injection process moved downgradient through the aquifer with natural groundwater flow. It was intercepted and sampled with 15port multilevel samplers (MLSs), located 4.0, 7.1, 10.1, and VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
197
15.2 m from the injection well (Figure 1). Water samples were collected daily from the sampling grid with a peristaltic pump and from the injection apparatus. Samples to be analyzed for (1) nitrate, nitrite, and formate were filtered (0.45 µm Gelman Supor Capsule filters) and frozen; (2) nitrous oxide was collected in a syringe (15 mL) and injected into stoppered 30 mL serum bottles that contained 200 µL of 12.5 N NaOH; (3) bromide was collected in plastic 60 mL bottles; and (4) oxygen was collected and measured in the field using evacuated Vacu-vials containing colorimetric reagents (Chemetrics, Inc, Calverton, VA). Laboratory Incubations. Aquifer sediments were collected with a wire-line piston core barrel through hollow stem augers (26). The core barrel was fitted with a 5.1 cm diameter, 1.5 m long aluminum liner, which was cut into 0.3-m segments upon retrieval. Each segment was capped, chilled to 4 °C, and shipped to the laboratory in Colorado. Formate-utilizing and denitrifying activity were measured in the laboratory using sediment slurry incubations. The sediment was extruded from the aluminum liners on a lab bench and mixed, and 25 g (wet wt) was added to three 150mL serum bottles, which were then immediately transferred to an anaerobic glovebox. Sterile, O2-free, artificial groundwater (100 mL; 1.5 mM NaCl, 15 µM KH2PO4, 150 µM Na2SO4, 0.9 mM NaHCO3; pH 6.5) was added to each bottle. The bottles were stoppered, crimped, and flushed for 20 min with O2-free N2. Sterile, O2-free solutions of sodium nitrate or sodium nitrite and sodium formate were then added to each bottle (0.7 and 1.5 mM, final concentration, respectively), and the bottles were incubated statically in a 15 °C incubator. Water samples were periodically collected from the bottles by syringe, filtered through 0.45 µm Metricel membrane filters (Gelman Sciences, Ann Arbor, MI), and frozen. An equal volume of O2-free N2 was added to each bottle prior to sampling to replace the volume of water removed. Model Formulation. A mathematical model was developed for the purpose of interpreting concentration data and quantifying nitrate, nitrite, and formate removal rates and the rate of nitrite production. A model by Widdowson et al. (28) describing the one-dimensional transport and biodegradation of three solutes (one electron donor and two aqueous-phase electron acceptors: oxygen and nitrate) and the growth and decay of a facultative heterotrophic microbial population was extended for two-dimensional transport and expanded to include the transport, production (due to the reduction of nitrate), and reduction of nitrite. The result is a system of four differential equations describing the fate and transport of formate (electron donor), nitrate, and nitrite (electron acceptors) and the growth and decay of the nitrate/ nitrite-reducing microbial population. The governing equations of mass balance are presented in the Supporting Information as well as the values used for the model parameters. The model was tested and verified using both laboratory and field data (27). Analytical Techniques. Nitrate and nitrite were analyzed colorimetrically with a flow-injection autoanalyzer using the cadmium reduction technique (29). Oxygen was determined with both an oxygen-specific probe and colorimetrically (30), while bromide was measured in the field with an ion-specific electrode. Nitrous oxide was measured with a gas chromatograph equipped with an electron capture detector (24). Formate was determined using a Dionex (Sunnyvale, CA) Model DX-300 Ion Chromatograph with a CDM-3 Conductivity Detector using an elution gradient of NaOH (0.75-86.4 mM) through Ionpac AS5A-5µ guard and analytical columns with a flow rate of 1.0 mL min-1. The limit of detection was approximately 5 µM formate. 198
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
FIGURE 3. Vertical profile of dissolved oxygen and nitrate in groundwater collected from the formate injection site in June 1993 (solid symbols) and September 1994 (open symbols).
Results and Discussion In Situ Tests. The groundwater contaminant plume that was selected for this study is typified by sharp vertical gradients of dissolved constituents (31). The contaminant source (discharged, treated sewage) has continuously loaded the aquifer with degradable organic carbon, nitrate, and ammonium. As the contaminants move downgradient, the plume becomes overlain by recharge water, creating vertical gradients between the uncontaminated water and the core of the contaminant plume. One predominant region within the plume is anoxic or suboxic and contains relatively high concentrations of nitrate (Figure 3). Previous studies at this field site have identified denitrification as the major electronaccepting process within this region and that the process is primarily electron-donor limited (20, 23, 25). This region is typical of the groundwater geochemistry often associated with subsurface nitrate contamination. If the electron donor limitation can be relieved, in situ denitrification should be capable of removing (or remediating) the contaminant nitrate by reducing it to nitrogen gas. Thus, this location was selected to test enhancement of in situ denitrification using formate as an added substrate. The in situ field tests were conducted by continuously withdrawing nitrate-contaminated groundwater, metering into it a sterile, anoxic, concentrated solution of sodium formate and sodium bromide, and then reinjecting the groundwater into the aquifer in the same vertical horizon from which it was withdrawn (Figure 2). This approach was selected to introduce the formate into a sufficient volume of groundwater to conduct the in situ test without altering the groundwater oxygen concentration and without requiring preparation of large volumes of the formate stock solution. The groundwater pumping rate was relatively slow (6-8 L h-1) (Table 1). Once back in the aquifer, the formate- and bromide-amended water was transported downgradient by regional groundwater flow (∼0.5 m day-1). The downgradient location of the injectate cloud and the selection of MLS ports to sample were based on bromide concentrations determined in the field. Two formate injection experiments were conducted, both at the same location using the same wells. The first test was a 16-day formate injection conducted in 1993 (Table 1).
FIGURE 4. Time course of nitrate, nitrite, and formate concentrations in groundwater collected 10 m downgradient from the injection well at an altitude of 9.08 m above sea level for the 1993 injection experiment. Samples were collected from MLSs located at 4, 7, and 10 m downgradient from the injection well. The tracers were detected in four ports at each MLS, representing a 0.8 m vertical interval that was slightly deeper than the injection well screen. There was relatively little consumption of the injected formate after 10 m of transport (Figure 4). At that location, the formate breakthrough curve was ∼22 days wide (sample collection was terminated before the concentration returned to background) with a peak concentration on day 35 that was 78% of the injectate. After day 35, the formate concentration decreased somewhat and was accompanied by a concomitant decrease in nitrate and increase in nitrite concentrations (Figure 4). Nitrate decreased from 1.6 mM to 0.4 mM in 8 days (the corresponding nitrite increase was 1 mM), before subsequently increasing as the formate levels declined at the tail of the tracer cloud. During the 35-43 day interval, the molar ratio of formate to nitrate consumed was ∼2.7. At the same time, the background nitrous oxide concentration decreased from 5 to 7 µM to undetectable levels (data not shown). The second formate injection experiment was conducted 15 months later. For the 1994 test, the background nitrate concentration had decreased to less than half the 1993 value (Figure 3, Table 1); hence, the formate addition was adjusted accordingly (Table 1). For this experiment, a second metering pump was used to add sodium bromide independently of the sodium formate, and a new row of MLSs was installed 15 m downgradient of the injection well to increase the travel distance that could be monitored. Significant decreases in nitrate levels were evident in the 1994 experiment, starting 29 days after the beginning of the formate injection. Nitrate concentrations decreased immediately upon the arrival of the formate cloud at the MLS 7 m downgradient (Figure 5A), dropping to a low value of 0.4 mM (53% of background at that sampling port). Nitrate concentrations were further depleted after 10 m of travel (Figure 5B) and nearly completely removed for a 10 day interval at the 15 m MLS (Figure 5C). Formate breakthrough curves demonstrated concentration decreases due to the combined effects of consumption, dispersion (evident in increased peak width with subsequent travel downgradient), and sinking. For the sampling altitude shown in Figure 5, peak formate concentrations were 102, 88, and 29% of the mean injectate levels after 7, 10, and 15 m of transport, respectively. The highest formate concentration at the 15 m MLS (56% of mean injectate concentration) was actually 1 port (0.25 m) deeper due to density-induced sinking (data not shown). As in 1993, nitrate consumption was accompanied by an increase in nitrite concentrations that nearly matched the amount of nitrate lost. However, there was a small but measurable and increasing amount of
FIGURE 5. Time course of nitrate, nitrite, and formate concentrations in groundwater collected from (A) 7, (B) 10, and (C) 15 m downgradient from the injection well at an altitude of 9.58 m above sea level for the 1994 injection experiment. nitrite consumed after 7 and 10 m of travel distance (Table 2). For the 15 m well, there was much less nitrite consumption. Even after 55 days, with excess formate still present and nitrate concentrations reduced below 0.2 mM, nitrite concentrations remained high (Figure 5C). It should be noted that subsurface heterogeneity and sinking of the formate cloud during transport make it unlikely that any two wells are collecting samples from the same “slice” of the tracer cloud. This can be particularly important if concentration heterogeneities within the cloud are greater than concentration losses over a certain travel distance. Finally, in 1994 background nitrous oxide concentrations were again completely depleted whenever the nitrite peak was present (data not shown). The results from the two field injection experiments clearly demonstrate that the formate additions to the aquifer VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
199
TABLE 2. Nitrite Consumption during the In Situ Formate Injection Experimentsa field test
travel distance, m
width of nitrite consumption peak, days
nitrite consumed,b mM
1993 1994 1994 1994
10 7 10 15
>11 4 12 3
0.5 0.2 0.5 0.4
a Calculated as the decrease in the sum total concentration of nitrate + nitrite. b Maximum amount.
stimulated in situ nitrate consumption. The onset of the enhanced nitrate consumption was faster the second year (Figures 4 and 5), suggesting that microbial acclimation to the first 10-day formate addition persisted 15 months later. Barbaro et al. (32) similarly found that subsurface microbial populations that had been acclimatized to in situ injections of BTEX (benzene, toluene, ethylbenzene and xylene) and nitrate retained an enhanced toluene-degradation capacity for several months in the absence of substrate. In this study, the combined acclimation from the two tests enabled the complete removal of nitrate during the second field test within 15 m of travel. On the other hand, significant concentrations of formate persisted throughout the sampled transport interval. This suggested that the formate additions were fairly specific in targeting nitrate-reducing microorganisms. There appears to have been relatively little formate degradation occurring that was not coupled to nitrate reduction. This is the result that might be expected, i.e., relatively little competition for the added substrate, if the injection process has stimulated a group of microorganisms that have a unique niche, such as the autotrophic hydrogen-oxidizing denitrifiers. In culture, the HOD strains isolated from this field site produce no or only transient levels of nitrite when growing on hydrogen (5); nitrite reduction readily occurs. This was not the case in the field tests, as only a small amount of nitrite was reduced (Table 2) within the sampled interval. Obviously, from a remediation perspective, nitrite consumption must continue to completion for the formate addition to be considered successful. The U.S. drinking water limit for nitrite is 71 µM (1 mg N L-1). Denitrification was active during these tests because the background nitrous oxide was completely consumed in the presence of formate. If denitrification was fueled in situ by extracellular hydrolysis of formate to hydrogen, then hydrogen consumption must have approximately equaled production because high levels of hydrogen (µM amounts) were not found (data not shown). The long-term persistence of formate suggests that if hydrogen production was occurring, it was much slower than nitrate consumption. Perhaps more likely, then, is that these field tests stimulated formate-utilizing denitrifiers (or nitrate reducers) rather than the hydrogen-oxidizers. Only one of the HOD strains previously studied was able to also grow on formate as a sole carbon and energy source. Isolation and characterization of formate-utilizing denitrifiers from this aquifer would be a useful next step for interpreting the results from this field study. Laboratory Incubations. To better understand the effect of formate on nitrate and nitrite reduction in the field tests, laboratory batch incubations were conducted with aquifer sediments collected from the denitrification zone. Nitrate consumption occurred almost immediately, while nitrite consumption occurred only after a 5-day lag (Figure 6A,B). Nitrite concentrations initially increased when nitrate was the substrate and then decreased as the nitrate was depleted. Transient nitrite accumulation occurred for both high (1.0 mM) and low (0.2 mM) nitrate levels (Table 3). This is a 200
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
FIGURE 6. Consumption of nitrate, nitrite, and formate by aquifer sediments collected from the denitrification zone (10.5-10.0 m altitude) when slurried with artificial, anoxic groundwater and amended with 4 mM sodium formate and 1 mM sodium nitrate (A) or 1 mM sodium nitrite (B). Incubations were conducted at ambient temperature (15 °C); error bars represent ( one standard deviation.
TABLE 3. Consumption of Formate, Nitrate, and Nitrite during Laboratory Incubations with Aquifer Sediment Collected from the Denitrification Zone consumption ratec [nmol (gr wet wt × day)-1] incubation conditiona -
1 mM NO3
0.2 mM NO31 mM NO2-
time periodb (day no.)
formate
nitrate
nitrite
3-14 14-22 1-5 5-9 5-25
463 (38) 346 (74) 93 (21) 305 (28) 180 (10)
186 (6)
107 63 (0.3) 71 48 (5) 81 (5)
104 (18)
a Sediment was slurried with artificial groundwater and incubated at 15 °C. Shown are initial nitrogen oxide concentrations; formate concentration was 4.3 mM in all incubations. b Time interval selected based on pre- and post-peak nitrite concentration. There was some nitrate remaining during the first 2-3 days of the second interval (see Figure 6). c Rates calculated using linear regression of triplicate incubations after accounting for mass lost due to sample collection. Brackets enclose standard error; when no bracket, rate computed from rate of nitrate disappearance minus rate of nitrite accumulation.
typical denitrification response attributed to the differential responses of nitrate and nitrite reductases to a variety of factors, including low levels of oxygen (33), different types of electron donors (17, 34, 35), an inhibitory effect of nitrate on nitrite reduction (33), pH and temperature (36), differential enzyme kinetics (37), or even the presence of light (38). In the sediment slurries, the rate of nitrate consumption exceeded the rate of nitrite consumption (after nitrate depletion) by 2-3-fold (Table 3). The rate of nitrite consumption also decreased by 40% after the nitrate was exhausted to roughly the same rate irrespective of the initial nitrate concentrations (Table 3). Nitrite levels never exceeded concentrations reported to be toxic for growth of denitrifiers in culture (39) in either the lab or the field experiments. Formate consumption by the sediment slurries initially exhibited a 3-5 day lag, then increased markedly (Figure 6). Rates of formate consumption were 2.6-fold higher when
nitrate was the initial electron acceptor as opposed to nitrite (Figure 6, Table 3). However, when the initial nitrate concentration was low, the nitrate was apparently reduced primarily via other electron donors, as nitrate was consumed before the formate consumption increased (Table 3). Relatively high formate:nitrite molar consumption ratios (5-6) were evident after nitrate was depleted, suggesting that other formate-degrading populations, in addition to nitrate- and nitrite-reducers, were growing during the incubations. This result differs somewhat from the field tests in which formate consumption appears to be more closely tied to nitrate reduction. However, the aquifer is an open system, with water flow transporting substrate and product away from adapted microbial communities, the majority of which are stationary and attached to solid surfaces. Thus, the situation is more analogous to removing the groundwater from the nitrateconsuming incubations at the peak nitrite concentrations, when nitrate is low (Figure 6A), and continually transferring it to unadapted sediments (t ) 0 in Figure 6B). Under those conditions, complete consumption of formate and nitrite would require much longer time periods than might be suggested from a single set of closed-bottle incubations. Model Simulations. The field experiment was simulated using a two-dimensional vertical profile model that was developed to quantify in situ rates of nitrate and formate utilization and to predict the long-term fate of nitrite. The model is spatially aligned in the main direction of groundwater flow, with the injection well simulated along the inflow boundary, where the mass flux of formate, nitrate, and nitrite is specified as a function of time. Local perturbation to the velocity field from the injection process is neglected. The vertical transect was positioned in the mean direction of groundwater flow and includes multiport sampling wells 7, 10, and 15 m downgradient of the injection well. Advectiondispersion transport parameters were calibrated from the results of a previous bromide tracer experiment. Figure 7 (parts A-C) shows simulated and measured breakthrough curves for formate, nitrate, and nitrite, respectively, at the same location shown in Figure 5C (sampling port 9.58 m above sea level at the 15 m MLS). Through trial-and-error adjustment of model input parameters, the model was first calibrated to the formate, nitrate, and nitrite concentration data at four sampling ports from the 7 m MLS and evaluated using an error analysis (27). Uniqueness of the calibrated model parameters was investigated as part of a parameter sensitivity study. The results shown in Figure 7 demonstrate that the model was verified and captured the major data trends. Consumption rates for formate and nitrate were calculated using the calibrated transport model (see Supporting Information for model parameters). Utilization rates in the transport equations, expressed using dual Monod kinetic terms, are concentration dependent and vary with space and time. Rates are calculated based on the peak concentrations and represent maximum values (27). The resulting rates for formate loss under nitrate and nitrite reduction were 19 and 5.1 nmol (gr wet wt × day)-1, respectively. Nitrate and nitrite utilization rates were 3.8 and 1.0 nmol (gr wet wt × day)-1, respectively, and the nitrite production rate was 3.9 nmol (gr wet wt × day)-1. Nitrite was produced in the field experiment at a rate approximately four times greater than the maximum nitrite consumption rate. Compared to the laboratory incubations, these field rates for nitrate and nitrite consumption are considerably lower (see Table 3); an observation typical of lab vs field activity assays for subsurface microbial populations (25). Extrapolation of nitrite transport and subsequent consumption during the 1994 field test were investigated using the model. Figure 8 shows simulated formate, nitrate, and nitrite concentration distributions, respectively, at 100 days
FIGURE 7. Simulated and measured time course of formate (A), nitrate (B), and nitrite (C) concentrations in groundwater collected at an altitude of 9.84 m above sea level from the 15 m downgradient well for the 1994 injection experiment. after the start of formate injection. The formate cloud is nearly dissipated, except for the formate delivered during the first several days of the injection period, but the nitrite plume has not been removed. Two factors at work are electron donor (formate) availability and the preferential utilization of nitrate as an electron acceptor. The model predicted that nitrite reduction is inhibited except in zones where the nitrate concentration falls below 70 µM. As shown in Figure 7, sediment-bound nitrite-reducing bacteria had a limited time frame at the 15 m MLS (∼12-15 days) to utilize formate and nitrite in the absence of nitrate. After 100 days of transport, these conditions apparently no longer existed within the nitrite plume (Figure 8) and suggest that the nitrite would continue to persist during further transport. A secondary factor contributing to this result may be the mass flux of nitrate into the periphery of the formate cloud due to hydrodynamic dispersion combined with the inhibiting effect of nitrate, which resulted in an environment for inefficient nitrite consumption. Implications and Further Feasibility Studies. There have been other attempts to stimulate denitrification in situ as a means to remediate nitrate contamination in groundwater (40-44). Many of these studies involve injecting a substrate, such as denatured ethanol or sucrose, into an aquifer and withdrawing “remediated” water using a Daisy wheel configuration of wells. Nitrate removal efficiencies have ranged from 10 to 70% of background levels (which have exceeded drinking water limits in all cases) (41-44), usually with significant nitrite production (41, 42, 44). In some cases, nitrite VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
201
FIGURE 8. Simulated concentration distribution for formate, nitrate, and nitrite at 100 days after the start of the 1994 injection experiment. levels subsequently decreased with continued treatment (41, 44). In a much different approach, Robertson and Cherry (45) used a porous media barrier composed of sawdust mixed with silt or sand to treat septic tank effluents. In that case, denitrification eliminated 60-100% of the nitrate (up to 8.9 mM) in the septic effluent for up to a year, though no mention was made regarding nitrite accumulation. In this study, the formate treatment was successful in enhancing the reduction of nitrate to nitrite. In fact, >80% of the background nitrate was consumed within 15 m of transport downgradient from the injection well during these two relatively short-term injection experiments. Nitrite reduction was also stimulated, though as indicated, the rate of reduction was much lower than that of nitrate, resulting in nitrite accumulation with transport. The laboratory results from this study clearly suggested that continued contact in the field experiments of formate with nitrite (after the nitrate was consumed) would have resulted in enhanced nitrite reduction (Figure 6). Continued contact could best be accomplished by continuous formate injection, which should allow in situ adaptation of nitrite reduction within the first few meters of transport. As well as the nature and effectiveness of the in situ adaptation to utilize nitrite, other aspects of this approach will require further examination before it would be acceptable as a remediation practice. The long-term effect of formate upon subsurface denitrifiers or other microorganisms is unknown. For example, clogging due to biomass buildup is a potential problem that has been encountered in many other groundwater studies. The utility of the approach when oxygen is present also needs to be examined. Given the capacity of denitrifiers to grow aerobically, it is likely that formate could be used to remove both oxygen (a requirement for denitrification to commence) and then nitrate. This would be a desirable attribute because most nitrate-laden groundwater also contains oxygen. Methods of formate delivery to the subsurface and the economic feasibility of using formate also need to be considered. Overall, this study is a first step toward demonstrating that formate-enhanced denitrification does have potential as a remediation approach for nitrate contamination. Largerscale, longer-term feasibility studies are needed to confirm this conclusion and to characterize the conditions within 202
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
which the approach will or will not function suitably as a remediation tool.
Acknowledgments We thank D. R. LeBlanc, coordinator of the Cape Cod site, for field and technical assistance and Barbara Bekins and Isabelle Cozzarelli for manuscript reviews. This study was funded by the U.S. Geol. Survey Toxic Substances Hydrology Program and by U.S. Department of Agriculture grant #9537101-1713. The use of trade or product names in this paper is for identification purposes only and does not constitute endorsement by the U.S. Geol. Survey.
Supporting Information Available Equations for the two-dimensional transport model that simulates formate-enhanced denitrification, description of model variables and model parameters, and a table of the parameter values used for the simulations presented in Figures 7 and 8. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Spalding, R. F.; Exner, M. E. In Nitrate Contamination: Exposure, Consequence and Control; Bogardi, I.; Kuzelka, R. D., Ennenga, W. G., Eds. Springer-Verlag: New York, 1991; pp 13-48. (2) Nolan, B. T.; Ruddy, B. C.; Hitt, K. J.; Helsel, D. R. Environ. Sci. Technol. 1997, 31, 2229-2236. (3) Gayle, B. P.; Boardman, G. D.; Sherrard, J. H.; Benoit, R. E. J. Environ. Eng. 1989, 115, 930-943. (4) Mateju, V.; Cizinska, S.; Krejci, J.; Janoch, T. Enzyme Microb. Technol. 1992, 14, 170-183. (5) Smith, R. L.; Ceazan, M. L.; Brooks, M. H. Appl. Environ. Microbiol. 1994, 60, 1949-1955. (6) Liessens, J.; Vanbrabant, J.; De Vos, P.; Kersters, K.; Verstraete, W. Microb. Ecol. 1992, 24, 271-290. (7) Gros, H.; Schnoor, G.; Rutten, P. Water Supply 1988, 6, 193198. (8) Peck, H. D.; Gest, H. J. Bacteriol. 1957, 73, 706-721. (9) Rossmann, R.; Sawers, G.; Bock, A. Mol. Microbiol. 1991, 5, 28072814. (10) Dijkhuizen, L.; Harder, W. J. Gen. Microbiol. 1984, 130, 447457. (11) Phelps, T. J.; Zeikus, J. G. Appl. Environ. Microbiol. 1985, 50, 1285-1291. (12) de Graaf, W.; Cappenberg, T. E. Appl. Environ. Microbiol. 1996, 62, 3535-3537.
(13) Hungate, R. E. Annu. Rev. Ecol. Syst. 1975, 6, 39-66. (14) Thiele, J. H.; Zeikus, J. G. Appl. Environ. Microbiol. 1988, 54, 20-29. (15) Schink, B. Microbiol. Mol. Biol. Rev. 1997, 61, 262-280. (16) van Verseveld, H. W.; Stouthamer, A. H. Arch. Microbiol. 1978, 118, 21-26. (17) van Rijn, J.; Tal, Y.; Barak, Y. Appl. Environ. Microbiol. 1996, 62, 2615-2620. (18) Nishimura, Y.; Kamihara, T.; Fukui, S. Arch. Microbiol. 1980, 124, 191-195. (19) LeBlanc, D. R. U.S. Geol. Survey Water Supply Paper 2218 1984, 28 pp. (20) Smith, R. L.; Howes, B. L.; Duff, J. H. Geochim. Cosmochim. Acta 1991, 55, 1815-1825. (21) Barber, L. B., II.; Thurman, E. M.; Schroeder, M. P.; LeBlanc, D. R. Environ. Sci. Technol. 1988, 22, 205-211. (22) Harvey, R. W.; Barber, L. B., II J. Contam. Hydrol. 1992, 9, 91103. (23) Smith, R. L.; Duff, J. H. Appl. Environ. Microbiol. 1988, 54, 10711078. (24) Brooks, M. H.; Smith, R. L.; Macalady, D. L. Appl. Environ. Microbiol. 1992, 58, 1746-1753. (25) Smith, R. L.; Garabedian, S. P.; Brooks, M. H. Environ. Sci. Technol. 1996, 30, 3448-3456. (26) Zapico, M. M.; Vales, S.; Cherry, J. A. Ground Water Monit. Rev. 1987, 74-82. (27) Killingstad, M. W.; Widdowson, M. A.; Smith, R. L. J. Environ. Eng., submitted for publication. (28) Widdowson, M. A.; Molz, F. J.; Benefield, L. D. Water Resour. Res. 1988, 24, 1553-1565. (29) Fishman, M. J.; Friedman, L. C. U.S. Geol. Survey Open-File Report 85-495 1985. (30) Kent, D. B.; Davis, J. A.; Anderson, L. C. D.; Rea, B. A.; Waite, T. D. Water Resour. Res. 1994, 30, 1099-1114.
(31) Smith, R. L.; Harvey, R. W.; LeBlanc, D. R. J. Contam. Hydrol. 1991, 7, 285-300. (32) Barbaro, J. R.; Barker, J. F.; Lemon, L. A.; Mayfield, C. I. J. Contam. Hydrol. 1992, 11, 245-272. (33) Korner, H.; Zumft, W. G. Appl. Environ. Microbiol. 1989, 55, 1670-1676. (34) Blaszczyk, M. Appl. Environ. Microbiol. 1993, 59, 3951-3953. (35) Kelso, B. H. L.; Smith, R. V.; Laughlin, R. J. Appl. Environ. Microbiol. 1999, 65, 61-66. (36) Wilderer, P. A.; Jones, W. L.; Dau, U. Water Res. 1987, 21, 239245. (37) Betlach, M. R.; Tiedje, J. M. Appl. Environ. Microbiol. 1981, 42, 1074-1084. (38) Barak, Y.; Tal, Y.; van Rijn, J. Appl. Environ. Microbiol. 1998, 64, 813-817. (39) Williams, D. R.; Rowe, J. J.; Romero, P.; Eagon, R. G. Appl. Environ. Microbiol. 1978, 36, 257-263. (40) Hiscock, K. M.; Lloyd, J. W.; Lerner, D. N. Water Res. 1991, 25, 1099-1111. (41) Hamon, M.; Fustec, E. J. Water Pollut. Contr. Fed. 1991, 63, 942-949. (42) Janda, V.; Rudovsky, J.; Wanner, J.; Marha, K. Water Sci. Technol. 1988, 20, 215-219. (43) Mercado, A.; Libhaber, M.; Soares, M. I. M. Water Sci. Technol. 1988, 20, 197-209. (44) McMahon, P. B.; Dennehy, K. F.; Bruce, B. W. U.S. Geol. Survey Water-Resour. Invest. Report 98-4046 1998, 34 pp. (45) Robertson, W. D.; Cherry, J. A. Ground Water 1995, 33, 99-111.
Received for review June 12, 2000. Revised manuscript received September 27, 2000. Accepted October 4, 2000. ES001360P
VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
203
