Remediation of Nitrate-Contaminated Surface Water Using Sulfur and Limestone Autotrophic Denitrification Processes

J.M. Flere and T.C. Zhang

125 B Engineering Hall, Dept. of Civil Engineering, University of Nebraska-Lincoln at Omaha Campus, Omaha, NE 68182-0178, Phone: (402) 554-3784, FAX: (402) 554-3288


The feasibility of using the sulfur/limestone autotrophic denitrification (SLAD) process as an in situ method for remediation of nitrate-contaminated surface water was investigated. Four bench-scale pond systems with working volumes of 21.4 liters each and hydraulic retention time (HRT) of 30 days were operated under mixed conditions. Under mixed (aerobic) conditions, with the addition of alkalinity to raise pH, NO3--N removal in the SLAD ponds was 85 - 100%, while the control reactor showed negative removal. Sulfate production under mixed conditions was between 1000 - 2500 mg/l SO42-, which shows that 40 - 60 mg/l of SO42- is produced for every 1 mg/l of NO3--N reduced. Although the system is very efficient in removing nitrates under simulated surface water conditions, the sulfate production makes the process questionable for use under aerobic conditions. However, batch experiments under anaerobic conditions demonstrate that system maybe very efficient in removing nitrate while not producing insufferable amounts of sulfates.

Keywords: autotrophic denitrification, nitrates, sulfur, in situ remediation, thiobacillus denitrificans


In the past several decades numerous studies have been conducted on in situ methods of reducing nitrogen from municipal, industrial, and agricultural wastewater through the use of constructed wetlands or stabilization ponds. In these systems the potential removal mechanisms include storage in the living biomass, detritus, and sediments; ammonia volatilization; and denitrification. Of these mechanisms the most efficient, cost-effective, and long-term method is biological denitrification. Biological denitrification can be broken down into two groups: heterotrophic and autotrophic. Heterotrophic denitrification is very efficient in nitrate removal provided that adequate amounts of organic carbon are available. However, in many situations, such as in groundwater, lakes, and polishing ponds for tertiary treatment, insufficient amounts of organic carbon may limit the application of in situ heterotrophic denitrification unless organic substances are added as external carbon sources [e.g., methanol (CH3OH), glucose, and glycerol]. Sulfur-based autotrophic denitrification utilizes autotrophic denitrificans, such as Thiobacillus denitrificans and Thiomicrospira denitrificans, to reduce nitrates to nitrogen gas while using sulfur as the electron donor. The process requires no external carbon source and produces low amounts of biomass. Therefore, sulfur-based autotrophic denitrification could be an efficient and cost-effective replacement for heterotrophic denitrification.

Sulfur and limestone autotrophic denitrification (SLAD) processes have been studied by many researchers in Europe (van der Hoek, et al., 1992a, 1992b; Kruithof, et al., 1988; Schippers, et al., 1987; Claus and Kuntzer, 1985; Baalaruud and Baalaruud, 1954) and a few in the U.S. (Lampe and Zhang, 1996; Davidson and Ridgeway, 1995; Batchelor and Lawrence, 1978a, 1978b, 1978c). All of these studies have concentrated on using the process for ex situ treatment of nitrate-contaminated groundwater. These SLAD systems are highly efficient and comparable to heterotrophic denitrification systems in effectiveness in this type of treatment. The SLAD process, however, has never been studied as an in situ treatment method for remediation of nitrate-contaminated surface water other than batch experiments conducted by Lampe and Zhang (1997, 1996). If successful, this treatment option would allow in situ remediation of nitrate-contaminated surface water and offer a safe and efficient method for nitrate removal.

The objective of this study was to evaluate the feasibility of using sulfur/limestone autotrophic denitrification pond systems for the in situ remediation of nitrate-contaminated surface water. Specifically, the feasibility of pond systems was evaluated based on reactor-nitrate removal efficiencies and the amount and type of by-products produced in the systems. The mechanisms of nitrate removal in these pond systems were also evaluated based on changes of operation modes, the monitoring of the operational parameters, and bacterial counts.

Materials and Methods

Pond reactors

Four bench-scale SLAD ponds were used as the main reactors for the project. Figure 1 shows the schematic setup of a SLAD pond reactor. All reactors were constructed with a baffle at the influent point to prevent the influent from short-circuiting the process and a weir system at the effluent line to allow a uniform collection of the effluent. Mixing mechanisms were also incorporated into the design to allow oxygen to enter the pond. Each pond reactor had a working volume of 21.1 liters with a hydraulic retention time (HRT) of 30 days. The flow rate of the system was 0.48 ml/min. The feed was pumped through a Masterflex L/S Peristaltic pump with four pump heads.

The four pond reactors were set up as follows: 1) reactors 1 and 4 contained 1.75 in. of sediment from a rural cattle pond covered with a layer of granular sulfur/limestone (S/L) (2.38 mm to 4.76 mm) with a S/L ratio of 3:1 (Reactor 4 was the same as reactor 1 except that 1/3 less S/L was placed into the system.); 2) reactor 2 was set up exactly like reactor 1 except that the reactor was seeded with 250 ml of seed sludge (Thiobacillus denitrificans) from an anaerobic completely-stirred tank reactor which had been cultured in our laboratory for more than two years; and 3) reactor 3 was a control reactor with only the sediment placed in the system.

The synthetic surface water composition was adapted from Lampe and Zhang (1996) for an influent concentration of 30 mg/l NO3--N or a volumetric loading rate of 1.0 mg/l/day. The substrate consisted of tap water with the addition of laboratory grade chemical, including 0.217 mg/l of KNO3 and 0.0361 mg/l KH2PO4. Sodium bicarbonate (Na2HCO3) was also used in the feed from day 77 to day 153 in the experiment to evaluate the effect of alkalinity and pH on nitrate removal and sulfate production.

Analytical methods

To measure the system efficiency, NO3--N, NO2--N and sulfate were measured duplicately based on HACH methods (1992). These tests were conducted using spectrophotometric methods. Nitrate-nitrogen was measured using a modified cadmium reduction method with readings taken at a wavelength of 500 nm. To insure that all nitrate was accounted for, 30 mg/l bromine water was added dropwise until a yellowish color persisted. This converts any nitrite to nitrate. To stop the bromine water from interfering in the readings, phenol was then added to remove the yellowish color. Nitrite-nitrogen was tested in the low range utilizing the diazotization method at a wavelength of 507 nm. Sulfate was also determined spectrophotometrically using the SulfaVer 4 method at 450 nm as described in HACH methods (1992). All measurements were taken on a Milton Roy Spectrophotometer 21 or a HACH Direct Reading (DR)/2000 spectrophotometer. pH and dissolved oxygen were measured using electrodes. pH was measured using two different meters. All effluent samples were tested with a Fisher Scientific Accu-pHast polymer combination electrode with an Accumet 925 pH meter; all pH data taken from the reactor interior was measured with a Beckman 100 F ISFET pH meter with standard Beckman pH probe. The pH measurements taken from the reactor interior were at the depth of 4 inches in the reactor center. Dissolved oxygen was determined using a Leeds and Northrup 7932 Portable dissolved oxygen meter with a probe. Data was collected from the effluent samples and from the center of the pond reactor at a depth of 2 inches. All samples were filtered with a 0.45 mm membrane filter before analysis and all samples were tested within 7 days of sampling. Any stored samples were placed in refrigeration until testing. In addition, all samples were allowed to warm to room temperature before testing was conducted.

Results and Discussion

Nitrate-nitrogen removal

Figure 2 shows the NO3--N removal efficiencies of the four pond reactors over the experimental duration. From day 1 to day 40, the reactors were not sampled regularly due to the start up stage involved. It can be seen that all of the SLAD reactors had approximately the same nitrate removal, ranging from 40% to 100%; while the control reactor (reactor 3) had nitrate removals fluctuating from +20% to -40%. Based on this preliminary experiment, the following observations need to be discussed. (1) The SLAD pond systems (reactors 1, 2 and 4) removed nitrates efficiently from the system. With the addition of Na2HCO3, the system alkalinity was controlled and the overall nitrate removal efficiencies increased slowly over time and could reach a 90 - 100% reduction of nitrates in all the SLAD pond systems. Without alkalinity control, the nitrate removal efficiencies of the SLAD were lower than that with alkalinity control, but were still as high as 83%. (2) The seed sludge of thiobacillus denitrificans added to reactor 2 did not cause the system to out-perform reactors 1 and 4 after the one-month start up period. The inoculation of reactor 2 did accelerate the autotrophic denitrification process in that reactor at the very beginning (data not shown), but autotrophic denitrificans in the sediment could quickly multiply to catch up to the population in the seeded reactor (reactor 2). This observation is consistent with Lampe and Zhang (1997). According to our additional batch experiments (Zhang and Lampe, 1997), different sediments sampled from different natural environments resulted in similar nitrate removal results, indicating that the autotrophic denitrifiers widely exist in natural sediments. Therefore, our experiments have demonstrated the possibility of using the SLAD pond system for in situ remediation of nitrate-contaminated surfaces. (3) Although reactor 4 contained one third less S/L, its performance was as good as reactors 1 (S/L) and 2 (S/L with seed). Therefore, there may exist some optimal dosages of S/L for the pond systems, which necessitates further studies. (4) The nitrate removal efficiency in the control reactor fluctuated between +20% and -20% for the first 75 days. The system then ended in a pattern of negative nitrate removal efficiencies near -40 to -50%. It is surmised that the system had nitrification occurring and was producing more nitrates than were introduced. Based on the initial nitrate-nitrogen concentration in the feed solution (30 mg N/l), it is estimated that approximately 30 mg/l to 50 mg/l NO3--N was produced through the nitrification process in all the pond systems (including reactors 1, 2, and 4). Accordingly, the real nitrate removal efficiencies in the SLAD pond may be higher than what are shown in Figure 2, which demonstrates the potential of using the SLAD pond system to treat surface water contaminated with higher nitrate concentrations (e.g., 100 mg N/L).

Effect of pH

Figure 3 shows the time courses of pH and NO3--N removal efficiency in the SLAD pond systems. The initial data points were collected after 41 days, which is slightly past one full retention time. From day 1 to day 60, the pH of the SLAD pond systems (Figure 3a, 3b, 3d) was approximately 4, which is about 2 to 3 pH units lower than the control reactor. The system pH in the reactors rose after initially being low and peaked near day 72. Following this day, the pH started to fall, and along with the pH the nitrate-nitrogen removal efficiency fell. The exact mechanism for this phenomenon was unknown until a later time. It was thought that the pH decrease was induced by the autotrophic denitrification process. According to Batchelor and Lawrence (1978a, 1978b): Thiobacillus denitrificans reduces nitrates using elemental sulfur based on the following stoichiometric equation: 55 S + 20 CO2 + 50 NO3- + 38 H2O + 4 NH4+ 4 C5H7O2N + 25 N2 + 55 SO42- + 64 H+. Hydrogen ions and sulfates are produced by the process. Therefore, on day 77, 1.36 mg/l of Na2HCO3 was added to the feed solution to raise alkalinity and pH in the system, and the pH slowly rose (see Figure 3a, 3b). On day 92, a large increase in pH occurred due to the amount of Na2HCO3 being raised to 1.64 mg/l. With the large increase in pH came enhanced removal efficiency in the pond systems. Reactors 1 and 2 even reached nitrate removal efficiencies up to 95 - 100%. Figure 3c shows the time courses of pH and NO--N removal efficiency in reactor 3, the control reactor. It should be pointed out that the pH in the control reactor never decreased as it did in reactors 1, 2, and 4. This shows that some process initiated by the sulfur and limestone caused the pH to decrease which called for the addition of alkalinity to the system to maintain an environment allowing the T. denitrificans to multiply.

Figure 3d shows the time courses of pH and NO3--N removal efficiency in reactor 4. The 4th reactor was started 2 weeks later than reactors 1 - 3. This pond system did not meet the same fate as reactors 1 and 2 with the radical pH changes. The pond behaved exactly like ponds 1 and 2 and started with a low pH until it spiked on day 72, but the pH did not decrease between days 72 to 84 as in reactors 1 and 2.

Sulfate production

Figure 4 shows the sulfate production in the pond reactors. The influent sulfate concentration was in almost all cases between 100 and 250 mg/l of SO42-. In the control reactor, reactor 3, the sulfate levels were slightly elevated, ranging from 300 to 500 mg/l of SO42-. Since no denitrification was evident in reactor 3, this increase in SO42- concentration could be attributed to leaching from the sediment. The sulfate concentration was continually growing and reached amounts of 1500 to 2500 mg/l of SO42- in reactors 1, 2 and 4. Based on the stoichiometric equation for autotrophic denitrification (Batechlor and Lawrence, 1978a, 1978b), for reduction of every 1 mg/l of NO3--N, approximately 7.1 mg/l of SO42- should be produced. However, the reduction of 1 mg/l of NO3--N was producing SO42- between 40 - 60 mg/l in our SLAD pond systems under the mixed conditions. This data shows that there is a problem with the process and not all of the transformation mechanisms in the experiment are known. The U.S. EPA has established the secondary maximum contamination level of SO42- in drinking water to be 250 mg/l. Therefore, the sulfate concentration in the effluent from SLAD pond systems becomes a critical issue in the process feasibility.

Batch reactors were also used to simulate the SLAD pond systems under both mixed (aerobic) and unmixed (anaerobic) conditions (for detailed conditions of batch experiments, see Zhang and Lampe, 1997). Figure 5 shows sulfate vs. time in batch pond systems. Under aerobic conditions (mixing conditions), sulfate was produced continuously; at the end of the test (275 hours), more than 660 mg/l sulfate was detected. Sulfate was produced rapidly before the nitrate was completely reduced (before 50 hours), and was produced slowly even after the nitrate was completely. Under anaerobic conditions, sulfate concentrations in all cases dipped in the first 10 to 12 hours, rebounded during the next 50 hours, and generally leveled off thereafter. These additional batch tests demonstrate the feasibility of the SLAD systems under unmixed or anaerobic conditions.


The feasibility of using the sulfur/limestone autotrophic denitrification process as an in situ method for remediation of nitrate-contaminated surface water was investigated in this study. Under mixed (aerobic) conditions, the nitrate-nitrogen removal efficiencies were as high as 90 to 100% when the pH in the SLAD ponds, was properly controlled by adding alkalinity in the systems. The SLAD ponds however, produced sulfate under mixed conditions, much higher than the MCL established by the U.S. EPA, which makes the application of the SLAD ponds under aerobic conditions questionable. Although further testing must be conducted to determine the actual sulfate concentration produced in a continuously-fed unmixed (anaerobic) SLAD system, the preliminary batch experiments conducted in this study indicate that the SLAD process may be a viable alternative to heterotrophic denitrification processes that can be used for nitrate control or removal in polishing ponds for tertiary treatment, and that the application of the SLAD ponds under unmixed conditions is feasible for in situ remediation of nitrate-contaminated surface water.


We gratefully acknowledge Mr. P. Ponugoti and Mr. D. G. Lampe of the University of Nebraska-Lincoln (UNL) for their assistance with batch experiments and bacterial counts. This work was funded by the National Water Research Institute, U.S.A. with matching funds provided by the Water Center/Environmental Programs of UNL and the Center for Infrastructure Research of UNL, which is greatly appreciated.


Baalaruud, K., and Baalaruud, K.S. (1954) Studies on Thiobacillus denitrificans. Archivfur Mikrobiologie. Vol. 20:34-62.

Batchelor, B., and Lawrence, A.W. (1978a) A kinetic model for autotrophic denitrification using elemental sulfur. Water Research. Vol. 12:1075-1084.

Batchelor, B., and Lawrence, A.W. (1978b) Autotrophic denitrification using elemental sulfur. J. WPCF. Vol. 50:1986-2001.

Batchelor, B. and Lawrence, A.W. (1978c) Stoichiometry of autotrophic denitrification using elemental sulfur. In: A.J. Rubin (Ed.), Chemistry of Wastewater Technology. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, pp. 421-440.

Claus, G., and Kuntzer, H.J. (1985) Autotrophic denitrification by Thiobacillus denitrificans in a packed bed reactor. Applied Microbiology and Biotechnology. Vol. 22:289-296.

Claus, G., and Kuntzer, H.J. (1985) Physiology and kinetics of autotrophic denitrification by thiobacillus denitrificans. Applied Microbiology and Biotechnology. Vol. 22:283-288

Davidson, M.S., and Ridgway, H.F. (1995) Biological denitrification of groundwater using sulfur-oxidizing bacteria: construction and evaluation of a 25 gpm pilot system. A cooperative project of Orange County Water District and the Metropolitan Water District of Southern California.

HACH Water Analysis Handbook. (1992) 2nd Ed., HACH Company, Loveland, Colorado.

Kruithof, J.C., van Bennekom, C.A., Dierx, H.A.L., Hijnen, W.A.M., van Paassen, J.A.M., and Schippers, J.C. (1988) Nitrate removal from groundwater by sulfur/limestone filtration. Water Supply. Vol. 6: 207-217.

Lampe, D.G., and Zhang, T.C. (1996) Evaluation of sulfur-based autotrophic denitrification. Proceedings of the HSRC/WERC Joint Conference on the Environment. pp. 444-458. Albuquerque, NM, May 21- 23, 1996.

Lampe, D.G., and Zhang, T.C. (1997) Sulfur-based autotrophic denitrification for remediation of nitrate-contaminated water. In Situ and On-Site Bioremediation: Volume 3, pp. 423-428. Papers from the Fourth International In Situ and On-Site Bioremediation Symposium. New Orleans, April 28-May 1, 1997. Bettelle Press.

Schippers, J.C., Kruithof, J.C., Mulder, F.G. and van Lieshout, J.W. (1987) Removal of nitrate by slow sulphur/limestone filtration. Journal of Water SRT-Aqua. Vol. 62:1045-1052.

van der Hoek, J.P., Hijnen, W.A.M., van Bennekom, C.A., and Mijnarends, B.J. (1992a) Optimization of the sulphur-limestone filtration process for nitrate removal from groundwater. Journal of Water SRT- Aqua. Vol. 41:209-218.

van der Hoek, J.P., Kappelhof, J.W.N.M., and Hijnen, W.A.M. (1992b) Biological nitrate removal from ground water by sulphur/limestone denitrification. J. Chemical Tech. Biotech.. Vol. 54:197-200.

Figure 1. Sulfur/limestone autotrophic denitrification pond reactor.

Figure 2. NO3-- removal efficiency over time in pond systems. The down arrow indicates the addition of Na2HCO3 to control alkalinity since day 77.

Figure 3. Relationship of pH to nitrate removal efficiencies over time. (a) Reactor 1; (b) reactor 2; (c) reactor 3; and (d) reactor 4. The down arrow indicates the addition of Na2HCO3 to control alkalinity since day 77.

Figure 4. Sulfate production in pond systems. The down arrow indicates the addition of Na2HCO3 to control alkalinity since day 77.

Figure 5. Time courses of sulfate production in batch pond reactors. The initial sulfate concentrations in tap water was 200 mg/l. (a) Under aerobic conditions; and (b) under anaerobic conditions.