- Email: [email protected]
Biological denitrification in a continuous-flow pilot bioreactor containing immobilized Pseudomonas butanovora cells
Bioresource Technology 87 (2003) 75–80
Biological denitrification in a continuous-flow pilot bioreactor containing immobilized Pseudomonas butanovora cells Peter Kesser} u *, Istvan Kiss, Zolt an Bihari, Bela Poly ak Bay Zolt an Foundation for Applied Research, Institute for Biotechnology, Derkovits fasor 2., P.O. Box 2337, Szeged H-6726, Hungary Received 13 May 2002; received in revised form 9 August 2002; accepted 14 August 2002
Abstract Pseudomonas butanovora, a novel denitrifying bacterium, was immobilized in composite beads and filled into a reactor system. The pilot bioreactor average denitrification activity was at ethanol–C:nitrate–N ratios of 3:1 and 1.5:1 0.88 and 0.54 kg NO 3– N m3 d1 , respectively. The denitrification was stable in spite of the relatively low hydraulic retention times of 2.47 and 3 h. The nitrate content of the influent was almost completely reduced at the first level of the bioreactor and the nitrite formed underwent reduction in the upper part of the reactor. The experimentally determined optimum ethanol–C:nitrate–N ratio was 1:41 0:41. In consequence of the aerobic conditions, the acetic acid produced by the oxygenation of ethanol was also detectable in the reactor effluent. The pH of the effluent (7.58) never exceeded the acceptable maximum (8.5). The nitrate removal efficiency of the cells was nearly 100% at both C:N ratios, and the nitrite content of the effluent was around the prescribed limit throughout the continuous operation. This continuous-flow pilot bioreactor containing immobilized P. butanovora cells proved an efficient denitrification system with a relatively low retention time. Ó 2002 Elsevier Science Ltd. All rights reserved. Keywords: Acetic acid; Cell immobilization; Denitrification; Ethanol; Pseudomonas butanovora
1. Introduction Nitrate (NO 3 ) contamination of groundwater has been documented worldwide (Mansell and Schroeder, 1999). The ingestion of high levels of NO 3 may have serious effects on human health in consequence of the reduction to nitrite (NO 2 ) and the formation of nitrosamines (Bouchard et al., 1992). This danger necessitates the removal of NO 3 from water reserves. Biological denitrification is an attractive treatment option, for the NO 3 is converted by the denitrifying bacteria to inert nitrogen gas and the waste product usually contains only biological solids. In recent years, very many different technologies have been developed in efforts to solve the problem of NO 3 in drinking, waste or groundwater, e.g. bio-electrochemical systems (Sakakibara and Kuroda, 1993; Kiss et al., 2000), membrane reactors (McCleaf and Schroeder, 1995; Fuchs et al., 1997; Mansell and Schroeder, 1999), fluidized bed reactors * Corresponding author. Address: Zoltan Bay Foundation for Applied Research, Institute for Biotechnology, P.O. Box 2337, Szeged 6701, Hungary. Tel.: +36-62-432-252; fax: +36-62-432-250. E-mail address: [email protected] (P. Kesser} u).
(Croll et al., 1985), a porous ceramic plate reactor (Nagadomi et al., 2000), a rotating of biological contactor (RBC) system (Mohseni-Bandpi and Elliott, 1998) or with the help of gel-immobilized bacteria (Chang et al., 1999; Chen et al., 1996). A number of factors must be considered in the choice of the bioreactor system and the carbon (C) source, such as cost, denitrification rate, kinetics, degree of utilization, handling and storage safety, stability, etc. RBC systems and those technologies where a biofilm-covered surface is responsible for the denitrification are commonly used in wastewater treatment. The application of membrane reactors and bacteria immobilized in a gel matrix is becoming popular in drinking water purification. One important aim during the selection of an appropriate bioreactor is to achieve minimal NO 3 and NO2 concentrations and the lowest possible C content of the effluent or to attain a high denitrification rate. Membrane reactors and a bio-electrochemical system fulfil the former requirements, though their effectiveness is relatively low. Other methods, such as the RBC system or fluidized bed reactors, have very high denitrification activities, but their effluents almost always need some post-treatment because of the biomass accumulation. During continuous
0960-8524/03/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 0 - 8 5 2 4 ( 0 2 ) 0 0 2 0 9 - 2
76
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
operation, the effectiveness and the stability of these reactors depends on the structure of the biofilm formed (Alves et al., 2002). The concentration of the substrates in the influent affects the growth and the composition of the biofilm. The growth and the activity of the bacteria increase if the C source is available in excess, but this leads to a higher cell accumulation as well. Nutrient limitation can result in activity loss and biofilm detachment. Destruction of the biofilm by the high velocity of the liquid phase (shear stress) is also frequent. All these effects can be avoided by the application of cells immobilized in a gel matrix (Chang et al., 1999; Chen et al., 1996; Kesser} u et al., 2002). The application of gel immobilization techniques provides a continuously high cell concentration in the bioreactor, independently of the influent flow rate. Although nutrient limitation also has an effect on gel-immobilized cells, it does not radically modify the biomass concentration and the recovery of the activity is usually faster. Consideration of the aspects listed above in connection with bioreactors containing gel-immobilized cells could lead to more balanced systems than technologies involving a surface biofilm. Methanol, ethanol and acetic acid are generally used as additional or individual C sources in these processes because they are easily available and give a high denitrifcation rate (Æsøy et al., 1998). The usage of ethanol is common not only in experimental pilot plants (Fuchs et al., 1997; Æsøy et al., 1998; Mohseni-Bandpi and Elliott, 1998), but also in full-scale technologies (Hallin et al., 1996; Hasselblad and Hallin, 1996). The present work was undertaken in order to investigate the stability and denitrification activity of a bioreactor containing gel-immobilized Pseudomonas butanovora cells. The bioreactor was driven with ethanol as a C source with a relatively low hydraulic retention time (HRT).
2. Methods 2.1. Culture conditions The Pseudomonas butanovora strain was isolated from the soil of highly contaminated land formerly used for military purposes. The strain was maintained on DSM 1 agar plates (5 g peptone, 3 g meat extract and 15 g agar in 1000 ml distilled water). The bacterium was enriched in shaken cultures in DSM 1 broth, pH ¼ 7 (200 rpm, at 37 °C). The bacterial biomass for cell immobilization was produced in the 80 l New Brunswick Mobile Pilot Plant fermentation system. 2.2. Cell immobilization The grown culture was collected by centrifugation (Sorvall RC 3B, 2600 g, 30 min, 10 °C). The pellet was
resuspended in composite gel solution (Kov acs and Polyak, 1991) at a final cell concentration of 5% (w/w). The composite solution was prepared by mixing 1.6% Na-alginate (Sigma) in 200 mg l1 Na2 -EDTA solution and 1.6% Gelrite (Sigma) in 200 mg l1 Na2 -EDTA solution in a volumetric ratio of 1:1. The two solutions were sterilized separately and mixed when hot. After the composite solution had cooled down, the collected bacterial cells were added. The mixed gel solution was added in drops to 1% CaCl2 solution. The average diameter of the resulting beads was 2 mm. The beads were stirred at 25 °C for 40 min and were then incubated at 4 °C for 12 h. 2.3. Bioreactor operation The total volume of the reactor up to the top level 4 was 4700 ml, with height 41 cm, and diameter 15 cm. The beads functioning as the active volume of the reactor were placed on four indented plates which were fixed at equal distance from each other, and the bottom of the reactor. The total volume of the beads loaded into the four levels of the reactor was 2900 ml. The NO 3 –N concentration of the influent solution was set at 50 mg l1 . A constant flow rate was applied, at which the average HRT of the influent referred to the total volume of the reactor was 2.5–3 h. During the first period of operation, the amount of influent ethanol used as C source was determined to attain an ethanol–C:NO 3 –N ratio in the influent solution of 3:1; this was later decreased to 1.5:1. A peristaltic pump was used to load the NO 3 and ethanol stock solutions (13% and 10%, respectively) in appropriate proportions into the main influent flow. The non-sterilized influent solution applied at the bottom of the reactor was supplemented with 0.05% CaCl2 to preserve the stability of the beads. The reactor was operated at 25 °C. Samples were taken from each level of the bioreactor every 24 h, and the NO 3 and NO2 concentrations of the samples were determined to study the spatial separation of the NO 3 and NO 2 reduction steps of the denitrification process. The ethanol concentration and the pH were also determined in each sample. The numbers of living cells in the samples were measured by viability counts on DSM 1 agar plates. The identification of P. butanovora was done by its colony morphology. 2.4. Calculation of denitrification rate 3 The eliminated kg NO 3 –N by the bioreactor m a day, were calculated as the following equation:
ðinput NO 3 –N concentration output NO 3 –N concentrationÞ flow rate=volume of beads
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
The beads served as the biologically active volume of the bioreactor.
2.5. Analysis NO 3 and NO2 analyses were performed in accordance with Hungarian Standard no. 448/12–82 (1983). The measurements of ethanol and acetic acid were carried out on a Gynkotek HPLC instrument, with a SARASEP CAR-H column (temperature 50 °C). The sample detection was carried out by a refractive index detector and a UV detector. The eluent was 0.01 N H2 SO4 and its flow rate was 0.9 ml min1 . Data were evaluated with CHROMELEON software.
2.6. Oxygen consumption of the cells During the measurement of the oxygen consumption, the ratio of the beads and the influent medium volume referred to the ratio set in the case of the 1 level of the 4.7-l bioreactor. The reactor was fed continuously with 1 influent containing 50 mg NO and ethanol in an 3 –N l ethanol–C:NO3 –N ratio of 3:1 for more than a week before the measurements. During the batch measurements, the medium was stirred to allow the diffusion of oxygen inside the beads. In the case of continuous operation, the oxygen electrode was placed 1 cm above the beads, but its position was changed horizontally. Data were collected 10 min after the electrode was placed into the medium in order to avoid the interfering effect of the N2 produced on the dissolved oxygen (DO) measurement. Fifty individual measurements were made and the average is given.
77
3.2. Utilization of the carbon source During the first operation, the ethanol concentration was set to attain a C:N ratio of 3:1 in the influent in order to provide the immobilized cells with ample reducing power to achieve complete denitrification (Kesser} u et al., 2002). A decrease in the ethanol concentration was measured in the samples taken from the different levels of the reactor. Fig. 1(a) presents the decline in ethanol concentration between the levels. The ethanol concentration decreased to the highest extent up to level 1, i.e. the part of the reactor where 80% of the influent NO 3 was reduced. At the other levels, the degree of decrease in the ethanol concentration was progressively less; this was definitely correlated with the lower efficiency of the reduction at these levels. The average difference between the ethanol concentrations of the influent and the effluent solutions was 42% (if the concentration in the influent was taken as 100%). From day 2 of operation, acetic acid could be detected by HPLC measurements at the different levels of the bioreactor. When the concentrations of ethanol and DO in the influent are high enough, ethanol can be oxidized to acetic acid (Mohseni-Bandpi and Elliott, 1998). The concentrations of acetic acid at the different
3. Results and discussion 3.1. Spatial separation of sequential steps of the denitrification process The spatial separation observed throughout the entire period of operation of the bioreactor is well represented by the average data. Nearly 80% of the NO 3 content of the influent had already been reduced in the layer of the first level beads, and the NO 3 concentration was below the detectable level at level 3 of the reactor on 95% of the sampling days. The reduction of the NO 3 was followed by the accumulation of NO , which was separated in 2 time and space because of the constant flow. The aver1 age NO 2 –N concentration at level 1 was 13 mg l , and the concentration progressively decreased up to level 4 of the reactor.
Fig. 1. The average ethanol concentration in the case of ethanol– C:NO 3 –N ratio of 3:1 of the samples at different levels of the reactor, and (a) the amounts of ethanol reacted at each level of the bioreactor, compared to the influent ethanol content. (b) The acetic acid contents of the samples and its utilization as a co-substrate at the different levels.
78
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
levels of the reactor are shown in Fig. 1(b). The acetic acid concentration was highest at level 1, which is correlated with the ethanol concentration. The DO content of the influent was likewise highest at level 1. The results indicated that the acetic acid produced from the ethanol was also utilized by the immobilized cells during the reduction processes. Less than 30% of the acetic acid produced at level 1 could be detected in the effluent (Fig. 1(b)). The relative declines in ethanol–C and NO 3 –N were determined at level 4 of the reactor. In the determination of the decline in C, the oxidation of ethanol to acetic acid was also taken into consideration. The experimentally determined optimum C:N ratio for the given bioreactor was 1:41 0:41, which is higher than the ratio (1.25) observed for the RBC system by Mohseni-Bandpi and Elliott (1998), but lower than the ratios measured by Fuchs et al. (1997) and McCleaf and Schroeder (1995) in membrane bioreactor systems (1.9–3 g g1 and 2.2 g g1 , respectively). During the second period of operation, when the influent ethanol content was decreased by half, acetic acid could not be detected and the concentration of residual ethanol in the effluent was 20–30 mg l1 . 3.3. Oxygen consumption of the immobilized cells The consumption of DO by the immobilized cells in the stirred medium (batch experiment) took 15 min (Fig. 2). In spite of the fact that the uptake is relatively fast in the batch system and the HRT of the influent was 10 times higher than the time which would have been sufficient for the process, the concentration of DO never fell below 1.7 mg l1 during continuous operation. The average level of residual DO in the effluent was 2:1 0:35 mg l1 . This result suggests that nearly 80% of the DO must have been utilized by the beads at level 1. After level 2, the DO concentration had to be further
Fig. 2. Decrease of the dissolved oxygen concentration of the stirred medium by the immobilized cells in the batch system. The composition of the medium was similar to that of the influent of the bioreactor. The experiment was repeated three times, and the average of the results is given.
decreased to zero or thereabouts. To verify this assumption, continuous reduction of the NO 2 formed was performed, which took place from level 1 of the reactor. 3.4. Denitrification activity of the bioreactor The aim was to attain a constantly high denitrification activity and a minimal NO 2 concentration in the effluent with a low retention time. Furthermore, the stability of the bioreactor at ethanol–C:NO 3 –N ratios of 3:1 and 1.5:1 was studied. The denitrification ability of the bioreactor at double the optimum C source (3:1 ratio) proved stable under relatively low HRT (2:45 0:12 h) and at a loading rate of 40–50 l influent day1 . The NO 3 –N content of the effluent was less (from 0 up to 5.21 mg l1 NO 3 –N), than the prescribed WHO limit (10 mg l1 NO 3 –N) in every case even though the average NO 3 –N concentration of the influent was 57 mg l1 (Fig. 3(a)). The average value of the pH (7:58 0:14) of the effluent never exceeded 8.5, which is the maximum acceptable value laid down by the international standard (EEC-Europen Economic Community, 1980). The average difference (DpH) between the pH levels of the influent and effluent solutions was 0:25 0:16. Fig. 3(b) illustrates the denitrification activity of the reactor and the effluent NO 2 concentration throughout the entire period of operation. The average denitrification activity was 0:88 0:06 kg NO 3– N m3 d1 . The NO 2 –N concentration of the effluent was below the prescribed limit of 1 mg l1 on half of the
Fig. 3. Changes in NO 3 concentration and the HRT values (a), and the denitrification rate and the NO 2 –N content of the bioreactor effluent in mg l1 (b) during operation at an ethanol–C:NO 3 –N ratio of 3:1. (–j–) NO 3 –N in, (––) NO3 –N out, (––) HRT, (––) NO2 –N out, (– –) denitrification rate.
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
sampling days, but values of 2 mg l1 were also detected occasionally. The number of living cells in the effluent (3 107 ) was higher than the maximum acceptable by the international standard. The excessive amount of C source in the effluent could be the reason for this. Besides P. butanovora, four other Gram-negative bacteria could be isolated from the samples. Although they were able to growth on ethanol, none of them proved to be capable of reducing NO 2 . Furthermore, their NO3 reduction abilities were negligible. The overcontamination of the bioreactor with these strains could be the cause of the relatively high optimum ethanol–C:NO 3 –N ratio detected (1:41 0:41). The result of the viable cell counting revealed that fewer than half of the detected cells were P. butanovora. When the influent ethanol concentration was halved (1.5:1 ratio), the number of living cells in the effluent decreased 100-fold (5 105 ). Furthermore, the relative amount of P. butanovora cells increased as compared with the contaminants. The denitrification rates of the bioreactor also changed (Fig. 4(b)). The activity at an ethanol–C:NO 3 –N ratio of 1.5:1 dropped to an average 3 1 of 0:54 0:05 from 0:88 0:06 kg NO d . The 3 –N m NO elimination activity after the first day was on the 3 average 95:2 2:1%, so the NO –N content of the ef3 fluent was still under the prescribed limit (from 0.7 up to 8.5 mg l1 NO 3 –N) (Fig. 4(a)). Although the NO2 –N concentration fluctuated more (from 0.37 up to 7.29 mg l1 NO 2 –N), similarly as in the previous period of
79
operation, with an average concentration of 1:1 0:62 1 (except on the day immediately after the mg NO 2 –N l change), its amount was always around the prescribed level throughout the experiment (Fig. 4(b)). The average of the pH of the effluent was 7:35 0:21. Dahab and Lee (1988) and Mohseni-Bandpi and Elliott (1999) reported that a nitrate removal efficiency of nearly 100% was achieved with HRTs of 9 and 8.8 h, respectively, using a bench-scale anoxic filter and the RBC system. With our bioreactor, the nitrate removal efficiency was 100 and 95.2% at ethanol–C:NO 3 –N ratios of 3:1 and 1.5:1, with HRTs of 2.45 and 3 h. In both cases, a low nitrite efflux was attained. The average difference between the influent and effluent pH levels was very low, in contrast with some autotrophic denitrifying processes (Chang et al., 1999; Kiss et al., 2000), where its value attained even 10. As concerns the amount of accumulated biomass in the present system (at the optimum C:N ratio) and the autotrophic electrobiochemical system reported by Kiss et al. (2000), no difference was detected. Although contamination of the bioreactor took place, which occurs naturally if a nonsterilized influent is used, this did not markedly influence the activity. The decrease in the denitrification rate could be explained by the slightly lower flow rate in the second stage of the experiment. The composition of the beads was stable: no destruction was detected, leading to a concentrated denitrifying biomass in the bioreactor, despite the high flow rate and the appearance of other strains. The composition of a biofilm in a denitrifying system is constantly changing (Szekeres et al., 2002), and the possibility arises that a bacterium unable to denitrify may overgrow the others in a heterotrophic system, causing a loss of activity. This effect can be avoided by the usage of gel-immobilized cells, which is the other advantage of the presented system. On the other hand, the amount and the composition of the biomass in the beads are unaffected by the flow rate, and therefore shear stress is negligible. This means that this technology allows the achievement of rapid and efficient denitrification at relatively low HRTs. Acknowledgements The authors wish to thank Maria T oth and Anik o Gargyan for their excellent technical assistance. This research was supported by Ph.D. fellowship grants from the Zoltan Bay Foundation for Applied Research for Peter Kesser} u, Istvan Kiss and Zoltan Bihari, and by a PHARE grant (HU9606-02-01-624).
Fig. 4. Changes in NO 3 concentration and the HRT values (a), and the denitrification rate and the NO 2 –N content of the bioreactor effluent in mg l1 (b) during operation at an ethanol–C:NO 3 –N ratio of 1.5:1. (–j–) NO 3 –N in, (––) NO3 –N out, (––) HRT, (––) NO2 –N out, (– –) denitrification rate.
References Æsøy, A., Ødegaard, H., Bach, K., Pujol, R., Hamon, M., 1998. Denitrification in a packed bed biofilm reactor (BIOFOR)-experiments with different carbon sources. Water Res. 32, 1463–1470.
80
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
Alves, C.F., Melo, L.F., Vieira, M.J., 2002. Influence of medium composition on the characteristics of a denitrifying biofilm formed by Alcaligenes denitrificans in a fluidised bed reactor. Process Biochem. 37, 837–845. Bouchard, D.C., Williams, M.K., Surampalli, R.Y., 1992. Nitrate contamination of groundwater: source and potential health effects. Journal AWWA 9, 85–90. Chang, C.C., Tseng, Sz.K., Huang, H.K., 1999. Hydrogenotrophic denitrification with immobilized Alcaligenes eutrophus for drinking water treatment. Bioresource Technol. 69, 53–58. Chen, K.C., Chen, S.J., Houng, J.Y., 1996. Improvement of gas permeability of denitrifying PVA gel beads. Enzyme Microb. Technol. 18, 502–506. Croll, B.T., Green, L.A., Hall, T., Whitford, C.J., Zabel, T.F., 1985. Biological fluidised bed denitrification for potable water. Paper presented at the International Conference on Advance in Water Engineering, Birmingham University, 15–19 July 1985. Dahab, M., Lee, Y.W., 1988. Nitrate removal from water supplies using biological denitrification. J. Water Pollut. Control Fed 60, 1670–1678. EEC-Europen Economic Community 1980. Council directive on the quality of water for human consumption. no. 80/778, Off J EEC 229, pp. 11–29. Fuchs, W., Schatzmayr, G., Braun, R., 1997. Nitrate removal from drinking water using a membrane-fixed biofilm reactor. Appl. Microbiol. Biotechnol. 48, 267–274. Hallin, S., Lindberg, C.F., Pell, M., Plaza, E., Carlsson, B., 1996. Microbial adaptation, process performance and a suggested control strategy in a pre-denitrifying system with ethanol dosage. Water Sci. Technol. 34, 91–99. Hasselblad, S., Hallin, S., 1996. Intermittent dosage of ethanol in a pre-denitrifying activated sludge process. Water Sci. Technol. 34, 387–389. Hungarian Standards for Drinking Water Analysis. Determination of nitrate and nitrite ions. 1983. Standards no. MSZ 448/12-82.
Kesser} u, P., Kiss, I., Bihari, Z., Polyak, B., 2002. Investigation of the denitrification activity of immobilized Pseudomonas butanovora cells in the presence of different organic substrates. Water Res. 36, 1565–1571. Kiss, I., Szekeres, Sz., Bejerano, T.T., Soares, M.I.M., 2000. Hydrogen-dependent denitrification: preliminary assessment of two bio-electrochemical systems. Water Sci. Technol. 42 (1–2), 373– 379. Kovacs, K., Polyak, B., 1991. Hydrogenase reactions and utilisation of hydrogen in biogas production and microbiological denitrification systems. Proceedings of the 4th IGT Symposium, Colorado Springs, pp. 1–16 (Chapter 5). Mansell, B.O., Schroeder, D.E., 1999. Biological denitrification in a continuous flow membrane reactor. Water Res. 33, 1845– 1850. McCleaf, P.R., Schroeder, E.D., 1995. Denitrification using a membrane-immobilized biofilm may offer a more economic nitrate removal method than conventional treatments. Am. Water Works Assoc. 95, 77–85. Mohseni-Bandpi, A., Elliott, D.J., 1998. Groundwater denitrification with alternative carbon sources. Water Sci. Technol. 38, 237– 243. Mohseni-Bandpi, A., Elliott, D.J., 1999. Denitrification of groundwater using acetic acid as a carbon source. Water Sci. Technol. 40, 53–59. Nagadomi, H., Takahasi, T., Sasaki, K., Yang, H.C., 2000. Simultaneous removal of chemical oxygen demand and nitrate in aerobic treatment of sewage wastewater using an immobilized photosynthetic bacterium of porous ceramic plates. World J. Microb. Biot. 16, 57–62. Sakakibara, Y., Kuroda, M., 1993. Electric prompting and control of denitrification. Biotechnol. Bioeng. 42, 535–537. Szekeres, Sz., Kiss, I., Kalman, M., Soares, M.I.M., 2002. Microbial population in a hydrogen-dependent denitrification reactor. Water Res. 36, 4088–4094.
Biological denitrification in a continuous-flow pilot bioreactor containing immobilized Pseudomonas butanovora cells Peter Kesser} u *, Istvan Kiss, Zolt an Bihari, Bela Poly ak Bay Zolt an Foundation for Applied Research, Institute for Biotechnology, Derkovits fasor 2., P.O. Box 2337, Szeged H-6726, Hungary Received 13 May 2002; received in revised form 9 August 2002; accepted 14 August 2002
Abstract Pseudomonas butanovora, a novel denitrifying bacterium, was immobilized in composite beads and filled into a reactor system. The pilot bioreactor average denitrification activity was at ethanol–C:nitrate–N ratios of 3:1 and 1.5:1 0.88 and 0.54 kg NO 3– N m3 d1 , respectively. The denitrification was stable in spite of the relatively low hydraulic retention times of 2.47 and 3 h. The nitrate content of the influent was almost completely reduced at the first level of the bioreactor and the nitrite formed underwent reduction in the upper part of the reactor. The experimentally determined optimum ethanol–C:nitrate–N ratio was 1:41 0:41. In consequence of the aerobic conditions, the acetic acid produced by the oxygenation of ethanol was also detectable in the reactor effluent. The pH of the effluent (7.58) never exceeded the acceptable maximum (8.5). The nitrate removal efficiency of the cells was nearly 100% at both C:N ratios, and the nitrite content of the effluent was around the prescribed limit throughout the continuous operation. This continuous-flow pilot bioreactor containing immobilized P. butanovora cells proved an efficient denitrification system with a relatively low retention time. Ó 2002 Elsevier Science Ltd. All rights reserved. Keywords: Acetic acid; Cell immobilization; Denitrification; Ethanol; Pseudomonas butanovora
1. Introduction Nitrate (NO 3 ) contamination of groundwater has been documented worldwide (Mansell and Schroeder, 1999). The ingestion of high levels of NO 3 may have serious effects on human health in consequence of the reduction to nitrite (NO 2 ) and the formation of nitrosamines (Bouchard et al., 1992). This danger necessitates the removal of NO 3 from water reserves. Biological denitrification is an attractive treatment option, for the NO 3 is converted by the denitrifying bacteria to inert nitrogen gas and the waste product usually contains only biological solids. In recent years, very many different technologies have been developed in efforts to solve the problem of NO 3 in drinking, waste or groundwater, e.g. bio-electrochemical systems (Sakakibara and Kuroda, 1993; Kiss et al., 2000), membrane reactors (McCleaf and Schroeder, 1995; Fuchs et al., 1997; Mansell and Schroeder, 1999), fluidized bed reactors * Corresponding author. Address: Zoltan Bay Foundation for Applied Research, Institute for Biotechnology, P.O. Box 2337, Szeged 6701, Hungary. Tel.: +36-62-432-252; fax: +36-62-432-250. E-mail address: [email protected] (P. Kesser} u).
(Croll et al., 1985), a porous ceramic plate reactor (Nagadomi et al., 2000), a rotating of biological contactor (RBC) system (Mohseni-Bandpi and Elliott, 1998) or with the help of gel-immobilized bacteria (Chang et al., 1999; Chen et al., 1996). A number of factors must be considered in the choice of the bioreactor system and the carbon (C) source, such as cost, denitrification rate, kinetics, degree of utilization, handling and storage safety, stability, etc. RBC systems and those technologies where a biofilm-covered surface is responsible for the denitrification are commonly used in wastewater treatment. The application of membrane reactors and bacteria immobilized in a gel matrix is becoming popular in drinking water purification. One important aim during the selection of an appropriate bioreactor is to achieve minimal NO 3 and NO2 concentrations and the lowest possible C content of the effluent or to attain a high denitrification rate. Membrane reactors and a bio-electrochemical system fulfil the former requirements, though their effectiveness is relatively low. Other methods, such as the RBC system or fluidized bed reactors, have very high denitrification activities, but their effluents almost always need some post-treatment because of the biomass accumulation. During continuous
0960-8524/03/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 0 - 8 5 2 4 ( 0 2 ) 0 0 2 0 9 - 2
76
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
operation, the effectiveness and the stability of these reactors depends on the structure of the biofilm formed (Alves et al., 2002). The concentration of the substrates in the influent affects the growth and the composition of the biofilm. The growth and the activity of the bacteria increase if the C source is available in excess, but this leads to a higher cell accumulation as well. Nutrient limitation can result in activity loss and biofilm detachment. Destruction of the biofilm by the high velocity of the liquid phase (shear stress) is also frequent. All these effects can be avoided by the application of cells immobilized in a gel matrix (Chang et al., 1999; Chen et al., 1996; Kesser} u et al., 2002). The application of gel immobilization techniques provides a continuously high cell concentration in the bioreactor, independently of the influent flow rate. Although nutrient limitation also has an effect on gel-immobilized cells, it does not radically modify the biomass concentration and the recovery of the activity is usually faster. Consideration of the aspects listed above in connection with bioreactors containing gel-immobilized cells could lead to more balanced systems than technologies involving a surface biofilm. Methanol, ethanol and acetic acid are generally used as additional or individual C sources in these processes because they are easily available and give a high denitrifcation rate (Æsøy et al., 1998). The usage of ethanol is common not only in experimental pilot plants (Fuchs et al., 1997; Æsøy et al., 1998; Mohseni-Bandpi and Elliott, 1998), but also in full-scale technologies (Hallin et al., 1996; Hasselblad and Hallin, 1996). The present work was undertaken in order to investigate the stability and denitrification activity of a bioreactor containing gel-immobilized Pseudomonas butanovora cells. The bioreactor was driven with ethanol as a C source with a relatively low hydraulic retention time (HRT).
2. Methods 2.1. Culture conditions The Pseudomonas butanovora strain was isolated from the soil of highly contaminated land formerly used for military purposes. The strain was maintained on DSM 1 agar plates (5 g peptone, 3 g meat extract and 15 g agar in 1000 ml distilled water). The bacterium was enriched in shaken cultures in DSM 1 broth, pH ¼ 7 (200 rpm, at 37 °C). The bacterial biomass for cell immobilization was produced in the 80 l New Brunswick Mobile Pilot Plant fermentation system. 2.2. Cell immobilization The grown culture was collected by centrifugation (Sorvall RC 3B, 2600 g, 30 min, 10 °C). The pellet was
resuspended in composite gel solution (Kov acs and Polyak, 1991) at a final cell concentration of 5% (w/w). The composite solution was prepared by mixing 1.6% Na-alginate (Sigma) in 200 mg l1 Na2 -EDTA solution and 1.6% Gelrite (Sigma) in 200 mg l1 Na2 -EDTA solution in a volumetric ratio of 1:1. The two solutions were sterilized separately and mixed when hot. After the composite solution had cooled down, the collected bacterial cells were added. The mixed gel solution was added in drops to 1% CaCl2 solution. The average diameter of the resulting beads was 2 mm. The beads were stirred at 25 °C for 40 min and were then incubated at 4 °C for 12 h. 2.3. Bioreactor operation The total volume of the reactor up to the top level 4 was 4700 ml, with height 41 cm, and diameter 15 cm. The beads functioning as the active volume of the reactor were placed on four indented plates which were fixed at equal distance from each other, and the bottom of the reactor. The total volume of the beads loaded into the four levels of the reactor was 2900 ml. The NO 3 –N concentration of the influent solution was set at 50 mg l1 . A constant flow rate was applied, at which the average HRT of the influent referred to the total volume of the reactor was 2.5–3 h. During the first period of operation, the amount of influent ethanol used as C source was determined to attain an ethanol–C:NO 3 –N ratio in the influent solution of 3:1; this was later decreased to 1.5:1. A peristaltic pump was used to load the NO 3 and ethanol stock solutions (13% and 10%, respectively) in appropriate proportions into the main influent flow. The non-sterilized influent solution applied at the bottom of the reactor was supplemented with 0.05% CaCl2 to preserve the stability of the beads. The reactor was operated at 25 °C. Samples were taken from each level of the bioreactor every 24 h, and the NO 3 and NO2 concentrations of the samples were determined to study the spatial separation of the NO 3 and NO 2 reduction steps of the denitrification process. The ethanol concentration and the pH were also determined in each sample. The numbers of living cells in the samples were measured by viability counts on DSM 1 agar plates. The identification of P. butanovora was done by its colony morphology. 2.4. Calculation of denitrification rate 3 The eliminated kg NO 3 –N by the bioreactor m a day, were calculated as the following equation:
ðinput NO 3 –N concentration output NO 3 –N concentrationÞ flow rate=volume of beads
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
The beads served as the biologically active volume of the bioreactor.
2.5. Analysis NO 3 and NO2 analyses were performed in accordance with Hungarian Standard no. 448/12–82 (1983). The measurements of ethanol and acetic acid were carried out on a Gynkotek HPLC instrument, with a SARASEP CAR-H column (temperature 50 °C). The sample detection was carried out by a refractive index detector and a UV detector. The eluent was 0.01 N H2 SO4 and its flow rate was 0.9 ml min1 . Data were evaluated with CHROMELEON software.
2.6. Oxygen consumption of the cells During the measurement of the oxygen consumption, the ratio of the beads and the influent medium volume referred to the ratio set in the case of the 1 level of the 4.7-l bioreactor. The reactor was fed continuously with 1 influent containing 50 mg NO and ethanol in an 3 –N l ethanol–C:NO3 –N ratio of 3:1 for more than a week before the measurements. During the batch measurements, the medium was stirred to allow the diffusion of oxygen inside the beads. In the case of continuous operation, the oxygen electrode was placed 1 cm above the beads, but its position was changed horizontally. Data were collected 10 min after the electrode was placed into the medium in order to avoid the interfering effect of the N2 produced on the dissolved oxygen (DO) measurement. Fifty individual measurements were made and the average is given.
77
3.2. Utilization of the carbon source During the first operation, the ethanol concentration was set to attain a C:N ratio of 3:1 in the influent in order to provide the immobilized cells with ample reducing power to achieve complete denitrification (Kesser} u et al., 2002). A decrease in the ethanol concentration was measured in the samples taken from the different levels of the reactor. Fig. 1(a) presents the decline in ethanol concentration between the levels. The ethanol concentration decreased to the highest extent up to level 1, i.e. the part of the reactor where 80% of the influent NO 3 was reduced. At the other levels, the degree of decrease in the ethanol concentration was progressively less; this was definitely correlated with the lower efficiency of the reduction at these levels. The average difference between the ethanol concentrations of the influent and the effluent solutions was 42% (if the concentration in the influent was taken as 100%). From day 2 of operation, acetic acid could be detected by HPLC measurements at the different levels of the bioreactor. When the concentrations of ethanol and DO in the influent are high enough, ethanol can be oxidized to acetic acid (Mohseni-Bandpi and Elliott, 1998). The concentrations of acetic acid at the different
3. Results and discussion 3.1. Spatial separation of sequential steps of the denitrification process The spatial separation observed throughout the entire period of operation of the bioreactor is well represented by the average data. Nearly 80% of the NO 3 content of the influent had already been reduced in the layer of the first level beads, and the NO 3 concentration was below the detectable level at level 3 of the reactor on 95% of the sampling days. The reduction of the NO 3 was followed by the accumulation of NO , which was separated in 2 time and space because of the constant flow. The aver1 age NO 2 –N concentration at level 1 was 13 mg l , and the concentration progressively decreased up to level 4 of the reactor.
Fig. 1. The average ethanol concentration in the case of ethanol– C:NO 3 –N ratio of 3:1 of the samples at different levels of the reactor, and (a) the amounts of ethanol reacted at each level of the bioreactor, compared to the influent ethanol content. (b) The acetic acid contents of the samples and its utilization as a co-substrate at the different levels.
78
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
levels of the reactor are shown in Fig. 1(b). The acetic acid concentration was highest at level 1, which is correlated with the ethanol concentration. The DO content of the influent was likewise highest at level 1. The results indicated that the acetic acid produced from the ethanol was also utilized by the immobilized cells during the reduction processes. Less than 30% of the acetic acid produced at level 1 could be detected in the effluent (Fig. 1(b)). The relative declines in ethanol–C and NO 3 –N were determined at level 4 of the reactor. In the determination of the decline in C, the oxidation of ethanol to acetic acid was also taken into consideration. The experimentally determined optimum C:N ratio for the given bioreactor was 1:41 0:41, which is higher than the ratio (1.25) observed for the RBC system by Mohseni-Bandpi and Elliott (1998), but lower than the ratios measured by Fuchs et al. (1997) and McCleaf and Schroeder (1995) in membrane bioreactor systems (1.9–3 g g1 and 2.2 g g1 , respectively). During the second period of operation, when the influent ethanol content was decreased by half, acetic acid could not be detected and the concentration of residual ethanol in the effluent was 20–30 mg l1 . 3.3. Oxygen consumption of the immobilized cells The consumption of DO by the immobilized cells in the stirred medium (batch experiment) took 15 min (Fig. 2). In spite of the fact that the uptake is relatively fast in the batch system and the HRT of the influent was 10 times higher than the time which would have been sufficient for the process, the concentration of DO never fell below 1.7 mg l1 during continuous operation. The average level of residual DO in the effluent was 2:1 0:35 mg l1 . This result suggests that nearly 80% of the DO must have been utilized by the beads at level 1. After level 2, the DO concentration had to be further
Fig. 2. Decrease of the dissolved oxygen concentration of the stirred medium by the immobilized cells in the batch system. The composition of the medium was similar to that of the influent of the bioreactor. The experiment was repeated three times, and the average of the results is given.
decreased to zero or thereabouts. To verify this assumption, continuous reduction of the NO 2 formed was performed, which took place from level 1 of the reactor. 3.4. Denitrification activity of the bioreactor The aim was to attain a constantly high denitrification activity and a minimal NO 2 concentration in the effluent with a low retention time. Furthermore, the stability of the bioreactor at ethanol–C:NO 3 –N ratios of 3:1 and 1.5:1 was studied. The denitrification ability of the bioreactor at double the optimum C source (3:1 ratio) proved stable under relatively low HRT (2:45 0:12 h) and at a loading rate of 40–50 l influent day1 . The NO 3 –N content of the effluent was less (from 0 up to 5.21 mg l1 NO 3 –N), than the prescribed WHO limit (10 mg l1 NO 3 –N) in every case even though the average NO 3 –N concentration of the influent was 57 mg l1 (Fig. 3(a)). The average value of the pH (7:58 0:14) of the effluent never exceeded 8.5, which is the maximum acceptable value laid down by the international standard (EEC-Europen Economic Community, 1980). The average difference (DpH) between the pH levels of the influent and effluent solutions was 0:25 0:16. Fig. 3(b) illustrates the denitrification activity of the reactor and the effluent NO 2 concentration throughout the entire period of operation. The average denitrification activity was 0:88 0:06 kg NO 3– N m3 d1 . The NO 2 –N concentration of the effluent was below the prescribed limit of 1 mg l1 on half of the
Fig. 3. Changes in NO 3 concentration and the HRT values (a), and the denitrification rate and the NO 2 –N content of the bioreactor effluent in mg l1 (b) during operation at an ethanol–C:NO 3 –N ratio of 3:1. (–j–) NO 3 –N in, (––) NO3 –N out, (––) HRT, (––) NO2 –N out, (– –) denitrification rate.
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
sampling days, but values of 2 mg l1 were also detected occasionally. The number of living cells in the effluent (3 107 ) was higher than the maximum acceptable by the international standard. The excessive amount of C source in the effluent could be the reason for this. Besides P. butanovora, four other Gram-negative bacteria could be isolated from the samples. Although they were able to growth on ethanol, none of them proved to be capable of reducing NO 2 . Furthermore, their NO3 reduction abilities were negligible. The overcontamination of the bioreactor with these strains could be the cause of the relatively high optimum ethanol–C:NO 3 –N ratio detected (1:41 0:41). The result of the viable cell counting revealed that fewer than half of the detected cells were P. butanovora. When the influent ethanol concentration was halved (1.5:1 ratio), the number of living cells in the effluent decreased 100-fold (5 105 ). Furthermore, the relative amount of P. butanovora cells increased as compared with the contaminants. The denitrification rates of the bioreactor also changed (Fig. 4(b)). The activity at an ethanol–C:NO 3 –N ratio of 1.5:1 dropped to an average 3 1 of 0:54 0:05 from 0:88 0:06 kg NO d . The 3 –N m NO elimination activity after the first day was on the 3 average 95:2 2:1%, so the NO –N content of the ef3 fluent was still under the prescribed limit (from 0.7 up to 8.5 mg l1 NO 3 –N) (Fig. 4(a)). Although the NO2 –N concentration fluctuated more (from 0.37 up to 7.29 mg l1 NO 2 –N), similarly as in the previous period of
79
operation, with an average concentration of 1:1 0:62 1 (except on the day immediately after the mg NO 2 –N l change), its amount was always around the prescribed level throughout the experiment (Fig. 4(b)). The average of the pH of the effluent was 7:35 0:21. Dahab and Lee (1988) and Mohseni-Bandpi and Elliott (1999) reported that a nitrate removal efficiency of nearly 100% was achieved with HRTs of 9 and 8.8 h, respectively, using a bench-scale anoxic filter and the RBC system. With our bioreactor, the nitrate removal efficiency was 100 and 95.2% at ethanol–C:NO 3 –N ratios of 3:1 and 1.5:1, with HRTs of 2.45 and 3 h. In both cases, a low nitrite efflux was attained. The average difference between the influent and effluent pH levels was very low, in contrast with some autotrophic denitrifying processes (Chang et al., 1999; Kiss et al., 2000), where its value attained even 10. As concerns the amount of accumulated biomass in the present system (at the optimum C:N ratio) and the autotrophic electrobiochemical system reported by Kiss et al. (2000), no difference was detected. Although contamination of the bioreactor took place, which occurs naturally if a nonsterilized influent is used, this did not markedly influence the activity. The decrease in the denitrification rate could be explained by the slightly lower flow rate in the second stage of the experiment. The composition of the beads was stable: no destruction was detected, leading to a concentrated denitrifying biomass in the bioreactor, despite the high flow rate and the appearance of other strains. The composition of a biofilm in a denitrifying system is constantly changing (Szekeres et al., 2002), and the possibility arises that a bacterium unable to denitrify may overgrow the others in a heterotrophic system, causing a loss of activity. This effect can be avoided by the usage of gel-immobilized cells, which is the other advantage of the presented system. On the other hand, the amount and the composition of the biomass in the beads are unaffected by the flow rate, and therefore shear stress is negligible. This means that this technology allows the achievement of rapid and efficient denitrification at relatively low HRTs. Acknowledgements The authors wish to thank Maria T oth and Anik o Gargyan for their excellent technical assistance. This research was supported by Ph.D. fellowship grants from the Zoltan Bay Foundation for Applied Research for Peter Kesser} u, Istvan Kiss and Zoltan Bihari, and by a PHARE grant (HU9606-02-01-624).
Fig. 4. Changes in NO 3 concentration and the HRT values (a), and the denitrification rate and the NO 2 –N content of the bioreactor effluent in mg l1 (b) during operation at an ethanol–C:NO 3 –N ratio of 1.5:1. (–j–) NO 3 –N in, (––) NO3 –N out, (––) HRT, (––) NO2 –N out, (– –) denitrification rate.
References Æsøy, A., Ødegaard, H., Bach, K., Pujol, R., Hamon, M., 1998. Denitrification in a packed bed biofilm reactor (BIOFOR)-experiments with different carbon sources. Water Res. 32, 1463–1470.
80
P. Kesser}u et al. / Bioresource Technology 87 (2003) 75–80
Alves, C.F., Melo, L.F., Vieira, M.J., 2002. Influence of medium composition on the characteristics of a denitrifying biofilm formed by Alcaligenes denitrificans in a fluidised bed reactor. Process Biochem. 37, 837–845. Bouchard, D.C., Williams, M.K., Surampalli, R.Y., 1992. Nitrate contamination of groundwater: source and potential health effects. Journal AWWA 9, 85–90. Chang, C.C., Tseng, Sz.K., Huang, H.K., 1999. Hydrogenotrophic denitrification with immobilized Alcaligenes eutrophus for drinking water treatment. Bioresource Technol. 69, 53–58. Chen, K.C., Chen, S.J., Houng, J.Y., 1996. Improvement of gas permeability of denitrifying PVA gel beads. Enzyme Microb. Technol. 18, 502–506. Croll, B.T., Green, L.A., Hall, T., Whitford, C.J., Zabel, T.F., 1985. Biological fluidised bed denitrification for potable water. Paper presented at the International Conference on Advance in Water Engineering, Birmingham University, 15–19 July 1985. Dahab, M., Lee, Y.W., 1988. Nitrate removal from water supplies using biological denitrification. J. Water Pollut. Control Fed 60, 1670–1678. EEC-Europen Economic Community 1980. Council directive on the quality of water for human consumption. no. 80/778, Off J EEC 229, pp. 11–29. Fuchs, W., Schatzmayr, G., Braun, R., 1997. Nitrate removal from drinking water using a membrane-fixed biofilm reactor. Appl. Microbiol. Biotechnol. 48, 267–274. Hallin, S., Lindberg, C.F., Pell, M., Plaza, E., Carlsson, B., 1996. Microbial adaptation, process performance and a suggested control strategy in a pre-denitrifying system with ethanol dosage. Water Sci. Technol. 34, 91–99. Hasselblad, S., Hallin, S., 1996. Intermittent dosage of ethanol in a pre-denitrifying activated sludge process. Water Sci. Technol. 34, 387–389. Hungarian Standards for Drinking Water Analysis. Determination of nitrate and nitrite ions. 1983. Standards no. MSZ 448/12-82.
Kesser} u, P., Kiss, I., Bihari, Z., Polyak, B., 2002. Investigation of the denitrification activity of immobilized Pseudomonas butanovora cells in the presence of different organic substrates. Water Res. 36, 1565–1571. Kiss, I., Szekeres, Sz., Bejerano, T.T., Soares, M.I.M., 2000. Hydrogen-dependent denitrification: preliminary assessment of two bio-electrochemical systems. Water Sci. Technol. 42 (1–2), 373– 379. Kovacs, K., Polyak, B., 1991. Hydrogenase reactions and utilisation of hydrogen in biogas production and microbiological denitrification systems. Proceedings of the 4th IGT Symposium, Colorado Springs, pp. 1–16 (Chapter 5). Mansell, B.O., Schroeder, D.E., 1999. Biological denitrification in a continuous flow membrane reactor. Water Res. 33, 1845– 1850. McCleaf, P.R., Schroeder, E.D., 1995. Denitrification using a membrane-immobilized biofilm may offer a more economic nitrate removal method than conventional treatments. Am. Water Works Assoc. 95, 77–85. Mohseni-Bandpi, A., Elliott, D.J., 1998. Groundwater denitrification with alternative carbon sources. Water Sci. Technol. 38, 237– 243. Mohseni-Bandpi, A., Elliott, D.J., 1999. Denitrification of groundwater using acetic acid as a carbon source. Water Sci. Technol. 40, 53–59. Nagadomi, H., Takahasi, T., Sasaki, K., Yang, H.C., 2000. Simultaneous removal of chemical oxygen demand and nitrate in aerobic treatment of sewage wastewater using an immobilized photosynthetic bacterium of porous ceramic plates. World J. Microb. Biot. 16, 57–62. Sakakibara, Y., Kuroda, M., 1993. Electric prompting and control of denitrification. Biotechnol. Bioeng. 42, 535–537. Szekeres, Sz., Kiss, I., Kalman, M., Soares, M.I.M., 2002. Microbial population in a hydrogen-dependent denitrification reactor. Water Res. 36, 4088–4094.