Glyphosate Degradation by Immobilized Bacteria: Field Studies with Industrial Wastewater Effluent

Transcription

1 APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Apr. 1992, p /92/ $02.00/0 Copyright ) 1992, American Society for Microbiology Vol. 58, No. 4 Glyphosate Degradation by Immobilized Bacteria: Field Studies with Industrial Wastewater Effluent LAURENCE E. HALLAS,1* WILLIAM J. ADAMS,2t AND MICHAEL A. HEITKAMP2 Monsanto Agricultural Company' and Environmental Sciences Center,2 800 N. Lindbergh Blvd., St. Louis, Missouri Received 21 August 1991/Accepted 14 January 1992 Immobilized bacteria have been shown in the laboratory to effectively remove glyphosate from wastewater effluent discharged from an activated sludge treatment system. Bacterial consortia in lab columns maintained a 99%o glyphosate-degrading activity (GDA) at a hydraulic residence time of <20 min. In this study, a pilot plant (capacity, 45 liters/min) was used for a field demonstration. Initially, activated sludge was enriched for microbes with GDA during a 3-week biocarrier activation period. Wastewater effluent was then spiked with glyphosate and NH4Cl and recycled through the pilot plant column during start-up. Microbes with GDA were enhanced by maintaining the ph at <8 and adding yeast extract (<10 mg/liter). Once the consortia were stabilized, the column capacity for glyphosate removal was determined in a 60-day continuous-flow study. Waste containing 50 mg of glyphosate per liter was pumped at increasing flow rates until a steady state was reached. A microbial GDA of >90%o was achieved at a 10-min hydraulic residence time (144 hydraulic turnovers per day). Additional studies showed that microbes with GDA were recoverable within (i) 5 days of an acid shock and (ii) 3 days after a 21-day dormancy (low-flow, low-maintenance) mode. These results suggest that full-scale use of immobilized bacteria can be a cost-effective and dependable technique for the biotreatment of industrial wastewater. N-Phosphonomethylglycine (glyphosate) is a widely used, broad-spectrum herbicide that is sold under several trade names by the Monsanto Co. (St. Louis, Mo.). It is synthesized at Monsanto manufacturing facilities in the United States. Waste streams from glyphosate production are biologically processed in secondary treatment (activated sludge) biosystems. These aerobic biosystems are capable of significantly reducing organic chemicals in these waste streams. Previous studies have shown that microbes in activated sludge possess a glyphosate-degrading activity (GDA) that transforms the herbicide to aminomethylphosphonic acid (AMPA), a metabolite which is readily biodegradable in the environment (1-3, 6, 7). In recent years, glyphosate chemical production rates have increased at the same time that more stringent water quality guidelines have been promulgated. This has required Monsanto waste treatment operators to establish tertiary treatment capabilities for specific wastewater components like glyphosate. Technology that uses immobilized bacteria can achieve high population levels (109 to 1010 cells per g) on inert supports, which results in high rates of chemical degradation. The technique of attaching bacteria offers several advantages over other biological and chemical oxidation techniques. For example, increased wastewater flow is possible with minimal washout of immobilized bacteria. Microbes can also penetrate the carrier, which protects them from system upsets. Finally, the technology appears cost effective when compared with chemical oxidations such as chlorination. In a lab feasibility study, Heitkamp et al. (4) demonstrated that microorganisms with GDA could be immobilized in a column, removing 50 mg of glyphosate per liter from wastewater with a residence time of 23 min. They also noted that * Corresponding author. t Present address: ABC Laboratories, Columbia, MO inorganic nitrogen enhanced the activity while ph changes, repeated fluidization, and extended oligotrophic conditions could adversely affect performance. The purpose of this field study was to confirm the feasibility of using immobilized cells for removing low levels of glyphosate from high-volume effluent discharges. In addition, the performance of microbial activity was monitored during surge loading of glyphosate as well as after long periods of dormancy in which the bacteria were exposed to minimal glyphosate concentrations. MATERIALS AND METHODS Pilot scale reactor. In early February 1989, a pilot plant was leased from John Manville Corp. (Denver, Colo.). The configuration for its use in this study is shown in Fig. 1. Clarified wastewater from a secondary (activated sludge) biotreatment system was then pumped into an 1,800-liter equalization (EQ) tank, where the ph was adjusted to 6.5 to 7.0 and the temperature was monitored. The immobilized cell column was configured as an upflow reactor, and a feed pump delivered wastewater into the bottom (capacity, 45 liters/min). The column size was 2,000 liters (in three sections), and the biocarrier used was Manville R-635 diatomaceous earth pellets. These pellets utilized 1,300 liters of the reactor volume. Separate lines sparged air and a glyphosateammonia mixture into the wastewater. The flow was then either recycled back to the EQ tank (during start-up) or discharged after a single pass through the column. Enriching for microbes with GDA. The existing biotreatment activated sludge systems were used as seed. Ten days prior to pilot plant start-up, clarifier sludge underflow (volatile suspended solids = 8,000 mg/liter) in two 19-liter containers was enriched for bacteria with GDA as previously described by Heitkamp et al. (4). The sludge was mixed with 500 mg of glyphosate per liter and 50 mg of NH4Cl per liter. Clarified effluent was used as a diluent. The sludge was

2 1216 HALLAS ET AL. APPL. ENVIRON. MICROBIOL. Pump FIG. 1. Schematic diagram of pilot plant (John Manville Corp., Denver, Colo.). SCFM, standard cubic feet per minute (1 ft3 = 2.83 x 10-2 m3). aerated, and HCl was used to adjust the ph to 7.5. Intermittent feeding of process wastewater provided an abundant source of biodegradable carbon and additional glyphosate. Once microbes with GDA were established, the sludge was transferred to a 200-liter drum containing 45 kg of Manville R-635 diatomaceous earth biocarrier. A center well was created in the middle of the drum by using a pipette washing tube (inside diameter, 10 cm) with a perforated bottom. The biocarrier, sludge, and clarified effluent surrounded the center well tube. By pumping liquid from the bottom of the center well to the top of the drum, the clarified effluent, spiked with glyphosate (500 mg/liter) and NH4Cl (50 mg/liter), was circulated through the biocarrier bed. HCI was periodically added to adjust the ph to 7.5. Once chemical analysis indicated that the glyphosate was degraded, the liquid was drained from the drum and fresh, clarified effluent, glyphosate, and NH4Cl were added. This cycle was repeated until the pilot plant was ready for start-up. Pilot plant start-up. Five to ten kilograms of new biocarrier was added to the bottom of the pilot plant to bring the bed level up to the sparge tubes (Fig. 1). The inoculated biocarrier from the 200-liter GDA-enriched batch culture was then added to the pilot plant. Sufficient biocarrier was added to bring the total bed height up to about 2 m (approximately 900 kg of biocarrier). In addition, clarifier underflow (19 liters; 1% volatile suspended solids) was added with each 45 kg of carrier. To promote microbial growth throughout the biocarrier during the start-up phase, the pilot plant column and EQ tank were filled with clarified wastewater. One kilogram of glyphosate (500 mg/liter) and 0.5 kg of aqueous ammonia (50 mg/liter) were added. The mixer was turned on, and the ph was adjusted to 6.5 with sulfuric acid. The column pump (Fig. 1) was then adjusted to a flow rate of 19 liters/min, and the system was recirculated. Glyphosate analysis of the recycled effluent was conducted every 4 h. Once glyphosate degradation was demonstrated in the initial batch, the EQ tank and the column were drained and fresh, clarified effluent was added to the system along with a respike of glyphosate and ammonia (ph adjusted to 6.5). The cycle was repeated every 24 h until rapid glyphosate removal rates were demonstrated. Yeast extract was added in batches 5 to 9 (25 to 50 mg/liter) to improve bacterial growth and performance on the immobilized cell column. Samples were collected every 4 h for glyphosate analysis as well as measurements of dissolved oxygen (DO), ph, and temperature. The mechanical performance of the bioreactor was monitored by measuring water and air flow rates and by visually inspecting the pump operation. Continuous-flow studies. Continuous-flow operation of the pilot plant was begun after 500 mg of glyphosate per liter was removed within 24 h of batch operation. The initial reactor flow rate was 4 liters/min. An attempt was made to keep influent glyphosate levels at 50 mg/liter, which would simulate worst-case upset conditions in the activated sludge system. The ammonia was maintained at 25 mg/liter, and the flow continued for 3 to 5 days to allow for growth of the microbes on the biocarrier bed. Laboratory and field studies had previously shown that an increase of about 1.5 ph units is characteristic for the biotreatment of glyphosate wastewater. Accordingly, pilot plant influent ph was maintained at 6.5 to 7.0 in the EQ tank to maintain an effluent ph of <8.5. Once adequate removal of glyphosate was demonstrated (effluent glyphosate, <5 mg/liter) at a flow rate of 4 liters/ min, the flow was increased to 19 liters/min. Yeast extract (25 to 50 mg/liter) was added every 24 h from days 22 to 29

3 VOL. 58, 1992 GLYPHOSATE DEGRADATION BY IMMOBILIZED BACTERIA 1217 to the EQ tank to provide an additional nutrient source for promoting bacterial growth. A fluidization of the biocarrier bed was conducted by using recycled effluent and a gasoperated pump (flow rate, 800 liters/min). Fluidization was conducted only long enough to mix the biocarrier and to scour the excess biomass off the bed. Glyphosate loading and dormancy studies. Once glyphosate removal was >95% at a flow rate of 19 liters/min, glyphosate loadings were increased by raising the flow rates through the biological reactor. The goal was to determine the maximum flow rate that could be used before significant and sustained breakthrough of undegraded glyphosate was observed. Accordingly, the flow rate was increased in increments of 5 to 10 liters/min, after which the glyphosate level was monitored for 2 to 4 days (or until complete removal of glyphosate was achieved). Most of the study was conducted with hydraulic residence times of between 8 and 20 min. Therefore, 70 to 170 column hydraulic turnovers were achieved over a 24-h period, which allowed for ample column stabilization. System fluidization, column recycling, and yeast extract amendments were used in various combinations to maintain enrichment for microbes with GDA when significant and sustained glyphosate breakthroughs were observed. This approach served to give the immobilized microbes the best chance to survive and maintain high GDA at the increased levels of chemical loadings. All amendments to the column feed were stopped once the flow rate study was completed. The column was then placed on a low-maintenance mode with only mechanical integrity monitored. After 21 days of dormancy (about 150 to 200 column turnovers), 10 mg of glyphosate per liter was added to the pilot plant feed and the concentration of glyphosate was increased in increments of 10 mg/liter for 4 days. Nitrogen additions and ph control were also initiated as described earlier. Once glyphosate levels stabilized, an additional glyphosate surge loading of 100 mg/liter was added and glyphosate removal was monitored for 5 days. Analytical support. Glyphosate removal was the primary criterion for determining the success of the pilot plant project. After start-up, glyphosate removal was measured twice daily in the influent (EQ tank) and effluent. A dedicated high-performance liquid chromatograph with a phosphorus-specific postcolumn detector enabled completion of analyses within 4 h of sample collection (8). The detection limit after sample dilution was 3 mg/liter. The need for quick sample turnaround did not allow for detection of AMPA (the major metabolite of glyphosate degradation). However, the lab studies confirmed the stoichiometric conversion of glyphosate to AMPA under these conditions (4). DO and ph from the column effluent were measured twice daily. Influent DO was maintained near saturation (8 to 10 mg/liter), and the effluent DO was maintained near 2 mg/liter. The oxygen reaeration rate was measured in the pilot system during the study. First, the DO probe was inserted into the biocarrier bed, the air was turned off, and the drop in DO due to microbial activity was measured. The reaeration rate was then measured by turning the air back on. Nitrogen (measured as ammonia) was monitored periodically during start-up on selected influent and effluent samples. During the remainder of the study, quantitative measurements were made by using an NH3 probe three times weekly on the influent and effluent. Samples for determining total suspended solids were taken periodically from select start-up batches of the influent and effluent. Throughout the remainder of the study, measurements of total suspended solids were made three times weekly on the influent and effluent. Ti Mm) FIG. 2. Glyphosate degradation in pilot plant batch mode. Filland-draw batches are labeled in sequential order. Total organic carbon was measured three times per week on the effluent and once per week on the influent for the pilot plant. Biological oxygen demand was measured once per week on the influent and effluent after the start-up phase was completed. RESULTS AND DISCUSSION Seed enrichment. Sludge batches were enriched for microbes with GDA by using the fill-and-draw technique described by Murthy et al. (6). The clarified effluent from the activated sludge systems of the production facility was the diluent, and the nitrogen was supplied as NH4Cl. Within 1 week, sludge enrichments exhibited the capability of degrading 500 to 1,500 mg of glyphosate per liter in 24 h (data not shown). After approximately 10 days, the sludges were combined and recirculated in a 200-liter drum along with the biocarrier. Analyses showed that glyphosate (500 mg/liter) was being cleared in 12 to 18 h during the last several batches, indicating that activity levels were established prior to column inoculation. Pilot plant start-up. The initial phase of the start-up was begun without automatic ph control. The column was initially recycled with clarified effluent amended with glyphosate and ammonia. In batch 1 (data not shown), 500 mg of glyphosate per liter was used and 30% of it remained at 72 h. For the remaining nine batches (Fig. 2), the glyphosate level was lowered to 100 to 300 mg/liter. The second and third batches cleared 80 and 185 mg of glyphosate per liter in 16 h, respectively. When the initial glyphosate level was increased to 250 mg/liter (batch 4), a decrease in GDA was noted. The ph of these batches rose above 8.0, which may have been harmful to microbes exhibiting GDA. Unfortunately, only limited ph control was possible. The addition of yeast extract (25 to 50 mg/liter per batch) dramatically improved performance; for example, in batch 9, 275 mg of glyphosate per liter was cleared in 12 h. The use of yeast extract here and in phase 2 suggested that critical nutrients that enhanced microbes with GDA were being provided. This observation has also been noted for pure laboratory cultures of glyphosate-degrading bacteria (5). Lab studies have also suggested that a yeast extract concentration of <5 mg/liter was sufficient to maintain microbial activity (data not shown). It is not known whether this

4 1218 HALLAS ET AL. APPL. ENVIRON. MICROBIOL. a 20' S9-39- I INFLUENT ~-.- -elm wki I R %- 1-1 o 0 * ) b I IOD 20' I69S 40 ISD. II XD EFFLUENT Time (days) FIG. 3. Glyphosate degradation in pilot plant continuous-flow mode. Arrow indicates hydraulic residence time in minutes. (a) Influent glyphosate levels (target concentration, 50 mg/liter); (b) effluent glyphosate levels (detection limit, 0 to 5 mg/liter) )S Tine (days) FIG. 4. Glyphosate degradation in pilot plant continuous-flow mode. Arrows indicates hydraulic residence time in minutes. (a) Influent glyphosate levels; (b) effluent glyphosate levels. phenomenon is due to yeast extract serving as a carbon source or to yeast extract serving as a supplier of trace elements critical to microorganisms exhibiting GDA. Continuous-flow studies. Continuous flow was begun with limited ph control and several operational problems (pump optimization and flow control). Figure 3a and b depict column performance over the 3 weeks following operation start-up. The detection limit for glyphosate was 3 to 5 mg/liter, so that removal of 90 to 95% glyphosate could be confirmed. The column flow rate was maintained at 19 liters/min (20-min hydraulic residence time [HRT]), which is about 70 column turnovers a day. Low glyphosate levels were maintained in the column for 3 days. A slow deterioration of glyphosate degradation was then seen during the first 7 days, with glyphosate removal approaching a low of 60%. On day 22, a feed batch of glyphosate and ammonia was spiked with yeast extract at 1 to 2 mg/liter. The consortia on the column responded well, with almost 1 week of 90 to 99% glyphosate removal levels. This turned out to be the most consistent mechanical and biological operation seen in the study. On days 33 and 34, the column was fluidized to remove excess biomass buildup. While increased turbidity was seen in the column overflow, no bed expansion or turnover was noted. Nevertheless, bacteria with 90 to 95% GDA were consistently maintained after fluidization. Loading studies. A series of flow rate increases were initiated over the next 30 days of the study. The purpose was to establish a maximum flow (minimum HRT) that would still maintain bacteria with GDA. As noted in Fig. 4a and b, there were three major step changes in flow resulting in HRTs of 15, 10, and 8 min. Some piping and valve changes were made to accommodate the higher flow rates. Yet, despite some intermittent problems with feed and column pumps, a >90% glyphosate removal was achievable down to 10-min HRTs (144 hydraulic turnovers per day). The bacteria on the column maintained an acceptable GDA for several hours at an 8-min HRT. However, the activity deteriorated after day 52. The activity loss was attributed to a blockage in the column flow. Subsequent increases in glyphosate levels above the 50-mg/liter goal were due to the continued input of glyphosate by the glyphosate-loading pump. In addition, the temperature of the system reached its highest level (29 C) during the study. By day 60, the column consortia were showing only 50% glyphosate removal levels. Although several factors may have led to the column inefficiencies, it was concluded that an 8-min HRT could not support microbes with GDA without further studies. However, a 10- min HRT would be adequate for design calculations. Several analyses of column influent and effluent were done throughout the study. The biological oxidation demand ranged between 1 and 6 mg/liter in both the EQ tank and the column overflow. This would be expected since the activated sludge treatment system that preceded the pilot plant was very efficient. Total organic carbon levels showed a small but discernible drop after column treatment, approximately 5 to 10 mg/liter (5% of the total loading). This may have been due to the degradation of glyphosate to AMPA, although not enough analyses were done to note a correlation. Total suspended solids were low (40 to 50 mg/liter) at the point of entry into the EQ tank. However, the pilot plant reduced the total suspended solids by about 40%, presumably as a result of nonspecific column adsorption. This suggests that an immobilized cell column would not contribute to a solids problem and would actually help reduce suspended solids. Levels of NH3-N were also monitored. During intervals when glyphosate degradation was present, NH3-N was removed from the system. However, when system upset conditions existed, the ammonia treatment efficiencies were sporadic. Nitrate levels were not monitored, so it is not known whether the NH3-N disappearance was due to assimilation by microbes or a nitrification process. Still, this is in line with previous studies since NH3-N flux is known to be strongly correlated with microbes with GDA (6). Surge studies. The final phase of the study was to demonstrate that the microbes on the column could recover their GDA after an extended period without exposure to glyphosate. First, microbes with GDA were restored by recycling three batches of outfall water, containing 250 to 500 mg of glyphosate per liter and 50 mg of ammonia per liter, through the column. The column was then placed in a low-maintenance, continuous-flow mode (5 to 10 liters/min). No chemical amendments (including ph control) were made to the waste stream that passed to the reactor. The purpose of this was to simulate conditions in a maintenance mode when the

5 VOL. 58, 1992 GLYPHOSATE DEGRADATION BY IMMOBILIZED BACTERIA 1219 I- D- 30- INFLUENT X) Time (days) FIG. 5. Glyphosate degradation after a 21-day column dormancy. immobilized cell column would not be needed. The goal was to determine how long it would take for the activity to recover after dormancy. The low-maintenance mode was held in place for 21 days. A surge study was then conducted for a subsequent 8-day period during which glyphosate levels were increased in the column feed from 0 to 50 mg/liter over 5 days (historical data from the manufacturing facility indicated that increasing glyphosate in this manner would be characteristic of a worst-case biosystem upset). Figure 5 shows glyphosate levels into and out of the column during the last 6 days of the low-maintenance mode and during the 8-day surge study. Glyphosate loading during the low-maintenance mode never exceeded 5 mg/liter. When the surge study began, both ammonia and ph control were reestablished and the column flow was started at 40 liters/min (10-min HRT). Two days was required to enrich for microbes with GDA, and an additional 2 days was required before column effluent levels of glyphosate dropped to <2 mg/liter (95% removal). These data suggest that only a few days would be required for initiating bacterial activity in a column that has bacteria not exhibiting GDA. Further laboratory studies are planned to establish the time interval for recovery of microbes with GDA after dormancy. However, unless a column treatment regime is developed to immediately restore the activity in a dormant, immobilized cell column, it is clear that some liquid storage capacity for untreated water will be required. Alternatively, glyphosate waste could be continuously supplied to the microbial column to maintain higher levels of microbes with GDA. Several conclusions can be made from the pilot plant study. First, microbes with GDA were capable of being immobilized on a Manville diatomaceous earth support and were maintained at a column hydraulic residence time of 10 min, treating 50 mg of glyphosate per liter. The effectiveness of treatment was dependent on the presence of a nitrogen source, a small amount of a nonglyphosate carbon source, and ph levels of between 6.0 and 8.0. It was also found that a 3-day recovery period for glyphosate removal was needed when the column was placed in a low-flow maintenance mode for 21 days. It appears that immobilized cells are a useful and cost-effective option for tertiary biotreatment of low levels of glyphosate in high-volume industrial wastewaters. ACKNOWLEDGMENTS Many people provided direction for this work, and we gratefully acknowledge them. We thank individuals from key institutions critical to the project, including the John Manville Corp. and the central waste treatment and analytical sections of the manufacturing facility. Harold Crouch helped in on-site coordination of the project. We also thank Ralph Portier (Louisiana State University, Baton Rouge, La.) for helpful discussions and the completion of some preliminary laboratory testing which generated interest for the full-scale investigation of immobilized cells. REFERENCES 1. Balthazor, T. M., and L. E. Hallas Glyphosate-degrading microorganisms from industrial activated sludge. Appl. Environ. Microbiol. 51: Cook, A. M., C. G. Daughton, and M. Alexander Phosphonate utilization by bacteria. J. Bacteriol. 133: Hallas, L. E., E. M. Hahn, and C. Korndorfer Characterization of microbial traits associated with glyphosate biodegradation in industrial activated sludge. J. Ind. Microbiol. 3: Heitkamp, M. A., W. J. Adams, and L. E. Hallas. Can. J. Microbiol., in press. 5. Kulpa, C. W. (University of Notre Dame) Personal communication. 6. Murthy, D. V. S., R. L. Irvine, and L. E. Hallas Principles of organism selection for the degradation of glyphosate in a sequencing batch reactor, p In J. M. Bell (ed.), 43rd Annual Purdue Industrial Waste Conference. Lewis Publishers, Chelsea, Mich. 7. Rueppel, M. L., B. B. Brightwell, J. Schafer, and J. T. Marvel Metabolism and degradation of glyphosate in soil and water. J. Agric. Food Chem. 25: U.S. Environmental Protection Agency Methods for nonconventional pesticide analyses of industrial and municipal wastewater: method 127-determination of glyphosate in wastewater, p U.S. Environmental Protection Agency publication no. 440/1-83/079-C. U.S. Environmental Protection Agency, Washington, D.C.