Environ. Sci. Technol. 2004, 38, 3449-3456
MtBE Biodegradation in a Gravity Flow, High-Biomass Retaining Bioreactor MAHER M. ZEIN,† M A K R A M T . S U I D A N , * ,† A N D ALBERT D. VENOSA‡ Department of Civil and Environmental Engineering, University of Cincinnati, Cincinnati, Ohio 45221, and U.S. Environmental Protection Agency National Risk Management Research Laboratory, Cincinnati, Ohio 45268
The aerobic biodegradation of methyl tert-butyl ether (MtBE), a widely used fuel oxygenate, was investigated using a pilot-scale biomass-retaining bioreactor called a Biomass Concentrator Reactor (BCR). The reactor was operated for a year at a flow rate of 2500 L/d on Cincinnati dechlorinated tap water and an influent MtBE concentration of 5 mg/L. Treatment efficiency of MtBE in the reactor during stable operations exceeded 99.9%. The upper 95% confidence levels of effluent MtBE concentrations and its degradation byproduct tert-butyl alcohol (TBA) were 2.9 and 0.9 µg/L, respectively, during these stable conditions. In addition, the effluent was found to be of better quality than the influent tap water as reflected by dissolved organic carbon analysis. Microbial community DNA profiling was carried out using denaturing gradient gel electrophoresis (DGGE) of polymerase chain reaction amplified 16s rDNA. The BCR was found to be inhabited by a wide spectrum of bacterial species, most notably microorganisms related to the genera Hydrogenophaga, Methylobacterium, Sphingomonas, and Pseudomonas. These organisms were previously reported to be associated with MtBE degradation. With the contamination of groundwater by MtBE being a wideranging problem throughout the United States, it is essential to develop a technology capable of effectively remediating such aquifers in order to protect public health and the environment. The BCR’s simple operation and low maintenance requirements may render it an economically attractive approach to remediating groundwater contaminated with MtBE.
Introduction The Clean Air Act Amendments of 1990 mandate seasonal or year-round use of oxygenated compounds in gasoline in certain areas of the country that exceed the National Ambient Air Quality Standards (NAAQS) for carbon monoxide (CO) and ozone (O3). Methyl tert-butyl ether (MtBE) was first introduced in the United States in 1979 primarily as an octane booster to replace organo-lead compounds and reduce air pollution. Currently, MtBE is the most widely used gasoline oxygenate additive in the United States where about one-
third of the gasoline sold contains MtBE in concentrations between 11 and 15 vol % (1). However, the use of MtBE has created a significant and unacceptable risk to drinking water and groundwater resources through its release mainly from leaking underground gasoline storage tanks in states such as California, New Jersey, Rhode Island, Illinois, Alaska, Texas, New York, Colorado, and others (2, 3). A study conducted by the U.S. Geological Survey (USGS) found that MtBE was detected in 5% of the wells monitored between 1993 and 1998 in urban areas nationwide and was the second most common volatile organic compound found in that study (4). The U.S. Environmental Protection Agency (EPA) classified MtBE as a possible human carcinogen without establishing any drinking water standards due to limited human toxicity studies (5). Consequently, the remediation of aquifers contaminated with MtBE has become an active research area in the past few years. The physicochemical characteristics of MtBE render most conventional remediation technologies, such as chemical oxidation, air stripping, and adsorption onto activated carbon, inefficient or impractical in treating MtBE-contaminated aquifers (6, 7). However, biological treatment of MtBE-contaminated groundwater appears to be the most economical, energy efficient, and environmentally sound approach. Recent in-situ studies have revealed the ability of several bacterial and fungal cultures to aerobically biodegrade MtBE either as the sole carbon and energy source or cometabolically while growing on other organic substrates. Pure cultures of only two bacterial strains, Rubrivivax gelatinosus PM1 (8, 9) and Hydrogenophaga flava ENV735 (10, 11), were able to completely utilize MtBE as the sole carbon and energy source. In addition, a phylogenetically diverse group of other pure bacterial cultures such as Methylobacterium, Rhodococcus, Arthrobacter (12), and Mycobacterium (13) was reported able to partially degrade MtBE. MtBE has also been reported to be cometabolized in the presence of pentane (14), propane (15), and ethanol (16). Hardison et al. (17) isolated a filamentous fungus (Graphium sp.) capable of cometabolically degrading MtBE in the presence of n-butane. Extensive research on the aerobic biodegradation of MtBE was also conducted from 1998 to 2003 at the University of Cincinnati. MtBE removal in a porous pot reactor, fed with an influent MtBE concentration of 150 mg/L, exceeded 99.99% (18). MtBE degradation was also investigated in two pilot-scale reactors: a fluidized bed reactor (FBR) and a membrane bioreactor (MBR). Both were operated with lower influent MtBE concentrations but higher flow rates than the porous pot reactor. The MBR performed better than the FBR, each attaining MtBE removal efficiencies of 99.99% and 99.61%, respectively (19, 20). The next step was to design and operate a bioreactor that is able to treat substantially higher water volumes without sacrificing the high MtBE removal efficiency of the porous pot. This paper reports the first phase of an evaluation of performance of a pilot-scale Biomass Concentrator Reactor (BCR) for the aerobic biodegradation of MtBE. The goal of the research was to demonstrate that the reactor was able to achieve final concentrations of MtBE < 5 µg/L on a consistent basis with no more than trace amounts of residual metabolic intermediates.
Materials and Methods * Corresponding author phone: (513)556-3695; fax: (513)556-2599; e-mail:
[email protected]. † University of Cincinnati. ‡ U.S. Environmental Protection Agency National Risk Management Research Laboratory. 10.1021/es030652y CCC: $27.50 Published on Web 05/15/2004
2004 American Chemical Society
Design and Operation of the BCR. The BCR is a pilot-scale, 1-m3 (1 m × 1 m × 1 m) completely stirred tank reactor (CSTR) made of 304 Stainless Steel and enclosing a 0.48-cm thick filter grade porous polyethylene membrane (Atlas VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3449
FIGURE 2. BCR setup.
TABLE 1. Target Nutrient Concentrations in the BCR nutrient
FIGURE 1. Fin design of the BCR. Minerals & Chemicals, Mertztown, PA) with a mean flow pore size of 18-28 µm. A fin design, composed of 30 parallel membrane panels with a total surface area of 25 m2, is used to maximize the permeable barrier surface-to-treatmentvolume ratio (Figure 1). Water flux through the membrane relies completely on gravity with vertical liquid level switches (Madison Co., Branford, CT) controlling the water level difference between the two sides of the membrane. Effluent water was discharged from the reactor using a 1/4 HP sump pump (ITT Jabsco, Costa Mesa, CA). The bulk water flow into the BCR was Cincinnati tap water dechlorinated in an F400 Granular Activated Carbon (GAC) column (Calgon Carbon Corporation, Pittsburgh, PA). The reactor was supplemented with an acidified nutrient solution containing salts and vitamins for biological growth and a sodium carbonate buffer solution dissolved in deionized water in two separate 20-L glass reservoirs (Table 1). Each solution was fed to the reactor using a 2-rpm Masterflex pump (Cole Palmer, Chicago, IL) to provide a flow rate of 10 L/d (0.4% of the total flow). MtBE was injected into the reactor at a rate of 0.7 mL/h using a Model 11 high precision syringe infusion pump (Harvard Apparatus, Inc., South Natick, MA) with a 50 mL Teflon-fluorocarbon-resin luer-lock syringe (Hamilton Co., Reno, NV). A recycle line, equipped with a 1/15 HP pump (March MFG Inc., Glenview, IL), was utilized to pump mixed liquor across the reactor to provide a uniform concentration of MtBE, buffer, and nutrients that were injected in this line. Building air was pumped into the reactor at a flow rate of 7 scfm (standard cubic feet per minute) or 198 L/min to maintain aerobic conditions and prevent biomass flocs from settling. The reactor setup is illustrated schematically in Figure 2. The BCR was operated for 1 year at a flow rate of approximately 2500 L/d of dechlorinated tap water spiked with MtBE at an influent feed concentration of 5 mg/L. The hydraulic retention time (HRT) of the reactor was maintained at approximately 4 h with an activated sludge volume of 400 3450
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
concn (mg/L)
essential nutrient
concn (mg/L)
K2HPO4 (NH4)2SO4 FeCl3 CaCl2‚2H2O MgSO4‚7H2O
Macronutrients 0.76835 5.64048 0.48083 0.76798 2.37819
P NH4+ Fe+3 Ca2+ Mg2+
0.13664 0.76998 0.16555 0.20936 0.23451
CuSO4‚5H2O Na2MoO4‚2H2O MnSO4‚H2O ZnCl2 CoCl2‚6H2O
Micronutrients 0.00384 0.00512 0.00444 0.00786 0.01435
Cu2+ Mo Mn2+ Zn2+ Co2+
0.00098 0.00203 0.00144 0.00377 0.00356
4-aminobenzoic acid biotin cyanocabalamine folic acid dihydrate nicotinic acid pantothenic acid pyridoxine hydrochloride riboflavin thiamine hydrochloride thioctic acid
Vitamins 0.00051 0.00020 0.00001 0.00020 0.00051 0.00051 0.00102 0.00051 0.00051 0.00051
L. The pH of the mixed liquor was maintained between 7.4 and 7.7. The dissolved oxygen concentration was maintained greater than 3 mg/L ensuring aerobic conditions. The temperature inside the reactor fluctuated over the seasons between 10 and 25 °C (50-77 °F) with the lower being a typical groundwater temperature. The reactor was seeded with 1 L of mixed liquor from a membrane bioreactor acclimated to MtBE (19), 1 L of mixed liquor from a porous pot reactor acclimated to MtBE, 0.5 L of mixed liquor from another porous pot reactor acclimated to MtBE and benzene + toluene + ethyl-benzene + p-xylene (BTEX) (21), and 5 L of activated sludge from the Sycamore Creek Wastewater Treatment Plant in Cincinnati. Chemicals. MtBE (>99.9% HPLC Grade) was purchased from the Fisher Scientific Co., Pittsburgh, PA. Tert-butyl alcohol (TBA) (>99.5% HPLC Grade) was obtained from Aldrich Chemical Co., Milwaukee, WI. All other chemicals used in this study were of the highest purity commercially available. Analytical Methods. Bulk water flow, MtBE, buffer, nutrient flow rates, pH, temperature, and dissolved oxygen were monitored on a daily basis. Effluent pH was measured using an Orion Model 720A pH meter (Orion Research Co., Boston, MA). Dissolved oxygen in the reactor was monitored
FIGURE 3. Performance of the BCR as effluent MtBE and TBA concentrations compared to influent MtBE concentration.
with a Corning Checkmate II dissolved oxygen sensor (Corning, NY). Analysis of the gas-phase levels of MtBE and its biodegradation intermediates was done three times a week using a Hewlett-Packard 5890 Series II gas chromatograph (GC) (Hewlett-Packard, Palo Alto, CA) equipped with a flame ionization detector (FID) and a 60/80 Carbopack B/5% Carbowax 20 M glass column (Supelco, Bellefonte, PA). The same GC was also used to measure high concentrations (>0.3 mg/L) of aqueous MtBE and its byproducts in the recycle/ feed line. Low concentrations of aqueous MtBE and its intermediates were measured daily using a Hewlett-Packard 6890 Series II GC/FID equipped with PTA-5 column (30 m, 0.53 mm i.d., and 3 µm film thickness) (Supelco, Bellefonte, PA) and coupled with a heated Purge and Trap (11 min at 40 °C of purging followed by 4 min at 225 °C of desorbing) consisting of a Tekmar Dohrmann 3100 Sample concentrator and a Tekmar Dohrmann AquaTek 70 Liquid Autosampler (Tekmar Dohrmann, Cincinnati, OH). The initial oven temperature of the GC was 35 °C held for 6 min, followed by a ramp of 12 °C/minute to 190 °C held for another 6 min. Dissolved organic carbon (DOC) of both the dechlorinated influent and the effluent, reported as nonpurgeable organic carbon (NPOC), was measured three times a week using a Shimadzu TOC-5050 Analyzer (Shimadzu, Kyoto, Japan). Volatile suspended solids (VSS) were measured weekly by drying a Whatman 934-AH Glass Microfiber filter (Clifton, NJ) at 550 °C for 1 h, filtering the sludge sample, drying at 105 °C for 2.5 h, and finally taking the difference in filter mass after baking at 550 °C for 2 h (22). All analyses were performed in triplicate to examine the quality of the data and ensure its precision. Community Structure Analysis. DNA Extraction. DNA extraction and isolation were performed in duplicate for all samples using a FastDNA kit and a FastPrep sample homogenizer provided by QBiogene (Vista, CA) according to the manufacturer’s protocol for liquid bacterial cultures. Polymerase Chain Reaction (PCR) Amplification. PCR was utilized to amplify a 200 base-pair (bp) portion of the V3 region of the 16S rDNA. Primers 534R and 341F (with a GC clamp), provided by Stratagene, Inc. (La Jolla, CA), were used. PCR reactions were done with 1.25 units of Expand Hi Fidelity DNA polymerase (Roche, Indianapolis, IN), 1% v/v formamide (Fisher, Fair Lawn, NJ), 0.25 mM dNTP (Roche), and 10 pmol of each primer in 25 µL total volume. A Robocycler PCR block (Stratagene) was then used for the temperature cycling as follows: 93 °C/2-min, 35 cycles of 92 °C/1-min,
55 °C/1-min, 68 °C/45-s, and finally one cycle at 72 °C/ 2-min. Denaturing Gradient Gel Electrophoresis (DGGE). DGGE was done with a D-code 16/16 cm acrylamide gel system (BioRad, Hercules, CA) kept at a constant temperature of 60 °C in 6 L of 0.5X TAE buffer (20 mM tris-acetate and 0.5 mM EDTA at pH 8.0). Fifteen percent and 55% denaturants, with 100% denaturant defined as 7 M urea and 40% v/v formamide, were used to produce the gradients needed for separating different DNA fragments. The gels were run at 35 V for 20 h and then stained in 1X TAE buffer with 1X SYBER Gold nucleic acid gel stain (Molecular Probes, Eugene, OR). Images of the resulting bands were documented using a GelDoc 2000 and Quantity One software (BioRad). The central 1-mm2 portion of each band was excised using a razor blade and stored in 36 µL of DNA-free water at 20 °C. This water was later used as a template for PCR amplification. DNA Sequencing. The PCR products were first purified using a GeneClean Spin Kit (Qbiogene). Then, DNA sequence analysis was performed off-campus by Davis Sequencing (Davis, CA) via an ABI Prism 377 DNA Sequencer (PerkinElmer, Foster City, CA). Phylogenetic Analysis. DNA sequences were compared to the GenBank database using the Blast facility of the National Center for Biotechnology Information (NCBI) and to the Ribosomal Database Project (RDP) using the “Similarity Rank” tool (23). The “Check Chimera” tool, available at RDP, was used to screen the sequences for chimeras.
Results and Discussion Reactor Performance. Figure 3 summarizes performance of the BCR for the entire experimental period (340 days). Because the BCR was inoculated with microbial cultures enriched on MtBE, an acclimation period of only 2 weeks was required before MtBE in the effluent dropped to the µg/L level. Effluent MtBE and TBA concentrations fluctuated in the first few months of the study due to operational problems that usually accompany the start-up of a new technology. The reactor was reseeded with the same microbial cultures on day 36 of the study when the biomass concentration was observed to be deteriorating and effluent MtBE concentrations started to increase. A recycle line, which was used to receive injected MtBE, was also installed to promote better MtBE distribution in the liquid phase and to minimize localized high MtBE concentrations and stripping to the air. However, the high VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3451
FIGURE 4. Biomass growth as VSS and TSS concentrations.
FIGURE 5. MtBE concentrations in the gas phase. recycle flow rate resulted in biomass being pushed into the MtBE syringe, thereby hindering its injection into the reactor. This problem was solved by reversing the recycle flow on day 60, thereby causing MtBE to be delivered to the BCR on the suction side of the recycle pump. This modification created a Venturi effect that pulled MtBE out of the syringe at a much higher flow rate than the syringe pump was set to deliver. The recycle line again was modified on day 158 to inject MtBE in the less pressurized buffer line as illustrated in Figure 2. As the biomass grew thicker, bacterial flocs were observed to settle, creating pockets of solids buildup in the reactor. On day 124, airflow into the reactor was raised from 4 scfm to 7 scfm to prevent the settling of flocs. This solved the biomass suspension problem, since biomass growth, quantified in terms of volatile suspended solids (VSS) and total suspended solids (TSS) concentrations, improved afterward, as illustrated in Figure 4. Nevertheless, high biomass concentrations caused the recycle line to clog and contributed to the inconsistency of MtBE levels in the feed. This problem was resolved by bleeding air out of the lines to create constant flow in the recycle line that is needed for proper functioning of the recycle pump. The major setback the reactor experienced occurred on day 239 of the study when failure of the effluent pump caused flooding of the reactor resulting in the loss of more than half the active biomass. This incident interrupted a period of operation that started on day 167 when the reactor performance had steadily improved as reflected in the decreasing trend in the effluent aqueous- and gas-phase MtBE levels shown in Figures 3 and 5, respectively. During that period, VSS and TSS concentrations rose to as high as 1000 mg/L and 1500 mg/L, respectively (Figure 4). Such problems were later avoided by performing periodic preventive main3452
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
tenance and installing backup water level sensors. Average VSS and TSS concentrations during the study period were 526 ((217) mg/L and 897 ((371) mg/L, respectively. Even though the biomass levels were lower at steady state than at the peak at day 221, the BCR was able to effectively and consistently degrade MtBE to its lowest effluent levels from day 278 to the end when stable conditions with respect to VOC concentrations were in effect. The ratio of VSS to TSS reported above (∼0.6) was lower than conventional activated sludge treatment, which is closer to 0.8. The reason is that the BCR was operated at an extremely long sludge age of >100 days (the only time solids were wasted was when the reactor contents were sampled). This results in a greater than normal biomass decay, leading to a larger fraction of inert solids in the mixed liquor. This did not diminish reactor performance in regards to treatment of MtBE. The microbial culture in the BCR was characterized by a slow growth rate, which is typical of biological degradation of MTBE (9, 19, 27, 31). The cellular yield during the steady-state period of operation was approximately 0.08 g cells produced per g MTBE consumed. TBA was the only byproduct of MtBE biodegradation detected in the effluent. tert-butyl formate (TBF) and acetone were monitored in the effluent and were consistently below detection limits. Nevertheless, despite all the problems discussed above, average effluent MtBE and TBA concentrations were low throughout the study duration, averaging 28.53 µg/L and 0.97 µg/L, respectively. As the reactor reached stable operational conditions in the last two months of the study, MtBE and TBA in the effluent declined to their lowest levels (Figure 3). The average effluent MtBE concentration during that period was 0.82 ((0.66) µg/L, while TBA levels were always lower, averaging 0.68 ((0.28) µg/L. The upper 95% confidence level (UCL95) of effluent MtBE concentrations for the times when conditions were stable (days 278 to 339) was 0.96 µg/L, which is much less than the 5 µg/L limit sought in the study. If the relatively stable period of operation before the BCR flooded due to the malfunctioning pump (days 167 to 239) is included in this calculation, the UCL95 is 2.9 µg/L, still 40% lower than the stated goal. Despite its high hydrophilicity, a small portion of the MtBE volatilized due to the airflow needed to provide sufficient mixing conditions for maintaining the microbial flocs in suspension and to ensure a homogeneous distribution of MtBE, nutrients, and buffer throughout the reactor. Gasphase MtBE concentrations during this steady-state phase averaged 0.89 ((0.38) µg/L with MtBE stripped to the air representing approximately 2% of the MtBE fed to the reactor. Figure 5 illustrates the decreasing trend of MtBE in the gas phase with time. From the mass balance, biological and total removal efficiencies of MtBE by the BCR averaged 97.93% and 99.98%, respectively, for the stable operating portion of the experimental period. Although the BCR was partially seeded with microbial cultures previously enriched on MtBE in a membrane bioreactor, the BCR and MBR produced effluents with different MtBE and TBA concentrations. The MBR was able to degrade an influent MtBE of 5 mg/L to lower levels (0.32 ( 0.39 µg/L) than the BCR; however, in the MBR, TBA was always detected at higher concentrations (1.38 ( 1.13 µg/L) than MtBE (19). The differences between the BCR and the MBR can be traced, as discussed later, to the development of a different bacterial community in each reactor. Similarly, the porous pot reactor produced effluents with MtBE never exceeding 0.62 µg/L but with TBA concentrations fluctuating between 0.15 and 59.20 µg/L (18). In addition, the BCR was observed to produce an effluent of better quality than the influent dechlorinated tap water as indicated by dissolved organic carbon (DOC) measurements. Non-Purgeable Organic Carbon (NPOC) analysis was performed to determine the DOC concentrations of the
TABLE 2. Summary of the BCR’s MtBE Removal Effectiveness throughout the Study Duration versus the Steady-State Period steady-state period
overall study period
5.57 ((0.64) 0.82 ((0.66) 0.89 ((0.38) 0.68 ((0.28) 97.93 ((0.77) 99.98 ((0.03) 0.92 ((0.22) 0.71 ((0.14)
5.44 ((0.58) 29 ((130) 7.21 ((9.34) 0.97 ((0.98) 90.63 ((11.64) 99.36 ((2.82) 0.86 ((0.29) 0.76 ((0.24)
influent MtBE (mg/L) effluent MtBE (µg/L) gas-phase MtBE (µg/L) effluent TBA (µg/L) biological MtBE removal (%) total MtBE removal (%) influent NPOC (mg/L) effluent NPOC (mg/L)
FIGURE 6. Influent and effluent NPOC concentrations plotted with influent MtBE. influent and effluent. Effluent NPOC concentrations were consistently less than those of the influent (0.76 ( 0.24 mg/L vs 0.86 ( 0.29 mg/L) throughout the study period. Influent and effluent NPOC concentrations, plotted with influent MtBE, are shown in Figure 6. NPOC levels in the influent tap water were observed to fluctuate over the study duration depending on the effectiveness of water treatment and the state of the activated carbon adsorption system at the Cincinnati Waterworks. Table 2 summarizes the reactor’s performance parameters during stable operation as well as throughout the operating period. Note that the overall study period includes those times when upset conditions occurred as well as the initial acclimation period. Temperatures also fluctuated between 10 and 25 °C over the study period without any detrimental effect on reactor performance. Numerous investigators have examined the ability of pure bacterial cultures (9-14, 24) or mixed consortia (25-27) to mineralize MtBE as a sole carbon and energy source. Fewer reports have been published describing the biodegradation of MtBE in continuous flow bioreactors operating at steady state for months at a time as this study does. Pruden et al. (20) operated a fluidized bed reactor (FBR) at an influent MtBE concentration of 7.8 mg/L for 250 days. At steady-state conditions, effluent MtBE concentrations averaged approximately 20 µg/L (99.7% removal). Stringfellow and Oh (28) examined the treatability of groundwater contaminated with MtBE in two pilot-field-scale FBRs and observed erratic removal efficiencies of MtBE in both. In one FBR, they attained 96% removal of MtBE with effluent concentrations of about 400-500 µg/L. However, in the second FBR, effective biodegradation was never achieved. In the laboratory, they studied cometabolic biodegradation using isopentane as the primary carbon source, and they were able to achieve 98% removal by day 80, which corresponded to an effluent MtBE concentration of 200-1000 µg/L MtBE. We have never experienced the need to supplement MtBE with another carbon source in any of our high-biomass reactors regardless of the influent concentrations. Kharoune et al. (29) operated
an upflow fixed-bed reactor for a 1-year period under varying conditions, using a combination of MtBE, ETBE, and TAME as the feed oxygenates. They showed that at an HRT of 24 h, they were able to achieve 98% removal of MtBE, but if the HRT were decreased to 13 h, biodegradation declined significantly. In the BCR, the important variables affecting performance are sludge age and high biomass solids, not HRT. Chang et al. (30) studied the biodegradation of MtBE in a packed bed GAC column inoculated with the MtBEdegrading culture PM-1. Using an influent feed concentration of 290-460 µg/L, they were able to achieve effluent concentrations ranging from 12 to 94 µg/L. Based upon our results, it is likely that a fluidized bed reactor cannot achieve consistently low levels of MtBE (
