Biological Hydrogen Production Using a Membrane Bioreactor Sang-Eun Oh,1 Prabha Iyer,1,2 Mary Ann Bruns,2 Bruce E. Logan1 1
COE Environmental Institute, Penn State University, University Park, Pennsylvania 16802; telephone: 814-863-7908; fax: 814-863-7304; e-mail: blogan@ psu.edu 2 Department of Crop and Soil Sciences, Penn State University, University Park, Pennsylvania Received 20 August 2003; accepted 9 March 2004 Published online 9 June 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20127
Abstract: A cross-flow membrane was coupled to a chemostat to create an anaerobic membrane bioreactor (MBR) for biological hydrogen production. The reactor was fed glucose (10,000 mg/L) and inoculated with a soil inoculum heat-treated to kill non-spore-forming methanogens. Hydrogen gas was consistently produced at a concentration of 57 – 60% in the headspace under all conditions. When operated in chemostat mode (no flow through the membrane) at a hydraulic retention time (HRT) of 3.3 h, 90% of the glucose was removed, producing 2200 mg/L of cells and 500 mL/h of biogas. When operated in MBR mode, the solids retention time (SRT) was increased to SRT=12 h producing a solids concentration in the reactor of 5800 mg/ L. This SRT increased the overall glucose utilization (98%), the biogas production rate (640 mL/h), and the conversion efficiency of glucose-to-hydrogen from 22% (no MBR) to 25% (based on a maximum of 4 mol-H2/mol-glucose). When the SRT was increased from 5 h to 48 h, glucose utilization (99%) and biomass concentrations (8,800 F 600 mg/L) both increased. However, the biogas production decreased (310 F 40 mL/h) and the glucose-tohydrogen conversion efficiency decreased from 37 F 4% to 18 F 3%. Sustained permeate flows through the membrane were in the range of 57 to 60 L/m2 h for three different membrane pore sizes (0.3, 0.5, and 0.8 Am). Most (93.7% to 99.3%) of the membrane resistance was due to internal fouling and the reversible cake resistance, and not the membrane itself. Regular backpulsing was essential for maintaining permeate flux through the membrane. Analysis of DNA sequences using ribosomal intergenic spacer analysis indicated bacteria were most closely related to members of Clostridiaceae and Flexibacteraceae, including Clostridium acidisoli CAC237756 (97%), Linmingia china AF481148 (97%), and Cytophaga sp. MDA2507 AF238333 (99%). No PCR amplification of 16S rRNA genes was obtained when archaea-specific primers were used. B 2004 Wiley Periodicals, Inc. Keywords: membrane bioreactor; hydrogen production; methanogen inhibition; cross-flow membrane; Clostridiaceae; Flexibacteraceae
Correspondence to: Bruce E. Logan Contract grant sponsor: National Science Foundation; US Filter Contract grant number: BES-0124674
B 2004 Wiley Periodicals, Inc.
INTRODUCTION Hydrogen-based fuel cells provide an excellent opportunity to reduce global emissions of carbon dioxide if the hydrogen can be produced from renewable resources (Kloeppel and Rogerson, 1991; US DOE Website, 2003). Hydrogen gas can be produced directly through fermentative routes at high concentrations (approximately 60% H2; Fang and Liu, 2002; Lay, 2000; Logan et al., 2002; Van Ginkle et al., 2001) from renewable substrates such as pure sugars (e.g., glucose and sucrose) or even wastewaters (Lay et al., 1998; Ueno et al., 1996). The industrial fermentation of sugars has been well studied for the production of alcohols and solvents, such as butanol, ethanol, and acetone (Bahl et al., 1986; Biebl, 1999; Dabrock et al., 1992; Jones and Woods, 1986) but not for hydrogen production. The maximum yield of hydrogen from glucose by a known biochemical route used by bacteria is 4 moles of hydrogen when acetic acid is produced (Gottschalk, 1986). Other products result in less hydrogen production. For example, the production of butyric acid results in only 2 moles of hydrogen. Pure cultures have been shown to produce hydrogen from defined substrates, such as glucose, at efficiencies of 22% to 57% (Kataoka et al., 1997; Rachman et al., 1998; Tanisho et al., 1989) (based on 4 moles of hydrogen per mole of glucose). The production of hydrogen from wastewaters creates new challenges because the waste materials are not sterile and it would be too costly to sterilize them and maintain aseptic conditions. In addition, wastewaters usually are composed of a variety of substrates that can be most efficiently used by different species of bacteria. Thus, hydrogen production from complex organics in wastewaters will require that different bacteria grow under mixed culture conditions. Unfortunately, some of the bacteria present in the microbial inoculum or wastewater will consume hydrogen, lowering the overall efficiency of hydrogen production. Of particular concern are methanogens that can convert hydrogen to methane, a gas that has only 42% of the energy content of hydrogen (mass basis). Strategies to control the growth of methanogens include maintaining a low pH in the
reactor (in the range of 5 –6.5; Fang and Liu, 2002; Van Ginkel et al., 2001), using an inoculum that is heat-treated to kill off non-spore-forming methanogens, and using short hydraulic retention times (HRTs) (Liu and Fang, 2002). Maintaining a low pH is only partly effective in limiting methane production (Mosey, 1983), and using a heat-treated inoculum does not necessarily prohibit the subsequent growth of methanogens already present in the wastewater. Because methanogens grow slowly, short detention times (less than a few days; Mosey, 1983) can limit the growth of methanogens in continuous flow reactors. However, short detention times also reduce the efficiency of substrate utilization by the bacteria and therefore overall process efficiency. One method of increasing biomass concentrations in a bioreactor is the use of a membrane in a reactor that allows fluid, but not bacteria, to leave the reactor. Membrane bioreactors (MBRs) have recently emerged as effective methods for treating wastewater under aerobic and anaerobic conditions (Beaubien et al., 1995; Defrance and Jaffrin, 1999; Wen et al., 1999). The advantages of MBRs include: higher biomass concentrations in the reactor that produces higher organics removal rates; decreased reactor volume due to higher removal rates; smaller excess sludge production due to biomass decay in the reactor; and a highquality effluent due to complete removal of bacteria by the membrane (Cicek et al., 1998; Defrance and Jaffrin, 1999). The main limitations of the technology are the high capital costs for the membrane, membrane fouling, and high operating costs due to energy needed to push the water through the membrane (Al-Malack and Anderson, 1997; Lee et al., 2001). In water and wastewater treatment systems, membrane fouling produces a need for frequent membrane cleaning and replacement (Defrance and Jaffrin, 1999; Field et al., 1995). A variety of techniques have been used to reduce membrane fouling, such as the addition of coagulants, operation of the system below a critical flux, backwashing, backpulsing, and (in aerobic systems) jet aeration (Ma et al., 2000). To try to increase the efficiency of hydrogen production in bioreactors using complex wastewaters, we examined hydrogen production using a MBR to maintain high concentrations of hydrogen-producing bacteria and short HRTs. Through the control of solids retention time (SRT), by controlling the rate solids are wasted from the reactor, and pH control, we hypothesized that it would be possible to omit methanogen growth in the system. The use of an MBR has not previously been examined for the purpose of hydrogen gas production, and therefore it was not known how the membrane would affect hydrogen gas concentration, production rates, or the composition of the microbial community. We therefore examined the effect of two different HRTs and several different SRTs on hydrogen production. In membrane systems, filtration performance is crucial to the overall design of the process. Therefore, we conducted a series of experiments with the solids produced in the bioreactor to determine the factors that affected the
120
permeate flux. These factors included membrane crossflow velocity (CFV), transmembrane pressure (TMP), mixed liquor suspended solid (MLSS) concentration, and membrane pore size. MATERIALS AND METHODS MBR System The hydrogen-producing membrane bioreactor (Fig. 1) consisted of a 2-L reactor (Bioflo 110, New Brunswick Scientific Co., Edison, NJ) constantly mixed at 200 rpm, flow level controller, pH controller, three peristaltic pumps (feed, waste, and recirculation-bioreactor side), a centrifugal pump (recirculation-membrane side), a membrane housing (259-mm long), liquid flow meter, three pressure gauges, and a backpulse unit with a control cabinet. The feed tank was pressurized using nitrogen gas. The recycle loop around the membrane module was operated at a flow rate of 378 L/h producing a membrane cross-flow velocity of 2.8 m/s. The recycle flow rate from the fermentor to the high recycle loop was set at 51 mL/min. The working volume of the fermentor (1 L) was maintained by a level probe connected to the feed pump. A pH of 5.5 was maintained using the pH controller and strong base (2M KOH). To adjust the SRT, solids were wasted from the bioreactor intermittently using a timer and the peristaltic pump. A medium without glucose was simultaneously pumped into the reactor at the same rate to maintain a constant liquid level in the reactor. Three MembraloxR (US Filter Co., Deland, FL) alumina membranes of various pore sizes (0.2, 0.5, and 0.8 Am) were used for filtration experiments (Table I). Pressures were measured at the inlet, outlet, and permeate side of the membrane to determine the transmembrane pressure (TMP). A constant permeate flow was obtained by adjusting the peristaltic pump on the permeate side of the membrane reactor.
Figure 1. Schematic diagram of the membrane bioreactor for hydrogen production. 1 = anaerobic reactor; 2 = cross-flow membrane; 3 = influent purged with nitrogen; 4 = feed pump; 5 = recirculation pump; 6 = high recirculation pump; 7 = flow meter; 8 = manometer; 9 = backpulsing; 10 = level controller; 11 = gas monitor; 12 = pH controller; 13 = motor; 14 = timer; 15 = waste; 16 = nitrogen gas; 17 = medium w/o organics; 18 = effluent.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 1, JULY 5, 2004
Table I. Characteristics of the membrane used. Characteristics
RISA Analysis Value
2
Membrane surface (m ) Type of module Membrane material Nominal pore size (Am) Flow direction Temperature limit, pressure limit pH range Membrane dimension
0.0055 Tubular Ceramic 0.2, 0.5, 0.8 Inside-out 225jC, 115 psi 0 – 14 ID = 7 mm; OD = 10 mm; length = 250 mm
Intermittent backpulsing (every 10 to 30 s) was used to limit membrane fouling (Sondhi and Bhave, 2000) unless stated otherwise. Backpulsing was achieved using a piston that injected nitrogen gas into the permeate side of the membrane module for very short period of time (0.5 – 1 s). This backpressure caused a reversal in flow through the membrane pushing solids off the membrane surface.
Experimental Procedures The MBR was seeded with heat-treated soil as previously described (Logan et al., 2002) and initially operated in a batch mode under anaerobic conditions. After 2 days, when hydrogen gas was produced, the reactor was switched to chemostat mode and operated at an initial HRT of 10 h without flow to the membrane. The HRT was subsequently reduced to 5 h and the SRT set at the same value in the absence of flow through the membrane. After reaching steady-state conditions (based on hydrogen production), the reactor was sampled at least 6 times (once per day). Steady state was defined as a period for which the MLSS, biogas production rate, and gas composition were stable (F 5%). The SRT was then sequentially increased to 12, 24, and 48 h using the membrane reactor. The HRT was changed to 3.3 h with SRTs of either 3.3 or 12 h. Glucose (10 g COD L) was used as a substrate in a nutrient solution containing (per liter of tap water): 1.0 g of NH4Cl, 0.2 g of KH2PO4, 0.2 g of K2HPO4, 30 mg MgCl2 6H2O, 2.5 mg of FeCl3, 1.6 mg of NiSO4, 2.5 mg of CaCl2, 1.15 mg of ZnCl2, 1.0 mg of CoCl 6H2O, 0.5 mg of CuCl2 2H2O, and 1.5 mg of MnCl2 4H2O. Membranes were cleaned by rinsing with 1% (wt/wt) HNO3 for 2 h, 2% (wt/wt) NaOCl for 2 h, and then rinsing with clean water until the baseline permeate flux was restored. If the baseline flux was not recovered, further washing with nitric acid at pH 2 was performed. The membranes were stored in NaOCl (6%) to prevent bacterial formation. The membrane was regenerated every 3 – 20 d depending on the HRT and SRT. The membrane was chemically cleaned before each test at a different MLSS concentration to restore the initial filtration performance.
During reactor operation at HRT of 5h, biomass was analyzed by Ribosomal Intergenic Spacer Analysis (RISA) at SRTs of 5, 12, 24, and 48 h. Microbial DNA was extracted from cell pellets obtained by centrifuging 2-mL samples. Cells were lysed with a beadbeater treatment (2 min) at high speed (BioSpec, Bartlesville, OK), and DNA was purified with the Ultraclean Soil DNA Kit (Mobio Laboratories, Carlsbad, CA). DNA was amplified by PCR using the bacterial 16S rDNA primer set 926f (16S rRNA) and 115r (23S rRNA) (Lane, 1991) or with the archaeal primer set 967f (16S rRNA) (Delong, 1992) and 14r (23S rRNA (Lane, 1991). PCR buffer (Promega, Madison, WI), 2.5 mM MgCl2, 2.5 U of Taq (Promega), 30 pmol of each primer, 400 AM of each deoxynucleoside triphosphate and 10 ng of template were used per reaction in a final volume of 50 AL. Amplification was performed in a GeneAmp PCR system 9600 (Perkin Elmer, Norwalk, CT): 5 min at 94jC; followed by 30 cycles of 30 s at 94jC, 30 s at 54jC, and 1 min at 72jC; with 5-min final extension at 72jC. PCR products obtained from different bacterial populations were separated by size via electrophoresis in 6% polyacrylamide gels. Gels were stained with SYBR-Green (Applied Biosystems, Foster City, CA) and imaged under UV light (Epi Chemi II Darkroom; UVP Laboratory Products, Upland, CA). Gel bands were excised and DNA was eluted from the gel by the crush and soak procedure (Sambrook et al., 1989). Eluted DNA from selected bands was cloned, and two clones were sequenced per band. DNA sequences for portions of the inserts corresponding to the 3V-ends of 16S rRNA genes (Escherichia coli positions 926-1542) were analyzed with the NCBI BLAST program (Altschul et al., 1990).
Resistance Analysis To characterize the origin of the membrane fouling, membrane resistances were calculated based on the resistancein-series model from the permeate flux J (L m2 h), calculated as: J¼
DP ARt
ð1Þ
where P is the applied transmembrane pressure difference (kPa) and A the permeate viscosity (0.89 mPa s). The total resistance, Rt (m1), is the sum of three resistances, or R t ¼ Rm þ R r þ R f
ð2Þ
where Rm is the intrinsic membrane resistance, Rr is the reversible resistance due to the cake layer; and Rf the internal fouling resistance due to irreversible adsorption and pore plugging that can only be removed by chemical, and not physical, cleaning (Li et al., 2002). Rm was measured using a clean membrane and water. Rt was calculated using Eq. (1)
OH ET AL.: BIOLOGICAL HYDROGEN PRODUCTION USING A MEMBRANE BIOREACTOR
121
from the final flux and TMP at the end of the operation. The sum Rm + Rf was measured after removing the cake layer by circulating tap water for 1 h (without pressurization) through the membrane, with backpulsing under previously described conditions. Rr is then calculated as the only unknown in Eq. (2).
Analytics The biogas production was continuously monitored with a bubble meter calibrated according to the manufacturer’s instructions (Challenge Environmental Systems AER-200 respirometer, Fayetteville, AR). The composition of biogas in the fermentor (hydrogen, carbon dioxide, and methane) was measured using gas chromatography (Logan et al., 2002). Glucose was analyzed using the phenol-sulfuric acid method for reducing sugars (Dubois et al., 1956). Mixed liquor suspended solid (MLSS) and mixed liquor volatile suspended solid (MLVSS) were measured according to Standard Methods 2540 (APHA, 1995). Permeate flux was measured gravimetrically.
Figure 3. MLSS concentrations at different HRTs and SRTs.
The MBR system was operated at hydraulic retention times of 3.3 and 5 h and SRTs ranging from 3.3– 48 h. Under all conditions, methanogenesis was completely eliminated as no methane gas was ever detected in the biogas. The hydrogen composition was always constant (57 –60%) with the remainder mostly CO2 and with minimal concentrations of N2 (1 – 5%) resulting from the initial gas sparging of the reactor with pure N2 (Fig. 2). The initial biogas production rate at an HRT of 3.3 h was 510 F 40 mL/h when the HRT and SRT were equal, under
conditions when the MBR was operated as a control (no permeate, but pressure was maintained across the membrane) (Fig. 2). Increasing the SRT from 3.3 to 12 h increased the biogas production rate by 25% to 640 F 40 mL/h. This increase in biogas production was due to the increase in biomass concentration in the reactor from 2200 to 5800 mg MLSS/L produced by increasing the SRT from 3.3 to 12 h (Fig. 3). Increasing the SRT also increased the overall utilization of glucose in the reactor. The influent glucose concentration was reduced from 10,000 mg COD/L to 940 mg COD/L at a SRT of 3.3 h, to 190 mg COD/L at a SRT of 12 h (Fig. 4), resulting in glucose utilization increasing from 90 to 98%. Overall, the conversion efficiency of glucose to hydrogen increased from 22% to 25% (Fig. 2) as the SRT was increased from 3.3 to 12 h. The biogas production rate at an HRT of 5 h, with no MBR retention of the solids (SRT = 5 h), was 640 F 40 mL/h and was therefore larger than that at 3.3 h with no MBR operation (Fig. 2). The longer HRT produced greater glucose utilization, producing an effluent glucose
Figure 2. Conversion efficiency from glucose to hydrogen, biogas production, and hydrogen and carbon dioxide percentage. The conversion efficiency was calculated based on a maximum of 4 mol H2/mol glucose.
Figure 4. Effluent glucose concentrations at different HRTs and SRTs.
RESULTS Hydrogen Production as a Function of Solids and Hydraulic Retention Times
122
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 1, JULY 5, 2004
concentration of only 30 –110 mg/L (99% glucose conversion efficiency) (Fig. 4). The efficiency of glucose conversion to hydrogen also increased to 37 F 4%, the highest conversion efficiency obtained under any reactor operating conditions. Increasing the SRT from 5 to 48 h at the HRT of 5 h further increased overall glucose utilization in the reactor, but did not increase overall hydrogen production. Glucose utilization increased from 99.0 to 99.5% (Fig. 4). However, the biogas production rate decreased from 640 F 40 mL/h to 310 F 40 mL/h as the SRT was increased from 5 to 48 h (Fig. 2). Overall, the glucose to hydrogen conversion efficiency decreased from 37 F 4% to 18 F 3%. This decrease was not due to a loss in total biomass, as biomass was efficiently retained in the reactor, increasing from 2400 F 500 mg/L to 8800 F 600 mg/L as the SRT was increased from 5 to 48 h (Fig. 3). Thus, the decrease in the hydrogen production rate must have been due to changes in the physiology or composition of the microbial community in the reactor as the SRT was increased. Microbial Community Analysis RISA profiles at SRTs of 5 h, 12 h, 24 h, and 48 h contained between 7 and 12 bands (Fig. 5). As SRT increased, 3 – 4 additional DNA bands appeared in the RISA profiles, demonstrating a slight increase in the number of detectable populations. DNA sequences from prominent bands were most closely related to members of Clostridiaceae and Flexibacteraceae, including Clostridium acidisoli CAC23
Figure 5. Inverted image of SYBR green-stained polyacrylamide gels showing DNA bands in RISA profiles at 5, 12, 24, and 48-h SRT. Last lane shows DNA marker. Positions of bands appearing at longer SRT but not at 5-h SRT are indicated with asterisks.
7756 (97%), Linmingia china AF481148 (97%), and Cytophaga sp. MDA2507 AF238333 (99%). No PCR amplification of 16S rRNA genes was obtained when archaea specific primers were used. Physical Performance of the MBR System The permeate flux through the membrane declined very rapidly upon introduction of the wastewater, and reached a low value within about 5 min. At all three solids concentrations tested, the flux across a clean membrane rapidly declined from 110 – 250 L/m2h to a range of 38– 50 L/m2h. The largest flux (50 L/m2h) was obtained at the lowest solids concentration of 1700 mg/L (Fig. 6). Increasing the solids concentration by a factor of 2 or 4 (3400 and 6800 mg/L) decreased the flux by an additional 12 L/m2h or 24%. Changing the membrane pore size did not substantially alter the permeate flux. Steady-state permeate fluxes were similar at 57 –60 L/m2h for three different pore size membranes (0.2, 0.5, and 0.8 Am) after an initial stabilization period (c 5 min) at a fixed solids concentration of 2500 mg/L solids concentration (Fig. 7). The pressure across the membrane, or the transmembrane pressure (TMP), can have a significant effect on the permeate flux. In general, as the TMP increases the flux increases up until a critical TMP is reached. The velocity of the water moving across the membrane also affects the permeate flux and TMP as shear can reduce fouling. To investigate the effect of TMP and cross-flow velocity, the TMP was set at 14. kPa (2.0 psi) and increased at defined pressures (22, 44, 65, and 83 kPa) every 10 min at three different cross-flow velocities (0.8, 1.7, and 2.8 m/s). At the lowest TMP of 14 kPa, each increase in the cross-flow velocity produced a proportional increase in the steadystate flux (Fig. 8). However, increasing the TMP to 22 kPa or higher did not increase the flux at a fixed cross-flow velocity. Increasing the cross-flow velocity from 0.8 to 1.7 m/s increased the steady flux from 20 to 50 L/m2h, but
Figure 6. Variation of permeate flux with time at different MLSS concentrations (cross-flow velocity = 2.2 m/s, membrane pore size = 0.2 Am, TMP = 2 psi).
OH ET AL.: BIOLOGICAL HYDROGEN PRODUCTION USING A MEMBRANE BIOREACTOR
123
Figure 7. Effect of membrane pore size on permeate flux (cross-flow velocity = 2.8 m/s, MLSS = 2500 F 100 mg/L, TMP = 14 kPa).
a further increase in the cross-flow velocity to 2.8 m/s did not affect the permeate flux. It appears from the above analysis that increasing the TMP up to 22 kPa and the cross-flow velocity to 1.7 m/s will produce the largest achievable permeate fluxes for the solids contained in this hydrogen reactor. However, this comparison does not take into consideration the type of fouling that occurs in the reactor. To further compare the characteristics of the reactor fouling, we compared the permeate flux in a reactor operated for 12 h at a low TMP of 14 kPa, with that produced at a high TMP of 90 kPa operated for only 1 h. Under these conditions, the flux in the 90 kPa system (20 – 23 L/m2h) had declined so that it was only about half that in the 14 kPa system (40 – 45 L/m2h) (Fig. 9A). This suggests that when the membrane is operated at high TMP, the flux decline rate increased meaning that fouling rate increased. To determine the dominant fouling mechanism (intrinsic membrane, internal fouling, cake fouling), the individual resistances were measured for the different pore-size mem-
Figure 8. Variation of permeate flux with time under step increments of transmembrane pressure (TMP) at different crossflow velocity (0.8, 1.7, and 2.8 m/s) (MLSS = 5000 mg/L, membrane pore size = 0.5Am).
124
branes. As the membrane pore size increased (from 0.2, 0.5, to 0.8 Am), the overall resistance, Rm, decreased (0.12 1012 m 1, 0.096 1012 m1 to 0.031 1012 m1, respectively) (Fig. 9B). In all cases, the intrinsic membrane resistances were only a small portion of the total resistance (0.7 – 6.3%). The main resistances were due to the internal fouling and cake formation, although the predominant type of resistance varied as a function of experimental conditions. It appeared that the internal fouling resistance (Rf) was highest for the high TMP operation while reversible or cake resistance, Rr, was greatest for low pressure conditions or for the smaller pore-sized membranes (0.5 and 0.8 Am). Backpulsing was essential for maintaining higher permeate flow rates. In the absence of backpulsing, permeate flux rapidly declined from 180 to 60 L/m2h over a 20 min period (Fig. 10; TMP 21 kPa). Initiating regular backpulsing (duration 0.5 s, interval 20 s, and forward filtration TMP 14 kPa) restored the permeate flow rate to 110 L/m2h. Regular backpulsing delayed, but could not prevent, a drop in the permeate flux back to 60 L/m2h over the next 50 min. Turning off the backpulsing after a total time of 70 min further decreased the permeate flux from 60 to 50 L/ m2h, but restoring a regular backpulse cycle at 90 min
Figure 9. The comparison between after 1 h of operation at the TMP of 90 kPa and after 12 h of operation at the TMP of 14 kPa. (A) The flux at three different pore sizes (0.2, 0.5, and 0.8 Am), (B) The resistances in the cross-flow membrane at the TMP of 22 kPa (MLSS: 6,000 F 200 mg/L, cross-flow velocity = 2.8 m/s).
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 1, JULY 5, 2004
Figure 10. Variation of permeate flux with time under the condition of with and without backpulse (cross-flow velocity = 2.8 m/s, pore size = 0.8Am, MLSS = 2500 F 100 mg/L, TMP = 14 kPa, backpulse amplitude = 21 kPa, backpulse duration = 0.5 s, backpulse interval = 20 s).
restored the flux to its steady value of 60 L/m2h. Different backpulse durations (0.5 – 1 s) and intervals (10 –30 s) did not substantially alter these trends (data not shown). DISCUSSION At a short hydraulic retention time of 3.3 h, use of the MBR increased hydrogen production from 510 F 40 mL/h to 640 F 40 mL/h, or by 25%. This increase was due to a greater retention of solids (2200 vs. 5800 mg/L) and greater utilization of glucose fed to the reactor (90 vs. 98%). However, at the longer retention time of 5 h, use of the MBR to retain solids (SRTs from 12– 48 h) decreased overall hydrogen production compared to that achieved with no increased solids retention (HRT = SRT). This indicates that the use of the MBR will only be helpful in cases where very short detention times or reactor sizes are needed. Otherwise, running a larger reactor in chemostat mode, with the SRT = HRT, will produce larger gas production rates. The reasons for the decrease in gas production rates in the MBR with long cell detention times are not known, but it is likely that several factors contributed to this decrease. One reason could be a shift in the physiology of the bacteria with detention time. Alternatively, as this was not a study with a pure culture, changes in gas production could also be due to changes in bacterial species in the reactor. Lower hydrogen production was associated with a slight increase in the number of DNA bands in RISA profiles, indicating that additional bacterial populations became detectable as biomass was retained for longer periods. Two bands yielded DNA sequences most closely related to Cluster Ic Clostridium spp. (Stackebrandt and Hippe, 2001). Another sequence was most closely related to the hydrogen-producing bacterium, Linmingia china, which is also classified in the Clostridiaceae. Closely related to the latter sequence was clone HPB-G1-15 obtained by Fang
et al. (2002a) from a hydrogen-producing sludge. Many species of Clostridium are known to be hydrogen producers (Nandi and Sengupta, 1998) and Clostridium spp. are consistently detected in waste-treating reactors optimized for hydrogen production (Fang et al., 2002a, 2002b; Ueno et al., 2001). Only Clostridium cluster I was detected here, in contrast to a study by Fang et al. (2002a), where sequences from Clostridium clusters I and IV were observed in a continuous-flow hydrogen bioreactor inoculated with activated sludge. Hydrogen gas was produced consistently at a concentration of 57 – 60% for all SRT and HRTs examined here. This gas concentration reflects changes in pH, which affects carbon dioxide dissolution and stoichiometric production of hydrogen. The pH was fixed in these experiments at 5.5. The conversion efficiencies for producing hydrogen from glucose, however, changed from 20 to 38% depending on SRT and HRT. These hydrogen gas concentrations and conversion efficiencies are comparable to those obtained by others using using glucose in a continuous-flow reactor with mixed cultures. For example, Lin and Chang (1999) measured H2 concentrations in the biogas of 43 –53% while Liu and Fang (2002) measured hydrogen concentration of 61 –66% regardless of the HRT. For mixed culture reactors, conversion efficiencies of glucose to hydrogen have been reported of 20.2% (Mizuno et al., 2000), 27.5% (Nakamura et al., 1993), 34% (Brosseau and Zajic, 1982), and 38% (Lin and Chang, 1999). For pure cultures, conversion efficiencies similarly range from 25% for Enterobacter aerogenes (Rachman et al., 1998) to as much as 53% for Clostridium butyricum (Kataoka et al., 1997). Physical Performance of the MBR System Although the permeate fluxes measured here were low, our values are typical of membrane systems having high solids concentrations. In our reactor, the solids concentrations ranged from 1,700 to 6,800 mg/L with corresponding permeate fluxes ranging from 50 to 38 L/m2h. In a study by Beaubien et al. (1996), the stabilized permeate flux was 46 L/m2h at 2500 mg/L, and it decreased to 25 L/m2h at a solids concentration of 22,000 mg/L. Others have found that the membrane flux decreased rapidly between at solids concentrations of 20,000 and 80,000 mg/L of TSS with a slower decrease observed between 80,000 and 150,000 mg/L of TSS (Cicek et al., 1998). However, the response of the membrane flux to solids concentration likely depends on operating variables such as TMP, crossflow velocity, and bacterial composition. For example, Defrance and Jaffrin (1999) found that the permeate flux did not substantially decrease when biomass concentration was increased from 20,000 to 60,000 mg/L using a ceramic membrane. The permeate flux is controlled by three different resistances: the clean membrane, internal membrane fouling, and cake fouling. At high solids concentrations typical of
OH ET AL.: BIOLOGICAL HYDROGEN PRODUCTION USING A MEMBRANE BIOREACTOR
125
MBRs, the fouling resistances are always larger than the intrinsic membrane resistance. Internal membrane fouling is of particular concern for MBR operation because the membrane must be chemically cleaned. We found that the internal fouling was greatest for high TMP operation while reversible, or cake, resistance was greatest for low TMP or the larger pore-sized membranes (0.5 and 0.8 Am). Our results therefore suggest that an MBR of the type used here should be run at low TMPs to minimize internal membrane fouling. Backpulsing was also essential to help maintain a stable permeate flux and the flux was not sensitive to either backpulse duration (0.5 – 2 s) or interval (10 – 30 s). Several studies have similarly shown that backpulsing has a positive effect on permeate flux, although most of these studies involved reactors with much lower MLSS concentrations than those used here. Sondhi and Bhave (2000) have shown that for a 50 mg SS/L water that with regular backpulsing there was no decline in permeate flux under conditions of 137 kPa TMP and a backpulse amplitude of 172 kPa, and backpulse duration and intervals of 0.5 s and 30 s, respectively. Vigneswaran et al. (1996) similarly found backpulsing (1 s duration, 1 min frequency) improved permeate flow and treatment of filter backwash water from a water treatment plant, while Parnham and Davis (1996) reported a 10-fold increase in permeate flux with backpulsing for a feed containing 0.0025 g cell debris/ g suspension. Thus, backpulse systems appear to be essential to the operation of MBRs used to treat wastewaters having a high solids concentration. CONCLUSIONS It was demonstrated that an MBR can be used to increase hydrogen production by fermentative bacteria grown at short (3.3 h) detention times. At a longer HRT, the MBR will not be necessary, as biogas production will decrease with any increase in SRT. However, HRTs of several days should be avoided to minimize the potential for methane gas generation. If an MBR is used, it should be operated at a low TMP with regular backpulsing to minimize internal fouling of the membrane. Biogas composition was not affected by any reactor operating condition (HRT or SRT) and was always in the range of 57 to 60% of hydrogen. References Al-Malack HM, Anderson GK. 1997. Crossflow microfiltration with dynamic membranes. Wat Res 31(8):1969 – 1979. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local alignment search tool. J Mol Biol 215(3):403 – 410. APHA. 1995. Standard methods for the examination of water and wastewater, 18th ed. Washington DC: American Public Health Association. Bahl H, Gottwald M, Kuhn A, Rale V, Andersch W, Gottschalk G. 1986. Nutritional factors affecting the ratio of solvents produced by Clostridium acetobutyricum. Appl Environ Microbiol 52:169 – 172. Beaubien AM, Baty FJ, Francoeur E, Manem J. 1996. Design and op-
126
eration of anaerobic membrane bioreactors: development of a filtration testing strategy. J Membr Sci 109:173 – 184. Biebl H. 1999. Clostridium acetobutylicum. In: Robison R, Batt C, Patel P, editors. Encyclopedia of food microbiology. New York: Academic Press. p 445 – 451. Brosseau JD, Zajic JE. 1982. Hydrogen gas production with Citrobacter intermedius and Clostridium pasteurianum. J Chem Tech Biotechnol 32:496. Cicek N, Winnen H, Suidan MT, Wrenn BE, Urbain V, Manem J. 1998. Effectiveness of the membrane bioreactor in the biodegradation of high molecular weight compounds. Wat Res 32(5):1553 – 1563. Dabrock B, Bahl H, Gottschalk G. 1992. Parameters affecting solvent production by Clostridiium pasteurianum. Appl Environ Microbiol 58:1233 – 1239. Defrance L, Jaffrin MY. 1999. Comparison between filtrations at fixed transmembrane pressure and fixed permeate flux: application to a membrane bioreactor used for wastewater treatment. J Membr Sci 152:203 – 210. Delong EF. 1992. Archaea in coastal marine environments. Proc Natl Acad Sci 91:6721 – 6728. Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F. 1956. Colorimetric method for determination of sugars and related substances. Analyt Chem 28(3):350 – 356. Fang HP, Liu H. 2002. Effect of pH on hydrogen production from glucose by a mixed culture. Bioresource Technol 82:87 – 93. Fang HH, Liu H, Zhang T. 2002a. Characterization of a hydrogenproducing granular sludge. Biotechnol Bioeng 78(1):44 – 52. Fang HH, Zhang T, Liu H. 2002b. Microbial diversity of a mesophilic hydrogen-producing sludge. Appl Microbiol Biotechnol 58(1): 112 – 118. Field RW, Wu D, Howell JA, Gupta BB. 1995. Critical flux concept for microfiltration fouling. J Membr Sci 100:259 – 272. Gottschalk G. 1986. Bacterial metabolism, 2nd ed. New York: SpringerVerlag. Jones DT, Woods DR. 1986. Acetone-butanol fermentation revisited. Microbiological Reviews 50(4):484 – 524. Kataoka N, Miya A, Kiriyama K. 1997. Studies on hydrogen production by continuous culture system of hydrogen-producing anaerobic bacteria. Wat Sci Tech 36(6/7):41 – 47. Kloeppel J, Rogerson S. 1991. The hydrogen economy. Electronics World + Wireless World 97:668 – 671. Lane DJ. 1991. 16S/23S rRNA sequencing. In: Stackebrandt E, Goodfellow M, editors. Nucleic acid techniques in bacterial systematics. New York: Wiley. p 115 – 175. Lay JJ. 2000. Modeling and optimization of anaerobic digested sludge converting starch to hydrogen. Biotechnol Bioeng 68(3):269 – 278. Lay JJ, Lee YJ, Noike T. 1998. Feasibility of biological hydrogen production from organic fraction of municipal solid waste. Wat Res 33(11):2579 – 2586. Lee JM, Ahn WY, Lee CH. 2001. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Wat Res 35(10):2435 – 2445. Li J, Sanderson RD, Jacobs EP. 2002. Ultrasonic cleaning of nylon microfiltration membranes fouled by Kraft paper mill effluent. J Membr Sci 205:247 – 257. Lin CY, Chang RC. 1999. Hydrogen production during the anaerobic acidogenic conversion of glucose. J Chem Technol Biotechnol 74(6): 498 – 500. Liu H, Fang HP. 2002. Hydrogen production from wastewater by acidogenic granular sludge. Wat Sci Technol 47(1):153 – 158. Logan BE, Oh SE, Kim IS, Van Ginkel S. 2002. Biological hydrogen production measured in batch anaerobic respirometers. Environ Sci Technol 36(11):2530 – 2535. Ma H, Brwman CN, Davis RH. 2000. Membrane fouling reduction by backpulsing and surface modification. J Membr Sci 173:191 – 200. Mizuno O, Dinsdale R, Hawkes FR, Hawkes DL, Noike T. 2000. Enhancement of hydrogen production from glucose by nitrogen gas sparging. Bioresource Technol 73(1):59 – 65.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 1, JULY 5, 2004
Mosey FE. 1983. Mathematical modeling of the anaerobic digestion process: Regulatory mechanisms for the formation of short-chain volatile acids from glucose. Wat Sci Tech 15:209 – 232. Nakamura M, Kanbe H, Matsumoto J. 1993. Fundamental studies on hydrogen production in the acid forming phase and its bacteria in anaerobic treatment processes—The effects of solids retention time. Wat Sci Tech 28(7):81 – 88. Nandi R, Sengupta S. 1998. Microbial production of hydrogen: an overview. Crit Rev Microbiol 24(1):61 – 84. Parnham CS, Davis RH. 1996. Protein recovery from bacterial cell debris using crossflow microfiltration with backpulsing. J Membr Sci 118: 259 – 268. Rachman MA, Nakashimada Y, Kakizono T, Nishio N. 1998. Hydrogen production with high yield and high evolution rate by self-flocculated cells of Enterbacter aerogenes in a packed-bed reactor. Appl Microbiol Biotechnol 49:450 – 454. Sambrook J, Fritsch EF, Maniatis T. 1989. Molecular Cloning: A laboratory manual, 2nd ed. Plainview, NY: Cold Spring Harbor Laboratory Press. Sondhi R, Bhave R. 2000. Role of backpulsing in fouling minimization in crossflow filtration with ceramic membranes. J Membr Sci 186: 41 – 52.
Stackebrandt E, Hippe H. 2001. Taxonomy and systematics. In: Bahl H, Durre P, editors. Clostridia: Biotechnology and medical applications. Weinheim, Germany: Wiley-VCH. p 19 – 48. Tanisho S, Kamiya N, Wakao N. 1989. Hydrogen evolution of Enterobacter aerogenes depending on culture pH: Mechanism of hydrogen evolution from NADH by means of membrane-bound hydrogenase. Biochimica et Biophysica Acta 973:1 – 6. Ueno Y, Otsuka S, Morimoto M. 1996. Hydrogen production from industrial wastewater by anaerobic microflora in chemostat culture. J Ferment Bioeng 82(2):194 – 197. Ueno Y, Haruta S, Ishii M, Igarashi Y. 2001. Microbial community in anaerobic hydrogen-producing microflora enriched from sludge compost. Appl Microbiol Biotechnol 57(4):555 – 562. US DOE Website. 2003. http://www.eren.doe.gov/hydrogen/faqs. html#cost. Van Ginkel S, Sung S, Lay JJ. 2002. Biohydrogen production as a function of pH and substrate concentration. Environ Sci Technol 35:4726 – 4730. Vigneswaran S, Boonthanon S, Prasanthi H. 1996. Filter backwash recycling using crossflow microfiltration. Desalination 106:31 – 38. Wen C, Huang X, Qian Y. 1999. Domestic wastewater treatment using an anaerobic bioreactor coupled with membrane filtration. Process Biochemistry 35:335 – 340.
OH ET AL.: BIOLOGICAL HYDROGEN PRODUCTION USING A MEMBRANE BIOREACTOR
127
