WHEC 16 / 13-16 June 2006 – Lyon France
Fermentative hydrogen production from anaerobic bacteria using a membrane bioreactor a*
a
a
Mi-Sun Kim , You-Kwan Oh , Young-Su Yun , Dong-Yeol Lee
b
a
Bioenergy Research Center, Korea Institute of Energy Research, 71-2, Jang-Dong, Yusung-Ku, Daejeon 305-343, Korea, *
[email protected] b
Woori Tech, 203-1, Pyongchon-Dong, Daeduk-Ku, Daejeon 306-130, Korea
ABSTRACT: Continuous H2 production from glucose was studied at short hydraulic retention times (HRT) of 4.69 – 0.79 h using a membrane bioreactor (MBR) with a hollow-fiber filtration unit and mixed cells as inoculum. The o
reactor was inoculated with sewage sludge, which were heat-treated at 90 C for harvesting spore-forming, H2-producing bacteria, and fed with synthetic wastewater containing 1% (w/v) glucose. With decreasing HRT, volumetric H2 production rate increased but the H2 production yield to glucose decreased gradually. The H2 content in biogas was maintained at 50 – 70% (v/v) and no appreciable CH4 was detected during the operation. The maximal volumetric H2 production rate and H2 yield to glucose were 1714 mmol H2/L·d and 1.1 mol H2/mol glucose, respectively. These results indicate that the MBR should be considered as one of the most promising systems for fermentative H2 production.
KEYWORDS : membrane bioreactor, hydrogen production, high cell density, hydraulic retention time, hollowfiber membrane
INTRODUCTION: Microbial H2 production can be either photosynthetic or non-photosynthetic. Photosynthetic H2 production is carried out by either algae or photosynthetic bacteria. Non-photosynthetic or fermentative H2 production is performed by either facultative or obligate anaerobes. The fermentative H2 production is generally faster than photosynthetic H2 production and does not rely on the availability of light. However, the H2 conversion yield (mol H2/mol substrate) is lower than that observed in photosynthetic processes (Oh et al., 2004). The economic feasibility of fermentative H2 production mainly depends on the rate and yield of H2 production with various substrates. Most researchers have focused on developing proper biocatalysts and/or efficient production processes. Both pure (Oh et al., 2003a,b) and mixed cultures (Chen et al., 2001; Chen and Lin, 2001) have been investigated as biocatalysts. When microbial consortia are used, mixed substrates of various sugars can be utilized. However, the H2 production yield is generally low since H2-consuming
1/9
WHEC 16 / 13-16 June 2006 – Lyon France bacteria and many non-H2 producing bacteria, such as lactate- and propionate-producing bacteria, coexist in the reactor. Continuous processes for fermentative H2 production can be divided into suspended and immobilized systems. Suspended systems, represented by a conventional stirred tank reactor (CSTR), are simple and easy to operate. However, when operated at a high dilution rate, the system is unstable and washout of the cells is often experienced. The immobilized system uses a support matrix onto which biofilms are established. The immobilized system maintains a higher cell concentration and H2 production rate than a suspended system. However, it has been reported that the biogas accumulation and excessive gas hold-up is one of the most challenging problems in the immobilized systems (Kumar and Das, 2001). Membrane bioreactors (MBR) have been used for treating wastewater under aerobic and anaerobic conditions for many years. The use of the membrane in a reactor allows fluid, but not bacteria, to leave the reactor. Although not extensively studied thus far, MBR is expected to offer many advantages to fermentative H2 production, such as high cell density, high organics removal rates, high-quality effluent by the membrane, and easy control of pH and temperature. This study focuses on the continuous H2 production in a MBR with a hollow-fiber filtration unit and mixed cells as inoculum at short hydraulic retention times (HRT). We compared the performance of the MBR to that of some other reactors, and attempted to evaluate the potential of the MBR for high-rate, fermentative H2 production. MATERIALS AND METHODS: Inoculum and Medium Composition Inoculum for the MBR was obtained from a sewage sludge digester in municipal wastewater treatment facility (Daejeon, Korea). In order to inactivate non-spore-forming, H2-consuming bacteria and to harvest sporeo
forming, H2-producing bacteria, the sewage sludge was heat-treated at 90 C for 20 min. The cells were o
acclimated to synthetic wastewater containing glucose in serum-bottle cultures at 37 C under anaerobic condition and then used as inoculum. The medium composition for the inoculum culture and MBR was (per liter): 10 g glucose, 10 g peptone, 5 g of a yeast extract, 0.5 g cystein·HCl, 4 g Na2CO3·10H2O, 0.9 g K2HPO4, 0.9 g KH2PO4, 0.9 g NaCl, 0.9 g (NH4)SO4, 0.09 g MgSO4, 0.09 g CaCl2, 0.1 mg of p-aminobenzoic acid and 0.01 mg biotin. MBR System and Operation Fig. 1 shows the schematic diagram of the MBR system used in this study. The reactor was a 2.2-L cylindrical Pyrex glass jar (working volume, 0.7 L), which was equipped with a headtop, outlet appendage, and water jacket, and constantly agitated at 100 rpm. There were 7 ports in the headtop, 1 for medium supply, 2 for pH control, 1 for biogas removal, and 3 for membrane filtration unit operation. The bundle of hollow-fiber membranes (see Table 1 for its characteristics; Woori Tech, Daejeon, Korea) was used to filtrate culture broth and packed in a 95-mL cylindrical flexi-glass column. Culture broth in the reactor was recirculated to the lower part of membrane filtration unit to generate cross-membrane flow, while the 2/9
WHEC 16 / 13-16 June 2006 – Lyon France headspace gas was introduced to the bottom part of the unit at 50 mL/min to reduce the fouling of membrane unless stated otherwise. The liquid broth and biogas collected from the upper part of the filtration unit were returned to the reactor. Pressure was measured at the permeate side of the filtration unit to determine the pressure across membrane. A permeate flow was controlled by adjusting the peristaltic pump on the o
permeate side of the unit. Fresh medium was stored at 4 C and continuously bubbled with an argon gas o
(99.999%) to maintain the anaerobic condition. The temperature was maintained at 37 C and pH at 5.5 by the automatic addition of 2 N NaOH. L G
B
F
H
K F
M J
E
B I B
B
N C A
D
Fig. 1. Schematic diagram of the MBR system used in this study : (A) feed tank ; (B) peristaltic pumps ; (C) reactor ; (D) magnetic stirrer ; (E) pH controller ; (F) gas meters ; (G) biogas out ; (H) membrane filtration unit ; (I) biogas recirculation ; (J) liquid recirculation ; (K) recirculation circuit of liquid and gas; (L) vacuum gauge; (M) permeate storage tank; (N) liquid waste stream. Table 1. Characteristics of the hollow-fiber membrane used. Characteristics Value 2 Membrane surface (m ) 0.047, 0.121 Membrane material polyvinylidene fluoride 0.1 Normal pore size (m) Membrane dimension (mm) Inner diameter, 1.1 – 1.2; Outer diameter, 2.0 – 2.1 Flow direction Out-inside Analytical Methods The volume of biogas produced was measured by a wet gas meter (W-NK, Shinagawa, Japan). The contents of H2, CO2 and CH4 in the biogas were analyzed by a gas chromatograph (14-B, Shimadzu, Japan), which was equipped with a thermal conductivity detector. A stainless steel column packed with a Molecular Sieve 5A (80/100 mesh; Alltech, Deerfield, USA) was used for H2 and the one packed with a Hayesep Q (80/100 mesh; Alltech) was used for CO2 and CH4, respectively. Glucose in the culture broth was measured by dinitrosalicylic acid (DNS) method. Biomass (dry cell weight, dcw) was measured according to standard methods (APHA, 1995). Permeate flow was measured gravimetrically. Microbial communities in the sewage sludge before and after heat-treatment were determined by PCRDGGE analysis. DNAs from the sludge were extracted by using an Ultraclean DNA kit (Mo Bio Labs. Inc., Solana Beach, USA). The 16S rDNA gene fragments were amplified with polymerase chain reaction (PCR) 3/9
WHEC 16 / 13-16 June 2006 – Lyon France o
primers 968f-GC and 1492r. PCR protocol was: initial denaturation for 5 min at 95 C and 30 cycles of o
o
o
denaturation for 30 s at 95 C, annealing for 30 s at 72 C, extension for 90 s at 72 C, followed by a final o
extension for 10 min at 72 C. DGGE was carried out using a Dcode Universal Mutation Detection System (Bio-Rad, Hercules, USA) in accordance with the manufacturer’s instructions. PCR products were separated o
at 100 V, 60 C for 6 h. The denaturing gradient in the gel was generated by mixing two stock solutions of 6% polyacrylamide containing 40% and 60% denaturant. After electrophoresis, the gel was stained with EtBr for 30 min, and DNA was visualized on a UV transilluminator. Major DNA bands were excised from DGGE gels and re-amplified by PCR. Nucleotide sequences were analyzed and screened against GenBank database using BLASTN to identify the most similar sequences in the database.
RESULTS AND DISCUSSION: Inoculum Preparation and Analysis of Microbial Community In order to use as inoculum for MBR, the sewage sludge from municipal wastewater treatment facility was o
heat-treated at 90 C for 20 min. The microbial communities in the fresh and heat-treated sludge were compared by PCR-DGGE analysis (Fig. 2). The number of DGGE bands detected from heat-treated sludge decreased compared to that of non-heat-treated sludge. The difference in DGGE banding pattern suggests that heat-treatment caused a change in the microbial community composition of the fresh culture. Most DGGE bands observed in the fresh sludge were affiliated with the Lactobacillus sp. (A-1 to A-7) and Bifidobacterium sp. (A-11 to A-14). In contrast, nucleotide sequences of the strongly stained bands (B-1 to B6) in the heat-treated sludge were most similar (92 – 100%) to the 16S rRNA gene of Clostridium perfringens, which is known as a spore-forming, H2-producing bacterium from carbohydrates. This result indicates that simple heat-shock treatment could be used for harvesting H2-producing bacteria from anaerobic microbial consortia. A
B A-1 A-3
B-1
Affiliation of DGGE fragments determined by their 16S rDNA sequence Band number
A-7 A-8
B-3 B-4
A-9
B-6
A-10 A-11 A-12
A-13 A-14
Affiliation
Similarity (%)
A-1
Lactobacillus sp.
95
A-3
Lactobacillus sp.
91
A-7
Lactobacillus sp.
95
A-8
Clostridium perfringens str.
94
A-9
Clostridium perfringens str.
99
A-10
Clostridium perfringens str.
96
A-11
Lactobacillus sp.
97
A-12
Lactobacillus sp.
99
A-13
Bifidobacterium sp.
100
A-14
Bifidobacterium sp.
100
B-1
Clostridium perfringens str.
92
B-3
Clostridium perfringens str.
99
B-4
Clostridium perfringens str.
99
B-6
Clostridium perfringens str.
100
Fig. 2. DGGE profiles and affiliation of 16S rRNA fragments obtained from DGGE gels. The fragments were PCR-amplified from the total DNA extracted from sludge. A, Fresh sludge; B, Heat-treated sludge. 4/9
WHEC 16 / 13-16 June 2006 – Lyon France H2 Production Using a CSTR Without and With Membrane Filtration Unit o
A CSTR was inoculated with the heat-treated sludge and initially operated in a batch mode at 37 C under anaerobic condition. When cell growth reached a late exponential phase at 6 h (data not shown), the culture was switched to a chemostat mode. When a CSTR was operated at a HRT of 4.69 h without a membrane filtration unit, volumetric H2 production rate and glucose removal efficiency were almost maintained at 0.23 L H2/Lh and 95%, respectively during the operation (Fig. 2A). However, the cell density in the reactor was gradually decreased after 2 h and then reached in 1.3 g dcw/L in 9 h. With a hollow-fiber membrane filtration 2
unit (effective surface area, 0.043 m ) at the same HRT, the CSTR could keep the cell density over 2 g dcw/L and showed similar H2 production and glucose removal rate for 5 h-operation, compared to that without a membrane unit. A further operation was not attempted since the pressure across membrane would be further increased. This result shows that a hollow-fiber membrane filtration unit could be used for maintaining the cell density of the CSTR without inhibiting H2-producing activity of microbial consortia in the CSTR.
2.5 1200 2.0 1.5
800
1.0 400 0.5 0 0
2
4
6
8
120 100 80 60 40 20
0.0
0
3.0
120
Glucose removal (%)
3.0
(A)
Cell concentration (g dcw/L)
Total H2 production (ml)
1600
10
1600
16 12 8 4 0
Total H2 production (ml)
Pressure across membrane (cm Hg)
(B) 2.5 1200 2.0 1.5
800
1.0 400 0.5 0
0.0 0
1
2
3
4
100 80 60 40 20
Glucose removal (%)
20
Cell concentration (g dcw/L)
Time (h)
0
5
Time (h)
Fig. 3. H2 production using a CSTR without (A) and with (B) hollow-fiber membrane filtration 2 unit. HRT was fixed at 4.69 h. The effective surface area of membrane was 0.043 m . Symbols: total H2 production (), cell concentration (), glucose removal (), and pressure across membrane (). Effect of HRT In order to evaluate the potential of MBR for high-rate, fermentative H2 production, bioreactor experiments were performed at very short HRTs of 0.79 – 1.24 h for 10 h (Fig. 4). The headspace gas in the reactor was 5/9
WHEC 16 / 13-16 June 2006 – Lyon France introduced to the bottom of the membrane unit at 50 mL/min to reduce the fouling of the membrane and increase the permeate flow across the membrane. The effective surface area of the membrane was also 2
2
increased from 0.043 m (Fig. 3B) to 0.121 m . As HRT decreased, volumetric H2 production rate increased from 1.37 0.13 L H2/Lh to 1.60 0.15 L H2/Lh, while the H2 yield decreased from 1.47 0.14 mol H2/mol glucose to 1.07 0.11 mol H2/mol glucose. The glucose removal rate was maintained at 95 – 99% regardless of HRT. The H2 content in biogas was kept at 50 - 70% (v/v) and no appreciable CH4 was detected during the experiments. Fig. 4B shows the changes in cell concentration, pressure across membrane, and permeate ratio in effluent after 10-h operation. Throughout operation of MBR, the fouling of membrane increased and it induced a lower permeate flow across the membrane. In order to improve permeate fluxes across membranes, the pressures across membranes at HRTs of 0.79, 1.05, and 1.24 h were gradually increased up to 58, 52, and 40 cm Hg, respectively by adjusting peristaltic pumps on the permeate sides of the membrane units. However, with decreasing HRT, the permeate ratio in the effluent decreased from 98% (v/v) to 52% (v/v). Cell concentrations in the reactor were in the range of 8 – 9 g dcw/L regardless of HRT. In a separated experiment at a HRT of 1.24 h, a biogas recirculation flowrate was increased to 800 mL/min and an antifoam agent was added in the feeding solution. With this, the fouling of membrane was considerably reduced and the pressure across membrane could be maintained below 30 cm Hg for more than 70 h (data not shown).
1.6
1.6
1.2
1.2
0.8
0.8
0.4
0.4
0.0 16
0.0 120 (B) 100
12 80 8
60 40
4 20 0
0.79
0
1.05
100 80 60 40
Glucose removal (%)
2.0
20 0 70
Permeate ratio in effluent (%)
Cell concentration (g dcw/L)
Volumetric H2 production rate (L H2/L.h)
2.0
120 H2 yield (mol H2/mol glucose)
2.4 (A)
60 50 40 30 20 10
Pressure across membrane (cmHg)
2.4
0
1.24
HRT (h)
Fig. 4. Effect of HRT on the performance of the MBR: (A) volumetric H2 production rate, ; H2 yield, ; and glucose removal, ; (B) cell concentration, ; permeate ratio in effluent, ; and pressure across membrane, . Continuous bioreactor experiments were performed for 10 h. The effective surface area of membrane and the gas recirculation flowrate were fixed at 0.121 2 m and 50 mL/min, respectively.
6/9
WHEC 16 / 13-16 June 2006 – Lyon France Comparison of Various H2-producing Bioreactors Various bioreactor systems have already been reported for the fermentative production of H2. Table 2 summarized the performance of some selected reactors including the MBR in the present study. Economics of fermentative H2 production depends on many factors such as H2 production rate, yield to carbon substrate, long-term stability of the reactor, scale-up, etc., and it is not easy to compare various reactors and draw a conclusion that a specific one is better than the others even under a specific set of conditions. Especially, the rate and yield of H2 production change significantly depending on experimental conditions including temperature, pH, the kind and concentration of substrate, and HRT, and these conditions have been poorly defined in many literatures. Nevertheless, it can be indicated that the MBR has a higher H2 production rate than most suspended and immobilized culture systems (Oh et al., 2004; Chang et al., 2002; Rachman et al., 1998; Yokoi et al., 1997; Chang and Lin, 2003; Chen and Lin, 2001; Mizuno et al., 2000; Lin and Chang, 1999; Nakamura et al., 1993; Lin and Jo, 2003). Some immobilized reactors (Oh et al., 2004; Chang et al., 2002; Kumar and Das, 2001; Rachman et al., 1998) have shown a high H2 production rate comparable to the MBR. However, in the work by Rachman et al. (1998) or Kumar and Das (2001), a pure culture was used as a biocatalyst and is hard to maintain in a large-scale H2 production. Chang et al., (2002) have used a mixed culture as a biocatalyst, but the H2 yield was very low as 1.1 mol H2/mol sucrose. Oh et al. (2004) developed o
a thermophilic (60 C) trickling biofilter reactor system, but thermophilic process can increase energy costs. In comparison, the MBR used in the present study exhibited a high H2 production rate (1714 mmol H2/Ld) and yield (1.1 mol H2/mol glucose). We suggest that the MBR is one of the most promising reactor systems for the fermentative production of H2. The superb performance of the MBR system is attributed to its distinctive characteristics. First of all, the MBR could maintain a high cell density of 8.3 g dcw/L at a very short HRT of 0.79 h, which is comparable to immobilized systems and considerably higher than chemostat-type digesters and CSTRs. Second, the quality of the biomass in the MBR seems to be outstanding. Although the microbial dynamics was not analyzed in this study, it is expected that the hydrogenotrophic methanogens, the major H2 consumers, do not exist in the MBR since no CH4 was detected in the biogas throughout operation of the MBR. Finally, compared to immobilized reactor systems, MBR system has no gas hold-up and, therefore, prevents the severe channeling of liquid and gas flows in the reactor.
Acknowledgements: This research was supported by the Hydrogen Energy R&D Center, one of the 21st Century Frontier R&D Programs, funded by the Ministry of Science and Technology of Korea. We would like to thank Eun-Kyung Kim for her technical assistance with operating MBR.
7/9
WHEC 16 / 13-16 June 2006 – Lyon France Table 2. Comparison of MBR with some selected bioreactors used for fermentative H2 production. Reactor type
Microorganism
Support matrix or membrane
HRT (h)
Carbon source (g/l)
Biomass (g VSS/l)
Volumetric H2 production rate (mmol H2/l·d)
H2 yield (mol H2/mol glucose)
MBR
Mixed culture
PVDF membrane
0.79
Glucose (10)
8.3 g dcw/L
1714
1.1
Present study
Thermophilic trickling biofilter Packed-bed
Mixed culture
Synthetic polymer Activated carbon
2
Glucose (20.6) Sucrose (17.8)
18 – 24
1050
1.1
Oh et al., 2004
14.6 a
1188 a
Enterobacter cloacae IIT-BT 08 E. aerogenes mutant AY-2 E. aerogenes strain HO-39 Mixed culture
Lignocellulosic agroresidues
1.08
Glucose (10) b
44 g dcw/L
1814
1.1 mol H2/mol sucrose a Nr
Self-flocculated cells Porous glass bead Biogranules
1.49
Glucose (15) Glucose (10) Sucrose (17.8)
17
1392
1.1
Nr
803 a
0.73
3.6 – 7.8
271
Mixed culture
Nu
13.3
3.6
184 a
1.5 mol H2/mol sucrose 1.63
CSTR
Mixed culture
Nu
8.5
Chemostat-type digester d Chemostat-type digester e
Mixed culture
Nu
6
1.45 (1.06) c 1.2 – 1.9
119 a (189) a,c 711
0.85 (1.43) c 1.7
Mixed culture
Nu
2
Nr
47 a
Nr
Anaerobic sequential batch reactor
Mixed culture
Nu
4
4.4 – 7.6
470
1.4 mol H2/mol sucrose
Packed-bed
Packed-bed Packed-bed Upflow anaerobic sludge blanket CSTR
Mixed culture
1
1 8
Glucose (18.7) Glucose (10) Glucose (18.7) Glucose (11.7) Sucrose (17.8)
Reference
Chang et al., 2002 Kumar and Das, 2001 Rachman et al., 1998 Yokoi et al., 1997 Chang and Lin, 2003 Chen and Lin, 2001 Mizuno et al., 2000 Lin and Chang, 1999 Nakamura et al., 1993 Lin and Jo, 2003
Nr, Not reported. Nu, Not used. b The value was estimated from experimental data of the reference. The feed solution contained 1% malt c d extract and 0.4% yeast extract. N2 was continuously sparged at a flow rate of 110 ml/min. The reactor was e completely mixed with produced biogas. The reactor was completely mixed with liquid circulation pump.
a
References: APHA (American Public Health Association), American Water Works Association, Water Environment th
Federation, Standard methods for the examination of water and wastewater, 19 ed., APHA, Washingto DC, 1995. Chang J, Lee K, Lin P, Biohydrogen production with fixed-bed bioreactors. Int J Hydrogen Energy, 27, 116774, 2002. Chen CC, Lin CY, Chang JS, Kinetics of hydrogen production with continuous anaerobic cultures utilizing sucrose as the limiting substrate, Appl Microbiol Biotechnol, 57, 56-64, 2001. Chen CC, Lin CY, Start-up of anaerobic hydrogen producing reactors seeded with sewage sludge, Acta Biotechnol, 21, 371-9, 2001. Chang F-Y, Lin C-Y, Biohydrogen production using an up-flow anaerobic sludge blanket reactor, Int J Hydrogen Energy, 29, 33-9, 2004. Kumar N, Das D, Continuous hydrogen production by immobilized Enterobacter cloacae IIT-BT 08 using lignocellulosic materials as solid matrices, Enzyme Microbiol Technol, 29, 280-7. 2001. Lin C-Y, Chang RC, Hydrogen production during the anaerobic acidogenic conversion of glucose, J Chem Technol Biotechnol, 74, 498-500, 1999. 8/9
WHEC 16 / 13-16 June 2006 – Lyon France Lin C-Y, Jo C-H, Hydrogen production from sucrose using an anaerobic sequencing batch reactor process, J Chem Technol Biotechnol, 78, 678-84, 2003. Mizuno O, Dinsdale R, Hawkes FR, Hawkes DL, Noike T, Enhancement of hydrogen production from glucose by nitrogen gas sparging, Bioresource Technol, 73, 59-65, 2000. Nakamura M, Kanbe H, Matsumoto J, 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, 81-8, 1993. Oh Y-K, Park MS, Seol E-H, Lee S-J, Park S, Isolation of hydrogen-producing bacteria from granular sludge of an upflow anaerobic sludge blanket reactor, Biotechnol Bioprocess Eng, 8, 54-7, 2003a. Oh Y-K, Seol E-H, Kim JR, Park S, Fermentative biohydrogen production by a chemoheterotrophic bacterium Citrobacter sp. Y19, Int J Hydrogen Energy, 28, 1353-9, 2003b. Oh Y-K, Kim SH, Kim M-S, Park S, Thermophilic biohydrogen production from glucose with trickling biofilter, Biotechnol Bioeng, 88, 690-8, 2004. Rachman MA, Nakashimada Y, Kakizono T, Nishio N, Hydrogen production with high yield and high evolution rate by self-flocculated cells of Enterobacter aerogenes in a packed-bed reactor, Appl Microbiol Biotechnol, 49, 450-4, 1998. Yokoi H, Tokushige T, Hirose J, Hayashi S, Takasaki Y, Hydrogen production by immobilized cells of aciduric Enterobacter aerogenes strain HO-39, J Ferment Bioeng 83, 481-484, 1997.
9/9
