- Email: [email protected]
Hydrogen production with effluent from an ethanol–H2-coproducing fermentation reactor using a single-chamber microbial electrolysis cell
Biosensors and Bioelectronics 24 (2009) 3055–3060
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Hydrogen production with effluent from an ethanol–H2 -coproducing fermentation reactor using a single-chamber microbial electrolysis cell Lu Lu a , Nanqi Ren a , Defeng Xing a,b,∗ , Bruce E. Logan b a State Key Laboratory of Urban Water Resource and Environment, School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China b Department of Civil and Environmental Engineering, 212 Sackett Building, The Pennsylvania State University, University Park, PA 16802, USA
a r t i c l e
i n f o
Article history: Received 1 February 2009 Received in revised form 15 March 2009 Accepted 17 March 2009 Available online 25 March 2009 Keywords: Biohydrogen production Microbial electrolysis cell (MEC) Effluent Ethanol-type fermentation
a b s t r a c t Hydrogen can be produced by bacterial fermentation of sugars, but substrate conversion to hydrogen is incomplete. Using a single-chamber microbial electrolysis cell (MEC), we show that additional hydrogen can be produced from the effluent of an ethanol-type dark-fermentation reactor. An overall hydrogen recovery of 83 ± 4% was obtained using a buffered effluent (pH 6.7–7.0), with a hydrogen production rate of 1.41 ± 0.08 m3 H2 /m3 reactor/d, at an applied voltage of Eap = 0.6 V. When the MEC was combined with the fermentation system, the overall hydrogen recovery was 96%, with a production rate of 2.11 m3 H2 /m3 /d, corresponding to an electrical energy efficiency of 287%. High cathodic hydrogen recoveries (70 ± 5% to 94 ± 4%) were obtained at applied voltages of 0.5–0.8 V due to shorter cycle times, and repression of methanogen growth through exposure of the cathode to air after each cycle. Addition of a buffer to the fermentation effluent was critical to MEC performance as there was little hydrogen production using unbuffered effluent (0.0372 m3 H2 /m3 /d at Eap = 0.6 V, pH 4.5–4.6). These results demonstrate that hydrogen yields from fermentation can be substantially increased by using MECs. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Dark-fermentation is a promising method for renewable biohydrogen production (Levin et al., 2004; Xing et al., 2008a,b). A low pH, ethanol-type fermentation process is one of the most successful dark-fermentation methods for producing hydrogen gas from sugars (Ren et al., 1997; Xing et al., 2008a). This process differs from traditional butyrate-type fermentation of clostridia due to the simultaneous production of high concentrations of acetic acid and ethanol, and a relatively high hydrogen yield of 1.93–2.81 mol H2 /mol glucose (Xing, 2006; Xing et al., 2008b). In addition, the low pH fermentation results in reduced concentrations of propionic acid compared to near neutral pH conditions (Ren et al., 1997). However, all dark-fermentation processes are limited to a maximum yield of 4 mol H2 /mol hexose. Thus, additional methods are needed to further convert organics in the fermentation effluent to hydrogen gas. Photo-fermentation can be used to increase hydrogen yields from dark-fermentation, but the rates and efficiencies are currently too low to make this process economically feasible (Fang et al., 2005).
∗ Corresponding author at: P.O. Box 2650, School of Municipal and Environmental Engineering, Harbin Institute of Technology, 202 Haihe Road, Nangang District, Harbin, Heilongjiang Province 150090, China. Tel.: +86 451 86282008; fax: +86 451 86282008. E-mail address: [email protected] (D. Xing). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.03.024
Hydrogen production by electrohydrogenesis in a microbial electrolysis cell (MEC) is a new method for generating hydrogen from acetate and other fermentation end products (Rozendal et al., 2006; Rozendal and Buisman, 2005; Liu et al., 2005; Call and Logan, 2008; Logan et al., 2008). In an MEC, exoelectrogens (Logan and Regan, 2006) oxidize a substrate and release electrons to the anode. Voltage is added to the circuit, allowing hydrogen production at the cathode. When acetate is used as a substrate, a voltage of >0.2 V in practice is required for hydrogen evolution (Cheng and Logan, 2007; Hu et al., 2008). This voltage is substantially less than the 1.8–2.0 V used in practice for hydrogen production via water electrolysis under alkaline conditions (Kim et al., 2002; Kinoshita, 1992). In most MECs, membranes have been used to keep the hydrogen produced at the cathode separate from the anode to prevent hydrogen losses due to methanogens (Ditzig et al., 2007; Liu et al., 2008; Rozendal et al., 2007, 2008; Chae et al., 2008; Tartakovsky et al., 2008). However, the use of a membrane can result in a large potential drop due to a pH gradient across the membrane (0.26–0.38 V; Rozendal et al., 2007). Furthermore, use of a membrane does not ensure complete recovery of hydrogen as the biogas produced on the cathode can diffuse through the membrane to the anode (Rozendal et al., 2006; Ditzig et al., 2007). Several MEC systems have been recently developed based on single-chamber MECs that lack a membrane (Call and Logan, 2008; Hu et al., 2008; Tartakovsky et al., 2009; Wagner et al., 2009). While the lack of a
3056
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
membrane increases hydrogen production rates, the gas produced has a lower concentration of hydrogen. Most MEC studies have examined the use of pure compounds (primarily acetate) as the substrate. While other substrates can be used, such as domestic or animal wastewaters (Ditzig et al., 2007; Wagner et al., 2009), hydrogen yields can be low or there can be appreciable methane gas production compared to tests with pure chemicals. In this study we examined hydrogen gas production in a membraneless MEC using the effluent from a low pH, ethanol-type continuous stirred-tank reactor (CSTR). Due to the low pH of this effluent, we examined hydrogen production using both buffered and unbuffered solutions. In addition, we examined the effect of removal of the microorganisms grown in the fermentation system on hydrogen production rates. 2. Materials and methods 2.1. Reactor construction Single-chamber membraneless MECs (liquid volume = 26 mL) with brush anodes (carbon fibers, TOHO TENAX Co., Ltd.) were constructed as previously described (Call and Logan, 2008) (Supplementary Fig. S1). The air cathode was wet-proofed (30%) carbon cloth (type B, E-TEK, 7 cm2 ) coated with 0.5 mg/cm2 of Pt (20 wt% Pt/C, JM) and a Nafion (5%, Dupont) binder on the water-facing side. The gas collection tube (total headspace volume = 8 mL), made of polymethyl methacrylate, was glued to the top of the reactor above the cathode. Copper wire was used to connect the electrodes to the circuit. The reactors were covered with aluminum foil to prohibit growth of photosynthetic bacteria. 2.2. Fermentation effluent The effluent produced from an ethanol-type fermentation CSTR (continuous stirred-tank reactor) that was used for hydrogen production (Ren et al., 2007) was used as the feed for the MECs. The CSTR was fed a molasses wastewater at an organic loading rate of 22.8 kg COD/m3 /d, and produced hydrogen gas at a maximum rate of 0.70 m3 H2 /m3 /d and yield of 0.017 g H2 /g COD (0.27 mol H2 /mol COD). The effluent had a pH in the range of 4.5–4.6, and a COD of 6500 ± 120 mg/L, with the following constituents identified (mg/L as COD): residual reducing sugars, 215 ± 38 (based on glucose); ethanol, 2379 ± 392; acetic acid, 740 ± 99; propionic acid, 351 ± 18; butyric acid, 204 ± 16; and valeric acid, 20 ± 6. Before being used as substrate, the effluent was centrifuged at 13,000 × g for 5 min to remove the fermentative bacteria in order to avoid the possibility of hydrogen production via fermentation (except as noted). In one test, non-centrifuged effluent was compared with centrifuged effluent. 2.3. Start-up and operation The MECs were first operated in electricity-production mode (i.e. as a microbial fuel cell, MFC), with the cathode exposed to air. The MFCs were inoculated with domestic wastewater, collected from the primary clarifier of the local wastewater treatment plant, in a nutrient buffer solution (NBS) (Na2 HPO4 , 4.58 g/L; NaH2 PO4 ·H2 O, 2.45 g/L; NH4 Cl, 0.31 g/L; KCl, 0.13 g/L; trace mineral; vitamin; 50:50 (v:v) wastewater and buffer) (Call and Logan, 2008) containing 1 g/L sodium acetate. During each fed-batch cycle, the solutions in the MFCs were replaced when the voltage decreased to <20 mV (external resistance of Rex = 1000 ). Once a maximum voltage of >0.5 V (Rex = 1000 ) was obtained for at least five successive fed-batch cycles, the cathodes were sealed and the reactors were switched to MEC mode.
The MECs were fed a 50:50 (v:v) mixture (pH 6.7–7.0) of fermentation effluent (pH 4.51–4.58) and NBS, or fermentation effluent. A fixed voltage (Eap ) of 0.2–0.8 V was applied to the MECs circuit using a power supply (3645A, Array Elec. Co., Ltd.) by connecting the positive pole of the power supply to the anode, and the negative to the cathode. The voltage across a high-precision resistor (Rex = 10 ) in the circuit was measured using a recorder (VX3000R, PanGu Auto. Sys. Co., Ltd.) at 5 s intervals to calculate current. An reference electrode (Ag/AgCl, 0.197 V vs NHE, model 218, Shanghai Preci. Sci. Instru. Co., Ltd.) was inserted into the top of the reactor (Supplementary Fig. S1), to measure the anode and cathode potentials. The volume of gas produced by the MECs was measured by water displacement using tubing and a needle to pierce the septum on the top of the gas collection tube. Experiments were conducted in fed-batch mode. After fresh medium was added to the reactors it was sparged with ultra high purity nitrogen gas (99.999%) for 10 min. All tests were conducted at room temperature (25 ◦ C). Two methods were used to suppress methanogen growth: (i) draining the reactors and exposing it to air for 30 min after each batch cycle (Call and Logan, 2008); (ii) reducing the reaction time () by replacing the solution in MECs when the voltage across the resistor decreased to less than 30% of the peak voltage. Applied voltage scans were obtained by varying the voltages stepwise from 0.2 to 0.8 V with 1 h intervals between each voltage change. The current densities (normalized to the liquid volume of the reactor or cathode surface area) and the potential of electrodes data of every applied voltage step were averaged over each time interval. 2.4. Analyses and calculations The gas was sampled (100 L) using a gastight syringe (500 L) and analyzed using a gas chromatograph (Agilent, 4890D; J&W Scientific, USA) with a thermal conductivity detector (TCD) and a column (19095P-QO3 HP-PLOT/Q, 15 m × 0.530 mm × 40.00 m, J&W Scientific, USA), with helium as the carrier gas. The cumulative volume for a specific gas (Vi , L) such as H2 , CH4 , CO2 is calculated as: Vi = (Vt + Vh )xi
(1)
where Vt and Vh are measured gas volume and the headspace volume of gas collection tube at sample time t, and xi is the specific gas fraction. The concentrations of volatile fatty acids (VFAs) and ethanol in the fermentation effluent were analyzed using a gas chromatograph (Agilent, 4890D; J&W Scientific, USA) equipped with a flame ionization detector (FID) and an appropriate column (19095N-123 HP-INNOWAX, 30 m × 0.530 mm × 1.00 m, J&W Scientific, USA) using a nitrogen carrier gas. Total chemical oxygen demand (COD) was measured following standard methods (Method 5220, APHA et al., 1995). The pH was measured using a pH meter (PHS-3C, Yangguang Lab. App. Co., Ltd.). Reducing sugars were quantified by 3,5-dinitrosalicylic acid (DNS) colorimetry using a spectrophotometer (DU800, Beckman). Hydrogen yields (YH2 , g H2 /g COD or mol H2 /mol COD) and Coulombic efficiencies (CE ) were based on COD removal, where a maximum of 0.125 g H2 / can be obtained per g of COD removed (2 mol H2 /mol COD on a molar basis). Cathodic hydrogen recoveries (rcat , e− to H2 in the cathode), overall hydrogen recovery (RH2 = CE rcat ), and maximum volumetric hydrogen production rates (Q, m3 H2 /m3 reactor/day) were calculated as previously described assuming standard biological conditions (T = 298.15 K, P = 1 bar, pH 7) (Logan et al., 2008). The over energy recovery based on both the electricity input and substrate (E+S ), the energy efficiency relative to the electrical input (E ) and substrate (S ), and the percentages of energy con-
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
3057
tributed by the power supply (eE ) and substrate (eS ) related to the total energy input were calculated as previously described under standard biological conditions (Logan et al., 2008). The amount of energy recovered in hydrogen and substrate was based on the upper heating value of combustion of hydrogen and substrate, using values of HH2 = −285.83 kJ/mol (Lide, 1995) and HS = −14.955 kJ/g COD (Lide, 1995). 3. Results 3.1. Hydrogen production Hydrogen production rates substantially increased from 0.05 ± 0.02 (at 0.2 V) to 1.41 ± 0.08 m3 H2 /m3 /d (0.6 V) using buffered fermentation effluent by increasing the applied voltage (Fig. 1). At 0.8 V the hydrogen production rate increased slightly to 1.52 ± 0.10 m3 H2 /m3 /d. To verify the hydrogen was produced primarily by electrohydrogenesis and not other biological reactions (e.g. fermentation of residual sugars), control tests were conducted without an applied voltage. In this case, hydrogen production rate was negligible (0.01 ± 0 m3 H2 /m3 /d, at 0 V). There was substantial COD in the influent not associated with the VFAs and other components in the fermentation effluent, such as cell biomass produced during fermentation. To determine if biomass and other particulate COD affected hydrogen production, we repeated the above experiments using buffered centrifuged or non-centrifuged samples. At an Eap = 0.6 V, the hydrogen production rate was 1.39 m3 H2 /m3 /d (135 A/m3 ) using a centrifuged sample, compared with 1.32 m3 H2 /m3 /d (123 A/m3 ) with a non-centrifuged sample (Supplementary Fig. S2). Thus, biomass and particulate substrate did not contribute to hydrogen production. Hydrogen production was accompanied by production of CO2 and CH4 (Fig. 2). The volume of CO2 produced did not vary with Eap , but the production of CH4 decreased from 1.3 ± 0.3 mL at 0.2 V, to 0.3 ± 0.1 mL at 0.8 V, likely due to the reduction of the time for the fed-batch cycle from 26 ± 0.5 to 19 ± 0 h. No CH4 was detected in tests using non-buffered fermentation effluent, likely due to the low pH (4.51–4.58) (Supplementary Table S1). When the original fermentation effluent was used as substrate, the hydrogen production rate was very low and decreased over successive batch cycles (Fig. 1 insert). Over four batch cycles, the hydrogen production rate declined from 0.0372 to 0.0058 m3 H2 /m3 /d at Eap = 0.6 V. Furthermore, whenever a non-buffered
Fig. 2. Biogas production at each applied voltage using buffered fermentation effluent. Reaction time () is shown on the secondary axis. (Error bars ± SD are based on duplicate measurements.)
fermentation effluent was used in an MEC test, the hydrogen production rate was always very low. After using a non-buffered effluent in a test, the reactor performed poorly in subsequent tests even with a buffered effluent, indicating that the bacteria suffered irreversible damage due to the low pH (4.5–4.6). 3.2. Hydrogen and energy recoveries At a Eap between 0.2 and 0.6 V, Coulombic efficiency, cathodic hydrogen recovery, and overall hydrogen recovery increased with the Eap to a maximum of CE = 87 ± 2%, rcat = 94 ± 4%, RH2 = 83 ± 4% (Eap = 0.6 V) using a buffered fermentation effluent (Fig. 3A). These values decreased when the applied voltage was further increased to 0.8 V. The energy efficiencies based on voltage input (E ) were all above 100%, and ranged from 135 ± 2% (at 0.2 V) to 252 ± 5% (at 0.4 V) (Fig. 3B). The fact that E was higher than 100% was due to the energy derived from the substrate. Energy inputs ranged from 0.098 to 1.39 kWh/m3 H2 (Eap = 0.2–0.6 V), which are much lower than that those typical of water electrolysis (5.6 kWh/m3 H2 ) (Ivy, 2004). The maximum energy efficiencies based on substrate (S = 99 ± 4%) and overall energy efficiency (E+S = 70 ± 3%) were achieved at Eap = 0.6 V. At the lower Eap , the substrate had a greater contribution to the energy captured in the hydrogen gas, whereas the power supply contributed more energy as the applied voltages were increased (Fig. 3B). 3.3. Voltage scans
Fig. 1. Hydrogen production rate (Q, m3 H2 /m3 /d) with buffered fermentation effluent as a function of applied voltage. (Error bars ± SD for buffered effluent are based on duplicate measurements.) Insert: hydrogen production using the original effluent (non-buffered) over four batch cycles at an applied voltage of 0.6 V (control is for 0 V applied).
There was a nearly linear increase in the current density for Eap from 0.2 to 0.8 V with an average slope of 225 A/m3 /V (8.33 A/m2 /V) (Fig. 4A). In comparison, the non-buffered effluent had an average slope of 32 A/m3 /V (1.18 A/m2 /V) between 0.2 and 0.8 V. In an MEC the hydrogen production rate is a direct function of the volumetric current density, so the low current densities were primarily responsible for the decreased performance of the system using the unamended fermentation effluent. There was a nominal current of 4–15 A/m3 (0.14–0.57 A/m2 /V) in the absence of the substrate, resulting in an average slope of 18 A/m3 /V (0.72 A/m2 /V) between 0.2 and 0.8 V. Fig. 4B shows the electrodes potentials measured during the applied voltage scans. Assuming the acetate was the main substrate utilized by the exoelectrogens, electrode overpotentials can be calculated by comparing the measured electrode potentials with theoretical potentials for acetate oxidation and proton reduction. Under standard biological conditions (T = 298.15 K, P = 1 bar, pH 7),
3058
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
Fig. 3. (A) Coulombic efficiency (CE ), cathodic recovery (rcat ), overall hydrogen recovery (RH2 ) and (B) input electricity efficiency (E ), substrate efficiency (S ), overall energy recovery (E+S ), electricity input contribution (eE ), substrate input contribution (eS ) with buffered fermentation effluent as a function of applied voltage. (Error bars ± SD are based on duplicate measurements.)
the theoretical anode potential for acetate oxidation is −0.476 V (vs Ag/AgCl), the theoretical cathode potential for proton reduction is −0.611 V (vs Ag/AgCl) (Fig. 4B). For buffered effluent, the overpotential of the anode increased slowly (0.051 ± 0.002 to 0.181 ± 0.002 V) at lower applied voltages, and then it sharply increased to 0.323 ± 0.001 and 0.403 ± 0 V at Eap = 0.7 and 0.8 V, respectively. The cathode overpotential had a maximum of 0.279 ± 0.003 V (at 0.6 V), and on average was 0.175 ± 0.002 V (Eap = 0.2–0.8 V) less than that of the anode (0.185 ± 0.002 V). When unamended effluent was used as substrate, the overpotential mainly occurred for the anode (average of 0.458 V) over four batch cycles, indicating the bacteria were adversely affected by the low pH solution. The cathode potential remained relatively constant, with no obvious overpotential due to the low current densities. The theoretical anode and cathode potentials are −0.310 and −0.464 V (vs Ag/AgCl, pH 4.51–4.58) for the unamended fermentation effluent.
Fig. 4. (A) Applied voltage scans and (B) measured electrodes potentials (vs Ag/AgCl) of buffered effluent during the applied voltage scans and original effluent at applied voltage of 0.6 V sustained four complete batch cycles (insert of figure, control is at applied of 0 V) (based on duplicate measurements for buffered effluent).
test (Eap = 0 V). Ethanol concentrations decreased with most applied voltages, while the concentration of propionic acid increased. The fermentation effluent contained residual reducing sugars as a result of incomplete utilization of the substrate in the fermentation reactor. In the MEC tests, the concentration of reducing sugars (based on glucose) decreased from 109 mg COD/L to an average of 11 mg COD/L (Eap = 0.2–0.8 V), resulting in an increase in some VFAs and ethanol. The pH of solutions at the end of each applied voltage batch cycle also slightly increased, but the buffering capacity of the solution prevented large pH changes (Fig. 5).
3.4. Substrate utilization Acetate was readily used as a substrate, as shown by a decrease in its concentration in most tests (Fig. 5). For example, acetate decreased from 370 mg COD/L in the original solution to 12 mg/L for Eap = 0.8 V. The concentrations of butyric and valeric acid remained relatively constant in all tests. It is clear that there was continued production of some constituents despite the overall decrease in total COD concentration. For example, acetate was evidently produced from other constituents in the fermentation effluent, as it increased in concentration from 370 to 604 mg COD/L in the control
Fig. 5. Total COD and specific constituents (bars), and solution pH (line), during a batch cycle using buffered fermentation effluent at different applied voltages.
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
3059
Table 1 Current density, Coulombic efficiency, overall hydrogen recovery, electrical efficiency, overall energy efficiency, and hydrogen production rates reported in the literatures versus these obtained in this study. Reactor system Single chamber with CEMa and AEMb Single chamber no membrane Two chamber with Nafion membrane Single chamber no membrane with brush anode Single chamber no membrane with brush anode Single chamber no membrane AEMb with granula anode Single chamber no membrane with brush anode Two chamber with Nafion membrane Single chamber no membrane with brush anode a b c d
Substrate
Eap (V)
CDc (A/m3 )
CE (%)
E (%)
E+S (%)
Q (m3 /m3 /d)
23 24b
1.0 0.8 0.8
68 97
53 93
194
75
0.33a 0.31b 6.32 0.05 3.12
Buffered effluentd
0.8
158
78
68
170
55
1.52
This study
Sodium acetate Acetic acid Buffered effluentd Effluentd Acetic acid Swine wastewater
0.6 0.6
52 99 135 23 2.8 106
75
62 91 83 4 53 17
204 261 246 4 169 190
58 82 70 4 53
0.53 1.10 1.41 0.04 0.02 0.9
Hu et al., 2008 Cheng and Logan, 2007 This study Rozendal et al., 2006 Wagner et al., 2009
0.6 0.5 0.5
87 24 92 29
a
Source
23 23b
1.0
a
RH2 (%)
29 26b 470 40 292
Sodium acetate Sodium acetate Acetate Sodium acetate
a
Rozendal et al., 2007 Tartakovsky et al., 2009 Chae et al., 2008 Call and Logan, 2008
Cation exchange membrane. Anion exchange membrane. Current density calculated based liquid volume of reactor. Effluent of a biohydrogen production reactor in ethanol-type dark-fermentation.
4. Discussion The use of an MEC for hydrogen production with a buffered ethanol-type fermentation effluent produced maximum hydrogen and energy recoveries of 83 ± 4% and 70 ± 3%, and a hydrogen production rate of 1.41 ± 0.08 m3 H2 /m3 /d (Eap = 0.6 V). These values compare reasonably well with other MEC studies (Table 1). Although higher hydrogen production rates could be achieved by increasing Eap (1.52 ± 0.10 m3 H2 /m3 /d at Eap = 0.8 V), this resulted in a greater percentage of the recovered energy being contributed by the power supply. The key factors influencing the hydrogen yield include current density (normalized to electrodes area or liquid volume), Coulombic efficiency, and cathodic hydrogen recovery. The high hydrogen production rate achieved here resulted from the high current density, which was obtained by not using a membrane in the MEC. In other MECs the cation or anion exchange membrane contributed 38% or 26% of the total resistance (Rozendal et al., 2007). The CE values were relatively high, ranging from 73 ± 3% to 87 ± 2%, for Eap of 0.4–0.8 V (Supplementary Table S2). At lower applied voltages of 0.2 and 0.3 V, CE values were much lower due to methanogenesis (Tartakovsky et al., 2008). This was supported by the high CH4 concentrations (2.2 ± 0.2% to 2.9 ± 0.3%) at low applied voltages, compared to much lower concentrations (0.6 ± 0.2%) at Eap = 0.8 V. However, all values are lower than CH4 concentrations of 26.4% obtained by Chae et al. (2008), likely due to exposure of the cathode to air and shorter cycle times. As the Eap was increased from 0.7 to 0.8 V, there was a more rapid increase in anode potential and the CE decreased (Fig. 4B). This more positive anode potential indicates that the activity of the exoelectrogens was impaired, giving other acetate-utilizing microorganisms a competitive advantage, and resulting in reduced performance. This same effect of applied voltage was observed by Chae et al. (2008). The rcat was also slightly reduced when the Eap was increased from 0.7 to 0.8 V, as was also observed by others in MEC tests using a high conductivity solution (Call and Logan, 2008), for reasons not well understood. The combined decrease of both CE and rcat resulted in a reduction of hydrogen recovery (RH2 ) at the highest applied voltage. Cathodic hydrogen recoveries may also have been adversely affected by long cycle times. Hydrogen gas can easily diffuse through tubing and reactor seals. The method used for measuring the volume of hydrogen gas required the reactor to build up pressure. This likely increased hydrogen losses, especially when cycle times were long. Alternative methods that should be used in the
future include respirometers (Call and Logan, 2008) or collection of the gas in a gas bag (Logan et al., 2008). While high hydrogen production rates were achieved using a buffered fermentation effluent, the reactor performance was poor using the actual fermentation effluent that had a pH between 4.5 and 5.0. Buffering an effluent using phosphate buffer would not be practical for an industrial process, so improvements in the process are needed. If a consortium of acidophilic exoelectrogens could be developed that could tolerate low pH environments, then buffering might not be necessary. Based on the MFC literature, bioelectricity can be generated under acidophilic conditions at pH 5.5 by using a selectively enriched hydrogen producing mixture culture (Mohan et al., 2008) and pH 5.0 using anaerobic sludge as inoculum (He et al., 2008). Using Shewanella oneidensis MR1, a power density of 6 W/m3 was obtained at a pH as low as 5 (Biffinger et al., 2008). Bicarbonate may also be a better choice than phosphate as a buffer (Fan et al., 2007). One advantage of operating the MEC at a low pH is that it might prohibit substantial growth of methanogens. Future research should focus on researching and optimizing microbial community structure in MEC anode to reach a balance between complex organic matter degradation and substrate utilization by exoelectrogens for hydrogen production through electrohydrogenesis. Compared to water electrolysis, the electricity costs needed for hydrogen generation using MECs are relatively low, particularly for a two-stage fermentation and MEC process train. The CSTR fermentation reactor produced hydrogen at a yield of 0.017 g H2 /g COD (0.27 mol H2 /mol COD), while the MEC achieved 0.103 g H2 /g COD (1.65 mol H2 /mol COD) using buffered effluent (Eap = 0.6 V). The total conversion is 0.120 g H2 /g COD (1.92 mol H2 /mol COD), corresponding to an overall hydrogen recovery of 96% compared to the maximum theoretical yield of 0.125 g H2 /g COD (2 mol H2 /mol COD). If the total hydrogen yield is used to calculate energy recovery, this is 17.2 kJ/g COD (550 kJ/mol COD). In the MEC process, the electricity energy input relative to COD removal is 6.0 kJ/g COD (192 kJ/mol COD) at applied voltage of 0.6 V. So, the energy efficiency of whole system based on electricity energy input is 287% (550/192), which is equivalent to the electricity energy demand of only 1.12 kWh/m3 H2 . This is much better than water electrolysis, which requires 5.6 kWh/m3 H2 . This energy demand translates into a cost of $0.83/kg H2 versus $4.16/kg H2 for water electrolysis (assuming $0.06/kWh; Energy Information Administration). Because the substrate used for hydrogen production is a waste stream, it is excluded from these calculations.
3060
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
5. Conclusions By using a single-chamber microbial electrolysis cell lacking a membrane, it was possible to produce hydrogen at high yields using effluent from an ethanol-type reactor for biohydrogen production. MEC achieved greater hydrogen yields and production rates than dark-fermentation as well as greater energy efficiencies. This twostage process could result in an electrical energy demand of only 1.12 kWh/m3 H2 , which is much less than that needed for water electrolysis (5.6 kWh/m3 H2 ). However, a buffer was needed as the unamended effluent resulted in poor performance at the pH of 4.5–4.6. Methane gas was produced in low amounts at higher applied voltages, but it could not be completely eliminated. Further improvements in the process should focus on developing an acid-tolerant electrogenic community, and additional methods to limit methane generation. Acknowledgements This research was supported by the National Natural Science Foundation of China (No. 30870037), National Renewable Energy Laboratory contract RFH-7-77623-01, the Paul L. Bush award administered by the Water Environment Research Foundation, and the KAUST Global Research Partnership. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bios.2009.03.024. References American Public Health Association, American Water Works Association, Water Pollution Control Federation, 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed., Washington, DC. Biffinger, J.C., Pietron, J., Bretschger, O., Nadeau, L.J., Johnson, G.R., Williams, C.C., Nealson, K.H., Ringeisen, B.R., 2008. Biosens. Bioelectron. 4, 900–905. Call, D., Logan, B.E., 2008. Environ. Sci. Technol. 42, 3401–3406.
Chae, K.J., Choi, M.J., Lee, J., Ajayi, F.F., Kim, I.S., 2008. Int. J. Hydrogen Energy 33, 5184–5192. Cheng, S., Logan, B.E., 2007. Proc. Natl. Acad. Sci. U.S.A. 104, 18871–18873. Ditzig, J., Liu, H., Logan, B.E., 2007. Int. J. Hydrogen Energy 32, 2296–2304. Energy Information Administration, Average Retail Price of Electricity to Ultimate Customers by End-Use Sector, 2006. Official Energy Statistics from the U.S. Government. http://www.eia.doe.gov/cneaf/electricity/epa/epat7p4.html. Fan, Y., Hu, H., Liu, H., 2007. Environ. Sci. Technol. 41, 8154–8158. Fang, H.H.P., Liu, H., Zhang, T., 2005. Int. J. Hydrogen Energy 30, 785–793. He, Z., Huang, Y., Manohara, A.K., Mansfeld, F., 2008. Bioelectrochemistry 74, 78–82. Hu, H., Fan, Y., Liu, H., 2008. Water Res. 42, 4172–4178. Ivy, J., 2004. National renewable energy laboratory. http://www.nrel.gov/hydrogen/. Kim, H.J., Park, H.S., Hyun, M.S., Chang, I.S., Kim, M., Kim, B.H., 2002. Enzyme Microbiol. Technol. 30, 145–152. Kinoshita, K., 1992. Electrochemical Oxygen Technology. Wiley, New York. Levin, D.B., Pitt, L., Love, M., 2004. Int. J. Hydrogen Energy 29, 173–185. Lide, D.R., 1995. CRC Handbook of Chemistry and Physics, 76 ed. CRC Press, Inc., Boca Raton. Liu, H., Grot, S., Logan, B.E., 2005. Environ. Sci. Technol. 39, 4317–4320. Liu, W., Wang, A., Ren, N., Zhao, X., Liu, L., Yu, Z., Lee, D., 2008. Energy Fuels 22, 159–163. Logan, B.E., Call, D., Cheng, S., Hamelers, H.V.M., Sleutels, T.H.J.A., Jeremiasse, A.W., Rozendal, R.A., 2008. Environ. Sci. Technol. 42, 8630–8640. Logan, B.E., Regan, J.M., 2006. Trends Biotechnol. 14, 512–518. Mohan, S.V., Mohanakrishna, G., Purushotham Reddy, B., Saravanan, R., Sarma, P.N., 2008. Biochem. Eng. J. 39, 121–130. Ren, N., Wang, B., Huang, J.C., 1997. Biotechnol. Bioeng. 54, 428–433. Ren, N., Xing, D., Rittmann, B.E., Zhao, L., Xie, T., Zhao, X., 2007. Environ. Microbiol. 9, 1112–1125. Rozendal, R.A., Buisman, C.J.N., 2005. Bio-Electrochemical Process for Producing Hydrogen. Patent WO-2005-005981. Rozendal, R.A., Hamelers, H.V.M., Euverink, G.J.W., Metz, S.J., Buisman, C.J.N., 2006. Int. J. Hydrogen Energy 31, 1632–1640. Rozendal, R.A., Hamelers, H.V.M., Molenkamp, R.J., Buisman, C.J.N., 2007. Water Res. 41, 1984–1994. Rozendal, R.A., Jeremiasse, A.W., Hamelers, H.V.M., Buisman, C.J.N., 2008. Environ. Sci. Technol. 42, 629–634. Tartakovsky, B., Manuel, M.F., Neburchilov, V., Wang, H., Guiot, S.R., 2008. J. Power Sources 182, 291–297. Tartakovsky, B., Manuel, M.F., Wang, H., Guiot, S.R., 2009. Int. J. Hydrogen Energy 34, 672–677. Wagner, R.C., Regan, J.M., Oh, S.-E., Zuo, Y., Logan, B.E., 2009. Water Res. 43, 1480–1488. Xing, D., 2006. Ph.D. Dissertation, Harbin Institute of Technology. Xing, D., Ren, N., Rittmann, B.E., 2008a. Appl. Environ. Microbiol. 74, 1232–1239. Xing, D., Ren, N., Wang, A., Li, Q., Feng, Y., Ma, F., 2008b. Int. J. Hydrogen Energy 33, 1489–1495.
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Hydrogen production with effluent from an ethanol–H2 -coproducing fermentation reactor using a single-chamber microbial electrolysis cell Lu Lu a , Nanqi Ren a , Defeng Xing a,b,∗ , Bruce E. Logan b a State Key Laboratory of Urban Water Resource and Environment, School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China b Department of Civil and Environmental Engineering, 212 Sackett Building, The Pennsylvania State University, University Park, PA 16802, USA
a r t i c l e
i n f o
Article history: Received 1 February 2009 Received in revised form 15 March 2009 Accepted 17 March 2009 Available online 25 March 2009 Keywords: Biohydrogen production Microbial electrolysis cell (MEC) Effluent Ethanol-type fermentation
a b s t r a c t Hydrogen can be produced by bacterial fermentation of sugars, but substrate conversion to hydrogen is incomplete. Using a single-chamber microbial electrolysis cell (MEC), we show that additional hydrogen can be produced from the effluent of an ethanol-type dark-fermentation reactor. An overall hydrogen recovery of 83 ± 4% was obtained using a buffered effluent (pH 6.7–7.0), with a hydrogen production rate of 1.41 ± 0.08 m3 H2 /m3 reactor/d, at an applied voltage of Eap = 0.6 V. When the MEC was combined with the fermentation system, the overall hydrogen recovery was 96%, with a production rate of 2.11 m3 H2 /m3 /d, corresponding to an electrical energy efficiency of 287%. High cathodic hydrogen recoveries (70 ± 5% to 94 ± 4%) were obtained at applied voltages of 0.5–0.8 V due to shorter cycle times, and repression of methanogen growth through exposure of the cathode to air after each cycle. Addition of a buffer to the fermentation effluent was critical to MEC performance as there was little hydrogen production using unbuffered effluent (0.0372 m3 H2 /m3 /d at Eap = 0.6 V, pH 4.5–4.6). These results demonstrate that hydrogen yields from fermentation can be substantially increased by using MECs. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Dark-fermentation is a promising method for renewable biohydrogen production (Levin et al., 2004; Xing et al., 2008a,b). A low pH, ethanol-type fermentation process is one of the most successful dark-fermentation methods for producing hydrogen gas from sugars (Ren et al., 1997; Xing et al., 2008a). This process differs from traditional butyrate-type fermentation of clostridia due to the simultaneous production of high concentrations of acetic acid and ethanol, and a relatively high hydrogen yield of 1.93–2.81 mol H2 /mol glucose (Xing, 2006; Xing et al., 2008b). In addition, the low pH fermentation results in reduced concentrations of propionic acid compared to near neutral pH conditions (Ren et al., 1997). However, all dark-fermentation processes are limited to a maximum yield of 4 mol H2 /mol hexose. Thus, additional methods are needed to further convert organics in the fermentation effluent to hydrogen gas. Photo-fermentation can be used to increase hydrogen yields from dark-fermentation, but the rates and efficiencies are currently too low to make this process economically feasible (Fang et al., 2005).
∗ Corresponding author at: P.O. Box 2650, School of Municipal and Environmental Engineering, Harbin Institute of Technology, 202 Haihe Road, Nangang District, Harbin, Heilongjiang Province 150090, China. Tel.: +86 451 86282008; fax: +86 451 86282008. E-mail address: [email protected] (D. Xing). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.03.024
Hydrogen production by electrohydrogenesis in a microbial electrolysis cell (MEC) is a new method for generating hydrogen from acetate and other fermentation end products (Rozendal et al., 2006; Rozendal and Buisman, 2005; Liu et al., 2005; Call and Logan, 2008; Logan et al., 2008). In an MEC, exoelectrogens (Logan and Regan, 2006) oxidize a substrate and release electrons to the anode. Voltage is added to the circuit, allowing hydrogen production at the cathode. When acetate is used as a substrate, a voltage of >0.2 V in practice is required for hydrogen evolution (Cheng and Logan, 2007; Hu et al., 2008). This voltage is substantially less than the 1.8–2.0 V used in practice for hydrogen production via water electrolysis under alkaline conditions (Kim et al., 2002; Kinoshita, 1992). In most MECs, membranes have been used to keep the hydrogen produced at the cathode separate from the anode to prevent hydrogen losses due to methanogens (Ditzig et al., 2007; Liu et al., 2008; Rozendal et al., 2007, 2008; Chae et al., 2008; Tartakovsky et al., 2008). However, the use of a membrane can result in a large potential drop due to a pH gradient across the membrane (0.26–0.38 V; Rozendal et al., 2007). Furthermore, use of a membrane does not ensure complete recovery of hydrogen as the biogas produced on the cathode can diffuse through the membrane to the anode (Rozendal et al., 2006; Ditzig et al., 2007). Several MEC systems have been recently developed based on single-chamber MECs that lack a membrane (Call and Logan, 2008; Hu et al., 2008; Tartakovsky et al., 2009; Wagner et al., 2009). While the lack of a
3056
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
membrane increases hydrogen production rates, the gas produced has a lower concentration of hydrogen. Most MEC studies have examined the use of pure compounds (primarily acetate) as the substrate. While other substrates can be used, such as domestic or animal wastewaters (Ditzig et al., 2007; Wagner et al., 2009), hydrogen yields can be low or there can be appreciable methane gas production compared to tests with pure chemicals. In this study we examined hydrogen gas production in a membraneless MEC using the effluent from a low pH, ethanol-type continuous stirred-tank reactor (CSTR). Due to the low pH of this effluent, we examined hydrogen production using both buffered and unbuffered solutions. In addition, we examined the effect of removal of the microorganisms grown in the fermentation system on hydrogen production rates. 2. Materials and methods 2.1. Reactor construction Single-chamber membraneless MECs (liquid volume = 26 mL) with brush anodes (carbon fibers, TOHO TENAX Co., Ltd.) were constructed as previously described (Call and Logan, 2008) (Supplementary Fig. S1). The air cathode was wet-proofed (30%) carbon cloth (type B, E-TEK, 7 cm2 ) coated with 0.5 mg/cm2 of Pt (20 wt% Pt/C, JM) and a Nafion (5%, Dupont) binder on the water-facing side. The gas collection tube (total headspace volume = 8 mL), made of polymethyl methacrylate, was glued to the top of the reactor above the cathode. Copper wire was used to connect the electrodes to the circuit. The reactors were covered with aluminum foil to prohibit growth of photosynthetic bacteria. 2.2. Fermentation effluent The effluent produced from an ethanol-type fermentation CSTR (continuous stirred-tank reactor) that was used for hydrogen production (Ren et al., 2007) was used as the feed for the MECs. The CSTR was fed a molasses wastewater at an organic loading rate of 22.8 kg COD/m3 /d, and produced hydrogen gas at a maximum rate of 0.70 m3 H2 /m3 /d and yield of 0.017 g H2 /g COD (0.27 mol H2 /mol COD). The effluent had a pH in the range of 4.5–4.6, and a COD of 6500 ± 120 mg/L, with the following constituents identified (mg/L as COD): residual reducing sugars, 215 ± 38 (based on glucose); ethanol, 2379 ± 392; acetic acid, 740 ± 99; propionic acid, 351 ± 18; butyric acid, 204 ± 16; and valeric acid, 20 ± 6. Before being used as substrate, the effluent was centrifuged at 13,000 × g for 5 min to remove the fermentative bacteria in order to avoid the possibility of hydrogen production via fermentation (except as noted). In one test, non-centrifuged effluent was compared with centrifuged effluent. 2.3. Start-up and operation The MECs were first operated in electricity-production mode (i.e. as a microbial fuel cell, MFC), with the cathode exposed to air. The MFCs were inoculated with domestic wastewater, collected from the primary clarifier of the local wastewater treatment plant, in a nutrient buffer solution (NBS) (Na2 HPO4 , 4.58 g/L; NaH2 PO4 ·H2 O, 2.45 g/L; NH4 Cl, 0.31 g/L; KCl, 0.13 g/L; trace mineral; vitamin; 50:50 (v:v) wastewater and buffer) (Call and Logan, 2008) containing 1 g/L sodium acetate. During each fed-batch cycle, the solutions in the MFCs were replaced when the voltage decreased to <20 mV (external resistance of Rex = 1000 ). Once a maximum voltage of >0.5 V (Rex = 1000 ) was obtained for at least five successive fed-batch cycles, the cathodes were sealed and the reactors were switched to MEC mode.
The MECs were fed a 50:50 (v:v) mixture (pH 6.7–7.0) of fermentation effluent (pH 4.51–4.58) and NBS, or fermentation effluent. A fixed voltage (Eap ) of 0.2–0.8 V was applied to the MECs circuit using a power supply (3645A, Array Elec. Co., Ltd.) by connecting the positive pole of the power supply to the anode, and the negative to the cathode. The voltage across a high-precision resistor (Rex = 10 ) in the circuit was measured using a recorder (VX3000R, PanGu Auto. Sys. Co., Ltd.) at 5 s intervals to calculate current. An reference electrode (Ag/AgCl, 0.197 V vs NHE, model 218, Shanghai Preci. Sci. Instru. Co., Ltd.) was inserted into the top of the reactor (Supplementary Fig. S1), to measure the anode and cathode potentials. The volume of gas produced by the MECs was measured by water displacement using tubing and a needle to pierce the septum on the top of the gas collection tube. Experiments were conducted in fed-batch mode. After fresh medium was added to the reactors it was sparged with ultra high purity nitrogen gas (99.999%) for 10 min. All tests were conducted at room temperature (25 ◦ C). Two methods were used to suppress methanogen growth: (i) draining the reactors and exposing it to air for 30 min after each batch cycle (Call and Logan, 2008); (ii) reducing the reaction time () by replacing the solution in MECs when the voltage across the resistor decreased to less than 30% of the peak voltage. Applied voltage scans were obtained by varying the voltages stepwise from 0.2 to 0.8 V with 1 h intervals between each voltage change. The current densities (normalized to the liquid volume of the reactor or cathode surface area) and the potential of electrodes data of every applied voltage step were averaged over each time interval. 2.4. Analyses and calculations The gas was sampled (100 L) using a gastight syringe (500 L) and analyzed using a gas chromatograph (Agilent, 4890D; J&W Scientific, USA) with a thermal conductivity detector (TCD) and a column (19095P-QO3 HP-PLOT/Q, 15 m × 0.530 mm × 40.00 m, J&W Scientific, USA), with helium as the carrier gas. The cumulative volume for a specific gas (Vi , L) such as H2 , CH4 , CO2 is calculated as: Vi = (Vt + Vh )xi
(1)
where Vt and Vh are measured gas volume and the headspace volume of gas collection tube at sample time t, and xi is the specific gas fraction. The concentrations of volatile fatty acids (VFAs) and ethanol in the fermentation effluent were analyzed using a gas chromatograph (Agilent, 4890D; J&W Scientific, USA) equipped with a flame ionization detector (FID) and an appropriate column (19095N-123 HP-INNOWAX, 30 m × 0.530 mm × 1.00 m, J&W Scientific, USA) using a nitrogen carrier gas. Total chemical oxygen demand (COD) was measured following standard methods (Method 5220, APHA et al., 1995). The pH was measured using a pH meter (PHS-3C, Yangguang Lab. App. Co., Ltd.). Reducing sugars were quantified by 3,5-dinitrosalicylic acid (DNS) colorimetry using a spectrophotometer (DU800, Beckman). Hydrogen yields (YH2 , g H2 /g COD or mol H2 /mol COD) and Coulombic efficiencies (CE ) were based on COD removal, where a maximum of 0.125 g H2 / can be obtained per g of COD removed (2 mol H2 /mol COD on a molar basis). Cathodic hydrogen recoveries (rcat , e− to H2 in the cathode), overall hydrogen recovery (RH2 = CE rcat ), and maximum volumetric hydrogen production rates (Q, m3 H2 /m3 reactor/day) were calculated as previously described assuming standard biological conditions (T = 298.15 K, P = 1 bar, pH 7) (Logan et al., 2008). The over energy recovery based on both the electricity input and substrate (E+S ), the energy efficiency relative to the electrical input (E ) and substrate (S ), and the percentages of energy con-
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
3057
tributed by the power supply (eE ) and substrate (eS ) related to the total energy input were calculated as previously described under standard biological conditions (Logan et al., 2008). The amount of energy recovered in hydrogen and substrate was based on the upper heating value of combustion of hydrogen and substrate, using values of HH2 = −285.83 kJ/mol (Lide, 1995) and HS = −14.955 kJ/g COD (Lide, 1995). 3. Results 3.1. Hydrogen production Hydrogen production rates substantially increased from 0.05 ± 0.02 (at 0.2 V) to 1.41 ± 0.08 m3 H2 /m3 /d (0.6 V) using buffered fermentation effluent by increasing the applied voltage (Fig. 1). At 0.8 V the hydrogen production rate increased slightly to 1.52 ± 0.10 m3 H2 /m3 /d. To verify the hydrogen was produced primarily by electrohydrogenesis and not other biological reactions (e.g. fermentation of residual sugars), control tests were conducted without an applied voltage. In this case, hydrogen production rate was negligible (0.01 ± 0 m3 H2 /m3 /d, at 0 V). There was substantial COD in the influent not associated with the VFAs and other components in the fermentation effluent, such as cell biomass produced during fermentation. To determine if biomass and other particulate COD affected hydrogen production, we repeated the above experiments using buffered centrifuged or non-centrifuged samples. At an Eap = 0.6 V, the hydrogen production rate was 1.39 m3 H2 /m3 /d (135 A/m3 ) using a centrifuged sample, compared with 1.32 m3 H2 /m3 /d (123 A/m3 ) with a non-centrifuged sample (Supplementary Fig. S2). Thus, biomass and particulate substrate did not contribute to hydrogen production. Hydrogen production was accompanied by production of CO2 and CH4 (Fig. 2). The volume of CO2 produced did not vary with Eap , but the production of CH4 decreased from 1.3 ± 0.3 mL at 0.2 V, to 0.3 ± 0.1 mL at 0.8 V, likely due to the reduction of the time for the fed-batch cycle from 26 ± 0.5 to 19 ± 0 h. No CH4 was detected in tests using non-buffered fermentation effluent, likely due to the low pH (4.51–4.58) (Supplementary Table S1). When the original fermentation effluent was used as substrate, the hydrogen production rate was very low and decreased over successive batch cycles (Fig. 1 insert). Over four batch cycles, the hydrogen production rate declined from 0.0372 to 0.0058 m3 H2 /m3 /d at Eap = 0.6 V. Furthermore, whenever a non-buffered
Fig. 2. Biogas production at each applied voltage using buffered fermentation effluent. Reaction time () is shown on the secondary axis. (Error bars ± SD are based on duplicate measurements.)
fermentation effluent was used in an MEC test, the hydrogen production rate was always very low. After using a non-buffered effluent in a test, the reactor performed poorly in subsequent tests even with a buffered effluent, indicating that the bacteria suffered irreversible damage due to the low pH (4.5–4.6). 3.2. Hydrogen and energy recoveries At a Eap between 0.2 and 0.6 V, Coulombic efficiency, cathodic hydrogen recovery, and overall hydrogen recovery increased with the Eap to a maximum of CE = 87 ± 2%, rcat = 94 ± 4%, RH2 = 83 ± 4% (Eap = 0.6 V) using a buffered fermentation effluent (Fig. 3A). These values decreased when the applied voltage was further increased to 0.8 V. The energy efficiencies based on voltage input (E ) were all above 100%, and ranged from 135 ± 2% (at 0.2 V) to 252 ± 5% (at 0.4 V) (Fig. 3B). The fact that E was higher than 100% was due to the energy derived from the substrate. Energy inputs ranged from 0.098 to 1.39 kWh/m3 H2 (Eap = 0.2–0.6 V), which are much lower than that those typical of water electrolysis (5.6 kWh/m3 H2 ) (Ivy, 2004). The maximum energy efficiencies based on substrate (S = 99 ± 4%) and overall energy efficiency (E+S = 70 ± 3%) were achieved at Eap = 0.6 V. At the lower Eap , the substrate had a greater contribution to the energy captured in the hydrogen gas, whereas the power supply contributed more energy as the applied voltages were increased (Fig. 3B). 3.3. Voltage scans
Fig. 1. Hydrogen production rate (Q, m3 H2 /m3 /d) with buffered fermentation effluent as a function of applied voltage. (Error bars ± SD for buffered effluent are based on duplicate measurements.) Insert: hydrogen production using the original effluent (non-buffered) over four batch cycles at an applied voltage of 0.6 V (control is for 0 V applied).
There was a nearly linear increase in the current density for Eap from 0.2 to 0.8 V with an average slope of 225 A/m3 /V (8.33 A/m2 /V) (Fig. 4A). In comparison, the non-buffered effluent had an average slope of 32 A/m3 /V (1.18 A/m2 /V) between 0.2 and 0.8 V. In an MEC the hydrogen production rate is a direct function of the volumetric current density, so the low current densities were primarily responsible for the decreased performance of the system using the unamended fermentation effluent. There was a nominal current of 4–15 A/m3 (0.14–0.57 A/m2 /V) in the absence of the substrate, resulting in an average slope of 18 A/m3 /V (0.72 A/m2 /V) between 0.2 and 0.8 V. Fig. 4B shows the electrodes potentials measured during the applied voltage scans. Assuming the acetate was the main substrate utilized by the exoelectrogens, electrode overpotentials can be calculated by comparing the measured electrode potentials with theoretical potentials for acetate oxidation and proton reduction. Under standard biological conditions (T = 298.15 K, P = 1 bar, pH 7),
3058
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
Fig. 3. (A) Coulombic efficiency (CE ), cathodic recovery (rcat ), overall hydrogen recovery (RH2 ) and (B) input electricity efficiency (E ), substrate efficiency (S ), overall energy recovery (E+S ), electricity input contribution (eE ), substrate input contribution (eS ) with buffered fermentation effluent as a function of applied voltage. (Error bars ± SD are based on duplicate measurements.)
the theoretical anode potential for acetate oxidation is −0.476 V (vs Ag/AgCl), the theoretical cathode potential for proton reduction is −0.611 V (vs Ag/AgCl) (Fig. 4B). For buffered effluent, the overpotential of the anode increased slowly (0.051 ± 0.002 to 0.181 ± 0.002 V) at lower applied voltages, and then it sharply increased to 0.323 ± 0.001 and 0.403 ± 0 V at Eap = 0.7 and 0.8 V, respectively. The cathode overpotential had a maximum of 0.279 ± 0.003 V (at 0.6 V), and on average was 0.175 ± 0.002 V (Eap = 0.2–0.8 V) less than that of the anode (0.185 ± 0.002 V). When unamended effluent was used as substrate, the overpotential mainly occurred for the anode (average of 0.458 V) over four batch cycles, indicating the bacteria were adversely affected by the low pH solution. The cathode potential remained relatively constant, with no obvious overpotential due to the low current densities. The theoretical anode and cathode potentials are −0.310 and −0.464 V (vs Ag/AgCl, pH 4.51–4.58) for the unamended fermentation effluent.
Fig. 4. (A) Applied voltage scans and (B) measured electrodes potentials (vs Ag/AgCl) of buffered effluent during the applied voltage scans and original effluent at applied voltage of 0.6 V sustained four complete batch cycles (insert of figure, control is at applied of 0 V) (based on duplicate measurements for buffered effluent).
test (Eap = 0 V). Ethanol concentrations decreased with most applied voltages, while the concentration of propionic acid increased. The fermentation effluent contained residual reducing sugars as a result of incomplete utilization of the substrate in the fermentation reactor. In the MEC tests, the concentration of reducing sugars (based on glucose) decreased from 109 mg COD/L to an average of 11 mg COD/L (Eap = 0.2–0.8 V), resulting in an increase in some VFAs and ethanol. The pH of solutions at the end of each applied voltage batch cycle also slightly increased, but the buffering capacity of the solution prevented large pH changes (Fig. 5).
3.4. Substrate utilization Acetate was readily used as a substrate, as shown by a decrease in its concentration in most tests (Fig. 5). For example, acetate decreased from 370 mg COD/L in the original solution to 12 mg/L for Eap = 0.8 V. The concentrations of butyric and valeric acid remained relatively constant in all tests. It is clear that there was continued production of some constituents despite the overall decrease in total COD concentration. For example, acetate was evidently produced from other constituents in the fermentation effluent, as it increased in concentration from 370 to 604 mg COD/L in the control
Fig. 5. Total COD and specific constituents (bars), and solution pH (line), during a batch cycle using buffered fermentation effluent at different applied voltages.
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
3059
Table 1 Current density, Coulombic efficiency, overall hydrogen recovery, electrical efficiency, overall energy efficiency, and hydrogen production rates reported in the literatures versus these obtained in this study. Reactor system Single chamber with CEMa and AEMb Single chamber no membrane Two chamber with Nafion membrane Single chamber no membrane with brush anode Single chamber no membrane with brush anode Single chamber no membrane AEMb with granula anode Single chamber no membrane with brush anode Two chamber with Nafion membrane Single chamber no membrane with brush anode a b c d
Substrate
Eap (V)
CDc (A/m3 )
CE (%)
E (%)
E+S (%)
Q (m3 /m3 /d)
23 24b
1.0 0.8 0.8
68 97
53 93
194
75
0.33a 0.31b 6.32 0.05 3.12
Buffered effluentd
0.8
158
78
68
170
55
1.52
This study
Sodium acetate Acetic acid Buffered effluentd Effluentd Acetic acid Swine wastewater
0.6 0.6
52 99 135 23 2.8 106
75
62 91 83 4 53 17
204 261 246 4 169 190
58 82 70 4 53
0.53 1.10 1.41 0.04 0.02 0.9
Hu et al., 2008 Cheng and Logan, 2007 This study Rozendal et al., 2006 Wagner et al., 2009
0.6 0.5 0.5
87 24 92 29
a
Source
23 23b
1.0
a
RH2 (%)
29 26b 470 40 292
Sodium acetate Sodium acetate Acetate Sodium acetate
a
Rozendal et al., 2007 Tartakovsky et al., 2009 Chae et al., 2008 Call and Logan, 2008
Cation exchange membrane. Anion exchange membrane. Current density calculated based liquid volume of reactor. Effluent of a biohydrogen production reactor in ethanol-type dark-fermentation.
4. Discussion The use of an MEC for hydrogen production with a buffered ethanol-type fermentation effluent produced maximum hydrogen and energy recoveries of 83 ± 4% and 70 ± 3%, and a hydrogen production rate of 1.41 ± 0.08 m3 H2 /m3 /d (Eap = 0.6 V). These values compare reasonably well with other MEC studies (Table 1). Although higher hydrogen production rates could be achieved by increasing Eap (1.52 ± 0.10 m3 H2 /m3 /d at Eap = 0.8 V), this resulted in a greater percentage of the recovered energy being contributed by the power supply. The key factors influencing the hydrogen yield include current density (normalized to electrodes area or liquid volume), Coulombic efficiency, and cathodic hydrogen recovery. The high hydrogen production rate achieved here resulted from the high current density, which was obtained by not using a membrane in the MEC. In other MECs the cation or anion exchange membrane contributed 38% or 26% of the total resistance (Rozendal et al., 2007). The CE values were relatively high, ranging from 73 ± 3% to 87 ± 2%, for Eap of 0.4–0.8 V (Supplementary Table S2). At lower applied voltages of 0.2 and 0.3 V, CE values were much lower due to methanogenesis (Tartakovsky et al., 2008). This was supported by the high CH4 concentrations (2.2 ± 0.2% to 2.9 ± 0.3%) at low applied voltages, compared to much lower concentrations (0.6 ± 0.2%) at Eap = 0.8 V. However, all values are lower than CH4 concentrations of 26.4% obtained by Chae et al. (2008), likely due to exposure of the cathode to air and shorter cycle times. As the Eap was increased from 0.7 to 0.8 V, there was a more rapid increase in anode potential and the CE decreased (Fig. 4B). This more positive anode potential indicates that the activity of the exoelectrogens was impaired, giving other acetate-utilizing microorganisms a competitive advantage, and resulting in reduced performance. This same effect of applied voltage was observed by Chae et al. (2008). The rcat was also slightly reduced when the Eap was increased from 0.7 to 0.8 V, as was also observed by others in MEC tests using a high conductivity solution (Call and Logan, 2008), for reasons not well understood. The combined decrease of both CE and rcat resulted in a reduction of hydrogen recovery (RH2 ) at the highest applied voltage. Cathodic hydrogen recoveries may also have been adversely affected by long cycle times. Hydrogen gas can easily diffuse through tubing and reactor seals. The method used for measuring the volume of hydrogen gas required the reactor to build up pressure. This likely increased hydrogen losses, especially when cycle times were long. Alternative methods that should be used in the
future include respirometers (Call and Logan, 2008) or collection of the gas in a gas bag (Logan et al., 2008). While high hydrogen production rates were achieved using a buffered fermentation effluent, the reactor performance was poor using the actual fermentation effluent that had a pH between 4.5 and 5.0. Buffering an effluent using phosphate buffer would not be practical for an industrial process, so improvements in the process are needed. If a consortium of acidophilic exoelectrogens could be developed that could tolerate low pH environments, then buffering might not be necessary. Based on the MFC literature, bioelectricity can be generated under acidophilic conditions at pH 5.5 by using a selectively enriched hydrogen producing mixture culture (Mohan et al., 2008) and pH 5.0 using anaerobic sludge as inoculum (He et al., 2008). Using Shewanella oneidensis MR1, a power density of 6 W/m3 was obtained at a pH as low as 5 (Biffinger et al., 2008). Bicarbonate may also be a better choice than phosphate as a buffer (Fan et al., 2007). One advantage of operating the MEC at a low pH is that it might prohibit substantial growth of methanogens. Future research should focus on researching and optimizing microbial community structure in MEC anode to reach a balance between complex organic matter degradation and substrate utilization by exoelectrogens for hydrogen production through electrohydrogenesis. Compared to water electrolysis, the electricity costs needed for hydrogen generation using MECs are relatively low, particularly for a two-stage fermentation and MEC process train. The CSTR fermentation reactor produced hydrogen at a yield of 0.017 g H2 /g COD (0.27 mol H2 /mol COD), while the MEC achieved 0.103 g H2 /g COD (1.65 mol H2 /mol COD) using buffered effluent (Eap = 0.6 V). The total conversion is 0.120 g H2 /g COD (1.92 mol H2 /mol COD), corresponding to an overall hydrogen recovery of 96% compared to the maximum theoretical yield of 0.125 g H2 /g COD (2 mol H2 /mol COD). If the total hydrogen yield is used to calculate energy recovery, this is 17.2 kJ/g COD (550 kJ/mol COD). In the MEC process, the electricity energy input relative to COD removal is 6.0 kJ/g COD (192 kJ/mol COD) at applied voltage of 0.6 V. So, the energy efficiency of whole system based on electricity energy input is 287% (550/192), which is equivalent to the electricity energy demand of only 1.12 kWh/m3 H2 . This is much better than water electrolysis, which requires 5.6 kWh/m3 H2 . This energy demand translates into a cost of $0.83/kg H2 versus $4.16/kg H2 for water electrolysis (assuming $0.06/kWh; Energy Information Administration). Because the substrate used for hydrogen production is a waste stream, it is excluded from these calculations.
3060
L. Lu et al. / Biosensors and Bioelectronics 24 (2009) 3055–3060
5. Conclusions By using a single-chamber microbial electrolysis cell lacking a membrane, it was possible to produce hydrogen at high yields using effluent from an ethanol-type reactor for biohydrogen production. MEC achieved greater hydrogen yields and production rates than dark-fermentation as well as greater energy efficiencies. This twostage process could result in an electrical energy demand of only 1.12 kWh/m3 H2 , which is much less than that needed for water electrolysis (5.6 kWh/m3 H2 ). However, a buffer was needed as the unamended effluent resulted in poor performance at the pH of 4.5–4.6. Methane gas was produced in low amounts at higher applied voltages, but it could not be completely eliminated. Further improvements in the process should focus on developing an acid-tolerant electrogenic community, and additional methods to limit methane generation. Acknowledgements This research was supported by the National Natural Science Foundation of China (No. 30870037), National Renewable Energy Laboratory contract RFH-7-77623-01, the Paul L. Bush award administered by the Water Environment Research Foundation, and the KAUST Global Research Partnership. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bios.2009.03.024. References American Public Health Association, American Water Works Association, Water Pollution Control Federation, 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed., Washington, DC. Biffinger, J.C., Pietron, J., Bretschger, O., Nadeau, L.J., Johnson, G.R., Williams, C.C., Nealson, K.H., Ringeisen, B.R., 2008. Biosens. Bioelectron. 4, 900–905. Call, D., Logan, B.E., 2008. Environ. Sci. Technol. 42, 3401–3406.
Chae, K.J., Choi, M.J., Lee, J., Ajayi, F.F., Kim, I.S., 2008. Int. J. Hydrogen Energy 33, 5184–5192. Cheng, S., Logan, B.E., 2007. Proc. Natl. Acad. Sci. U.S.A. 104, 18871–18873. Ditzig, J., Liu, H., Logan, B.E., 2007. Int. J. Hydrogen Energy 32, 2296–2304. Energy Information Administration, Average Retail Price of Electricity to Ultimate Customers by End-Use Sector, 2006. Official Energy Statistics from the U.S. Government. http://www.eia.doe.gov/cneaf/electricity/epa/epat7p4.html. Fan, Y., Hu, H., Liu, H., 2007. Environ. Sci. Technol. 41, 8154–8158. Fang, H.H.P., Liu, H., Zhang, T., 2005. Int. J. Hydrogen Energy 30, 785–793. He, Z., Huang, Y., Manohara, A.K., Mansfeld, F., 2008. Bioelectrochemistry 74, 78–82. Hu, H., Fan, Y., Liu, H., 2008. Water Res. 42, 4172–4178. Ivy, J., 2004. National renewable energy laboratory. http://www.nrel.gov/hydrogen/. Kim, H.J., Park, H.S., Hyun, M.S., Chang, I.S., Kim, M., Kim, B.H., 2002. Enzyme Microbiol. Technol. 30, 145–152. Kinoshita, K., 1992. Electrochemical Oxygen Technology. Wiley, New York. Levin, D.B., Pitt, L., Love, M., 2004. Int. J. Hydrogen Energy 29, 173–185. Lide, D.R., 1995. CRC Handbook of Chemistry and Physics, 76 ed. CRC Press, Inc., Boca Raton. Liu, H., Grot, S., Logan, B.E., 2005. Environ. Sci. Technol. 39, 4317–4320. Liu, W., Wang, A., Ren, N., Zhao, X., Liu, L., Yu, Z., Lee, D., 2008. Energy Fuels 22, 159–163. Logan, B.E., Call, D., Cheng, S., Hamelers, H.V.M., Sleutels, T.H.J.A., Jeremiasse, A.W., Rozendal, R.A., 2008. Environ. Sci. Technol. 42, 8630–8640. Logan, B.E., Regan, J.M., 2006. Trends Biotechnol. 14, 512–518. Mohan, S.V., Mohanakrishna, G., Purushotham Reddy, B., Saravanan, R., Sarma, P.N., 2008. Biochem. Eng. J. 39, 121–130. Ren, N., Wang, B., Huang, J.C., 1997. Biotechnol. Bioeng. 54, 428–433. Ren, N., Xing, D., Rittmann, B.E., Zhao, L., Xie, T., Zhao, X., 2007. Environ. Microbiol. 9, 1112–1125. Rozendal, R.A., Buisman, C.J.N., 2005. Bio-Electrochemical Process for Producing Hydrogen. Patent WO-2005-005981. Rozendal, R.A., Hamelers, H.V.M., Euverink, G.J.W., Metz, S.J., Buisman, C.J.N., 2006. Int. J. Hydrogen Energy 31, 1632–1640. Rozendal, R.A., Hamelers, H.V.M., Molenkamp, R.J., Buisman, C.J.N., 2007. Water Res. 41, 1984–1994. Rozendal, R.A., Jeremiasse, A.W., Hamelers, H.V.M., Buisman, C.J.N., 2008. Environ. Sci. Technol. 42, 629–634. Tartakovsky, B., Manuel, M.F., Neburchilov, V., Wang, H., Guiot, S.R., 2008. J. Power Sources 182, 291–297. Tartakovsky, B., Manuel, M.F., Wang, H., Guiot, S.R., 2009. Int. J. Hydrogen Energy 34, 672–677. Wagner, R.C., Regan, J.M., Oh, S.-E., Zuo, Y., Logan, B.E., 2009. Water Res. 43, 1480–1488. Xing, D., 2006. Ph.D. Dissertation, Harbin Institute of Technology. Xing, D., Ren, N., Rittmann, B.E., 2008a. Appl. Environ. Microbiol. 74, 1232–1239. Xing, D., Ren, N., Wang, A., Li, Q., Feng, Y., Ma, F., 2008b. Int. J. Hydrogen Energy 33, 1489–1495.