Optimization of membrane stack configuration for efficient hydrogen ...
Bioresource Technology 140 (2013) 399–405
Contents lists available at SciVerse ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Optimization of membrane stack configuration for efficient hydrogen production in microbial reverse-electrodialysis electrolysis cells coupled with thermolytic solutions Xi Luo a, Joo-Youn Nam b, Fang Zhang c, Xiaoyuan Zhang c, Peng Liang a, Xia Huang a,⇑, Bruce E. Logan a,c,⇑ a b c
State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, PR China Jeju Global Research Center, Korea Institute of Energy Research, 200 Haemajihaean-ro, Gujwa-eup, Jeju 695-971, Republic of Korea Department of Civil and Environmental Engineering, Penn State University, 212 Sackett Building, University Park, PA 16802, USA
h i g h l i g h t s Membrane stack configuration was optimized for the MREC utilizing NH4HCO3 solutions. The optimum number of cell pairs was determined to be five. Increasing the number of cell pairs did not appreciably affect anode performance. Adding an LC chamber reduced ammonia crossover and improved hydrogen production.
a r t i c l e
i n f o
Article history: Received 25 March 2013 Received in revised form 22 April 2013 Accepted 25 April 2013 Available online 3 May 2013 Keywords: Microbial reverse-electrodialysis electrolysis cell Ammonium bicarbonate Hydrogen Configuration Optimization
a b s t r a c t Waste heat can be captured as electrical energy to drive hydrogen evolution in microbial reverse-electrodialysis electrolysis cells (MRECs) by using thermolytic solutions such as ammonium bicarbonate. To determine the optimal membrane stack configuration for efficient hydrogen production in MRECs using ammonium bicarbonate solutions, different numbers of cell pairs and stack arrangements were tested. The optimum number of cell pairs was determined to be five based on MREC performance and a desire to minimize capital costs. The stack arrangement was altered by placing an extra low concentration chamber adjacent to anode chamber to reduce ammonia crossover. This additional chamber decreased ammonia nitrogen losses into anolyte by 60%, increased the coulombic efficiency to 83%, and improved the hydrogen yield to a maximum of 3.5 mol H2/mol acetate, with an overall energy efficiency of 27%. These results improve the MREC process, making it a more efficient method for renewable hydrogen gas production. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction A microbial electrolysis cell (MEC) is a bio-electrochemical system which can achieve hydrogen production from various types of renewable biomass (Cheng and Logan, 2007). An applied voltage (>0.3 V in practice) is required to overcome the thermodynamic limit for hydrogen evolution at the cathode (Logan and Rabaey, 2012). To eliminate the need for electrical grid energy, a sustainable method for hydrogen production was recently proposed based on integrating a small reverse electrodialysis (RED) stack into the MEC, which was called a microbial reverse-electrodialysis electrol⇑ Corresponding authors. Address: Department of Civil and Environmental Engineering, Penn State University, 212 Sackett Building, University Park, PA 16802, USA. Tel.: +1 814 863 7908; fax: +1 814 863 7304 (B.E. Logan), tel.: +86 10 62772324; fax: +86 10 62771472 (X. Huang). E-mail addresses: [email protected] (X. Huang), [email protected] (B.E. Logan). 0960-8524/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2013.04.097
ysis cell (MREC) (Kim and Logan, 2011a). A RED stack comprises an alternating series of anion (AEMs) and cation exchange membranes (CEMs) typically separated by porous spacers (Veerman et al., 2009; Vermaas et al., 2011a). When high concentration (HC) and low concentration (LC) solutions flow through alternating chambers in the stack, cations and anions in HC chamber migrate into the LC chamber through the membranes with opposing-charge functional groups due to concentration gradient, resulting in a potential difference across the membranes (Długołe˛cki et al., 2010; Vermaas et al., 2011b, 2013). Thus, renewable salinity gradient energy is converted to electrical energy to drive hydrogen evolution in an MREC, without the need for an external power source. The use of RED stacks in MRECs can be limited to estuaries or coastal areas when river water and seawater are used for the HC and LC solutions. The use of these natural waters also requires substantial and energy intensive pre-treatment to minimize membrane fouling (McGinnis et al., 2007). To avoid these limitations,
400
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
thermolytic solutions (Elimelech and Phillip, 2011) such as ammonium bicarbonate (NH4HCO3), have been proposed as the source of the salinity gradient energy for RED stacks (Cusick et al., 2012; Luo et al., 2012; Nam et al., 2012). With a thermolytic solution the ionic species can be volatilized and captured into HC solutions at temperatures below that needed to boil water. Heating ammonium bicarbonate solutions at 60 °C and 1 atm, for example, volatilizes ammonia and carbon dioxide, which can be condensed to form the HC solution (McGinnis and Elimelech, 2007). NH4HCO3 solutions have been shown to work in MRECs (Nam et al., 2012), and the HC and LC solutions can be regenerated using waste heat and conventional distillation technologies. Thus, waste heat can be captured as electrical energy, enabling hydrogen evolution in an MREC. The RED stack configuration and performance is critical to hydrogen production in an MREC. The number of cell pairs used can affect the electrochemical potential available to drive hydrogen production. Increasing cell pairs should improve the potential difference across the membrane stack (Długołe˛cki et al., 2009; Post et al., 2008), but adding cell pairs can increase the internal resistance through the addition of extra HC and LC chambers (Veerman et al., 2008, 2010). In addition, the use of more cell pairs can increase capital costs as the cost of the stack is dominated by cost of the ion exchange membranes (Post et al., 2010; Ramon et al., 2011; Turek and Bandura, 2007). Thus, it is essential to ascertain an optimum number of cell pairs for balancing cost and MREC performance. In addition, the application of higher potentials could adversely affect electrochemical performance by changing the anode potentials to values unfavorable for microbial oxidation of substrate. In this study, different numbers of cell pairs and stack arrangements were tested to determine the optimum membrane stack configuration for an MREC utilizing NH4HCO3 solutions. The number of cell pairs was optimized based on measuring current and hydrogen production rates, and calculating energy recoveries and efficiencies. Anode, cathode and stack performance were evaluated by galvanostatic polarization during variation of numbers of cell pairs. The stack arrangement was also changed to minimize ammonia crossover into the anode chamber. Previous studies used an AEM in the stack adjacent to the anode chamber (Kim and Logan, 2011a; Nam et al., 2012). This configuration minimized internal resistance by having a HC chamber adjacent to the anode chamber, but it resulted in high rates of ammonia transfer into the anolyte. In one series of tests, as much as 540 mg/L of total ammonia nitrogen (TAN) was transferred into the anode chamber over a single cycle (initial conductivity of the HC NH4HCO3 solution of 103 mS/cm) (Nam et al., 2012), which could inhibit current generation by the anode microorganisms (Nam et al., 2010) and therefore hydrogen gas production. To avoid this situation, the effect of adding an LC chamber adjacent to the anode chamber was examined to reduce TAN crossover.
catalyst layer (5 mg/cm2 10% Pt on carbon black) on the side facing membrane stack, and carbon black layer (5 mg/cm2) on the other side. A RED stack with up to 7 cell pairs was sandwiched between the anode and cathode chambers (Fig. 1). One cell pair consisted of a pair of HC and LC chambers, and a pair of AEM and CEM (Selemion AMV and CMV, Asashi glass, Japan). Except as noted otherwise, one additional AEM was used to close the last chamber at the end of the stack next to the electrode. The effective area of each membrane was 8 cm2 (4 cm 2 cm). Membranes were separated by polyethylene woven spacers and silicone gaskets with a thickness of 1.3 mm. 2.2. Solutions The anolyte was 1 g/L sodium acetate in a nutrient buffer solution containing 8.4 g/L NaHCO3, 0.31 g/L NH4Cl, 0.13 g/L KCl, 0.05 g/L Na2HPO4, 0.03 g/L NaH2PO4H2O, trace elements and minerals. A 1 M NaHCO3 solution was used as catholyte (Nam et al., 2012). Based on the solubility of NH4HCO3 at room temperature (approx. 2 M), a NH4HCO3 solution of 1.7 M was used (conductivity of 103 mS/cm) as the HC solution. The LC solution was prepared by dilution of the HC solution with distilled water to produce a salinity ratio of 75 (Luo et al., 2012).
2. Methods 2.1. MREC construction The MREC was composed of an anode chamber, a cathode chamber, and a RED membrane stack (Fig. 1). Two cubic Lexan blocks with a cylindrical cavity were used as the anode and cathode chamber (30 mL liquid volume each). A glass tube was glued to the top of cathode chamber for hydrogen collection (Mehanna et al., 2010). The anode was a heat treated graphite fiber brush (25 mm diameter 25 mm length; fiber type: PANEX 33 160 K, ZOLTEK). The cathode was stainless steel mesh (projected area: 7 cm2; Type 304, #60 mesh, McMaster-Carr) coated with platinum
Fig. 1. (a) Schematic and (b) photograph of microbial reverse-electrodialysis electrolysis cell (CEM, cation exchange membrane; AEM, anion exchange membrane; HCin, high concentration solution inlet; HCout, high concentration solution outlet; LCin, low concentration solution inlet; LCout, low concentration solution outlet).
401
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
2.3. MREC operation Brush anodes were inoculated with the effluent from an existing microbial fuel cell (MFC) and pre-acclimated in a single chamber air–cathode MFC. After reproducible peak voltages were obtained for at least three successive batch cycles, anodes were transferred to the MREC. The MREC was operated in fed-batch mode. Anolyte and catholyte were replaced when the produced current decreased to less than 0.5 mA, forming one complete cycle. HC and LC solutions were continuously fed into the stack at a fixed flow rate of 0.8 mL/min during one batch cycle. In order to investigate the influence of number of cell pairs on the performance of MREC, the number of cell pairs was varied from 1 to 7. In these tests, LC solution entered the membrane stack from the bottom of the LC chamber near the anode and flowed serially through the other LC chambers, exiting from the top of the chamber next to cathode chamber (Fig. 1). Similarly, the HC solution flowed serially through the HC chambers, but in the opposite direction (from cathode to anode side). In order to reduce transfer of ammonia nitrogen into the anolyte, in some experiments (as indicated) a CEM was added between the stack (containing 5 cell pairs) and the anode chamber, creating an extra LC chamber. Under this condition, LC solution entered the stack from the top of this extra chamber while the flow path of HC solution remained unchanged. All experiments were carried out at 30 °C in a constant temperature room. 2.4. Analyses Two Ag/AgCl reference electrodes (RE-5B, BASi) were separately inserted into the anode and cathode chambers to measure electrode potentials and stack voltages. The anode and cathode were connected to a 10 X external resistor to measure the current based on the voltage drop using a multimeter (Keitheley Instruments, OH, USA). Galvanostatic polarization was carried out with a potentiostat (VMP3, BioLogic, Claix, France) to determine the internal resistance of the MREC. Prior to the tests, the MREC was held under open circuit conditions for three hours. The anode was used as the working electrode while the cathode was used as the counter and reference electrode. Diverse current steps were set for different number of cell pairs according to the current production range in the fedbatch cycle: 2 pairs: 0.4, 0.5, 0.6, 0.7, 0.8, 1 and 1.2 mA; 3 pairs: 0.3, 0.6, 0.9, 1.2, 1.5, 1.8 and 2.1 mA; 5 pairs: 0.5, 0.8, 1, 1.5, 2, 2.5 and 3 mA; 7 pairs: 0.5, 1, 1.5, 2, 2.5, 3 and 3.5 mA. The current was set for 50 min for the first point, and at 20 min intervals thereafter. In order to evaluate the performance of anode, cathode and stack, electrode potentials and stack voltage were also measured using the two Ag/AgCl reference electrodes in anode and cathode chambers during polarization tests. Anode, cathode, and stack resistances were determined from the slopes of the polarization data measured for each of the electrodes, and from the stack polarization data. A respirometer (AER-200, Challenge Environmental) was used to measure the produced gas volume. Gas was collected for analysis in a gas bag (100 mL capacity; Cali-5-Bond, Calibrated Instruments Inc.). The compositions of cathode headspace and the gas bag were determined using gas chromatographs (models 8610B and 310, SRI Instruments, CA, USA) as previously described (Cheng and Logan, 2011). Total chemical oxygen demand (COD) was measured according to standard methods (COD Reagent, HACH Company). TAN concentrations in the anolyte were determined before and after each cycle (nitrogen ammonia reagent (salicylate), HACH Company). Conductivity and pH of anolyte, catholyte, HC and LC solutions were measured using a conductivity-pH meter (SevenMulti, Mettler Toledo, OH, USA). The conductivity of HC and LC
solutions was converted into concentration through a second order calibration curve. Coulombic efficiency (gCE, %), cathodic hydrogen recovery (rcat, %), hydrogen yield (Y, mol H2/mol acetate) and maximum volumetric hydrogen production rate (Q, m3 H2/m3 anolyte/d) were determined as previously described (Kim and Logan, 2011a). Energy recovery (rE) was the ratio of the combustion energy of produced hydrogen to the total energy provided to the MREC (substrate energy and salinity gradient energy), calculated (Nam et al., 2012) as:
rE ¼
DHh nh
ð1Þ
in DHs nin s þX
where nh and nin s are the amounts of produced hydrogen and supplied substrate respectively (mol), DHh and DHs the heat of combustion for hydrogen and substrate respectively (J/mol), Xin the free energy released from the total mixing of the influent HC and LC solutions (J). Energy efficiency (gE) was defined as the combustion energy of produced hydrogen divided by the input energy extracted by MREC (extracted substrate energy and extracted salinity gradient energy), and calculated as:
gE ¼
DH h n h
ð2Þ
in out out DHs ðnin s ns Þ þ X X
where nout is the amount of remaining substrate at the end of a s batch cycle (mol), Xout the free energy released from the total mixing of the effluent HC and LC solutions (J). Xin was determined (Cusick et al., 2012) using in
X ¼ RT
X i
V H C in i;H
ln
ain i;H ain i;m
þ
V L C in i;L
ln
ain i;L
!
ain i;m
ð3Þ
where R is the gas constant [J/(mol K)], T the absolute temperature (K), V the volume of solution, C the concentration of ionic species i in the solution, a the activity of ionic species i. The subscripts indicate H = HC, L = LC, and m is the mixture of the two solutions. The superscript in indicates influent. Xout was defined in an analogous manner to Xin by changing the concentrations and activities of influents into those of effluents in Eq. (3). Concentrations and activities of the ionic species in NH4HCO3 solution were estimated by OLI Stream Analysis software (OLI Systems Inc., Morris Plains, NJ) as previously described (Cusick et al., 2012). 3. Results and discussion 3.1. Current generation and hydrogen production using different numbers of cell pairs Increasing the number of cell pairs from 1 to 7 improved current production (Fig. 2a). Peak currents were observed at the beginning of the fed-batch cycle, and the currents decreased over time due to the substrate depletion. The peak current for 7 cell pairs (4 mA) was nearly thirteen times that of 1 cell pair (0.3 mA). This increase in current with the use of more cell pairs was mainly attributed to the sharper increase of stack open circuit voltage (OCV) than changes in the total internal resistance (see below). Peak current increased by only 29% with an increase in cell pairs from 5 to 7, compared to a 77% improvement going from 2 to 3 cell pairs, and a 35% improvement going from 3 to 5 cell pairs. This smaller enhancement with more cell pairs suggested that further addition of cell pairs would not substantially improve the current. When 1 cell pair was used, there was essentially no current production as the peak current of the MREC was lower than 0.5 mA, which was defined as the endpoint of a fed-batch cycle.
402
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
higher stack voltage. Anode and cathode potentials were essentially unchanged by using different numbers of cell pairs in polarization tests (Fig. 3b), demonstrating that the changes in performance were due to solely the RED stack. The maximum current density in polarization tests increased from 0.17 mA/cm2 (2 cell pairs) to 0.5 mA/cm2 (7 cell pairs) (Fig. 3a and b). Anode, cathode, and stack resistances were estimated from the slopes of the polarization data given in Fig. 3. The stack resistance predominated, and it increased with the use of more cell pairs, accounting for 63% (2 cell pairs) to 88% (7 cell pairs) of the total internal resistance (Fig. 4). It was expected that adding cell pairs would produce larger membrane and ohmic resistances of the salt solutions, and increase stack resistance (Veerman et al., 2008, 2010). Compared with stack resistance, cathode or anode resistances only decreased slightly with the increase of cell pairs, illustrating that the internal resistance change of MREC was mainly caused by the variation in stack resistances. The addition of cell pairs brought about a sharper increase in stack OCV than the total internal resistance (Fig. 4). For example, the total internal resistance increased by only 27% using 5 cell pairs (145 X) to 7 cell pairs (185 X), compared to a 54% improvement in the stack OCV. This sharper increase of stack OCV therefore primarily explains why adding cell pairs improved the current generation for MREC.
Fig. 2. (a) Current generation and (b) hydrogen volume and maximum volumetric hydrogen production rate (Q) of MREC using different number of cell pairs.
As an increase in current results in faster oxidation of substrate, the cycle time of the MREC decreased from 36 h (2 cell pairs) to 22.5 h (7 cell pairs) (Fig. 2a). Total coulombs produced in one fed-batch cycle increased gradually from 103 C (2 cell pairs) to 192 C (5 cell pairs), but it did not further increase with additional cell pairs. A larger hydrogen gas volume was obtained by adding more cell pairs (Fig. 2b). When the number of cell pairs was larger than 5, the volume of hydrogen produced stabilized at about 27 mL (Y = 2.9 mol H2/mol acetate), or 0.9 L H2/L anolyte volume. Little hydrogen (0.7 mL) was generated using 1 cell pair, again indicating a single cell pair was insufficient for hydrogen evolution in an MREC. The variation of hydrogen volumes recovered was consistent with that of the total coulombs produced in one cycle as a function of the number of cells pairs. This is expected due to the proportional relationship between hydrogen volume and total coulombs (Call and Logan, 2008), and the high cathodic hydrogen recoveries (around 100%) obtained here. As a result, the volume of hydrogen gas recovered was unchanged when 5 or more cell pairs were used. Increasing the number of cell pairs improved the maximum volumetric hydrogen production rate (Q) (Fig. 2b). As Q mainly depends on the peak current (Logan et al., 2008), it was expected that more cell pairs would result in larger peak currents, and improve Q. 3.2. Galvanostatic polarization tests Adding cell pairs increased the stack voltages and slopes of the stack polarization curve (Fig. 3a). More cell pairs created larger potential differences across the membrane stack, producing the
Fig. 3. (a) RED stack voltage and (b) anode (A) and cathode (C) potentials vs. current density for the MREC using different number of cell pairs. Electrode potentials were reported vs. Ag/AgCl reference electrode (+211 mV vs. standard hydrogen electrode).
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
403
3.3. Energy recovery, energy efficiency, coulombic efficiency and COD removal The energy that could be recovered based on the hydrogen gas produced (Wh) increased gradually with the addition of cell pairs and finally reached a plateau of 309 J (Fig. 5a). The substrate energy extracted using the MREC (Ws) was essentially unchanged by varying the number of cell pairs, and it stabilized at about 326 J, which was mainly due to the similar COD removals in the different experiments (see below). The salinity gradient energy extracted using the RED stack (WSGE) decreased slightly when the number of cell pairs rose from 2 to 3, and increased gradually with the further addition of cell pairs (Fig. 5a). More cell pairs resulted in not only the improved current generation for MREC, but also more extensive co-ion transport due to the longer hydraulic retention time, resulting in larger concentration changes in the HC and LC solutions after flowing through the membrane stack. The conductivity of effluent LC solution increased from 2.7 mS/cm (2 cell pairs) to 4.6 mS/cm (7 cell pairs) while that of effluent HC solution decreased from 99 mS/ cm (2 cell pairs) to 96 mS/cm (7 cell pairs). The larger concentration change signified a higher degree of mixing between HC and LC solutions, implying more salinity gradient energy was consumed in the MREC per unit time (Kim and Logan, 2011a, 2011b). Thus, raising the number of cell pairs from 3 to 7 led to the increase of WSGE due to similar cycle times. As the cycle time for 2 cell pairs (36 h) was much longer, the WSGE under this condition was slightly larger than that obtained with 3 cell pairs. A greater amount of energy was recovered (energy recovered versus total applied) when using more cell pairs (Fig. 5b). This was due to the larger Wh, as well as the shorter cycle time which decreased the total salinity gradient energy input during one fedbatch cycle when adding cell pairs. Energy recovery overall, however, remained very low (2–5%), mainly due to the large amounts of salinity gradient energy provided (5335–8751 J), which was more than 93% of the total energy supplied to the MREC. The greatest energy efficiency (recovered relative that entering and leaving) of 27% was achieved using 5 cell pairs (Fig. 5b). This efficiency is comparable to that of traditional RED systems (10–35% when ignoring power losses due to electrode reactions) using NaCl for the HC and LC solutions (Veerman et al., 2009). When the number of cell pairs decreased from 5 to 2, the reduction of Wh was more significant than WSGE, resulting in a lower energy efficiency. Comparing 5 and 7 cell pairs, the WSGE for 7 cell pairs was larger while Wh remained nearly unchanged, resulting in a smaller energy efficiency for 7 cell pairs.
Fig. 5. (a) Extracted input energy (Ws, extracted substrate energy; WSGE, extracted salinity gradient energy) and output energy (Wh, combustion energy of hydrogen), (b) energy recovery (rE), energy efficiency (gE), coulombic efficiency (gCE) and COD removal of MREC at different number of cell pairs.
Coulombic efficiency substantially improved with more cell pairs, increasing from 36% for 2 cell pairs to 66% for 5 cell pairs, but it remained nearly constant with additional cell pairs (Fig. 5b). More cell pairs resulted in shorter cycle times, increasing the coulombic efficiencies. Shorter cycle times are produced by higher current densities, resulting in less oxygen leaking into the anode chamber and thus less loss of substrate by aerobic processes (Kim and Logan, 2011a,b). COD removal was not affected by the number of cell pairs, ranging from 92% to 94% in all of the tests (Fig. 5b). This high COD removal mainly resulted from the neutral anolyte pH (final pHs of 7.3–7.6), which was maintained by the use of high concentration bicarbonate buffer (100 mM) and the flow of bicarbonate ions from the adjacent HC chamber into anolyte. Taken together, these results suggest that 5 cell pairs represent the most useful compromise in performance and minimization of costs. The maximum hydrogen volume and energy recoveries were obtained with 5 or 7 cell pairs. The highest energy efficiency was achieved using 5 cell pairs, but the largest peak current was obtained for 7 cell pairs. Considering the capital costs for additional membranes, 5 cell pairs provided the best overall stack architecture. 3.4. Ammonia nitrogen crossover into anolyte without an additional LC chamber
Fig. 4. Effect of the number of cell pairs on internal resistance of MREC and stack open circuit voltage (OCV).
Transport of ammonia nitrogen into the anolyte occurred in all the above tests. The TAN concentration in the anolyte of MREC
404
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
ranged from 450 mg/L to 800 mg/L after one fed-batch cycle (Fig. 6). The main nitrogen forms in HC solution are ammonium ðNHþ 4 Þ, ammonia (NH3), and carbamate ðNH2 CO2 Þ (Cusick et al., þ 2012). Negatively charged carbamate and NH4 , due to co-ion transport, migrated from the HC solution and through the AEM adjacent into the anode chamber. The MREC with 2 cell pairs had the highest TAN concentration (789 mg/L) due to low current densities and the longest cycle times. The MREC with 7 cell pairs had the shortest cycle time, and thus the lowest TAN concentration. Although the cycle time for 5 cell pairs was shorter than that for 3 cell pairs, the final TAN concentration in anolyte for 5 cell pairs was higher, probably due to the transfer of more carbamate into the anolyte resulting from the higher current production for 5 cell pairs than that for 3 cell pairs. The anolyte conductivity increased from 9.2 mS/cm to 12– 14 mS/cm after one cycle (Fig. 6), mainly due to the ionic flux from HC solution into the anode chamber. The variation of final anolyte conductivity was consistent with that of the final TAN concentration, suggesting anolyte conductivity change was dominated by the ammonia nitrogen migration.
3.5. Reducing ammonia nitrogen crossover by adding an LC chamber To reduce the migration of ammonia nitrogen into anolyte, an extra LC chamber was added between the anode chamber and the RED stack with 5 cell pairs. As a result of this change, there were significant decreases in both ammonia nitrogen crossover (60%) and final anolyte conductivity (66%) (Fig. 6). The introduction of this additional LC chamber increased the cationic flux from the anolyte to the LC solution, decreasing the anolyte conductivity from 9.2 mS/cm to 4.5 mS/cm after one cycle. In addition, only a small quantity of NHþ 4 was transferred into the anolyte, with a final TAN concentration of 311 mg/L. The presence of the extra LC chamber increased the cycle time (from 23 h to 29 h) but did not appreciably alter the peak current (Fig. S1a). There was an increase of 26% of the total coulombs produced, which resulted from the improvement of coulombic efficiency (see below). The hydrogen gas yield increased (32 mL, Y = 3.5 mol H2/mol acetate) due to the increased coulombic recovery (Fig. S1b). The maximum volumetric hydrogen production rate was not affected by adding the LC chamber (Fig. S1b), mainly due to the constant peak current. Energy recovery, energy efficiency and COD removal remained nearly constant with the additional LC chamber, while coulombic
efficiency increased appreciably from 66% to 83% (Fig. S1c). The higher coulombic efficiency was most likely due to the low TAN concentration in the anolyte, which would have avoided ammonia inhibition of exoelectrogenic microorganisms. Adding the LC chamber prolonged the cycle time, which increased both the WSGE (from 813 J to 1127 J) and total salinity gradient energy input (from 5587 J to 7033 J), thus resulting in the nearly unchanged energy recovery and efficiency. Anode potentials or cathode potentials remained totally the same in polarization tests (Fig. S1d), demonstrating stable anode and cathode performance. Consequently, nearly identical anode and cathode resistances were observed (Fig. S1e). The existence of the extra LC chamber resulted in similar increase for stack OCV (by 7%) and the total internal resistance (by 8%) (Fig. S1e), explaining why the peak current remained nearly constant. These results demonstrate that adding an LC chamber was an effective strategy to reduce the migration of ammonia nitrogen into anolyte and improve the hydrogen production performance for MREC. 4. Conclusions Membrane stack configuration was optimized to achieve a maximum hydrogen yield of 3.5 mol H2/mol acetate, energy efficiency of 27%, and coulombic efficiency of 83% using the MREC. Based on performance and a desire to minimize membrane total area, the optimum number of cell pairs was five. Varying the number of cell pairs did not appreciably affect anode and cathode performance. Adding an LC chamber adjacent to anode chamber greatly reduced ammonia crossover into anolyte, increased coulombic efficiency and further improved hydrogen production. Thus, an extra LC chamber is recommended when using NH4HCO3 solutions in MREC applications. Acknowledgements This research was supported by Award KUS-I1-003-13 from the King Abdullah University of Science and Technology (KAUST), the National High Technology Research and Development Program of China (863 Program) (No. 2011AA060907) and a scholarship from the China Scholarship Council (CSC). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.biortech. 2013.04.097. References
Fig. 6. Final total ammonia nitrogen (TAN) concentration and anolyte conductivity of MREC following one fed-batch cycle before and after the addition of an extra low concentration chamber (LCC) (5 + LCC: the extra LCC was added to the stack consisting of 5 cell pairs). Initial TAN concentration in anolyte was about 74 mg/L.
Call, D., Logan, B.E., 2008. Hydrogen production in a single chamber microbial electrolysis cell lacking a membrane. Environ. Sci. Technol. 42, 3401–3406. Cheng, S., Logan, B.E., 2011. High hydrogen production rate of microbial electrolysis cell (MEC) with reduced electrode spacing. Bioresour. Technol. 102, 3571–3574. Cheng, S., Logan, B.E., 2007. Sustainable and efficient biohydrogen production via electrohydrogenesis. Proc. Natl. Acad. Sci. USA 104, 18871–18873. Cusick, R.D., Kim, Y., Logan, B.E., 2012. Energy capture from thermolytic solutions in microbial reverse-electrodialysis cells. Science 335, 1474–1477. Długołe˛cki, P., Da˛browska, J., Nijmeijer, K., Wessling, M., 2010. Ion conductive spacers for increased power generation in reverse electrodialysis. J. Membr. Sci. 347, 101–107. Długołe˛cki, P., Gambier, A., Nijmeijer, K., Wessling, M., 2009. Practical potential of reverse electrodialysis as process for sustainable energy generation. Environ. Sci. Technol. 43, 6888–6894. Elimelech, M., Phillip, W.A., 2011. The future of seawater desalination: energy, technology, and the environment. Science 333, 712–717. Kim, Y., Logan, B.E., 2011a. Hydrogen production from inexhaustible supplies of fresh and salt water using microbial reverse-electrodialysis electrolysis cells. Proc. Natl. Acad. Sci. USA 108, 16176–16181. Kim, Y., Logan, B.E., 2011b. Microbial reverse electrodialysis cells for synergistically enhanced power production. Environ. Sci. Technol. 45, 5834–5839.
X. Luo et al. / Bioresource Technology 140 (2013) 399–405 Logan, B.E., Call, D., Cheng, S., Hamelers, H.V.M., Sleutels, T.H.J.A., Jeremiasse, A.W., Rozendal, R.A., 2008. Microbial electrolysis cells for high yield hydrogen gas production from organic matter. Environ. Sci. Technol. 42, 8630–8640. Logan, B.E., Rabaey, K., 2012. Conversion of wastes into bioelectricity and chemicals by using microbial electrochemical technologies. Science 337, 686–690. Luo, X., Cao, X., Mo, Y., Xiao, K., Zhang, X., Liang, P., Huang, X., 2012. Power generation by coupling reverse electrodialysis and ammonium bicarbonate: implication for recovery of waste heat. Electrochem. Commun. 19, 25–28. McGinnis, R.L., Elimelech, M., 2007. Energy requirements of ammonia–carbon dioxide forward osmosis desalination. Desalination 207, 370–382. McGinnis, R.L., McCutcheon, J.R., Elimelech, M., 2007. A novel ammonia–carbon dioxide osmotic heat engine for power generation. J. Membr. Sci. 305, 13–19. Mehanna, M., Kiely, P.D., Call, D.F., Logan, B.E., 2010. Microbial electrodialysis cell for simultaneous water desalination and hydrogen gas production. Environ. Sci. Technol. 44, 9578–9583. Nam, J.-Y., Cusick, R.D., Kim, Y., Logan, B.E., 2012. Hydrogen generation in microbial reverse-electrodialysis electrolysis cells using a heat-regenerated salt solution. Environ. Sci. Technol. 46, 5240–5246. Nam, J.-Y., Kim, H.-W., Shin, H.-S., 2010. Ammonia inhibition of electricity generation in single-chambered microbial fuel cells. J. Power Sources 195, 6428–6433. Post, J.W., Goeting, C.H., Valk, J., Goinga, S., Veerman, J., Hamelers, H.V.M., Hack, P.J.F.M., 2010. Towards implementation of reverse electrodialysis for power generation from salinity gradients. Desalin. Water Treat. 16, 182–193.
405
Post, J.W., Hamelers, H.V.M., Buisman, C.J.N., 2008. Energy recovery from controlled mixing salt and fresh water with a reverse electrodialysis system. Environ. Sci. Technol. 42, 5785–5790. Ramon, G.Z., Feinberg, B.J., Hoek, E.M.V., 2011. Membrane-based production of salinity-gradient power. Energy Environ. Sci. 4, 4423–4434. Turek, M., Bandura, B., 2007. Renewable energy by reverse electrodialysis. Desalination 205, 67–74. Veerman, J., de Jong, R.M., Saakes, M., Metz, S.J., Harmsen, G.J., 2009. Reverse electrodialysis: comparison of six commercial membrane pairs on the thermodynamic efficiency and power density. J. Membr. Sci. 343, 7–15. Veerman, J., Post, J.W., Saakes, M., Metz, S.J., Harmsen, G.J., 2008. Reducing power losses caused by ionic shortcut currents in reverse electrodialysis stacks by a validated model. J. Membr. Sci. 310, 418–430. Veerman, J., Saakes, M., Metz, S.J., Harmsen, G.J., 2010. Electrical power from sea and river water by reverse electrodialysis: a first step from the laboratory to a real power plant. Environ. Sci. Technol. 44, 9207–9212. Vermaas, D.A., Kunteng, D., Saakes, M., Nijmeijer, K., 2013. Fouling in reverse electrodialysis under natural conditions. Water Res. 47, 1289–1298. Vermaas, D.A., Saakes, M., Nijmeijer, K., 2011a. Doubled power density from salinity gradients at reduced intermembrane distance. Environ. Sci. Technol. 45, 7089– 7095. Vermaas, D.A., Saakes, M., Nijmeijer, K., 2011b. Power generation using profiled membranes in reverse electrodialysis. J. Membr. Sci. 385–386, 234–242.
Contents lists available at SciVerse ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Optimization of membrane stack configuration for efficient hydrogen production in microbial reverse-electrodialysis electrolysis cells coupled with thermolytic solutions Xi Luo a, Joo-Youn Nam b, Fang Zhang c, Xiaoyuan Zhang c, Peng Liang a, Xia Huang a,⇑, Bruce E. Logan a,c,⇑ a b c
State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, PR China Jeju Global Research Center, Korea Institute of Energy Research, 200 Haemajihaean-ro, Gujwa-eup, Jeju 695-971, Republic of Korea Department of Civil and Environmental Engineering, Penn State University, 212 Sackett Building, University Park, PA 16802, USA
h i g h l i g h t s Membrane stack configuration was optimized for the MREC utilizing NH4HCO3 solutions. The optimum number of cell pairs was determined to be five. Increasing the number of cell pairs did not appreciably affect anode performance. Adding an LC chamber reduced ammonia crossover and improved hydrogen production.
a r t i c l e
i n f o
Article history: Received 25 March 2013 Received in revised form 22 April 2013 Accepted 25 April 2013 Available online 3 May 2013 Keywords: Microbial reverse-electrodialysis electrolysis cell Ammonium bicarbonate Hydrogen Configuration Optimization
a b s t r a c t Waste heat can be captured as electrical energy to drive hydrogen evolution in microbial reverse-electrodialysis electrolysis cells (MRECs) by using thermolytic solutions such as ammonium bicarbonate. To determine the optimal membrane stack configuration for efficient hydrogen production in MRECs using ammonium bicarbonate solutions, different numbers of cell pairs and stack arrangements were tested. The optimum number of cell pairs was determined to be five based on MREC performance and a desire to minimize capital costs. The stack arrangement was altered by placing an extra low concentration chamber adjacent to anode chamber to reduce ammonia crossover. This additional chamber decreased ammonia nitrogen losses into anolyte by 60%, increased the coulombic efficiency to 83%, and improved the hydrogen yield to a maximum of 3.5 mol H2/mol acetate, with an overall energy efficiency of 27%. These results improve the MREC process, making it a more efficient method for renewable hydrogen gas production. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction A microbial electrolysis cell (MEC) is a bio-electrochemical system which can achieve hydrogen production from various types of renewable biomass (Cheng and Logan, 2007). An applied voltage (>0.3 V in practice) is required to overcome the thermodynamic limit for hydrogen evolution at the cathode (Logan and Rabaey, 2012). To eliminate the need for electrical grid energy, a sustainable method for hydrogen production was recently proposed based on integrating a small reverse electrodialysis (RED) stack into the MEC, which was called a microbial reverse-electrodialysis electrol⇑ Corresponding authors. Address: Department of Civil and Environmental Engineering, Penn State University, 212 Sackett Building, University Park, PA 16802, USA. Tel.: +1 814 863 7908; fax: +1 814 863 7304 (B.E. Logan), tel.: +86 10 62772324; fax: +86 10 62771472 (X. Huang). E-mail addresses: [email protected] (X. Huang), [email protected] (B.E. Logan). 0960-8524/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2013.04.097
ysis cell (MREC) (Kim and Logan, 2011a). A RED stack comprises an alternating series of anion (AEMs) and cation exchange membranes (CEMs) typically separated by porous spacers (Veerman et al., 2009; Vermaas et al., 2011a). When high concentration (HC) and low concentration (LC) solutions flow through alternating chambers in the stack, cations and anions in HC chamber migrate into the LC chamber through the membranes with opposing-charge functional groups due to concentration gradient, resulting in a potential difference across the membranes (Długołe˛cki et al., 2010; Vermaas et al., 2011b, 2013). Thus, renewable salinity gradient energy is converted to electrical energy to drive hydrogen evolution in an MREC, without the need for an external power source. The use of RED stacks in MRECs can be limited to estuaries or coastal areas when river water and seawater are used for the HC and LC solutions. The use of these natural waters also requires substantial and energy intensive pre-treatment to minimize membrane fouling (McGinnis et al., 2007). To avoid these limitations,
400
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
thermolytic solutions (Elimelech and Phillip, 2011) such as ammonium bicarbonate (NH4HCO3), have been proposed as the source of the salinity gradient energy for RED stacks (Cusick et al., 2012; Luo et al., 2012; Nam et al., 2012). With a thermolytic solution the ionic species can be volatilized and captured into HC solutions at temperatures below that needed to boil water. Heating ammonium bicarbonate solutions at 60 °C and 1 atm, for example, volatilizes ammonia and carbon dioxide, which can be condensed to form the HC solution (McGinnis and Elimelech, 2007). NH4HCO3 solutions have been shown to work in MRECs (Nam et al., 2012), and the HC and LC solutions can be regenerated using waste heat and conventional distillation technologies. Thus, waste heat can be captured as electrical energy, enabling hydrogen evolution in an MREC. The RED stack configuration and performance is critical to hydrogen production in an MREC. The number of cell pairs used can affect the electrochemical potential available to drive hydrogen production. Increasing cell pairs should improve the potential difference across the membrane stack (Długołe˛cki et al., 2009; Post et al., 2008), but adding cell pairs can increase the internal resistance through the addition of extra HC and LC chambers (Veerman et al., 2008, 2010). In addition, the use of more cell pairs can increase capital costs as the cost of the stack is dominated by cost of the ion exchange membranes (Post et al., 2010; Ramon et al., 2011; Turek and Bandura, 2007). Thus, it is essential to ascertain an optimum number of cell pairs for balancing cost and MREC performance. In addition, the application of higher potentials could adversely affect electrochemical performance by changing the anode potentials to values unfavorable for microbial oxidation of substrate. In this study, different numbers of cell pairs and stack arrangements were tested to determine the optimum membrane stack configuration for an MREC utilizing NH4HCO3 solutions. The number of cell pairs was optimized based on measuring current and hydrogen production rates, and calculating energy recoveries and efficiencies. Anode, cathode and stack performance were evaluated by galvanostatic polarization during variation of numbers of cell pairs. The stack arrangement was also changed to minimize ammonia crossover into the anode chamber. Previous studies used an AEM in the stack adjacent to the anode chamber (Kim and Logan, 2011a; Nam et al., 2012). This configuration minimized internal resistance by having a HC chamber adjacent to the anode chamber, but it resulted in high rates of ammonia transfer into the anolyte. In one series of tests, as much as 540 mg/L of total ammonia nitrogen (TAN) was transferred into the anode chamber over a single cycle (initial conductivity of the HC NH4HCO3 solution of 103 mS/cm) (Nam et al., 2012), which could inhibit current generation by the anode microorganisms (Nam et al., 2010) and therefore hydrogen gas production. To avoid this situation, the effect of adding an LC chamber adjacent to the anode chamber was examined to reduce TAN crossover.
catalyst layer (5 mg/cm2 10% Pt on carbon black) on the side facing membrane stack, and carbon black layer (5 mg/cm2) on the other side. A RED stack with up to 7 cell pairs was sandwiched between the anode and cathode chambers (Fig. 1). One cell pair consisted of a pair of HC and LC chambers, and a pair of AEM and CEM (Selemion AMV and CMV, Asashi glass, Japan). Except as noted otherwise, one additional AEM was used to close the last chamber at the end of the stack next to the electrode. The effective area of each membrane was 8 cm2 (4 cm 2 cm). Membranes were separated by polyethylene woven spacers and silicone gaskets with a thickness of 1.3 mm. 2.2. Solutions The anolyte was 1 g/L sodium acetate in a nutrient buffer solution containing 8.4 g/L NaHCO3, 0.31 g/L NH4Cl, 0.13 g/L KCl, 0.05 g/L Na2HPO4, 0.03 g/L NaH2PO4H2O, trace elements and minerals. A 1 M NaHCO3 solution was used as catholyte (Nam et al., 2012). Based on the solubility of NH4HCO3 at room temperature (approx. 2 M), a NH4HCO3 solution of 1.7 M was used (conductivity of 103 mS/cm) as the HC solution. The LC solution was prepared by dilution of the HC solution with distilled water to produce a salinity ratio of 75 (Luo et al., 2012).
2. Methods 2.1. MREC construction The MREC was composed of an anode chamber, a cathode chamber, and a RED membrane stack (Fig. 1). Two cubic Lexan blocks with a cylindrical cavity were used as the anode and cathode chamber (30 mL liquid volume each). A glass tube was glued to the top of cathode chamber for hydrogen collection (Mehanna et al., 2010). The anode was a heat treated graphite fiber brush (25 mm diameter 25 mm length; fiber type: PANEX 33 160 K, ZOLTEK). The cathode was stainless steel mesh (projected area: 7 cm2; Type 304, #60 mesh, McMaster-Carr) coated with platinum
Fig. 1. (a) Schematic and (b) photograph of microbial reverse-electrodialysis electrolysis cell (CEM, cation exchange membrane; AEM, anion exchange membrane; HCin, high concentration solution inlet; HCout, high concentration solution outlet; LCin, low concentration solution inlet; LCout, low concentration solution outlet).
401
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
2.3. MREC operation Brush anodes were inoculated with the effluent from an existing microbial fuel cell (MFC) and pre-acclimated in a single chamber air–cathode MFC. After reproducible peak voltages were obtained for at least three successive batch cycles, anodes were transferred to the MREC. The MREC was operated in fed-batch mode. Anolyte and catholyte were replaced when the produced current decreased to less than 0.5 mA, forming one complete cycle. HC and LC solutions were continuously fed into the stack at a fixed flow rate of 0.8 mL/min during one batch cycle. In order to investigate the influence of number of cell pairs on the performance of MREC, the number of cell pairs was varied from 1 to 7. In these tests, LC solution entered the membrane stack from the bottom of the LC chamber near the anode and flowed serially through the other LC chambers, exiting from the top of the chamber next to cathode chamber (Fig. 1). Similarly, the HC solution flowed serially through the HC chambers, but in the opposite direction (from cathode to anode side). In order to reduce transfer of ammonia nitrogen into the anolyte, in some experiments (as indicated) a CEM was added between the stack (containing 5 cell pairs) and the anode chamber, creating an extra LC chamber. Under this condition, LC solution entered the stack from the top of this extra chamber while the flow path of HC solution remained unchanged. All experiments were carried out at 30 °C in a constant temperature room. 2.4. Analyses Two Ag/AgCl reference electrodes (RE-5B, BASi) were separately inserted into the anode and cathode chambers to measure electrode potentials and stack voltages. The anode and cathode were connected to a 10 X external resistor to measure the current based on the voltage drop using a multimeter (Keitheley Instruments, OH, USA). Galvanostatic polarization was carried out with a potentiostat (VMP3, BioLogic, Claix, France) to determine the internal resistance of the MREC. Prior to the tests, the MREC was held under open circuit conditions for three hours. The anode was used as the working electrode while the cathode was used as the counter and reference electrode. Diverse current steps were set for different number of cell pairs according to the current production range in the fedbatch cycle: 2 pairs: 0.4, 0.5, 0.6, 0.7, 0.8, 1 and 1.2 mA; 3 pairs: 0.3, 0.6, 0.9, 1.2, 1.5, 1.8 and 2.1 mA; 5 pairs: 0.5, 0.8, 1, 1.5, 2, 2.5 and 3 mA; 7 pairs: 0.5, 1, 1.5, 2, 2.5, 3 and 3.5 mA. The current was set for 50 min for the first point, and at 20 min intervals thereafter. In order to evaluate the performance of anode, cathode and stack, electrode potentials and stack voltage were also measured using the two Ag/AgCl reference electrodes in anode and cathode chambers during polarization tests. Anode, cathode, and stack resistances were determined from the slopes of the polarization data measured for each of the electrodes, and from the stack polarization data. A respirometer (AER-200, Challenge Environmental) was used to measure the produced gas volume. Gas was collected for analysis in a gas bag (100 mL capacity; Cali-5-Bond, Calibrated Instruments Inc.). The compositions of cathode headspace and the gas bag were determined using gas chromatographs (models 8610B and 310, SRI Instruments, CA, USA) as previously described (Cheng and Logan, 2011). Total chemical oxygen demand (COD) was measured according to standard methods (COD Reagent, HACH Company). TAN concentrations in the anolyte were determined before and after each cycle (nitrogen ammonia reagent (salicylate), HACH Company). Conductivity and pH of anolyte, catholyte, HC and LC solutions were measured using a conductivity-pH meter (SevenMulti, Mettler Toledo, OH, USA). The conductivity of HC and LC
solutions was converted into concentration through a second order calibration curve. Coulombic efficiency (gCE, %), cathodic hydrogen recovery (rcat, %), hydrogen yield (Y, mol H2/mol acetate) and maximum volumetric hydrogen production rate (Q, m3 H2/m3 anolyte/d) were determined as previously described (Kim and Logan, 2011a). Energy recovery (rE) was the ratio of the combustion energy of produced hydrogen to the total energy provided to the MREC (substrate energy and salinity gradient energy), calculated (Nam et al., 2012) as:
rE ¼
DHh nh
ð1Þ
in DHs nin s þX
where nh and nin s are the amounts of produced hydrogen and supplied substrate respectively (mol), DHh and DHs the heat of combustion for hydrogen and substrate respectively (J/mol), Xin the free energy released from the total mixing of the influent HC and LC solutions (J). Energy efficiency (gE) was defined as the combustion energy of produced hydrogen divided by the input energy extracted by MREC (extracted substrate energy and extracted salinity gradient energy), and calculated as:
gE ¼
DH h n h
ð2Þ
in out out DHs ðnin s ns Þ þ X X
where nout is the amount of remaining substrate at the end of a s batch cycle (mol), Xout the free energy released from the total mixing of the effluent HC and LC solutions (J). Xin was determined (Cusick et al., 2012) using in
X ¼ RT
X i
V H C in i;H
ln
ain i;H ain i;m
þ
V L C in i;L
ln
ain i;L
!
ain i;m
ð3Þ
where R is the gas constant [J/(mol K)], T the absolute temperature (K), V the volume of solution, C the concentration of ionic species i in the solution, a the activity of ionic species i. The subscripts indicate H = HC, L = LC, and m is the mixture of the two solutions. The superscript in indicates influent. Xout was defined in an analogous manner to Xin by changing the concentrations and activities of influents into those of effluents in Eq. (3). Concentrations and activities of the ionic species in NH4HCO3 solution were estimated by OLI Stream Analysis software (OLI Systems Inc., Morris Plains, NJ) as previously described (Cusick et al., 2012). 3. Results and discussion 3.1. Current generation and hydrogen production using different numbers of cell pairs Increasing the number of cell pairs from 1 to 7 improved current production (Fig. 2a). Peak currents were observed at the beginning of the fed-batch cycle, and the currents decreased over time due to the substrate depletion. The peak current for 7 cell pairs (4 mA) was nearly thirteen times that of 1 cell pair (0.3 mA). This increase in current with the use of more cell pairs was mainly attributed to the sharper increase of stack open circuit voltage (OCV) than changes in the total internal resistance (see below). Peak current increased by only 29% with an increase in cell pairs from 5 to 7, compared to a 77% improvement going from 2 to 3 cell pairs, and a 35% improvement going from 3 to 5 cell pairs. This smaller enhancement with more cell pairs suggested that further addition of cell pairs would not substantially improve the current. When 1 cell pair was used, there was essentially no current production as the peak current of the MREC was lower than 0.5 mA, which was defined as the endpoint of a fed-batch cycle.
402
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
higher stack voltage. Anode and cathode potentials were essentially unchanged by using different numbers of cell pairs in polarization tests (Fig. 3b), demonstrating that the changes in performance were due to solely the RED stack. The maximum current density in polarization tests increased from 0.17 mA/cm2 (2 cell pairs) to 0.5 mA/cm2 (7 cell pairs) (Fig. 3a and b). Anode, cathode, and stack resistances were estimated from the slopes of the polarization data given in Fig. 3. The stack resistance predominated, and it increased with the use of more cell pairs, accounting for 63% (2 cell pairs) to 88% (7 cell pairs) of the total internal resistance (Fig. 4). It was expected that adding cell pairs would produce larger membrane and ohmic resistances of the salt solutions, and increase stack resistance (Veerman et al., 2008, 2010). Compared with stack resistance, cathode or anode resistances only decreased slightly with the increase of cell pairs, illustrating that the internal resistance change of MREC was mainly caused by the variation in stack resistances. The addition of cell pairs brought about a sharper increase in stack OCV than the total internal resistance (Fig. 4). For example, the total internal resistance increased by only 27% using 5 cell pairs (145 X) to 7 cell pairs (185 X), compared to a 54% improvement in the stack OCV. This sharper increase of stack OCV therefore primarily explains why adding cell pairs improved the current generation for MREC.
Fig. 2. (a) Current generation and (b) hydrogen volume and maximum volumetric hydrogen production rate (Q) of MREC using different number of cell pairs.
As an increase in current results in faster oxidation of substrate, the cycle time of the MREC decreased from 36 h (2 cell pairs) to 22.5 h (7 cell pairs) (Fig. 2a). Total coulombs produced in one fed-batch cycle increased gradually from 103 C (2 cell pairs) to 192 C (5 cell pairs), but it did not further increase with additional cell pairs. A larger hydrogen gas volume was obtained by adding more cell pairs (Fig. 2b). When the number of cell pairs was larger than 5, the volume of hydrogen produced stabilized at about 27 mL (Y = 2.9 mol H2/mol acetate), or 0.9 L H2/L anolyte volume. Little hydrogen (0.7 mL) was generated using 1 cell pair, again indicating a single cell pair was insufficient for hydrogen evolution in an MREC. The variation of hydrogen volumes recovered was consistent with that of the total coulombs produced in one cycle as a function of the number of cells pairs. This is expected due to the proportional relationship between hydrogen volume and total coulombs (Call and Logan, 2008), and the high cathodic hydrogen recoveries (around 100%) obtained here. As a result, the volume of hydrogen gas recovered was unchanged when 5 or more cell pairs were used. Increasing the number of cell pairs improved the maximum volumetric hydrogen production rate (Q) (Fig. 2b). As Q mainly depends on the peak current (Logan et al., 2008), it was expected that more cell pairs would result in larger peak currents, and improve Q. 3.2. Galvanostatic polarization tests Adding cell pairs increased the stack voltages and slopes of the stack polarization curve (Fig. 3a). More cell pairs created larger potential differences across the membrane stack, producing the
Fig. 3. (a) RED stack voltage and (b) anode (A) and cathode (C) potentials vs. current density for the MREC using different number of cell pairs. Electrode potentials were reported vs. Ag/AgCl reference electrode (+211 mV vs. standard hydrogen electrode).
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
403
3.3. Energy recovery, energy efficiency, coulombic efficiency and COD removal The energy that could be recovered based on the hydrogen gas produced (Wh) increased gradually with the addition of cell pairs and finally reached a plateau of 309 J (Fig. 5a). The substrate energy extracted using the MREC (Ws) was essentially unchanged by varying the number of cell pairs, and it stabilized at about 326 J, which was mainly due to the similar COD removals in the different experiments (see below). The salinity gradient energy extracted using the RED stack (WSGE) decreased slightly when the number of cell pairs rose from 2 to 3, and increased gradually with the further addition of cell pairs (Fig. 5a). More cell pairs resulted in not only the improved current generation for MREC, but also more extensive co-ion transport due to the longer hydraulic retention time, resulting in larger concentration changes in the HC and LC solutions after flowing through the membrane stack. The conductivity of effluent LC solution increased from 2.7 mS/cm (2 cell pairs) to 4.6 mS/cm (7 cell pairs) while that of effluent HC solution decreased from 99 mS/ cm (2 cell pairs) to 96 mS/cm (7 cell pairs). The larger concentration change signified a higher degree of mixing between HC and LC solutions, implying more salinity gradient energy was consumed in the MREC per unit time (Kim and Logan, 2011a, 2011b). Thus, raising the number of cell pairs from 3 to 7 led to the increase of WSGE due to similar cycle times. As the cycle time for 2 cell pairs (36 h) was much longer, the WSGE under this condition was slightly larger than that obtained with 3 cell pairs. A greater amount of energy was recovered (energy recovered versus total applied) when using more cell pairs (Fig. 5b). This was due to the larger Wh, as well as the shorter cycle time which decreased the total salinity gradient energy input during one fedbatch cycle when adding cell pairs. Energy recovery overall, however, remained very low (2–5%), mainly due to the large amounts of salinity gradient energy provided (5335–8751 J), which was more than 93% of the total energy supplied to the MREC. The greatest energy efficiency (recovered relative that entering and leaving) of 27% was achieved using 5 cell pairs (Fig. 5b). This efficiency is comparable to that of traditional RED systems (10–35% when ignoring power losses due to electrode reactions) using NaCl for the HC and LC solutions (Veerman et al., 2009). When the number of cell pairs decreased from 5 to 2, the reduction of Wh was more significant than WSGE, resulting in a lower energy efficiency. Comparing 5 and 7 cell pairs, the WSGE for 7 cell pairs was larger while Wh remained nearly unchanged, resulting in a smaller energy efficiency for 7 cell pairs.
Fig. 5. (a) Extracted input energy (Ws, extracted substrate energy; WSGE, extracted salinity gradient energy) and output energy (Wh, combustion energy of hydrogen), (b) energy recovery (rE), energy efficiency (gE), coulombic efficiency (gCE) and COD removal of MREC at different number of cell pairs.
Coulombic efficiency substantially improved with more cell pairs, increasing from 36% for 2 cell pairs to 66% for 5 cell pairs, but it remained nearly constant with additional cell pairs (Fig. 5b). More cell pairs resulted in shorter cycle times, increasing the coulombic efficiencies. Shorter cycle times are produced by higher current densities, resulting in less oxygen leaking into the anode chamber and thus less loss of substrate by aerobic processes (Kim and Logan, 2011a,b). COD removal was not affected by the number of cell pairs, ranging from 92% to 94% in all of the tests (Fig. 5b). This high COD removal mainly resulted from the neutral anolyte pH (final pHs of 7.3–7.6), which was maintained by the use of high concentration bicarbonate buffer (100 mM) and the flow of bicarbonate ions from the adjacent HC chamber into anolyte. Taken together, these results suggest that 5 cell pairs represent the most useful compromise in performance and minimization of costs. The maximum hydrogen volume and energy recoveries were obtained with 5 or 7 cell pairs. The highest energy efficiency was achieved using 5 cell pairs, but the largest peak current was obtained for 7 cell pairs. Considering the capital costs for additional membranes, 5 cell pairs provided the best overall stack architecture. 3.4. Ammonia nitrogen crossover into anolyte without an additional LC chamber
Fig. 4. Effect of the number of cell pairs on internal resistance of MREC and stack open circuit voltage (OCV).
Transport of ammonia nitrogen into the anolyte occurred in all the above tests. The TAN concentration in the anolyte of MREC
404
X. Luo et al. / Bioresource Technology 140 (2013) 399–405
ranged from 450 mg/L to 800 mg/L after one fed-batch cycle (Fig. 6). The main nitrogen forms in HC solution are ammonium ðNHþ 4 Þ, ammonia (NH3), and carbamate ðNH2 CO2 Þ (Cusick et al., þ 2012). Negatively charged carbamate and NH4 , due to co-ion transport, migrated from the HC solution and through the AEM adjacent into the anode chamber. The MREC with 2 cell pairs had the highest TAN concentration (789 mg/L) due to low current densities and the longest cycle times. The MREC with 7 cell pairs had the shortest cycle time, and thus the lowest TAN concentration. Although the cycle time for 5 cell pairs was shorter than that for 3 cell pairs, the final TAN concentration in anolyte for 5 cell pairs was higher, probably due to the transfer of more carbamate into the anolyte resulting from the higher current production for 5 cell pairs than that for 3 cell pairs. The anolyte conductivity increased from 9.2 mS/cm to 12– 14 mS/cm after one cycle (Fig. 6), mainly due to the ionic flux from HC solution into the anode chamber. The variation of final anolyte conductivity was consistent with that of the final TAN concentration, suggesting anolyte conductivity change was dominated by the ammonia nitrogen migration.
3.5. Reducing ammonia nitrogen crossover by adding an LC chamber To reduce the migration of ammonia nitrogen into anolyte, an extra LC chamber was added between the anode chamber and the RED stack with 5 cell pairs. As a result of this change, there were significant decreases in both ammonia nitrogen crossover (60%) and final anolyte conductivity (66%) (Fig. 6). The introduction of this additional LC chamber increased the cationic flux from the anolyte to the LC solution, decreasing the anolyte conductivity from 9.2 mS/cm to 4.5 mS/cm after one cycle. In addition, only a small quantity of NHþ 4 was transferred into the anolyte, with a final TAN concentration of 311 mg/L. The presence of the extra LC chamber increased the cycle time (from 23 h to 29 h) but did not appreciably alter the peak current (Fig. S1a). There was an increase of 26% of the total coulombs produced, which resulted from the improvement of coulombic efficiency (see below). The hydrogen gas yield increased (32 mL, Y = 3.5 mol H2/mol acetate) due to the increased coulombic recovery (Fig. S1b). The maximum volumetric hydrogen production rate was not affected by adding the LC chamber (Fig. S1b), mainly due to the constant peak current. Energy recovery, energy efficiency and COD removal remained nearly constant with the additional LC chamber, while coulombic
efficiency increased appreciably from 66% to 83% (Fig. S1c). The higher coulombic efficiency was most likely due to the low TAN concentration in the anolyte, which would have avoided ammonia inhibition of exoelectrogenic microorganisms. Adding the LC chamber prolonged the cycle time, which increased both the WSGE (from 813 J to 1127 J) and total salinity gradient energy input (from 5587 J to 7033 J), thus resulting in the nearly unchanged energy recovery and efficiency. Anode potentials or cathode potentials remained totally the same in polarization tests (Fig. S1d), demonstrating stable anode and cathode performance. Consequently, nearly identical anode and cathode resistances were observed (Fig. S1e). The existence of the extra LC chamber resulted in similar increase for stack OCV (by 7%) and the total internal resistance (by 8%) (Fig. S1e), explaining why the peak current remained nearly constant. These results demonstrate that adding an LC chamber was an effective strategy to reduce the migration of ammonia nitrogen into anolyte and improve the hydrogen production performance for MREC. 4. Conclusions Membrane stack configuration was optimized to achieve a maximum hydrogen yield of 3.5 mol H2/mol acetate, energy efficiency of 27%, and coulombic efficiency of 83% using the MREC. Based on performance and a desire to minimize membrane total area, the optimum number of cell pairs was five. Varying the number of cell pairs did not appreciably affect anode and cathode performance. Adding an LC chamber adjacent to anode chamber greatly reduced ammonia crossover into anolyte, increased coulombic efficiency and further improved hydrogen production. Thus, an extra LC chamber is recommended when using NH4HCO3 solutions in MREC applications. Acknowledgements This research was supported by Award KUS-I1-003-13 from the King Abdullah University of Science and Technology (KAUST), the National High Technology Research and Development Program of China (863 Program) (No. 2011AA060907) and a scholarship from the China Scholarship Council (CSC). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.biortech. 2013.04.097. References
Fig. 6. Final total ammonia nitrogen (TAN) concentration and anolyte conductivity of MREC following one fed-batch cycle before and after the addition of an extra low concentration chamber (LCC) (5 + LCC: the extra LCC was added to the stack consisting of 5 cell pairs). Initial TAN concentration in anolyte was about 74 mg/L.
Call, D., Logan, B.E., 2008. Hydrogen production in a single chamber microbial electrolysis cell lacking a membrane. Environ. Sci. Technol. 42, 3401–3406. Cheng, S., Logan, B.E., 2011. High hydrogen production rate of microbial electrolysis cell (MEC) with reduced electrode spacing. Bioresour. Technol. 102, 3571–3574. Cheng, S., Logan, B.E., 2007. Sustainable and efficient biohydrogen production via electrohydrogenesis. Proc. Natl. Acad. Sci. USA 104, 18871–18873. Cusick, R.D., Kim, Y., Logan, B.E., 2012. Energy capture from thermolytic solutions in microbial reverse-electrodialysis cells. Science 335, 1474–1477. Długołe˛cki, P., Da˛browska, J., Nijmeijer, K., Wessling, M., 2010. Ion conductive spacers for increased power generation in reverse electrodialysis. J. Membr. Sci. 347, 101–107. Długołe˛cki, P., Gambier, A., Nijmeijer, K., Wessling, M., 2009. Practical potential of reverse electrodialysis as process for sustainable energy generation. Environ. Sci. Technol. 43, 6888–6894. Elimelech, M., Phillip, W.A., 2011. The future of seawater desalination: energy, technology, and the environment. Science 333, 712–717. Kim, Y., Logan, B.E., 2011a. Hydrogen production from inexhaustible supplies of fresh and salt water using microbial reverse-electrodialysis electrolysis cells. Proc. Natl. Acad. Sci. USA 108, 16176–16181. Kim, Y., Logan, B.E., 2011b. Microbial reverse electrodialysis cells for synergistically enhanced power production. Environ. Sci. Technol. 45, 5834–5839.
X. Luo et al. / Bioresource Technology 140 (2013) 399–405 Logan, B.E., Call, D., Cheng, S., Hamelers, H.V.M., Sleutels, T.H.J.A., Jeremiasse, A.W., Rozendal, R.A., 2008. Microbial electrolysis cells for high yield hydrogen gas production from organic matter. Environ. Sci. Technol. 42, 8630–8640. Logan, B.E., Rabaey, K., 2012. Conversion of wastes into bioelectricity and chemicals by using microbial electrochemical technologies. Science 337, 686–690. Luo, X., Cao, X., Mo, Y., Xiao, K., Zhang, X., Liang, P., Huang, X., 2012. Power generation by coupling reverse electrodialysis and ammonium bicarbonate: implication for recovery of waste heat. Electrochem. Commun. 19, 25–28. McGinnis, R.L., Elimelech, M., 2007. Energy requirements of ammonia–carbon dioxide forward osmosis desalination. Desalination 207, 370–382. McGinnis, R.L., McCutcheon, J.R., Elimelech, M., 2007. A novel ammonia–carbon dioxide osmotic heat engine for power generation. J. Membr. Sci. 305, 13–19. Mehanna, M., Kiely, P.D., Call, D.F., Logan, B.E., 2010. Microbial electrodialysis cell for simultaneous water desalination and hydrogen gas production. Environ. Sci. Technol. 44, 9578–9583. Nam, J.-Y., Cusick, R.D., Kim, Y., Logan, B.E., 2012. Hydrogen generation in microbial reverse-electrodialysis electrolysis cells using a heat-regenerated salt solution. Environ. Sci. Technol. 46, 5240–5246. Nam, J.-Y., Kim, H.-W., Shin, H.-S., 2010. Ammonia inhibition of electricity generation in single-chambered microbial fuel cells. J. Power Sources 195, 6428–6433. Post, J.W., Goeting, C.H., Valk, J., Goinga, S., Veerman, J., Hamelers, H.V.M., Hack, P.J.F.M., 2010. Towards implementation of reverse electrodialysis for power generation from salinity gradients. Desalin. Water Treat. 16, 182–193.
405
Post, J.W., Hamelers, H.V.M., Buisman, C.J.N., 2008. Energy recovery from controlled mixing salt and fresh water with a reverse electrodialysis system. Environ. Sci. Technol. 42, 5785–5790. Ramon, G.Z., Feinberg, B.J., Hoek, E.M.V., 2011. Membrane-based production of salinity-gradient power. Energy Environ. Sci. 4, 4423–4434. Turek, M., Bandura, B., 2007. Renewable energy by reverse electrodialysis. Desalination 205, 67–74. Veerman, J., de Jong, R.M., Saakes, M., Metz, S.J., Harmsen, G.J., 2009. Reverse electrodialysis: comparison of six commercial membrane pairs on the thermodynamic efficiency and power density. J. Membr. Sci. 343, 7–15. Veerman, J., Post, J.W., Saakes, M., Metz, S.J., Harmsen, G.J., 2008. Reducing power losses caused by ionic shortcut currents in reverse electrodialysis stacks by a validated model. J. Membr. Sci. 310, 418–430. Veerman, J., Saakes, M., Metz, S.J., Harmsen, G.J., 2010. Electrical power from sea and river water by reverse electrodialysis: a first step from the laboratory to a real power plant. Environ. Sci. Technol. 44, 9207–9212. Vermaas, D.A., Kunteng, D., Saakes, M., Nijmeijer, K., 2013. Fouling in reverse electrodialysis under natural conditions. Water Res. 47, 1289–1298. Vermaas, D.A., Saakes, M., Nijmeijer, K., 2011a. Doubled power density from salinity gradients at reduced intermembrane distance. Environ. Sci. Technol. 45, 7089– 7095. Vermaas, D.A., Saakes, M., Nijmeijer, K., 2011b. Power generation using profiled membranes in reverse electrodialysis. J. Membr. Sci. 385–386, 234–242.
