Enzyme and Microbial Technology 60 (2014) 56–63
Contents lists available at ScienceDirect
Enzyme and Microbial Technology journal homepage: www.elsevier.com/locate/emt
Membrane filtration biocathode microbial fuel cell for nitrogen removal and electricity generation Guangyi Zhang a , Hanmin Zhang a,∗ , Yanjie Ma b , Guangen Yuan a , Fenglin Yang a , Rong Zhang a a Key Laboratory of Industrial Ecology and Environmental Engineering (MOE), School of Environmental Science and Technology, Dalian University of Technology, Linggong Road 2, Dalian 116024, China b Environmental Management College of China, Hebei Road 73, Qinhuangdao 066004, China
a r t i c l e
i n f o
Article history: Received 17 February 2014 Received in revised form 1 April 2014 Accepted 7 April 2014 Available online 18 April 2014 Keywords: Microbial fuel cell Membrane filtration biocathode Bioelectrochemical denitrification Carbon felt Stainless steel mesh
a b s t r a c t Conductive materials with attached biofilms, were used as membrane filtration biocathodes to filter the effluent and supply electrons for denitrification. Stainless steel mesh and carbon felt were employed to fabricate membrane modules, and the two MFC systems were termed as M1 and M2, respectively. High effluent quality was obtained with M1 and M2 in terms of turbidity, COD and ammonium. In M1, no bioelectrochemical denitrification took place, while nitrate decreased from 35.88 ± 4.15 to 27.33 ± 5.32 mg-N/L through the membrane in M2, causing a removal efficiency of 23.3 ± 6.5% with respect to cathodic nitrate. The denitrification ceased without electricity. The maximum power densities of M1 and M2 were 121 and 1253 mW/m3 , respectively. Micrococcus bacteria and rod-shaped bacteria covered the surface of carbon felt and fewer bacteria were found on stainless steel mesh. According to fluorescence in situ hybridization, the putative bacteria affiliated with Paracoccus genus and Pseudomonas spp. dominated in the interior biofilm on carbon felt for denitrification. Results demonstrate that the carbon felt system can perform bioelectrochemical denitrification to polish the effluent. © 2014 Elsevier Inc. All rights reserved.
1. Introduction Microbial fuel cells (MFCs), a promising approach to wastewater treatment, can convert organic substrates contained in wastewaters into electricity [1]. Since nitrate reduction was proved in biocathode MFCs, bioelectrochemical denitrification has attracted increasing attentions [2–4]. Compared to the conventional nitrification/denitrification process for nitrogen removal, bioelectrochemical denitrification can uncouple carbon source and nitrate spatially, making the denitrification an energy-recovering process. The specific MFCs have been established to remove various nitrogen compounds and recover electricity [5–9]. Also, the related processes are well understood in terms of biofilm stratification, microbial communities and analysis of electron fluxes [10–12]. A dual-cathode MFC can carry out successive nitrification and bioelectrochemical denitrification in their respective cathodes [13]. In a rotating biocathode MFC, as much as 25% of nitrogen removal resulted from bioelectrochemical denitrification [14].
∗ Corresponding author. Tel.: +86 411 84706173; fax: +86 411 84708083. E-mail addresses: [email protected], [email protected] (H. Zhang). http://dx.doi.org/10.1016/j.enzmictec.2014.04.005 0141-0229/© 2014 Elsevier Inc. All rights reserved.
MFCs may not be sufficient as a stand-alone wastewater treatment technology to achieve high effluent quality. Membrane bioreactors (MBRs) as a compact treatment technology has several advantages, such as high biomass content for pollutants removal and low effluent turbidity by membrane filtration. More importantly, MBRs provide better retention of slowly growing microorganisms (like nitrifiers, enhancing nitrification) [15]. The idea of combining MFCs with MBRs for wastewater treatment has recently been considered. The systems in which membranes simultaneously functioned as filtration components and cathodes were developed to treat wastewater and recover energy [16–19]. However, previous MFC-MBR studies focused on cathodic oxygen reduction, rather than cathodic denitrification and nitrogen removal. There is little knowledge about the effectiveness of membrane cathode for bioelectrochemical denitrification. Thus, the feasibility of bioelectrochemical denitrification using membrane cathodes as electron donor for nitrogen removal needs to be investigated. A novel bioreactor combining MFC and MBR was designed. Anode anaerobic chamber primarily removed majority of organic substrates, decreasing the organic load of the next aerobic chamber (i.e. MBR/cathode chamber) and establishing a more suitable environment for nitrifiers. In the cathode, cake layers (biofilm)
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
57
Fig. 1. (A) Schematic of the integrated system, (B) ammonium evolution in the cathode chamber, (C, E) raw/used SS mesh membrane module and (D, F) raw/used carbon felt membrane module. (1) Anodic chamber; (2) cathodic chamber; (3) air diffusers; (4) membrane modules (biocathodes); (5) stainless steel mesh separator; (6) suspended sludge; (7) biofilm; (8) membrane material (stainless steel mesh or carbon felt).
on the membranes simultaneously filtrated the effluent and performed denitrification. The aim of this study was to investigate the feasibility of simultaneous filtration and bioelectrochemical denitrification by employing conductive membrane modules as the cathodes. Two different matrix electrode materials (stainless steel mesh and carbon felt) were investigated in terms of treatment efficiencies of wastewater, electrochemical performances, catalytic behaviors, biofilm morphologies and community structures.
treatment plant (Dalian, China) was used to inoculate the MBR (cathode chamber), and the anodic inoculum was taken from another MFC reactor in the laboratory. The feeding medium contained (1 L of tap water): 6 g Na2 HPO4 , 3 g KH2 PO4 , 0.5 g NaCl, 0.1 g MgSO4 ·7H2 O, 0.015 g CaCl2 , and 1 mL trace nutrient solution [20]. COD and ammonium were supplied to the medium in the form of glucose and ammonium chloride according to demand. Each feed had a final pH of around 7.1 and a conductivity of around 5 mS/cm. The effluent was pumped out constantly with a fixed membrane flux of 11 L/(m2 h), resulting in a hydraulic retention time (HRT) of ∼9.6 h for the whole system (3.9 h for anode and 5.7 h for cathode, respectively). A pressure sensor was interfaced with the computer for measuring transmembrane pressure (TMP).
2. Materials and methods 2.1. Construction of MFC reactor The MFC setup is illustrated in Fig. 1. The total volume of the reactor was 960 mL (16[L] × 12[H] × 5[W]). The reactor was equally divided into two chambers (an anodic anaerobic zone and a cathodic aerobic zone) by a Plexiglas baffle (10 cm high), with a 2 cm gap on the top for completing electrical circuit and mass flow. The anode chamber (480 mL) was filled with granular graphite (3–5 mm diameter) and a graphite rod (8 mm diameter) was used as the electron collector (Sanye Carbon Co.), leaving a net liquid volume of 200 mL. A mm-sized stainless steel mesh (SS mesh) was located at the gap to hold granular graphite. The cathode chamber (480 mL) was constructed like a membrane bioreactor with a net liquid volume of 320 mL. Two gas diffusers were placed at the bottom of cathode chamber. The whole experiment was divided into two stages, stage I (1–93 days) for SS mesh MFC (M1) and stage II (94–143 days) for carbon felt MFC (M2). The used SS mesh was 1000 mesh, corresponding to about 13 m pore size. In each stage, two membrane modules were installed through the experiments and the effective filtration area of each module membrane was 30 cm2 . The modules were connected with copper wires across an external resistance to receive electrons from the anode. The joints were sealed with superglue to prevent potential corrosion. The MFC was shielded from light to prevent phototrophic reactions. 2.2. Operation conditions The MFC was operated at a room temperature (20–25 ◦ C) under a continuous feeding regime. Aerobic sludge from the Lingshui Ziguang wastewater
2.3. Electrochemical and chemical measurements The voltages generated in the experiment were collected using a data acquisition system (PISO–813, ICP–DAS) [21]. Power density (P, mW/m3 ) was normalized 2 /(Rex Vr ), where by the liquid volume of the cathodic chamber according to P = Vcell Vcell (V) was the voltage across the external resistance at a defined time interval, Rex () the external resistance, and Vr (m3 ) the cathodic solution volume. Coulombic efficiencies (CEs) based on COD removal (between the influent and anodic effluent) and/or nitrate reduction (between cathode and the final effluent) were calculated according to the previously described method [22,23]. Following stable power generation, a stepwise change of Rex was performed to obtain polarization and power curves using a three-electrode system. The specific method was: opening circuit for 60 min was initiated, and then the voltage over Rex at 30 min interval per resistor was recorded. Cyclic voltammograms were carried out using a CHI760E electrochemical workstation (CH Instruments, Chenhua Instrument Co., China) at a scan rate of 5 mV/s. At least 2 cycles were performed and the last cycle was shown. Hg/HgCl2 reference electrode (+0.242 V vs. SHE) was employed in all electrochemical tests, mounted in the cathode. The concentrations of COD and nitrogen compounds were measured according to standard methods [24]. The samples were filtered through a 0.45 m membrane before analysis. The standard parameters pH, conductivity and dissolved oxygen (DO) were determined potentiometrically with a digital, portable multiline meter (Multi 3430, SET F, WTW, Germany). The diameter of suspended sludge was measured by a Malvern laser particle size analyzer (Mastersizer 2000, UK) and D10 meant the grain size than which 10% of the gains were finer.
58
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Fig. 2. Effluent turbidities of SS mesh MFC (M1) and carbon felt MFC (M2).
2.4. Scanning electron microscope (SEM) and fluorescence in situ hybridization (FISH) Thin plates of SS mesh/carbon felt were cut off from membrane modules in experimental and control groups and were rinsed with the sterile medium. The samples were immediately immersed in a fixative solution of 4% glutaraldehyde (pH 7.4) for 90 min, and were subsequently rinsed three times (10 min each time) in phosphate buffer (0.1 mol L−1 , pH 7.0), and dehydrated by a graded ethanol series (40%, 70%, 90%, and 100%) each for 10 min and isoamyl acetate twice (10 min each time). The electrode pieces were dried at the CO2 -critical point from isoamyl acetate transitional solvent for 60 min. After lyophilization, samples were coated with 15nm Au particles, and viewed under a scanning electron microscope (SEM, Quanta 450, FEI, USA). Fluorescence in situ hybridization (FISH) was employed to further determine community structure of the biofilm on the carbon felt. At the end of the experiment, the exterior biofilm was scraped off from the carbon felt with a sterilized razor. Then the carbon felt was put into a centrifuge tube together with some feeding medium. The tube was shaken on a vortex shaker and centrifuged for 2 min at 8000 revolutions per minute (rpm) to obtain the remaining bacteria, termed as interior biofilm. FISH was conducted according to the method described by Neef et al. [25]. Probes used in this study included EUB338 for bacteria, NSO190 for ammonium oxidizing bacteria (AOB), NIT3 for nitrite oxidizing bacteria (NOB) and PAR1457&PAE997 for denitrifying bacteria [10]. The samples were detected by a confocal laser scanning microscope (CLSM, Olympus FV1000, Japan).
3. Results and discussion 3.1. Turbidity, COD and nitrogen removal in the systems Effluent turbidity, a relevant parameter to assess membrane filtration performance, was obtained as soon as the reactor was started up. As can be seen in Fig. 2, the effluent turbidities dropped fast to the stable values within the first several days (2.82 ± 1.54 NTU for M1 and 4.98 ± 1.53 NTU for M2, respectively), which was comparable with other studies (0–5 NTU) [16,18]. Low turbidities could be obtained with both membrane materials in a short time because of their inherent smaller pore sizes than the diameter of suspended sludge (D10 = 33.4 m). Since the membrane modules were directly exposed to the suspended activated sludge, microorganisms and/or their metabolites could aggregate on the membranes, forming that enhance the rejection effect [26,27]. The cathode was inoculated with the aerobic phase sludge. After one month, the effluent ammonium decreased to as low as 2.94 mgN/L (Fig. 3A), indicating that nitrifying bacteria were acclimated and enriched. However, there was no significant difference in nitrate concentration between the cathode and the effluent (Fig. 3B). The biofilm might be washed off from the mesh by intense aeration and oxygen was probably accessible to the deeper layer of the biofilm, imposing detrimental effect on denitrification. To alleviate washout of air bubbles and create an anoxic microenvironment within the biofilm, the aeration flow rate was cut down from 0.15 to 0.01 m3 /h on the 31st day. However, no nitrate was reduced yet, instead, the newly enriched nitrifiers got inhibited because of insufficient aeration, as was indicated by as high as 35.57 mg-N/L of effluent ammonium on the 58th day. After that, dissolved oxygen
Fig. 3. Evolution of nitrogen species at variable sites over time in SS mesh MFC (M1). (A) Ammonium and (B) nitrate. The left line represents the day when aeration rate was lowered and the right one represents the beginning of DO control (0.5–1.0 mg/L).
concentration was controlled at 0.5–1.0 mg/L and nitrification was restored until the 93rd day. In stage I (1–93 days), nitrite was always below 0.1 mg-N/L. The nitrate content increased slightly after passing through the membrane instead of decreasing (Fig. 3B), implying that nitrification rather than denitrification happened on the surface due to the fact that little biomass could be attached to the mesh and shed easily (Fig. 1E). Based on nitrogen balance (31–93 days) between the influent and cathode in M1, 20–30% of total nitrogen (sum of ammonium and nitrate) was removed. This part of nitrogen sinking might result from biomass growth and potential simultaneous nitrification and denitrification in the anode and cathode [18]. However, nitrate removal relying on bioelectrochemical denitrification by receiving electrons from the SS electrode could be hardly achieved. Carbon felt is probably a good biofilm retainer. Also, the relatively high conductivity of carbon felt is fit for electrode material [28]. The SS mesh was replaced with carbon felt on the 93rd day. Experimental data is illustrated in Fig. 4. The influent ammonium was 52.39 ± 3.85 mg-N/L. The average cathodic and effluent concentrations of nitrite still stayed at a low level (below 0.1 mg-N/L), and those of ammonium were 0.84 ± 0.6 and 0.76 ± 1.01 mg-N/L, respectively. From the 101 st day to 143rd day, nitrate obviously decreased through the membrane, with concentrations of 35.88 ± 4.15 mg-N/L in the cathodic solution and 27.73 ± 5.32 mg-N/L in the effluent, respectively. Nitrate of 8.15 ± 2.00 mg-N/L was removed, corresponding to a removal efficiency of 23.3 ± 6.5% based on cathodic nitrate. Open circuit mode was conducted to check if nitrate loss was caused by electricity generation. When the circuit was disconnected, the difference of nitrate in the cathode and effluent disappeared. The results clearly indicated that bioelectrochemical denitrification with electrodes as electron donor could be achieved to improve effluent quality, although the performance needed to be further improved.
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
59
Table 1 COD changes and CEs of SS mesh MFC (M1) and carbon felt MFC (M2). Cathode material
Mesh Carbon felt
COD (mg/L) Influent
Anode
Cathode
Effluent
296.6 ± 50.2 291.2 ± 87.4
82.9 ± 10.3 71.6 ± 26.4
12.1 ± 8.9 4.25 ± 10.3
10.8 ± 7.9 13.1 ± 13.5
CECOD (%)
CEnitrate (%)
1.8 ± 0.5 8.5 ± 3.6
n 116 ± 28
n represents no value.
A summary of data on CODs and CEs was given in Table 1. Contents of COD at various sites had no notable changes between M1 and M2. 66.8–75.4% of COD was eliminated after the stream passed through the anode compartments, causing the anodic CEs of 1.8 ± 0.5% and 8.5 ± 3.6% for M1 and M2, respectively. The low CEs might result from uncontrolled oxygen intrusion, and a more stringent anaerobic environment would be helpful. The CE based on nitrate reduction was 116 ± 28%, indicating that the electrons delivered from the anode were not sufficient to completely accomplish denitrification from nitrate into nitrogen gas. Considering that nitrite had a minor accumulation in the open circuit mode (Fig. S1), it was suspected that some intermediate nitrogen oxides, such as nitrite, NO and N2 O were released during the denitrification process [7]. Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.enzmictec.2014. 04.005.
Fig. 4. Evolution of nitrogen species at variable sites over time in carbon felt MFC (M2) with closed/open circuit. The period after the segregating line is in open circuit mode.
3.2. Electricity generation performances Fig. 5A shows the power density and polarization curves of the two MFC systems when the currents were stable. The maximum power density of M2 was 10-fold higher than that of M1 (1253 mW/m3 vs. 121 mW/m3 , normalized to the liquid volume of cathode). The voltages of M1 and M2 fell linearly with the increasing current density, and the former was much faster than the latter. The maximum current density of M2 was more than four times higher than that of M1 (6.7 A/m3 vs. 1.4 A/m3 , normalized to the liquid volume of cathode). The individual electrode potentials were collected to verify the performances of the cathodic biofilms (Fig. 5B). The potential of each anode was similar and relatively stable, maintained at −0.25 to −0.20 V vs. SHE, and thus the differences in power and current production from the two MFC systems resulted from the difference in cathode potential. Compared to SS mesh in M1, the biocompatibility of carbon felt in M2 was relatively high and bacteria could attach to the electrode more easily (Fig. 1E and F). The results implied that the carbon-based biofilm (bio-catalyst), including maximum thickness, surface coverage and biomass amount, was superior to the metal-based biofilm for cathode reactions. As a result, M1 was not
Fig. 5. Power generations and polarization curves (A) and electrode potentials (B) of SS mesh MFC (M1) and carbon felt MFC (M2) as function of current density (normalized by the liquid volume of cathode).
60
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
competitive with M2 in terms of power density and current density. The electrochemical performance of the carbon felt biocathode were improved by enhanced growth of cathodic biofilm [29]. The cathodic membrane materials as well as the attached biofilms were closely associated with the power output. 3.3. Catalytic behaviors of the biocathodes Cyclic voltammetry tests were carried out to study the effects of the two different materials on catalytic behaviors toward cathode reactions. CV tests were carried out under air or nitrogen gas flushing conditions. Because the time of each test was no more than 20 min, the conditions during the CV tests were thought to be constant. CVs of abiotic membrane modules (cathodes) were obtained by replacing suspended sludge with the modified influent medium (ammonium-N was substituted with 40 mg-N/L nitrate). As to air flushing condition (i.e. the routine operation), both oxygen and nitrate (from nitrification) were acted as electron acceptors; under nitrogen flushing condition, nitrate remaining in catholyte (about 40 mg/L) was the only oxidant. As shown in Fig. 6A and B, compared with their respective abiotic counterparts, each corresponding reductive current had an elevation, indicating the biocathodes presented better catalytic behaviors. Regarding the mesh, the reductive current bubbled with air was higher than that with nitrogen gas and the maximum values were 2.85 and 2.42 A/m2 , respectively. The presence of oxygen significantly increased the reductive current, suggesting that the cathode reaction with oxygen as electron acceptor could take place on the mesh. It was deduced that the membrane matrix was directly exposed to the sparged solution and no favorable site for nitrate reduction existed on the membrane. The result was in agreement with the previous observation that the biofilm on the mesh was sparse (Fig. 1E). As to carbon felt, very similar voltammograms for air and nitrogen flushing were obtained; even the latter had better performance, with the
Fig. 6. Cyclic voltammograms (vs. Hg/HgCl2 ) of (A) the SS mesh cathode and (B) the carbon felt cathode under different conditions. The arrows indicate the onset potentials.
maximum reductive currents of 7.84 A/m2 and 7.92 A/m2 , respectively. The similar CV curves suggested that the cathode reactions catalyzed by the biofilms were the same under aerobic or anoxic conditions and had no relationship with the presence of oxygen.
Fig. 7. SEMs for raw electrodes (A, C) and electrodes with the attached biofilms (B, D); SS mesh (A, B); carbon felt (C, D).
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
61
Fig. 8. Fluorescence in situ hybridization of bacteria on the carbon felt membrane with the probes for Eubacteria (green), ammonium-oxidizing bacteria and nitrite-oxidizing bacteria (red) and denitrifying bacteria (blue). Each third-photograph is a composite of the former two. (A) AOBs of exterior biofilm, (B) NOBs of exterior biofilm, (C) denitrifying bacteria of exterior biofilm, (D) AOBs of interior biofilm, (E) NOBs of interior biofilm, (F) denitrifying bacteria of interior biofilm. The scale bars correspond to 280 m.
62
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Oxygen penetration was limited by the thriving biofilm on the carbon felt. These results verified that the anoxic microenvironment within the dynamic biofilm was plausible and favorable for bioelectrochemical denitrification. Additionally, when comparing the observed reductive currents between the mesh and carbon felt, the carbon felt biocathode system possessed higher electro-catalytic activity, about four times higher than the SS mesh with respect to the maximum reductive current density, which was consistent with the data from polarization curves. The biofilms also affected the onset potentials at which the reductive currents started to occur. The onset potentials of the carbon felt biocathode under aerobic and anoxic conditions were around 0.12 and 0.10 V, while those of the mesh biocathode were 0.17 and 0.11 V, respectively (Fig. 6A and B). Concerning carbon felt, electrochemical reactions primarily took place in close proximity of electrodes. The presence of oxygen had little impact on the interior bacteria of the biofilm, and consequently caused similar onset potentials. As to the mesh, the onset potential for nitrogen flushing was 0.11 V, which was consistent with carbon felt biocathode, since their working sites were similarly anaerobic; the significant increase of the onset potential under the air flushing condition suggested that oxygen reduction could happen, consistent with above results. The electrode materials were critical to the properties of the biofilms (biocatalysts) such as coverage, thickness and stratification (Fig. 1E and F). Catalytic capacity of the biocathodes was closely associated with the electrode materials and the attached biofilms. Electricity generation in the case of SS mesh was severely limited by the poor biocatalyst and high cathodic overpotential, although the redox potential of oxygen reduction was higher than nitrate (0.82 V vs. 0.74 V vs. SHE, pH = 7) [4]. As to carbon felt, higher current could result from the better biofilm. Finally, better denitrification was achieved because of the proper microenvironment and more electrons (current). 3.4. SEM and FISH characterizations of biofilms In Section 3.1, nitrate reduced through the carbon-membrane and the reduction ceased during the open circuit; in Section 3.3, reductive reaction with nitrate as electron acceptor was achieved within the thriving biofilm on the carbon. All the results implied that anaerobic condition was preserved in the deeper layer of the biofilm and denitrifying bacteria could capture electrons from electrodes to reduce nitrate. The morphology of the bacteria on the two different materials was detected by SEM, especially the interior denitrifying bacteria on the carbon felt (Fig. 7). High magnification scanning electron microscopic observation showed that carbon felt was covered with a majority micrococcus bacteria and a minority rod-shaped bacteria in close proximity of the electrode (the exterior bacteria were removed in the pretreatment process, Fig. 1F), while the control electrode was not colonized. Fewer bacteria were found on the mesh electrode. A layer of polymer excreted from microbial activity appeared to cover the surface of the mesh, which possibly improved the biocompatibility of the mesh for further bacterial adherence, but did not benefit the conductivity and electrochemical reactions. This explained why the resultant denitrification was different when employing carbon felt or SS mesh from microbial morphology point of view. Denitrification is a biological process and denitrifying microorganisms play an indispensable role in this process. The presented bacteria associated with the carbon felt electrode were postulated to take charge of denitrification by receiving electrons from the electrode. Florescence in situ hybridization (FISH) was employed to characterize the communities of the exterior and interior biofilms on the carbon felt. The compositions of the biofilms were shown in the confocal micrographs, which used probe NSO190 for AOB, probe
Fig. 9. TMPs of SS mesh MFC (M1) and carbon felt MFC (M2).
NIT3 for NOB and probes PAR1457 and PAE997 for denitrifying bacteria belonging to the Paracoccus genus (micrococcus shaped) and Pseudomonas spp. (rod shaped), respectively (Fig. 8). Denitrifying bacteria were the majority population in the interior layer, implying that the micrococcus and rod-shaped bacteria observed in the SEM were probably affiliated with the described genus. The exterior biofilm of cathode was mainly occupied by the putative nitrifying bacteria. These results further confirmed that bioelectrochemical denitrification using the electrode as the electron donor was feasible in close proximity of the cathode [10,30,31]. All evidences supported a scenario where nitrifying microorganisms dominated in the exterior biofilm and created a DO concentration gradient within the whole biofilm, which allowed for bioelectrochemical denitrification by scavenging for electrons from solid electrode. This was further verified by the effluent DO (without detection). 3.5. Transmembrane pressure In the system with biofilm as a major filtration component, TMP was dependent on the formation of dynamic biofilm (Fig. 9). The TMP of M1 rose slowly and was maintained at a relatively low level of 0.95 kPa. When carbon felt was employed as the membrane material, the TMP exponentially increased up to 6.02 kPa with the growth of bacteria on the membrane in 10 days. The distinction of TMP between M1 and M2 was attributed to the characteristics of electrode matrix material. Carbon felt was prone to adherence for microorganisms in comparison with SS mesh. After the TMP of M2 reached 6.0 kPa, the increasing rate slowed down. Through the whole experiment, the TMP of M2 was always kept below 7.0 kPa. An equilibrium biofilm state was obtained between the growth and the washing-off by aerated bubbles, leading to a constant biofilm thickness and a stable TMP [32]. 4. Conclusions This study demonstrated that the membrane filtration biocathode MFC could effectively produce electricity and filter the effluent. More importantly, bioelectrochemical denitrification could further improve nitrogen removal in carbon felt MFC. Both the employed membranes presented high removal efficiencies in turbidity, COD and ammonium-N. However, bioelectrochemical denitrification was achieved only with carbon felt biocathode. Open circuit mode verified the phenomenon of bioelectrochemical denitrification. Micrococcus bacteria and rod-shaped bacteria were closely attached to the carbon felt electrode. AOBs and NOBs predominated in the exterior biofilm performing ammonium oxidation, whereas denitrifying microorganisms were more abundant in close proximity to the cathode for capturing electrons.
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Acknowledgements This research work has been financially supported by the Open Project of State Key Laboratory of Urban Water Resource and Environment (Grant No. ESK201301) and the Fundamental Research Funds for the Central Universities. References [1] Zhang F, Ge Z, Grimaud J, Hurst J, He Z. Long-term performance of liter-scale microbial fuel cells treating primary effluent installed in a municipal wastewater treatment facility. Environ Sci Technol 2013;47:4941–8. [2] Clauwaert P, Rabaey K, Aelterman P, De Schamphelaire L, Ham TH, Boeckx P, et al. Biological denitrification in microbial fuel cells. Environ Sci Technol 2007;41:3354–60. [3] Lefebvre O, Al-Mamun A, Ng HY. A microbial fuel cell equipped with a biocathode for organic removal and denitrification. Water Sci Technol 2008;58:881–5. [4] He Z, Angenent LT. Application of bacterial biocathodes in microbial fuel cells. Electroanalysis 2006;18:2009–15. [5] Virdis B, Rabaey K, Yuan Z, Keller J. Microbial fuel cells for simultaneous carbon and nitrogen removal. Water Res 2008;42:3013–24. [6] Virdis B, Rabaey K, Rozendal RA, Yuan Z, Keller J. Simultaneous nitrification, denitrification and carbon removal in microbial fuel cells. Water Res 2010;44:2970–80. [7] Puig S, Serra M, Vilar-Sanz A, Cabre M, Baneras L, Colprim J, et al. Autotrophic nitrite removal in the cathode of microbial fuel cells. Bioresour Technol 2011;102:4462–7. [8] Puig S, Coma M, Desloover J, Boon N, Colprim J, Balaguer MD. Autotrophic denitrification in microbial fuel cells treating low ionic strength waters. Environ Sci Technol 2012;46:2309–15. [9] Tong Y, He Z. Nitrate removal from groundwater driven by electricity generation and heterotrophic denitrification in a bioelectrochemical system. J Hazard Mater 2013;262:614–9. [10] Virdis B, Read ST, Rabaey K, Rozendal RA, Yuan ZG, Keller J. Biofilm stratification during simultaneous nitrification and denitrification (SND) at a biocathode. Bioresour Technol 2011;102:334–41. [11] Virdis B, Rabaey K, Yuan ZG, Rozendal RA, Keller J. Electron fluxes in a microbial fuel cell performing carbon and nitrogen removal. Environ Sci Technol 2009;43:5144–9. [12] Wrighton KC, Virdis B, Clauwaert P, Read ST, Daly RA, Boon N, et al. Bacterial community structure corresponds to performance during cathodic nitrate reduction. ISME J 2010;4:1443–55. [13] Zhang F, He Z. Integrated organic and nitrogen removal with electricity generation in a tubular dual-cathode microbial fuel cell. Process Biochem 2012;47:2146–51. [14] Zhang G, Zhang H, Zhang C, Zhang G, Yang F, Yuan G, et al. Simultaneous nitrogen and carbon removal in a single chamber microbial fuel cell with a rotating biocathode. Process Biochem 2013;48:893–900. [15] Judd S. The status of membrane bioreactor technology. Trends Biotechnol 2008;26:109–16.
63
[16] Wang YK, Sheng GP, Shi BJ, Li WW, Yu HQ. A novel electrochemical membrane bioreactor as a potential net energy producer for sustainable wastewater treatment. Sci Rep 2013;3. [17] Wang YP, Liu XW, Li WW, Li F, Wang YK, Sheng GP, et al. A microbial fuel cell-membrane bioreactor integrated system for cost-effective wastewater treatment. Appl Energy 2012;98:230–5. [18] Wang YK, Sheng GP, Li WW, Huang YX, Yu YY, Zeng RJ, et al. Development of a novel bioelectrochemical membrane reactor for wastewater treatment. Environ Sci Technol 2011;45:9256–61. [19] Su X, Tian Y, Sun Z, Lu Y, Li Z. Performance of a combined system of microbial fuel cell and membrane bioreactor: wastewater treatment, sludge reduction, energy recovery and membrane fouling. Biosens Bioelectron 2013;49: 92–8. [20] Lu HB, Oehmen A, Virdis B, Keller J, Yuan ZG. Obtaining highly enriched cultures of Candidatus Accumulibacter phosphates through alternating carbon sources. Water Res 2006;40:3838–48. [21] Wang X, Feng Y, Ren N, Wang H, Lee H, Li N, et al. Accelerated start-up of two-chambered microbial fuel cells: effect of anodic positive poised potential. Electrochim Acta 2009;54:1109–14. [22] Logan BE, Hamelers B, Rozendal RA, Schrorder U, Keller J, Freguia S, et al. Microbial fuel cells: methodology and technology. Environ Sci Technol 2006;40:5181–92. [23] Cheng KY, Ginige MP, Kaksonen AH. Ano-cathodophilic biofilm catalyzes both anodic carbon oxidation and cathodic denitrification. Environ Sci Technol 2012;46:10372–8. [24] APHA. Standard methods for the examination of water and wastewater. 20th ed. USA: United Book Press; 1998. [25] Neef A, Zaglauer A, Meier H, Amann R, Lemmer H, Schleifer KH. Population analysis in a denitrifying sand filter: conventional and in situ identification of Paracoccus spp. in methanol-fed biofilms. Appl Environ Microbiol 1996;62:4329–39. [26] Ersahin ME, Ozgun H, Dereli RK, Ozturk I, Roest K, van Lier JB. A review on dynamic membrane filtration: materials, applications and future perspectives. Bioresour Technol 2012;122:196–206. [27] Fan B, Huang X. Characteristics of a self-forming dynamic membrane coupled with a bioreactor for municipal wastewater treatment. Environ Sci Technol 2002;36:5245–51. [28] Wei JC, Liang P, Huang X. Recent progress in electrodes for microbial fuel cells. Bioresour Technol 2011;102:9335–44. [29] De Schamphelaire L, Boeckx P, Verstraete W. Evaluation of biocathodes in freshwater and brackish sediment microbial fuel cells. Appl Microbiol Biotechnol 2010;87:1675–87. [30] Li YZ, He YL, Ohandja DG, Ji J, Li JF, Zhou T. Simultaneous nitrification–denitrification achieved by an innovative internalloop airlift MBR: comparative study. Bioresour Technol 2008;99: 5867–72. [31] Park HI, Kim Dk, Choi Y-J, Pak D. Nitrate reduction using an electrode as direct electron donor in a biofilm-electrode reactor. Process Biochem 2005;40:3383–8. [32] Chu L, Li S. Filtration capability and operational characteristics of dynamic membrane bioreactor for municipal wastewater treatment. Sep Purif Technol 2006;51:173–9.
Contents lists available at ScienceDirect
Enzyme and Microbial Technology journal homepage: www.elsevier.com/locate/emt
Membrane filtration biocathode microbial fuel cell for nitrogen removal and electricity generation Guangyi Zhang a , Hanmin Zhang a,∗ , Yanjie Ma b , Guangen Yuan a , Fenglin Yang a , Rong Zhang a a Key Laboratory of Industrial Ecology and Environmental Engineering (MOE), School of Environmental Science and Technology, Dalian University of Technology, Linggong Road 2, Dalian 116024, China b Environmental Management College of China, Hebei Road 73, Qinhuangdao 066004, China
a r t i c l e
i n f o
Article history: Received 17 February 2014 Received in revised form 1 April 2014 Accepted 7 April 2014 Available online 18 April 2014 Keywords: Microbial fuel cell Membrane filtration biocathode Bioelectrochemical denitrification Carbon felt Stainless steel mesh
a b s t r a c t Conductive materials with attached biofilms, were used as membrane filtration biocathodes to filter the effluent and supply electrons for denitrification. Stainless steel mesh and carbon felt were employed to fabricate membrane modules, and the two MFC systems were termed as M1 and M2, respectively. High effluent quality was obtained with M1 and M2 in terms of turbidity, COD and ammonium. In M1, no bioelectrochemical denitrification took place, while nitrate decreased from 35.88 ± 4.15 to 27.33 ± 5.32 mg-N/L through the membrane in M2, causing a removal efficiency of 23.3 ± 6.5% with respect to cathodic nitrate. The denitrification ceased without electricity. The maximum power densities of M1 and M2 were 121 and 1253 mW/m3 , respectively. Micrococcus bacteria and rod-shaped bacteria covered the surface of carbon felt and fewer bacteria were found on stainless steel mesh. According to fluorescence in situ hybridization, the putative bacteria affiliated with Paracoccus genus and Pseudomonas spp. dominated in the interior biofilm on carbon felt for denitrification. Results demonstrate that the carbon felt system can perform bioelectrochemical denitrification to polish the effluent. © 2014 Elsevier Inc. All rights reserved.
1. Introduction Microbial fuel cells (MFCs), a promising approach to wastewater treatment, can convert organic substrates contained in wastewaters into electricity [1]. Since nitrate reduction was proved in biocathode MFCs, bioelectrochemical denitrification has attracted increasing attentions [2–4]. Compared to the conventional nitrification/denitrification process for nitrogen removal, bioelectrochemical denitrification can uncouple carbon source and nitrate spatially, making the denitrification an energy-recovering process. The specific MFCs have been established to remove various nitrogen compounds and recover electricity [5–9]. Also, the related processes are well understood in terms of biofilm stratification, microbial communities and analysis of electron fluxes [10–12]. A dual-cathode MFC can carry out successive nitrification and bioelectrochemical denitrification in their respective cathodes [13]. In a rotating biocathode MFC, as much as 25% of nitrogen removal resulted from bioelectrochemical denitrification [14].
∗ Corresponding author. Tel.: +86 411 84706173; fax: +86 411 84708083. E-mail addresses: [email protected], [email protected] (H. Zhang). http://dx.doi.org/10.1016/j.enzmictec.2014.04.005 0141-0229/© 2014 Elsevier Inc. All rights reserved.
MFCs may not be sufficient as a stand-alone wastewater treatment technology to achieve high effluent quality. Membrane bioreactors (MBRs) as a compact treatment technology has several advantages, such as high biomass content for pollutants removal and low effluent turbidity by membrane filtration. More importantly, MBRs provide better retention of slowly growing microorganisms (like nitrifiers, enhancing nitrification) [15]. The idea of combining MFCs with MBRs for wastewater treatment has recently been considered. The systems in which membranes simultaneously functioned as filtration components and cathodes were developed to treat wastewater and recover energy [16–19]. However, previous MFC-MBR studies focused on cathodic oxygen reduction, rather than cathodic denitrification and nitrogen removal. There is little knowledge about the effectiveness of membrane cathode for bioelectrochemical denitrification. Thus, the feasibility of bioelectrochemical denitrification using membrane cathodes as electron donor for nitrogen removal needs to be investigated. A novel bioreactor combining MFC and MBR was designed. Anode anaerobic chamber primarily removed majority of organic substrates, decreasing the organic load of the next aerobic chamber (i.e. MBR/cathode chamber) and establishing a more suitable environment for nitrifiers. In the cathode, cake layers (biofilm)
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
57
Fig. 1. (A) Schematic of the integrated system, (B) ammonium evolution in the cathode chamber, (C, E) raw/used SS mesh membrane module and (D, F) raw/used carbon felt membrane module. (1) Anodic chamber; (2) cathodic chamber; (3) air diffusers; (4) membrane modules (biocathodes); (5) stainless steel mesh separator; (6) suspended sludge; (7) biofilm; (8) membrane material (stainless steel mesh or carbon felt).
on the membranes simultaneously filtrated the effluent and performed denitrification. The aim of this study was to investigate the feasibility of simultaneous filtration and bioelectrochemical denitrification by employing conductive membrane modules as the cathodes. Two different matrix electrode materials (stainless steel mesh and carbon felt) were investigated in terms of treatment efficiencies of wastewater, electrochemical performances, catalytic behaviors, biofilm morphologies and community structures.
treatment plant (Dalian, China) was used to inoculate the MBR (cathode chamber), and the anodic inoculum was taken from another MFC reactor in the laboratory. The feeding medium contained (1 L of tap water): 6 g Na2 HPO4 , 3 g KH2 PO4 , 0.5 g NaCl, 0.1 g MgSO4 ·7H2 O, 0.015 g CaCl2 , and 1 mL trace nutrient solution [20]. COD and ammonium were supplied to the medium in the form of glucose and ammonium chloride according to demand. Each feed had a final pH of around 7.1 and a conductivity of around 5 mS/cm. The effluent was pumped out constantly with a fixed membrane flux of 11 L/(m2 h), resulting in a hydraulic retention time (HRT) of ∼9.6 h for the whole system (3.9 h for anode and 5.7 h for cathode, respectively). A pressure sensor was interfaced with the computer for measuring transmembrane pressure (TMP).
2. Materials and methods 2.1. Construction of MFC reactor The MFC setup is illustrated in Fig. 1. The total volume of the reactor was 960 mL (16[L] × 12[H] × 5[W]). The reactor was equally divided into two chambers (an anodic anaerobic zone and a cathodic aerobic zone) by a Plexiglas baffle (10 cm high), with a 2 cm gap on the top for completing electrical circuit and mass flow. The anode chamber (480 mL) was filled with granular graphite (3–5 mm diameter) and a graphite rod (8 mm diameter) was used as the electron collector (Sanye Carbon Co.), leaving a net liquid volume of 200 mL. A mm-sized stainless steel mesh (SS mesh) was located at the gap to hold granular graphite. The cathode chamber (480 mL) was constructed like a membrane bioreactor with a net liquid volume of 320 mL. Two gas diffusers were placed at the bottom of cathode chamber. The whole experiment was divided into two stages, stage I (1–93 days) for SS mesh MFC (M1) and stage II (94–143 days) for carbon felt MFC (M2). The used SS mesh was 1000 mesh, corresponding to about 13 m pore size. In each stage, two membrane modules were installed through the experiments and the effective filtration area of each module membrane was 30 cm2 . The modules were connected with copper wires across an external resistance to receive electrons from the anode. The joints were sealed with superglue to prevent potential corrosion. The MFC was shielded from light to prevent phototrophic reactions. 2.2. Operation conditions The MFC was operated at a room temperature (20–25 ◦ C) under a continuous feeding regime. Aerobic sludge from the Lingshui Ziguang wastewater
2.3. Electrochemical and chemical measurements The voltages generated in the experiment were collected using a data acquisition system (PISO–813, ICP–DAS) [21]. Power density (P, mW/m3 ) was normalized 2 /(Rex Vr ), where by the liquid volume of the cathodic chamber according to P = Vcell Vcell (V) was the voltage across the external resistance at a defined time interval, Rex () the external resistance, and Vr (m3 ) the cathodic solution volume. Coulombic efficiencies (CEs) based on COD removal (between the influent and anodic effluent) and/or nitrate reduction (between cathode and the final effluent) were calculated according to the previously described method [22,23]. Following stable power generation, a stepwise change of Rex was performed to obtain polarization and power curves using a three-electrode system. The specific method was: opening circuit for 60 min was initiated, and then the voltage over Rex at 30 min interval per resistor was recorded. Cyclic voltammograms were carried out using a CHI760E electrochemical workstation (CH Instruments, Chenhua Instrument Co., China) at a scan rate of 5 mV/s. At least 2 cycles were performed and the last cycle was shown. Hg/HgCl2 reference electrode (+0.242 V vs. SHE) was employed in all electrochemical tests, mounted in the cathode. The concentrations of COD and nitrogen compounds were measured according to standard methods [24]. The samples were filtered through a 0.45 m membrane before analysis. The standard parameters pH, conductivity and dissolved oxygen (DO) were determined potentiometrically with a digital, portable multiline meter (Multi 3430, SET F, WTW, Germany). The diameter of suspended sludge was measured by a Malvern laser particle size analyzer (Mastersizer 2000, UK) and D10 meant the grain size than which 10% of the gains were finer.
58
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Fig. 2. Effluent turbidities of SS mesh MFC (M1) and carbon felt MFC (M2).
2.4. Scanning electron microscope (SEM) and fluorescence in situ hybridization (FISH) Thin plates of SS mesh/carbon felt were cut off from membrane modules in experimental and control groups and were rinsed with the sterile medium. The samples were immediately immersed in a fixative solution of 4% glutaraldehyde (pH 7.4) for 90 min, and were subsequently rinsed three times (10 min each time) in phosphate buffer (0.1 mol L−1 , pH 7.0), and dehydrated by a graded ethanol series (40%, 70%, 90%, and 100%) each for 10 min and isoamyl acetate twice (10 min each time). The electrode pieces were dried at the CO2 -critical point from isoamyl acetate transitional solvent for 60 min. After lyophilization, samples were coated with 15nm Au particles, and viewed under a scanning electron microscope (SEM, Quanta 450, FEI, USA). Fluorescence in situ hybridization (FISH) was employed to further determine community structure of the biofilm on the carbon felt. At the end of the experiment, the exterior biofilm was scraped off from the carbon felt with a sterilized razor. Then the carbon felt was put into a centrifuge tube together with some feeding medium. The tube was shaken on a vortex shaker and centrifuged for 2 min at 8000 revolutions per minute (rpm) to obtain the remaining bacteria, termed as interior biofilm. FISH was conducted according to the method described by Neef et al. [25]. Probes used in this study included EUB338 for bacteria, NSO190 for ammonium oxidizing bacteria (AOB), NIT3 for nitrite oxidizing bacteria (NOB) and PAR1457&PAE997 for denitrifying bacteria [10]. The samples were detected by a confocal laser scanning microscope (CLSM, Olympus FV1000, Japan).
3. Results and discussion 3.1. Turbidity, COD and nitrogen removal in the systems Effluent turbidity, a relevant parameter to assess membrane filtration performance, was obtained as soon as the reactor was started up. As can be seen in Fig. 2, the effluent turbidities dropped fast to the stable values within the first several days (2.82 ± 1.54 NTU for M1 and 4.98 ± 1.53 NTU for M2, respectively), which was comparable with other studies (0–5 NTU) [16,18]. Low turbidities could be obtained with both membrane materials in a short time because of their inherent smaller pore sizes than the diameter of suspended sludge (D10 = 33.4 m). Since the membrane modules were directly exposed to the suspended activated sludge, microorganisms and/or their metabolites could aggregate on the membranes, forming that enhance the rejection effect [26,27]. The cathode was inoculated with the aerobic phase sludge. After one month, the effluent ammonium decreased to as low as 2.94 mgN/L (Fig. 3A), indicating that nitrifying bacteria were acclimated and enriched. However, there was no significant difference in nitrate concentration between the cathode and the effluent (Fig. 3B). The biofilm might be washed off from the mesh by intense aeration and oxygen was probably accessible to the deeper layer of the biofilm, imposing detrimental effect on denitrification. To alleviate washout of air bubbles and create an anoxic microenvironment within the biofilm, the aeration flow rate was cut down from 0.15 to 0.01 m3 /h on the 31st day. However, no nitrate was reduced yet, instead, the newly enriched nitrifiers got inhibited because of insufficient aeration, as was indicated by as high as 35.57 mg-N/L of effluent ammonium on the 58th day. After that, dissolved oxygen
Fig. 3. Evolution of nitrogen species at variable sites over time in SS mesh MFC (M1). (A) Ammonium and (B) nitrate. The left line represents the day when aeration rate was lowered and the right one represents the beginning of DO control (0.5–1.0 mg/L).
concentration was controlled at 0.5–1.0 mg/L and nitrification was restored until the 93rd day. In stage I (1–93 days), nitrite was always below 0.1 mg-N/L. The nitrate content increased slightly after passing through the membrane instead of decreasing (Fig. 3B), implying that nitrification rather than denitrification happened on the surface due to the fact that little biomass could be attached to the mesh and shed easily (Fig. 1E). Based on nitrogen balance (31–93 days) between the influent and cathode in M1, 20–30% of total nitrogen (sum of ammonium and nitrate) was removed. This part of nitrogen sinking might result from biomass growth and potential simultaneous nitrification and denitrification in the anode and cathode [18]. However, nitrate removal relying on bioelectrochemical denitrification by receiving electrons from the SS electrode could be hardly achieved. Carbon felt is probably a good biofilm retainer. Also, the relatively high conductivity of carbon felt is fit for electrode material [28]. The SS mesh was replaced with carbon felt on the 93rd day. Experimental data is illustrated in Fig. 4. The influent ammonium was 52.39 ± 3.85 mg-N/L. The average cathodic and effluent concentrations of nitrite still stayed at a low level (below 0.1 mg-N/L), and those of ammonium were 0.84 ± 0.6 and 0.76 ± 1.01 mg-N/L, respectively. From the 101 st day to 143rd day, nitrate obviously decreased through the membrane, with concentrations of 35.88 ± 4.15 mg-N/L in the cathodic solution and 27.73 ± 5.32 mg-N/L in the effluent, respectively. Nitrate of 8.15 ± 2.00 mg-N/L was removed, corresponding to a removal efficiency of 23.3 ± 6.5% based on cathodic nitrate. Open circuit mode was conducted to check if nitrate loss was caused by electricity generation. When the circuit was disconnected, the difference of nitrate in the cathode and effluent disappeared. The results clearly indicated that bioelectrochemical denitrification with electrodes as electron donor could be achieved to improve effluent quality, although the performance needed to be further improved.
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
59
Table 1 COD changes and CEs of SS mesh MFC (M1) and carbon felt MFC (M2). Cathode material
Mesh Carbon felt
COD (mg/L) Influent
Anode
Cathode
Effluent
296.6 ± 50.2 291.2 ± 87.4
82.9 ± 10.3 71.6 ± 26.4
12.1 ± 8.9 4.25 ± 10.3
10.8 ± 7.9 13.1 ± 13.5
CECOD (%)
CEnitrate (%)
1.8 ± 0.5 8.5 ± 3.6
n 116 ± 28
n represents no value.
A summary of data on CODs and CEs was given in Table 1. Contents of COD at various sites had no notable changes between M1 and M2. 66.8–75.4% of COD was eliminated after the stream passed through the anode compartments, causing the anodic CEs of 1.8 ± 0.5% and 8.5 ± 3.6% for M1 and M2, respectively. The low CEs might result from uncontrolled oxygen intrusion, and a more stringent anaerobic environment would be helpful. The CE based on nitrate reduction was 116 ± 28%, indicating that the electrons delivered from the anode were not sufficient to completely accomplish denitrification from nitrate into nitrogen gas. Considering that nitrite had a minor accumulation in the open circuit mode (Fig. S1), it was suspected that some intermediate nitrogen oxides, such as nitrite, NO and N2 O were released during the denitrification process [7]. Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.enzmictec.2014. 04.005.
Fig. 4. Evolution of nitrogen species at variable sites over time in carbon felt MFC (M2) with closed/open circuit. The period after the segregating line is in open circuit mode.
3.2. Electricity generation performances Fig. 5A shows the power density and polarization curves of the two MFC systems when the currents were stable. The maximum power density of M2 was 10-fold higher than that of M1 (1253 mW/m3 vs. 121 mW/m3 , normalized to the liquid volume of cathode). The voltages of M1 and M2 fell linearly with the increasing current density, and the former was much faster than the latter. The maximum current density of M2 was more than four times higher than that of M1 (6.7 A/m3 vs. 1.4 A/m3 , normalized to the liquid volume of cathode). The individual electrode potentials were collected to verify the performances of the cathodic biofilms (Fig. 5B). The potential of each anode was similar and relatively stable, maintained at −0.25 to −0.20 V vs. SHE, and thus the differences in power and current production from the two MFC systems resulted from the difference in cathode potential. Compared to SS mesh in M1, the biocompatibility of carbon felt in M2 was relatively high and bacteria could attach to the electrode more easily (Fig. 1E and F). The results implied that the carbon-based biofilm (bio-catalyst), including maximum thickness, surface coverage and biomass amount, was superior to the metal-based biofilm for cathode reactions. As a result, M1 was not
Fig. 5. Power generations and polarization curves (A) and electrode potentials (B) of SS mesh MFC (M1) and carbon felt MFC (M2) as function of current density (normalized by the liquid volume of cathode).
60
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
competitive with M2 in terms of power density and current density. The electrochemical performance of the carbon felt biocathode were improved by enhanced growth of cathodic biofilm [29]. The cathodic membrane materials as well as the attached biofilms were closely associated with the power output. 3.3. Catalytic behaviors of the biocathodes Cyclic voltammetry tests were carried out to study the effects of the two different materials on catalytic behaviors toward cathode reactions. CV tests were carried out under air or nitrogen gas flushing conditions. Because the time of each test was no more than 20 min, the conditions during the CV tests were thought to be constant. CVs of abiotic membrane modules (cathodes) were obtained by replacing suspended sludge with the modified influent medium (ammonium-N was substituted with 40 mg-N/L nitrate). As to air flushing condition (i.e. the routine operation), both oxygen and nitrate (from nitrification) were acted as electron acceptors; under nitrogen flushing condition, nitrate remaining in catholyte (about 40 mg/L) was the only oxidant. As shown in Fig. 6A and B, compared with their respective abiotic counterparts, each corresponding reductive current had an elevation, indicating the biocathodes presented better catalytic behaviors. Regarding the mesh, the reductive current bubbled with air was higher than that with nitrogen gas and the maximum values were 2.85 and 2.42 A/m2 , respectively. The presence of oxygen significantly increased the reductive current, suggesting that the cathode reaction with oxygen as electron acceptor could take place on the mesh. It was deduced that the membrane matrix was directly exposed to the sparged solution and no favorable site for nitrate reduction existed on the membrane. The result was in agreement with the previous observation that the biofilm on the mesh was sparse (Fig. 1E). As to carbon felt, very similar voltammograms for air and nitrogen flushing were obtained; even the latter had better performance, with the
Fig. 6. Cyclic voltammograms (vs. Hg/HgCl2 ) of (A) the SS mesh cathode and (B) the carbon felt cathode under different conditions. The arrows indicate the onset potentials.
maximum reductive currents of 7.84 A/m2 and 7.92 A/m2 , respectively. The similar CV curves suggested that the cathode reactions catalyzed by the biofilms were the same under aerobic or anoxic conditions and had no relationship with the presence of oxygen.
Fig. 7. SEMs for raw electrodes (A, C) and electrodes with the attached biofilms (B, D); SS mesh (A, B); carbon felt (C, D).
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
61
Fig. 8. Fluorescence in situ hybridization of bacteria on the carbon felt membrane with the probes for Eubacteria (green), ammonium-oxidizing bacteria and nitrite-oxidizing bacteria (red) and denitrifying bacteria (blue). Each third-photograph is a composite of the former two. (A) AOBs of exterior biofilm, (B) NOBs of exterior biofilm, (C) denitrifying bacteria of exterior biofilm, (D) AOBs of interior biofilm, (E) NOBs of interior biofilm, (F) denitrifying bacteria of interior biofilm. The scale bars correspond to 280 m.
62
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Oxygen penetration was limited by the thriving biofilm on the carbon felt. These results verified that the anoxic microenvironment within the dynamic biofilm was plausible and favorable for bioelectrochemical denitrification. Additionally, when comparing the observed reductive currents between the mesh and carbon felt, the carbon felt biocathode system possessed higher electro-catalytic activity, about four times higher than the SS mesh with respect to the maximum reductive current density, which was consistent with the data from polarization curves. The biofilms also affected the onset potentials at which the reductive currents started to occur. The onset potentials of the carbon felt biocathode under aerobic and anoxic conditions were around 0.12 and 0.10 V, while those of the mesh biocathode were 0.17 and 0.11 V, respectively (Fig. 6A and B). Concerning carbon felt, electrochemical reactions primarily took place in close proximity of electrodes. The presence of oxygen had little impact on the interior bacteria of the biofilm, and consequently caused similar onset potentials. As to the mesh, the onset potential for nitrogen flushing was 0.11 V, which was consistent with carbon felt biocathode, since their working sites were similarly anaerobic; the significant increase of the onset potential under the air flushing condition suggested that oxygen reduction could happen, consistent with above results. The electrode materials were critical to the properties of the biofilms (biocatalysts) such as coverage, thickness and stratification (Fig. 1E and F). Catalytic capacity of the biocathodes was closely associated with the electrode materials and the attached biofilms. Electricity generation in the case of SS mesh was severely limited by the poor biocatalyst and high cathodic overpotential, although the redox potential of oxygen reduction was higher than nitrate (0.82 V vs. 0.74 V vs. SHE, pH = 7) [4]. As to carbon felt, higher current could result from the better biofilm. Finally, better denitrification was achieved because of the proper microenvironment and more electrons (current). 3.4. SEM and FISH characterizations of biofilms In Section 3.1, nitrate reduced through the carbon-membrane and the reduction ceased during the open circuit; in Section 3.3, reductive reaction with nitrate as electron acceptor was achieved within the thriving biofilm on the carbon. All the results implied that anaerobic condition was preserved in the deeper layer of the biofilm and denitrifying bacteria could capture electrons from electrodes to reduce nitrate. The morphology of the bacteria on the two different materials was detected by SEM, especially the interior denitrifying bacteria on the carbon felt (Fig. 7). High magnification scanning electron microscopic observation showed that carbon felt was covered with a majority micrococcus bacteria and a minority rod-shaped bacteria in close proximity of the electrode (the exterior bacteria were removed in the pretreatment process, Fig. 1F), while the control electrode was not colonized. Fewer bacteria were found on the mesh electrode. A layer of polymer excreted from microbial activity appeared to cover the surface of the mesh, which possibly improved the biocompatibility of the mesh for further bacterial adherence, but did not benefit the conductivity and electrochemical reactions. This explained why the resultant denitrification was different when employing carbon felt or SS mesh from microbial morphology point of view. Denitrification is a biological process and denitrifying microorganisms play an indispensable role in this process. The presented bacteria associated with the carbon felt electrode were postulated to take charge of denitrification by receiving electrons from the electrode. Florescence in situ hybridization (FISH) was employed to characterize the communities of the exterior and interior biofilms on the carbon felt. The compositions of the biofilms were shown in the confocal micrographs, which used probe NSO190 for AOB, probe
Fig. 9. TMPs of SS mesh MFC (M1) and carbon felt MFC (M2).
NIT3 for NOB and probes PAR1457 and PAE997 for denitrifying bacteria belonging to the Paracoccus genus (micrococcus shaped) and Pseudomonas spp. (rod shaped), respectively (Fig. 8). Denitrifying bacteria were the majority population in the interior layer, implying that the micrococcus and rod-shaped bacteria observed in the SEM were probably affiliated with the described genus. The exterior biofilm of cathode was mainly occupied by the putative nitrifying bacteria. These results further confirmed that bioelectrochemical denitrification using the electrode as the electron donor was feasible in close proximity of the cathode [10,30,31]. All evidences supported a scenario where nitrifying microorganisms dominated in the exterior biofilm and created a DO concentration gradient within the whole biofilm, which allowed for bioelectrochemical denitrification by scavenging for electrons from solid electrode. This was further verified by the effluent DO (without detection). 3.5. Transmembrane pressure In the system with biofilm as a major filtration component, TMP was dependent on the formation of dynamic biofilm (Fig. 9). The TMP of M1 rose slowly and was maintained at a relatively low level of 0.95 kPa. When carbon felt was employed as the membrane material, the TMP exponentially increased up to 6.02 kPa with the growth of bacteria on the membrane in 10 days. The distinction of TMP between M1 and M2 was attributed to the characteristics of electrode matrix material. Carbon felt was prone to adherence for microorganisms in comparison with SS mesh. After the TMP of M2 reached 6.0 kPa, the increasing rate slowed down. Through the whole experiment, the TMP of M2 was always kept below 7.0 kPa. An equilibrium biofilm state was obtained between the growth and the washing-off by aerated bubbles, leading to a constant biofilm thickness and a stable TMP [32]. 4. Conclusions This study demonstrated that the membrane filtration biocathode MFC could effectively produce electricity and filter the effluent. More importantly, bioelectrochemical denitrification could further improve nitrogen removal in carbon felt MFC. Both the employed membranes presented high removal efficiencies in turbidity, COD and ammonium-N. However, bioelectrochemical denitrification was achieved only with carbon felt biocathode. Open circuit mode verified the phenomenon of bioelectrochemical denitrification. Micrococcus bacteria and rod-shaped bacteria were closely attached to the carbon felt electrode. AOBs and NOBs predominated in the exterior biofilm performing ammonium oxidation, whereas denitrifying microorganisms were more abundant in close proximity to the cathode for capturing electrons.
G. Zhang et al. / Enzyme and Microbial Technology 60 (2014) 56–63
Acknowledgements This research work has been financially supported by the Open Project of State Key Laboratory of Urban Water Resource and Environment (Grant No. ESK201301) and the Fundamental Research Funds for the Central Universities. References [1] Zhang F, Ge Z, Grimaud J, Hurst J, He Z. Long-term performance of liter-scale microbial fuel cells treating primary effluent installed in a municipal wastewater treatment facility. Environ Sci Technol 2013;47:4941–8. [2] Clauwaert P, Rabaey K, Aelterman P, De Schamphelaire L, Ham TH, Boeckx P, et al. Biological denitrification in microbial fuel cells. Environ Sci Technol 2007;41:3354–60. [3] Lefebvre O, Al-Mamun A, Ng HY. A microbial fuel cell equipped with a biocathode for organic removal and denitrification. Water Sci Technol 2008;58:881–5. [4] He Z, Angenent LT. Application of bacterial biocathodes in microbial fuel cells. Electroanalysis 2006;18:2009–15. [5] Virdis B, Rabaey K, Yuan Z, Keller J. Microbial fuel cells for simultaneous carbon and nitrogen removal. Water Res 2008;42:3013–24. [6] Virdis B, Rabaey K, Rozendal RA, Yuan Z, Keller J. Simultaneous nitrification, denitrification and carbon removal in microbial fuel cells. Water Res 2010;44:2970–80. [7] Puig S, Serra M, Vilar-Sanz A, Cabre M, Baneras L, Colprim J, et al. Autotrophic nitrite removal in the cathode of microbial fuel cells. Bioresour Technol 2011;102:4462–7. [8] Puig S, Coma M, Desloover J, Boon N, Colprim J, Balaguer MD. Autotrophic denitrification in microbial fuel cells treating low ionic strength waters. Environ Sci Technol 2012;46:2309–15. [9] Tong Y, He Z. Nitrate removal from groundwater driven by electricity generation and heterotrophic denitrification in a bioelectrochemical system. J Hazard Mater 2013;262:614–9. [10] Virdis B, Read ST, Rabaey K, Rozendal RA, Yuan ZG, Keller J. Biofilm stratification during simultaneous nitrification and denitrification (SND) at a biocathode. Bioresour Technol 2011;102:334–41. [11] Virdis B, Rabaey K, Yuan ZG, Rozendal RA, Keller J. Electron fluxes in a microbial fuel cell performing carbon and nitrogen removal. Environ Sci Technol 2009;43:5144–9. [12] Wrighton KC, Virdis B, Clauwaert P, Read ST, Daly RA, Boon N, et al. Bacterial community structure corresponds to performance during cathodic nitrate reduction. ISME J 2010;4:1443–55. [13] Zhang F, He Z. Integrated organic and nitrogen removal with electricity generation in a tubular dual-cathode microbial fuel cell. Process Biochem 2012;47:2146–51. [14] Zhang G, Zhang H, Zhang C, Zhang G, Yang F, Yuan G, et al. Simultaneous nitrogen and carbon removal in a single chamber microbial fuel cell with a rotating biocathode. Process Biochem 2013;48:893–900. [15] Judd S. The status of membrane bioreactor technology. Trends Biotechnol 2008;26:109–16.
63
[16] Wang YK, Sheng GP, Shi BJ, Li WW, Yu HQ. A novel electrochemical membrane bioreactor as a potential net energy producer for sustainable wastewater treatment. Sci Rep 2013;3. [17] Wang YP, Liu XW, Li WW, Li F, Wang YK, Sheng GP, et al. A microbial fuel cell-membrane bioreactor integrated system for cost-effective wastewater treatment. Appl Energy 2012;98:230–5. [18] Wang YK, Sheng GP, Li WW, Huang YX, Yu YY, Zeng RJ, et al. Development of a novel bioelectrochemical membrane reactor for wastewater treatment. Environ Sci Technol 2011;45:9256–61. [19] Su X, Tian Y, Sun Z, Lu Y, Li Z. Performance of a combined system of microbial fuel cell and membrane bioreactor: wastewater treatment, sludge reduction, energy recovery and membrane fouling. Biosens Bioelectron 2013;49: 92–8. [20] Lu HB, Oehmen A, Virdis B, Keller J, Yuan ZG. Obtaining highly enriched cultures of Candidatus Accumulibacter phosphates through alternating carbon sources. Water Res 2006;40:3838–48. [21] Wang X, Feng Y, Ren N, Wang H, Lee H, Li N, et al. Accelerated start-up of two-chambered microbial fuel cells: effect of anodic positive poised potential. Electrochim Acta 2009;54:1109–14. [22] Logan BE, Hamelers B, Rozendal RA, Schrorder U, Keller J, Freguia S, et al. Microbial fuel cells: methodology and technology. Environ Sci Technol 2006;40:5181–92. [23] Cheng KY, Ginige MP, Kaksonen AH. Ano-cathodophilic biofilm catalyzes both anodic carbon oxidation and cathodic denitrification. Environ Sci Technol 2012;46:10372–8. [24] APHA. Standard methods for the examination of water and wastewater. 20th ed. USA: United Book Press; 1998. [25] Neef A, Zaglauer A, Meier H, Amann R, Lemmer H, Schleifer KH. Population analysis in a denitrifying sand filter: conventional and in situ identification of Paracoccus spp. in methanol-fed biofilms. Appl Environ Microbiol 1996;62:4329–39. [26] Ersahin ME, Ozgun H, Dereli RK, Ozturk I, Roest K, van Lier JB. A review on dynamic membrane filtration: materials, applications and future perspectives. Bioresour Technol 2012;122:196–206. [27] Fan B, Huang X. Characteristics of a self-forming dynamic membrane coupled with a bioreactor for municipal wastewater treatment. Environ Sci Technol 2002;36:5245–51. [28] Wei JC, Liang P, Huang X. Recent progress in electrodes for microbial fuel cells. Bioresour Technol 2011;102:9335–44. [29] De Schamphelaire L, Boeckx P, Verstraete W. Evaluation of biocathodes in freshwater and brackish sediment microbial fuel cells. Appl Microbiol Biotechnol 2010;87:1675–87. [30] Li YZ, He YL, Ohandja DG, Ji J, Li JF, Zhou T. Simultaneous nitrification–denitrification achieved by an innovative internalloop airlift MBR: comparative study. Bioresour Technol 2008;99: 5867–72. [31] Park HI, Kim Dk, Choi Y-J, Pak D. Nitrate reduction using an electrode as direct electron donor in a biofilm-electrode reactor. Process Biochem 2005;40:3383–8. [32] Chu L, Li S. Filtration capability and operational characteristics of dynamic membrane bioreactor for municipal wastewater treatment. Sep Purif Technol 2006;51:173–9.
