Bioelectrochemically-assisted nitrogen removal in osmotic membrane bioreactor

As an emerging technology of wastewater treatment, osmosis membrane bioreactor (OMBR) is considered as a promising alternative process for high-quality water extraction and nutrient-energy-water (‘NEW’) recovery (Qin & He 2017). In an OMBR, semi-permeable forward osmosis (FO) membranes are used for physical separation combined with biological activated sludge for organic matter and nutrient removal (Achilli et al. 2009). Water passes through the FO membrane from feed solution (i.e. bioreactor) to draw solution (i.e. high salinity) driven by osmotic pressure difference. OMBRs have several advantages over conventional membrane bioreactors (MBRs), such as lower energy consumption because of the low driven force produced by osmotic pressure, higher pollutants rejection due to the smaller pore diameter of FO membrane (Holloway et al. 2014), and lower potential of membrane fouling (Kim et al. 2014; Luo et al. 2015). In the past few years, OMBRs have been widely used for seawater desalination, wastewater treatment (Estella et al. 2019), power generation due to the salinity gradient difference between seawater and freshwater, and biogas production (i.e. methane) from an anaerobic OMBR (AnOMBR) (Li et al. 2017).

However, there is a fundamental problem of OMBRs to be processed urgently, which is low nitrogen removal efficiency in the feed side resulted from low FO membrane rejection (i.e. 70%). Conventional biological nitrogen removal consists of two processes: ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB) oxidize ammonium (NH₄⁺) to nitrate (NO₃⁻) under aerobic condition, and subsequently nitrate is reduced to nitrogen gas by the denitrification bacteria under anaerobic condition (Chen et al. 2010). Such processes require a suitable dissolved oxygen concentration (DO) and a readily biodegradable organic matter (Yu et al. 2011). However, the DO concentration in the feed side of OMBRs is over 2.0 mg L⁻¹, and that of AnOMBRs is below 0.5 mg L⁻¹, which can only remove partial total nitrogen (TN) in the bioreactor. A tested OMBR could achieve high chemical oxygen demand (COD) and phosphorus removal efficiency (i.e. 100% and 96%, respectively), but ammonium removal efficiency was only 43% (Huang et al. 2015). An AnOMBR reported that it only had an ammonium removal efficiency of 62% (Chen et al. 2014). This is because of the significant impact of SRT on both salinity accumulation due to reversal solute flux and microbial activity (Wang et al. 2014). In other words, low SRT operation can control the salinity build-up in OMBRs but results in adverse effects on the denitrification microorganisms due to their relatively long generation period. As for the OMBRs at high SRT, the elevated salinity may not be favorable for the growth of microorganisms and results in poor organics and nitrogen removal. Therefore, it is necessary to explore a novel method to alleviate the salinity build-up and enhance nitrogen removal in OMBRs.

One method to reduce salt accumulation and thus enhance nitrogen removal is to integrate membrane filtration such as ultrafiltration or microfiltration into an OMBR, but this requires high energy consumption (Holloway et al. 2014; Wang et al. 2017). An energy-efficiency strategy for enhancing nitrogen removal is the incorporation of bioelectrochemical systems (BESs) into the feed side of OMBRs to complete both simultaneous nitrification and denitrification (SND) in one integrated system and convert the energy contained in wastewater to electricity. A BES is using electricity production bacteria in an anode to oxidize organic matter for electrons production that then reduces a terminal electron acceptor in a cathode for electricity generation. BESs have been synergistically linked to nitrification and denitrification to achieve enhanced nitrogen removal. It was demonstrated that ammonium removal efficiency varies from 80 to 90% in traditional two-chamber BESs, and TN removal efficiency could reach to 93.9% in the single-chamber microbial fuel cell (MFC) (Yan et al. 2012; Wu et al. 2017). Thus, it is feasible to enhance OMBRs nitrogen removal by integrating with BES theoretically. On the one hand, a BES integration with low sludge production can reduce FO membrane fouling of an OMBR (Zhang et al. 2017), and the generated current from a BES can reduce the salinity accumulation in the feed side (Yang et al. 2018). On the other hand, the effluent quality of a BES, which used to be unsatisfied (Logan et al. 2015; Zhang et al. 2015), could be improved by high rejection of FO membrane, and the concentrating effect in the feed side of an OMBR would increase the electricity generation in a BES (Hou et al. 2016).

In this study, bioelectrochemically-assisted (BEA) operation was first proposed to integrate into the feed side of the OMBR to enhance pollutant removal (i.e. nitrogen and organic carbon). The integrated system consisted of three equal-size compartments, and the anode and cathode chambers were separated by anion exchange membrane (AEM). Ammonium is oxidized to nitrate in the anode compartment by using inadvertently diffused oxygen from aerated cathode and subsequently achieves denitrification under anaerobic conditions. The sodium acetate (NaAc) was served as a draw solute, and the reversed acetate due to inevitable reversal solute flux was used as a supplementary carbon source for both denitrification and electricity generation in the feed/anode side. Moreover, the BESs remove organic matters with small biomass production, which may reduce FO membrane fouling as pre-treatment by improving the mixed liquor properties. The specific objectives of this study were to (1) evaluate nitrogen removal in an integrated BEA-OMBR system and (2) explore the pathways of nitrogen removal in this integrated system.

The integrated BEA-OMBR consisted of three equal-size compartments with a liquid volume of 400 mL/each, which were draw side, feed side/anode, and cathode (Figure 1). The draw and feed/anode were separated by a thin-film-composite forward membrane (Guochu Technology Inc., Xia Meng, China) with an effective membrane area 0.0079 m². The active layer of the FO membrane faced the feed solution and the support layer faced the draw solution. The feed/anode and the cathode were separated by anion exchange membrane (AMI-7001, Membranes International Inc., Ringwood, NJ, USA) with the same effective membrane area 0.0079 m². The anode electrode was made of non-wet proofed carbon brushes (Toray Industries, Japan), and the cathode electrode was made of wet-proofed carbon cloth (10 cm × 10 cm, WOS1002, Ce Tech. Inc., Taiwan, China). The carbon brushes were submerged in acetone for 24 h and then in ethyl alcohol for 12 h followed by pure water for 6 h, ultrasonic washing for 30 min, and stabilized with 370 °C for 30 min in a muffle furnace. The carbon cloth was catalyzed by evenly painting active carbon powder with 5 mg cm⁻² and then stabilized the same as carbon brush.

The integrated BEA-OMBR was operated at 25 ± 1 °C. The feed/anode was fed in synthetic wastewater included 780 ± 16 mg L⁻¹ sodium acetate (1,000 ± 20 mg L⁻¹ COD), 15 mg L⁻¹ MgSO₄, 20 mg L⁻¹ CaCl₂, 500 mg L⁻¹ NaCl, 100 mg L⁻¹ NaHCO₃, 5.35 mg L⁻¹ K₂HPO₄, 2.65 mg L⁻¹ KH₂PO₄, and trace elements (Angenent & Sung 2001). The feed/anode compartment was inoculated with 40 mL anaerobic sludge from the laboratory anaerobic digester. In addition, the cathode compartment was filled by 0.2 M NaHCO₃ with air aeration at a flow rate of 10 L min⁻¹. Both anolyte and catholyte were recirculated at the flow rate of 96 mL min⁻¹. The anode of the BES was operated in a batch mode, and the hydraulic retention time was 120 h. The draw solution was 0.25-M NaAc with the recirculation rate of 96 mL/min. The integrated reactor was operated in two-stages: in the first stage, the ammonium (NH₄⁺-N) concentration was 39 mg L⁻¹ in the influent under open and closed-circuit (external resistance of 1,000 Ω) conditions, and in the second stage, three different NH₄⁺ concentrations (39 mg L⁻¹, 80 mg L⁻¹, and 120 mg L⁻¹) were studied in the influent under closed-circuit condition.

The feed/anode mix liquid samples were collected in open/closed-circuit, and corresponding samples were stored at −20 °C. Then, the samples were extracted for DNA using the PowerSoil DNA Isolation Kit (MoBio Laboratories, Inc.). The extracted DNA of two samples were sequenced by high-throughput sequencing analysis of 16 s RNA gene amplicon and metagenomics analysis, respectively. The extracted DNA of open-circuit employed 16 s RNA amplifying V3-V4 hypervariable regions by polymerase chain reaction (PCR) with the bacterial universal primers 341F (CCTACGGGNGGCWGCAG) and 806R (GGACTACNVGGGTWTCTAAT). The PCR products were sequenced with an Illumina MiSeq platform. PICRUSt analysis was performed to predict the nitrogen removal pathway from the bacterial community data based on the Greengenes database.

The extracted DNA of open-circuit was examined by 1% agarose gel electrophoresis and by spectrophotometry (260 nm/280 nm optical density ratio). For qualified DNA samples, the Covaris ultrasonic breaker was used to randomly interrupt small fragments of genomic DNA with a growth of about 300 bp, and the whole library was prepared through such steps as terminal repair, a-tail addition, sequencing joint addition, purification, and PCR amplification. After the construction of the library, Qubit2.0 was used for initial quantification, the library was diluted to 2 ng µL⁻¹, and the inserted fragments of the library were detected by Agilent 2100. After the inserted fragments met the expectation, the qPCR method was used for accurate quantification of the effective concentration of the library (the effective concentration of the library was >3 nM) to ensure the library quality. After the library passed the quality inspection, different libraries were pooled according to the requirements of effective concentration and target disembarkation data volume, and then Illumina HiSeq sequencing was performed.

Because of the current generation, the nitrate in the catholyte would move across AEM into anode, thereby cause denitrification in the anolyte (or the feed), while at the same time, the organic carbon removal efficiency would be enhanced by the electrochemically-active bacteria; because of the RSF, the reversed NaAc could be served as supplementary carbon source to support the growth of both denitrifying and electrogenesis bacteria in the anolyte. This was investigated by comparing the system performance with the influent NH₄⁺-N concentration of 39 mg L⁻¹ in closed-circuit (with current generation) and open-circuit (no electricity generation; a conventional OMBR) (Figure 2). The voltage in open-circuit was 0.14–0.49 V, and the power density in closed-circuit was 0–1.12 W m⁻³. The COD concentration in the feed/anode effluent under closed-circuit condition was lower than that under open-circuit condition, and thus the COD removal efficiency in closed-circuit was improved by current production (p < 0.05, one-tailed two-sample t-test with unequal variance at α = 0.05 for all the following statistical tests). For 84-h operation for example, COD removal efficiency in the feed/anode effluent was increased from 49.96% in open-circuit to 90.76% in closed-circuit (Figure 2(a)). The activity of heterotrophic exoelectrogenesis increased potential for COD degradation via increased metabolism of bacteria under the closed-circuit condition (Tian et al. 2014). The integrated system in closed-circuit had a slightly higher water flux of 0.21–0.51 LMH than that in open-circuit (0.14–0.49 LMH), which resulted in a higher final water recovery of 227 mL in the closed-circuit (Figure 2(b)).

The open and closed-circuit had a similar trend of ammonium removal rate in the feed/anode effluent, permeate, and catholyte. For example, the ammonium (NH₄⁺-N) concentration in the feed/anode effluent under open-circuit condition was gradually decreased from 40.44 to 2.08 mg L⁻¹ over 84-h operation, and that under the closed-circuit condition was gradually decreased from 44.20 to 0 mg L⁻¹; the NH₄⁺-N concentration in the permeate and catholyte under both open and closed-circuit condition was below 4.46 mg L⁻¹ during 84-h operation (Figure 3(a)). This result has two implications: first, the ammonium almost did not move across the AEM and FO membrane; and second, ammonium could be removed in the feed/anode chamber because the oxygen diffusion from an aerated cathode was in favor of the growth of AOB or NOB (Min et al. 2005). The NO₃^–-N and NO₂^–-N concentration in three chambers under open and closed-circuit conditions was below 0.78 mg L⁻¹ (Figure 3(b)), indicated that generated NO₃^–-N and NO₂^–-N from the nitrification process in the anode were reduced by denitrifying bacteria under the anode anoxic condition.

The nitrogen removal performance of the integrated system in the open and closed-circuit was further investigated by the influent NH₄⁺-N concentration of 120 mg L⁻¹ (Figure 4). The NH₄⁺-N concentration in the feed/anode effluent under the open-circuit condition was decreased from 132.87 to 13.77 mg L⁻¹ over 120-h operation, and that for open-circuit had a similar trend (Figure 4(a)). Because of high NH₄⁺-N concentration in the mixed liquor of the feed/anode, the NH₄⁺-N concentration in the permeate under closed-circuit was increased from 0.02 to 16.39 mg L⁻¹ over 120-h operation, and that under open-circuit condition was increased from 0 to 8.68 mg L⁻¹. Thus, the overall ammonium removal efficiency had a similar value of 78.87–84.50% in closed and open-circuit. The NO₃^–-N and NO₂^–-N concentration in three chambers for the open and closed-circuit was below 1.28 mg L⁻¹ (Figure 4(b)). Thus, the limited step for nitrogen removal in integrated system was nitrification, and BES coupling OMBR has the potential to enhance nitrification in the cathode chamber by looping anolyte effluent.

The above results have demonstrated that the integrated system may achieve effective nitrogen removal from a high ammonium strength wastewater. The ammonium strength is critically important to the integrated system, and this was estimated in the closed-circuit with different influent ammonium strength (i.e. 39 mg L⁻¹, 80 mg L⁻¹, and 120 mg L⁻¹) (Figure 5). The time needed to remove influent NH₄⁺-N concentration of 39 mg L⁻¹ in the feed/anode was 48-h, and that of 80 mg L⁻¹ was 120-h. Because of the low NH₄⁺-N concentration in the permeate and catholyte, the time for overall ammonium removal efficiency to reach 95.83% was the same as achieving ammonium removal in the feed/anode. The integrated system with influent NH₄⁺-N concentration of 120 mg L⁻¹ could not achieve complete ammonium removal in the feed/anode, the NH₄⁺-N concentration in the feed/anode still decreased from 124.97 to 17.87 mg L⁻¹ over 120-h operation. The NH₄⁺-N concentration in the permeate increased to 16.39 mg L⁻¹ due to the bidirectional solute flux in the FO membrane, and thus the overall ammonium removal efficiency was 76.32% in the integrated system. These results indicated that the ammonium removal from a high ammonium strength wastewater by the integrated system was limited by the ammonium oxidation process in the feed/anode side.

To better understand the mechanism of nitrogen removal in the integrated system, two amplicon libraries for sequencing analysis were obtained in open and closed-circuit (Figure 6). At the genus level, a total of 16 genera were detected >1% taxa in open-circuit and that in close-circuit were only six genera. More than 13.91% and 53.13% of the sequences were not identified, indicating the diversity of the microbial community in both open and closed-circuit. Thirteen genus of organisms (Thauera, Azonexus, Comamonas, Ignavibacterium, Bacillus, Thiobacillus, Geobactor, Thermomonas (0.05%), Dechloromonas (0.07%), Paracoccus (0.08%), Runrivivax (0.01%), Nitrosomonas, and Nitrospira) associated with nitrogen conversion were identified in closed-circuit, and only three (Thauera, Azonexus, and Simplicispira) existed in open-circuit, which had the capacity of aerobic denitrification (Figure 6(a)) (Claus & Kutzner 1985). The abundance of Thauera in the close-circuit (8.01%) was higher than that in open-circuit (1.36%), and the abundance of Azonexus in the closed-circuit (0.56%) was lower than that in open-circuit (0.20%). Comamonas (0.02%), Ignavibacterium (0.02%), Bacillus (0.12%), Thiobacillus (0.03%), Azonexus (0.20%) and Geobactor (0.9%) were only identified in the closed-circuit, which have been reported as denitrifiers that utilized substrate as growth conditions and carbon sources for denitrification. These results showed that the Thauera genus was the predominant denitrifier in the closed-circuit, which synergistically worked with AOB to facilitate electron transport from ammonium to nitrite, and thus stimulated current generation (He et al. 2009). A group of nitrifiers was detected in the closed-circuit, such as Bacillus (NOB, 0.12%), Nitrosomonas (AOB, 0.04%) and Nitrospira (NOB, 0.04%). Thauera, Nitrosomonas, Thiobacillus, and Azonexus could remove nitrogen via simultaneous nitrification and denitrification, and the Thauera and Nitrosomonas were considered as electroactive bacteria to improve electricity generation in the closed-circuit (Quan et al. 2006; Liu et al. 2013).

Overall, the abundance of nitrogen conversion bacteria in the closed-circuit was 8.85%, therefore, the nitrifying genus was 0.20%, and the denitrifying genus was 8.65%. However, the total abundance of nitrogen conversion bacteria (i.e. denitrifying bacteria) under open-circuit condition was 1.98% without nitrifying bacteria (Figure 6(b)). This result demonstrated the potential of improving nitrogen removal in the feed side of an OMBR by bioelectrochemically-assisted operation and explained the pathway of the proposed system.

This study has demonstrated the feasibility of using bioelectrochemically-assisted (BEA) operation to enhance both nitrogen and organic carbon removal in an osmotic membrane bioreactor (OMBR). The diversity of the microbial community in the feed side of an OMBR was increased under the BEA operation and thus improved nutrient and organic carbon removal efficiency. The sequencing results showed that AOB, NOB, and denitrifying bacteria were abundant in the feed/anode side of the integrated BEA-OMBR and explained the pathway of enhancing nitrogen removal. Future research will need to improve nitrogen removal from a high ammonium strength wastewater by looping anolyte effluent to the cathode. These results have provided an effective operation mode for improving nitrogen removal in OMBRs.

This work was supported by National Natural Science Foundation of China (51908292, 51828801, 51978148), National Major Science and Technology Projects of China (2017ZX07202004), Natural Science Foundation of Jiangsu Province (BK20190716), Natural Science Foundation of the Jiangsu Higher Education Institutions of China (19KJB610003), and startup fund of Nanjing Normal University (184080H202B179). Hai-Liang Song would like to acknowledge the Qing Lan Project of Jiangsu Province.