Simultaneous current generation and ammonia recovery from real urine using nitrogen-purged bioelectrochemical systems

Abstract

Bioelectrochemical systems (BESs) offer a strategy for treating source-separated urine with current generation, but the high content of ammonia is still a challenge for sustainable maintenance of BESs due to ammonia inhibition. Therefore, an integrated BES setup was developed to overcome this problem by ammonia recovery. This setup, working in closed circuit mode with nitrogen purging (CN), allowed for the produced ammonia to be continuously channeled to an absorption bottle. In addition, control reactors in closed circuit (CC) or in open circuit mode (OC) were also run for comparison. A maximum power density of 310.9 ± 1.0 mW m⁻² was obtained for the CN reactor, and 127.1 ± 0.9 mW m⁻² was obtained for the CC reactor. Total nitrogen (TN) removal efficiency (84.9% ± 2.2%) from urine was considerably higher in the CN reactor than it was in the CC (29.7% ± 6.7%) or OC (30.0% ± 8.2%) reactor. In the CN reactor, 52.8% ± 3.6% of the TN was recovered in the form of NH₃-N, with a NH₃ recovery rate of 435.7 ± 29.6 gN m⁻³ d⁻¹. The improved performance of the CN reactor was attributed to the mitigation of ammonia inhibition to the anode electro-activity. 16S rDNA sequencing showed that no Anammox and nitrifiers were detected on the anodes and cathodes. Overall, nitrogen purging provides the urine-fed BESs with a useful approach for maintaining the system performance by ammonia recovery.

---

Introduction

In a municipal wastewater treatment plant (WWTP), approximately 80% of nitrogen (N) and 50% of phosphorous (P) come from urine, although it constitutes only 1% of the total wastewater volume.^1,2 In another aspect, nitrogen and phosphorus are two main elements that lead to eutrophication in natural water bodies. Therefore, a process, either physical/chemical (e.g., using activated carbon, flocculation and membrane technologies) or biological (e.g., nitrification/denitrification and Anammox) should be applied to remove them from wastewater.^3,4 All these practices has its own disadvantages, which include the high operation costs of chemical addition or the requirement of large volume reactors to create the desired operating conditions.⁵ Besides N and P, there exist a large amount of other nutrients and organics in urine. If urine is separated at the source instead of being directly discharged into WWTPs or natural water bodies, it can be used for different purposes, such as nutrient recovery.^6–8 Among them, phosphorous and nitrogen recoveries have received the most attention, because both elements are not only essential nutrients for plant growth, but they also offer great potential for fertilizer production.

Struvite precipitation offers a strategy for the simultaneous recovery of nitrogen and phosphorus from source-separated urine. In natural conditions, urea hydrolysis can readily occur and then reach a NH₄⁺/NH₃ equilibrium. This will cause an increase in pH and trigger the precipitation of ammonium, phosphate and magnesium in the form of struvite. Being an effective slow-release fertilizer, struvite is of great value to the fertilizer market. Its production can be achieved by the following chemical reaction:

Mg²⁺ + NH₄⁺ + PO₄³⁻ + 6H₂O → MgNH₄PO₄·6H₂O↓ (1)

where the molar ratio of NH₄⁺ : PO₄³⁻ : Mg²⁺ is 1 : 1 : 1 in theory. In hydrolyzed urine, ammonium is sufficient for struvite production with the ratio of 260 : 6 : 1 (NH₄⁺ : PO₄³⁻ : Mg²⁺).⁶ Therefore, if more struvite is expected to be produced, additional phosphate and magnesium (e.g., MgO and MgCl₂) are needed.^9,10

In comparison, the technologies for nitrogen recovery are more economically attractive; thus, the practice for phosphate or magnesium addition can be obviated. It has been reported that various technologies such as selective ion exchange and ammonia stripping, which are usually used to recover nitrogen from high-strength ammonium wastewater, can also be applied to urine treatment.^11,12 However, these technologies have to consider the extra energy input for cation transport or the addition of caustics for pH increase, whereas these challenges can be circumvented by applying bioelectrochemical systems (BESs). Recently, BESs have been intensively investigated as an alternative to ammonia recovery. In a BES equipped with a cation exchange membrane (CEM), ammonium can be transported from the anode to the cathode chamber, which is a process driven by electro-migration and diffusion. It has been reported that the NH₄⁺/NH₃ migration could account for about 90% of the ionic flux.¹³ With increasing pH, the concentrated ammonium in the cathode chamber is readily transformed into volatile ammonia, which can be subsequently removed through NH₃ stripping.¹⁴ Following up on the proof-of-concept for BESs to achieve ammonia recovery, the feasibility of recovering ammonia from real urine was first demonstrated and evaluated using a two-chamber microbial fuel cell (MFC) with a gas diffusion layer cathode, from which volatile ammonia was carried away with the gas stream and absorbed by an acid solution.¹⁵

Although ammonia recovery from urine is a promising concept for application of BESs, effects of ammonia on the sustainable maintenance of BESs remain controversial. Some reports show that ammonia is oxidized with the anode as an electron acceptor in BESs,^16–18 while others support adverse effects of ammonia on anode electro-activity.¹⁹ These contradictory conclusions are believed to be dependent on the types of dominant bacterial species enriched on the anodes. Generally, the production of electricity from ammonium is only found on anodes enriched with nitrifiers such as Nitrosomonas europaea,¹⁸ but in most cases, they are rarely detected. Therefore, it is critical to identify and circumvent the potential challenges if a BES is expected to recover ammonia from urine.

In this study, a hypothesis has been proposed that if urine, which has been treated inside a BES, is continuously purged with nitrogen gas (purity, 99.999%) and then channeled to an absorption bottle, we would not only recover the desired ammonia, but also alleviate the possible ammonia inhibition to the BES performance. Therefore, we developed an integrated BES setup, and its performance was examined in terms of current generation and nitrogen conversion, which was compared to those of controls in closed and open circuit modes. The difference in performance between the setup and the controls was further evaluated from the perspective of the electrochemical characteristics of the electrode. In addition, the possibilities for nitrification and denitrification in these reactors were also assessed using 16S rDNA pyrosequencing to target specific bacteria associated with these biological processes.

Materials and methods

Experimental setup construction

The integrated BES reactor (Fig. 1), operated in closed circuit mode with nitrogen purging (CN), was constructed based on an air-cathode MFC as described in a previous study.²⁰ Graphite fiber was chosen as the anode and was wound onto two twisted titanium wires and heat treated according to a previous report.²¹ The cathode was carbon cloth, with one side as the catalytic layer and the other side as the diffusion layer.²² The anode was connected to the cathode via an external circuit containing a resistor of 1000 Ω. There were two ports on the top of the CN reactor, which were sealed with rubber stoppers. Two custom-made needles (N-1, N-2) were vertically inserted through the rubber stopper (close to the cathode) into the reactor chamber at different depths, which allowed the nitrogen gas to flow in (N-1) and out (N-2) of the chamber. The N-1 needle hub was connected via a peristaltic pump (LongerPump BT100-1L, Baoding) to the washing bottle, from which moist nitrogen gas streams were pumped (at a flow rate of 1.35 mL min⁻¹) into the chamber of the CN reactor. The N-2 needle hub was connected via a reflux tube to the absorption bottle, where ammonia was absorbed by 0.4 M sulfuric acid. For the washing bottle, there were also two needles as the inlet and outlet. When the nitrogen gas was continuously pumped into the chamber of the CN reactor, it was necessary to keep the pressure in the washing bottle slightly positive compared to that of the atmosphere. Another two sets of identical air-cathode MFCs as controls, one operating in open circuit mode (OC) and the others in closed circuit mode (CC), were also used to compare the differences in performance or microbial community with those of the CN reactors.

Fig. 1 Schematic diagram of the integrated BES (CN) for recovery of ammonia from human urine.

Inoculation and operation

All the reactors (in triplicate) were inoculated with a 30 : 70 (v/v) mixture of domestic wastewater and fresh urine. After 3 batch cycles, all the reactors, except for the OC reactors, produced maximum currents of more than 0.2 mA. Subsequently, the influents were completely changed to urine. Urine was provided by a healthy male volunteer. The reactors were operated in batch mode and fed with fresh urine once the voltage decreased to 50 mV. The CN reactors were connected with the needles when the peak currents increased to 0.35 mA. Similar enrichment was also conducted in our previous study, where electrochemical characteristics of electrodes over time in air-cathode MFCs were investigated based on polarization data, while the urine used in each cycle was same. All the reactors were operated in a temperature-controlled chamber (30 °C). All the experiments were performed in triplicate, and mean values or typical results were presented.

Physical and chemical analyses

To compare the maximum power densities as well as the electrochemical characteristics of the electrodes, power density and polarization curves were obtained for the nitrogen-purged and non-purged reactors following the single-cycle method.²³ In our previous study, to investigate the impact of real urine on MFC performance during long-term operation, power density and polarization curves at different times were also obtained on single-chamber air-cathode MFCs. The external resistance was varied from OCV to 30 Ω, with each resistor being connected for 20 min. The oxygen flux into the chamber over one full batch cycle was monitored for the CN and CC reactors using a non-consumptive oxygen probe (NeoFox, Ocean Optics, Inc., Dunedin, FL). The rubber stopper next to the air-cathode was used to fix the probe. All the urine samples for chemical analysis were centrifuged (Eppendorf Centrifuge 5418; 14 000 × g, 5 min) to remove suspended solids. TN was determined using a HACH T-N kit (10–150 mg L⁻¹). Total ammonia nitrogen (TAN, the sum of NH₄⁺-N and NH₃-N) was measured by the salicylate method using a Test N′ Tube kit (HACH, Loveland, CO, USA). Nitrite and nitrate concentrations were measured using ion chromatography (Dionex ICS-3000 system with an IonPac AS11-HC Analytical column, 4 × 250 mm). Na, K, Mg, and P were analyzed using inductively coupled plasma-atomic emission spectrometry (ICP-AES) (Perkin Elmer Optima 5300DV, Waltham, Massachusetts) and are presented in Table S2.† Precipitates on the graphite fiber anodes were characterized by X-ray diffraction (XRD) using a diffractometer (Bruker D8 ADVANCE) operated at 40 kV and 30 mA. Qualitative analysis was performed using PDXL software.

Electrochemical analysis

Cyclic voltammetry (CV) was performed on the anode of the CC and CN reactors with the cathode as the counter electrode and Ag/AgCl electrode as the reference electrode. The reference electrode was inserted into the port close to the anode side. Both reactors were fed with fresh urine. When the current output approached the maximum (based on average values) in either of the reactors, the circuits were disconnected from the data acquisition system and connected to a potentiostat/galvanostat (Autolab, Metrohm, model PGSTAT 128N). For the CN reactor, nitrogen purging was stopped prior to the CV test. Subsequently, the potential was scanned positively from −0.7 V to 0.2 V at a rate of 5 mV s⁻¹. The catalytic feature of the biofilms was characterized by plotting the first derivative of each sweep as a function of potential.²⁴

Microbial community analysis

To compare the microbial communities on the anode and cathode, six samples were collected from the three types of reactors for 16S rDNA pyrosequencing after 152 days of operation. Total genomic DNA was extracted from biofilms using a Bacteria DNA Mini Kit (Watson Biotechnologies, Inc., Shanghai) according to the manufacturer's instructions and assessed by electrophoresis in 1% agarose gels. The V3 and V4 regions of 16S rDNA genes were amplified by PCR using the fusion primers: 341F (5′-CCTACGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGGTATCTAATCC-3′).²⁵ Amplification was carried out as previously described.²⁶ After PCR amplification, amplicons were purified with an agarose gel DNA purification kit and cloned using a T-carrier cloning kit (Sangon Biotech Shanghai Co., Ltd, China) according to the manufacturer's instructions. Pyrosequencing of amplicons was performed by an instrument provided by Sangon Biotech (Shanghai) Co., Ltd using a 3730 type DNA sequencing system. Low-quality sequences (<25) or those with lengths lower than 200 base pairs (bp) were removed according to standard protocols. Representative sequences from each OTU were phylogenetically assigned to a taxonomic identity using the RDP Naïve Bayesian rRNA classifier at a confidence threshold of 80%.²⁷

Calculations

Current was calculated by monitoring the voltage (U) across the external resistor (R) in the circuit using a data acquisition system (PISO-813, ICP DAS Co., Ltd). Power density, P (mW m⁻²), was calculated from the measured voltage as P = U²/A × R, where A is the projected surface area of the cathode (7 cm²).

NH₃ concentration in urine was estimated from the following equation:²⁸

where [NH₃] is the concentration of free ammonia, [TAN] is the total ammonia concentration, including the concentrations of ammonia (NH₃) and ammonium (NH₄⁺), and T is the temperature (Kelvin).

The ammonia recovery rate (gN m⁻³ d⁻¹) was calculated as

where ϖ is the amount of ammonia recovered daily from urine and v is the volume of liquid in the reactor.

Results

Current and power generation

After 50 days of operation, all the reactors began to produce a stable current, showing that MFCs are applicable to high-strength human urine. As can be seen in four successive feeding cycles (Fig. 2), the peak current reached 0.43 mA for the CN reactor and 0.32 mA for the CC reactor. The difference in peak current between both systems was observed to be within a stable range of 0.10 ± 0.02 mA.

Fig. 2 The current generation over four successive feeding cycles in the CC and CN reactors fed with fresh urine after 50 days of operation.

On day 62, the performances of these reactors were evaluated by the polarization and power density curves. A maximum power density of 310.9 ± 1.0 mW m⁻² and 127.1 ± 0.9 mW m⁻² was obtained for the CN reactor and the CC reactor, respectively (Fig. 3A). Accordingly, the electrode potentials varied differentially along the resistance decrease. When the anode (or cathode) potential increased (or decreased) to the final potentials (30 Ω), the current density was 0.22 ± 0.02 mA for the CN reactor, but 0.11 ± 0 mA for the CC reactor (Fig. 3B). The sharp convergence of electrode potentials with the increasing current density suggests that limitation or inhibition occurred, especially on the anodes in the CC reactors.

Fig. 3 (A) Power density and (B) electrode potential curves for the CC and CN reactors.

Nitrogen conversion and recovery

In one of the complete feeding cycles, the urine samples were taken for chemical analysis (Table 1). The initial concentration of TAN was 270 mg N L⁻¹, but the finial concentration increased to 3020 ± 214 mg N L⁻¹ (10 days) for the OC reactor, to 2890 ± 156 mg N L⁻¹ (10 days) for the CC reactor, and to 570 ± 87 mg N L⁻¹ (6.4 days) for the CN reactor. In fresh urine, most of the nitrogen existed in the form of urea. The increase of TAN was attributed to urea hydrolysis, while the slight difference in TAN concentration between the OC and CC reactors was possibly due to ammonium transport to the cathode with current generation in the CC but not in the OC reactor. It is assumed that ammonium was subsequently converted to volatile ammonia at the elevated pH.¹⁴ Although the increase of TAN was found in all the reactors, it was not followed by the increase of NO₂⁻-N or NO₃⁻-N. On the contrary, NO₂⁻-N was undetected in both the influent and effluent, and NO₃⁻-N (initial concentration, 0.58 mg L⁻¹) decreased by one order of magnitude, with 0.02 ± 0 mg L⁻¹ remaining in the CC and CN reactors. Nitrate might be reduced by denitrifying bacteria or the cathode because nitrate is also an alternative electron acceptor or oxidant.²⁹

Table 1 Key parameters measured for real urine before and after treatment

Unit	Reactor type	TN (mg N L^−1)	TAN (mg N L^−1)	NO2^−-N (mg N L^−1)	NO3^−-N (mg N L^−1)	TOC (mg N L^−1)	COD (mg N L^−1)
Fresh urine		5273	270	—	0.58	3613	8500
Treated urine	OC	3688 ± 432	3020 ± 214	—	—	1158 ± 130	1500 ± 100
	CC	3706 ± 360	2890 ± 156	—	0.02	538 ± 20	1400 ± 141
	CN	708 ± 230	570 ± 87	—	0.02	325 ± 59	1600 ± 182

---

TN removal efficiency (84.9% ± 2.2%) from urine was considerably higher in the CN reactor than it was in the CC (29.7% ± 6.7%) or OC (30.0% ± 8.2%) reactor. In the CN reactor, 52.8% ± 3.6% of TN was recovered in the form of NH₃-N, suggesting that about 32% of TN was probably lost through ammonia volatilization or bacterial nitrification/denitrification. The NH₃-N recovery rate for the CN reactor was 435.7 ± 29.6 gN m⁻³ d⁻¹, 2.3 times higher than that (131.6 gN m⁻³ d⁻¹) achieved in another study, where ammonia was recovered from the air-cathode via volatilization and subsequent absorption into an acid solution.¹⁵ Furthermore, COD removal rates of these reactors were also compared, which were 1078.0 ± 40.2 g m⁻³ d⁻¹ for the CN reactor, 710.0 ± 19.9 g m⁻³ d⁻¹ for the CC reactor and 700.0 ± 14.1 g m⁻³ d⁻¹ for the OC reactor.

Microbial community analysis

The microbial communities at the genus level were considerably diverse (Table S1†) on the anodes (A-OC; A-CC; and A-CN), with the species relatively evenly distributed (Fig. 4A). Generally, the composition and structure of the microbial community are closely related to the substrates used in MFCs. Compared to the non-fermentable substrates such as acetate, the composition or content of urine is more complex. According to the Urine Metabolome database (http://www.urinemetabolome.ca), the most abundant organic constituents (based on average values) of urine are urea (22.5 ± 4.4 mM mM⁻¹ creatinine), creatinine (10.4 ± 0.2 mM), hippuric acid (298.5 ± 276.8 μM mM⁻¹ creatinine) and citric acid (280.6 ± 115.2 μM mM⁻¹ creatinine). Members of Clostridia, Tissierella (25.0%; 13.4%; 12.8%), Finegoldia (8.1%; 17.6%; 4.8%), Anaerosphaera (2.1%; 6.2%; 2.9%) and Clostridium XII (5.3%; 4.4%; 4.0%) are four dominant genera detected on the anodes, and most of them are responsible for the fermentation of various organic acids, proteins and their hydrolytic products.^30–33 Among these, Tissierella species have been known to ferment urine creatinine to acetate, ammonia and other products.³⁴ Atopostipes (18.3%; 15.5%; 9.5%) and Facklamia (6.8%; 10.9%; 5.4%) are two dominant genera belonging to the class Bacilli, which can use carbohydrates to produce various fatty acids as well as urease.^35,36 They, therefore, are believed to be related to urea hydrolysis. Petrimonas, the main genus belonging to class Bacteroidia, was found on the anodes (8.8%; 1.2%; 0.8%). It can ferment carbohydrates and some organic acids and reduce nitrate to ammonium.³⁷ Another two genera, Oligella and Serpens, were found with the proportion increasing in the following sequence: A-OC (0; 0.5%), A-CC (2.0%; 9.7%), and A-CN (12.1%; 23.1%). Oligella is a member of the class Betaproteobacteria and has been reported to be capable of hydrolyzing urea.³⁸ Serpens, belonging to the class of Gammaproteobacteria, is considered to be the most likely exoelectrogen, because most bacterial species affiliated with class Gammaproteobacteria are electrochemically active bacteria. This may explain which bacteria further consumed the end products of fermentation or hydrolysis for current generation.

Fig. 4 Microbial community distribution for biofilms developed on electrodes (anode; cathode) at genus levels (A and B) of these three types of reactors (OC, open circuit; CC, closed circuit; CN, closed circuit and nitrogen purged). “A” and “C” indicate the anode and cathode samples, respectively.

Most genera present on the anodes were also found on the cathodes, but at least five dominant genera were uniquely on the cathodes (C-OC; C-CC; C-CN), namely, Pusillimonas (1.5%; 8.4%; 2.8%), Azorhizophilus (4.9%; 1.9%; 2.2%), Erysipelothrix (6.2%; 2.1%; 1.4%), Dokdonella (1.9%; 3.0%; 6.6%), and Sporanaerobacter (5.4%; 4.9%; 1.2%) (Fig. 4B). Most of them show an aerobic respiratory metabolism, and none of them have been reported to be related to nitrification or denitrification.

Discussion

Without nitrogen purging, the impact of real urine on MFC performance during long-term operation was investigated in our previous study, with the maximum power density and polarization curves measured at different times. The maximum power density decreased from 143.5 ± 4.9 mW m⁻² on day 32 to 51.1 ± 1.0 mW m⁻² on day 118 (Fig. 5A). Correspondingly, the anode potential at the specific resistance shifted toward more positive values over time (e.g., from −0.38 ± 0.01 V on day 32 to −0.28 ± 0.01 V on day 118, 1000 Ω) with simultaneous decrease of current density, suggesting that remarkable polarization occurred on the anode. However, there was no appreciable change in cathode potential (Fig. 5B). Given the ammonia accumulating in the chamber due to urea hydrolysis and its toxicity to bacterial activity, the decrease of maximum power density should be associated with long-term exposure of anodic biofilms to high concentration of ammonia, leading to the decline of anodic electro-activity over time. Therefore, the anodic electro-activity was the major limitation, leading to the decline in performance. In this study, the enhanced performance exhibited on the CN reactors was believed to be attributed to nitrogen purging, leaving less ammonia in the chamber and allowing for higher biological activity.

Fig. 5 (A) Power density and (B) electrode potential curves over time for urine-fed air-cathode MFCs.

The CVs and their first derivatives indicated the differences in bacterial electro-activity between the nitrogen-purged and non-purged reactors (Fig. S1†). CV revealed that the anodic current from the CN reactor was considerably higher than that from the CC reactor over a broad potential range (i.e., from −0.70 V to −0.05 V). The increase in anodic current from the CN reactor was consistent with the polarization data, indicating that purging the urine being treated inside the chamber with nitrogen could improve the system performance (Fig. 3). First derivative analysis of the CV data showed that with nitrogen purging, the peak intensity was more evident, especially at −0.67 V, compared to that without nitrogen purging, suggesting that redox activity of anodic biofilms in the CN reactor was enhanced due to the involvement of nitrogen purging.

The improved performance of the CN reactor was believed to be independent of dissolved oxygen (DO), because DO concentration in the CN reactor was slightly higher than in the CC reactor over time (Fig. 6). When the chamber was continuously purged with nitrogen gas, a portion of the DO would be carried away with volatile ammonia, causing a difference of oxygen partial pressure between the inside and outside. In this case, oxygen would continuously diffuse through the air-cathode into the chamber. The turbulent flow was believed to intensify oxygen diffusion. Furthermore, because the microbial community on the cathode of the CC reactor was dominated by aerobes (Fig. 4B), more oxygen was likely to be consumed before entering the non-purged chamber in comparison to that in the nitrogen-purged reactor. These may explain the difference of DO in both reactors.

Fig. 6 DO evolution over time in the CC and CN reactors in one complete batch cycle.

Generally, weak acids and bases are toxic for bacterial growth, because their undissociated forms dissipate proton motive force, leading to disturbance of energy metabolism.³⁹ Acting as a base, a certain fraction of ammonia remains undissociated depending on the pH at constant temperature and pressure. In this study, the pH of the effluent from the CN reactor reached up to the highest value of 9.1, followed by the CC (8.9) or OC (8.9) reactor, but this does not mean that the treated urine in the CN reactor had a NH₃-N concentration (214 mg L⁻¹) considerably higher than that in the CC (822 mg L⁻¹) or OC (1087 mg L⁻¹) reactor. Although the threshold ammonia concentrations that cause an inhibitory effect are not in agreement with published studies, which usually range from 500–3500 mg N L⁻¹,^15,40 the inhibition of exoelectrogenic activities is often observed in BESs.⁴¹ According to eqn (2), the ratio (δ) of NH₄⁺-N to NH₃-N could be simplified as a function (eqn (4)) of pH:

The ratio (δ_CN = 1.00) for the CN reactor was considerably smaller than that (δ_CC = δ_OC = 1.58) for the CC and OC reactors. The significant difference in ratio suggests that the equilibrium of urea hydrolysis (eqn (5) and (6)) in the CN reactor accelerated the conversion of urea by continuously carrying volatile ammonia out of the chamber in comparison to that in the OC or CC reactor. Thus, the lower ammonia content was the reason for the improved electro-activity or performance in the CN reactor.

NH_3(gas) + H₂O ⇄ NH₄⁺ + OH⁻ (6)

In addition, the nitrogen removed from the OC and CC reactors, as well as that lost in the CN reactor, is believed to be mainly related to volatilization through the air-cathode.⁴² As noted above, the TN removal efficiency for the OC reactor was similar to that of the CC reactor. It appears that the CC reactor failed to enhance ammonia volatilization through the air-cathode, although current generation from BESs could increase NH₄⁺/NH₃ transport to the cathode, where the conversion of NH₄⁺ to volatile NH₃ occurs as a result of an elevated pH near the cathode.⁴³ There is a possibility that after a long-term operation, the ammonium migration process that is driven by current generation was undermined for the non-purged (CC) reactors, because the biofilm activities were reduced over time. In another aspect, although NH₃/NH₄⁺ transport in the OC reactor just depended on diffusion, a process slower than ammonium migration with current generation,¹⁵ i.e. the enrichment of functional bacteria on the anode without current generation might have facilitated urea hydrolysis, leading to a high ammonia concentration gradient and hence the increase of ammonia transport. As can be seen in Fig. 4A, the genera such as Tissierella, Atopostipes and Petrimonas made up a large proportion of the microbial community on the anode of the OC reactor, which contrasted with that on the anode of the CC reactor. This may explain why the TN removal efficiencies in both types of reactors were comparable to each other. Given the salinity of real urine, there exists another possibility that ammonium, compared to the other ions, might contribute little to the charge balance in the CC reactor and therefore the NH₄⁺/NH₃ transport, making the ammonia lost through the air-cathode comparable to that in the OC reactor. Furthermore, the TN lost through the air-cathode of the CN reactor also showed similar removal efficiency with the other two types of reactors, but the cause remains unclear. In the CN reactor, nitrogen purging not only carried away a great part of volatile ammonia, but also improved bacterial activities, especially of exoelectrogens (e.g., the uncharacterized Serpens). The combination of BES with nitrogen purging offered a strategy for simultaneous recovery of ammonia and maintenance of BESs. These factors may explain why large amounts of nitrogen were removed from the CN reactor through nitrogen purging.

Besides ammonia content, salt precipitation might be another limiting factor responsible for the decreased anode electro-activities. The XRD analysis showed that the precipitates on the OC or CC reactors were mainly comprised of KMgPO₄·6H₂O and MgNH₄PO₄·6H₂O (Fig. 7). In contrast, few precipitates were formed on the anodes of the CN reactors. On an average, the molar ratios in urine are 50–100 : 10 : 1 for N : P : Mg, whereas the formation of MgNH₄PO₄·6H₂O needs equimolar amounts of Mg, NH₄⁺ and PO₄.^3–10 In this study, the dramatic decrease in NH₄⁺-N might not facilitate salt precipitation.

Fig. 7 X-ray diffraction patterns of precipitates deposited on the graphite fiber anode of the OC or CC reactors.

As the major product of urea hydrolysis, ammonia was not further converted into nitrite or nitrate. This was reflected by the microbial community analysis that no nitrifiers or Anammox bacteria were found on the anodes. Generally, the nitrification process only occurs on anodes enriched with some nitrifiers such as Nitrosomonas europaea. The lack of nitrifiers in this study might be related to the inoculums. Actually, biological hydrolysis of urea is an unfavourable process for nitrification.⁴⁴ There is gaining evidence supporting the claim that ammonia is a major factor limiting the activity of nitrifiers.^45,46 Generally, nitrite oxidizing bacteria are sensitive to NH₃-N in the range of 0.1–1.0 mg L⁻¹, while ammonia oxidizing bacteria are inhibited in the range of 10–150 mg L⁻¹.⁴⁷ Herein, the NH₃-N concentration in the CN reactor remained at 214 mg L⁻¹, although more ammonia was recovered in the absorption bottle. This may explain why no nitrifiers were detected on the electrode samples. The decrease of nitrate should be associated with a chemical reduction reaction, because no denitrifying bacteria were detected on the cathodes.

Conclusions

This study presents a possibility that ammonia inhibition of anode electro-activities could be mitigated concomitantly with ammonia recovery from urine in a nitrogen-purged BES. With nitrogen purging, the peak current from the CN reactor reached up to 0.43 mA, which contrasted to that (0.32 mA) of the CC reactor. Correspondingly, TN removal efficiency was 84.9% ± 2.2% for the CN reactor but was only 29.7% ± 6.7% for the CC reactor. In the CN reactor, 52.8% ± 3.6% of the TN was recovered in the form of NH₃-N, with a NH₃ recovery rate of 435.7 ± 29.6 gN m⁻³ d⁻¹. Ammonia content in the hydrolyzed urine was the determining element affecting anode electro-activities or performance of urine-fed BESs. The results show that nitrogen-purging offers a strategy for sustainable maintenance of BESs with simultaneous ammonia recovery.

Acknowledgements

This study was supported by the State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology (Grant No. 2015DX05) and by the National Natural Science Fund for Distinguished Young Scholars (Grant No. 51125033) and the National Natural Science Fund of China (Grant No. 51209061 and No. 51408156). The authors also acknowledged the supports from the International Cooperating Project between China and European Union (Grant No. 2014DFE90110).

Notes and references

1. J. Hanaeus, D. Hellstrom and E. Johansson, Water Sci. Technol., 1997, 35, 153–160.
2. T. A. Larsen and W. Gujer, Water Sci. Technol., 1996, 34, 87–94.
3. Z. R. Hu, S. Sotemann, R. Moodley, M. C. Wentzel and G. A. Ekama, Biotechnol. Bioeng., 2003, 83, 260–273.
4. S. M. Kotay, B. L. Mansell, M. Hogsett, H. Pei and R. Goel, Biotechnol. Bioeng., 2013, 110, 1180–1192.
5. T. A. Larsen, M. Maurer, K. M. Udert and J. Lienert, Water Sci. Technol., 2007, 56, 229–237.
6. J. A. Wilsenach, C. A. H. Schuurbiers and M. C. M. van Loosdrecht, Water Res., 2007, 41, 458–466.
7. B. Beler-Baykal, A. D. Allar and S. Bayram, Water Sci. Technol., 2011, 63, 811–817.
8. W. Pronk, M. Biebow and M. Boller, Environ. Sci. Technol., 2006, 40, 2414–2420.
9. K. Hirooka and O. Ichihashi, Bioresour. Technol., 2013, 137, 368–375.
10. G. L. Zang, G. P. Sheng, W. W. Li, Z. H. Tong, R. J. Zeng, C. Shi and H. Q. Yu, Phys. Chem. Chem. Phys., 2012, 14, 1978–1984.
11. S. Basakcilardan-Kabakci, A. N. Ipekoglu and I. Talini, Environ. Eng. Sci., 2007, 24, 615–624.
12. B. Beler-Baykal, S. Bayram, E. Akkaymak and S. Cinar, Water Sci. Technol., 2004, 50, 149–156.
13. R. Cord-Ruwisch, Y. Law and K. Y. Cheng, Bioresour. Technol., 2011, 102, 9691–9696.
14. P. Kuntke, M. Geleji, H. Bruning, G. Zeeman, H. V. M. Hamelers and C. J. N. Buisman, Bioresour. Technol., 2011, 102, 4376–4382.
15. P. Kuntke, K. M. Smiech, H. Bruning, G. Zeeman, M. Saakes, T. H. Sleutels, H. V. Hamelers and C. J. Buisman, Water Res., 2012, 46, 2627–2636.
16. H. Chen, P. Zheng, J. Q. Zhang, Z. F. Xie, J. Y. Ji and A. Ghulam, Bioresour. Technol., 2014, 161, 208–214.
17. B. Qu, B. Fan, S. K. Zhu and Y. L. Zheng, Environ. Microbiol. Rep., 2014, 6, 100–105.
18. Z. He, J. J. Kan, Y. B. Wang, Y. L. Huang, F. Mansfeld and K. H. Nealson, Environ. Sci. Technol., 2009, 43, 3391–3397.
19. J. Desloover, A. A. Woldeyohannis, W. Verstraete, N. Boon and K. Rabaey, Environ. Sci. Technol., 2012, 46, 12209–12216.
20. Y. Feng, X. Wang, B. E. Logan and H. Lee, Appl. Microbiol. Biotechnol., 2008, 78, 873–880.
21. Y. J. Feng, Q. Yang, X. Wang and B. E. Logan, J. Power Sources, 2010, 195, 1841–1844.
22. X. T. Zhou, Y. P. Qu, B. H. Kim, H. N. Li, J. Liu, Y. Du, D. Li, Y. Dong, N. Q. Ren and Y. J. Feng, RSC Adv., 2015, 5, 14235–14241.
23. V. J. Watson and B. E. Logan, Electrochem. Commun., 2011, 13, 54–56.
24. H. C. Angove, J. A. Cole, D. J. Richardson and J. N. Butt, J. Biol. Chem., 2002, 277, 23374–23381.
25. E. Shipitsyna, A. Roos, R. Datcu, A. Hallen, H. Fredlund, J. S. Jensen, L. Engstrand and M. Unemo, PLoS One, 2013, 8, 1–10.
26. C. H. Song, M. X. Li, X. Jia, Z. M. Wei, Y. Zhao, B. D. Xi, C. W. Zhu and D. M. Liu, Microb. Biotechnol., 2014, 7, 424–433.
27. Z. J. Wang, T. Lee, B. Lim, C. Choi and J. Park, Biotechnol. Biofuels, 2014, 7, 1–10.
28. K. H. Hansen, I. Angelidaki and B. K. Ahring, Water Res., 1998, 32, 5–12.
29. C. Fang, B. Min and I. Angelidaki, Appl. Biochem. Biotechnol., 2011, 164, 464–474.
30. Z. Zaybak, J. M. Pisciotta, J. C. Tokash and B. E. Logan, J. Biotechnol., 2013, 168, 478–485.
31. T. Goto, A. Yamashita, H. Hirakawa, M. Matsutani, K. Todo, K. Ohshima, H. Toh, K. Miyamoto, S. Kuhara, M. Hattori, T. Shimizu and S. Akimoto, DNA Res., 2008, 15, 39–47.
32. A. Ueki, K. Abe, D. Suzuki, N. Kaku, K. Watanabe and K. Ueki, Int. J. Syst. Evol. Microbiol., 2009, 59, 3161–3167.
33. A. S. Finch, T. D. Mackie, C. J. Sund and J. J. Sumner, Bioresour. Technol., 2011, 102, 312–315.
34. C. Harms, A. Schleicher, M. D. Collins and J. R. Andreesen, Int. J. Syst. Evol. Microbiol., 1998, 48(3), 983–993.
35. M. A. Cotta, T. R. Whitehead, M. D. Collins and P. A. Lawson, Anaerobe, 2004, 10, 191–195.
36. M. D. Collins, P. A. Lawson, R. Monasterio, E. Falsen, B. Sjoden and R. R. Facklam, J. Clin. Microbiol., 1998, 36, 2146–2148.
37. A. Grabowski, B. J. Tindall, V. Bardin, D. Blanchet and C. Jeanthon, IInt. J. Syst. Evol. Microbiol., 2005, 55, 1113–1121.
38. R. Rossau, K. Kersters, E. Falsen, E. Jantzen, P. Segers, A. Union, L. Nehls and J. Deley, Int. J. Syst. Bacteriol., 1987, 37, 198–210.
39. J. B. Russell, Appl. Environ. Microbiol., 1987, 53, 2379–2383.
40. H. W. Kim, J. Y. Nam and H. S. Shin, J. Power Sources, 2011, 196, 6210–6213.
41. R. C. Tice and Y. Kim, J. Power Sources, 2014, 271, 360–365.
42. J. R. Kim, Y. Zuo, J. M. Regan and B. E. Logan, Biotechnol. Bioeng., 2008, 99, 1120–1127.
43. J. R. Kim, Y. Zuo, J. M. Regan and B. E. Logan, Biotechnol. Bioeng., 2008, 99, 1120–1127.
44. B. Krogulska, H. Rekosz and R. Mycielski, Acta Microbiol. Pol., 1983, 32, 373–380.
45. J. Surmacz-Gorska, A. Cichon and K. Miksch, Water Sci. Technol., 1997, 36, 73–78.
46. A. C. Anthonisen, R. C. Loehr, T. B. S. Prakasam and E. G. Srinath, J. - Water Pollut. Control Fed., 1976, 48, 835–852.
47. D. J. Kim, D. I. Lee and J. Keller, Bioresour. Technol., 2006, 97, 459–468.

---

Footnote

† Electronic supplementary information (ESI) available: Estimators for evaluation of community diversity and richness; electrochemical analysis; key components measured in real urine. See DOI: 10.1039/c5ra11556f

---