Anoxic-biocathode microbial desalination cell as a new approach for wastewater remediation and clean water production

Due to global warming, population growth, urbanization and increasing consumption of water and energy, the world is ever more focused on the conservation of these, two resources. Although wastewaters are known to have high pollution potential, with the presence of pathogens, hydrocarbons, metals and nutrients, they are a potential source for renewable energy generation due to their high energetic value. In this context emerges the concept of bioelectrochemical technologies (BES). BES applications are attractive as a complement to traditional wastewater treatment technologies, reducing energy requirements as well as recovering resources and synthesizing new products by using wastes as raw material. However, the small amount of energy generated would be sufficient only for low-power applications. Alternatively, it would be an advantage to utilize the electricity to conduct desalination (Al-Mamun et al. 2018) or even to synthesize new products. In this context, we emphasize microbial desalination cells (MDCs). These devices enable conversion of the energy stored in wastewater directly into electricity by microorganism activity and utilization of it in situ to drive the desalination process, producing clean water (Wang & Ren 2013; Dong et al. 2017).

Numerous studies are proving the concept of MDCs using chemicals, e.g. potassium ferricyanide, as electron acceptor (Saba et al. 2017). However, due to their toxicity, new alternatives have been developed, such as biocathodes (Kokabian et al. 2018a, 2018b; Zuo et al. 2018) enabling nutrients removal and recovery and synthesis of valuable bioproducts. Therefore, biocathodes appear as a promising alternative, due to their potential for self-regeneration, scalability and sustainable nature (Al-Mamun et al. 2018). In biocathodes, the microorganisms accept the electrons directly from the electrode surface reducing the compounds of interest and, thus, improving the coulombic and desalination efficiencies (Wen et al. 2012).

According to the electron acceptor, they are classified into oxic or anoxic (Santoro et al. 2017; Al-Mamun et al. 2018). In oxic biocathodes, O₂ is the most popular electron acceptor due to its high redox potential (+0.82 V) (Logan et al. 2006), having bacterial or microalgae consortia as biocatalysts (Meng et al. 2017; Arana & Gude 2018; Zhang et al. 2019). However, its main disadvantage is associated with dissolved oxygen supply, increasing system operational costs. Due to these limitations, new research has been directed towards the development of anoxic biocathodes. In the absence of O₂, a variety of compounds (e.g. NO₃⁻, NO₂⁻, SO₄⁻, Fe, Mn, selenate, arsenate, urinate, fumarate and CO₂) could be employed as the final electron acceptor, in which nitrate presents redox potential (+0.74 V) comparable to O₂.

The use of denitrifying bacteria to reduce nitrate in BES dates back to 1966 (Lewis 1966). However, this concept has been experimentally validated only in recent years (Clauwaert et al. 2007; Srinivasan et al. 2016; Wang et al. 2016). Recently, Kokabian et al. (2018b) developed an MDC using a consortium of anammox bacteria as biocatalysts achieving a maximum current density of 0.814 A m⁻³ (0.092 W m⁻³), and ammonium nitrogen conversion higher than 90% with nitrate accumulation and desalination efficiency of 25.5%.

Because of the above-mentioned points, the development of biocathode MDCs is emerging as a promising alternative; however, this process needs to be better understood and exploited, particularly when it comes to the establishment of anoxic biocathodes for nitrate reduction via the autotrophic denitrification process (Cecconet et al. 2019). In this study, we established a novel MDC operating with an anoxic biocathode to remediate carbon- and nitrogen-rich wastewaters, generating bioelectricity with additional salt removal, aiming for the reuse and conservation of water resources. Investigations in this field are strongly helpful, bringing new possibilities to couple such systems with those already existing, reducing energy requirements as well as recovering resources and synthesizing new products by using wastes as raw material.

The MDC operating with an anoxic biocathode (Anox-Bio-MDC) reactor was constructed of polyvinyl chloride, consisting of three chambers with the following net liquid volume: anode (0.8 L), cathode (0.8 L) and desalination (0.31 L). The desalination chamber was separated from anode and cathode by a compartmental anion exchange membrane (AEM; AMI-7001S) and cation exchange membrane (CEM; CMI-7000S), spaced 5 cm from each other. Commercial granular activated carbon (GAC, ∼670 μm) produced from coconut shell (Smart Carbon, Brazil) packed in a stainless-steel metallic fabric (mesh 200 – Telas Rocha Ltd, Brazil) cartridge (length: 15.0 cm, diameter: 2.0 cm, GAC mass: 15 g) was used to build electrodes. Six packaged cartridges were added to each chamber. This strategy was chosen aiming to increase the electrode contact area available for microorganisms and also to avoid loss of GAC mass during system operation. To collect electrons, we used stainless-steel fabric in the cartridges and connected it to an external cable collector. The bioreactor configuration is shown in Figure 1.

The anode was inoculated with a mix of anaerobic sludge from a municipal wastewater treatment plant and denitrifying sludge from a shrimp wastewater treatment pilot system (SWTP) (1:1, v/v, volatile suspended solids 1.70 ± 0.2 g L⁻¹), while the cathode was inoculated only with denitrifying sludge from the SWTP with the same biomass concentration. This inoculum source was selected as the bacteria were already adapted to salinity >15‰. Nutrient medium was based on Lovley & Phillips (1988), containing (per L): 2.5 g NaHCO₃, 0.1 g CaCl₂.2H₂O, 0.1 g KCl, 0.6 g NaH₂PO₄.H₂O, 1.87 g Na₂HPO₄.12H₂O, 0.1 g NaCl, 0.1 g MgCl₂.6H₂O, 0.1 g MgSO₄.7H₂O, 0.005 g MnCl₂.4H₂O and 0.05 g of yeast extract. The same medium composition was adopted for anode and cathode, at pH 7.0; meanwhile, acetate (2.0 g L⁻¹) was the electron donor and nitrate (0.05 g L⁻¹) was the electron acceptor.

To promote the electroactive biofilm growth and acclimatization (Babauta et al. 2012), the system was initially operated under open-circuit voltage (OCV), enabling the development of a stable biofilm on the electrode surface through natural redox processes. After voltage stabilization, external resistors (560, 100 and 22 Ω) were connected between the anode and cathode, closing the circuit (CCV) and improving the electron transfer through the circuit. In both operational stages, anode and cathode chambers were operated under fed-batch mode with cycles of 12 hours and a feed rate of 0.6 L d⁻¹ (hydraulic retention time, HRT of 1.33 days), while the desalination chamber was operated under fed-batch mode with cycles of 8 days and a feed rate of ∼0.039 L d⁻¹ (HRT of 8 days). It is worth mentioning that HRT was computed by the relation V/Q, where V represents the volume of the system (L) and Q is the influent flow-rate (L d⁻¹).

Kinetic assays were performed to achieve substrate conversion rates. Here acetate is expressed in terms of COD (chemical oxygen demand), nitrate in terms of NO₃^–-N, and saline solution in terms of electrical conductivity (EC). For anode and cathode kinetics, the concentration of the saline solution was fixed at 10 g NaCl L^−1. When the system achieved operational stabilization at a 22 Ω resistor (condition that gave us the best carbon and nitrogen conversion rates (‘Organic matter conversion’ and ‘Nitrogen conversion’ sections)), the influence of salt concentration on desalination performance was investigated. The following saline solutions were evaluated (g NaCl L⁻¹): A (10.00), B (20.00) and C (35.0). Additional tests were performed by replacing the synthetic solution with seawater (D).

COD and nitrogen species (NH₄⁺-N, NO₂⁻-N and NO₃⁻-N) were determined through the colorimetric method (APHA 2012). For voltage and EC measurements, a device based on an Arduino microcontroller was constructed. For data acquisition and monitoring in real-time (sampling interval of one second), custom software was developed using Python (available at https://github.com/simoneperazzoli/dracarys-project). For EC measurements, an analog EC probe designed for Arduino microcontrollers (Analog EC Meter, DFRobot^©) was used. Polarization curve tests, which are used to characterize current as a function of voltage, were carried out when the system reached stability at OCV condition, according to Watson & Logan (2011) methodology. Each resistance was tested for two consecutive operational cycles (24 hours) to ensure a stable voltage response. The following resistors were applied (Ω): 10,000, 5,600, 1,000, 560, 220, 100, 47, 22 and 4.6.

To investigate the biofilm formation, scanning electron microscopy (SEM) analysis was performed. Anode and biocathode samples were prepared according to Perazzoli et al. (2018). Samples were examined under a JEOL microscope (SEM JSM-6390LV), operated at 15 kV (LCME/UFSC, Brazil). The microbiological profile of the biofilm developed on the surface of the electrodes was analyzed through 16S rDNA sequencing analysis, according to Christoff et al. (2017). Detailed information about the sample preparation and procedures are presented in the Supplementary Information.

All data obtained were post-processed using Python software with NumPy and SciPy libraries. In the case of voltage data (obtained in digital form), the Savitzky–Golay smooth filter (Savitzky & Golay 1964) was applied, to minimize random noise effects.

The first stage of Anox-Bio-MDC reactor operation was under OCV condition, where current is null and the voltage produced is the maximum. As presented in Figure 2(a), initially, there is an adaptation or lag phase that goes until the third day. The higher bacterial activity (maximum voltage rate of 91 mV d⁻¹) was observed between the fourth and eighth days of inoculation. After this period, the system reached the steady-state phase, and its maximum potential was achieved in the twenty-sixth day (563 mV). After that, the voltage remained stable at ∼544.5 mV. These values are higher than previously reported for bio-MDCs. By incorporating Nanochloropsis salina into the cathode of an MDC, Girme (2014) observed a maximum OCV potential of 98.2 mV. Here, our strategy proved to be efficient, allowing the development of a stable microbial community (Babauta et al. 2012).

The polarization curve is presented in Figure 2(b). The maximum power density observed was 0.425 Wm⁻³, which is 2.8 times higher than reported for an MDC operating with Chlorella vulgaris sp. biocathode (Kokabian & Gude 2013) and 4.6 times higher than reported for an MDC operating with anammox biocathode (Kokabian et al. 2018b). Internal resistance (Ri) was 115.7 Ω, which is close to that reported by Meng et al. (2017) (94.2 Ω) and 26 times lower than observed by Kokabian et al. (2018b) (3,101 Ω), indicating good performance of our system.

According to Ohm's Law, the external resistance (Re) controls the electron flow from anode to cathode, affecting directly the potential and the current passing through the system. As reported in Figure 3, as the Re load decreases, there is a significant drop in voltage values, contrary to current density behavior, which gradually increases up to 2.61 A m⁻³ (Table 1). These results show that the ohmic losses were dominant in this operational stage (Lefebvre et al. 2008). At 22 Ω resistance, there were significant variations in the voltage, current and power density values (Figure 3(a) and 3(b)). This is due to the conductivity of the saline solution. Thus, as the conductivity in the desalination chamber increases, the higher is the ion migration between the chambers and, therefore, there is a higher internal current consumption to perform the desalination process. Also, a significant drop in power density was observed. This is because the system was operated with Re lower than Ri, as previously reported by Logan (2008). Electric coefficients obtained during reactor operation are presented in Table S1 (Supplementary Information).

According to the kinetic assays, the highest acetate consumption rate was observed in the first hours of the cycle (Figure 4(a)). From these assays, the maximum COD conversion rates ( −r_COD) were obtained. As presented in Figure 4(b), for OCV condition, the −r_COD was 212.4 mg L⁻¹h⁻¹, while it was 37.95% higher (293.01 mg L⁻¹h⁻¹) for 22 Ω condition. COD removal efficiencies were improved from 75.7% to 99.85%. The pollutant removal was improved with the load resistor decrease. This strategy allows the reduction of the overpotentials, increasing, therefore, the electron transfer rate through the circuit (Katuri et al. 2011). These values are in agreement with Zhang et al. (2019) and higher than those previously reported for biocathode MDCs (see Table 2). COD removal efficiencies of between 40 and 60% were achieved by Kokabian et al. (2018a, 2018b). Meng et al. (2014, 2017) evaluated two MDC configurations to treat dewatered sludge, with organic matter removal efficiencies between 14.7 and 25.7%.

Coulombic efficiency was improved from 5.01 to 28.32% by reducing Re load. Besides the improvement, 71.68% of oxidized organic matter was not converted into electricity, indicating the COD removal by non-electrogenic mechanisms (Cecconet et al. 2018; Molognoni et al. 2018). According to Vilajeliu-Pons et al. (2016), an alternative for improving the CE could be the application of a variable resistance control to enhance the electron transfer, as intermittent electric connection allows higher current production since both capacitive and faradaic currents are harvested.

Electroautotrophic denitrification processes have gained widespread attention in recent years (Chen et al. 2016; Xu et al. 2017; Cecconet et al. 2019). In the present study, during the startup and acclimatization, there was observed a gradual increase in denitrification rates, indicating the establishment of electroactive biofilm. From the kinetic assays, the maximum nitrate conversion rates (r_NO3) were obtained (Figure 4(c)). The r_NO3 for OCV condition (Figure 4(d)) was 3.65 mg L⁻¹h⁻¹, while the same for 22 Ω condition was 96.44% higher (7.18 mg L⁻¹h⁻¹). NO₃^–-N conversion efficiencies were improved from 60.84% to 92.11%, which is 1.31 times higher than observed by Kizilet et al. (2015) in a biocathode MFC.

Production and accumulation of nitrite (Figure 4(e)) was also observed. In the OCV conditions, an amount of 2.04 mg L⁻¹ (r_NO2 = 0.217 mg L⁻¹h⁻¹) accumulated at the end of an operational cycle. For CCV conditions, the nitrite accumulation reached 2.74 mg L⁻¹ (r_NO2 = 0.365 mg L⁻¹h⁻¹) for 560 Ω and 6.11 mg L⁻¹ (r_NO2 = 0.895 mg L⁻¹h⁻¹) for 100 Ω, respectively (Figure 4(f)). These results indicate the denitrification process occurred incompletely (Desloover et al. 2011; Srinivasan et al. 2016). However, at 22 Ω resistance, nitrite accumulation was not observed. According to Clauwaert et al. (2007), high resistor load contributes to the accumulation of intermediate compounds in the denitrification process, such as nitrite and nitrous oxide. Thus, an alternative to solve it is to operate the system with lower resistor load, as reported here. It should be mentioned that ammoniacal nitrogen was not detected.

In MDCs, the desalination process occurs by two main phenomena: osmotic pressure and ionic concentration difference between compartments. Here, the EC for anolyte and catholyte was 8.51 ± 0.01 and 6.49 ± 0.08 mS cm⁻¹, while the saline solution had an initial EC of 16.58 ± 0.11, 38.84 ± 0.12, 61.35 ± 0.32 and 61.89 ± 0.22 mS cm⁻¹ for solutions A, B, C and D, which are higher, compared to the anolyte and catholyte at the beginning of the experiment. This difference could lead to the occurrence of natural osmosis. Thus, the liquid of the lowest conductivity solution migrates to the solution with higher conductivity until the balance between the solutions is established.

The second factor is related to the ion migration between cell compartments. As biological degradation of organic compounds in the anode results in the release of protons and electrons, the anionic species migrate from the desalination chamber to the anode through the AEM. Similarly, the cationic species migrate from the desalination chamber to the biocathode through the CEM. Therefore, the catalytic activity of bacteria, substrate utilization and formation of metabolites play a vital role in forming an electrochemical gradient that will establish the rate of the desalination process (Ashwaniy & Perumalsamy 2017). To assess it, desalination kinetic assays were conducted. As presented in Figure 5, by increasing EC from solution A to D, the desalination rate ( −r_d) was increased 6.78 times (from 0.95 to 6.43 mS cm⁻¹ d⁻¹).

A similar trend was observed for desalination efficiencies and these values agree with those previously reported for biocathode MDCs, as presented in Table 2. Desalination efficiency was improved from 24.42% to 39.71% for solution A to C, respectively. When NaCl solution was replaced by seawater, an increase of 10.02% in desalination efficiency (43.69%) was observed. These results are promising; however, improvements are still needed, especially concerning the increase of the desalination efficiency and reduction of the HRT of the desalination chamber. According to Gude et al. (2013), desalination performance in MDCs could be improved by inserting multiple pairs of ion-exchange membranes between the anode and cathode chambers. This strategy improves the charge transfer efficiency and allows the saline water to flow through a series of MDCs leading to more salt removal. In addition, increasing the number of cell pairs reduces the voltage required in each cell, allowing an energy gain.

Moreover, with the use of thin ion exchange membranes and desalination chambers, the internal resistance is reduced and more efficient separation of ions and water desalination can be achieved (Kim & Logan 2011). Kinetic coefficients are presented in Table S1. It is worth mentioning that, during OCV operation, no changes were observed in the conductivity of the desalination chamber solution. This result agrees with that previously reported by Cao et al. (2009) and is because under these conditions, there is no current passing through the circuit and, therefore, the ion migration is minimal.

In MDCs, the biofilm developed on the anodic electrode acts as a catalyst, aiding the bacterial respiration, which in turn produces more current (Baranitharan et al. 2015). These bacteria transfer extracellularly the electrons obtained in the respiration process towards an exogenous electron acceptor. Thus, the highest power densities are produced by inoculating the anode with a rich and diverse source of bacteria, such as sludge, soil or sediments (Logan 2009). On the other hand, the biofilm developed in the biocathode accepts directly the electrons from the electrode surface, thus reducing compounds (Lovley & Nevin 2011). However, in contrast to exoelectrogens, there are few studies related to electrotrophic characterization (Vilajeliu-Pons et al. 2016).

According to literature, the microbial community composition of biofilm is affected by the inoculum source, type of microorganisms (Gram-positive or Gram-negative), nature of microbial culture (pure or mixed) and pre-enrichment and startup strategies (Molognoni et al. 2014; Saratale et al. 2017). Thus, to investigate the morphology of bacterial biofilm developed on the electrode surface, SEM was conducted.

In Figure 6, we can observe the development and enrichment of exoelectrogens (Figure 6(a) and 6(b)) and electrotrophs (Figure 6(c) and 6(d)) with biofilm formation on granular activated carbon electrodes. Electron micrographs show unique biofilm structures and cell shapes. The anode was covered with short-rod and coccoid bacteria, which are similar in cell shape and arrangement to exoelectrogenic bacteria (Bond & Lovley 2003). On the other hand, the nitrate-fed biocathode presented a lower diversity, with a predominance of coccoid-shaped bacteria.

A relatively higher bacterial abundance was developed in the anode chamber than in the anoxic biocathode chamber. The dominant phyla were Proteobacteria (88.45%), where α-Proteobacteria (53.24%), β-Proteobacteria (15.01%), δ-Proteobacteria (8.86%), γ-Proteobacteria (7.6%) and ɛ-Proteobacteria (3.75%) were the classes. Other phyla presented in minor proportions were Thermotogae (2.87%), Bacteroidetes (2.74%) Euryarcheota (2.41%), Actinobacteria (1.75%) and Firmicutes (1.72%) (Figure S2(a)). These results are in agreement with the literature, as there are several reports on BES containing diverse microbial communities, in which electrical current generation has been shown by Proteobacteria, Firmicutes and Acidobacteria phyla (Li et al. 2014; Zhang et al. 2015, 2016).

Herein, the main genera identified in the anode chamber are presented in Figure 7(a). Paracoccus showed the highest number of sequences (2,531), followed by Stappia (1,414), Desulfomicrobium (1,268), Rhizobium (1,124), Alcaligenes (997) and Aquamicrobium (963). Paracoccus and Stapia genera belong to the Rhodobacterales order. Paracoccus spp. include common types of denitrifier species (e.g. P. pantotrophus, P. denitrificans) which can grow under aerobic or anaerobic conditions. However, these bacteria have also the ability to oxidize carbon and sulfur compounds and produce bioelectricity, as previously reported (Jothinathan & Wilson 2018).

Desulfomicrobium is a strictly anaerobic genus of sulfate-reducing bacteria, which has been reported to consume organic substances in bioelectrochemical systems (Cao et al. 2018; Gacitúa et al. 2018). The Rhizobiales order was also identified, having Rhizobium spp. as the main representative (7.8%). These bacteria were previously identified in the anode biofilm enriched from rice paddy soil (Ishii et al. 2008) and also in a microbial fuel cell for Congo red decolorization (Hou et al. 2011), suggesting a possible involvement of these microorganisms in the current generation.

Alcaligenes composed 6.9% of the microbial community diversity. These are facultative anaerobic bacteria that are able to produce hydrogen, denitrify and also produce electrical current. In the latter case, the electron transfer is associated with the presence of plastocyanin, a mediator excreted by these species (Young et al. 2011). Aquamicrobium spp. was identified in proportions similar to that reported for BES treating recalcitrant organics with simultaneous electricity generation (Cheng et al. 2015). Several other microorganisms related to the current production were identified in a minor proportion, such as Brevundimonas (3.6%), Thauera (3.4%), Pseudochrobactrum (3.3%), Devosia (3.3%), Delftia (3.2%), Mesotoga (3.0%), Methanosarcina (2.5%), Halothiobacillus (2.4%), Macellibacteroides (2.0%), Pseudomonas (1.7%), Alcanivorax (1.6%) and Achromobacter (1.5%) (Wenzel et al. 2017; Zhang et al. 2017). Studies have shown that current generation by Delfia species is associated with the excretion of secondary metabolites, as occurs with Pseudomonas spp. through phenazines excretion (Qiao et al. 2017).

Halothiobacillus neapolitanus presence suggests potential for CO₂ fixation (Menon et al. 2008), while the presence of bacteria belonging to the genera Thauera, Pseudochrobactrum, Alcanivorax and Achromobacter suggests the potential for remediation of environments contaminated by metals, phenols, hydrocarbons and quinones with concomitant bioelectricity generation and desalination (Wang et al. 2017). On the other hand, the presence of Macellibacteroides, Mesotoga and Methanosarcina genera indicates the presence of fermentative and methanogenic bacteria competing with the exoelectrogenic ones for the substrate, because a large fraction of the added substrate (71.68%) was not converted into electricity (Molognoni et al. 2018). It is worth mentioning that Geobacter and Shewanella species, model microbes for exoelectrogenic studies (Rotaru et al. 2014), were not observed in this study, probably due to the inoculum nature.

Denitrifier identification is essential to reveal the microbiological mechanisms and to optimize the denitrification process (Xing et al. 2018). Proteobacteria phylum was dominant in the anoxic biocathode (97.13%), where α-Proteobacteria (75.16%) and β-Proteobacteria (20.89%) were the dominant classes. Firmicutes phylum was also identified (see Figure S2(b)). These results confirm that α-Proteobacteria classes can improve current production and nitrate removal performance (Qiao et al. 2017; Sun et al. 2017). The dominant species were those belonging to the Paracoccus genus with 30,542 sequences (Figure 7(b)).

Unlike the anode, the anoxic biocathode showed less diversity, and Paracoccus pantotrophus was the dominant species (70.98%), confirming the predominance of coccoid bacteria observed in SEM micrographs (Figure 6(d)). Cheng et al. (2017) reported the ability of Paracoccus species to receive electrons from the surface of the cathodic electrode in MFCs conducting electroautotrophic denitrification, as occurs with Alicycliphilus species (Tang et al. 2017), identified in less proportion (1.4%). Paracoccus pantotrophus is among the dominant mixotrophs, which can grow autotrophically, heterotrophically or mixotrophically, both under aerobic or anaerobic conditions, which justifies its presence in the anode. When these microorganisms grow autotrophically (as in the biocathode), the respiratory metabolism can use nitrate, nitrite or nitrous oxide as the final electron acceptor enabling the conversion of most of the oxidation products into nitrogen gas (Liu et al. 2017).

A considerable amount of Nitrosomonas spp. was identified, including N. europaea (16.30%) and N. eutropha (2.30%). These species can grow in aerobic or even anaerobic ammonia oxidation. However, under anoxic conditions, these bacteria can also grow via denitrification. In this case, nitrite is used as the final electron acceptor (Schmidt et al. 2004). This is explained by the prevalence of cytL and cytS genes, suggesting their involvement in the oxidation/reduction and electron transfer reactions for energy generation (Caranto et al. 2016). Bacillus species were identified in the proportion of 2.6% (1,044 sequences). Studies have shown that these microorganisms can grow under anoxic conditions, using nitrate or nitrite as a final electron acceptor, in addition to producing electric current (Yoganathan & Ganesh 2015). Other genera related to the denitrifying activity (3.9%) were identified in smaller proportions (>100 sequences each). The main microorganism species identified are presented in Figure S3.

This study demonstrates the potential of MDCs operating with anoxic biocathode for electroautotrophic nitrate reduction as an eco-friendly technology to remediate carbon- and nitrogen-rich wastewaters, enabling energy recovery and in situ desalination. Our results show an additional power generation of 0.425 W m⁻³ with suitable salt removal efficiencies. Furthermore, carbon and nitrate conversions higher than 90% were observed.

Despite the challenges related to the implementation of such technologies, this study brings new opportunities for the development and application of bioelectrochemical reactors for wastewater treatment and cleaner water production.

Authors thank financial support from CAPES and CNPQ.

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/wst.2020.134.