Long-term bio-power of ceramic Microbial Fuel Cells in individual and stacked configurations

In order to improve the potential of Microbial Fuel Cells (MFCs) as an applicable technology, the main challenge is to engineer practical systems for bioenergy production at larger scales and to test how the prototypes withstand the challenges occurring during the prolonged operation under constant feeding regime with real waste stream. This work presents the performance assessment of low cost ceramic MFCs in the individual, stacked (modular) and modular cascade (3 modules) configurations during 19 months utilising neat human urine as feedstock. During 1 year, the performance of the individual MFC units reached up to 1.56 mW (22.3 W/m3), exhibiting only 20 % power loss on day 350 which was significantly smaller in comparison to conventional proton or cation exchange membranes. The stack module comprising of 22 MFCs reached up to 21.4 mW (11.9 W/m3) showing power recovery to the initial output levels after 580 days, whereas the 3-module cascade reached up to 75 mW (13.9 W/m3) of power, showing 20 % power loss on day 446. In terms of chemical oxygen demand (COD) removal, the 3-module cascade configuration allowed for a cumulative reduction >92 %, higher than that observed in the single module (56 %).


Introduction
Microbial Fuel Cells (MFCs) utilise organic feedstocks such as urine as a fuel for direct electricity production by employing anode respiring microbes that convert organic matter into electrons while treating waste. In terms of effective applicability of MFC systems and the reduction of production costs, there is still much room for improvement in the reactor design and scalability process (Janicek et al., 2014). Many operational and design parameters, that might affect the power output by MFCs, can only be effectively tested in large-scale systems.
Real-world implementation of MFCs requires that power generation and treatment efficiency be obtained with large-scale reactors, operated under realistic conditions (Li et al., 2014;Logan et al., 2015). Different approaches can be employed for optimising MFC technology to allow their scaled-up for practical applications and one being miniaturisation and multiplication of small-scale units (Ieropoulos et al., 2010) as it has been shown that higher energy density levels and optimum biofilm/electrode surface area-to-volume ratios reside within smaller scale MFCs. In order to scale-up the MFC technology towards real-world applications and reach usable power levels, the MFCs units can be operated in collectives (stacks) of small-scale MFC units (Ieropoulos et al., 2008(Ieropoulos et al., , 2009Mateo et al., 2018) using affordable and durable materials (Choudhury et al., 2017;Li et al., 2017). Ceramic materials have shown to be suitable and costeffective separators for MFCs (Behera et al., 2010;Winfield et al., 2013Winfield et al., , 2016. Economic optimisation and the selection of the best stack structure becomes more essential where largescale power production using MFC systems. It is necessary to conduct sufficiently long-term experiments to determine and understand the long-term behaviour, stability and potential challenges. The choice of ceramic as a separator seems suitable for larger applications as over 60 % of the material cost of the MFCs in the recent large scale application was due to the cation exchange membrane (Ge and He, 2016) therefore further study into the durability, performance and properties in long-term operation is much needed.
Additionally to the generation of electricity, MFC are used for breaking down and removing organic waste material from the processed substrate (Delaney et al., 2008;Habermann and Pommer, 1991). In this relationship between power and treatment, the higher the power output, the higher the rate of electron-abstraction from organic substrate that has been processed by the electroactive biofilm and the greater degree of waste removal (Bourgeois et al., 2001;Zhang et al., 2019). Over the recent years more practical demonstrations have been reported, however, sustainable processing of given wastewater under various operational conditions must take the longevity of the system into account. The practical application of MFCs is currently restricted by the poor long-term stability of air cathodes, which has proven closely related to the scaling and biofilm growth on the cathodic catalyst layer (An et al., 2017a;Babauta et al., 2013;Santini et al., 2015). As well as the cathode, also the separator suffers from blockages (Flimban et al., 2018) due to precipitates causing transfer limitation of cations and decreased diffusion coefficients. However, this can be prevented by appropriate design and moisture control to allow for the catholyte production and extraction that is sustaining the electroosmotic drag washing the deposits away from the cathode surface (Gajda et al., 2015(Gajda et al., , 2018b and simultaneously producing antimicrobial agents . In ceramic based pilot studies this approach showed potential for the remote power generation (Business Leader News, 2017; and implemented in small scale stack prototypes (Gajda et al., 2018c). However, it is little known about the longevity of the system, its components and peripherals, and this study is aiming to address challenges ahead of the practical implementation of the MFC technology in real-world scenarios in the future. The novel approach in this work uses low cost materials in MFC and performs a long-term evaluation which is rarely described in the literature. For this purpose, the long-term performance of the following MFC configurations: individual units, 22-MFC module and 3module cascade in terms of power output and urine treatment capacity has been assessed. The tests were running for up to 19 months in fed-batch mode with real human urine and the stability of the systems and their resilience against adverse conditions were analysed.

Individual MFC units configuration
The individual MFC units used in this study were constructed from terracotta cylinders as described previously (Gajda et al., 2018a) using terracotta cylinders (H:50mm) sealed at one end. The anode was made of 594 cm 2 piece of carbon veil fibre (20g/m 2 , PRF Composites, UK) which was folded and wrapped around the ceramic cylinder. Carbon veil is used here as flexible substrate to support electrochemically active biofilm. The cathodes were prepared using activated carbon paste applied onto wet proofed carbon veil fibre as previously described (Gajda et al., 2015). The cathode with a total area of 22.5 cm 2 was inserted into the inner chamber of the cylinder and connected with stainless steel crocodile clip. The MFC was hosted into the plastic lid that was holding the cylinder in place avoiding anolyte overflow (Fig 1) and the lid was placed inside the polycarbonate plastic container. The anode chamber had 4 mm plastic inlet connectors attached at the lower end of the plastic containers and T-connectors attached at the outlet to allow passive overflow of the anolyte. MFCs effective volume was 70 mL in each bioreactor. The anode and air-cathode were connected with stainless steel wire with an external resistance of 100 Ω. Continuous flow was maintained using a peristaltic pump (205U, Watson Marlow, UK) which was feeding urine from the inlet tank into the inflow tubes of the MFCs at a flowrate of 9 mL/h.

MFC Modules
MFC-stack was assembled using Euro stacking container (Plastor, UK) of the dimensions 300 x 200 x 118 mm. The container was used as the chassis with attached pipes as the inlet and outlet to allow the electrolyte flow. 22-MFC units were installed inside the module using sealed acrylic lid holding each MFC in position with all the anode wires connected underneath to the main anodic connection on the outer side of the module. The MFCs were arranged in four rows where all the anodes and the cathodes were connected in parallel electrical configuration. The total volume of the anodic chamber was 1.8 L. All the cathodes were connected above the lid using stainless steel wires and two main connection leads cables on the sides of the module connected towards the main cathodic connection outside of the module (Fig. 1).

Modular Cascade
Three modules as described above were stacked vertically using T-connectors and flexible tubing to allow the anolyte flow. The outlet of the first module was connected to the inlet of the second, whereas the outlet of the second module was connected to the inlet of the third module in the stack. The outlet of the third module was led into the outlet collection tank.
Electrically, the stack was in parallel configuration using copper wires for the cathode parallel connection and stainless steel wire for the anodic connection. Main output from the 3-module stack was set up using a banana to alligator clip test lead cables connected to the decade box (ELC DR 05 Decade Box, RS, UK) and to the Agilent data acquisition equipment (Agilent LXI 34972A; Farnell, UK) and a PC to record voltage output.

Operation of MFC stacks and Analysis
Anaerobic activated sludge, obtained from the Wessex Water wastewater treatment plant in Saltford, UK, was mixed 1:1 with urine and used as the inoculum. Urine was collected anonymously from healthy volunteers, pooled together and stored in a 40 L collection tank (pH

Long-term performance of the Individual MFC-units
Polarisation of the systems were performed once the MFCs reached stability after 3 months of operation. The open circuit voltage (OCV) was 692 mV (Fig 2A), reaching a maximum power output by the system of 1.63 mW at a current level of 4.81 mA. Regarding the temporal performance, the maximum value of power achieved was 1.56 mW corresponding to 22.3 W/m 3 (Fig. 2B), which is similar value to previously reported data [23] in the same type and size of the MFC reactors. Individual MFC units were operated under continuous flow conditions and reached 1.55 mW of power on the day 50, achieving stable output. On the day 60, the power diminished due to feedstock depletion, after which the inlet tank was replenished with urine and the performance recovered to previous level, which is in agreement with previous work conducted in a continuous flow using human urine, reporting no power variance during 3 months of operating time (Salar Garcia et al., 2019). As the feedstock flow and lab temperature were kept constant, the fluctuations must be related to the periodical feedstock depletion in the feeding tank as well as changes in feedstock composition, which is dependent on diet, time of the year and other factors. On the day 350, the performance reached up to 1.24 mW, which is only 20 % lower than the peak power output. These results suggest good reliability of the ceramic architecture in comparison to a commercial proton exchange membranes that reported 55 % loss in performance in 6 months (Miskan et al., 2016a).
Membrane biofouling is an inevitable process in two-chambered MFCs utilising proton or cation exchange membranes (Choi et al., 2011) leading to physical blockage of charge transfer (Xu et al., 2012). As it was shown, ceramic membranes offer robust and prolonged stability to perform to a similar level to CEM by the eighth month (Winfield et al., 2013). Moreover, in this study high performance, above 1 mW of power, is maintained even after 1 year of operation. Previous long-term studies have reported deteriorating performance levels also due of the ceramic cylinder that was exposed to air. However, the cathode chambers showed accumulation of liquid catholyte as previously described (Gajda et al., , 2018a(Gajda et al., , 2018b that kept the inner cathodes sufficiently hydrated yet not completely flooded. The production of the catholyte maintained the cathode clean of any deposits in the long term operation of 1 year [22] and it might be due to antimicrobial properties  of the electrochemically produced liquid. It was also observed that the liquid kept the inner electrode clean of the salt deposits apart from the top part of the cylinder that was exposed to air. This might be the reason for good longevity of the system preventing from biofouling, scaling and clogging of the cathode.

Long term performance of the Module
The performance of the MFC module in long-term operation was assessed during 19 months of fed-batch mode. The polarisation curve shows an OCV of 671 mV and the peak power at  After the start of the experiment, the performance reached up to 19.0 mW after which point diary wastewater was tested (pH 5.5) for the period of two months and low power production was recorded in comparison to urine as a feedstock. When urine was reintroduced, the power output immediately increased and stabilised at the maximum level of approximately 19 mW after 100 days of operation. After 188 days of operating time the stack was removed from the data logger and connected to the 5 module cascade (data not shown) in the adverse conditions which might be the reason for the module underperformance that followed after day 235 when the stack was reconnected to the logger (Fig.3). It showed some gradual recovery over the next 250 days, however the stack did not reach the initial power level. On the day 395 the Constant Voltage Load (CVL) external circuitry was connected to the module where the voltage level was kept at 400 mV, dynamically adjusting the load as previously described (Walter et al., 2019), however this also did not improve the output. It might indicate the effect of the adverse conditions that changed the microbial diversity and the presence of non-electroactive microbes metabolising substrates to survive rather than generating current (Nimje et al., 2012). The stack was left for prolonged period of starvation after which it was refilled with fresh urine (Fig 3,  achieving steady state of the system, however the operation of the module and the modular stack was in the batch-feeding mode where the retention time was usually 24h (fed daily) up to 72h (i.e. when not fed during the weekends) with visible periods of starvation. As the continuous operation requires a motorised pump for the feedstock supply, it is more suitable to implement the batch-mode feeding regime into the practical demonstrations and large scale set-ups.
Throughout the long-term study, it was observed that the cathodic wiring used for the parallel connections as well as stainless steel crocodile clips were corroded and needed to be replaced on multiple occasions. Corrosion was observed however it was not a part of the current generation throughout the experiment. Corrosion and subsequent malfunction of the external wiring, as well as junctions, bolts and clips is a common issue in the long-term prototype testing (Ge and He, 2016) and it needs addressing in the development of future prototypes.

Long-term performance of the 3-Module cascade
The cascade was fed in batch mode and the external resistance was manually controlled starting with the 50 Ω, whereas 2 Ω was the optimum and 1.5 Ω was the heaviest load applied.

Urine treatment capacity
The   lowering the anode to cathode ratio from 24.9 to 12.4. The performance of the stacked modules and their calculated volumetric power density shows improved output up to 13.9 W/m 3 , which is higher than the values obtained in similar configuration but using larger cylinders  . This result also double the power reported in a previous work with similar size cylinder but different type of ceramic materials (Gajda et al., 2018d). The reduction of the size of the individual units, variation the type of ceramic as well as the improvement of the external circuitry result in an enhancement of the power density output by the system. The effect of the parallel connection on the microbial catalytic activity in a MFC stack suggests that it is a good strategy for long-term stability of stacked MFC systems. The information in Schematic 1 shows the importance of peripherals in MFC scaled-up systems and the need for careful consideration of essential parts (like resistive loads) to avoid significant detriment to the MFC performance. As can be seen in the example below, the effect of simply connecting the wires, which have to be corrosion-resistant and bio-compatible, can range from 1% to 33%, depending on scale and configuration, and this must be taken into account when designing appropriate circuitry for field applications.

System characteristics and Challenges
Schematic 1. Electrical losses originating from the external wiring connected to the resistor load in all configurations.
Maximising power and improving longevity at decreased cost, is the key to promoting the MFC technology as a real product that can add value, through a range of practical applications (Liu et al., 2014;Wei et al., 2011;Zhou et al., 2011) in a future with new markets.

CONCLUSIONS
Most of works in literature report MFC short-term assays, however for the purpose of practical application, it is crucial to test the performance of the system during prolonged operating times which would allow to address the potential challenges that might appear during the process. The results show that all MFC set-ups studied here are suitable for long-term processes, reporting lower loss of power compared with commercial membranes. In the case of the 3modules cascade, the cascade configuration not only increased the power output but also the COD removal rate. Both good long-term stability as well as the resilience of the system against changes in the operating conditions support the suitability of ceramic membranes for being used as a MFC separator, boosting the real-implementation of this technology.