Bioelectrochemical upgrading of anaerobic digester biogas under thermophilic conditions

. Bioelectrochemical systems (BESs) can be integrated in situ into anaerobic digesters for increasing methane (CH 4 ) content of biogas. Using BES ex situ for improving biogas quality has recently been gaining attention. However, information on the process under thermophilic conditions is very limited. In this study, we placed a BES cathode in-line at the exit gas from a thermophilic anaerobic digester to convert carbon dioxide (CO 2 ) in the biogas into CH 4 . The performance of the ex situ BES reactor under thermophilic conditions was evaluated. When poising the cathode at (cid:1) 1.1V versus Ag/AgCl in the ex situ BES reactor, CH 4 content increased from 50% to 85%. Of the incoming CO 2 73% was biologically converted to CH 4 and 23% absorbed by alkalinity generated in the cathode. The energy output as additional CH 4 as a percentage of the energy input to operate the BES was calculated at 56%. The biocathode of the BES reactor was dominated by Methanothermobacter spp., which are thermophilic hydrogen consuming methanogens. This study con ﬁ rms that thermophilic BES can be used as an ex situ treatment process for enriching the CH 4 content of biogas. However, energy ef ﬁ ciency of the process was found to be limited by the lack of an energetically ef ﬁ cient anodic reaction. For industrial applications, optimisation of energy ef ﬁ ciency is an area for further research.


Introduction
Biogas is a product of anaerobic digestion that can be used in gas turbines to produce electricity. It typically contains 50-70% methane (CH 4 ), 30-50% carbon dioxide (CO 2 ) and trace amounts of hydrogen (H 2 ), hydrogen sulfide (H 2 S) and ammonia (NH 3 ) [1]. Typically, the calorific value of biogas is approximately 22 MJ/m 3 , whereas that of natural gas is 36 MJ/m 3 . Because the energy content of biogas is directly proportional to its CH 4 content, removing the CO 2 would increase the calorific value of the biogas. Hence, to achieve its full potential the removal of CO 2 and other impurities from the biogas is necessary [2]. The purified methane-enriched biogas can, for example, be injected into a natural gas network for town gas supply.
Conventional methods for removing CO 2 from biogas are largely physical/ chemical based, including for example pressure swing adsorption, water scrubbing, organic physical scrubbing, and chemical scrubbing [3]. Despite having excellent performance in enriching the methane content in biogas, these physical and chemical methods typically involve multiple and complicated steps with high energy input and corresponding operational costs [2]. On the other hand, biological gas treatment has been considered as a possible less reagent-intensive and energy-intensive biogas upgrading alternative. Ghosh and Klass were the first to suggest that biogas from anaerobic digestion could be passed through a secondary methanogenic digester to remove CO 2 and H 2 S from biogas [4]. Following this idea, Strevett et al. studied the ability of a chemo-autotrophic archaea (Methanobacterium thermoautotrophicum) to upgrade biogas by reducing CO 2 to CH 4 using H 2 as an electron donor [5]. The biogas (50-60% CH 4 , 40-50% CO 2 and 1-2% H 2 S) was mixed with H 2 and passed through a hollow fibre membrane colonised by the microorganisms capable of catalysing the hydrogenotrophic methanogenesis reaction (reaction 1). The results showed that the anaerobic chemo-autotrophic process could effectively remove CO 2 , increasing the CH 4 content from 50-60% to 96%.
Recently, integrating bioelectrochemical systems (BES) with anaerobic digestion (BES-AD) has been considered a promising way to enhance biogas production [6][7][8]. We investigated and compared the effectiveness of two BES-AD configurations, single-chamber (membraneless) and two-chamber (membrane-separated), on biogas production under thermophilic conditions [9]. However, the results showed that operating a BES directly in an AD reactor led to a suppression of organics conversion into biogas, which is understandable as increased H 2 partial pressures are known to interfere with organics conversion to biogas. Therefore, it would be worth retrofitting the process with the BES being deployed downstream of the anaerobic digester as a standalone unit process, such that it serves as a biogas converter (CO 2 to CH 4 ). Such an ex situ biogas upgrading approach has the potential to facilitate upgrading of biogas produced from multiple (small-scale) digesters. Hence, the aims of this study were: (1) to determine if the ex situ thermophilic BES reactor could increase the methane content of biogas; (2) to characterise the microbial community on the cathode surface and the bulk medium to gain insight into the biogas upgrading mechanisms of the BES reactor; (3) to evaluate the energy efficiency of the BES reactor for identifying areas of further improvement towards practical application. Investigations were conducted under thermophilic conditions as no previous studies had been carried out under these conditions and performance at thermophilic conditions is expected to be higher than at lower temperatures.

Setup of ex situ BES reactor
The ex situ BES reactor was set up based on the twochamber BES reactor described in our previous study [9]. The two-chamber reactor consisted of two identical acrylic half-cell cylindrical vessels (500 mL capacity in each vessel) with several sampling ports. The two half-cells were separated by a cation-exchange membrane (CEM) (30 cm 2 , AS2S, A2-11, Membranes International Inc., US). The counter electrode (anode) chamber contained 350 mL of 150 mM NaHCO 3 , and the counter graphite electrode (length Â diameter: 15 cm Â 2.5 mm, Progresso 8911/8B) was inserted at the middle of the counter chamber. The working volume of the cathodic chamber was 350 mL. A sheet of carbon felt (width Â length: 6 cm Â 22 cm) (MGM-Carbon Industrial, Ltd. Co., China) was used as the working electrode (WE), which was folded into a cylindrical form within the cathodic chamber. The distance between the working (the side that was closest to the CEM) and the counter electrodes was approximately 10 cm. Electrical connection between the carbon felt and the external circuit was made via a titanium wire, with one end tightly fastened to the felt. A silver silver chloride (Ag/ AgCl) reference electrode (saturated KCl), pH and redox potential (ORP) probes were tightly mounted (sealed with epoxy glue) into the cathodic chamber to allow real-time monitoring. Instead of carrying out simultaneous AD and BES within the reactor as was the case in our previous paper, AD was conducted in a separate reactor. Biogas from the AD reactor or synthetic biogas was then used to feed into the ex situ BES reactor for testing the conversion of CO 2 to CH 4 .

Experimental conditions 2.2.1 Coupling of an anaerobic digestor with the ex situ BES reactor
A semi-continuous anaerobic digester (400 mL) was fed with 3.5 mL of concentrated glucose solution (1 M) daily after withdrawing the same volume of bulk solution. The biogas produced was monitored. Once the biogas production pattern was established, the biogas from the AD was directly fed into the BES reactor via a Tygon tube (diameter 3.1 mm, Masterflex ® 06409-16). The outflow gas from the BES reactor was collected in an inverted glass measuring cylinder (500 mL) to determine gas volume. A schematic diagram of the AD-ex situ BES reactor set-up is shown in Figure 1. Compositions of biogas from both AD and ex situ BES reactors were monitored daily. Working electrode (WE) potential in the ex situ BES reactor was applied at different levels (-0.9, À1.0, À1.1, À1.2 V vs. Ag/ AgCl) to determine the optimum WE potential for CO 2 to CH 4 conversion.
All the reactors in this study were operated under thermophilic conditions in a thermostat water bath controlled at 55°C. The WE potentials reported in this study refer to values against Ag/AgCl reference electrode (ca. +197 mV vs. standard hydrogen electrode). The current and pH in the BES reactor were recorded by a computer using a LabJack USC interface and National Instrument LabView software. pH in the BES reactor was maintained at around 8 by adding HCl (4 M).

Synthetic gas as feed to ex situ BES reactor
Two sets of experiments using synthetic biogas with known gas composition (50:50 (v/v) CO 2 :CH 4 , or 100% CO 2 ) were also used to evaluate the BES CO 2 reduction process. The synthetic gas was transferred from a gas-bag (3L, TEDLAR BAG, CEL Scientific Corp. USA) via pumping (MasterFlex ® , Peristaltic Pump, Cole-Parmer Instrument Company, USA) at a fixed volumetric loading rate (3.4 L/ L/d). The set up was similar to Figure 1, except the gas bag with the peristaltic pump replaced the anaerobic digester. The CO 2 -converting electrode (i.e. the cathode) of the BES reactor was poised at a potential of À1.1 V versus Ag/ AgCl. The outflow gas from the BES reactor was collected in another gasbag for gas composition analysis.

Role of cathodic biofilm on CH 4 enhancement
To determine the role of the biofilm attached to the carbon felt cathode (i.e. WE) in the bioelectrochemical conversion process (CO 2 ! CH 4 ), after the aforementioned experiments (Sect. 2.2.2) the bulk solution in the ex situ BES reactor was completely removed from the cathodic chamber and replaced with a cell-free bicarbonate buffer (150 mM NaHCO 3 , which was the same as the anolyte). The WE potential was again maintained at À1.1 V versus Ag/AgCl in the BES reactor to determine to what extent, in the absence of planktonic cells, the microbial communities in the cathodic biofilm could sustain the bioelectrochemical conversion.

Gas analyses
All biogas samples were taken using a glass syringe (100 mL, GASTIGHT ® #1710). Biogas compositions (methane, hydrogen and carbon dioxide concentrations) were immediately analyzed by a Varian Star 3400 gas chromatograph (GC) equipped with a thermal conductivity detector. The carrier gas was set at a flow rate of 30 mL/min. The headspace samples (100 mL) were manually injected into a GC column (2 m Â 5 mm, Pora-PakQ) maintained at 40°C. The temperatures of the detector and inlet were maintained at 40 and 120°C respectively.

Microbial community analysis
To determine microbial community in the BES reactor, three samples were taken from the initial thermophilic anaerobic sludge inoculum, the biofilm on the cathode and the bulk solution in the BES cathodic reactor. DNA was extracted from these samples using the PowerSoil ® DNA Isolation Kit as per the manufacturer's instruction. The  3 Results and discussion 3.1 Transferring biogas from the anaerobic digester to the BES A steady biogas production was first established for the AD reactor prior to coupling the AD reactor with the BES reactor. Results showed that after approximately one month of operation, the AD reactor reproducibly and rapidly responded to glucose additions, allowing biogas production at a maximal rate of approximately 4 L/L/d (Fig. 2). The average biogas production rate recorded (∼1.2 L/L/d) also matched the theoretical value (1.3 L/L/d) estimated based on the application rate of glucose. Thereafter, the cathodic half-cell of the two-chamber BES reactor was coupled via a Tygon tube to the headspace of the AD reactor.
The capability of the BES reactor in upgrading the AD biogas was evaluated by poising the carbon felt WE at different potentials, ranging between À0.9 V and À1.2 V (Fig. 3). Results showed that when the WE was operated at a more negative potential of À1.1 V and À1.2 V, a higher cathodic current was generated (Fig. 3A), which also coincided with a rapid pH rise of the catholyte (from 8 to 9) (Fig. 3B). The gas outflow from the BES reactor was found to have a significantly higher CH 4 content than the biogas from the AD (Fig. 4). When the applied WE potential was at À1.1 V, the CH 4 content in the gas outflow from the BES reactor reached 90%, accounting for a 60% improvement. These results confirmed that the ex situ BES reactor was effective in augmenting the CH 4 content in the AD biogas. This could be due to two reasons, microbial conversion of CO 2 to CH 4 and/ or CO 2 adsorption by catholyte. As the varying gas production rate and CO 2 :CH 4 ratio in the AD biogas interfered with the investigation of the BES reactor, separate experiments using synthetic AD biogas steams were performed in subsequent tests.

Performance of the ex situ BES reactor with the use of synthetic biogas
Synthetic gas (CO 2 :CH 4 50:50) was continuously fed into the BES reactor at a rate of 3.4 L/L/d and the WE was poised at À1.1 V for 6 days (Fig. 5). As noted before, the catholyte pH increased soon after the onset of the cathodic current. To prevent the pH from becoming too alkaline for methanogens, it was controlled to approximately 8 using 4M HCl. The CH 4 content of the resulting outflow gas was determined and compared with that in the inflow gas stream. Results showed that the average CH 4 content (over the 6 days) in the outflow gas was significantly increased from 50% to approximately 85% (Fig. 5A). This suggested an effective CO 2 conversion and/or trapping had occurred in the BES reactor. When switching the BES to an open circuit mode operation, the CH 4 content in the outflow returned to approximately 50%, confirming that the cathodic current sustained by the bioelectrochemical reaction(s) within the BES was responsible for the CH 4 enrichment in the biogas.
To further determine if the increased CH 4 content was indeed generated from the cathodic bioconversion of CO 2, another synthetic gas experiment was conducted with pure CO 2 gas as the inlet feed. As in the previous experiment, the pure CO 2 gas was continuously fed into the BES reactor at a similar rate (3.4 L/L/d). The results showed that a notable concentration of methane (∼20%) was recorded in the outflow gas of the BES reactor (Fig. S3), suggesting that the increase of CH 4 in the biogas was likely caused by a bioelectrochemical reduction of CO 2 .

Role of cathodic biofilm on methane enhancement
In order to eliminate any possible CH 4 produced from residual organics in the digester solution around the cathode of the BES, the bulk solution of the cathodic chamber was replaced by a fresh bicarbonate buffer (150 mM). The activity of hydrogenotrophic methanogens on the biocathode was evaluated by poising WE potential at À1.1 V. Gas inflow (CO 2 :CH 4 = 50:50) to the BES reactor was maintained from a gas-bag and the resulting outflow gas content was monitored for three days. Analysis of the outflow gas in this experiment indicated a similar increase of CH 4 content (peaked at around 80% at the end of the third day) that was obtained from the BES reactor before the replacement of the bulk solution (Fig. 6).
Gas monitoring over a 24 h experiment showed that for the 636 mL of CO 2 removed, about 462 mL extra CH 4 was produced (Fig. S4). Of the CO 2 that was removed, 73% was biologically converted into CH 4 , whereas the remaining 27% was trapped (absorbed) by the alkaline bicarbonate solution. This result confirmed that the bioelectrochemical reaction catalysed by the cathodophilic biofilm was the main CO 2 removal mechanism, supporting the finding of Cheng et al. (2009) who first demonstrated that electromethanogenesis can be catalysed by a biofilm in a MEC [10]. Nonetheless, it should be mentioned that without acid dosing for pH control, our system should facilitate an even greater proportion of CO 2 absorption by the alkaline catholyte.

Microbial community in the BES
To identify the major groups of microorganisms in the population on the cathode, the biofilm was scratched off from the carbon felt on the cathode (sample MG11) and investigated by 16S Metabarcoding analysis. For comparison, the initial inoculum (sample MG2: thermophilic anaerobic sludge) and the bulk solution in BES reactor (sample MG12) were also examined. The results suggested that the majority of the microbes in the samples were bacteria (see Supplementary Materials). Given that methanogenic microorganisms are largely under the domain of archaea, in this section, only the changes of archaea in different samples were discussed and shown in Table 1: (1) a small amount of Methanosarcina (0.2%) presented in the initial inoculum; (2) after one week anaerobic digestion and 7 weeks BES operation, the hydrogenotrophic methanogen Methanothermobacter developed in the solution and also on the cathode of the BES reactor; (3) the fraction of Methanothermobacter on the cathode population was remarkably higher than that in bulk solution.
The results suggest that the BES under the tested conditions facilitated the growth of hydrogenotrophic methanogens of the genus Methanothermobacter, whereas the start-up inoculum (anaerobic sludge) was dominated by Methanosarcina, a genus of methanogens specialised on acetate oxidation that can also use hydrogen. The abundance of Methanothermobacter in the bulk solution was very nominal (about 5% of that in the cathodic  8, 8 (2023) biofilm), likely due to the lower availability of the key substrate hydrogen in the solution compared to the proximity of the cathode (where hydrogen was evolved). The overall results show that under thermophilic conditions, the growth of Methanothermobacter was favourable, leading to their high abundance recorded in the cathodic biofilm. So far only a few AD-BES integration studies have been carried out under thermophilic conditions, and the same finding was reported by Fu et al. (2015) that Methanothermobacter was the major group of archaea on the cathode that involved in CO 2 reduction to CH 4 [11].

Calculation of electron balance
To further investigate the relative CO 2 removal by bioelectrochemical conversion and catholyte trapping, an electron balance of the BES reactor was calculated. The data on day 6 from the experiment shown in Figure 5 was taken as an example for this calculation (Tab. 2). The total electron (e -) flow was calculated from the current, and the theoretical CH 4 generated based on the eflow was estimated by considering that one CH 4 molecule accepts 8 electrons.
A second example from the experiment shown in Figure 3 shows the electron balance for different WE potentials by the same calculation (Tab. 3). When applying WE potentials at À0.9, À1.0 and À1.1 V, the actual CH 4 content in the outflow gas was higher than the theoretical contents. This may be attributed to CO 2 trapping in the BES reactor. However, when applying WE potential at À1.2 V, the actual methane content was less than the theoretical content. Upon inspection it was found that there was erosion on the anode that interfered with the BES reactor operation. Therefore, this condition was not considered technically feasible.
The results above show that CO 2 was both absorbed and bioelectrochemically converted to CH 4 . In some cases, absorption was more predominant than conversion (Tabs. 2 and 3) and in other cases conversion was more predominant (Fig. S4). Based on the stoichiometry of the (reaction 4) According to this stoichiometry, CO 2 absorption can contribute up to 8 times more than the CO 2 conversion in methane upgrading (Fig. S5). The likely reasons why our experiments showed lower CO 2 absorption contribution   8, 8 (2023) was due to the pH control carried out which neutralised the cathodic hydroxide (reaction 2) and minimised CO 2 trapping by absorption.

Analysis of energy balance
To determine the feasibility of our system for industrial application, we evaluated the net energy balance of the system. The data from the experiment shown in Figure 5 was used as an example for the evaluation. The energy efficiency of the system was evaluated by comparing the energy input (as invested electricity through the potentiostat) with energy output as the CH 4 produced from CO 2 . The heating energy for maintaining thermophilic conditions (55°C) and energy input for chemicals such as acid dosing was not considered in this calculation, as an insulated thermophilic BES would not need to be heated for treating the gas coming from a thermophilic AD. The results show that the energy input to operate the BES for 6 days was 102.3 kJ, and the final energy output as extra CH 4 was 57.1 kJ (Tab. 4). This suggests a 56% energy efficiency or a 44% energy loss during the energy conversion (i.e. from electrical to thermal combustion energy of CH 4 ).
To understand how the energetic performance of the tested process can be optimised, we attempted to determine the applied cell voltage at which the energy input (as electricity invested via the potentiostat) and the energy output (as the extra CH 4 produced via the bioelectrochemical conversion) can be balanced (i.e. 100% efficiency). When the cathode potential was poised at À1.1 V versus Ag/AgCl, and the current was 0.047 A (i.e. the settings reported in Sect. 3.7, Tab. 4), such a threshold cell voltage and the corresponding anode potential are 1.15 V and +54 mV versus Ag/AgCl, respectively (Fig. 7). Figure 7 also reveals that over 100% of energy efficiency may be achieved if the anodic potential of the BES could be maintained at below +54 mV versus Ag/AgCl. This suggests that effective strategies should be devised to lower the anodic potential. One such strategy could be to operate the anode as a bio-anode, whereby a suitable microbial culture is used for catalysing the anodic oxidation reaction. Suitable electron donors for the bio-anode would ideally be low value substrates contained in waste streams (e.g. wastewater or anaerobic digestate). This approach should be considered in future studies. Table 2. Electron balance of the BES process (data derived from the day 6 operation as shown in Fig. 5  (1) Theoretical CH 4 content calculated from reduction of CO 2 by electron assuming no CO 2 absorbed in the process; (2) Actual CH 4 content in the off-gas.
It should also be noted that the BES technology for biogas upgrading has only been explored in lab-scale, and no study has been conducted to optimise the energy efficiency of an ex situ BES reactor for biogas upgrading [12]. Hence, it would be meaningful to investigate how energy efficiency can be improved and to develop strategies for minimising energy losses (such as ionic losses) in this emerging biogas upgrading process.

Conclusions
Based on the results, the following conclusions are made: (1) biogas to biomethane conversion by using ex situ themophilic BES reactor is theoretically and technically possible; (2) the operation of an ex situ BES CO 2 reduction system selectively enriched Methanothermobacter (95.4%) which was not detectable in the initial inoculum. Further   5) that WE potential was applied at À1.1 V; inflow gas À 50% methane and 50% CO 2 ; 1 mole of methane = 24.5 L; heat of combustion of methane = 891 kJ/mol. testing and optimisation of the BES reactor needs to be conducted, in particular with respect to replacing acid dosing by improved CO 2 absorption, as well as minimising the energy requirement of the process. Figure S1: Microbial community analysis in the BES reactor at Phylum level. Only phylum comprising at least 0.1% relative abundances is shown, others are included in the "other" category. Figure S2: Microbial community analysis in the BES reactor at Genus level. Only genus comprising at least 0.1% relative abundances is shown, others are included in the "other" category.