Microbial fuel cells (MFCs) which implement the concept of green chemistry are an emerging renewable energy technology that directly use waste biomass to produce electrical energy by the catalytic reaction of exoelectrogenic bacteria, oxidization of organic compounds to CO₂ and conduction of electrons to an cathode through an external circuit ^{[1, 2]}. As potential technology, MFCs have drawn increasing attention in recent years, and progress has been achieved with the promise of practical application, especially for wastewater treatment ^{[3, 4, 5]}, chemical synthesis ^[6], biochemical oxygen demand (BOD) detection ^[7], and seawater desalination ^[8]. However, a high capital cost and unsteady power behavior hinder the development of MFCs on a large scale due to the occurrence of current overshoot. The anode performance is one critical factor due to its low electrochemical capacitance ^[9] and non-maturing biofilm ^[10]. To address these problems, efforts had been made in previous studies ^{[11, 12, 13]} for getting steady MFC performance.

Anode materials with different capacitance for biomass retention were investigated in a MFC process study. It was found that there was no current overshoot phenomenon with an ultra-capacitor activated carbon (UAC) which store electrons as an electro-chemical mediator ^[9]. However, the high cost and complex preparation procedure of the UAC limit MFC application in practice. The weak conductivity is also a limit for electron transfer between the anode and bacteria. The future MFC anode requires a highly efficient and low cost electrode material. Carbon black (CB) is more attractive because of its excellent electrical property, chemical stability and low production cost. However, normal CB has a low capacitance, resulting in the occurrence of power overshoot with unstable power output ^[9]. It has been reported that the enhancement of capacitance can be achieved by surface chemical activation ^{[14, 15, 16, 17, 18]} or the addition of metal oxides (eg. Fe₃O₄) ^[19]. Here, we used CB as the matrix for the electrode active material. To further increase the capacitance of the CB anode to achieve overshoot elimination, chemical activation with HNO₃ and KOH at low temperature was employed (denoted as CBN and CBK, respectively). Subsequently, Fe₃O₄ was added to the host with a content of 5 wt.% for the untreated and treated CB anodes (denoted hereafter as CBFe, CBNFe and CBKFe). The electricity-generating behavior of the chemically modified CB as the MFC anode was evaluated and compared with a CB anode as the control. The effects of structure and texture on the physical characteristics and electrochemical capacitance were discussed.

KOH and HNO₃ were individually utilized for the pretreatment of CB (T₉₀, Tianjin Jinqiushi Chemical Co. Ltd, China). First, the CB powder (1.0 g) was dispersed in KOH-ethanol solution (5.0 g KOH dissolved in 40 mL ethanol) or concentrated HNO₃ (14 mol/L, 100 mL) and treated ultrasonically for 0.5 h, and then continuously stirred for 24 h at 80 °C. After cooling to room temperature, the activated powder was soaked in HCl or NaOH solution (1 mol/L) for 24 h, and then rinsed with distilled water using a suction filter to remove residual alkaline or acid. Finally, the product was dried overnight in a vacuum oven at 60 °C.

Fe₃O₄ was prepared by the co-precipitation method according to reported details ^[20]. The sheet comprising the electrode matrix and polytetrafluoroethylene (PTFE, 60 wt.%; Horizon LID, Shanghai, China) emulsion with the ratio of 19:1 was roll-pressed onto the current collector of an SSM (type 304L, 80 × 80 mesh, thickness of 0.2 mm, Hebei Anping County Shengze Screen Co. Ltd, China). This was used as the MFC anode. The detailed process was described elsewhere ^[21]. A cathode with a sandwich structure was made with a Pt/C catalytic layer (0.5 mg/cm²), carbon fiber mesh (10 wt.% wet proof, Jilin Carbon Plant Co. Ltd, China) base layer and four PTFE diffusion layer faced to air as shown previously ^[22].

The cube shaped membrane-less MFC was a 28 mL cylindrical chamber (4 cm long by 3 cm diameter; 7 cm² project area; electrode spacing of 4 cm). It was run in batch mode at 30 ± 1 °C in a controlled biochemical incubator (SPX150BIII, Tianjin Taisite Co. Ltd, China) under 1 kΩ except when stated otherwise. All the anodes were inoculated and pre-acclimated with the effluent from other parent MFCs that had been running for one year. The anode medium was the same as the prevous study with an initial COD 780 mg/L, which was replenished when the cell voltage ≤ 20 mV. Trace minerals (12.5 mL/L) and vitamins (5 mL/L) were added, and the pH was adjusted to 7.0 with 50 mmol phosphate buffer solution (PBS; Na₂HPO₄, 4.09 g/L; NaH₂PO₄·H₂O, 2.93 g/L; KCl, 0.13 g/L; NH₄Cl, 0.31 g/L) ^[12] when the output voltages were ≥ 500 mV during four consecutive periods, marked as 1st, 2nd, 3rd, 4th cycle, respectively.

The surface functional groups were observed with Fourier Transform Infrared spectroscopy (FTIR, KBr pellet, 4000 ~ 400 cm^-1) using a Magna560 (Thereto Nicolet Co. Ltd., USA) apparatus. The element mapping of the sample surface was performed by energy dispersive X-ray spectroscopy (EDX, Nova NanoSEM 430, FEI Company, USA). The specific surface area (S_BET) and pore size distribution (PSD) were calculated by the Brunauer-Emmett-Teller (BET) and Barret-Joyner-Halenda (BJH) equations. The total pore volume (V_total) was obtained from the N₂ adsorption isotherm at p/p₀ = 0.99, and the micropore surface area (S_micro) and micropore volume (V_micro) were extracted by the t-plot method ^[24]. These analyses were carried out by a TriStar 3000 (Micromeritics, ASAP2020, USA) at -196 °C with N₂ as adsorbate. The sample surface pH was determined by soaking 0.45 g of dried CB powder in 20 mL deioned water and vigorously magnetically stirred for 18 h. The pH (S20P, SevenEasy Plus, Mettler Toledo, Switzerland) of the filtrate was referred to as the pH of the CB sample for simplication ^[25]. The apparent contact angle was measured using a 1 mL of distilled water drop and a Harke-SPCA (Beijing Hake Co. Ltd, China) apparatus to characterize the surface wettability of the electrode. The scanning electron microscope (SEM) images of the microbe on the MFC anode were obtained using a Shimadzu SS-550 operated at 15.0 kV.

The voltage across the resistor was monitored at 30 min intervals using a date acquisition system (PISO-813, ICP DAS Co., Ltd.). The polarization and power density curves were acquired by varying the external resistance from 1000 to 10 Ω at a time interval of 30 min. Before and after bacteria inoculation, the capacitance determination of the MFC anode was peformed by a cyclic voltammetry (CV) scan recorded from -0.8-0 V (vs. Ag/AgCl) at a rate of 0.1 mV/s using a CHI660D potentiostat (CH Instruments Inc., China). The abiotic anode was soaked in 50 mmol/L PBS in the MFC reactor for 48 h before the CV test. All the electrochemical measurements were performed with the anode serving as the working electrode, the air-cathode as the counter electrode and an Ag/AgCl electrode close to the anode as the reference electrode (+197 mV, 3.5 mol/L KCl, vs. SHE).

The tests of electrode potential and power density curves were started in Cycle 2 (Fig. 1(a)). The CB anode gave a MPD of 821 ± 2 mW/m² at a current density (i) of 2.42 A/m² with subsequent current-fluctuation which first increased to 2.68 A/m² (150 Ω), then returned to 1.98 A/m² (50 Ω), and then recovered to 2.12 A/m² (10 Ω). After chemical activation, the MPDs of the CBN and CBK anodes were unexpectedly decreased by 2.4% and 1.1% to 801 ± 5 mW/m² at i = 2.76 A/m² and 812 ± 13 mW/m² at i = 2.78 A/m², respectively. Interestingly, the MPDs of the Fe₃O₄-added MFC were 3.0%, 7.7% and 9.7% higher than those of those without added Fe₃O₄ with the values of 845 ± 52 mW/m² at 2.45 A/m² for CBFe-MFC (CBNFe-MFC: 863 ± 7 mW/m² at 2.85 A/m²; CBKFe-MFC: 891 ± 31 mW/m² at 2.91 A/m²). Different from CBN, CBNFe and CBKFe that did not show overshoot, there was a current overshoot in CBK-MFC when the resistance was switched from 1000 to 50 Ω, and a rapid sharp drop of current density for CBFe appeared when the external resistance was varied from 100 to 50 Ω. The electrode potential curves indicated that the overshoot was from the anodes but not the cathodes, as all the MFCs cathodes maintained similar working potentials (Fig. 1(b)). The abnormal rapid increases in anode potential of CBK and CBFe were observed from -311 to -44 mV and -283 to -49 mV (vs. Ag/AgCl), while the CB anode potential appeared as a current hysteresis over the potential window of -319 to -50 mV (vs. Ag/AgCl) ^[12].

Fig. 1. Power density curves ((a), (c), (e) and electrode potential curves ((b), (d), (f), open symbols represent anode potential, closed symbols are cathode potential) of the MFCs in Cycle 2, 3 and 4. Error bars ±SD are based on averages measured in duplicate. (CB: carbon black; CBFe: Fe3O4-added CB; CBN: HNO3-treated CB; CBNFe: Fe3O4-added CBN; CBK: KOH-treated CB; CBKFe: Fe3O4-added CBK).

In the following Cycle 3 (Fig. 1(c)), the MPD of CBK-MFC had a 2.7% increase of 834 ± 31 mW/m² at a 1.1% increase of i (2.81 A/m²) with overshoot eliminationg. This was possibly due to the maturation of the anodic biofilm, which implied the successful startup of CBK-MFC as discussed previously ^[9]. However, the overshoot still existed with the MPDs and increased by 5.7% and 6.9% to 868 ± 45 mW/m² (2.87 A/m²) for CB-MFC and 903 ± 18 mW/m² (2.93 A/m²) for CBFe-MFC. The abnormal increase of the anode potential became smooth for CB and CBFe (Fig. 1(d)). Notably, the overshoot disappeared for both CB-MFC and CBFe-MFC in Cycle 4, and there was a 10% increase in the MPD to 903 ± 15 mW/m² (2.93 A/m²) for CB and 927 ± 15 mW/m² (2.97 A/m²) for CBFe (Fig. 1(e)). These values were in accordance with the changes of the anode potential curve (Fig. 1(f)).

For each CB sample, a broad FTIR absorption band at 3500-3100 cm^-1 was ascribed to the stretching vibration of an OH group. The band at 2067-2043 cm^-1 corresponded to C =O vibrations (Fig. 2). After activation treatment, the peak appearing at 670 cm^-1 was attributed to the bending vibration of O-H, which demonstrated that the intensity associated with the OH group was strongly increased. As expected, there was a new peak at 1384 cm^-1 for CBN which displayed a characteristic sharp absorption band of the asymmetric stretch of NO₃^-. There was another new stretching vibration peak of C-O at 1172 cm^-1 for CBK. The EDX analysis showed the presence of C (0.28 keV), O (0.52 keV), S (2.3 keV) atoms in each CB sample. However, it was noticed that there was an increase in the oxygen content and a decrease in the percentage of carbon after chemical activation comparing with the control (Table 1). The existence of N (0.39 keV) demonstrated that the nitrogen-associated functionality was anchored on the surface after HNO₃ activation. The K signal (3.31 keV) was due to adsorbed potassium from KOH activation. The FTIR and EDX analyses confirmed that the oxygen (nitrogen)-containing functionalities were introduced to the activated CB surface by the chemical treatment, which would make more active sites to improve the electrochemical response and wettability of the MFC anode.

Fig. 2. FTIR spectra of carbon black before and after chemical activation.

Table 1 EDX analysis of carbon black before and after chemical activation. Sample CB CBN CBK w(C)/% 95.21 86.47 89.67 w(O)/% 4.41 13.05 9.83 w(S)/% 0.38 0.09 0.14 w(N)/% — 0.39 — w(K)/% — — 0.36

Table 1 EDX analysis of carbon black before and after chemical activation.

According to the IUPAC classification, the N₂ adsorption isotherms for each CB sample were Type IV with a Type H3 hysteresis loop as noted in Fig. 3. In the range of p/p₀ > 0.95, the slope of the hysteresis loops followed the order of CBK > CB > CBN, which were related to the pore width (D, 4V/S by BET) as listed in Table 2. Both S_BET (585.8 cm²/g) and V_total (0.89 cm³/g) of CB were decreased by 57% and 60% to 251.2 cm²/g and 0.36 cm³/g after the acid treatment. There was a similar reduction of 28% (424.9 cm²/g) and 16% (0.75 cm³/g) after the alkaline treatment, which may have resulted from hole wall collapse and aperture blockage caused by the functional groups from the chemical activation as discussed by Chen et al. ^[26]. It should be noted that there were obvious changes in the PSD for the activated CB matrix as presented in Table 2. The S_micro increased by 29.7% to 48.6 cm²/g for the HNO₃ activation CB, while it was slightly decreased by 2.7% to 36.0 cm²/g for the KOH activation CB. Meanwhile, the S_meso decreased by 63% for the acidic pretreated CB (CBN, 202.6 cm²/g) and 29% for the alkaline pretreated CB (CBK, 388.9 cm²/g). In particular, the V_micro values increased by 2 times to 0.02 cm³/g after the chemical treatment, while the V_meso values decreased by 63% to 0.34 cm³/g for CBN and 18% to 0.73 cm³/g for CBK. The higher micropo rosity contributed to the electrochemically accessible surface area ^[29], which was verified by the following capacitance behavior analysis.

Fig. 3. N2 adsorption/desorption analysis of carbon black before and after chemical activations (solid: adsorption, hollow: desorption).

Table 2 Porosity characteristics of the carbon black before and after chemical activation. Sample CB CBN CBK SBET (cm2/g) 585.8 251.2 424.9 Smicro (cm2/g) 37.1 48.6 36.0 Smeso (cm2/g) 548.7 202.6 388.9 Vtotal (cm3/g) 0.89 0.36 0.75 Vmicro (cm3/g) 0.01 0.02 0.02 Vmeso (cm3/g) 0.88 0.34 0.73 Vmicro/Vexter 1.0 6.0 3.0 D a /nm 6.1 5.7 7.7 a Average pore diameter.

Table 2 Porosity characteristics of the carbon black before and after chemical activation.

Table 3 showed that the pH of CB (6.30) was close to that of deionized water (6.29). After acid treatment, the pH value slightly decreased by 2.2% to 6.16, while the pH increased by 11% to 7.11 after alkaline treatment. The variation of the pH can be ascribed to the adsorption of soluble reagents in the micropores and mesopores of the CB matrixes. The wettability of an electrode surface is routinely elucidated by its water contact angle. The value (94.7°) of the contact angle of the CB anode was higher than 90° as shown in Table 3, suggesting the hydrophobicity characteristics of its surface. This was attributed to its non-polar structure. The KOH activated CB anode displayed a lower contact angle with the value of 62.7°, and the lowest contact angle of 56.9° was found with the HNO₃ activated CB anode, indicating that surface hydrophiliciy was increased by the activation treatment, which was conducive for ion adsorption on the surface and bacterial adhesion in the MFC anode.

Table 3 The comparisons of surface pH and contact angle of carbon black before and after chemical activations. (DI: deioned water). Sample DI CB CBN CBK pH 6.29 6.30 6.16 7.11 Contact angle (°) — 94.7 56.9 62.7

Table 3 The comparisons of surface pH and contact angle of carbon black before and after chemical activations. (DI: deioned water).

Electrochemical studies of the CB anodes were conducted using the specific capacitance (C_m, F/cm²) before and after the bacteria colonization, which was calculated by

C_m =It/A∆E

where I (A) is the the charge-discharge current of the CV record, t (s) is the time of CV scan, A (7 cm²) is the projection area of the electrode, and ∆E (V) is the operating potential window (Fig. 4).

Fig. 4. CVs of each CB anode before and after the exoelectrogenic bacterial adherence. (a) CVs of abiotic anodes; (b) CVs of biotic anodes in Cycle 2; (c) CVs of biotic anodes in Cycle 3; (d) CVs of biotic anodes in Cycle 4.

It was found that the acid-pretreated CB (CBN) had the maximum abiotic capacitance (C_m^abiotic) of 1.52 F/cm², with a value 46% and 53% higher than those of the alkaline-pretreated CB (CBK, 0.81 F/cm²) and untreated control (CB, 0.71 F/cm²). As we know, the micropores play a crucial role in ion accumulation and enhancing the electrical double layer ^[27]. It has been reported that micropores can be created by a heating treatment with KOH at 800 °Cin argon atmosphere ^[28] or HNO₃ oxidation at 120 °C ^[29]. As shown in Table 2, there were obvious increases of 2 times in the volume of the micropores for the activated CB. The acid-pretreated CB exhibited a superior capacitance characteristic compared to the alkaline-pretreated CB due to its good electrolyte affinity, which can be explained by the contact angle measurement ^[30]. This was also shown by the FTIR results that the oxygen (nitrogen)-containing functional groups improved the surface hydrophilicity of HNO₃-activated CB anode. Different from Deeke’s report ^[31], the improvement of abiotic capacitance did not have a positive effect on the power output of the MFC after chemical activation with HNO₃ and KOH, which probably resulted from the decrease of actual contact area between the bacteria and the electrode ^[32] as shown by the reduction of the specific surface area following the order of CB > CBK > CBN. Thus it was inferred that the surface area of the electrode material plays a key role in the capacitance performance. This will need to be addressed in future. Furthermore, it was found that the C_m^abiotic of Fe₃O₄-added anodes was higher than those of Fe₃O₄-unadded ones as shown in Table 3. This was in fair agreement with the report that Fe₃O₄ can deliver remarkable pseudo-capacitive activity ^[33].

Table 4 Capacitance of CB anode before and after chemical modification based on CV analysis (2c, 3c and 4c represented Cycle 2, 3 and 4 individually). Anode Cmabiotic (F/cm2) Cmbiotic·2c (F/cm2) Cmbiotic·3c (F/cm2) Cmbiotic·4c (F/cm2) CB 0.71 0.79 0.86 1.12 CBFe 0.77 0.82 0.9 1.17 CBN 1.52 1.76 — — CBNFe 1.77 1.84 — — CBK 0.81 0.9 1.1 — CBKFe 1.14 1.26 — —

Table 4 Capacitance of CB anode before and after chemical modification based on CV analysis (2c, 3c and 4c represented Cycle 2, 3 and 4 individually).

With the enrichment of anodic bacteria, the biotic CV was performed with the cell voltage ≤ 20 mV. C_m for CBN (CBNFe and CBKFe) increased to 1.76 (1.84 and 1.26) F/cm² in Cycle 2. For CBK, C_m increased from 0.9 F/cm² in Cycle 2 to 1.1 F/cm² in Cycle 3. C_m of CB (CBFe) was consecutively increased from 0.79 (0.82) F/cm² (Cycle 2) to 0.86 (0.9) F/cm² (Cycle 3) and then to 1.12 (1.17) F/cm² (Cycle 4). It was clear that the biotic capacitance (C_m^biotic) was increased with the values of 0.12-0.4 F/cm² by different degrees, indicating that the living microorganisms accelerated ion transport from the electrolytic solution to the electrode between each CB particle and bacterial conductive appendage. This would be due to the transient charge storage of oxidoreductase on the cell membrane and in the cytoplasm ^[34]. SEM pictures suggested the crosslinking among the indigenous microorganisms. However, there was no significant difference in morphology and amount of bacterial cells (Fig. 5). Associating the anodic capacitance with the current overshoot, it can be found that the higher C_m of the MFC anode gave the better alleviation of the current return. Our previous research provided the proof-of-concept that a steady state was achieved with a concomitant elimination of MFC overshoot ^[9].

Fig. 5. SEM images with 2 K magnification of the anodes with the exoelectrogenic bacterial adherence before and after chemical activations.

This was shown by the power density curves as plotted in Fig. 1. So, we can interpret this result as evidence that, either from the physical property of the electrode matrix or the capacitance compensation of the anodic biofilm, the system using CB as the MFC anode can be considered to be stabilized so long as the anodic capacitance is no less than 1.1 F/cm².

This study investigated the power generating behavior of MFCs using chemically modified CB as the MFC anode. The steady power output obtained followed the order of CBN > CBK > CB. This was due to the improved anodic capacitance for electron transfer from the surface modification. Electrochemical analysis revealed that the chemically modified CB anode exhibited favorable power stability, and can therefore be recommended for use as the MFC anode.

Acknowledgments

Thanks go to Associate Prof. Xin Wang of Nankai University, Tianjin for his kind help. The authors thank the anonymous reviewers for their instructive comments.