Effect of MWCNT-modified graphite felts on hexavalent chromium removal in biocathode microbial fuel cells

Multi-walled carbon nanotubes (MWCNTs) and oxidative acid pretreated MWCNTs (oxidized MWCNTs, OMWCNTs) were used tomodify graphite felts as biocathode electrodes in Cr(VI)-reducingmicrobial fuel cells (MFCs). The results showed that both MWCNTmodifications improved the efficiency of the Cr(VI)-reducing biocathode. In particular, the O-MWCNT modification led to a better performance due to the induced oxygen-containing functional groups on the O-MWCNTs. The O-MWCNT-modified graphite felt significantly promoted the Cr(VI) removal and electricity generation of the MFC. The Cr(VI) removal rate increased to 2.00 0.10 mg L 1 h , which was 2.05 times higher than that of the unmodified control. The improvement was ascribed to the strong affinity and capacity of the O-MWCNTs towards microorganisms and Cr(VI) ions. In addition, this study further confirmed that the ex situ biocathode acclimatization method could be an efficient way to screen potential biocathode materials for Cr(VI)reducing MFCs.


Introduction
Hexavalent chromium (Cr(VI)) is a known priority toxic chemical and carcinogen, which has gained attention due to its serious threat to humans. 1,2 Recently, microbial fuel cells (MFCs), a sustainable and effective technology for the reduction of oxidative heavy metals, [3][4][5] have been proposed to provide promising results for Cr(VI) removal. The abiotic cathode MFCs for Cr(VI) removal obtained the maximum removal rate as high as 8.12 mg L À1 h À1 , but this high removal in the abiotic cathode MFC strongly relies on a low pH. 3,6-8 Therefore, the biocathode MFCs show a great potential in Cr(VI) bioremediation, which can use the catalysis of self-regenerating electrochemically active bacteria to provide a sustainable and economic operation under a neutral environment. [9][10][11] However, the low efficiency of Cr(VI) removal and electricity generation in biocathode MFCs becomes the main bottleneck to limit their practical and largescale applications. [12][13][14] Since electrode material is a signicant factor to affect the performance of biocathodes by acting as a bacterial carrier and electron exchange site, 15,16 suitable biocathode materials are worth developing to improve the performance of Cr(VI)-reducing MFCs. So far, most of biocathode electrodes applied in Cr(VI)-reducing MFCs were conventional carbon-based materials, such as graphite plates, graphite granules, graphite felts and graphite bers. 1,10,12,17 Recently, the nanostructured materials modication on conventional carbon-based materials has been demonstrated its high efficiency to improve the MFC performance. [18][19][20] However, little effort has been spent on seeking for surface modiers for carbon-based materials to enhance the biocathode performance in Cr(VI)-reducing MFCs. 11,15 Our previous work found that the NaX zeolite-modied graphite felt improved the biocathode performance in Cr(VI)reducing MFCs due to the affinity of the NaX zeolite towards microorganisms and Cr(VI) ions; however, the high resistance of NaX zeolite would affect the conductivity of the modied electrode. 15 Multi-walled carbon nanotubes (MWCNTs), another kind of popular nanostructured material, possess unique physical and chemical properties (e.g. high porosity, high electrical conductivity, high surface area, good electrochemical stability and biocompatibility), which have been applied as the surface modier for carbon-based materials to improve the electricity generation of MFCs. [21][22][23][24] Considering the MWCNTs have a strong adsorption capacity for heavy metals like NaX zeolite, 25,26 the MWCNTs-modied electrode also might be a promising biocathode material for Cr(VI)-reducing MFCs. To the best of our knowledge, the MWCNTs-modied electrode has not been applied as biocathode material in Cr(VI)-reducing MFCs.
For preventing the self-aggregation of MWCNTs, an oxidative acid pretreatment is oen used for improving the dispersion of MWCNTs and nanotube purication. 27 As a result of oxidizing acid attack, oxygen-containing groups can be attached to the ends and surfaces of MWCNTs to improve the dispersion of MWCNTs in the aqueous solution. [28][29][30] Furthermore, the oxidized MWCNTs-modied electrode, compared to the original MWCNTs-modied one, caused a higher efficiency of power generation and wastewater treatment in MFCs. 31 The aim of this study was to investigate the effect of two kinds of MWCNTs (original and oxidized MWCNTs)-modied graphite felts as biocathode electrodes in Cr(VI)-reducing MFCs. An ex situ acclimatization method was used to obtain the Cr(VI)-reducing biocathode: 32 namely, the Cr(VI)-reducing biocathode was acclimatized in the anode chamber rst to enrich the electrochemically active biolm, and then directly transferred to the cathode chamber for Cr(VI) removal. The performance of different electrodes was studied during both ex situ acclimatization and Cr(VI) removal periods in terms of the efficiency of electricity generation and Cr(VI) removal. The effect of the MWCNTs-modied electrodes on Cr(VI) removal was deeply elucidated by conducting the extensive analyses of electrode characteristics, surface morphology, and biocatalytic activity of the bio-electrodes.

Electrode preparation
The bare electrode was a piece of graphite felt (5.0 Â 5.0 Â 0.5 cm; Hunan Jiuhua Carbon Hi-Tech Co., Ltd., China), which was soaked in 33% nitric acid overnight to remove impurities and then dried at 100 C. This bare graphite felt was dubbed the "control electrode".
To prepare the original MWCNTs-modied electrode, the original MWCNTs (outer diameter 20-40 nm, bundle length > 5 mm, purity > 95%, Shenzhen Nanotech Port Co., Ltd) were mixed with the sodium dodecyl benzene sulfonate (SDBS) and deionized water. The nal mass fraction of MWCNTs and SDBS was achieved to 0.18% and 1%, respectively. This mixture was dispersed by sonication and agitation for 1 h. Aerwards, a bare graphite felt was placed into the MWCNTs/SDBS mixture for 20 min to facilitate physical adsorption and then dried in a vacuum at 100 C for 24 h. This step was repeated thrice to maximize the adsorption of MWCNTs on the graphite felt. Subsequently, the MWCNTs-modied graphite felt was washed thrice with deionized water to remove the unadsorbed MWCNTs and dried again. This modied electrode was labelled as the "MWCNTs electrode".
The original MWCNTs were pretreated with the mixture of nitric acid and sulfuric acid (3 : 1 v/v) as described by Qu et al. 29 to obtain the oxidized MWCNTs (O-MWCNTs). Then, the modication of the O-MWCNTs on a bare graphite felt was conducted by following the same steps mentioned above. This modied electrode was labelled as the "O-MWCNTs electrode". Besides, the nal adsorption masses of the original and oxidized MWCNTs on the corresponding modied electrodes in this study were controlled to be similar (0.21-0.22 g) in order to compare the effect of these two modiers.

MFC construction and operation
The dual-chamber MFC used was the same as described in our previous study. 15 Two cubic plexiglass chambers (net volume of 70 mL each) were separated by a proton exchange membrane (38.5 cm 2 ; Naon 117, Dopont Co., USA). A sheet of graphite felt (5.0 Â 5.0 Â 0.5 cm) was used as the anode electrode, and also as the cathode electrode unless specied. The anodic inoculation was from anaerobic digester sludge. During the Cr(VI) removal experiment, an acclimatized mature bioanode with stable potential was used in the MFC, and the bioanode acclimatization method was described in Section 2.3. The anode chamber was lled with glucose culture medium during both ex situ acclimatization and Cr(VI) removal experiments. 15 The MFC was operated with an external resistance of 1000 U in a batchfed mode at 25 AE 0.5 C. All experimental reactors were performed in triplicate to ensure reproducibility.

Ex situ biocathode acclimatization method
According to the ex situ biocathode acclimatization method described previously, 15 the testing electrodes were acclimatized as anodes in MFCs. During the ex situ acclimatization period, the cathode chamber was lled with 40 mM ferricyanide and 50 mM phosphate buffer solution, 15 while the other conditions were the same as those mentioned in Section 2.2. The anolyte and catholyte were refreshed every 4-5 days. When the MFCs achieved their steady-state for voltage generation, these bioanodes were anaerobically transferred to the cathode chambers to function as the biocathodes for Cr(VI) removal.

Cr(VI) removal experiment
The medium (11.53 g L À1 Na 2 HPO 4 $12H 2 O, 2.772 g L À1 NaH 2 -PO 4 $2H 2 O, 0.28 g L À1 NH 4 Cl, 0.78 g L À1 KCl, 0.2 g L À1 NaHCO 3 ; pH 7.0) containing 20 mg L À1 Cr(VI) (prepared by dissolving K 2 Cr 2 O 7 in deionized water) was added to the cathode chamber of MFCs for testing the biocathodes obtained by the ex situ acclimatization method. In order to study the Cr(VI) removal mechanisms, the MFCs with different biocathodes were operated 24 h under both open-circuit and closed-circuit conditions. Meanwhile, the three testing electrodes before and aer acclimatization (without and with biomass) were also conducted a 24 h-Cr(VI) adsorption experiment in the same cathode medium.

Analytical techniques and calculations
The voltage generated by the MFCs was automatically collected every 10 min by a data acquisition system (Keithley Instruments 2700, USA). The power density as a function of the current density was obtained from the polarization curve, which was conducted at the time point of the maximum voltage generation in the MFC; the current and power densities were calculated based on the total surface area of the bare electrode substrate; the current (I) was calculated using Ohm's law: U ¼ IR, where U is the voltage and R is the external resistance; the power (P) was calculated according to P ¼ IU, while the internal resistance of the MFC was calculated using the polarization slope method. 15 Cyclic voltammetry (CV) was performed using a potentiostat (CHI660D, Shanghai Chenhua Instruments Co., Ltd.) at a scan rate of 5 mV s À1 over a range between À600 mV and +600 mV in a three-electrode system (Ag/AgCl as the reference electrode, the anode as the working electrode and the cathode as the counter electrode). Electrochemical impedance spectroscopy (EIS) measurement was performed to determine the different internal resistances of the MFC, namely: ohmic resistance (R s ), charge-transfer resistance (R ct ) and diffusion resistance (W). The EIS measurement was conducted in a three-electrode system (same as CV) at the potential amplitude of 10 mV with frequency range between 100 kHz and 5 mHz.
The morphology of the electrode surface was investigated by scanning electron microscopy with coupled energy dispersive spectroscopy (SEM-EDS, Hitachi S-4800, Japan). The bacteria attached to the electrode were stabilized based on previously described procedures. 33 The Fourier transform infrared (FT-IR) spectra were obtained with a FT-IR spectrometer (NICOLET, NEXUS 670). 15 The specic surface area (SSA) of the electrode was determined from a N 2 adsorption-desorption experiment with an ASAP2020 surface area analyzer (Micromeritics, USA). Contact-angle measurement was performed on the electrode with a contact angle meter (SL200B, Shanghai Solon Information Technology Co., Ltd.). The surface resistance of the electrode was determined with a digital multimeter (UT61B, Shanghai Uni-Trend Co., Ltd.). 15 At the end of ex situ acclimatization, the biomass on the anode electrode was removed into 5 mL of deionized water by ultrasonic treatment (100 W, 30 min), and the supernatant was separated from the solution by centrifugation at 2000 rpm for 2 min. Aerwards, 0.5 mL of the supernatant was mixed with 0.5 mL of NaOH solution (0.1 mol L À1 ). The mixture was boiled for 20 min until the solution was clear. Finally, the biomass protein was determined by modied Lowry method. 34 A colorimetric 1,5-diphenylcarbazide method was used for Cr(VI) analysis, and total chromium was analyzed by reoxidizing any reduced form of chromium by potassium permanganate aer acid digestion with concentrated HNO 3 and H 2 SO 4 . 35 Samples for Cr(VI) and total chromium analyses were ltered through 0.45 mm membrane syringe lters.

Electrode characterization
The surface morphology of the three electrodes was analyzed by SEM before acclimatization. As shown in Fig. 1a-c, the control electrode showed a smooth and clean surface (Fig. 1a), while the other two modied electrodes exhibited rougher surfaces caused by some affixed substances, which have the typical appearance of carbon nanotubes. 31 In particular, the tubular graphite bers of the O-MWCNTs electrode (Fig. 1c) were completely enwrapped by a layer of the evenly distributed attachments, whereas the MWCNTs electrode (Fig. 1b) displayed an irregular attachments distribution and even some chunks of agglomerate on the surface. This difference might be attributed to the oxidative acid pretreatment, which could be able to form large amount of oxygen-containing groups on the surface of MWCNTs 29 and also create porous out layer with much abundant nanoscale defects. 30 The MWCNTs with oxygencontaining groups and nanoscale defects achieved better dispersibility and stability 29,30 in aqueous solution, which might  FT-IR was further employed to investigate the functional groups induced by the oxidative acid pretreatment of MWCNTs. Fig. 2 shows the FT-IR spectra of MWCNTs and O-MWCNTs powders. The peak at about 3410 cm À1 corresponding to the stretching vibrations of -OH within -COOH increased considerably aer acid treatment in O-MWCNTs, indicating the quantity of carboxyl groups on the O-MWCNTs surface increased remarkably. The peaks at about 1400 cm À1 and 1710 cm À1 corresponding to the stretching vibration of C]O from carboxyl groups also signicantly increased, conrming that carboxyl containing functional groups were greatly generated on the O-MWCNTs. Besides, the broad peak at 1220-1080 cm À1 representing the stretching vibration of C-O from carboxyl groups increased as well. 29,30 These intensied peaks and newly generated peaks indicated that the acid pretreatment offered more carboxyl containing functional groups on the O-MWCNTs surface, which could effectively improve the dispersibility of O-MWCNTs in aqueous systems. This was also veried in other studies. [28][29][30]36 The surface characteristic analyses of the three electrodes were conducted in terms of SSA, contact angle, and surface resistance ( Table 1). The MWCNTs modications improved the SSA of the graphite felt dramatically, as the SSA values of the MWCNTs (29.84 AE 0.56 m 2 g À1 ) and O-MWCNTs (66.98 AE 0.43 m 2 g À1 ) electrode were almost 4 and 9 times higher than that of the control electrode, respectively. Large surface area can facilitate an extensive biolm formation throughout the material and efficient transport of nutrients and wastes at the same time. 37 In addition, the MWCNTs modications enhanced the wettability of the graphite felt noticeably, as the MWCNTs and O-MWCNTs electrode both became a superhydrophilic material (contact angle: 0 ) from the original hydrophobic graphite felt (contact angle: 115.81 AE 1.26 ). High hydrophilicity of materials can accelerate the bacterial adhesion and electrochemical reactions. 38 Furthermore, the MWCNTs modications decreased the surface resistance of the graphite felt, and the O-MWCNTs electrode possessed the lowest surface resistance among the three electrodes.
3.2. Electrode performance during ex situ acclimatization 3.2.1. Electricity generation. Since the three electrodes worked as anodes to enrich the biolms during ex situ acclimatization, the successful formation of biolms on the electrodes can be determined by the electricity production of the MFCs. 15 As shown in Fig. 3a, all MFCs achieved a stable maximum voltage for two consecutive cycles aer four batch cycles, indicating the successful acclimatization of biolms. Comparatively, the MFC with the O-MWCNTs anode produced the highest voltage of 0.674 V, which was 1.07 and 1.20 times as high as that of the MWCNTs (0.629 V) and control (0.564 V) anode, respectively. The MFC with the O-MWCNTs anode also generated the highest power density (0.192 AE 0.004 W m À2 , Fig. 3b), which was 1.11 and 1.41 times as high as that of the MWCNTs (0.172 AE 0.005 W m À2 ) and control (0.136 AE 0.003 W m À2 ) anode, respectively. According to Fig. 3c, the lowest internal resistance (133.52 AE 2.72 U) was obtained in the MFC with the O-MWCNTs anode. The results revealed that the utilization of MWCNTs and O-MWCNTs to modify carbon-based materials as anodes enhanced the electricity generation of MFCs, especially O-MWCNTs modication showing a better performance, which was consistent with the earlier work. 20 Furthermore, the MFCs with the MWCNTs and O-MWCNTs anode had higher current and power densities also because of the increased specic surface areas of these two electrodes, since the current and power densities are generally calculated based on the total surface area of the bare electrode substrate.
3.2.2. Electrochemical and SEM analyses. CV was conducted to understand the electrocatalytic activities of the bio-lms on the different electrodes aer the ex situ acclimatization (Fig. 4a). The O-MWCNTs bioanode showed the highest oxidative peak current of 0.013 A at À0.169 V, and the highest reductive peak current of À0.015 A at À0.254 V. In contrast, the control bioanode showed the lowest redox peak currents. Generally speaking, the oxidation-reduction peak position  indicates the redox potential of the electron transfer components of the bacteria, and the size of the oxidation-reduction peak reects the electrochemical activity of the biolm. 39 The results demonstrated that the electrochemical activity of the biolm on the O-MWCNTs electrode was the strongest, which might be ascribed to the largest amount of electrochemically active bacteria on the electrode. 15 EIS was also performed to investigate the microbial electrocatalytic activity and its interaction with electrodes (Fig. 4b).
The impedance spectra is composed of semicircle at the high frequencies (show the charge transfer) and straight line at low frequencies (show Warburg diffusion limitations). 40 Compared with the control and MWCNTs bioanode, the semicircle of the O-MWCNTs bioanode notably was the smallest. Since the diameter of the semicircle is normally tted to the interfacial electron transfer resistance (R ct ), 41 the O-MWCNTs bioanode was estimated to receive the lowest R ct of 1.1 U, while the MWCNTs and control bioanode obtained the R ct of 16.7 U and 32.6 U, respectively. This suggested superior charge transfer  between the O-MWCNTs electrode and the biolm, which also indicated that the organic matters could easily reach the O-MWCNTs bioanode surface and then react with bacteria fast. 42 The impedance decrease of the O-MWCNTs bioanode might be not only due to the improvement of the material characteristics (Table 1), but also due to the large amount of bacteria on the electrode, since the enrichment of electroactive bacteria on anode could reduce the anodic overpotential in MFCs. 43,44 Therefore, SEM and biomass protein determination were conducted to conrm the quantity of bacteria on each electrode aer the ex situ acclimatization. As seen from Fig. 1d-f, the biolm on the O-MWCNTs electrode was continuous, even and thick, while the biolm on the control electrode was sparse, uneven and thin. The total amounts of biomass protein on the three bio-electrodes are summarized in Table 1 Besides, all MFCs produced signicant voltages only before 0.4 d (10 h) operation time, and then the voltages continuously decreased until the end of the experiment (Fig. 5a), which is the typical voltage generation pattern in Cr(VI)-reducing MFCs. 1,15,17 The MFC with the O-MWCNTs biocathode still produced a background current even though the Cr(VI) was not detected in the liquid aer 0.4 d (10 h) operation time (Fig. 6), which was probably due to the Cr(VI) physically adsorbed by the biocathode as the residual electron acceptor. In addition, the small amount of oxygen leaking in through the thief hole every sampling time also might act as another electron acceptor in the cathodes to produce a part of current, especially when Cr(VI) was depleted. The calculation for the electron balances at the end of 10 h operation time (Fig. S1 †) proved the existence of other electron acceptors apart from the soluble Cr(VI) in the liquid, as the total coulombs transferred (C t ) were more than the coulombs required (C r ) for the reduction of soluble Cr(VI) in all MFCs.  [45][46][47] due to their large SSA, high porous and hollow structure, and strong interaction with the pollutant molecules. The bare O-MWCNTs electrode displayed the strongest adsorption capacity probably due to the introduction of the oxygen-containing functional groups on the O-MWCNTs which increased electrode SSA as well as the potential interaction sites for Cr(VI) ions. 48 Furthermore, the large amount of bacteria on the O-MWCNTs electrode also improved the Cr(VI) removal (35.41 AE 1.3%). 1 When the circuit was connected, the Cr(VI) removal rates in all MFCs were much higher than those observed under open circuit condition, indicating that the bioelectrochemical process in the cathode chamber accelerated the removal of Cr(VI). In other words, the Cr(VI) removal in the MFC was mainly a bioelectrochemical reduction rather than a simple adsorption. The highest Cr(VI) removal rate of 2.00 AE 0.10 mg L À1 h À1 was obtained in the MFC with the O-MWCNTs biocathode (Cr(VI) was completely removed in 10 h), which was 1.54 and 2.05 times higher than those in the MFCs with the MWCNTs (1.30 AE 0.06 mg L À1 h À1 ) and control (0.97 AE 0.02 mg L À1 h À1 ) biocathode, respectively. Among the reported studies which used a dual-chamber MFC reactor and mixed culture biocathode (as listed in Table 2), the maximum Cr(VI) removal rate in this study is 1.41 times higher than that from a selfassembling graphene modied graphite felt, 11 but lower than that from a HNO 3 -NaX modied graphite felt in our previous work. 15 This might be due to the stronger Cr(VI) adsorption capacity of the HNO 3 -NaX zeolite on graphite felt. 49 This work further conrmed that the electricity generation performance of an electrode as anode during ex situ acclimatization period could be used to estimate the Cr(VI) removal performance of it as biocathode due to the similar dominant bacteria catalyzing these electrochemical reactions. 32 Therefore, the ex situ biocathode acclimatization method could be an efficient way to screen potential biocathode materials for Cr(VI)reducing MFCs only based on the level of electricity generation at the preliminary acclimatization stage. 1,15,32 3.3.3. SEM analysis. Aer the Cr(VI) removal experiment, the three biocathodes were re-analyzed by SEM (Fig. 1g-i). Compared with the corresponding electrodes aer ex situ acclimatization, some noticeable precipitates were observed on all of the three biocathodes, which were mainly composed of Cr(OH) 3 according to the EDS detection (data not shown here) and previous research results. 1,10,12,14,17,32 The largest amount of Cr(OH) 3 deposits on the O-MWCNTs biocathode was attributed to its substantial reduction of Cr(VI), while the control biocathode had a relatively clean surface due to its lowest Cr(VI) reduction mass.
In summary, the improvement mechanisms of the O-MWCNTs electrode for Cr(VI) removal in the MFC include: (1) the O-MWCNTs and biomass on the electrode adsorbed a part of the Cr(VI) ions; (2) the easy accessibility of oxidized Cr(VI) ions onto the electrode decreased the mass transport resistance; 50 (3) the large number of bacteria on the electrode accelerated the bioelectrochemical Cr(VI) reduction.

Conclusions
This study demonstrated that the MWCNTs-modied graphite felts as biocathode materials improved the Cr(VI) removal and electricity generation in MFCs. In particular, the O-MWCNTs (MWCNTs pretreated by the oxidative acid) further promoted the performance of the modied electrode due to the induced oxygen-containing functional groups. The highest Cr(VI)