Water-activated graphite felt as a high-performance electrode for vanadium redox flow batteries

Journal of Power Sources 341 (2017) 270e279

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Water-activated graphite felt as a high-performance electrode for vanadium redox flow batteries Daniel Manaye Kabtamu, Jian-Yu Chen, Yu-Chung Chang, Chen-Hao Wang* Department of Materials Science and Engineering, National Taiwan University of Science and Technology, 10607, Taipei, Taiwan

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


The effect of water activation on the electrochemical activity of GF is investigated.
A high content of oxygen-containing groups improves the performance of VRFB.
Improved wettability due to increased surface-active oxygen functional groups.
WA-GF-5 min improves battery energy efficiency from 69.84% to 78.12% at 80 mA cm-2.
WA-GF-5 min acts as more powerful positive electrode for the VO2þ/VOþ 2 redox couple.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 August 2016 Received in revised form 28 November 2016 Accepted 2 December 2016 Available online 9 December 2016

A simple, green, novel, time-efficient, and potentially cost-effective water activation method was employed to enhance the electrochemical activity of graphite felt (GF) electrodes for vanadium redox flow batteries (VRFBs). The GF electrode prepared with a water vapor injection time of 5 min at 700  C exhibits the highest electrochemical activity for the VO2þ/VOþ 2 couple among all the tested electrodes. This is attributed to the small, controlled amount of water vapor that was introduced producing high contents of oxygen-containing functional groups, such as eOH groups, on the surface of the GF fibers, which are known to be electrochemically active sites for vanadium redox reactions. Chargeedischarge tests further confirm that only 5 min of GF water activation is required to improve the efficiency of the VRFB cell. The average coulombic efficiency, voltage efficiency, and energy efficiency are 95.06%, 87.42%, and 83.10%, respectively, at a current density of 50 mA cm-2. These voltage and energy efficiencies are determined to be considerably higher than those of VRFB cells assembled using heat-treated GF electrodes without water activation and pristine GF electrodes. © 2016 Elsevier B.V. All rights reserved.

Keywords: Vanadium redox flow battery Electrode Electrochemical activity Water activation Water vapor

1. Introduction Redox flow batteries (RFBs) are considered one of the most

* Corresponding author. E-mail address: [email protected] (C.-H. Wang). http://dx.doi.org/10.1016/j.jpowsour.2016.12.004 0378-7753/© 2016 Elsevier B.V. All rights reserved.

effective grid-scale electrochemical energy storage systems currently available because of their design flexibility, uncoupled energy and power capacities, high safety, quick response, low maintenance cost, long life cycle, and high energy efficiency [1e6]. In RFBs, energy is stored entirely within electrolytes rather than in electrode materials [7]. They store electrical energy through the chemical reaction of a

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

pair of reduced and oxidized species dissolved in two separate liquid electrolytes [8]. Among the various types of RFBs, all-vanadium redox flow batteries (VRFBs) have attracted considerable attention because they use active species of the same element as the positive 2þ 3þ electrolyte (VO2þ/VOþ 2 ) and negative electrolyte (V /V ), which substantially minimizes the problem of active component crossover contamination across the ion exchange membrane [9,10]. A standard cell potential of 1.255 V is produced through the following electrochemical reactions in a VRFB: þ  Positive electrode : VO2þ þ H2 O4VOþ 2 þ 2H þ e

¼ þ1:00 V Negative electrode : V 3þ þ e 4V 2þ E0 ¼ 0:255 V

E0 (1) (2)

Because the redox reaction of vanadium ions occur at the electrode surface in each half cell, the energy efficiency of a VRFB mainly depends on the physicochemical properties of the electrode materials, which thus must be selected carefully. Graphite felt (GF) is a typical VRFB electrode material because of its wide range of operating potentials, corrosion resistance in the acidic solution, good electrical conductivity, high mechanical and chemical stability, porous three-dimensional network structure, and low cost [11e13]. However, GF without modification has a naturally hydrophobic surface and low specific surface area, which results in poor electrochemical activity in the VRFB [14,15]. Therefore,

271

modifications are required to enhance the performance of GF [16]. Various methods have been reported by several groups, including acid treatment, electrochemical oxidation, thermal treatment, modification using nitrogen doping, and modification using metals [10,12,16e18]. The main aim of these methods was to increase the number of active sites at which the VO2þ/VO2 þ redox reaction could occur by introducing more nitrogen- or oxygen-containing functional groups onto the GF surface, such as hydroxyl (eOH), carbonyl (eCO), and carboxyl (eCOOH), which facilitates electron transfer and thus reduces the overpotential. However, these methods are unsuitable for commercial application because they use precious metals or dangerous concentrated acids, or involve difficult and time-consuming processing steps [19,20]. Thus, a new, simple, and time-efficient modification method with a reasonable cost is required to produce robust and abundant oxygen-containing functional groups. Previous studies have demonstrated that introducing a small, controlled amount of water vapor into the tube furnace enhances the catalytic activity of the materials because H2O produces large numbers of oxygen containing functional groups, such as eOH groups on the surface of the GF fibers [21e28]. Moreover, using an optimal amount of water vapor led to the elimination of impurities adhered to the surface. In this work, we report a simple, green, novel, time-efficient, and potentially cost-effective water activation (WA) method for improving the electrochemical activity of GF for use in VRFBs for the first time. Within 5 min water activation of the GFs, the VRFB

Fig. 1. SEM images at high magnification of (a) pristine GF, (b) GF-without WA, and WA-GF with the water vapor injection time of (c) 1 min, (d) 3 min, (e) 5 min, and (f) 10 min.

272

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

Fig. 2. CV curves showing (a) WA-GF electrode with various temperatures, (b) various samples as electrodes in 0.05 M VOSO4 þ 2 M H2SO4 solutions at a scan rate of 5 mV s1, and (c) the plots of peak current densities vs square root of scan rates for VO2þ/VOþ 2 redox couples for various samples.

Table 1 Electrochemical properties of various samples obtained from CV results of Fig. 2(b). Electrode

Jpa (mA cm2)

Jpc (mA cm2)

Epa (V)

Epc (V)

Jpa/Jpc

EpaEpc (V)

Pristine GF GF-without WA WA-GF-1 min WA-GF-3 min WA-GF-5 min WA-GF-10 min

41.81 44.12 48.83 52.29 58.77 54.18

20.97 29.64 32.67 34.76 41.08 35.92

1.38 1.40 1.39 1.37 1.39 1.39

0.68 0.69 0.70 0.71 0.73 0.71

1.99 1.49 1.49 1.50 1.43 1.51

0.70 0.71 0.69 0.66 0.66 0.68

using these GFs shows improved efficiencies, in which the average coulomb efficiency (CE), voltage efficiency (VE), and energy efficiency (EE) are 95.06%, 87.42%, and 83.10%, respectively at a current density of 50 mA cm-2. These voltage and energy efficiencies are considerably higher than those of VRFB cells that used GF-without WA and pristine GF electrodes. 2. Experimental section 2.1. Preparation of modified GF electrodes In this study, 6.5-mm-thick commercial GF samples (CeTech) were used. A modified GF electrode was prepared using thermal treatment of GF in a tube furnace at various temperatures from 600 to 900  C through WA. The optimum temperature was 700  C. Prior to the heat treatment, a pristine GF sample was first placed in a quartz boat located in the middle of a quartz tube furnace. The tubular furnace was purged with N2 gas for 25 min at a flow rate of

100 sccm to eliminate air. Subsequently, the furnace temperature was raised to 700  C at a heating rate of 5  C min1 during N2 flow (50 sccm) prior to the introduction of water vapor into the chamber. When the temperature reached 700  C, the water vapor was introduced by fluxing additional N2 gas (200 sccm) through a water bubbler. To optimize the amount of water vapor entering the chamber, the injection time was varied from 1 to 10 min, and after the chosen period, N2 bubbling through water was turned off. The temperature was maintained at 700  C for 2 h under N2 flow (50 sccm) to prevent air from entering the system. Finally, the furnace was naturally cooled to room temperature. For comparison, heattreated GF was prepared by a similar process but without the introduction of water vapor into the chamber. The experimental setup is shown in the supporting information (Fig. S1). 2.2. Characterization of materials The morphologies of the samples were examined using field

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

273

electrolyte volume of 60 mL each. The concentration of electrolyte in each side was 1.6 M VOSO4 in a 2.5 M H2SO4 solution. The VRFB became fully charged by applying a constant current density of 50 mA cm-2 followed by a constant voltage of 1.6 V in the first charging process. After the first charging step, VO2þ was converted to V2þ and VOþ 2 at the negative and positive electrodes, respectively. The flow rate of the electrolyte was maintained at 30 mL min1. The chargeedischarge profiles of the VRFB were evaluated using a potentionstat/galvanostat (Bio-Logic, EC-Lab® software) at a constant current density within a fixed potential range of 0.7e1.6 V. 3. Results and discussion

Fig. 3. EIS using various GFs as electrodes in 0.05 M VOSO4 þ 2 M H2SO4 solutions.

Table 2 Fitting data of the Nyquist plot obtained from EIS results of Fig. 3. Electrode

Rs (U)

Rct (U)

Pristine GF GF-without WA WA-GF-1 min WA-GF-3 min WA-GF-5 min WA-GF-10 min

1.50 1.46 1.48 1.46 1.36 1.51

32.50 20.28 18.92 15.83 11.79 12.68

emission scanning electron microscopy (FESEM, JSM 6500F, JEOL). X-ray photoelectron spectroscopy (XPS, Thermal K-Alpha) was used to identify the functional groups of the samples. The wettability of the samples was analyzed using contact angle measurement (FTA125). 2.3. Electrochemical measurements Electrochemical measurements of the samples were conducted using a typical three-electrode system at room temperature. A standard Hg/Hg2SO4/sat. K2SO4 (0.658 V vs. NHE) and a platinum wire were used as the reference and counter electrodes, respectively. A sample with a geometric area of 1.58 cm2 was used as the working electrode; it was placed in a holder and connected with a golden wire to the electrochemical instrument. All electrochemical measurements were conducted using a Bio-Logic (SP-240) potentiostat/galvanostat controlled by EC-Lab® software. Cyclic voltammetry (CV) was conducted on 0.05 M VOSO4 in a 2.0 M solution of H2SO4 in the approximate potential range of 0.36e1.66 V versus NHE at various scan rates (1e10 mV s1). The electrolyte was N2purged to prevent unnecessary oxidation. Furthermore, electrochemical impedance spectroscopy (EIS) was performed, wherein an AC voltage of 10 mV in the frequency range of 105e102 Hz was applied at the open circuit potential. 2.4. Evaluation of a VRFB single cell A VRFB single cell was evaluated using a 25 cm2 (5 cm  5 cm) area of the modified electrodes on both the positive and negative sides. A Nafion® 117 membrane was used as a separator. The single cell was connected to two glass tanks containing a balanced

The surface morphology changes of the GF samples were investigated using SEM. Fig. 1 depicts the SEM images of the pristine GF, heat-treated GF without water activation (GF-without WA), and water-activated GF (WA-GF) electrodes when various water vapor injection times were used. The pristine GF (Fig. 1(a)) has a relatively smooth surface and some impurities that may hinder the electron transfer and adsorption of vanadium ions [29]. The heattreated GF without WA has a slightly rough surface (Fig. 1(b)). More noticeable morphology changes are observed on the GF surface after WA. When the GF samples are water-treated for long periods, their surfaces are cleaner and rougher than those of the pristine GF and GF-without WA (Fig. 1(c)e(f)). This is due to the removal of impurities that adheres to the surface of the GF samples. Moreover, in the case of WA-GF-5 min electrode (Fig. 1(e)), more uniform small holes appear on the surface. These small holes may substantially increase the specific surface area of the GF sample [15]. A CV test was used to assess the electrochemical activity of the WA-GF electrode toward the VO2þ/VOþ 2 redox couple reaction in 0.05 M VOSO4 and 2 M H2SO4 solutions. WA-GF at 700  C exhibits the best electrochemical activity (Fig. 2(a) and S2) among all the treatment temperatures used (600e900  C). Therefore, 700  C was selected henceforth. Fig. 2(b) displays the CV curves when the pristine GF, GF-without WA, and WA-GF electrodes are used, at a scan rate of 5 mV s1. The oxidation and reduction peak current densities (Jpa and Jpc), ratio of redox peak current densities (Jpa/Jpc), peak potentials (Epa and Epc), and redox peak potential separations of all samples are summarized in Table 1. The oxidation and reduction peaks corresponding to a VO2þ/VOþ 2 couple appear at around 1.39 and 0.68 V (vs. NHE), respectively. The Jpa and Jpc determined for the samples, from high to low, are as follows: WAGF-5 min > WA-GF-10 min > WA-GF-3 min > WA-GF-1 min > GFwithout WA > pristine GF. Therefore, the oxidation and reduction peak current densities are substantially enhanced when the GF electrode is WA for any period compared with the GF-without WA and pristine GF electrodes, suggesting that the electron transfer kinetics of the VO2þ/VOþ 2 couple are greatly improved. Specifically, the WA-GF-5 min electrode exhibits the highest redox peak current density and the lowest peak potential separation (DEp) for the VO2þ/VOþ 2 redox couple, indicating that it has the best electrochemical activity and reversibility toward the VO2þ/VOþ 2 redox reaction. The CV curves of various electrodes at different scanning rates for the VO2þ/VOþ 2 redox couple are plotted in Fig. S3, in which Jpa/Jpc is almost constant for all scan rates used because of increased electron transfer on the surface of the electrodes. Furthermore, WA-GF-5 min electrode has superior reversibility for the VO2þ/VOþ 2 redox couple than the other electrodes at all scanning rates, because the peak potential separation is lower than that of the other electrodes at each scanning rate. As expected, there is a shift toward a lower potential during reduction and a higher potential during oxidation when the scan rate is increased. According to the RandlesSevcik equation, mass transfer can be assessed by plotting

274

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

Fig. 4. XPS C1s spectra of (a) pristine GF, (b) GF-without WA, (c) WA-GF-5 min, (d) WA-GF-10 min, and (e) chemical composition ratio of functional groups from curve fitting of C1s XPS spectra.

the peak current density versus the square root of the scan rate [9]. As illustrated in Fig. 2(c), the cathodic and anodic peak current densities of the VO2þ/VOþ 2 redox pair are linearly proportional to the square root of the scan rate on each electrode, indicating that the electrochemical behavior of the redox couple at each electrode

was controlled by diffusive mass transport [14]. Fig. 3 presents Nyquist plots of various electrodes in 0.05 M VOSO4 þ 2 M H2SO4 solutions with 10 mV amplitude in the frequency range from 105 to 102 Hz at the open circuit potential. All the Nyquist plots contain one semicircle in the high frequency

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

275

Fig. 5. XPS O1s spectra of (a) pristine GF, (b) GF-without WA, (c) WA-GF-5 min, (d) WA-GF-10 min, and (e) chemical composition ratio of functional groups from curve fitting of O1s XPS spectra.

range arising from charge transfer reactions at the electrolyteeelectrode interface. The radius of the semicircle reflects the charge transfer resistance, with a smaller radius indicating a lower charge transfer resistance, which in turn indicates a faster electron transfer reaction. The numerical fitting results are listed in Table 2, where Rct is the charge transfer resistance, and Rs is the bulk solution resistance of the electrolyte and electrode. The Rs values of all electrodes are nearly equal, which indicates that all samples are

properly measured under the same conditions [30]. The charge transfer resistances of the electrodes, in descending order, are as follows: pristine GF > GF-without WA > WA-GF-1 min > WA-GF-3 min > WA-GF-10 min > WA-GF-5 min. The Rct value of the pristine GF electrode is higher than those of the other electrodes, which indicates that it has poorer electrochemical activity. The Rct values of the WA-GF electrodes are considerably lower, which implies that the electrochemical activity of the electrodes is greatly enhanced by

276

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

Fig. 6. Photographs of the contact angle measurements on (a) pristine GF, (b) GF-without WA, and WA-GF with the water vapor injection time of (c) 1 min, (d) 3 min, (e) 5 min, and (f) 10 min.

the introduction of water vapor. Furthermore, Rct has a strong dependence on the length of the water vapor injection period, with the WA-GF-5 min electrode exhibiting the lowest charge transfer resistance and thus the optimal electrochemical activity toward the VO2þ/VOþ 2 redox reaction. This is consistent with the CV results. To discover which functional groups are chiefly responsible for the activity enhancement of the electrodes, XPS was performed for the pristine GF, GF-without WA, WA-GF-5 min, and WA-GF-10 min electrodes. Fig. 4(a)e(d) illustrate the high-resolution C1s XPS spectra of various samples and their corresponding atomic concentrations. The XPS peaks are fitted using a sufficient number of GaussianeLorentzian curves. The C1s spectra can be deconvoluted into four peaks at the binding energies of approximately 284.4, 285.3, 286.6, and 288.9 eV, corresponding to C]C, CeC, CeO, and O]CeO functional groups, respectively [16,21,31]. Compared with the pristine GF, GF-without WA, and WA-GF-10 min electrodes, WA-GF-5 min electrode has more CeC, CeO, and O]CeO groups (Fig. 4(e), Table S1) but lower C]C functional groups, indicating that the sp2 carbon hybridization is transformed into the sp3 mode by the WA process [21]. This confirms that the functional groups are grafted onto the surface of GF for WA-GF-5 min electrode. The O1s spectra can be also deconvoluted into four peaks at binding energies of approximately 531.8, 532.5, 533.1, and 534.2 eV, corresponding to C]O, eOH, CeC]O, and HeOeH bonds, respectively (Fig. 5(a)e(d)) [9,15,16,31]. More eOH functional groups but fewer C]O functional groups are found on the WA-GF-5 min electrode (Fig. 5(e), Table S2), which demonstrates that the eOH groups chiefly originated from the breakage of C]O bonds during the modification process [15]. Additionally, more CeC]O functional groups are found on the WA-GF-5 min electrode than on the others, which are directly associated with vanadium redox sites [31]. We can thus conclude that many more oxygen-containing functional

groups on the surface of GF fibers are eOH groups, which are known to be more electrochemically active sites for vanadium redox reaction than other groups [20]. Consequently, the high contents of oxygen-containing functional groups in the WA-GF-5 min electrode are responsible for improving the electrode's electrochemical activity toward the VO2þ/VOþ redox reaction. 2 Furthermore, the amount of hydroxyl and carboxyl groups on the GF fiber surface enhances its hydrophilicity, which makes it more favorable for electrochemical reactions. To investigate the effect of surface treatment on the hydrophilicity of GF, contact angle measurements were performed on each sample by using the water droplet method. The contact angles of water on the pristine GF, GF-without WA, WA-GF-1 min, WA-GF-3 min, WA-GF-5 min, and WA-GF-10 min surfaces are 133.3 , 123.1, 116.6 , 100.3 , 60.8 , and 80.1, respectively (Fig. 6). The wettability of the WA-GF electrodes is substantially improved compared with the GF electrode without WA and pristine GF. This may be attributed to the hydrophilicity of the oxygen-containing functional groups on the surface, such as hydroxyl and carboxyl groups, which are favorable to electrochemical reactions [32]. A chargeedischarge test was performed using a VRFB single cell to further demonstrate the effect of GF on the electrochemical performance of the cell before and after WA. Fig. 7(a) presents the chargeedischarge curves for the cells with pristine GF, GF-without WA, and WA-GF-5 min electrodes at a current density of 50 mA cm2 . The VRFB cell with the WA-GF-5 min electrode exhibits a longer chargeedischarge time, higher discharge voltage plateau, and lower charge voltage plateau than the other cells, which results in higher voltage and energy efficiency. This is because introducing a small, controlled amount of water vapor into the furnace during thermal treatment produced large amounts of oxygen-containing functional groups on the GF surface and promoted faster charge

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

277

Fig. 7. Electrochemical performances of VRFBs cells with pristine GF, GF-without WA, and WA-GF-5 min: (a) Charge-discharge curves at the current density of 50 mA cm-2; (b) CE, (c) VE, and (d) EE of VRFBs as a function of cycle number at different charge/discharge current densities (50e80 mA cm2).

Table 3 Summary of the efficiencies of VRFB cells recorded at various current densities. Cell

Pristine GF

GF-without WA

WA-GF-5 min

Current density (mA cm2)

50 60 70 80 50 60 70 80 50 60 70 80

5 cycles average efficiency (%) CE

VE

EE

94.71 95.24 95.83 96.40 95.12 95.42 95.82 96.40 95.06 95.53 96.02 96.32

83.02 79.50 76.64 72.46 86.13 83.67 80.47 77.73 87.42 85.49 82.84 81.10

78.63 75.72 73.45 69.84 81.93 79.84 77.10 74.78 83.10 81.67 79.54 78.12

transfer, leading to improved electrode performance. Fig. 7(b)e(d) display the CE, VE, and EE of cells with different electrodes as a function of cycle number at current densities varying from 50 to 80 mA cm-2. The average CE, VE, and EE of the cells for five cycles at each current density are summarized in Table 3. For all cells, as the current density increases, the CE increased slightly owing to the reduced time of vanadium ion crossover through membranes caused by the reduction in the chargeedischarge time. However, both the VE and EE are lower because a fast chargeedischarge rate causes large chargeedischarge overpotential [33]. The average EE of the cells that used the WA-GF-5 min electrode reached 83.10%

when a current density of 50 mA cm-2 is used, which is 1.17% and 4.47% more efficient than those of the cells that used the GFwithout WA and pristine GF electrodes, respectively. These differences become more substantial when a higher current density is used. For instance, a VRFB assembled with the WA-GF-5 min electrode has an EE of 78.12% even at a high current density of 80 mA cm-2, which is around 3.34% and 8.28% more efficient than the cells assembled with GF-without WA and pristine GF, respectively. Fig. 8 shows the CE, VE, and EE after 30 chargeedischarge cycles at a current density of 50 mA cm-2. The average CE values of the cells assembled with the pristine GF, GF-without WA, and WA-GF-5 min are nearly equal (Fig. 8(a)), which is due to a similar overall self-discharge effect on each electrode [34]. However, the average VE (Fig. 8(b)) and EE (Fig. 8(c)) of the cell with the WA-GF-5 min electrode are much higher than those of the cells with the GFwithout WA and pristine GF electrodes. Stability tests reveal no apparent decay in each efficiency even after 30 cycles, implying that the WA-GF-5 min electrode has high stability during the redox reaction of vanadium ions under highly acidic conditions. From all the experimental results, we can conclude that the use of a small, controlled amount of water vapor to modify GF led to a substantial improvement in the electrochemical performance of the VRFB that used it as an electrode. This is because H2O produces numerous oxygen-containing functional groups (such as eOH groups) on the surface of the GF fibers, which are known to be electrochemically active sites for vanadium redox reaction. Furthermore, an increase of hydroxyl and carboxyl groups on the GF fiber surface enhances its hydrophilicity, which makes it

278

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

Fig. 8. The (a) CE, (b) VE, and (c) EE of 30 charge-discharge cycles of the cell with pristine GF, GF-without WA, and WA-GF-5 min at a current density of 50 mA cm2.

Table 4 Comparison of the CE, VE and EE of WA-GF-5 min material with previously reported GF materials modified by thermal treatment. Material

Method

Electrolyte concentrations

WA-GF-5 min WA in tube furnace (700  C, 5 min)

Flow rate

Current density (mA cm2)

CE (%)

1.6 M VOSO4 þ 2.5 M H2SO4 30 mL min1 6 mm GF

50 50

95.06 87.42 83.10 This work 96.9 75.5 73.2 [29]

50

95.0

81.3

77.2

[17]

20

94.2

92.89 87.5

[35]

30

94.9

87.8

83.3

[34]

20

94.5

85.2

80.5

[36]

50

97.3

75.7

73.6

[19]

10 40

81.3 80.2

94.7 93.5

77.0 75.0

[12] [2]

GF-15 min

Microwave (400  C, 15 min)

0.5 M VOSO4 þ 1 M H2SO4

e

GF

modified Hummers method

0.5 M VOSO4 þ 1 M H2SO4

30 L/h

N-doped GF

CVD (900  C, 2.5 h)

1.2 M VOSO4 þ 3 M H2SO4

e

CVD (400e420  C in N2 for 3 h and H2 reduction at 450e460  C) PAN-based GF CVD (500  C, 24 h)

1.2 M VOSO4 þ 3.0 M H2SO4 e

CNF-GF

1.0 M VOSO4 þ 3 M H2SO4 

PAN-based GF Hydrothermal ammoniated treatment (180 C, 1.5 M VOSO4 þ 3 M H2SO4 15 h) 1 M VOSO4 þ 2 M H2SO4 N-CNT/GF CVD (800  C) CVD (500  C for 5 h) mild 2 M VOSO4 þ 2.5 M H2SO4 oxidation of CF

favorable to electrochemical reaction. A WA-GF-5 min is identified as the most powerful positive electrode for use in a VRFB. Finally, we summarized the findings of several studies (Table 4) to compare the electrochemical performance of our material with previously reported VRFBs using thermally treated GF materials. The VRFB cell using the WA-GF-5 min electrode has an EE of 83.10% at a high current density of 50 mA cm-2, which is superior to those achieved by previously reported VRFBs using GF electrodes

Electrode thickness

3 mm PANGF 3 mm PANGF 5 mm PANGF PAN- GF

10 cm3 min1 6 mm PANGF e 6 mm PANGF 15 mL min1 GF e CF

VE (%)

EE (%)

Ref.

modified by thermal treatment. 4. Conclusions Modified GF electrodes were prepared through thermal treatment in a tube furnace and using WA. Introducing a small, optimal amount of water vapor into the furnace modified the GF electrode substantially and improved the electrochemical performance of the

D.M. Kabtamu et al. / Journal of Power Sources 341 (2017) 270e279

VRFB into which the GF electrode was inserted. Among all the tested electrodes, the WA-GF-5 min electrode exhibits the highest electrochemical activity and reversibility for the VO2þ/VOþ 2 redox reaction. This is because numerous oxygen-containing functional groups, such as eOH groups, are produced on the surface of GF fibers during WA, which are known to be electrochemically active sites for vanadium redox reactions. Furthermore, the high contents of hydroxyl and carboxyl groups on the GF fiber surface enhance its hydrophilicity, which make it favorable to electrochemical reaction. Only 5 min of GF water activation is required to improve the average CE, VE, and EE of the VRFB that used the GF electrode. The average CE, VE, and EE are 95.06%, 87.42%, and 83.10%, respectively, at a current density of 50 mA cm-2. The VE and EE are considerably higher for the VRFB cells containing the WA-GF electrodes than for the cells containing the GF-without WA and pristine GF electrodes. Notably, these high efficiencies are maintained even at higher current densities. Stability tests reveal no apparent decay in efficiency after 30 cycles, implying that the WA-GF-5 min has high stability during the redox reaction of vanadium ions under highly acidic conditions. Simple surface treatment of GF using WA is thus promising for the assemblage of high-performance VRFBs, and we believe that this method is suitable for large-scale production of economical GF electrodes. Acknowledgments The authors would like to thank the Institute of Nuclear Energy Research, Atomic Energy Council supporting this research. Also, the authors deeply appreciate Prof. Li-Chyong Chen of National Taiwan University and Prof. Kuei-Hsien Chen of Academia Sinica of their great help. Technical support from the Core facilities for nanoscience and nanotechnology at Academia Sinica in Taiwan, is acknowledged. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.12.004. References [1] A. Ejigu, M. Edwards, D.A. Walsh, ACS Catal. 5 (2015) 7122e7130. [2] K.J. Kim, Y.-J. Kim, J.-H. Kim, M.-S. Park, Mater. Chem. Phys. 131 (2011) 547e553. [3] J. Liu, S. Liu, Z. He, H. Han, Y. Chen, Electrochim. Acta 130 (2014) 314e321. [4] Z. Gonzalez, C. Botas, P. Alvarez, S. Roldan, C. Blanco, R. Santamaria, M. Granda, R. Menendez, Carbon 50 (2012) 828e834.

279

[5] S.-K. Park, J. Shim, J.H. Yang, C.-S. Jin, B.S. Lee, Y.-S. Lee, K.-H. Shin, J.-D. Jeon, Electrochim. Acta 116 (2014) 447e452. [6] D.M. Kabtamu, J.-Y. Chen, Y.-C. Chang, C.-H. Wang, J. Mater. Chem. A 4 (2016) 11472e11480. [7] Z. Gonzalez, C. Botas, C. Blanco, R. Santamaria, M. Granda, P. Alvarez, R. Menendez, J. Power Sources 241 (2013) 349e354. [8] K.J. Kim, M.-S. Park, Y.-J. Kim, J.H. Kim, S.X. Dou, M. Skyllas-Kazacos, J. Mater. Chem. A 3 (2015) 16913e16933. [9] M. Park, Y.J. Jung, J. Kim, H. Lee, J. Cho, Nano Lett. 13 (2013) 4833e4839. [10] A. Parasuraman, T.M. Lim, C. Menictas, M. Skyllas-Kazacos, Electrochim. Acta 101 (2013) 27e40. [11] S. Park, H. Kim, J. Mater. Chem. A 3 (2015) 12276e12283. [12] S. Wang, X. Zhao, T. Cochell, A. Manthiram, J. Phys. Chem. Lett. 3 (2012) 2164e2167. [13] H.J. Lee, H. Kim, J. Electrochem. Soc. 162 (2015) A1675eA1681. [14] P. Han, H. Wang, Z. Liu, X. Chen, W. Ma, J. Yao, Y. Zhu, G. Cui, Carbon 49 (2011) 693e700. [15] L. Zhang, Z.-G. Shao, X. Wang, H. Yu, S. Liu, B. Yi, J. Power Sources 242 (2013) 15e22. [16] T. Liu, X. Li, H. Nie, C. Xu, H. Zhang, J. Power Sources 286 (2015) 73e81. [17] X. Wu, H. Xu, Y. Shen, P. Xu, L. Lu, J. Fu, H. Zhao, Electrochim. Acta 138 (2014) 264e269. [18] M.H. Chakrabarti, N.P. Brandon, S.A. Hajimolana, F. Tariq, V. Yufit, M.A. Hashim, M.A. Hussain, C.T.J. Low, P.V. Aravind, J. Power Sources 253 (2014) 150e166. [19] T. Wu, K. Huang, S. Liu, S. Zhuang, D. Fang, S. Li, D. Lu, A. Su, J. Solid State Electrochem. 16 (2011) 579e585. [20] K.J. Kim, S.W. Lee, T. Yim, J.G. Kim, J.W. Choi, J.H. Kim, M.S. Park, Y.J. Kim, Sci. Rep. 4 (2014) 6906. [21] M. Ran, W. Sun, Y. Liu, W. Chu, C. Jiang, J. Solid State Chem. 197 (2013) 517e522. [22] S.P. Patole, P.S. Alegaonkar, H.-C. Lee, J.-B. Yoo, Carbon 46 (2008) 1987e1993. [23] K.-Y. Lee, W.-M. Yeoh, S.-P. Chai, S. Ichikawa, A.R. Mohamed, J. Ind. Eng. Chem. 18 (2012) 1504e1511. [24] M. Bansal, C. Lal, R. Srivastava, M.N. Kamalasanan, L.S. Tanwar, Phys. B Condens. Matter 405 (2010) 1745e1749. [25] S. Hussain, R. Amade, E. Bertran, Mater. Chem. Phys. 148 (2014) 914e922. [26] A. Okamoto, I. Gunjishima, T. Inoue, M. Akoshima, H. Miyagawa, T. Nakano, T. Baba, M. Tanemura, G. Oomi, Carbon 49 (2011) 294e298. [27] D.N.F. Satoshi Yasuda, Takeo Yamada, Junichi Satou, Akiyoshi Shibuya, Hirokazu Takai, M.Y. Kouhei Arakawa, Kenji Hata, ACS Nano 3 (2009) 4164e4170. , Materials 3 (2010) [28] A. Magrez, J.W. Seo, R. Smajda, M. Mioni c, L. Forro 4871e4891. [29] X. Wu, H. Xu, P. Xu, Y. Shen, L. Lu, J. Shi, J. Fu, H. Zhao, J. Power Sources 263 (2014) 104e109. [30] Y. Shen, H. Xu, P. Xu, X. Wu, Y. Dong, L. Lu, Electrochim. Acta 132 (2014) 37e41. [31] M. Park, J. Ryu, Y. Kim, J. Cho, Energy Environ. Sci. 7 (2014) 3727e3735. [32] W. Wang, Z. Wei, W. Su, X. Fan, J. Liu, C. Yan, C. Zeng, Electrochim. Acta 205 (2016) 102e112. [33] B. Li, M. Gu, Z. Nie, Y. Shao, Q. Luo, X. Wei, X. Li, J. Xiao, C. Wang, V. Sprenkle, W. Wang, Nano Lett. 13 (2013) 1330e1335. [34] Z. He, L. Liu, C. Gao, Z. Zhou, X. Liang, Y. Lei, Z. He, S. Liu, RSC Adv. 3 (2013) 19774e19777. [35] Z. He, L. Shi, J. Shen, Z. He, S. Liu, Int. J. Energy Res. 39 (2015) 709e716. [36] C. Flox, M. Skoumal, J. Rubio-Garcia, T. Andreu, J.R. Morante, Appl. Energy 109 (2013) 344e351.