MoO₂ nanocrystals interconnected on mesocellular carbon foam as a powerful catalyst for vanadium redox flow battery

Abstract

To increase the performance of vanadium redox flow battery (VRFB), the rate of the reaction of the VO²⁺/VO₂⁺ redox couple, known as the rate determining reaction, should be increased. To increase the rate of this reaction, molybdenum dioxide nanocrystals interconnected on mesocellular carbon foam (MoO₂/MSU-F-C) are suggested as a new catalyst. Initially, the optimal amount of MoO₂ embedded on MSU-F-C is determined, whereas its activity, reversibility and charge–discharge behavior are investigated. The specific surface area, crystal structure, surface morphology and component analysis of the composite are also measured using BET, XRD, TEM, TGA, EELS and XPS. As a result, the MoO₂/MSU-F-C results in a high peak current, small peak potential difference and high electron transfer rate constant, confirming that the composite is an excellent catalyst for the VO²⁺/VO₂⁺ redox reaction. In terms of multiple charge–discharge tests, a VRFB single cell, including MoO₂/MSU-F-C, induces high voltage and energy efficiencies with high specific capacity and a low capacity loss rate. These results are attributed to the intercalation of MoO₂ by metal cations such as VO²⁺ and VO₂⁺ and the existence of hydrophilic functional groups on the surface of MoO₂/MSU-F-C. The intercalated MoO₂ plays an excellent conductor role in promoting fast ionic and electron transfer and reducing overpotential, whereas the hydrophilic functional groups improve the VO²⁺/VO₂⁺ redox reaction by lowering its activation energy.

---

1 Introduction

As demand for the continuous development and exploitation of renewable energies, such as solar cell and wind power increases, the deployment of large scale energy storage systems (ESS) is becoming increasingly important. Redox flow batteries (RFBs) have been acknowledged as a viable candidate for large scale ESS.^1–4 RFBs are electrochemical systems that conserve and store energy by the reactions of redox couples, and many efforts have been made to determine the most promising redox couples and RFB systems. Among the numerous types of RFBs explored to date, all-vanadium redox flow batteries (VRFBs) have attracted much attention due to their various desirable characteristics.^3–9 In a VRFB, both electrodes are filled with solutions, including four different vanadium ions. In the positive electrode, the reservoir is filled with the VO²⁺/VO₂⁺ redox couple, whereas in the negative electrode, the reservoir is filled with V³⁺/V²⁺. With the use of two different redox couples that both contain vanadium ions, advantages such as low cross contamination of metal cations, low environmental impact and long lifetime are possible.^10,11

However, in spite of these benefits, mass production of VRFBs has not been yet realized because their electrochemical activity is still low, resulting in poor performance of the VRFBs.¹² Because the electrochemical activity and performance of VRFBs are related to the competition between the reactions occurring in both electrodes, it is critical to discover which of the reactions occurring at the two electrodes is the rate determining step, and to increase the rate of that reaction. Regarding the rate determining reaction, it is known that the reaction rate of the redox couple (VO²⁺/VO₂⁺) occurring at the positive electrode is lower, and the reaction is more complicated than that (V²⁺/V³⁺) occurring at the negative electrode.^13,14 Previous study results showed that the low VO²⁺/VO₂⁺ reaction rate can be mainly attributed to (i) the unfavorable surface conditions of the electrode material and (ii) the complicated reaction steps (one electron and two proton transfers).¹⁵

Extensive attempts have been made to overcome these problems. Among these, graphite felt was selected as a baseline electrode material due to its high surface area and good stability in acidic solution.¹² However, the hydrophobic property of the graphite felt surface prevented further facilitation of the VO²⁺/VO₂⁺ reaction because the VO²⁺/VO₂⁺ redox reaction was prohibited at the hydrophobic surface and more activated at the hydrophilic surface. To fortify the hydrophilicity and increase the reactivity of the VO²⁺/VO₂⁺ reaction, two ideas have been recently considered, (i) oxidation of the graphite felt and (ii) utilization of a noble metal catalyst.

Oxidized graphite is receiving attention because its structure contains an increased number of oxygen functional groups that can facilitate the vanadium reaction, while retaining the unique merits of graphite.¹⁶ Moreover, noble metals, such as Pt, Pd and Ir, have been used to enhance the catalytic activity of the vanadium redox reaction.^4,17–19 However, in the case of graphite oxide, the issue of reduction of the electron/ion conductivity should be addressed; moreover, noble metal catalysts are very expensive and are sensitive to undesirable hydrogen evolution reactions. As another approach to promote the vanadium redox reaction, the adoption of inexpensive metal oxides, such as Mn₃O₄, WO₃ and Nb₂O₅, have been suggested.^20,21 It was also proved that the use of metal oxides induced an improvement in VRFB performance, although the degree of the improvement was not significant.

In this study, to pave the way for enhancing the VO^2+/VO₂⁺ reaction and VRFB performance, we suggest a new catalyst that is configured as a combinational structure of mesocellular carbon foam (MSU-F-C) and metal oxide nanocrystal (MoO₂). The MSU-F-C consists of large interconnected pores and is open in all directions. Therefore, ions and electrons can diffuse rapidly through the pores of the MSU-F-C. Moreover, the MSU-F-C contains oxygen functional groups which play a role in enhancing the vanadium redox reaction, whereas the other features of carbon remain unchanged. To explain in detail, the C–O and C=O bonds of the oxygen functional groups are the main components that influence the vanadium redox reaction,^5,22 namely, the carbon–oxygen bonds act as active sites, catalyzing the VO²⁺/VO₂⁺ redox reaction. For example, during the charge process (VO²⁺ → VO₂⁺), VO²⁺ ions diffuse to the electrode and react with hydrogen that is adjacent to the C–O bonds on the electrode surface. An electron is then transferred from the VO²⁺ ion to the electrode along with the C–O–V bond. An oxygen atom released from the C–O on the surface bonds with VO²⁺ to form VO₂⁺. Finally, VO₂⁺ ions are replaced with H⁺ ions provided by the electrolyte and move back to the electrolyte. By this mechanism, electron transfer is facilitated from the C–O–V bond, and VO²⁺ ion binds to the oxygen atom released from C–O to generate VO₂⁺. In the discharge process (VO₂⁺ → VO²⁺), the opposite reactions to the charge process occur.

Unlike the other metal oxide candidates studied to date, molybdenum dioxide (MoO₂) nanocrystal has a distorted rutile structure and is known for its low electrical resistivity, high chemical stability in acidic solution, high volume capacity and high surface acidity. Due to these advantages, catalysts such as MoO₂ have been used in the fields of lithium-ion batteries and hydrogen storage. Thus, it is anticipated that the use of MoO₂ may promote the vanadium redox reaction.^23–25

Taken together, the addition of the new catalyst consisting of MSU-F-C and MoO₂ to graphite felt would be expected to result in enhancement of the vanadium redox reaction and in the performance of the VRFB. Following the employment of MSU-F-C and MoO₂ on graphite felt, the catalytic activity and VRFB performance were investigated to estimate whether the catalyst facilitated the VO²⁺/VO₂⁺ redox reaction and enhanced the performance of the VRFB.

2 Experimental

2.1 Synthesis of catalysts

The synthesis of mesocellular carbon foam, MSU-F-C, was already reported, and we followed the procedure for this study.^26,27 To generate acidic sites on the silica template, the MSU-F silica was mixed with AlCl₃ in DI water solution (the Si/Al weight ratio was 20). After drying, the mixture was heat-treated at 550 °C under air. As a carbon source, 1 g of furfuryl alcohol was impregnated in the pores of 1 g of aluminated MSU-F silica by the incipient wetness method. After polymerization at 85 °C for 5 h under vacuum, the material was carbonized at 850 °C for 2 h under Ar flow. To remove the aluminated silica, the product was stirred in 5 wt% HF solution for 8 h, followed by washing with DI water.

For the MoO₂ impregnation, 0.1 g of MSU-F-C was dispersed in 20 ml of ethanol. To impregnate the molybdenum precursors, different weights (0.05, 0.1, and 0.15 g) of phosphomolybdic acid (PMA) hydrate were added to MSU-F-C/ethanol solution. After stirring for 30 min, the mixture was dried at 60 °C and calcined at 500 °C in H₂/N₂ (4% H₂) atmosphere for 3 h (1 °C min⁻¹).

2.2 Catalytic activity and reversibility measurements

The electrochemical measurements such as cyclic voltammetry (CV) were performed using a computer connected potentiostat (CHI 720D, CH Instruments, USA). CV was performed to estimate (i) the catalytic activity and reaction reversibility of the VO²⁺/VO₂⁺ reaction and (ii) the electron transfer rate constant. Pt wire was used as a counter electrode and Ag/AgCl (soaked in 3 M KCl) acted as a reference electrode. To build the working electrode, catalytic powder was mixed with isopropanol and 5 wt% Nafion solution under sonication for 3 min. After mixing, the catalytic ink was dropped on a glass carbon disk electrode (GCE) with a loading amount of 7 μL. After loading, the working electrode was dried for 1 h at room temperature.^28,29

1 M VOSO₄ + 1 M H₂SO₄ mixture was used as the electrolyte for the VRFB tests. The vanadium solution used was prepared by dissolving VOSO₄ (Sigma Aldrich, 97%) in H₂SO₄ with stirring, and all tests were performed at room temperature. In particular, the potential scan rate was used to determine both the rate determining process and the electron transfer rate constant using Laviron's model. In all the related tests, the potential scan rate of the CVs was increased from 5 to 50 mV s⁻¹ by increase of 5 mV s⁻¹.

2.3 VRFB single cell fabrication and cyclic tests of the charge–discharge sequence

The VRFB single cells had the same configuration as previously reported cells.^4–6 A 5 mm thick PAN graphite felt (XF30A, Toyobo, Japan) was used as both the positive and negative electrode. The total active area of the single cell was 4 cm², and Nafion 212 was used as a membrane. The tanks placed in both the negative and positive electrodes were filled with a solution containing 1 M VOSO₄ + 1 M H₂SO₄. The VOSO₄ and H₂SO₄ solutions were mixed under vigorous stirring until the blue color of V⁴⁺ appeared.⁶

To evaluate the effect of catalyst coated on the graphite felt on VRFB performance, the catalysts were sprayed onto the surface of the graphite felt. 10 mg of the catalyst powder was initially mixed with 750 μL isopropanol and 20 μL Nafion, and the ink was sonicated for 3 min and then coated on the graphite felt using air spraying. The VRFB single cell tests consisted of a pre-activation step, a charging step and a discharging step. We used the same procedure as in previously reported studies.^4,5

2.4 Characterization of catalysts

The structures of MoO₂/MSU-F-C and pure MSU-F-C were characterized by scanning electron microscopy (SEM, S-4200 field emission SEM, Hitachi) and transmission electron microscopy (TEM, JEM-1011, JEOL LTD). Electron energy loss spectroscopy (EELS) images were obtained by high-resolution TEM (HR-TEM, JEOL EM-2010). Powder X-ray diffraction (XRD) patterns were obtained using a Bruker D8 Advance X-ray diffractometer (Cu Kα radiation). Nitrogen physisorption tests were conducted using a Tristar II 3020 instrument at 77 K (Micromeritics Instrument Co.). A NETZSCH STA 449C thermobalance was used for the thermogravimetric analysis (TGA, heating rate: 2 °C min⁻¹). To characterize the surface states and degrees of the functional groups, XPS (ESCA LAB250, VG Scientific) analysis was performed. Water contact angle analysis was conducted using a contact angle measurement system (Krüss, Model DSA-10, Hamburg, Germany) with 7 μl deionized water. The drop of water was recorded with a web-camera (@info, ALC-M1000).

3 Results and discussion

3.1 Structural characterizations of the corresponding catalysts

A schematic explaining the synthesis of the MoO₂/MSU-F-C samples is shown in Scheme 1. MSU-F-C was prepared by a hard template method using mesocellular silica foam as a template.²⁷ The structure of the MSU-F-C consists of (i) cellular pores with diameters of ∼30 nm and (ii) small pores additionally generated by the removal of the silica template. The large and 3-D open pores facilitate access of the electrolyte to the carbon surface, whereas the high surface area of MSU-F-C provides abundant active sites for high loading of MoO₂ particles.^30,31 For MoO₂ catalyst loading, different weights of the molybdenum precursors (phosphomolybdic acid hydrate, PMA) were impregnated inside the pores. The ratios of precursor to MSU-F-C were 0.5 : 1, 1 : 1, and 1.5 : 1, and the catalysts are denoted as MoO₂/MSU-F-C-1, MoO₂/MSU-F-C-2, and MoO₂/MSU-F-C-3, respectively. During heat-treatment under H₂/N₂ (4%) conditions, PMA was decomposed to nano-sized molybdenum oxides and the particles were well-dispersed inside the small and large pores of MSU-F-C.

Scheme 1 Schematic of the synthesis of MoO₂-mesocellular carbon foam composite (MoO₂/MSU-F-C).

Nitrogen physisorption was used to analyze the structural characterizations of the samples (see Fig. 1). Isotherms of the nitrogen physisorption show two distinct capillary condensations at ∼0.6 and ∼0.9 P/P₀ (Fig. 1a). This result indicates that two different sized mesopores are uniformly produced inside the carbons. The Brunauer–Emmett–Teller (BET) surface area and pore volume of the MSU-F-C were 880 m² g⁻¹ and 1.64 cm³ g⁻¹, respectively. After impregnation with the MoO₂ catalyst, the surface area and pore volume decreased with increasing catalyst loading. The values of all the catalysts are summarized in Table 1. According to the Barrett–Joyner–Halenda (BJH) pore size distribution (PSD), calculated from the adsorption branches of the isotherms, MSU-F-C acquired bimodal pores with average pore sizes of 5 and 30 nm (Fig. 1b). The PSD peak intensities of both the small and large pores decreased as the catalyst loading increased. This may be attributed to the formation of small nanoparticles that are a few nanometers in size. The SEM images displayed in Fig. S1† also support that there are neither separated particles outside the MSU-F-C nor noticeable MoO₂ particles inside the pores.

Fig. 1 (a) Nitrogen physisorption isotherms and (b) BJH pore size distribution graphs of the MSU-F-C and MoO₂/MSU-F-C samples.

Table 1 Specific surface areas and total pore volumes of the MSU-F-C and MoO₂/MSU-F-C samples^a

	MSU-F-C	MoO2/MSU-F-C-1	MoO2/MSU-F-C-2	MoO2/MSU-F-C-3
a Measured at P/P0 = 0.99.
Specific surface area [m^2 g^−1]	880	641	469	384
Total pore volume [cm^3 g^−1]	1.64	1.21	0.85	0.69

---

The crystal structures of the samples were investigated by X-ray diffraction (XRD), and the results are represented in Fig. 2a. The XRD pattern of MSU-F-C shows broad two peaks at 23.1° and 43.7°, which correspond to the (002) and (101) diffractions of typical amorphous carbons, whereas in the case of MoO₂/MSU-F-C, all peaks coincide with those of monoclinic MoO₂ (JCPDS #: 86-0135). The peaks of MoO₂/MSU-F-C-1 and MoO₂/MSU-F-C-2 appeared unclear. The data indicate that ultrafine MoO₂ nanoparticles are well dispersed on the carbon surface or inside the carbon walls.³² In contrast, MoO₂/MSU-F-C-3, which has a high Mo content, shows a clear XRD pattern for monoclinic MoO₂, indicating that large, exposed MoO₂ nanocrystals were formed; this will be discussed with the TEM analysis.

Fig. 2 (a) X-ray diffraction patterns of MSU-F-C and MoO₂/MSU-F-C samples and (b) thermogravimetric (TG) curves for MoO₂/MSU-F-C samples (heating rate: 2 °C min⁻¹).

The ratio of metal oxide to carbon was estimated using thermogravimetric analysis (TGA) under air (Fig. 2b). The weights of the MoO₂/MSU-F-C samples all decreased at 350–400 °C, meaning that the amorphous carbon in the composite materials was oxidized. The residual weights of MoO₂/MSU-F-C-1, -2 and -3 are 28%, 41%, and 57%, respectively. The theoretical weight increase that occurs during the oxidation of MoO₂ to MoO₃ is 12.5%. Based on the theoretical value, the MoO₂ contents in MoO₂/MSU-F-C-1, -2, and -3 were calculated to be 26, 38, and 54 wt%, respectively.

Transmission electron microscopy (TEM) was used to investigate the surface morphology of the MoO₂/MSU-F-C catalysts, and the TEM images were compared with the XRD results. According to the TEM images (Fig. 3a), MSU-F-C shows a typical mesocellular structure containing large pores (∼30 nm), whereas MoO₂/MSU-F-C-1 and -2 have identical porous structures (Fig. 3b and c) even after MoO₂ loading, indicating that the MoO₂/MSU-F-C-based structures are advantageous for electrolyte penetration inside the carbon. The MoO₂/MSU-F-C samples also contain well-dispersed and small-sized MoO₂ particles inside the pores (magnified images are displayed in Fig. S2†). This may affect the vanadium redox reactions in a positive way. In contrast, in the MoO₂/MSU-F-C-3 sample, large-sized or agglomerated nanoparticles (<20 nm) may have formed inside the main pores of MSU-F-C (Fig. 3d).

Fig. 3 TEM images of (a) MSU-F-C, (b) MoO₂/MSU-F-C-1, (c) MoO₂/MSU-F-C-2, and (d) MoO₂/MSU-F-C-3 samples.

To evaluate whether small MoO₂ nanoparticles are embedded well in the pores of MSU-F-C, electron energy loss spectroscopy (EELS) measurements of MoO₂/MSU-F-C-2 of the three samples were performed (Fig. 4). Although noticeable MoO₂ nanoparticles were not observed in the TEM images in Fig. 4a, the signals of Mo and O elements were detected around the porous carbon matrix by EELS. The high resolution TEM images in Fig. S3† also support the XRD, TEM, and EELS data; no noticeable nanocrystals were found in the pores or carbon matrix in the MSU-F-C-1 sample, while nanocrystals with a large particle size (∼20 nm) were formed in the mesopores of the MSU-F-C-3 sample. With these results, it was confirmed that small sized MoO₂ nanoparticles were well embedded in the pores of MSU-F-C for the MoO₂/MSU-F-C-1 and -2 samples. In the same manner, it is expected that in the MoO₂/MSU-F-C-3 sample, relatively large MoO₂ nanoparticles formed and partly agglomerated inside the main pores of MSU-F-C, indicating that the MoO₂ nanoparticles do not play a key role in catalyzing the vanadium redox reactions.

Fig. 4 (a) TEM image of MoO₂/MSU-F-C-2 and corresponding EELS images of (b) molybdenum, (c) carbon, and (d) oxygen.

To investigate the surface structure of the catalysts in detail, X-ray photoelectron spectroscopy (XPS) measurements were conducted; the peaks of C 1s and Mo 3d and their deconvolution results are represented in Fig. 5. In MSU-F-C, the C 1s peak corresponds to typical amorphous carbon structures such as C–C and C=C bonds (C1), alcohol or ether (C2), carbonyl (C3), carboxylic acid or ester (C4), and carbonate (C5).³³ After MoO₂ loading, the MoO₂/MSU-F-C-2 sample shows a similar carbon C 1s peak (Fig. 5b), proving that the surface functionalities of MSU-F-C are preserved under reductive gas conditions. These functional groups of mesoporous carbon can serve as catalytic sites for VO²⁺/VO₂⁺ redox reactions.⁵ On the other hand, the three peaks of Mo 3d observed in Fig. 5c show that the oxidation states of Mo vary in the MoO₂/MSU-F-C-2 sample. Although the crystal structure was expected to be monoclinic MoO₂ in the XRD measurements, the deconvolution data show that the major oxidation state of the MoO₂ surface is Mo⁶⁺, whereas Mo⁴⁺ and Mo⁵⁺ were the next major oxidation states. The multiple oxidation states observed in the reduced MoO₂ surface were compatible with those observed in a previous study.³⁴

Fig. 5 XPS spectra of (a) C 1s in MSU-F-C, (b) C 1s and (c) Mo 3d in MoO₂/MSU-F-C-2.

Although surface oxidation states of Mo can be produced from Mo to Mo⁶⁺ depending on the calcination conditions, Mo⁴⁺ to Mo⁶⁺ mainly formed in our experiments due to the absorption of oxygen and hydroxyl groups into MoO₂.³¹ Moreover, the unsaturated sites created by the calcination of MoO₂, such as oxygen vacancies and hydroxyl groups, acquire hydrophilic properties, resulting in improved absorption ability of the vanadium ions for activating the vanadium redox reaction.^34,35 This fact was quantified by water contact angle measurements. As shown in Fig. S4a,† the water contact angle of MoO₂/MSU-F-C (112°) was lower than that of MSU-F-C (139°). This indicates that MoO₂ and mesoporous carbon composite induce better wettability than pristine MSU-F-C. This result is also supported by previous studies. According to Jo et al.³⁶ and Zhou et al.,³⁷ the reduced forms of tungsten and molybdenum oxides are stable during electrochemical reactions, and the multiple oxidation states of the transition metal could act as additional catalytic sites due to the increased electron/ion conductivities.

3.2 Optimization of MoO₂/MSU-F-C structure

Three types of MoO₂/MSU-F-C, MoO₂/MSU-F-C-1, -2 and -3, were prepared to determine the optimal amount of MoO₂ embedded on MSU-F-C. To this end, their CV curves were evaluated and are represented in Fig. 6a. According to Fig. 6a, the oxidation and reduction peak currents (I_pc and I_pa) of MoO₂/MSU-F-C-2 were far superior to those of the other two catalysts, demonstrating that the catalytic activity of MoO₂/MSU-F-C-2 was the best for activation of the VO²⁺/VO₂⁺ redox reaction. On the other hand, the values of other factors, such as the ratio of reduction to oxidation peak current (I_pc/I_pa) and the difference in the reduction and oxidation peak potentials (ΔE_p), were similar in all the samples. This result is in good agreement with the TEM and EELS data, showing (i) an insufficient amount of MoO₂ embedded on the surface of MSU-F-C (MoO₂/MSU-F-C-1) and (ii) agglomeration of MoO₂ inside the MSU-F-C pores (MoO₂/MSU-F-C-3). Due to the compatibility of its structural and electrochemical characterizations, we considered MoO₂/MSU-F-C-2 for the subsequent processes.

Fig. 6 CV curves of (a) MoO₂/MSU-F-C-1, MoO₂/MSU-F-C-2, and MoO₂/MSU-F-C-3 catalysts and (b) MoO₂/MSU-F-C-2 and MSU-F-C catalysts measured in electrolyte of 0.1 M VOSO₄ + 0.1 M H₂SO₄. The potential scan rates for the (a) and (b) data were 70 and 100 mV s⁻¹, respectively, and the potential scan range of the CVs was from 0.4 to 1.0 V vs. Ag/AgCl.

3.3 Effects of MoO₂/MSU-F-C catalyst on the catalytic activity of the redox reaction of VO²⁺/VO₂⁺

To investigate the effect of MoO₂ on the activity and reversibility of the VO²⁺/VO₂⁺ reaction, the CV curves of the MSU-F-C and MoO₂/MSU-F-C-2 catalysts were measured, and the electrochemical properties such as (i) I_pc and I_pa, (ii) I_pc/I_pa and (iii) ΔE_p, of the two catalysts were compared (see Fig. 6b). According to Fig. 6b, the peak currents and ΔE_p of MoO₂/MSU-F-C-2 (I_pc and I_pa are 10.0 and 8.1 mA cm⁻², respectively, whereas ΔE_p is 256 mV) were better than those of MSU-F-C (I_pc and I_pa are 6.8 and 6.2 mA cm⁻², respectively, whereas ΔE_p is 265 mV), implying that MoO₂/MSU-F-C-2 facilitated the catalytic activity and reversibility of the VO²⁺/VO₂⁺ redox reaction.

Following the evaluation of the effect of MoO₂ on the VO²⁺/VO₂⁺ redox reaction (positive electrode), the effect of MoO₂ on the V²⁺/V³⁺ redox reaction (negative electrode) was also estimated to examine whether the MoO₂/MSU-F-C-2 catalyst affects (i) the V²⁺/V³⁺ redox reaction or (ii) the undesirable hydrogen evolution reaction because the two reactions occur at a similar potential range. To this end, the CV curves of the MSU-F-C and MoO₂/MSU-F-C-2 catalysts were measured in the corresponding potential range (Fig. S5†).

According to Fig. S5,† MoO₂/MSU-F-C-2 did not induce a redox peak at −0.4 to −0.6 V vs. Ag/AgCl; this is the potential range of both the V²⁺/V³⁺ and hydrogen evolution reactions. On the other hand, MSU-F-C showed V²⁺/V³⁺ reaction peaks in this potential range, indicating that MoO₂ embedded on MSU-F-C did not lead to the V²⁺/V³⁺ reaction, whereas MSU-F-C alone could promote the V²⁺/V³⁺ reaction. From the CV curves in Fig. S5,† it can be observed that MoO₂/MSU-F-C-2 was an effective catalyst for only the VO²⁺/VO₂⁺ redox reaction, confirming that MoO₂/MSU-F-C-2 could be considered as a catalyst for the positive electrode of VRFB.

The electron transfer rate constants, k_s, of the MoO₂/MSU-F-C-2 and MSU-F-C catalysts were measured because k_s is a key index (i) to quantify how rapidly electrons are transferred during the VO²⁺/VO₂⁺ redox reaction and (ii) to verify whether MoO₂/MSU-F-C-2 induces a faster electron transfer rate than MSU-F-C. To this end, CV curves of the two catalysts were measured under different scan rates using Laviron's formula.^4,5 Based on these results, the k_ss of the MoO₂/MSU-F-C-2 and MSU-F-C catalysts were calculated as 0.33 and 0.11 s⁻¹, respectively, showing that the k_s of MoO₂/MSU-F-C-2 was three times higher than that of MSU-F-C. This result is compatible with the catalytic activity and reversibility tendencies shown in Fig. S6.†

We also investigated the effect of the potential scan rate on the catalytic activity of the MoO₂/MSU-F-C-2 layer, and the result is represented in Fig. S7.† In these measurements, the redox peak currents linearly increased as the scan rate increased, demonstrating that MoO₂/MSU-F-C-2 is controlled by a surface reaction confined process.

3.4 Effect of MoO₂/MSU-F-C catalyst on VRFB performance

To inspect whether the catalytic activity and reversibility of the VO²⁺/VO₂⁺ redox reaction have any influence on VRFB performance, the charge–discharge curves of (i) VRFB single cells, including two catalysts (MoO₂/MSU-F-C-2 and MSU-F-C) and (ii) a VRFB single cell without catalyst were measured. Because the catalysts mostly affected the VO²⁺/VO₂⁺ redox reaction, they were only embedded at the positive electrode. The different VRFBs are denoted as pristine graphite felt (pristine-VRFB), MSU-F-C-deposited graphite felt (MSU-F-C-VRFB) and MoO₂/MSU-F-C-2-deposited graphite felt (MoO₂/MSU-F-C-2-VRFB).

According to the results (Fig. S8†), MoO₂/MSU-F-C-2-VRFB had higher discharging and lower charging potentials than MSU-F-C-VRFB and pristine-VRFB. This means that the overall overpotential of the MoO₂/MSU-F-C catalyst is low. A vertical potential drop was also estimated to measure the internal resistance of the corresponding VRFB.³³ Similarly to the overpotential trend, MoO₂/MSU-F-C-2-VRFB exhibited the lowest potential drop, followed by the lowest internal resistance. With these data, it is anticipated that MoO₂/MSU-F-C-2-VRFB will induce higher voltage efficiency (VE) than other VRFBs because the VE relies on overpotential and internal resistance. This also affects the reactivity (catalytic activity and reversibility) of the VO²⁺/VO₂⁺ redox reaction. Taken together, it is obvious that the MoO₂/MSU-F-C-2-VRFB, which shows (i) excellent reactivity of the VO²⁺/VO₂⁺ redox reaction and (ii) low overpotential and potential drop, improves VRFB performance, while MoO₂ plays a dominant role in eliciting excellent VRFB performance.

Fig. S9† shows the specific discharge capacities of the three VRFBs that were measured after every five cycles. According to Fig. S9,† the discharge capacity of MoO₂/MSU-F-C-2-VRFB was higher than that of MSU-F-C-VRFB and pristine-VRFB, whereas its capacity loss rate was better than that of MSU-F-C-VRFB and comparable to that of pristine-VRFB. This implies that the discharge capacity and capacity loss of MoO₂/MSU-F-C-2-VRFB are viable.

To evaluate the performance of the three VRFBs, the three efficiencies (charge efficiency (CE), VE and EE) were measured, and the results are represented in Fig. 7. According to Fig. 7, MoO₂/MSU-F-C-2-VRFB demonstrated higher VE and EE than the other VRFBs (MSU-F-C-VRFB and pristine-VRFB), while the difference in CE was not noticeably large. The difference in VE is mainly due to the difference in the potential losses of overpotential and internal resistance, as explained previously. Because the charge–discharge behavior of MoO₂/MSU-F-C-2-VRFB had a low potential loss, the difference in the charge and discharge potentials of the VRFBs was also low, resulting in an increase in VE that was a ratio of the charge and discharge potentials.³⁸ Unlike the VE, CE is determined by the difference in the amount of charges transferred during the charge and discharge step that crossed over the Nafion 212 membrane. This CE result means that the difference in the amount of charges that permeated the membrane is not large in all three VRFBs. With this explanation, it is expected that EE was more affected by VE.

Fig. 7 Charge efficiency, power efficiency and energy efficiency of VRFB single cells consisting of MoO₂/MSU-F-C-2, MSU-F-C catalysts, and graphite felt as the positive electrode at a current density of 40 mA cm⁻².

To further estimate the CE and VE effects on the EE of VRFBs, the efficiencies of the VRFBs were again measured with increases in current density; the result is represented in Fig. 8. For the tests, the applied current densities ranged from 40 to 100 mA cm⁻² with increase of 20 mA cm⁻². For the efficiency data, there are two things to note. First, CE increased with increasing current density. This means that with increasing current density, (i) the amount of charge produced increases and (ii) the amount of charge transferred during the charge and discharge steps becomes equalized, resulting in an increase in CE. Furthermore, at high current densities (80 and 100 mA cm⁻²), the CEs of MSU-F-C-VRFB and pristine-VRFB were unstable, whereas the CE of MoO₂/MSU-F-C-2-VRFB remained stable. In contrast to the stability, the increasing rates of CE with increasing current density were similar in all three VRFBs. Based on this result, it can be induced that the MoO₂ in the MoO₂/MSU-F-C-2 catalyst leads to superior CE stability.

Fig. 8 Charge efficiency, power efficiency and energy efficiency of VRFB single cells consisting of MoO₂/MSU-F-C-2, MSU-F-C catalyst, and graphite felt as the positive electrode at different current densities of 40 to 100 mA cm⁻².

The second thing to note is that VE decreased with increasing current density. This indicates that with an increase in current density, the overpotential and internal resistance increase due to Ohm's law, reducing VE. As with the CE, at high current densities, the VEs of MSU-F-C-VRFB and pristine-VRFB were unstable, whereas the VE of MoO₂/MSU-F-C-2-VRFB remained stable. With increasing current density, the VE loss rate in MoO₂/MSU-F-C-2-VRFB is lower than that in the other VRFBs. This result indicates that the MoO₂ results in excellent VE stability and a low VE loss rate with increasing current density. Regarding EE, the EE showed a similar trend to VE, confirming that VE played a more dominant role than CE in determining the EE of the VRFBs.

3.5 Role of MoO₂ and MSU-F-C in VRFB performance

Based on the half-cell and single cell data, it can be explained that the MoO₂/MSU-F-C-2 catalyst results in superior catalytic activity and reversibility for the activation of the VO²⁺/VO₂⁺ redox reaction, whereas the MoO₂/MSU-F-C-2-VRFB has excellent VE and EE, more discharge capacity and less capacity loss even in a high current density range. It was induced that the employment of MoO₂ as the catalyst led to these differences in the half-cell and single cell data.

With the adoption of MoO₂, there are three advantages to note. First, the interconnected structure between MoO₂ and MSU-F-C is supposed to promote fast ionic and electron transfers as well as reductions in the overpotential and internal resistance. According to Chernova et al.,³⁹ MoO₂ can be intercalated by metal cations, such as VO²⁺ and VO₂⁺, and the intercalated structure is a good conductor. This phenomenon is also called as “mixed conductor”. Indeed, the structure facilitates (i) superior ionic and electron conductivity and (ii) low charge resistance. Due to this reason, it is likely that the MoO₂/MSU-F-C-2 structure shows excellent electron conductivity as well as low resistance. This was confirmed by the high k_s, VE and EE values. This result was also in good agreement with previously published studies.⁴⁰

Second, MoO₂ induced more hydrophilic functional ions, such as oxygen vacancies and hydroxyl groups, at the catalyst surface, as observed in the XPS data (Fig. 5). Therefore, these hydrophilic groups played a role in reducing the activation barrier for the VO²⁺/VO₂⁺ redox reaction, producing faster charge transfer.^18,41

Third, the mesoporous structure, including MoO₂ facilitates the diffusion of liquid electrolyte into the carbon felt electrode using capillary action. This means that the MoO₂/MSU-F-C-2 structure induces fast conductive ion transport channels for conductive ions such as VO²⁺ and VO₂⁺.^42–44

4 Conclusions

A new MoO₂/MSU-F-C catalyst was used for VRFBs to improve the slow reaction of the VO²⁺/VO₂⁺ redox couple. A few nanometer-sized MoO₂ particles are loaded on the surface of MSU-F-C. The 3-D structured large pores (∼30 nm) facilitate electrolyte wetting and promote the redox reaction in VRFBs. Due to the additional catalytic effect of the hydrophilic MoO₂, the MoO₂/MSU-F-C-2 composite exhibited improved catalytic properties. When the activity and reversibility of the MoO₂/MSU-F-C catalyst were evaluated, the catalyst led to high peak currents (5.4 mA cm⁻² and 4.0 mA cm⁻²), a small peak potential difference (256 mV) and a high electron transfer rate constant (0.33 s⁻¹), indicating that the catalyst catalyzed the VO²⁺/VO₂⁺ redox reaction favorably.

During multiple charge/discharge cycle tests, MoO₂/MSU-F-C-VRFB produced high VE and EE (VE: 89% and EE: 76%), high specific capacity, low capacity loss rate and no significant changes in efficiencies even at a high current density of 100 mA cm⁻².

These superior half-cell and single cell results gained using the MoO₂/MSU-F-C catalyst are ascribed to the following three reasons. First, the MoO₂ was intercalated by metal cations such as VO²⁺ and VO₂⁺; second, a large portion of hydrophilic functional groups were doped on the surface of MoO₂/MSU-F-C; and third, the mesoporous structure containing MoO₂ enabled facile diffusion of the liquid electrolyte into the carbon felt electrode. All the desirable characteristics obtained by the employment of MoO₂ or MoO₂/MSU-F-C promote the VO²⁺/VO₂⁺ redox reaction by improving the conductivity/diffusion of the carbon felt electrode and increasing the reactivity. Eventually, it was found that the adoption of MoO₂/MSU-F-C resulted in an improvement in VRFB performance by promoting the slow VO²⁺/VO₂⁺ redox reaction.

Acknowledgements

This study was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2013R1A1A2006494).

Notes and references

1. Y. Shao, X. Wang, M. Engelhard, C. Wang, S. Dai, J. Liu, Z. Yang and Y. Lin, J. Power Sources, 2010, 195, 4375–4379.
2. A. Z. Weber, M. M. Mench, J. P. Meyers, P. N. Ross, J. T. Gostick and Q. H. Liu, J. Appl. Electrochem., 2011, 41, 1137–1164.
3. Z. González, C. Botas, P. Álvarez, S. Roldán, C. Blanco, R. Santamaría, M. Granda and R. Menéndez, Carbon, 2012, 50, 828–834.
4. S. Jeong, S. Kim and Y. Kwon, Electrochim. Acta, 2013, 114, 439–447.
5. S. Jeong, S. An, J. Jeong, J. Lee and Y. Kwon, J. Power Sources, 2015, 278, 245–254.
6. S. Jeong, L. Kim, Y. Kwon and S. Kim, Korean J. Chem. Eng., 2014, 31, 2081–2087.
7. W. Li, J. Liu and C. Yan, Carbon, 2013, 55, 313–320.
8. D. Chen, M. A. Hickner, E. Agar and E. C. Kumbur, ACS Appl. Mater. Interfaces, 2013, 5, 7559–7566.
9. Z. Li, W. Dai, L. Yu, L. Liu, J. Xi, X. Qiu and L. Chen, ACS Appl. Mater. Interfaces, 2014, 6, 18885–18893.
10. G. Oriji, Y. Katayama and T. Miura, J. Power Sources, 2005, 139, 321–324.
11. C. Yao, H. Zhang, T. Liu, X. Li and Z. Liu, J. Power Sources, 2012, 218, 455–461.
12. W. Zhang, J. Xi, Z. Li, H. Zhou, L. Liu, Z. Wu and X. Qiu, Electrochim. Acta, 2013, 89, 429–435.
13. P. Han, H. Wang, Z. Liu, X. Chen, W. Ma, J. Yao, Y. Zhu and G. Cui, Carbon, 2011, 49, 693–700.
14. S. Zhong and M. Skyllas-Kazacos, J. Power Sources, 1992, 39, 1–9.
15. M. Gattrell, J. Park, B. MacDougall, J. Apte, S. McCarthy and C. W. Wu, J. Electrochem. Soc., 2004, 151, A123–A130.
16. J. Xi, W. Zhang, Z. Li, H. Zhou, L. Liu, Z. Wu and X. Qiu, Int. J. Electrochem. Sci., 2013, 8, 4700–4711.
17. T. M. Tseng, R. H. Huang, C. Y. Huang, K. L. Hsueh and F. S. Shieu, J. Electrochem. Soc., 2013, 160(4), A690–A696.
18. W. H. Wang and X. D. Wang, Electrochim. Acta, 2007, 52, 6755–6762.
19. H. M. Tsai, S. L. Yang, C. C. M. Ma and X. Xie, Electrochim. Acta, 2012, 77, 232–236.
20. K. J. Kim, M. S. Park, J. H. Kim, U. Hwang, N. J. Lee, G. Jeong and Y. J. Kim, Chem. Commun., 2012, 48, 5455–5457.
21. B. Li, M. Gu, Z. Nie, X. Wei, C. Wang, V. Sprenkle and W. Wang, Nano Lett., 2014, 14, 158–165.
22. B. Sun and M. Skyllas-Kazacos, Electrochim. Acta, 1992, 52, 1253–1260.
23. Z. Wang, J. S. Chen, T. Zhu, S. Madhavi and X. W. Lou, Chem. Commun., 2010, 46, 6906–6908.
24. L. Yang, L. Liu, Y. Zhu, X. Wang and Y. Wu, J. Mater. Chem., 2012, 22, 13148–13152.
25. X. Zhang, X. Song, S. Gao, Y. Xu, X. Cheng, H. Zhao and L. Huo, J. Mater. Chem. A, 2013, 1, 6858–6864.
26. S. S. Kim, T. R. Pauly and T. J. Pinnavaia, Chem. Commun., 2000, 17, 1661–1662.
27. J. Lee, K. Sohn and T. Hyeon, Chem. Commun., 2002, 22, 2674–2675.
28. K. Hyun, J. H. Lee, C. W. Yoon and Y. Kwon, Int. J. Electrochem. Sci., 2013, 8, 11752–11767.
29. K. Hyun, J. H. Lee, C. W. Yoon, Y. H. Cho, L. H. Kim and Y. Kwon, Synth. Met., 2014, 190, 48–55.
30. E. Kang, Y. S. Jung, A. S. Cavanagh, G. H. Kim, S. M. George, A. C. Dillon, J. K. Kim and J. Lee, Adv. Funct. Mater., 2011, 21, 2430–2438.
31. C. Jo, S. An, Y. Kim, J. Shim, S. Yoon and J. Lee, Phys. Chem. Chem. Phys., 2012, 14, 5695–5704.
32. K. H. Seng, G. D. Du, L. Li, Z. X. Chen, H. K. Liu and Z. P. Guo, J. Mater. Chem., 2012, 22, 16072–16077.
33. S. D. Gardner, C. S. K. Singamsetty, G. L. Booth, G. R. He and C. U. Pittman, Carbon, 1995, 33, 587–595.
34. J. Baltrusaitis, B. Mendoza-Sanchez, V. Fernandez, R. Veenstra, N. Dukstiene, A. Roberts and N. Fairley, Appl. Surf. Sci., 2015, 326, 151–161.
35. J. G. Choi and L. T. Thompson, Appl. Surf. Sci., 1996, 93, 143–149.
36. C. Jo, J. Hwang, H. Song, A. H. Dao, Y. T. Kim, S. H. Lee, S. W. Hong, S. Yoon and J. Lee, Adv. Funct. Mater., 2013, 23, 3747–3754.
37. Y. Zhou, C. W. Lee and S. Yoon, Electrochem. Solid-State Lett., 2011, 14, A157–A160.
38. D. Chen, M. A. Hickner, E. Agar and E. C. Kumbur, J. Membr. Sci., 2013, 437, 108–113.
39. N. A. Chernova, M. Roppolo, A. C. Dillon and M. S. Whittingham, J. Mater. Chem., 2009, 19, 2526–2552.
40. L. Zhou, H. B. Wu, Z. Wang and X. W. Lou, ACS Appl. Mater. Interfaces, 2011, 3, 4853–4857.
41. J. S. Lee, T. Lee, H. K. Song, J. Cho and B. S. Kim, Energy Environ. Sci., 2011, 4, 4148–4154.
42. Y. Shi, B. Guo, S. A. Corr, Q. Shi, Y. S. Hu, K. R. Heier, L. Chen, R. Seshadri and G. D. Stucky, Nano Lett., 2009, 9(12), 4215–4220.
43. L. J. Fu, T. Zhang, Q. Cao, H. P. Zhang and Y. P. Wu, Electrochem. Commun., 2007, 9, 2140–2144.
44. Y. Ye, C. Jo, I. Jeong and J. Lee, Nanoscale, 2013, 5, 4584–4605.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra24626a

‡ HTT Pham and C Jo contributed equally to this work and should be co-first authors.

---