Among the different types of fuel cells available, direct methanol fuel cells (DMFCs) are promising power sources for various applications, such as vehicles, cell phones, and laptops, because of their relatively low operating temperature and high power density ^[1-4]. However, a lack of highly efficient and inexpensive anodic catalysts remains an encumbrance to commercial application of DMFCs. We show herein that tungsten carbide (WC) sandwiched in reduced graphene oxide (RGO) layers improves the mass and charge transfer ability, and thus, application to catalysis.

Pt and Pt alloy supported on carbon are excellent electrocatalysts for methanol oxidation reaction (MOR) ^[5], but the high cost, poor utilization coefficient ^[4], and ease of CO poisoning of Pt seriously narrows their practical application in DMFCs. WC as a non-noble catalyst is confirmed to have "platinum-like" properties because of its similar structure with Pt ^[6]. To reduce the amount of Pt required in DMFCs, WC is considered to be a promising co-catalyst of Pt metal because of its strong interaction with Pt ^[7]. Surface science and electrochemical studies indicate that, in the presence of Pt, WC is active in MOR and water decomposition, which is crucial for removing CO from the Pt surface ^{[8, 9]}. However, the low conductivity and small surface area of WC limits its catalytic activity in MOR.

To improve the conductivity and synergistic effect between WC and Pt, a support with higher electroconductivity and greater surface area, and significant interfaces are required. RGO is selected as an ideal support for loading WC to promote the conductivity and surface area of the catalyst. In general, the formation of a graphene intercalation compound can not only inhibit the stacking of graphene to obtain few-layer structures, but also increase the interlayer spacing to improve the mass and charge transfer ^[10], which greatly enhances the electrocatalytic activity. The intercalated layered materials used in this work were designed and studied by our group previously ^[11]. However, WC particles with a diameter of 20 nm were inclined to load onto the edges of the RGO, which lowered the electrocatalytic activity towards MOR.

Intercalation compounds of graphene sandwiched with transition metal carbides (TMCs) are not easily obtained because the TMCs are generally prepared by a direct reduction carbonization method that inevitably leads to large sized molecules and poor dispersion of the catalysts ^[12]. However, in the case of WC, direct attachment of metatungstate ions, [H₂W₁₂O₄₀]^6-, onto graphene oxide (GO) is difficult to realize because the negative charges of both [H₂W₁₂O₄₀]^6- and GO are unfavorable for assembly of the particles. Thiourea is a known, remarkably simple neutral receptor for anion recognition because of its strong ability to bind anions to form stable complexes through hydrogen bonds ^[13-16]. Therefore, we might assume that thiourea can be used as a bridge to connect two anions. Thiourea is also used to form transition metal sulfides, which have graphene-like sheet structures that can be assembled with graphene to develop intercalation compounds ^[17]. Subsequently, a WC-RGO intercalation compound can be obtained by in situ reduction carbonization method from tungsten disulfide (WS₂) sandwiched in RGO layers. To the best of our knowledge, the use of thiourea as an anchoring and inducing reagent for the preparation of a carbide-graphene intercalation compound has not been reported to date.

In this work, ultra-small-sized tungsten carbide sandwiched in RGO layers, forming a WC-RGO intercalation compound (WC-RGO), was prepared by a facile and efficient method by adding thiourea as an anchoring and inducing reagent. We propose that the excellent performance of Pt/WC-RGO is related to the ultra small size and high dispersion of WC on the few-layer RGO. The intercalated structure, which increases the probability of WC coming into contact with Pt, promotes synergistic effects and improves the mass and charge transfer ability.

GO was prepared from flake graphite using a modified Hummers method ^[18]. Specific steps are as follows. Flake graphite (3.0 g), KMnO₄ (18.0 g), and concentrated phosphoric acid (40 mL) were put into a 1000-mL flask under magnetic stirring in an ice bath for 2 h while concentrated H₂SO₄ (360 mL) was added to the mixture slowly. Then, the mixture was stirred in a water bath at 50 ℃ for 12 h followed by dropwise addition of 10 mL H₂O₂. Finally, the mixture was centrifuged and washed with HCl aqueous solution and deionized water several times. The resulting solid was dispersed in 600 mL deionized water and sonicated for 2 h to form GO aqueous dispersion. GO was derived by freeze-drying the GO aqueous dispersion. For preparation of Pt/WC-RGO, ammonium metatungstate (AMT) was added after thiourea and the GO dispersion was stirred for some time. Then, the mixture was stirred for another 12 h and freeze-dried to obtain the precursor that was then put into a ceramic boat and placed into a tubal furnace. The furnace temperature was preserved at 700 ℃ for 1 h in Ar (100 mL/min) and 900 ℃ for 3 h in CO (100 mL/min) and H₂ (20 mL/min). The obtained black powder was denoted as WC-RGO. 10 wt% Pt/WC-RGO was obtained by the method published in our previous work ^[11]. Pt/C catalyst (20% Pt on Vulcan XC-72R, Johnson Matthey Corp.) was used as a comparative sample.

The crystallite structures were investigated by X-ray diffraction (XRD) with an X'Pert PRO X-ray system (PANalytical, Netherlands) at room temperature. Cu K_α radiation (40 kV, 40 mA) was applied and an angle range from 10° to 80° was recorded at 0.05 increments. The morphologies of the samples were observed by scanning electron microscope (SEM) (Hitachi S-4700II, Hitachi, Japan) using Cu K_α radiation (λ = 0.154 nm). Transmission electron microscopy (TEM) and scanning transmission electron microscopy (STEM) were performed with a Tecnai G2 F30S-Twin microscope (FEI, Netherlands), coupled with energy dispersive X-ray spectrometry (EDX, Thermo NORAN VANSTAGE ESI), using Cu K_α radiation. The element distribution of the catalyst was characterized by EDX. Raman spectroscopy was performed using a Lab RAM HR UV800 Raman microscope (JOBIN YVON, France) with a laser emission wavelength of 632.81 nm from a frequency range of 2000 to 600 cm^-1.

Electrochemical properties were measured on a CHI660D (Chen-hua, Shanghai, China) using a three-electrode electrochemical cell. Platinum electrodes and saturated calomel electrodes (SCEs) were used as the counter and reference electrodes, respectively. The working electrode was prepared as in our previous work ^[11]. The cyclic voltammetry (CV) and chronoamperometry (CA) was tested in 0.5 mol/L H₂SO₄ + 0.5 mol/L CH₃OH, at a scan rate of 50 mV/s at 50 ℃. For CO stripping measurements, 0.5 mol/L H₂SO₄ solution was bubbled with N₂ for 0.5 h, then the CO absorption was performed at a constant potential of -0.14 V vs. SCE. The excess CO in the electrolyte was removed by bubbling N₂ vigorously. Finally, the stripping voltammograms were recorded from -0.1 to 1 V at a scan rate of 20 mV/s at room temperature. Accelerated durability testing (ADT) was carried out by cycling the electrode potential, under nitrogen saturation, for 100 cycles between 0.6 and 1.0 V (vs. SCE) in 0.5 mol/L H₂SO₄ at a scan rate of 50 mV/s at 50 ℃.

Fig. 1(a) shows the XRD patterns of WC-RGO and Pt/WC-RGO. It can be seen that both samples have strong C (002) diffraction peaks at 25.6°, characteristic of parallel RGO sheets. The peaks are 0.8° less than that of graphite (26.4°) ^[11], which may be the result of the introduction of WS₂ to form WS₂-RGO intercalation compound, and final product WC that prevents the aggregation of RGO while increasing the interlayer spacing. In the XRD patterns of Pt/WC-RGO, the distinct diffraction peaks located at 39.7°, 46.1°, and 67.5° are indexed to (111), (200), and (220) planes, respectively, of Pt (JCPDS 01-089-7382). For WC-RGO, the peaks located at 31.3°, 35.5°, 48.0°, 64.1°, 73.0°, and 75.9° can be identified as the (001), (100), (101), (110), (111), and (102) reflections, respectively, of hexagonal WC phase (JCPDS 00-025-1047). However, the peaks belonging to WC weaken for Pt/WC-RGO, which may suggest that WC is too small in size and is uniformly dispersed on the RGO layers. Similar results are also observed for Pt-WC/carbon and Pt-Mo₂C/carbon ^{[19, 20]}. The observation can be explained by the strong interaction of Pt with WC NPs and epitaxial growth of Pt on WC that results from the similar electronic structures of these species ^[21]. No other diffraction peaks are detected, implying a high purity of the WC phase.

Raman spectroscopy was used to study the evolution of the crystal structure and the nature of the chemical bonds in the samples during assembly from graphite and GO to WC-RGO. Fig. 1(b) shows the Raman results of graphite, GO, and WC-RGO. All three samples have a characteristic D band and G band at about 1333.2 and 1578.2 cm^-1, respectively. However, the G band of WC-RGO intercalation compound is shifted marginally to a higher wavenumber, which is probably caused by the WC that is sandwiched in the RGO layers forming an intercalated structure. An increased D/G intensity ratio (I_D/I_G) of graphite and GO relative to WC-RGO is also observed. I_D/I_G is used to quantify defects and disorder in graphene-related systems, which include a large family of sp² carbon structures ^[22]. Samples with a higher I_D/I_G value tend to have more defects and disorder in their structure. Compared with graphite, GO has a higher I_D/I_G value as a result of the oxygen-containing group introduced by the oxidation reaction. For WC-RGO, the increased I_D/I_G value confirms the success of loading WC between the layers of RGO.

From the SEM image shown in Fig. 2(a), thin and transparent RGO layers can be clearly observed at low magnification, suggesting few layers are evident. The RGO sheets also have the characteristic shape of a crumpled silk veil, which is reported to be part of the intrinsic nature of graphene sheets ^[23]. From the high magnification SEM image of WC-RGO (Fig. 2(b)), crumpled RGO sheets with an extraordinarily large interlayer spacing are presented, which is consistent with the XRD results. The formation of few-layer RGO sheets with a large interlayer spacing should be related to growth of WS₂ developing an intercalation compound with RGO, which acts as a space for separating RGO and prevents stacking like a sandwich. However, the WC particles cannot be found on the RGO sheets from SEM directly, probably because of the small size and good dispersion that are beyond the resolution limit of our SEM instrument.

TEM and high-resolution TEM (HRTEM) images of WC-RGO and Pt/WC-RGO are provided to give more detailed information. From Fig. 3(a), WC NPs of uniform size and with homogeneous distribution on the RGO sheet can be clearly observed. The inset image of Fig. 3(a) shows the size distribution of WC NPs, which confirms that WC is homogeneously dispersed on the support without agglomeration, with a mean particle diameter of 1.5 nm (one of the smallest WC sizes reported in the literature) ^[24-29]. The smaller size and better dispersion of WC on the RGO sheets significantly increases the probability of contact with Pt NPs, which can eventually enhance the synergistic effects between Pt and WC for MOR. The HRTEM image in Fig. 3(b) shows that the distances of two adjacent planes of a single WC and Pt particle are about 0.188 and 0.229 nm, corresponding to the (101) and (111) crystal faces of the hexagonal WC and Pt, respectively. The intimate contact between Pt and WC is also confirmed, and this can improve the electro-oxidation activity towards MOR by promoting the synergistic effects. From the inset image of Fig. 3(b), the carbon sheets that are about 5 layers thick confirm the successful synthesis of few-layer RGO. The distribution of the Pt, WC, and RGO phase is displayed by EDX and the corresponding result is shown in Fig. 3(c). The top right corner is the mapped images of the elements Pt, W, and C, respectively. The characterization of W and C shows that most of the selected area is filled with WC and RGO. The reason for this may be that WC NPs are small enough and sufficiently dispersed that they cover the RGO sheet uniformly. The characterization of Pt indicates that Pt particles also disperse well on the support. The small-sized and highly dispersed WC is consistent with the results from SEM.

The electrochemical surface area (ECSA) not only determines the number of catalytically active sites available for an electrochemical reaction, but is also a significant parameter to compare different electrocatalytic supports by accounting for the conductive path available for electron transfer ^[30]. The CV of Pt/WC-RGO and commercial Pt/C tested at a scan rate of 50 mV/s in 0.5 mol/L H₂SO₄ solution at 50 ℃ is shown in Fig. 4(a). In the range between -0.2 and 0.1 V (vs. SCE), typical hydrogen adsorption and desorption peaks are observed, from which the ECSA of the as-prepared samples can be estimated by employing the equation ^[30]:

The result indicates that the ECSA value of Pt/WC-RGO is promoted and is 3.66 times greater than that of commercial Pt/C (Table 1), suggesting better performance for MOR. This extraordinary performance can be explained as the result of the intimate contact between Pt and WC, anchoring on the few-layer RGO support with the intercalated structure.

The CV, after stabilization, of methanol electro-oxidation on Pt/WC-RGO and commercial Pt/C tested at a scan rate of 50 mV/s in 0.5 mol/L H₂SO₄ + 0.5 mol/L CH₃OH solution at 50 ℃ is presented in Fig. 4(b). The two CV curves show similar methanol oxidation current peaks in the positive and negative potentials. However, both the forward anodic peak current (I_f) and the backward anodic peak current (I_b) on Pt/WC-RGO are much higher than those on commercial Pt/C and other electrocatalysts reported in the literature (Table 1). The typical forward and backward anodic peak at around 0.7 and 0.4 V is attributed to the electrooxidation of methanol and a Faradaic oxidation reaction on the catalysts by residual intermediate species such as CH₂OH, CH₂O, HCOOH, and CO ^[32]. The value of I_f/I_b can be used to infer the tolerance of a catalyst to the intermediate carbonaceous species formed via decomposition of methanol and accumulated on the electrode surface ^[33]. Therefore, a higher value of I_f/I_b is indicative of, e.g., enhanced CO-tolerance. It is notable that the Pt/WC-RGO catalyst gives a much higher value of I_f/I_b, which means that the Pt-WC/RGO may be much more tolerant to the intermediate carbon species and, consequently, can much more efficiently facilitate the oxidization of methanol to carbon dioxide. Additionally, the onset potential (E_o) for MOR on Pt/WC-RGO starts at 0.242 V, which is 57 mV more negative compared with that on commercial Pt/C. These results indicate that the methanol can be easily oxidized and the oxidative removal of the intermediates generated from MOR can also occur more easily on Pt/WC-RGO compared with commercial Pt/C. The higher catalytic activity demonstrated by Pt/WC-RGO can be attributed to the increased interfaces provided by the intercalated structure of WC-RGO, where CO can be easily removed from Pt sites as a result of the synergistic effect between WC and Pt ^{[34, 35]}.

The excellent CO tolerance of Pt/WC-RGO is further demonstrated by CO stripping tests at a scan rate of 20 mV/s in 0.5 mol/L H₂SO₄ solution at room temperature (Fig. 4(c)). The oxidation of CO requires H₂O dissociation or activation ^[36]. The WC nanocrystallites are active in water decomposition and can provide the OH_ad species to oxidize the adsorbed CO on Pt surfaces. Furthermore, the enhanced contact between Pt and WC profiting from the intercalated nanostructure could favor a decrease in the mass and charge transfer limits, which, consequently, brings the synergistic effect into play ^[10]. The Pt/WC-RGO has a peak potential of 0.540 V, which is 104 mV more negative than that of commercial Pt/C. This means that CO can be more easily oxidized at a lower potential on the surface of Pt/WC-RGO compared with commercial Pt/C. The easier removal of CO on the Pt/WC-RGO directly leads to a refreshment of Pt sites that subsequently results in the slow degradation of the catalyst. It is notable that the removal of intermediate species benefited from the intercalated structure with well dispersed WC and Pt on it because this layer-by-layer construction is a functional system to assemble the support RGO, promoter WC, and active component Pt without aggregation. This result is in agreement with the good mass and charge transfer ability and excellent CO poisoning-tolerance of WC-RGO intercalation compound.

The catalytic stability of a catalyst is a key factor in determining its applicability on a commercial scale. To estimate the stability of the catalyst, chronoamperometry (CA) was performed in 0.5 mol/L H₂SO₄ + 0.5 mol/L CH₃OH solution (Fig. 4(d)). As is shown, the performances of both catalysts show that the polarization current initially decreases rapidly because of the formation of intermediates during the methanol oxidation; however, Pt/WC-RGO decreases relatively slowly. After 6000 s, the current value of Pt/WC-RGO is much higher than that of commercial Pt/C, which indicates that the Pt/WC-RGO catalyst has much better long-term stability compared with commercial Pt/C. The enhanced stability may be attributed to the improved synergistic effect between Pt and WC from the intercalated structure that prevents the small Pt particles from being poisoned by CO. WC-RGO intercalation compound provides not only intimate contact between Pt and WC to make full use of the synergistic effect, but also few-layer RGO to increase the mass and charge transfer ability. Thus, a smaller amount of Pt is needed to generate a higher current and a substantial amount of noble metal can be saved.

To further investigate catalyst durability, accelerated durability testing (ADT) is commonly employed to simulate the harsh, potentiodynamic, and corrosive conditions encountered at the cathode of proton exchange membrane fuel cells (PEMFCs) during operation ^[37]. For the three-electrode system, 0.6 and 1.0 V (vs. SCE) were chosen as the potential for the degradation of the catalysts (Fig. 4(c)). The Pt/WC-RGO and commercial Pt/C catalysts were subjected to 100 cycles using multi-potential steps between 0.6 and 1.0 V in nitrogen-saturated 0.5 mol/L H₂SO₄ with CVs collected after every 10 cycles of ADT. The stabilities of commercial Pt/C and Pt/WC-RGO were compared after 100 cycles of ADT, as shown in Fig. 5(a) and 5(b), respectively. Compared with commercial Pt/C, Pt/WC-RGO clearly provides a higher double layer current density, attributed to the special intercalated structure of our catalyst. Furthermore, the platinum oxide reduction peak for commercial Pt/C almost disappears after 100 cycles, which implies an increase in particle size as a result of aging ^[38] and surface restructuring of Pt particles, resulting in a passivated surface and a reduction in the number of highly exposed Pt atoms ^[29]. However, the platinum oxide reduction peak for Pt/WC-RGO remains relatively unchanged, which indicates that the Pt/WC-RGO maintains its original structure upon reaction. ECSA values were confirmed based on the calculated charge for hydrogen adsorption-desorption initially and after each 10 subsequent cycles. As is shown in Fig. 5(c), the ECSA value of Pt/WC-RGO catalyst decreased relatively fast for the first 10 cycles and then degraded slowly until the 50th cycle before maintaining a value of 153.5 m²/g Pt after 100 cycles. However, in the case of commercial Pt/C, the ECSA value decreased rapidly throughout the ADT process and dropped to 9.2 m²/g Pt after 100 cycles. This result indicates that the Pt/WC-RGO catalyst, to some extent, has the ability to resist the degradation caused by the harsh conditions. From Fig. 5(d), it can be clearly seen that the ECSA retention for Pt/WC-RGO is superior to that for commercial Pt/C, retaining 62.3% and 13.7% of their initial ECSA after 100 cycles, respectively. The excellent long-term stability of the former is primarily attributed to the intercalated structure that not only prevents the intrinsic structure from being poisoned, but also enhances the mass and charge transfer ability for MOR. From the electrochemistry measurements it is observed that Pt/WC-RGO catalyst demonstrates a higher electrochemical activity, which is related to the special intercalated structure of Pt/WC-RGO. In the well-organized catalysts studied, WC NPs are highly dispersed on the thiourea-modified graphene oxide, and are assembled to obtain the WC-RGO intercalated structure using thiourea as an anchoring reagent. Compared with simple methods, the means of assembly through use of an anchoring reagent in this work is inclined to acquire an intercalated structure with a larger ECSA value and intimate contact of WC and Pt. The intercalated structure facilitates the dispersal and anchoring of both Pt and WC, which is beneficial for the synergistic effect between Pt and WC. The promoted synergistic effect accelerates the removal of CO on the surface of Pt and decomposition of the intermediate carbonaceous species formed during the process of methanol electro-oxidation. As a result, the catalyst we have prepared has improved tolerance of CO and long-term stability during the catalytic process.

WC-RGO intercalation compound has been constructed by anchoring WC NPs (about 1.5 nm in diameter) on few-layer RGO sheets based on a controllable assembly. The use of thiourea as the anchoring and inducing reagent to bridge the [H₂W₁₂O₄₀]^6- and GO has not been reported before. The intercalation compound has a characteristic structure with ultra-small-sized WC NPs sandwiched in few-layer RGO sheets. This improves the conductivity for fast mass and charge transfer. That gives the intercalation compound great potential as a support of Pt for electrocatalytic application. This structure also provides intimate contact between WC and Pt after loading low amounts of Pt. Compared with commercial Pt/C, the Pt/WC-RGO catalyst possesses higher electrochemical activity. Possible applications may include many areas, such as hydrogenation, dehydrogenation, and hydrogenolysis as WC-RGO intercalation compound has more active sites and a larger specific surface area than does commercial Pt/C. The WC-RGO and other TMC intercalation compounds can be continually exploited and beneficial outcomes may be achieved for other chemical purposes.