Three-dimensional TiO₂@C nano-network with high porosity as a highly efficient Pt-based catalyst support for methanol electrooxidation

Abstract

The development of highly active and durable Pt-based catalysts is an important issue for methanol electrooxidation. Research into highly efficient supports provides a feasible method to achieve such catalysts. In this paper, carbon-coated TiO₂ nanowires with a unique three-dimensional network structure are prepared as a Pt-based catalyst support by the carbonization of a resorcinol–formaldehyde polymer. Physical characterization confirms that this unique structure can provide a large specific surface area, high porosity and efficient transport channels. The abundant heterogeneous interfaces between the TiO₂ nanowires and carbon provide numerous Pt loading sites, which greatly improves the utilization of Pt. Strikingly, electrochemical measurements show that the as-prepared catalyst has much better activity and durability than commercial Pt/C for methanol electrooxidation. Furthermore, the single direct methanol fuel cell test demonstrates that the as-prepared catalyst has higher power density and polarization current. We attribute this excellent catalytic performance to the unique nano-network structure, the numerous Pt anchoring sites, and the synergetic effect of the different components.

---

1. Introduction

Although direct methanol fuel cells (DMFC) have great potential to serve as power sources for 3C electronic devices, the promise of their practical application is primarily hindered by their high content of the noble metal Pt.^1–3 In recent years, non-noble metal catalysts have been significantly developed for cathodes; thus, the cost is greatly reduced.^4–8 However, it is still a fact that catalysts for anodes cannot function without Pt.^9,10 Among various Pt-based catalysts, the commercial Pt/C catalyst is most commonly used for methanol oxidation.¹¹ However, due to carbon corrosion and the poisoning of intermediates during the methanol oxidation process,¹² the activity and durability of Pt/C drastically decreases.^3,13 Therefore, the development of effective strategies is highly desirable to improve the durability and activity of Pt-based catalysts for methanol oxidation.^3,14–17

One effective approach is to substitute carbon with metal oxide.¹⁸ Metal oxide supports can not only resist the corrosion in acidic and oxidative conditions but can also produce a strong metal-support interaction with platinum.¹⁹ Accordingly, significant studies have been performed to control the synthesis of metal oxide supports, such as TiO₂, SnO₂, and CeO₂.^20–22 Among various reported metal oxide supports, titania has attracted much attention because of its low price and excellent corrosion resistance.²³ However, compared to carbon materials, metal oxides have almost no availability for Pt deposition due to their high crystallinity and small specific surface area.²⁴ Therefore, numerous efforts have been triggered in the development of controlled special morphologies to obtain large surface areas.^25–27 Titania nanotubes obtained from the hydrothermal method have been proven to have large a specific surface area.²⁸ However, it should be pointed out that titania nanotubes undergo crystal reunion very readily, particularly after calcination, which greatly weakens their advantages.¹¹ In addition, the poor electric conductivity of titania also restricts its applications. Therefore, the synthesis of titania with a large surface area and high electric conductivity is still an important scientific problem.

Considering the special structure of a 3D-network composed of nanowires, such as high porosity, large surface area and efficient transport channels, such a network would be expected to be a highly efficient support for Pt-based catalysts. For this reason, the synthesis of 3D-network structured titania should be conducive to enhance electro-catalytic performance. In this work, we successfully constructed a 3D-network structure with titania nanowires coated with carbon (TCN) derived from resorcinol–formaldehyde polymer (RF) carbonization. The presence of carbon not only suppresses the aggregation of titania but stabilizes the network structure, facilitating the final morphology of the 3D-network structure. Moreover, carbon encapsulation greatly improves the electronic conductivity of titania. Strikingly, the obtained Pt-based catalyst exhibits far better electro-catalytic performance than the commercial Pt/C catalyst (20 wt% Pt) for methanol electrooxidation. More importantly, the single DMFC test shows that the obtained Pt-based catalyst has higher power density and polarization current than the Pt/C catalyst. This work demonstrates that this 3D-network structured Pt/TCN catalyst is indeed a promising electro-catalyst with improved electro-catalytic performance for future applications.

2. Experimental

2.1 Synthesis of TCN support and Pt/TCN electro-catalysts

Na₂Ti₂O₅ nanotubes were firstly synthesized by a typical hydrothermal method. 0.5 g anatase titania powders were dispersed into 10 mol L⁻¹ NaOH aqueous solution and then reacted at 130 °C for 24 h under hydrothermal conditions. The products were washed and the pH was adjusted to ≈7.0 with deionized water and dilute HNO₃. 3D-network structured TCN were obtained through a hydrothermal process and subsequent high-temperature calcination. Firstly, RF was synthesized in situ on the surface of Na₂Ti₂O₅ nanotubes by the hydrothermal method. A certain amount of resorcinol and hexamethylenetetramine (molar ratio = 3 : 1) was dissolved into 150 mL of the above Na₂Ti₂O₅ suspension, which was then heated at 100 °C for 6 h under hydrothermal conditions. Afterwards, the obtained suspension was adjusted to pH ≈ 2.5 with dilute HNO₃ for 12 h and then was filtered and dried. Lastly, the samples were calcinated at 800 °C under N₂ for 2 h. The products prepared with different amounts of resorcinol (0.6 g, 0.8 g, 1.0 g and 1.2 g) were named TCN-5v6, TCN-5v8, TCN-5v10 and TCN-5v12, respectively. The Pt/TCN electro-catalysts were obtained by the microwave-assisted ethylene glycol method. The detailed procedure of Pt deposition is provided in the ESI.†

2.2 Preparation of membrane electrode assembly (MEA)

To conduct the single fuel cell tests, the MEA was firstly prepared using 40 wt% Pt/TCN-5v10 as the anode catalyst and the as-prepared 40 wt% Pt/C as the cathode catalyst. The detailed procedure of MEA preparation is provided in the ESI.†

2.3 Physical characterizations

The structures and morphologies of the samples were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), inductively coupled plasma-atomic emission spectrometry (ICP-AES), energy-dispersive X-ray spectroscopy (EDX), nitrogen adsorption–desorption isotherms (BET), Raman spectroscopy (Renishaw inVia, Ar ion laser), transmission electron microscopy (TEM), high-angle annular dark-field scanning TEM (HAADF-STEM), high-resolution TEM (HRTEM) and X-ray photoelectron spectroscopy (XPS). The conductivity of the as-prepared support material was measured by the four-probe method.

2.4 Electrochemical measurements

Electrochemical experiments were conducted in a typical cell with three electrodes at 25 °C using a CHI 650E instrument. The experimental methods included cyclic voltammograms (CV), electrochemical impendence spectroscopy (EIS), amperometric i–t curves and CO-stripping measurements. The single fuel cells were tested using the Fuel Cell Testing System. The detailed methods are provided in the ESI.†

3. Results and discussion

3.1 Physical characterization and discussion

The schematic of the preparation process of the 3D-network structured Pt/TCN catalyst is shown in Scheme 1. Firstly, Na₂Ti₂O₅ nanotubes were obtained by the typical hydrothermal reaction. The as-prepared Na₂Ti₂O₅ nanotubes were dispersed into the resorcinol and hexamethylenetetramine solution. Through the second hydrothermal reaction, the resorcinol–formaldehyde polymer (RF) was synthesized on the surface and in the gaps of the cross-linked Na₂Ti₂O₅ nanotubes. Na₂Ti₂O₅ was then converted into H₂Ti₂O₅ by adjusting the pH to about 2.5. After heat treatment, the internal structure of the H₂Ti₂O₅ nanotubes collapsed to form TiO₂ nanowires. At the same time, the RF was carbonized into porous carbon, which was uniformly distributed among the TiO₂ nanowires, forming the 3D-network structure. Lastly, Pt nanoparticles were deposited on the 3D-network structured support using a microwave-assisted ethylene glycol method.

Scheme 1 Schematic of the 3D-network structured Pt/TCN catalyst preparation process.

The morphologies of the obtained products were determined by electron microscopy. Fig. 1 shows the SEM and TEM images of the obtained samples. The SEM image (Fig. 1a) shows that the product obtained from the first hydrothermal reaction presents a cross-linked network structure consisting of uniform nanotubes. The nanotubes are highly uniform and very flexible, and the cross-linked network obviously exhibits a folding phenomenon, forming a multilayer network structure. The morphology of the nanotubes was confirmed by TEM (the inset of Fig. 1a). The product after carbon coating can be seen in the high resolution SEM image (Fig. 1b). It is obvious that the 3D-network structure was successfully synthesized with high porosity. The framework of the 3D-network is composed of interleaved nanowires. In addition, single nanowires can be clearly seen, indicating that the good dispersion of nanowires is retained after calcination. Among the nanowires, some loose structures can be observed, which is attributed to the carbon derived from the RF carbonization. The network structure is clearly presented in the TEM images (Fig. 1c and d). The interleaved nanowires are obviously enveloped by a layer of tulle-like carbon. Importantly, these nanowires are well separated from each other, with diameters of about 5 to 10 nm. The elemental distribution of the network structure was analysed by HAADF-STEM images (Fig. 1e), which show that the Pt, Ti, C and O elements are all homogeneously distributed throughout the whole network. The homogeneous distribution of carbon element demonstrates the uniform coating of carbon. Additionally, it can be distinctly observed that the Pt nanoparticles are uniformly deposited on the whole network structure, with a mean size of only 1.7 nm (inset of Fig. 1d).

Fig. 1 Morphology analysis for the as-prepared products: (a) SEM image of Na₂Ti₂O₅ nanotubes (the inset of (a) shows the TEM image of the Na₂Ti₂O₅ nanotubes); (b) SEM image of TCN-5v10; (c) low-magnification TEM image of Pt/TCN-5v10; (d) high-magnification TEM image of Pt/TCN-5v10 (the inset of (d) shows the mean size of the Pt nanoparticles); (e) elemental mapping of the as-prepared Pt/TCN-5v10 catalyst from HAADF-STEM.

Numerous tools were used to analyse the structure and components of the catalysts, as shown in Fig. 2. The HRTEM images (Fig. 2a and b) of the catalysts show obvious lattice fringes, indicating that the catalysts have good crystallization. Importantly, it is expected that some amorphous phase attributed to carbon can be discerned, which exists not only on the surface of nanowires but also between the nanowires. A multilayer stack structure can be observed at the same time. As shown in Fig. 2b, the crystal plane distances confirm the presence of the TiO₂ (101) plane (0.35 nm) and the Pt (111) plane (0.23 nm). In addition, it can be clearly seen that the TiO₂ nanowires are encircled by a carbon layer with a thickness of about 1.5 nm. The crystal structures of the products were further analysed by XRD, as shown in Fig. 2c. The XRD pattern of the support (TCN-5v10) mainly displays the anatase TiO₂ crystal structure, accompanied by a small amount of rutile TiO₂. As is known, the temperature of the TiO₂ phase transition from anatase to rutile is about 600 °C.²⁹ Nevertheless, in our work, the transition temperature seems to be greatly increased. Obviously, it is very likely that the higher thermal stability originates from the inhibition of the carbon layer.³⁰ In the XRD of the catalyst (Pt/TCN-5v10), the Pt peaks are submerged by the strong TiO₂ peaks. However, due to the overlapping of the Pt (111) plane diffraction peak at 2θ = 37.8°, the difference between the catalyst and the support can still be seen, which confirms the presence of Pt. More direct evidence is obtained from the HRTEM image (Fig. 2b) and EDX spectrum (Fig. 2d). In addition, carbon can be observed by EDX, although it cannot be seen in the XRD pattern. The content of each element can be roughly estimated by EDX. The Pt loading (21.3 wt%) is almost equal to the theoretical value of 20 wt%. In order to obtain the exact content of Pt in the catalysts, ICP-AES analysis was performed; the content of Pt was found to be 20.6 wt%.

Fig. 2 (a) and (b) HRTEM images of Pt/TCN-5v10; corresponding (c) XRD patterns and (d) EDX spectra.

Control experiments using different amounts of carbon were carried out. The changes in morphology were characterized in detail by TEM. When RF was absent in the precursors (no carbon), the network structure could not be obtained and significant reunion of the nanowires occurred (Fig. S1†), demonstrating that the presence of carbon is important to generate the morphology of the 3D-network structure. To investigate the influence of the carbon content, the products were prepared by simply controlling the mass ratios of titania and resorcinol, respectively, indexed as TCN-5v6, TCN-5v8, TCN-5v10 and TCN-5v12. The carbon contents of these products were roughly estimated by EDX (Fig. S2†) as 18.5%, 22.5%, 25.7% and 30.4% with increasing resorcinol. The increase of carbon not only improves the electronic conductivity but enhances the effect of inhibition of the titania phase transition. This explains the phenomenon where the rutile phase gradually weakens with increasing resorcinol (Fig. 3d). In addition, the XRD pattern of Pt/TCN exhibits the same phenomenon as the XRD pattern of TCN, as shown in Fig. S3.† Fig. 3a–c and 1d show TEM images of these products. It can be seen that, although the network structure is obtained in all the samples, their dispersions and porosities are quite different. According to the TEM images (Fig. 1d and 3a–c), the dispersion of titania nanowires becomes more and more uniform with increasing carbon content. Uniform dispersion of titania nanowires is an efficient approach for increasing the porosity and enhancing the interaction between Pt and TiO₂. However, there must be a maximum value. When the amount of carbon increases to this value, the dispersion of titania nanowires will not change with increasing carbon. Our experiment confirms the presence of the maximum value because the dispersion of titania nanowires for TCN-5v10 and TCN-5v12 is almost the same. More importantly, excessive carbon will clog the pore structure derived from the interleaved nanowire structure, resulting in a decrease of porosity.

Fig. 3 TEM images of Pt/TCN with different mass ratios of TiO₂/RF: (a) 5v6, (b) 5v8 and (c) 5v12. (d) Corresponding XRD patterns of the TCN supports.

The changes in the specific surface areas and porosities of these samples were studied by adsorption–desorption isotherms of nitrogen on the different TCN supports, as shown in Fig. S4† and Table 1. The results indicate that all the TCN supports are mesoporous materials. It can be seen from Table 1 that with increasing carbon content, the specific surface area of the TCN supports increases while the changes of the pore volume and the average pore size result in a volcano shape. The increase of the specific surface area can be easily understood, resulting from the more uniform dispersion of TiO₂ nanowires and the increase of carbonized mesoporous carbon. The larger the specific surface area, the more easily Pt nanoparticles can be loaded. The change of porosity is interesting and is consistent with the above prediction. When the carbon content is low, the amount of carbon is insufficient to stabilize the network structure. TiO₂ nanowires aggregate, leading to a decrease of porosity. When the carbon content is too high, the excess carbon will block the pore structure. Thus, the pore volume and the average pore size will both decrease. Through a comprehensive comparison, it can be found that the TCN-5v10 support has the best structure. The dispersion of Pt nanoparticles on the network can indirectly reflect the change of structure. On the basis of statistics derived from the TEM images, the average particle sizes of Pt on TCN-5v6, 5v8, 5v10 and 5v12 are 2.3 nm, 2.1 nm, 1.7 nm and 2.5 nm, respectively (inset of Fig. 1d and 3a–c). Obviously, the average particle size of Pt on TCN-5v10 is the smallest, indicating that the network structure of TCN-5v10 can provide a larger specific surface area, better porosity and a simpler interface of carbon and TiO₂.

Table 1 Textures of the different TCN supports

Sample	SBET (m^2 g^−1)	Pore volume (cm^3 g^−1)	Average pore size (nm)
TCN-5v6	95.8	0.48	20.2
TCN-5v8	156.4	0.90	23.1
TCN-5v10	236.1	1.39	22.2
TCN-5v12	287.7	0.93	12.9

---

It is worth noting that, although the surface of TiO₂ is coated by carbon, the Pt nanoparticles can still be directly deposited on TiO₂ when the carbon content is appropriate. It is well known that carbonized carbon has an abundant pore structure. The schematic of the carbon layer on the surface of TiO₂ is shown in Scheme 2, including proper carbon content (Scheme 2a) and excess content (Scheme 2b). In fact, the carbon coating on the surface of TiO₂ is not complete because the polymerization reaction of resorcinol and formaldehyde is very rapid. Because of the rapid polymerization reaction, the carbon layer is incomplete. Therefore, when the carbon content is suitable, the exposed TiO₂ surface, as shown in Scheme 2 by the arrows, and the holes in the carbon layer, as shown in Scheme 2a, can contact the Pt nanoparticles. However, when the carbon content is too high, the carbon layer on the surface of TiO₂ is complete, and most of the holes are masked, as shown in Scheme 2b. The contact of Pt nanoparticles and TiO₂ becomes difficult. Therefore, the contact of Pt nanoparticles and TiO₂ requires suitable carbon content. This suitable carbon content can be obtained from our experiments, and is consistent with TCN-5v10.

Scheme 2 Schematic of the carbon coating on the surface of TiO₂: (a) proper carbon content and (b) too much carbon content.

XPS was adopted to analyze the chemical state information of the C and Pt elements. The curve fittings of the C 1s and Pt 4f peaks of the XPS for as-prepared Pt/TCN-5v10 and commercial Pt/C are shown in Fig. 4. It is very important to study the difference between the carbon originating from RF carbonization and normal carbon black. From the deconvolution results of the C 1s peaks for as-prepared Pt/TCN-5v10 and commercial Pt/C (Table 2), the content of sp²-C in carbonized carbon is lower, while the content of oxygen-containing species is much higher; these can function as anchoring sites for Pt deposition. The binding energies of Pt 4f with their relative contents are shown in Table 3. It is expected that the Pt(0) binding energy of as-prepared Pt/TCN-5v10 will show a shift of about 0.3 eV in the direction of lower energy compared with that of commercial Pt/C. This can be ascribed to the strong metal-support interaction (SMSI) between Pt and TiO₂. Furthermore, the Pt(0) content increases by 11.5%, while the Pt(IV) content decreases by 15.1%. According to the Pt dissolving mechanism,³¹ Pt oxidation is the required condition for Pt dissolution. Compared with Pt(IV), the corrosion of Pt(0) firstly requires oxidization to Pt(IV) at high potential. Therefore, Pt(0) has higher stability than Pt(IV) in H₂SO₄ solution. Therefore, the Pt/TCN-5v10 catalyst can exhibit higher stability on account of the higher Pt(0) content and the lower Pt(IV) content.

Fig. 4 XPS analysis for Pt/TCN-5v10 and commercial Pt/C. (a) and (b) Curve fitting of C 1s peak and (c) and (d) curve fitting of Pt 4f peak. The data in (d) is reproduced with permission.³³ Copyright 2015, RSC.

Table 2 Results of curve fitting of C 1s peaks for Pt/TCN-5v10 and Pt/C

Sample	sp^2-C 284.5 eV	C 285.0 eV	–C–OR 285.7 eV	[double bond splayed left]C=O 287.8 eV	–COOR 288.9 eV
Pt/TCN-5v10	49.6%	19.3%	20.4%	5.6%	5.1%
Pt/C	57.5%	19.7%	14.1%	4.2%	4.5%

---

Table 3 Result of curve fitting of Pt 4f peak for Pt/TCN-5v10 and Pt/C. The data for Pt/C is reproduced with permission.³³ Copyright 2015, RSC

Sample	Pt species	Binding energy	Relative content
Pt/TCN-5v10	Pt(0)	71.1 eV	52.5%
	Pt(II)	72.2 eV	31.3%
	Pt(IV)	74.2 eV	16.2%
Pt/C	Pt(0)	71.4 eV	41.0%
	Pt(II)	72.4 eV	27.7%
	Pt(IV)	74.1 eV	31.3%

---

The XPS regarding the different catalysts with different amounts of carbon are carried out under the same conditions. The deconvoluted C 1s and Pt 4f peaks are shown in Fig. S5 and S6,† and the results are listed in Tables S1 and S2.† The content ratio of sp²-C and C represents the degree of graphitization. It can be seen from Table S1† that the ratio decreases with increasing carbon content, indicating that the products with lower carbon content generate a more ordered carbon structure during the carbonization process. The high degree of graphitization results in low porosity and few defects, which may be why the mean sizes of the Pt nanoparticles are larger for the catalysts with low amounts of carbon. According to the deconvolution results of the Pt 4f peaks (Table S2†), the binding energies of Pt(0) for all as-prepared Pt/TCN catalysts shift down compared with that of commercial Pt/C, by 0.1 eV, 0.3 eV, 0.3 eV and 0.2 eV, respectively, for Pt/TCN-5v6, 5v8, 5v10 and 5v12. It is easy to see that the interaction between Pt and TiO₂ is different. The weaker interaction may be due to the higher degree of graphitization of Pt/TCN-5v6 and the thicker carbon layer of Pt/TCN-5v12. As is known, the strength of the metal-support interaction depends on the degree of contact between Pt and TiO₂. When the carbon content is too low, the degree of carbon graphitization is too high. The ordered carbon layer structure hinders the contact of Pt and TiO₂. When the carbon content is too high, the contact of Pt and TiO₂ is also hindered. Therefore, there should be a maximum interaction with increasing carbon content. When the mass ratio of TiO₂ and resorcinol is 5v10, the interaction between Pt and TiO₂ is stronger, further indicating that the contact of Pt nanoparticles and TiO₂ is better.

Conductivity is an important parameter for a catalyst support. Therefore, the four-probe method was used to measure the conductivity of the different carbon coated TiO₂ nanowires. In order to ensure accuracy, every sample was measured three times to obtain an average. Due to the very low conductivity of pure TiO₂ (about 10⁻¹⁰ S cm⁻¹), it is difficult to measure its conductivity using our test instrument, which agrees with our previous finding that pure Pt/TiO₂ catalyst has no electrochemical performance.³² However, the conductivity of the different carbon coated TiO₂ nanowires could be measured. Their conductivities are 7.5 × 10⁻⁴, 5.2 × 10⁻², 2.4 × 10⁻¹ and 2.0 × 10⁻¹ S cm⁻¹ for TCN-5v6, TCN-5v8, TCN-5v10 and TCN-5v12, respectively. It can be seen from the experiment results that the amount of carbon has a great influence on the conductivity. When the amount of carbon is too little, the conductivity of the support is insufficient. Therefore, the activity or stability of the catalyst will be poor. With increasing carbon content, the conductivity of the support gradually increases. Thus, the performance of the catalysts will be further improved. However, when the amount of carbon is too great, the conductivity of the support decreases due to the decreased degree of carbon graphitization. In addition, excessive carbon can impede the contact of Pt and TiO₂, leading to a decrease of the interaction between Pt and TiO₂ (which can be seen from the XPS results). Therefore, the performance of the catalysts will decrease when the amount of carbon is too great. This analysis is consistent with the electrochemistry results below.

Through the above analysis, the interaction between 3D-network carbon, titania nanowires and Pt nanoparticles can be determined. 3D-network carbon can distinctly increase the electronic conductivity of titania and inhibit the crystal reunion of titania. The large surface area and numerous surface oxygen groups of the 3D-network carbon provide many anchoring sites for Pt deposition. In addition, the TiO₂ nanowires can increase the hydrophilicity of the carbon network and generate SMSI with Pt nanoparticles. Obviously, the performance of the catalysts can improve due to interaction between the 3D-network carbon, titania nanowires and Pt nanoparticles. Moreover, the special 3D-network structure can offer high porosity, large surface area and efficient transport channels, which can further enhance the performance of the catalysts. In order to verify the enhancing effect, electrochemical measurements were carried out.

3.2 Electrochemical measurements and discussion

CV tests were conducted to evaluate the electrochemical active surface areas (ESA), which can reflect the number of available active sites. The ESA is calculated by the integral charge of the hydrogen adsorption–desorption region. The Pt/TCN-5v10 catalyst exhibits an outstanding ESA of 91.4 m² g⁻¹, which is far higher than those of Pt/TCN-5v6 (58.2 m² g⁻¹), 5v8 (66.4 m² g⁻¹) and 5v12 (72.7 m² g⁻¹) (Fig. S7a†). To measure the methanol electrooxidation reaction (MOR), the ratio between the peak current of the respective positive and negative potential scans (I_f/I_b) is an inadequate parameter to gauge the catalytic activity of Pt-based catalysts for MOR.³⁴ However, there is no doubt that the mass normalized value of I_f can reflect the MOR activity. As can be seen from Fig. S7b,† the Pt/TCN-5v10 catalyst exhibits the highest I_f, indicating the best activity for MOR. The EIS and i–t tests at a given electrode potential (close to the working potential in a DMFC) provide better measures of intrinsic activity for MOR. In our work, the EIS and i–t were measured at 0.6 V vs. NHE. In the EIS (Fig. S7c†), the semicircle of Pt/TCN-5v10 is much smaller than in the other samples, demonstrating that its electro-oxidation rate is much faster. In order to intuitively survey their impedance behavior, a simulation was carried out to obtain their resistance by using an equivalent (the inset of Fig. S7c†). The simulated results display that the charge-transfer resistance (R_ct) of Pt/TCN-5v10 (65.7 Ω cm²) is much lower than those of Pt/TCN-5v6 (99.1 Ω cm²), 5v8 (81.8 Ω cm²) and 5v12 (72.8 Ω cm²). In the i–t curves (Fig. S7d†), among the four catalysts, the mass current density of Pt/TCN-5v10 catalyst is the highest during 3600 s. Moreover, the durability of Pt/TCN catalysts can be reflected by the ratio of the final current and the maximum current. After 3600 s, their ratios were 47.4%, 45.0%, 63.1% and 47.2% for the Pt/TCN-5v6, 5v8, 5v10 and 5v12 catalysts, respectively, indicating that the durability of the Pt/TCN-5v10 catalyst is better than that of the other samples. From the physical analysis, the combination of the large surface area, good electronic conductivity and high porosity endows the Pt/TCN-5v10 catalyst with enhanced electro-catalytic performance.

To further evaluate the electro-catalytic performance of the as-prepared Pt/TCN-5v10 catalyst, commercial Pt/C was adopted as a reference, as shown in Fig. 5. The ESA of the Pt/TCN-5v10 catalyst is 91.4 m² g⁻¹, much higher than that of for commercial Pt/C (57.3 m² g⁻¹, Fig. 5a). With respect to MOR, the I_f of Pt/TCN-5v10 is 1.5 times higher than that of commercial Pt/C (Fig. 5b), and the R_ct of Pt/TCN-5v10 catalyst (65.7 Ω cm²) is much lower than that of commercial Pt/C (134.5 Ω cm², Fig. 5c). In other words, Pt/TCN-5v10 catalyst using only 66.7% of commercial Pt/C can achieve the same catalytic effect. Additionally, the durability of Pt/TCN-5v10 catalyst is 63.1%, much higher than that of for commercial Pt/C (11.0%, Fig. 5d). Overall, the electro-catalytic performance of the Pt/TCN-5v10 catalyst is distinctly improved in comparison with that of commercial Pt/C.

Fig. 5 Electrocatalytic performance of Pt/TCN-5v10 and commercial Pt/C catalysts: (a) CV curves in H₂SO₄ solution. (b) CV curves of methanol electro-oxidation. (c) Nyquist plots of methanol electro-oxidation. (d) Chronoamperometric current.

To survey the reason for the improved performance, the CO-stripping experiment was implemented, as shown in Fig. 6. It can be clearly observed that the active area of CO oxidation for Pt/TCN-5v10 is much larger than that of Pt/C, which is in agreement with the abovementioned CV results. However, the onset potentials of CO oxidation on the Pt/TCN-5v10 and Pt/C catalysts are almost the same, indicating that the presence of TiO₂ does not play a role in improving the CO tolerance of Pt-based catalysts. It is well known that CO oxidation on Pt requires the adsorption of OH. Although the core levels for Pt/TCN-5v10 shift to a lower binding energy relative to that for Pt/C, the metal-support interaction simultaneously weakens the Pt–CO bond and the Pt–OH bond. The opposite effect may be the reason for the same onset potentials of CO oxidation on the Pt/TCN-5v10 and Pt/C catalysts. If this is case, what causes the improved performance of methanol oxidation on Pt/TCN-5v10? The hydrogen spillover effect between Pt and TiO₂ may be an important reason.³⁵ The protons generated on Pt nanoparticles can be quickly transferred to TiO₂, thus promoting the dehydrogenation reaction on Pt nanoparticles, which can create more clean active Pt sites for methanol oxidation. It is worth mentioning that there is another possible reason. The pathway of methanol oxidation reaction via adsorbed CO is known as the CO path, while the pathway via a reactive intermediate is called the non-CO path.³⁶ The reaction pathway on Pt/TCN-5v10 may occur via the non-CO path. Currently, the specific reaction pathway is being investigated by in situ IR studies.

Fig. 6 CO-stripping voltammetry of the catalysts in 0.5 mol L⁻¹ H₂SO₄. The data for Pt/C is reproduced with permission.³³ Copyright 2015, RSC.

In addition, we assembled and tested a single DMFC which used 40 wt% Pt/TCN-5v10 as the anode catalyst and as-prepared 40 wt% Pt/C as the cathode catalyst. As a comparison, another single DMFC was assembled with as-prepared 40 wt% Pt/C as the anode and cathode catalysts. The polarization and power density curves of the two single cells are shown in Fig. 7. Generally, the polarization curve can be approximately divided into three regions, the activation polarization loss region, the ohmic polarization loss region and the concentration polarization loss region, with increasing polarization current.^31,37,38 It is worth noting that the three kinds of polarization losses all exist during the whole process; however, at different stages, the dominant polarization loss is different. The voltage loss in the range of 10 to 20 mA cm⁻² is mainly due to the activation polarization losses, which originate from the electron transfer resistance from the electrode to the reactants and products. The activation polarization losses are mainly used to overcome the reaction activation energy.³⁹ Therefore, they can reflect the activity of catalysts in the catalytic reaction. The significant voltage loss indicates that the electrochemical polarization of the catalytic reaction is severe.

Fig. 7 Performance of single DMFCs using as-prepared 40 wt% Pt/C and 40 wt% Pt/TCN-5v10 as anode catalysts with the same cathode catalyst of as-prepared 40 wt% Pt/C. Operating temperature: 80 °C. Anodic feed: 1.5 mol L⁻¹ CH₃OH solution with a flow rate of 3.0 mL min⁻¹. Cathodic feed: oxygen at ambient pressure with a flow rate of 200 mL min⁻¹.

In other words, the activity of the catalysts is still insufficient. However, the voltage loss of Pt/TCN-5v10 is less than that of Pt/C, and the single DMFC with 40 wt% Pt/TCN-5v10 shows initial activity at a higher voltage, demonstrating that the electro-catalytic activity of Pt/TCN-5v10 is higher. Moreover, the maximum polarization current of the single DMFC with 40 wt% Pt/TCN-5v10 is 272.9 mA cm⁻², higher than that with 40 wt% Pt/C (189.0 mA cm⁻²). The maximum polarization current primarily reflects the diffusion of fuel. The result demonstrates that the diffusion of methanol on Pt/TCN-5v10 is better, which is consistent with the unique 3D-network structure. As can be seen from the power density curves, the power density of Pt/TCN-5v10 is higher than that of Pt/C in the whole process, and the maximum power densities are 33.6 and 21.6 mW cm⁻², respectively. The higher single cell performance is ascribed to the higher electro-catalytic activity of Pt/TCN-5v10 and the 3D-network structure, which is conducive to the diffusion of fuel. In general, several combined features of the Pt/TCN-5v10 catalyst should contribute to the enhanced performance: (1) the 3D-network structured Pt/TCN catalyst offers high porosity, large surface area and efficient transport channels; (2) the carbonized carbon not only greatly improves the electronic conductivity but provides many anchoring sites for Pt deposition; (3) the synergetic effect of the different components enhances the electro-catalytic performance effectively.

4. Conclusions

In summary, our work has shown an efficient strategy to synthesize unique 3D nano-network structured TiO₂@C nanowires as Pt-based catalyst supports. The presence of carbon derived from resorcinol–formaldehyde polymer (RF) carbonization has been demonstrated to be the most critical factor for the generation of the porous 3D-network structure. The resorcinol–formaldehyde content of the polymer was optimized for the synthesis of high-quality supports. The 3D-network structured Pt/TCN-5v10 catalyst acted as a highly efficient catalyst for MOR, with much better performance than commercial Pt/C. Our catalyst can achieve the same catalytic effect using only 66.7% of commercial Pt/C. The improved performance can be ascribed to the highly porous 3D nano-network structure, the abundant sites of Pt deposition and the interaction of the different components. In addition, the single DMFC tests indicated that the Pt/TCN-5v10 catalyst has higher electro-catalytic performance and a better diffusion path for methanol. The results suggest that excellent catalysts can be developed by designing unique support structures and compositions.

Acknowledgements

This research is financially supported by the National Natural Science Foundation of China (Grant No. 21273058), China postdoctoral science foundation (Grant No. 2012M520731 and 2014T70350), Heilongjiang postdoctoral foundation (LBH-Z12089).

Notes and references

1. A. Schroder, K. Wippermann, J. Mergel, W. Lehnert, D. Stolten, T. Sanders, T. Baumhofer, D. U. Sauer, I. Manke, N. Kardjilov, A. Hilger, J. Schloesser, J. Banhart and C. Hartnig, Electrochem. Commun., 2009, 11, 1606–1609.
2. M. Winter and R. J. Brodd, Chem. Rev., 2004, 104, 4245–4269.
3. X. Zhao, M. Yin, L. Ma, L. Liang, C. Liu, J. Liao, T. Lu and W. Xing, Energy Environ. Sci., 2011, 4, 2736–2753.
4. A. Morozan, B. Jousselme and S. Palacin, Energy Environ. Sci., 2011, 4, 1238–1254.
5. H. R. Byon, J. Suntivich and Y. Shao-Horn, Chem. Mater., 2011, 23, 3421–3428.
6. R. Othman, A. L. Dicks and Z. H. Zhu, Int. J. Hydrogen Energy, 2012, 37, 357–372.
7. L. Chen, X. Cui, M. Wang, Y. Du, X. Zhang, G. Wan, L. Zhang, F. Cui, C. Wei and J. Shi, Langmuir, 2015, 31, 7644–7651.
8. J. Wu, D. Zhang, H. Niwa, Y. Harada, M. Oshima, H. Ofuchi, Y. Nabae, T. Okajima and T. Ohsaka, Langmuir, 2015, 31, 5529–5536.
9. Y. Cao, Y. Yang, Y. Shan and Z. Huang, ACS Appl. Mater. Interfaces, 2016, 8, 5998–6003.
10. S. Manivannan, I. Kang and K. Kim, Langmuir, 2016, 32, 1890–1898.
11. C. L. Hui, X. G. Li and I. M. Hsing, Electrochim. Acta, 2005, 51, 711–719.
12. Z. Yang, C. Kim, S. Hirata, T. Fujigaya and N. Nakashima, ACS Appl. Mater. Interfaces, 2015, 7, 15885–15891.
13. P. J. Ferreira, G. J. la O', Y. Shao-Horn, D. Morgan, R. Makharia, S. Kocha and H. A. Gasteiger, J. Electrochem. Soc., 2005, 152, A2256–A2271.
14. S. Wu, J. Liu, Z. Tian, Y. Cai, Y. Ye, Q. Yuan and C. Liang, ACS Appl. Mater. Interfaces, 2015, 7, 22935–22940.
15. S. S. Li, J. J. Lv, Y. Y. Hu, J. N. Zheng, J. R. Chen, A. J. Wang and J. J. Feng, J. Power Sources, 2014, 247, 213–218.
16. L. L. He, J. N. Zheng, P. Song, S. X. Zhong, A. J. Wang, Z. J. Chen and J. J. Feng, J. Power Sources, 2015, 276, 357–364.
17. P. Song, S. S. Li, L. L. He, J. J. Feng, L. Wu, S. X. Zhong and A. J. Wang, RSC Adv., 2015, 5, 87061–87068.
18. Y. J. Wang, D. P. Wilkinson and J. J. Zhang, Chem. Rev., 2011, 111, 7625–7651.
19. S. Sharma and B. G. Pollet, J. Power Sources, 2012, 208, 96–119.
20. Z. Z. Jiang, Z. B. Wang, Y. Y. Chu, D. M. Gu and G. P. Yin, Energy Environ. Sci., 2011, 4, 728–735.
21. J. W. Magee, W. P. Zhou and M. G. White, Appl. Catal., B, 2014, 152, 397–402.
22. Y.-Y. Chu, Z.-B. Wang, Z.-Z. Jiang, D.-M. Gu and G.-P. Yin, Adv. Mater., 2011, 23, 3100–3104.
23. Q. Lv, M. Yin, X. Zhao, C. Y. Li, C. P. Liu and W. Xing, J. Power Sources, 2012, 218, 93–99.
24. C. Subban, Q. Zhou, B. Leonard, C. Ranjan, H. M. Edvenson, F. J. DiSalvo, S. Munie and J. Hunting, Philos. Trans. R. Soc., A, 2010, 368, 3243–3253.
25. S. K. Meher and G. R. Rao, J. Phys. Chem. C, 2013, 117, 4888–4900.
26. C. Zhang, H. Yu, Y. Li, L. Fu, Y. Gao, W. Song, Z. Shao and B. Yi, Nanoscale, 2013, 5, 6834–6841.
27. L. Chevallier, A. Bauer, S. Cavaliere, R. Hui, J. Roziere and D. J. Jones, ACS Appl. Mater. Interfaces, 2012, 4, 1752–1759.
28. Y. Lan, X. P. Gao, H. Y. Zhu, Z. F. Zheng, T. Y. Yan, F. Wu, S. P. Ringer and D. Y. Song, Adv. Funct. Mater., 2005, 15, 1310–1318.
29. M. Inagaki, Y. Hirose, T. Matsunaga, T. Tsumura and M. Toyoda, Carbon, 2003, 41, 2619–2624.
30. L. W. Zhang, H. B. Fu and Y. F. Zhu, Adv. Funct. Mater., 2008, 18, 2180–2189.
31. Y. Sugawara, T. Okayasu, A. P. Yadav, A. Nishikata and T. Tsuru, J. Electrochem. Soc., 2012, 159, F779–F786.
32. X. L. Sui, Z. B. Wang, M. Yang, L. Huo, D. M. Gu and G. P. Yin, J. Power Sources, 2014, 255, 43–51.
33. X.-L. Sui, Z.-B. Wang, Y.-F. Xia, M. Yang, L. Zhao and D.-M. Gu, RSC Adv., 2015, 5, 35518–35523.
34. A. M. Hofstead-Duffy, D. J. Chen, S. G. Sun and Y. Y. J. Tong, J. Mater. Chem., 2012, 22, 5205–5208.
35. S. J. Yoo, K.-S. Lee, Y.-H. Cho, S.-K. Kim, T.-H. Lim and Y.-E. Sung, Electrocatalysis, 2011, 2, 297–306.
36. K. Kunimatsu, H. Hanawa, H. Uchida and M. Watanabe, J. Electroanal. Chem., 2009, 632, 109–119.
37. J. G. Carton and A. G. Olabi, Energy, 2010, 35, 2796–2806.
38. Q. Yang, A. Kianimanesh, T. Freiheit, S. S. Park and D. Xue, J. Power Sources, 2011, 196, 10640–10651.
39. J. O. M. Bockris and A. K. N. Reddy, Modern Electrochemistry, Plenum, 1970.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra17542b

---