Study of the Relationship between Metal–Support Interactions and the Electrocatalytic Performance of Pt/Ti₄O₇ with Different Loadings

Abstract

The application potential of Pt/Ti₄O₇ has been reported, but the lack of research on the relationship between Pt loading, MSI, and catalytic activity hinders further development. Micron-sized Ti₄O₇ powders synthesized by a thermal reduction method under an H₂ atmosphere were used as a support material for Pt-based catalysts. Using a modified polyol method, Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 with different mass ratios (Pt to Pt/Ti₄O₇ is 0.05, 0.1, 0.2) were successfully synthesized. Uniformly dispersed platinum nanoparticles exhibit disparate morphologies, rod-like for Pt/Ti₄O₇-5 and approximately spherical for Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20. Small-angle deflections and lattice reconstruction induced by strong metal–support interactions were observed in Pt/Ti₄O₇-5, which indicated the formation of a new phase at the interface. However, lattice distortions and dislocations for higher loading samples imply the existence of weak metal–support interactions. A possible mechanism is proposed to explain the different morphologies and varying metal–support interactions (MSI). With X-ray photoelectron spectroscopy, spectrums of Pt and Ti display apparent shifts in binding energy compared with commercial Pt-C and non-platinized Ti₄O₇, which can properly explain the changes in absorption ability and oxygen reduction reaction activity, as described in the electrochemical results. The synthetic method, Pt loading, and surface coverage of the support play an important role in the adjustment of MSI, which gives significant guidance for better utilizing MSI to prepare the target catalyst.

1. Introduction

Advantages like high efficiency, environment friendly, and reactants bounteously make fuel cells a promising candidate for powering vehicles, as well as energy storage stations and portable power generation [1]. Although numerous studies have been carried out on non-precious metal catalysts, carbon-supported platinum nanoparticles (Pt/C) are still the most practical catalysts because of their intrinsic outstanding catalytic activities [2]. Carbon materials including Vulcan XC-72R, Ketjen Black, and Black Pearl are usually employed as supports in Pt/C catalysts. These carbon supports, however, are extremely susceptible to corrosion under the electrochemical condition, and such corrosion can lead to the agglomeration, dissolution, and detachment of Pt NPs (platinum nanoparticles) from carbon supports, which is identified as one of the major causes of performance degradation and a lifetime reduction of Pt/C catalysts [3]. Substantial efforts have been devoted to the exploration of alternative support materials for Pt NPs. Non-carbon materials such as nitrides, carbides, oxides, sulfides, borides, and metals have been used to support Pt NPs [4]. Compared to Pt/C catalysts, these non-carbon materials-supported Pt NPs exhibit improved stability/durability, which is ascribed to the resistance of non-carbon materials to electrochemical corrosion. Among the various non-carbon materials, transition metal oxides have received extensive attention for Pt NPs supporting because of their high electrochemical stability and thermal stability, as well as low-cost and easy manufacturing. Of all kinds of transition metal oxides, titanium dioxide (TiO₂) has been the most extensively studied because of its abundance, non-toxicity, and wide application in photocatalysis, photovoltaics, and water splitting [5]. There have been several reports on preparing TiO₂-supported Pt NPs catalysts (Pt/TiO₂) [6,7,8,9,10]. These Pt/TiO₂ catalysts, however, did not exhibit a comparable oxygen reduction reaction (ORR) catalytic performance to Pt/C catalysts, and the poor electroconductivity of TiO₂ was considered the main reason [11] leading to the decreased catalytic activity of Pt/TiO₂ catalysts. Although strategies such as doping TiO₂ with Nb [12,13] or N [14,15] and compositing TiO₂ with carbon material [16,17] have been proposed to improve the conductivity of TiO₂ support, the ORR catalytic activities of Pt/TiO₂ catalysts were still not as high as those of commercial Pt/C catalysts.

Ti₄O₇ was employed as the support for Pt NPs because Ti₄O₇ has electrical conductivity up to 1500 S·cm⁻¹ at room temperature, which is comparable to graphitized carbon [18]. In addition, as a member of Magnéli phases titanium suboxides with the general formula of Ti_nO_2n−1 (3 < n < 10) [19], Ti₄O₇ exhibits excellent chemical and thermal stability, and especially high resistance to electrochemical corrosion [20]. Many researchers have explored synthesizing Pt/Ti₄O₇ to elevate the catalytic activity and stability for ORR in the cathodes of fuel cells [21,22,23,24]. All of the results illustrate that Pt/Ti₄O₇ displays superior stability whether in the rotating disk electrode or membrane electrode assembly test, while the electrochemical surface area (ECSA) and mass activity (MA) of Pt/Ti₄O₇ are inferior to the Pt/C catalyst. The intrinsic stability of Ti₄O₇ and its strong metal–support interactions (SMSI) are recognized to be important reasons for this superior stability. As for the relatively poor catalytic performance, the researchers attribute it to the low conductivity of metal oxides and low platinum utilization due to the aggregation of Pt NPs [24]. Based on the above shortcomings, after 2013, there was a blank period for studies of the Pt/Ti₄O₇ catalyst. However, in the last decade, there have been many research papers on metal oxide-supported catalysts used in propylene hydrogenation, hydrolytic ammoniation, the water-gas shift reaction, and other chemical fields [25,26,27,28]. Metal–support interactions (MSI) play an important role in catalytic activity and selectivity. This provides a new possible explanation for the poor electrochemical activity of the Pt/Ti₄O₇ catalyst, that is the MSI lead to the formation of a Pt-Ti bond at the interface, which reduces the active platinum content and leads to a decrease in catalytic activity. This hypothesis was first verified by the experiments in this article. Initially, low Pt loading (<5 wt%) was reported as limited by the low surface area of Ti₄O₇ [22]. Next, in order to improve the performance of Pt/Ti₄O₇ catalysts, a series of tasks were carried out, including the preparation of a high Ti₄O₇ -specific surface area [24,29] and a high number of Pt loading (40 wt%) Pt/Ti₄O₇ catalysts [30]. The Pt loading is often limited by the properties of the support, loading method, etc. None of the previous articles have explored the relationship between the electrocatalytic performance of Pt/Ti₄O₇ with different Pt loadings and MSI. It’s still a problem to tune the MSI to acquire an ideal catalyst with desirable properties.

In our experiments, Ti₄O₇-supported Pt NP (Pt/Ti₄O₇) catalysts with different Pt loading were prepared according to the modified polyol method. It was found that the MSI between Pt and the Ti³⁺ played a significant role during the formation of Pt nuclei and the subsequent growth of Pt NPs, and thus greatly affected the size and shape of Pt NPs on Ti₄O₇ support. Although the Ti₄O₇ support has very high electrical conductivity, Pt/Ti₄O₇ catalysts still displayed ORR catalytic activity worse than commercial Pt/C catalysts. A new possible explanation that the formation of Pt-Ti at the interface induced by the MSI is responsible for the reduction in catalytic performance. However, after the portion of Pt atoms which had MSI with Ti were excluded, we found that the Pt NPs exhibit an active surface area and catalytic activity that are comparable to commercial Pt/C catalysts. Our research results manifest that the catalytic activity of transition metal oxides-supported Pt NPs was affected by not only the poor electroconductivity of the support, but also the MSI.

2. Results and Discussion

2.1. Preparation of Ti₄O₇

Ti₄O₇ was prepared by calcination of P25 (Aeroxide^® TiO₂) powder under H₂ atmosphere. Figure 1a is a digital image of the pristine P25, in which the P25 powder looks like white flour. However, after being calcined under an H₂ atmosphere at high temperature, the pristine P25 turns from white flour to dark granules. Figure 1b–d presents digital photos of the P25 powder calcined at different temperatures. As seen, after being calcined at 1123, 1173 and 1223 K, the P25 turned from white powder into grey-blue, black-blue, and black granules, respectively. XRD (X-ray diffraction) measurement was conducted to characterize the crystalline structures of the calcined P25 granules. For comparison, the XRD pattern of P25 was also recorded. As shown in Figure 1e, the P25 is composed of anatase and rutile TiO₂, and the molar ratio of anatase to rutile can be obtained from the K-value method. The ratio of anatase to rutile is about 3:1, which is in accordance with the reference [31]. Figure 1f presents the XRD patterns of the calcined products of P25 at 1173 K. It can be seen from Figure S1 that P25-1123 consisted of Ti₆O₁₁, Ti₅O₇, and Ti₄O₇, while the diffraction peaks on the XRD pattern of P25-1173 are well indexed to the characteristic diffraction peaks of Ti₄O₇ (Figure 1f), indicating that P25-1173 contains only Ti₄O₇ and no other Magnéli phases. However, when the calcination temperature rises to 1223 K, the diffraction peaks on the XRD pattern match well with the JCPDS card NO. 76-1066, as shown in Figure S1, suggesting P25-1223 is composed of Ti₃O₅. It is reported Ti₄O₇ is very stable. In our work, we found that even after 4 months of storage in capped vials at room temperature, our prepared P25-1173 maintained its color and granular shape. XRD measurement proved that P25-1173 keeps its Ti₄O₇ crystalline structure after 4 months of storage (Figure S2).

A SEM (scanning electron microscope), TEM (transmission electron microscope), and HRTEM (high-resolution transmission electron microscope) were employed to investigate the morphology and lattice structure of P25-1173, and the TEM and HRTEM images of P25 were also recorded for comparison. As shown in Figure 2a,b, P25 exhibits a particular morphology with an average particle size of 20 nm, and the interplanar spacings of 0.35 nm and 0.24 nm correspond to the (1 0 1) and (1 0 3) crystal faces of anatase TiO₂. Figure 2c,d show the SEM and HRTEM images of P25-1173, in which P25-1173 exhibits a micron to sub-micron irregular pore structure and the interplanar spacings of 0.34 nm and 0.26 nm are attributed to the (1 −2 0) and (2 0 0) crystalline faces of Ti₄O₇. The TEM image of P25-1173 is presented as an insert in Figure 2e, which indicates that coral-structured P25-1173 is composed of Ti₄O₇ submicron-plates. In the TEM and HRTEM images of P25-1173, no mesopores or micropores were observed, which accounted for the very small specific surface area of P25-1173. This specific surface area of P25-1173 was determined to be only 2.16 m²·g⁻¹ by the Brannuaer-Emmet-Teller (BET) method.

2.2. Synthesized Pt/Ti₄O₇ Catalysts

2.2.1. Morphologies and Crystalline Structures

Pt/Ti₄O₇ with different Pt loading levels were synthesized using the modified polyol method by varying the mass ratio of P25-1173 to Pt. XRD measurements were obtained to investigate the crystalline structures of the Pt/Ti₄O₇ samples with different Pt loadings. As shown in Figure 3a, even after 12 h of the reduction reaction at 433 K, the diffraction peaks of all the three Pt/Ti₄O₇ samples were still well indexed to the characteristic diffraction peaks of Ti₄O₇, implying that P25-1173 maintained its crystalline structure even after 12 h of reduction reaction. In addition to the diffraction peaks of Ti₄O₇, four new peaks emerged on the XRD patterns of the Pt/Ti₄O₇ samples. The four peaks at 40°, 46°, 67°, and 81° corresponded to the (1 1 1), (2 0 0), (2 2 0), and (3 1 1) reflections of Pt, suggesting successful loading of Pt on Ti₄O₇ in the three samples. Inductively coupled plasma-atomic emission spectrometry (ICP-AES) was used to measure the actual Pt loading in the samples, and the Pt loadings for Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 were determined to be 3.48 wt%, 6.95 wt%, and 12.60 wt%, respectively. Figure S3 displays the SEM images of Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20. It can be seen from Figure S3 that P25-1173 kept its morphology after hours of thermal reduction. Energy dispersive X-ray (EDX) elemental mapping analysis shows that the Pt element is homogeneously dispersed on the Ti₄O₇ substrate in all three samples, despite the Pt loading. The particle size of Pt on a Ti₄O₇ substrate could be calculated according to the diffraction line of Pt (1 1 1) using the Scherrer equation:

$$\tau =K\lambda /\beta \cos \theta ,$$

(1)

where $\tau$ is the mean size of the crystalline domains, K is the dimensionless shape factor (typical value is 0.9, varies with the actual shape), $\lambda$ is the X-ray wavelength, $\beta$ is the full width at half the maximum of the diffraction peak, and $\theta$ is the Bragg angle. The particle sizes of Pt for Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 are calculated to be 10.7, 7.7, and 5.8 nm, respectively.

It is reported that as Ti³⁺ with electron configuration [Ar] 3d¹ can induce d-orbital electrons’ transformation from TiO_x (x < 2) to Pt (electron configuration: [Xe] 4f¹⁴5d⁹4s¹), TiO_x can easily combine with Pt [32,33]. Figure 3b exhibits the crystalline structure of Ti₄O₇ as its crystalline structure has been revealed. The blue octahedra indicate edge-sharing sites of rutile structure TiO₂, and the green ones suggest face-sharing sites of corundum structure Ti₂O₃ [34]. It is because of the Ti³⁺ in Ti₂O₃ that Pt can be easily loaded on the Ti₄O₇ substrate.

A TEM and HRTEM were applied to further investigate the morphologies and crystalline structures of the Pt/Ti₄O₇ samples. Figure 4a is the TEM image of the Pt/Ti₄O₇-5, in which Pt nanorods of ca., 9.5 nm long and 5.3 nm wide, are observed on the Ti₄O₇ substrate. The size of the Pt nanorods obtained from the TEM image is very close to the size of the Pt particles calculated from the XRD pattern. It is seen in Figure 4a that the Pt nanorods are aggregated, and the aggregations of Pt nanorods are not evenly dispersed on the substrate. Figure 4b is the HRTEM image of Pt/Ti₄O₇-5, and the interplanar spacings of 0.34, 0.22, and 0.19 nm can be assigned to the Bragg reflections from Ti₄O₇ (1 −2 0), Pt (1 1 1), and Pt (2 0 0), respectively. The HRTEM image clearly shows that the Pt nanorods are confined in the lattice of the Ti₄O₇ rather than being adsorbed. An Inverse Fast Fourier Transform (IFFT) of the HRTEM image (Figure 4c) was applied to investigate the interface of the Pt nanorods and Ti₄O₇. The selected IFFT image shows that the Ti₄O₇ lattice and Pt lattice merge at the interface, which implies that Pt and Ti₄O₇ are fused at the interface.

Figure 4d is the TEM image of Pt/Ti₄O₇-10, in which the size of the Pt NPs is about 7.2 nm, which is almost consistent with the particle size obtained from the XRD pattern. Different from the Pt nanorods in Pt/Ti₄O₇-5, the Pt NPs in Pt/Ti₄O₇-10 are not aggregated. Instead, only several Pt NPs are agglomerated. The Pt agglomerates are dispersed more evenly on Ti₄O₇ than the aggregated Pt nanorods. The HRTEM image of Pt/Ti₄O₇-10 is illustrated in Figure 4e, in which the interplanar spacings corresponding to the Bragg reflections of Ti₄O₇ and Pt are observed. However, it can be seen from the HRTEM image that in Pt/Ti₄O₇-10, the Pt NPs seem to be adsorbed on the substrate rather than confined in the substrate lattice. The selected IFFT image (Figure 4f) reveals no lattice fusion, but lattice distortion is observed at the Pt and Ti₄O₇ interface. The lattice distortion comes from strong spin-split coupling, in which verifying Pt NPs are adsorbed on Ti₄O₇ in Pt/Ti₄O₇-10. Figure 4g,h shows the TEM and HRTEM images of Pt/Ti₄O₇-20, in which highly dispersed Pt NPs are spread on the Ti₄O₇ substrate. The size of the Pt NPs is about 5.9 nm, which is consistent with the particle size obtained from the XRD pattern. The selected IFFT image (Figure 4i) discloses the lattice distortions and dislocations [35] at the Pt and Ti₄O₇ interface, which indicate that in Pt/Ti₄O₇-20, the Pt NPs are adsorbed on Ti₄O₇.

To explain the reason why the particle size of Pt NPs on Ti₄O₇ decreases with increases in the mass ratio of Pt to Ti₄O₇, while the dispersion of Pt NPs increases with the mass ratio, a possible mechanism for deposition of Pt on Ti₄O₇ is proposed. As described in Figure 5, in a typical synthetic process, the mixture of diethylene glycol (DEG) and Ti₄O₇ is heated to 433 K. During the heating, reductive molecules such as diols, alkyds, aldehydes, carboxylic acids are generated due to the degradation and oxidation of DEG [36]. Except for the reducing agent, DEG performs as a solvent and stabilizer in the system. As a solvent, it has a relatively large viscosity, which is beneficial to the dispersion of relatively dense Ti₄O₇. Tightly adsorbed on the surface of nanoparticles, it provides electrostatic repulsion and steric hindrance between particles to inhibit aggregation and agglomeration, and thus acts as a stabilizer. The reductive molecules mentioned above can reduce Pt⁴⁺ into Pt²⁺ or Pt⁰. Thus, when the first several drops of H₂PtCl₆·6H₂O/DEG solution are added into the heated DEG/Ti₄O₇ system, Pt nuclei are immediately formed due to the reduction of Pt⁴⁺. The formed Pt nuclei can either bind with the Ti₄O₇ substrate or remain in the DEG/Ti₄O₇ system. If the concentration of Ti₄O₇ is low, many nuclei will stay in the DEG/Ti₄O₇ system and grow into large Pt NPs with the continuous addition of a Pt precursor (Reaction Ⅰ). For the minimum support system (Pt/Ti₄O₇-5), since most of the nuclei are in the solution, the Reaction Ⅰ dominates, tending to generate larger Pt particles before combining with Ti₄O₇. The large nanoparticles will undergo deformation due to the SMSI and large gaps between Pt NPs, turning from nanoparticles into nanorods, and such deformation will easily result in the aggregation of nanorods [37,38,39]. However, if the concentration of Ti₄O₇ is high, most of the formed nuclei can be deposited on Ti₄O₇ (Reaction ⅠⅠ) because of the electron donation from Ti³⁺ to Pt. As almost no Pt nuclei existed in the DEG/Ti₄O₇ system, the continuous addition of the Pt precursor will lead to the generation of new Pt nuclei, and these Pt nuclei will also combine with the Ti₄O₇ substrate. Therefore, in Pt/Ti₄O₇-20, small Pt NPs are observed to be evenly dispersed on the substrate. Restricted by the limited gap between particles or the extremely high surface coverage of the support, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 only suffered weak metal–support interactions (WMSI), which allowed the existence of slightly deformed, approximately spherical particles (Figure 4e,h) and the distortion of lattice fringes at the interface area (Figure 4f,i).

2.2.2. Electrocatalytic Performances of Pt/Ti₄O₇ Samples

Cyclic voltammetry (CV) was performed in a 0.5 M H₂SO₄ solution between −0.70 and 0.45 V (vs. Hg/Hg₂SO₄) at 0.05 V·s⁻¹ to evaluate the ECSA of the Pt/Ti₄O₇ catalysts. Before the CV measurements, the 0.5 M H₂SO₄ solution was deaired by bubbling nitrogen for 20 min. Figure 6a displays the CV curves of Pt/Ti₄O₇ catalysts. For comparison, the CV curve of Ti₄O₇ is also recorded. As shown in Figure 6a, the Pt/Ti₄O₇ catalysts all present typical CV curves of polycrystalline Pt with clear hydrogen adsorption/desorption and oxidation/reduction regions, whereas Ti₄O₇ only shows the CV curve of a featureless quasi-rectangle shape. The ECSA of Pt/Ti₄O₇ catalysts can be obtained from the hydrogen underpotential deposition (H_upd) region by integrating the hydrogen adsorption/desorption region after double-layer current correction, assuming a pseudo-capacity of 210 μC·cm_Pt⁻² for polycrystalline Pt. The H_upd [cm_Pt²] was normalized by the actual Pt loading used to express the ECSA [m²·g_Pt⁻¹] [40]. Figure 6c shows the ECSA for the Pt/Ti₄O₇ catalysts based on the CV curves (Figure 6a). As seen, the ECSAs for Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 were calculated to be 4.56, 15.86, and 25.65 m²·g_{total Pt}⁻¹, respectively, which means that for Pt/Ti₄O₇ catalysts, the lower the Pt loading, the smaller the ECSA. The ECSA is three times as high compared to around 7 m²·g_Pt⁻¹ (Pt loading: 20 wt%) in Senevirathne’s article [23], while only possessing a half value of that fiber-like, nanostructured Ti₄O₇-loaded Pt catalyst [24]. We also investigated the ECSA of commercial Pt/C catalysts with different Pt loadings. Figure S4 shows the CV curves of the commercial JM Pt/C with 20% and 40% Pt loading, and the ECSAs are calculated to be 79.0 and 58.1 m²·g_Pt⁻¹ for JM Pt/C-40 and JM Pt/C-20, separately, which suggest that for Pt/C catalysts, the ECSA is kept at ca. 70 m²·g_Pt⁻¹ no matter how much Pt is loaded.

The electrocatalytic activity of catalysts towards ORR was examined by a linear sweep voltammetry (LSV) technique in an oxygen-saturated 0.1 M KOH solution at a scan rate of 10 mV·s⁻¹. The LSV curves for the support Ti₄O₇ and Vulcan XC-72R in Figure 6b snd Figure S7 exhibit similar ORR activity with an onset potential around 0.70 V and a peak potential around 0.45 V. As shown in Figure 6b, Pt/Ti₄O₇-5 catalyst, to a large extent, performs the characteristics of Ti₄O₇ rather than Pt, and the onset potential of ORR on an electrode is ca. 0.91 V, which is about 21 mV positive to the Ti₄O₇ electrode. Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20 catalysts exhibit a comparable ORR activity to Pt/C catalyst. The onset potentials for ORR of Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20 are ca. 0.99 and 1.01 V, which is about 4 and 2 mV negative to commercial Pt/C, and the half-wave potential towards ORR of the two are ca. 0.83 and 0.85 V, which is about 1 and 3 mV positive to the Pt/C catalyst.

Why can Pt loading affect the ECSA of Pt/Ti₄O₇ catalysts so much, yet have little effect on the ECSA of commercial Pt/C catalysts? Figure S5 presents the TEM images of JM Pt/C-20 and JM Pt/C-40. It is seen in both TEM images that the Pt NPs present a particular morphology with sizes around 3 nm, implying that the Pt NPs have similar specific surface areas in JM Pt/C-40 and JM Pt/C-20. Thus, we calculated the specific surface areas of the Pt NPs in the Pt/Ti₄O₇ catalysts. As shown in Figure S6, the specific surface area of Pt NPs were calculated to be 0.97, 0.83, and 1.0 ${\rho }_{Pt}{}^{-1}$ for Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20, respectively, which indicates that the specific surface area of Pt NPs has nearly no effect on the ECSA of Pt/Ti₄O₇ catalysts.

Measurements were obtained by X-ray photoelectron spectroscopy (XPS) to analyze the surface electron structures of the catalysts. Figure 7a,b shows the high-resolution Ti 2p and Pt 4f XPS spectra of the Pt/Ti₄O₇ catalysts. The Ti 2p XPS spectrum of support Ti₄O₇ and the Pt 4f XPS spectrum of JM Pt/C-20 were also recorded for comparison. As shown in Figure 7a, Ti₄O₇ shows two peaks, at 458.63 and 464.39 eV, which are attributed to the Ti 2p_3/2 and Ti 2p_1/2 orbits, respectively [41,42,43]. However, on the spectra of the Pt/Ti₄O₇ catalysts, the Ti 2p_3/2 and Ti 2p_1/2 peaks both shift positively to a higher binding energy (BE), suggesting that there are should be fewer electrons surrounding the Ti atoms in Pt/Ti₄O₇ catalysts. Figure 7b presents the Pt 4f XPS spectra of Pt/Ti₄O₇ catalysts. The Pt 4f XPS spectra of a commercial Pt/C catalyst is also shown as the benchmark. There are two Pt 4f peaks, Pt 4f_5/2 and Pt 4f_7/2, located on 71.84 and 75.14 eV [44,45], shown on the Pt 4f XPS spectrum of the commercial Pt/C while the Pt 4f_5/2 and Pt 4f_7/2 peaks of the Pt/Ti₄O₇ catalysts all shift negatively to a lower BE. This suggests that there should be more electrons surrounding the Pt atoms in Pt/Ti₄O₇ catalysts compared to Pt/C. The shift of Ti 2p peaks to a higher BE and Pt 4f peaks to a lower BE discloses that in Pt/Ti₄O₇ catalysts, electrons are transferred from Ti atom to Pt atom [46,47]. The absolute values of ΔBE for Pt/Ti₄O₇ catalysts are shown in Table S1. It can be seen that Pt/Ti₄O₇-5 shows the biggest ΔBE (1.10 eV) for the Pt element, while Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20 exhibit diminished values, 0.87 and 0.88 eV, indicating a stronger interaction between Pt NPs and Ti₄O₇ for Pt/Ti₄O₇-5.

As shown in Figure 7b, the spectrum for Pt/C consists of three pairs of overlapping curves: Pt (0), located at 71.83 and 75.11 eV; Pt (II), located at 72.97 and 76.18 eV; and Pt (IV), located at 75.51 and 78.75 eV, which are classified as “active Pt” and used as benchmarks. Zhao et al. [48] investigated the MSI between Pt NPs and TiO₂ support. They found that during the formation of the surface Pt-Ti alloy, Ti atoms donate their electrons to Pt, and such a MSI reduced the affinity of the Pt atoms to H₂. Therefore, the Pt 4f peaks are deconvoluted into the Pt 4f_5/2 and Pt 4f_7/2 peaks of “active Pt” and the Pt 4f_5/2 and Pt 4f_7/2 peaks of Pt_x-Ti. The “active Pt” area was divided according to the commercial JM Pt/C-20 catalyst with the peak positions of 71.84 and 75.14 eV and the area ratio of 4f_5/2 to 4f_7/2 is 0.75 [49]. The relative amounts of active Pt and Pt_x-Ti for three Pt/Ti₄O₇ catalysts can be calculated separately, as shown in Figure 7b. As the H₂ adsorption on Pt atoms in the Pt-Ti phase is suppressed, we think that only the active Pt provides active sites for H₂ adsorption/desorption. Thus, correct the ECSA calculation by using the active Pt loading instead of the actual Pt loading and the corrected ECSAs for Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 are 52.11, 42.38, and 59.78 m²·g_{active Pt}⁻¹, as shown in Figure 6c, which suggests that for Ti₄O₇ catalysts, the corrected ECSA is kept at ca. 50 m²·g_Pt⁻¹ no matter how much Pt is loaded, complying with the general rules for Pt catalysts.

For a more accurate evaluation of the electrocatalytic activity of ORR, the kinetic current ($i_k$) was calculated through the Koutecky–Levich equation:

$$\frac{1}{i}=\frac{1}{i_L}+\frac{1}{i_k},$$

(2)

where $i$ is the actual current at 0.9V vs. RHE, $i_L$ is the limit current, and the MA was calculated using actual and active Pt loading normalizations, respectively. In Figure 6b, the shape of Pt/Ti₄O₇-5 deviates far from the four-electron reaction of normal platinum catalysts, and a reasonable $i_L$ cannot be determined, therefore only the MA of Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20 were calculated. The MA of JM Pt/C-20 was also recorded according to Figure S7 for comparison. As shown in Figure 6d, Pt/Ti₄O₇ catalysts perform only half the activity of JM Pt/C-20 catalysts affected by MSI. However, it is slightly higher than the MA values for similar systems in the related literature, which are 22 [24] and 32 A·g_Pt⁻¹ [11], respectively. Correct the mass activity with active Pt loading instead of actual Pt loading and the corrected mass activity values are around 90 A·g_{active Pt}⁻¹, which reaches the normal standard of Pt catalysts.

3. Materials and Methods

3.1. Chemicals

Aeroxide^® TiO₂ P25 was purchased from Degussa (Evonik, Shanghai, China). Carbon blacks (Vulcan XC-72R) were purchased from Cabot (Boston, MA, America), and Pt/C (20 wt% Pt on Vulcan XC-72, Hispec 3000) was commercially acquired from Johnson Matthey (London, England). A 5% Nafion^® solution (D 520) from Dupont (Wilmington, DE, America) was used as ionomer for catalyst ink. DEG, chloroplatinic acid hexahydrate (H₂PtCl₆·6H₂O, Pt ≥ 37.5%), ethanol, isopropanol (IPA), and potassium hydroxide (KOH) were supplied by Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Sulfuric acid (H₂SO₄) and acetone were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Ultrapure water (18.2 MΩ·cm @ 25 °C) and H₂, nitrogen N_2, and O₂ atmospheres with high-purity (>99.999%) were employed in the experiment. All the reagents were used as received without further purification.

3.2. Preparation of Catalysts

Ti₄O₇ was prepared by the reduction of P25 TiO₂. A certain amount of P25 powder was placed in a quartz crucible with a lid, which was heated in a tube furnace under H₂ atmosphere. After two hours of thermal reduction at 1273 K, the dark blue-black powder was obtained, which would be used as the support for catalysts. The prepared Ti₄O₇ powder was ground and stored at room temperature in a capped glass bottle.

The catalyst Pt/Ti₄O₇ (Pt nanoparticles deposited on Ti₄O₇ support) was synthesized by the modified polyol method. A 7.72 mM Pt precursor solution was obtained by dissolving H₂PtCl₆·6H₂O in DEG. A certain amount of Ti₄O₇ support was dispersed in 130 mL DEG in a flask, which was sonicated for 1h to obtain a homogeneous black suspension. The black suspension was heated to 433 K. Half an hour later, the Pt precursor solution was then added to the black suspension drop by drop with a feed rate of two drops per second under stirring till 0.5 mM. After refluxing for 3 h, the suspension was cooled down to room temperature naturally. The catalysts were collected by centrifugation and washed with acetone, ethanol, and water successively. Pt/Ti₄O₇ with different Pt loading levels, Pt/Ti₄O₇-5, Pt/Ti₄O₇-10, and Pt/Ti₄O₇-20 could be obtained by varying the mass ratio of Pt to Pt/Ti₄O₇ by 0.05:1, 0.1:1, and 0.2:1.

3.3. Characterization

XRD was carried out at a scan rate of 20°/min on a high-resolution X-ray diffractometer (RIGAKU Smartlab, Tokyo, Japan) with Cu-K_α radiation. SEM (Hitachi S-4800, Tokyo, Japan), EDX, TEM (JEOL JEM-2100F, operate at 200 KV, Tokyo, Japan), and HRTEM were conducted to investigate the morphology. N₂ adsorption-desorption measurements were carried out on a surface area analyzer (Micromeritics ASAP 2460, Norcross, GA, America) at liquid N₂ temperature. The BET method was used to calculate the specific surface area of Ti₄O₇. The actual Pt loadings of synthesized catalysts were determined by ICP-AES on a PerkinElmer Optima 7300 DV instrument (Waltham, MA, America). The evaluation of the surface properties and electronic structure of the catalysts was carried out by XPS (Thermo Scientific K-Alpha, Waltham, MA, America) with an Al K_a X-ray source.

3.4. Electrochemical Measurements

Electrochemical measurements were conducted at room temperature in a typical three-electrode cell using a CHI 660e electrochemical workstation (CH Instruments Ins., China). A catalyst modified glassy carbon (GC) electrode (5 mm in diameter) and a Pt foil (1 cm²) electrode were used as the working electrode and counter electrode, respectively.

The active surface areas of catalysts were evaluated by the CV method in a 0.5 M H₂SO₄ aqueous solution using an Hg/Hg₂SO₄/saturated K₂SO₄ electrode as the reference electrode. ORR catalytic activities of the catalysts were evaluated by LSV test on a Metrohm ATU-rotating disk electrode in a 0.1 M KOH aqueous solution with an Hg/HgO/1 M KOH reference electrode. The electrolyte solution was bubbled with pure N₂ or O₂ for 30 min before electrochemical measurement.

Catalyst ink was prepared by dispersing 2.5 mg catalyst (Pt/Ti₄O₇) in 990 μL IPA under sonication for 30 min, followed by adding 10 μL 5 wt% Nafion^® solution to form the catalyst ink. A catalyst-loaded GC electrode was constructed by casting 20 μL catalyst ink on the surface of a GC electrode, and then dried in air at room temperature. Prior to use, the GC electrode was polished with a 50 nm α-Al₂O₃ slurry and ultrasonically cleaned in ethanol and water successively.

4. Conclusions

Single-phase micron-sized Ti₄O₇ powder was successfully obtained with high-temperature reduction by H₂. SEM results indicate that Ti₄O₇ shows a coral-like structure without micro/mesoporous material, resulting in small specific surface area (BET surface area: 2.16 m²·g⁻¹). Through the modified polyol method, the ICP-validated Pt/Ti₄O₇ catalysts with platinum contents at 3.48 wt%, 6.95 wt%, and 12.60 wt% were synthesized respectively. From the EDX and TEM results, Pt NPs were uniformly distributed on supports despite the verified loadings. The mean particle sizes calculated from the XRD and TEM are similar, 5 nm < particle size < 11 nm, and negatively correlated with the load. Pt/Ti₄O₇-5 possesses aggregated Pt nanorods, while Pt/Ti₄O₇-10 and Pt/Ti₄O₇-20 exhibit relatively fine dispersed Pt NPs. This can be explained by the intensity differences in the MSI of three catalysts. SMSI and sufficient unoccupied surfaces in the lowest Pt loading catalyst lead to the flattening phenomenon of Pt NPs. With the increase in Pt, the gap between particles decreased, and the MSI gradually becomes WMSI. This can be verified by the values of ΔBE in the Pt 4f spectrums. Only 30–50 percent Pt can perform characteristic adsorption properties and catalytic activities. This provides a reasonable explanation for the decrease in ECSA and the ORR activity of the Pt/Ti₄O₇ catalysts. By paying attention to synthesis method choice, controlling the particle size of Pt, expanding the specific surface area of the support, and increasing the Pt loading, adjusting the MSI is an indispensable step to promoting the application of the Pt/Ti₄O₇ catalyst.