Improving the Electrocatalytic Activities and CO Tolerance of Pt NPs by Incorporating TiO2 Nanocubes onto Carbon Supports

Designing efficient anode CO‐tolerant electrocatalysts is critical in low‐temperature fuel cell catalysts fueled either by H2/CO or alcohol. We demonstrate that the incorporation of TiO2 nanocubes (TiO2NCs) on Carbon Vulcan supports, followed by the synthesis of Pt NPs at their surface (Pt/TiO2NCs‐C material), led to improvements in performance towards the electrooxidation of carbon monoxide, ethanol, methanol, ethylene glycol, and glycerol in acidic media relative to the commercial Pt/C and Pt/TiO2‐C counterparts employing commercial TiO2. The nanocubes enabled changes in the electronic properties of Pt NPs while contributing to the bifunctional mechanism as compared to Pt/C and Pt/TiO2‐C with commercial TiO2. Fuel cell experiments fed with H2/CO steam showed that Pt/TiO2NCs‐C employing nanocubes was resistant to CO‐poisoning, yielding superior performance in operational conditions. The results reported herein have important implications for developing electrocatalysts with superior performances in PEMFCs.


Introduction
Direct Alcohol Fuel Cells (DAFC) have emerged as a promising alternative device to the H 2 -fueled Proton Exchange Membrane Fuel Cell (PEMFC) for power generation in portable applications. [1,2] The electrooxidation of short-chained organic molecules, such as alcohols, have been extensively studied over the past decades. [3][4][5][6][7][8] These molecules represent a viable option relative to the use of pure H 2 from the expensive process of water electrolysis. [9] Among all types of alcohol, ethanol presents important advantages as compared to H 2 and other low-molecular alcohols. These including cheap large-scale production from biomass, non-toxicity, easy storage in liquid tanks at ambient temperature, and high molecular energy density. [3,10,11] Nevertheless, finding efficient catalysts, which can balance the performance, selectivity, and stability tripod is the main drawback for the commercialization of both DAFCs and the so-called CO-contaminated H 2 (H 2 /CO-fed PEMFCs).
Carbon supported platinum nanoparticles (Pt NPs) are the most active materials for both anode and cathode electrodes in PEMFC or DAFCs devices. [12][13][14] However, incomplete alcohol fragments such as CO and CH 3 -species strongly adsorb onto the Pt surface slowing down the kinetics due to poisoning effects. [15,16] Specifically, CO strongly hinders the Pt active sites and limits the catalytic turnover, [17] thereby reducing the overall efficiency of practical systems supplied with alcohol or H 2 /CO steam by causing overpotential losses. [10] It is worth mentioning that the CO-poisoning of Pt-based catalysts in fuel cell devices is a common effect faced by either alcohol or contaminated H 2 produced by hydrocarbons steam reforming processes. Therefore, the development of more active and CO-tolerant catalysts in oxidizing alcohols and H 2 /CO steam represents an important challenge. In this context, many studies have focused on combining Pt with other CO-resistant elements like Ru, Sn, Cu, and Ni (generating bimetallic NPs) and on the control over the NPs shape in supported Pt and Pt-based nanomaterials. [16,[18][19][20][21][22][23][24][25][26] In addition to electrocatalytic activity, the stability of PEMFC electrodes is a significant hurdle to their application as vehicular and stationary/portable suppliers, especially when such devices are operated upon the cyclic regime. The deterioration of Pt/Cbased catalysts results either from the reduction in the number of Pt active sites (due to leaching and/or agglomeration) or the corrosion of carbon support. [27][28][29][30][31] For example, Pt NPs aggrega-tion originated from the dissolution/re-deposition of the Pt, and the motion of the Pt atoms onto the carbon surface is regarded as the main reason for the reduction of the catalyst activity. [29] Meanwhile, the electrochemical mechanism of the support corrosion in acid fuel cells shows that carbon electrolyzes water in the presence of Pt to yield CO and CO 2 as products. [32] The optimization of metal-support interactions represents a promising strategy to tackle and enhance the stability of Pt NPs in electrocatalytic applications. [33] This has been pursued, for example, by synthesizing shape-controlled carbon surfaces such as sheets, tubes, and horns, [34] by doping carbon with heteroatoms, [35][36][37] and by the addition of metal oxides in the carbon supports, including TiO 2 , CeO 2 , WO 3 . [38][39][40][41] Interestingly, metal oxides can also act as sources of oxygen, which is involved in promoting the carbon monoxide removal from the Pt NPs surface during the electrooxidation reactions as well as promoting changes in the metal-support interactions. Inspired by these guidelines, we report on the effect of the control over the shape of the metal oxide added to the carbon support in Pt-based electrocatalytic materials. Precisely, we targeted the addition of TiO 2 nanocubes (TiO 2 NCs) onto carbon and employed the resulting material as hybrid support for the deposition of Pt NPs, which was subsequently evaluated as an electrocatalyst towards carbon monoxide, methanol, ethanol, ethylene glycol, glycerol, and CO-contaminated H 2 electrooxidation. Surprisingly, we found that the use of TiO 2 NCs in the support enables higher electrocatalytic activities and stability as compared to the conventional Pt/C material and the Pt/TiO 2 -C electrocatalyst counterpart prepared by using commercial TiO 2 NPs.

Synthesis of TiO 2 NCs
The TiO 2 nanocubes were synthesized using a modified hydrothermal procedure. [42] Typically, a mixture containing 1 mL of [bmin][BF 4 ], 40 mL of acetic acid, and 2.5 mL of DI water was stirred in a beaker. After 10 minutes, 5 mL of Ti(OBu) 4 were added dropwise to the reaction mixture. This system was kept under stirring for 30 minutes. After 5 minutes of ultrasonic treatment, the mixture was put in a 220 mL Teflon-lined autoclave and heated up to 180°C. After 18 hours, the autoclave was allowed to cool to room temperature under ambient conditions. The solid product was isolated from the reaction mixture and washed with water and ethanol by successive rounds of centrifugation (5 minutes at 6500 rpm), removal of the supernatant, and re-dispersion. The TiO 2 NCs were then dried under a high vacuum at room temperature. [42] Synthesis of Pt NPs supported on TiO 2 NCs-C (Pt/TiO 2 NCs-C) and Pt/TiO 2 -C employing ethylene glycol (EG) reducing agent Here, three kinds of supports were evaluated: Pt NPs supported on Vulcan Carbon (from commercial Pt/C, Etek); Pt NPs supported on TiO 2 nanocubes and C (Pt/TiO 2 NCs-C); and Pt NPs supported on commercial TiO 2 and C (Pt/TiO 2 -C). Pt NPs supported on TiO 2 NCs-C and TiO 2 -C materials were prepared by a modified alcohol reduction method. [43] In a typical synthesis, a suspension containing EG/H 2 O (3 : 1 volume), TiO 2, and C supports (corresponding to 80 wt.%) was sonicated for 10 min. The prepared suspensions were heated at 150°C under vigorous stirring. After the temperature was reached, 2.12 mL of H 2 PtCl 6 · 6H 2 O solution (0.05 g mL À 1 ) was quickly added, and the resulting mixtures were kept under reflux for another 3 h. The nominal Pt loading was fixed in 20 wt.% in both materials. The products were isolated from the reaction mixture and washed with water by successive rounds of centrifugation, removal of the supernatant, and re-dispersion. Finally, the produced electrocatalysts were dried at 80°C for 2 h (Scheme S1, Supporting Information). The content of cubic TiO 2 NCs was varied from 5 to 40 wt.% and different Carbon amounts were employed (75-40 wt.%) to keep the platinum load in 20 %. For Pt/TiO 2 -C commercial preparation, it was chosen 10 % of TiO 2 and 70 % of Vulcan Carbon.

Characterization
The size and morphology of the prepared materials were characterized by Scanning Electron Microscopy (SEM, using a JEOL JSM 6330F operating at 5 kV) and Transmission Electron Microscopy (TEM, JEOL JEM 2100F operated at 200 kV). SEM sample was prepared by drop-casting an aqueous suspension of TiO 2 NCs material onto a silicon wafer, followed by drying at ambient temperature. For TEM analysis, a small amount of the catalyst was suspended in isopropyl alcohol and dropped in a TEM copper grid coated with colloidal film. Size was measured by end-to-end of particle extension. Composition analysis was conducted by X-ray dispersive spectroscopy (EDS), coupled to an SEM microscope JEOL JSN 6010LA, operated at 20 kV. X-ray diffraction analyses were performed using a Bruker D2 Phaser equipment with a Cu Kα source. The integration time was set to 1 minute using 0.05°step, ranging from 20°to 90°. The sample was placed on an acrylic holder and the slit used measured 1.0 mm. X-ray photoelectron spectra (XPS) were taken with a Thermo Scientific K-Alpha spectrometer using a 72 W monochromated Al K alpha source (hν = 1486.6 eV. The pass energy was set to 40 eV, with 0.1 eV step and 1 s per point of acquisition time. The sample holders were transferred to the XPS pre-chamber in an inert atmosphere and held under a vacuum. All XPS were referenced according to the adventitious C 1s peak (285 eV). All binding energies were referenced to the C 1s core level at 285 eV. Simulation of the experimental photopeaks was carried out using a mixed Gaussian/Lorentzian peak fit procedure according to the software supplied by CasaXPS. The semiquantitative analysis accounted for a nonlinear Shirley background subtraction.

Electrochemical measurements
All the electrochemical measurements were performed by 910 PSTAT (Metrohm) potentiostat. A three-electrodes cell apparatus with an ultrathin vitreous carbon layer previously polished was used as a working electrode (WE). For the WE preparation, a sonicated aliquot (5 μL) of a catalytic dispersion containing 0.9 mL of water, 0.1 mL of isopropyl alcohol, 20 μL of Nafion solution and was dropped in the tip of the vitreous carbon layer and dried (totalizing about 2 μg of Pt on the WE). Sulfuric acid (0.5 mol L À 1 ) was employed as a supporting electrolyte. Cyclic voltammograms (CV) were recorded after N 2 gas was bubbled for 30 min into the electrolyte. CV recordings had a scan rate of 0.05 V s À 1 at ambient temperature. Cyclic voltammetry experiments of methanol, ethanol, ethylene glycol and glycerol electro-oxidation reactions were conducted in 1 mol L À 1 . For the CO-stripping experiment, the WE was submitted to 0.35 V polarization and carbon monoxide was bubbled for 500 s into the electrolyte, followed by 800 s of N 2 . For further comparison purposes, the electroactive area (ECSA) was calculated based on the hydrogen desorption profile through the voltammetry of the CO-stripping experiment data by Equation [1].
where Q H is the hydrogen desorption charge from 0.05-0.4 V, considering the electrical double layer contribution; [Pt] is the platinum loading and q 0 H is the charge required to oxidize a hydrogen monolayer adsorbed over platinum (210 μC cm À 2 ).

Fuel Cell performance
For Membrane Electrode Assembly (MEA) preparation, Pt/C and Pt/ TiO 2 NCs-C were evaluated as anode and Pt/C as the cathode, both electrodes loaded with 0.6 mg Pt cm À 2 . The 5 cm 2 electrodes were hot-pressed to a pre-treated Nafion 115 membrane (Dupont) at 125°C for 10 min. All prepared MEAs were inserted in a single cell, operating at 80°C, fueled with H 2 and H 2 /CO gases (100 ppm CO -300 mL min À 1 ) on the anodic side and O 2 supplying the cathodic side (200 mL min À 1 ). No pressure was used in the cathode. Polarization curves were employed to evaluate the catalyst's electrical performance. Figure 1a depicts the SEM images for the TiO 2 NCs that were employed as substrates for the deposition of Pt at their surface. It can be observed that the TiO 2 nanocubes are well defined, relatively monodisperse, and their edge lengths corresponded to 72 � 9 nm. The nanocubes were then mixed with carbon (20 mg of TiO 2 NCs and 140 mg of C) to generate our support material (TiO 2 NCs-C), which was employed as physical templates for the deposition of Pt at their surface. Figure 1b and Figure 1c show HRTEM images for the Pt/TiO 2 NCs-C material obtained after Pt deposition. The 0.36 nm lattice spacings characteristic of TiO 2 anatase can be detected, corresponding to the growth along the (100) crystallographic direction and the exposure of (100) surface facets in agreement with the nano- cube morphology. [44] It can also be observed that the Pt NPs were deposited at the surface of the TiO 2 nanocubes had a spherical shape and narrow particle size distribution. The histogram of the particle size distribution (Figure 1d) indicated that the Pt NPs size corresponded to 3.7 � 0.5 nm. The EDS spectra revealed that the Pt content corresponded to 20 wt.%, in agreement with the nominal Pt loading ( Figure S1 and Table S1). The presence of Vulcan Carbon at the surface of the TiO 2 NCs and Pt nanoparticles can be observed from Figure S3.

Results and discussion
We also prepared similar Pt/TiO 2 -C materials in which commercial TiO 2 , instead of nanocubes, was employed to generate the TiO 2 -C support and therefore served as a reference to evaluate the importance of the TiO 2 shape control over the synthesis and performance of the produced electrocatalysts. HRTEM images for the Pt/C and Pt/TiO 2 -C materials in which commercial TiO 2 (instead of nanocubes) was employed as reference support are shown in Figure S2a and S2b, respectively. For Pt/C, the Pt NPs were spherical and uniformly dispersed over the support. Their diameter corresponded to 3.2 � 0.5 nm ( Figure S2c). For Pt/TiO 2 -C material in which commercial TiO 2 was employed as support, Pt NPs 4.0 � 0.6 nm in size were obtained ( Figure S2d). However, they were not uniformly dispersed at the surface of the TiO 2 , and significant agglomeration could be detected. Therefore, our data indicate that the utilization of TiO 2 NCs was essential to achieve better dispersion of Pt NPs at the TiO 2 surface (relative to the commercial counterpart). Figure 2a shows the XRD patterns of Pt/TiO 2 NCs-C, Pt/TiO 2 -C, and Pt/C. The diffractograms show characteristic peaks for fcc Pt in all cases at 2theta = 39°, 46°, 67°, 81°, 86°. A broad peak in 25°can be assigned to the presence of amorphous carbon, while characteristic peaks assigned to the anatase structure of TiO 2 (JCPDS, no. 21-1272) could be detected in the materials containing TiO 2 . The Pt crystallite size estimated by Scherrer equation (using the peak at 2theta = 39°) corresponded to about 4 nm, in agreement with the HRTEM results. Figure 2b depicts the Pt 4f core-level XPS for the Pt/ TiO 2 NCs-C, Pt/TiO 2 -C, and the Pt/C (top to bottom traces, respectively). The binding energy (BE) values, full width at half maximum, and Pt surface percentages are given in Table 1. For   [38,45,46] It has been reported that the interfacial effect between Pt and TiO 2 is dependent on the nature of the TiO 2 surface facets, in which (100) surface facets have been found to provide stronger charge transfer values in Pt/TiO 2 electrocatalysts having controlled shapes. [47] In this case, it is plausible that the TiO 2 NCs present a larger fraction of (100) surface facets relative to the commercial TiO 2 NPs. Therefore, this increased fraction of (100) surface facets enables a higher charge transfer from TiO 2 to Pt NPs and thus stronger metal-support interactions relative to the commercial NPs.
To gain insights on the oxidation states of the Pt species over the surface of the catalysts, we examined the full width at half maximum and the atomic percentages of each Pt surface species obtained from the XPS data. The full width at half maximum values is higher for the Pt/C sample suggesting that more oxidized Pt (II) species are present on the surface of this material contributing to 53 % of the registered signal. Interestingly, it can be detected that the Pt/TiO 2 NCs-C sample displayed the highest percentage of Pt 0 relative to Pt II as compared to the other samples. The Pt 0 content in this sample was increased by approximately 22 % as compared to the Pt/C. These observations agree with an increased electron density at Pt as a result of enhanced metal-support interactions with the TiO 2 nanocubes in the support. [45,46] It is important to note that the precise estimation of the percentage of Pt loaded on TiO2 and C is challenging. Therefore, in our study, we made the approximation that Pt NPs are deposited over both TiO 2 and C during the synthesis procedure without any preferential deposition of one of the supports. Figure 3 displays the CO-stripping profile for the electrooxidation features of a one-absorbed carbon monoxide layer on Pt/TiO 2 -C (blue trace), Pt/TiO 2 NCs-C (red trace), and the Pt/C (black trace). It is established that Pt has a structural-sensitive interaction with carbon monoxide, thus this molecule has been employed as a molecular probe to study the properties of Pt catalytic surfaces. [48] As can be observed from Figure 3, the COstripping features were different and thus sensitive to the nature of the Pt-based catalysts, indicating the presence and shape of TiO 2 support influenced the Pt surface properties. In the voltammetry curves for Pt/TiO 2 NCs-C and Pt/TiO 2 -C, the main CO electrooxidation peaks were negatively-shifted (ca. 0.86 V vs RHE) relative to Pt/C (bottom trace, 0.88 V vs RHE). [49] This indicates that the CO removal from the Pt surface is facilitated in the materials containing TiO 2 . Moreover, differently from Pt/C, both TiO 2 -based materials exhibited a split-peak profile. Particularly, the CO-stripping voltammograms revealed the lower onset potential for Pt/TiO 2 NCs-C (0.56 V vs RHE) and Pt/TiO 2 -C (0.58 V vs RHE) relative to Pt/C (0.75 V vs RHE). Interestingly, for Pt/TiO 2 NCs-C, the CO-stripping profile showed a visible pre-peak (highlighted in the inset) centered at ca. 0.68 V vs RHE, which could be a result of the strong metalsupport interaction between the Pt NPs and the TiO 2 nanocubes (this pre-peak was not observed for the samples containing commercial TiO 2 , the same region is highlighted as in the corresponding inset). This strong metal-support interaction contributes to the weakening of the Pt-CO interaction. [50][51][52] The calculated electroactive surface areas (ECSA) after the COstripping voltammograms corresponded to 72, 64, and 76 m 2 g Pt À 1 for the Pt/TiO 2 NCs-C, Pt/TiO 2 -C, and Pt/C, respectively, indicating the values were relatively close for all catalysts in agreement with the similar Pt NPs sizes in the samples.
To reveal the influence of TiO 2 NCs loading on the performance of the catalyst towards ethanol electrooxidation reaction (EOR), we synthesized Pt/TiO 2 NCs-C with different TiO 2 NCs loadings. Figure S4 shows the maximum current density obtained from the forward scan in the corresponding CVs for Pt/TiO 2 NCs-C as a function of the TiO 2 NCs loading. Here, the Pt loading was maintained at 20 wt.% in all materials, while the composition of the support was varied. It can be observed that a volcano type relationship between maximum current density and TiO 2 NCs loading was observed, with the peak in current density centered at 10 wt.% in mass of TiO 2 NCs. Therefore, the TiO 2 NCs loading was kept at 10 wt.% for all subsequent studies. The decrease in the current density values at higher TiO 2 NCs loadings may be assigned to the decrease in the conductivity of the support. It can also be observed that the maximum current density for the Pt/TiO 2 NCs-C (10 wt.% TiO 2 NCs loading) was higher relative to the Pt/TiO 2 -C sample (408 vs. 353 A g Pt À 1 ). Such a difference in catalytical performance demonstrates that the proper choice of support morphology can lead to an improvement in the DAFC activity. [8] We then focused on the investigation of the electrocatalytic activity of the Pt/TiO 2 NCs-C and the state-of-art Pt/C towards the EOR. The EOR activity was evaluated by cyclic voltammetry, recorded in 0.5 mol L À 1 of H 2 SO 4 electrolyte containing 1 mol L À 1 of ethanol. It can be observed from Figure 4a that Pt/TiO 2 NCs-C showed higher activity than Pt/C. The maximum current density of the forward scan was 640 and 475 μA cm À 2 for Pt/TiO 2 NCs-C and Pt/C, respectively. Moreover, the onset potential (seen in the inset) for Pt/TiO 2 NCs-C was lower than on the Pt/C material (0.41 vs 0.49 V, respectively), demonstrating the promotion effect of TiO 2 NCs on the EOR. Figure 4b exhibits the maximum current density of the forward scan normalized by Pt mass (top panel) and specific area (bottom panel). It can be observed that the maximum current density by Pt load on Pt/TiO 2 NCs -C was also superior, being around 15 % greater than on Pt/C. These results indicate that Pt/TiO 2 NCs-C has a higher catalytic activity towards EOR than Pt/C, in agreement with previous reports. [38] In terms of stability, the performance of the Pt/TiO 2 NCs-C and Pt/C is shown in Figure 4c. This Figure shows the maximum current density of the forward scan collected by changing the potential exhaustively by up to 70 consecutive voltammogram cycles from 0.05 to 1.0 V vs RHE. This limited window of potential was chosen to preserve the morphologic characteristics of the materials. Both electrocatalysts exhibited a slight drop in their current density. However, Pt/TiO 2 NCs-C maintained higher current densities throughout the experiment. Considering the evaluation of medium-term performance, Pt/TiO 2 NCs-C lost about 5 % of its maximum current density after the 70 th CV cycle, while the traditional Pt/C decreased its performance by 10 %. Such a decrease in performance could be attributed to the poisoning effect of the incomplete ethanol fragments such as CO and/or to the loss of Pt active sites. To evaluate the possibility of Pt agglomeration and/or loss of Pt electroactive sites, we monitored the region of hydrogen desorption charge after 70 CV cycles with H 2 SO 4 (0.5 mol L À 1 ) as illustrated in Figure 4c inset. In the case of Pt/TiO 2 NCs-C, about an 18 % decrease in ECSA value was detected after 70 cycles, while for Pt/C, this value corresponded to 22 %. In the case of Pt/TiO 2 -C, the decrease in ECSA corresponded to 21 %. Therefore, these data indicate that Pt/TiO 2 NCs-C, in addition to better performances, also displays better stability. [53] The improved mediumterm stability of the material containing TiO 2 NCs may be related to the more poisoning-resistant characteristics and/or to prevent the Pt electroactive sites loss as well as the fact that Vulcan Carbon is susceptible to support corrosion, thus leading to the detachment and agglomerating of Pt NPs (the formation of oxidizing species such PtO and PtOH can also contribute to the carbon corrosion). [28] In this case, we believe that the utilization of TiO 2 and the strong metal-support interaction between Pt NPs and the TiO 2 NCs support can contribute to mitigating this effect. Figure S5 shows the activity towards methanol, ethylene glycol, and glycerol electrooxidation reaction catalyzed by Pt/C and Pt/TiO 2 NCs-C, respectively. For all three alcohols, Pt/ TiO 2 NCs-C exhibited higher current densities than Pt/C. The maximum current densities for Pt/TiO 2 NCs-C were 2.05, 1.26, and 1.47 folds higher for the electrooxidation of methanol, ethylene glycol, and glycerol, respectively. Besides, the onset potential for the Pt/TiO 2 NCs-C electrocatalyst was lowered compared with those of Pt/C. Interestingly, we found that the TiO 2 NCs can also be added to commercial electrocatalysts to improve their performances. We could further improve the electrocatalytic activity of the commercial PtSn/C by adding 10 wt.% of TiO 2 NCs simply through mixing. The evaluation EOR towards PtSn/C and PtSn/ C + TiO 2 NCs 10 % is shown in Figure S6. PtSn-based materials are known as the best catalysts for the EOR. [54,55] The maximum current densities described in Figure S6A were obtained during 70 CV cycles. Along with the test, higher current density values were obtained when TiO 2 NCs were added. From the first CV cycle, it is possible to verify that the current density was about 565 A g Pt À 1 and 550 A g Pt À 1 for PtSn/C + TiO 2 NCs 10% and PtSn/ C, respectively. Over the CV tests, the activity difference has begun to increase and after 70 CV cycles, the current density loss for PtSn/C (~100 A g Pt À 1 ) doubled as compared to PtSn/C + TiO 2 NCs 10% material (~50 A g Pt À 1 ). The relative activity loss, given in percentage, is shown in Figure S6B-C. The current density loss over the 70 th CV cycle for PtSn/C + TiO 2 NCs 10% was approximately 10 %, while for PtSn/C was about 18 %. This indicates that the addition of a small amount of TiO 2 NCs to Pt/C and PtSn/C would help to preserve oxidative characteristics even with the extensive potential cycling. Another potential explanation for the EOR enhancement derivates from a more complete ethanol electrooxidation to CO 2 . This could be understood by the fact that the hydroxyl species from TiO 2 NCs in Pt/ TiO 2 NCs-C and PtSn/C + TiO 2 NCs 10% electrocatalysts improve the conversion of CO into CO 2 through the bifunctional mechanism. Similar results were also reported by Colpan et al [56] by evaluating PtRu/C-TiO 2 catalyst, in which Ru assumes the same role as Sn co-catalyst.
Finally, we employed the developed electrocatalysts as applied anodes for a PEMFC device, testing their electrical performance by polarization and power density curves (V vs jnormalized per electrode area) for the electrocatalytic activity of CO-contaminated H 2 in operational fuel cell conditions (Figure 5). Here, the Pt/TiO 2 NCs-C and Pt/C were compared. The PEMFC anode devices were both fed with an H 2 /CO mixture (100 ppm CO, Figure 5a) and pure hydrogen (Figure 5b, inset) at practical operating temperature (80°C). Oxygen was chosen as a cathodic oxidizing gas, catalyzed by commercial Pt/C. PEMFC containing both the Pt/C and Pt/TiO 2 NCs-C anodes presented similar Open Circuit Voltages (OCV), at approximately  As can be observed in the graphical inset, the maximum current densities measured were about 2.3 A cm À 2 . As expected, such electrical performances decreased when both anodes were supplied with CO-contaminated hydrogen. More importantly, higher performances were achieved for the Pt/TiO 2 NCs-C material. The maximum current density drained of Pt/TiO 2 NCs-C was about 1.2 A cm À 2 , while for Pt/C catalyst this value was only 0.9 A cm À 2 in H 2 /CO/O 2 fueled tests. Power density curves (expressed by open squares) were calculated based on polarization curves. The maximum power density reached by Pt/ TiO 2 NCs-C was 130 mW cm À 2 . This value is 65 % higher than the obtained by Pt/C electrocatalyst. It is noteworthy that both materials have the same Pt mass, hence both PEMFC electrodes are equally loaded in terms of the active metal. Then, the increased electrical performance of Pt/TiO 2 NCs-C was related to the presence of TiO 2 nanocubes, which enables the oxidation of CO at lower overpotentials in comparison to the Pt/C. This tolerance to CO poisoning effect on Pt/TiO 2 NCs-C catalyst is further evidenced by comparing the single-cell anode overpotential (ŋ CO ) as a function of the measured PEMFC current density. Figure 5b demonstrate, as also observed in Figure 3, that CO is oxidized at lower potential values on Pt/TiO 2 NCs-C relative Pt/C. At the PEMFC operating temperature conditions, the gap of potential between Pt/TiO 2 NCs-C and Pt/C observed in Figure 5b was at about 0.08V, in agreement with CO stripping tests performed at ambient temperature, which further suggests that the catalytic activity difference is primarily structural/ compositional.

Conclusions
We reported on the development of hybrid support for the deposition of Pt NPs comprising of TiO 2 nanocubes and carbon (TiO 2 NCs-C). Pt/TiO 2 NCs-C electrocatalyst was synthesized and investigated towards the electrooxidation of carbon monoxide and alcohol. The performances were compared relative to the use of commercial TiO 2 as well as the use of only carbon (Pt/ TiO 2 -C and Pt/C). In these materials, the Pt NPs size and distribution over the supports were similar. However, the surface characteristics were remarkably changed in terms of Pt-TiO 2 interaction. Electrochemical investigation of the COstripping clearly showed that CO interacts differently with Pt/C, Pt/TiO 2 NCs-C, and Pt/TiO 2 -C. In particular, CO oxidation on Pt/ TiO 2 NCs-C was greatly facilitated in lower onset potentials, which was due to the strong metal-support interaction and/or bifunctional mechanism provided by Pt nanoparticles and TiO 2 nanocubes. The electrocatalytic activity and the stability of Pt/ TiO 2 NCs-C for ethanol electrooxidation were also improved relative to the other materials. The medium-term evaluation showed that the loss of Pt active sites phenomena could be suppressed considering the ECSA examination in Pt/TiO 2 NCs-C. It is important to note that this system may be further optimized by controlling the size and nature of the Pt species at the TiO 2 surface (NPs, clusters, and single atoms), the shape of Pt NPs (deposition of cubes, octahedra, or shapes enabling the exposure of high index facets), the size of the TiO 2 nanocubes, and the amount of surface defects such as oxygen vacancies. We believe our data indicates that the shape control over the support material can play an important role and can be applied as a promising approach to optimizing electrocatalytic activity and stability in the design of PEMFC and DAFC anodes.