Recently, direct alcohol fuel cells (DAFCs) have drawn increasing interest for high energy density, high efficiency and environmental friendliness and become promising power sources of portable electronic devices [1, 2]. Intensive research has been focused on direct methanol fuel cell for the past few decades [3, 4]. However, due to toxicity and volatility of methanol, direct ethanol fuel cell (DEFC) has been considered a more appealing candidate because of the intrinsic advantages of ethanol such as relatively high boiling point (78 ℃) and non-toxicity [5]. Furthermore, as a renewable biofuel, ethanol could be manufactured in a large scale from fermentation of agricultural products or other biomass resources.

Despite the abovementioned advantages, the development of DEFC has been limited owing to the sluggish electro-oxidation process of ethanol [6]. The complete electro-oxidation ethanol to CO₂ refers to a complex multi-step process involving the C–C bond cleavage and dehydrogenation, which is a 12-electron transfer pathway [7]. Pt and Pt-based electro-catalysts have been found to the most suitable materials for ethanol oxidation reaction (EOR) [8]. Unfortunately, Pt active center is readily poisoned by the main intermediate CO, which originates from the dissociation of ethanol and other organic species thus reduces the activity and long-term stability of Pt catalysts [9]. This detrimental influence has become the bottleneck to realize commercialization for Pt-based electrocatalysts. One useful method to alleviate the poisoning effect is to introduce a second metal (Sn [10], Ru [11], Rh [12], Cu [13], Ni [14], etc) to form bimetallic catalysts. Pt-Sn co-catalysts have been intensively studied due to their excellent performances toward ethanol oxidation [15, 16]. According to related reports, Sn atom adjacent to Pt favors the dissociation of H₂O adsorbed in its surface and forming active OH_ads species, which could facilitate the oxidation of CO intermediate on the surface of Pt and release the active sites [17, 18].

However, most studies focused on the above geometric effect from second metal, while the electronic effect in Platinum was rarely discussed. Zhu et al. [19] investigated the effect of degree of alloying in Pt/Sn/C catalysts and elucidated that the strengthened electronic effect between Pt-Sn alloys promote the entire EOR activity. Qu et al. [20] found that the incorporation of TiO₂ to Pt-based electrocatalysts changes the electronic properties of Pt and weakens the interaction between metal and poisoning CO intermediate, thus the catalytic performance towards MOR was greatly improved. Recently, Camacho and his coworkers [21, 22] have reported that the enhancement of catalytic behavior and durability both in oxygen reduction reaction and methanol oxidation reaction (MOR) was mainly due to the electronic changes of Pt active centers caused from strong interaction between Pt and M (M = Sn, Ti) in metal oxide (SnO_x, TiO₂). Among the metal oxides, tin oxide (SnO_x) has been widely used as co-catalyst in DAFCs due to its enhancement in terms of activity and stability in acidic electrolyte [23-26]. Thus, having a deep understanding of electronic interaction between Pt and SnO_x is of great significance to improve the electrocatalytic behavior of ethanol.

In the present work, Pt/Graphene and Pt/SnO_x/Graphene electrocatalysts were synthesized by using a facile ethylene glycol (EG) method. For comparison, conventional Pt/XC-72 catalyst was also synthesized using the same method. A series of physical characterization and electrochemical tests were employed to evaluate the structural and electrocatalytic performances of the obtained catalysts. The Pt/SnO_x heterostructure on graphene shows a remarkable enhancement in both the activity and stability towards the EOR. We find a significant electronic influence of SnO_x and graphene in Pt active sites for ethanol oxidation.

The Carbon Black (Vulcan XC-72) used in this work was obtained from Cabot Corporation and the graphene was prepared by a thermal expansion method reported in our previous work [27]. All the chemicals (H₂PtCl₆, SnCl₂·2H₂O, etc) used were supplied from sigma-aldrich and utilized directly without any purifications. Ultrapure water (MilliQ, Millipore, USA) was used throughout the whole experimental operations.

The Pt/C, Pt/Graphene and Pt/SnO_x/Graphene catalysts were synthesized based on a facile ethylene glycol (EG) method, which was described in detail in our previous work [28]. The synthesis procedure of Pt/SnO_x/Graphene catalysts is as follows: Firstly, appropriate amount of SnCl₂·2H₂O was dissolved in 200 mL EG with traced amount of water. Then the mixture was reflexed at 190 ℃ for 3 h with continuous agitation to form a uniform pale yellow colloid solution. Undergoing the reflux process, Sn (Ⅱ) was gradually converted into SnO_x nanoparticles through a series of complex reactions [29]. After the resulting colloid solution cooled down to room temperature, calculated amount of H₂PtCl₆-EG solution and graphene were added and sonicated for 30 min. Afterwards, NaOH-EG solution (1 mol/L) was utilized to adjust the pH of the mixture to about 12. The mixture was then heated to 140 ℃ and kept for 4 h at this temperature to ensure that the compete Pt reduction was realized. After that, HNO₃ solution (1 mol/L) was used to adjust the mixture pH to about 3. After vigorously stirring for 48 h, the obtained sample was ultrafiltrated and washed with deionized water until the Cl^‒ was not detected, and then dried at 90 ℃ for 4 h in vacuum. The Pt/C and Pt/Graphene catalysts were also synthesized using the abovementioned procedure. The nominal loading amount for all the as-prepared catalysts is 20 wt%.

The X-ray diffraction (XRD) measurements was conducted to investigate the crystallopraphic structure of the catalysts using a D8 Advance diffractometer (Bruker) with a Cu K_α radiation (λ = 1.5406 Å). The diffractometer was operated at 40 kV and 30 mA under atmospheric pressure over a range of 15°–90° with a scan rate of 5°/min and the angular resolution was kept at 0.02° in the 2θ scans. The morphologies and particle size distribution were characterized in a field emission transmission electron microscope (FE-TEM, JEOL 2100, the accelerating voltage is 200 kV) equipped with an X-ray dispersive spectroscopy (EDS) apparatus. For the TEM measurements, an amount of catalyst sample was dispersed in ethanol, and sonicated for 10 min. An aliquot of suspension was dropped in a Cu grid coating with carbon film and dried for 15 min to allow the solvent evaporate. The surface compositions and electronic information of the catalysts were acquired from X-ray photoelectron spectroscopy (XPS) measurements using a photoelectron spectrometer (PHI 5700) with a monochromatic Al K_α X-ray (1486.6eV) source and operating at a power of 250 W. Before each analysis, the sample was dried and deposited in a carbon tape. A survey spectrum was firstly collected before the data acquisition of high resolution spectrum of Sn and Pt. The Charge correction for all the data was referenced to C 1s peak with a binding energies of 284.6 eV. The deconvolution and processing of the spectrum was carried out employing a XPSPEAK 41 software and the Shirley method was utilized to correct the background of the spectra.

The working electrode was prepared through a general method, which has been presented in detail in our previous work [29]. Typically, a Glassy Carbon electrode (GC) with a geometric surface around 0.07 cm² was pre-polished using alumina powder with size of 0.3 and 0.05 μm, respectively, until a mirror-like surface was obtained. Subsequently, the pre-polished GC was sonicated for 10 min and washed with deionized water and dried. The as-prepared sample was dispersed in isopropanol (IPA) and ultrasonicated to form a uniform catalyst slurry with a concentration of 2 mg/mL. A total of 7.5 μL (15 μg catalyst) of above slurry was dropped on the surface of the pre-polished GC. After drying at ambient temperature, 5 μL of 0.05 wt% Nafion-Ethanol solution was coated on the surface of catalyst layer to form a Nafion membrane, which serves as a protection layer as well as a proton conductor. The well-prepared GC was employed as working electrode for each electrochemical measurement after drying at room temperature overnight.

The electrochemical measurements were carried out at ambient temperature in a standard three-electrode system comprising of a working electrode (the as-prepared GC), a reference electrode (Hg/Hg₂SO₄ electrode) and a counter electrode (Pt wire electrode) controlled by a CHI 760E electrochemical workstation. Cyclic voltammograms (CVs) were recorded in Ar-purged 0.5 mol/L H₂SO₄ within a potential range from 0.05 to 1.10 V at a scan rate of 50 mV/s. The activity measurements of the as-synthesized catalysts towards EOR was conducted in 1 mol/L ethanol and 0.5 mol/L H₂SO₄ electrolyte with the same voltage range and sweep rate as CVs. Before each test, the Ar was bubbled into the electrolyte for 30 min to remove the air. For the CO-stripping experiments, the procedure was performed in the following way: CO was firstly adsorbed on the working electrode in Ar-purged 0.5 mol/L H₂SO₄ at a constant voltage of 0.05 V for 10 min to reach saturation, and then the excess CO dissolved in electrolyte was eliminated via Ar bubbling for another 30 min. Subsequently, the CO-stripping voltammograms were obtained within the potential range of 0.05 to 1.20 V at 50 mV/s. The stability of the as-prepared electrocatalysts were evaluated by chronoamperometry (CA) at 0.5 V for 9000 s in 1 mol/L ethanol and 0.5 mol/L H₂SO₄ electrolyte. Electrocatalytic activity was all presented in terms of current per mass of Pt and all the potentials in this work are versus reversible hydrogen electrode (RHE).

The morphologies and sizes of the nanoparticles have essential influence on the activity of the catalysts. We firstly prepared different Pt-based catalysts using EG method, and then the size distribution of Pt nanoparticles along with their electronic interaction with support and metal oxide were carefully investigated. As shown in TEM images in Fig. 1 and Fig. S1, the Pt and SnO_x spherical particles were well-dispersed on the surface of the carbonaceous support (carbon black and graphene), which can be attributed to the advantages of the EG method [19, 30]. More than 200 different particles on TEM images were randomly counted to calculate the arithmetic average particle diameter and the size distribution profiles of Pt and SnO_x were presented in Fig. S2. The mean size for each catalyst was also reported in Table 1. As shown in Fig. S2 and Table 1, all the samples have an approximate narrow size distribution of Pt nanoparticles in the range of 1-3 nm and similar average size. In spite of slight agglomerate of SnO_x particles in Pt/SnO_x/Graphene catalyst, most of particles are uniformly dispersed on the surface of graphene with a mean size of 3.97 nm.

HR-TEM images of Pt/SnO_x/Graphene in Fig. 1(c) and Fig. S1(a) clearly show the configuration of Pt and SnO_x in nanocomposite: Pt and SnO_x nanoparticles contact with each other partially at various regions, which are well dispersive on the graphene support. There are two kinds of lattice fringes with interplanar spacing of 0.225 and 0.335 nm can be recognized clearly from Fig. 2(c), which correspond to the (111) plane of Pt (fcc) crystal and (110) plane of SnO_x with tetragonal rutile phase, respectively, showing good consistency with the XRD results below. The high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) image (Fig. 1(d)) shows the distribution of Pt and SnO_x nanoparticles on graphene support, where the brighter spots and darker spots stands for the Pt and SnO_x particles, respectively, and the black background denotes the graphene support. As can be seen from this micrograph, Pt nanoparticles stay close to SnO_x nanoparticles and both of them are uniformly deposited on the graphene surface, which further confirming the abovementioned heterostructure.

Fig. 2(a) displays the XRD results of the obtained samples. The first broad diffraction peak for the Pt/C at around 25° can be assigned to the (002) plane of the hexagonal structure of the carbon, whereas for the other two catalysts the peaks at 2θ ~26° correspond to the (002) facets of the graphene. The peaks for each sample located at 39.7°, 46.2°, 67.5° and 81.3° are associated to the (111), (200), (220) and (311) planes of Pt with a face centered cubic (fcc) structure (JCPDS 04-0802) and the broad peaks can be attributed to the small crystallize size. On the other hand, the Pt/SnO_x/Graphene sample exhibits two diffraction peaks at 33.8° and 51.7°, which can be assigned to the characteristic peaks of (110) and (211) facets of tetragonal rutile SnO_x (JCPDS 41-1445), respectively. In addition, a pronounced shift towards the lower angle values can be observed for the diffraction peaks of Pt (220) and Pt (311) facets for the Pt/SnO_x/Graphene sample, which may be due to the strong metal-metal oxide electronic interaction. The Gaussian fitted Pt (111) and SnO_x (110) peaks were used to calculate the average crystalline size of Pt and SnO_x nanoparticles, respectively. As shown in Table 1, all the samples have similar mean size of Pt nanoparticles.

To confirm whether the electronic structure of Pt has been tuned by carbon support and metal oxide, XPS characterization was carried out and the results were shown in Fig. 2(b) and (c). Fig. 2(b) displays the survey spectra, where O 1s, Pt 4d, C 1s and Pt 4f peaks can be clearly indentified and no other impurities were detected. The high-resolution Pt 4f and Sn 3d peaks are presented in Fig. 2(c). The binding energies (BE) for Pt 4f and Sn 3d components of the prepared catalysts are summarized in Table S1. As displayed in Table S1, the binding energies of Pt 4f_7/2 located at around 71‒72 eV can be assigned to zero-valence state Pt, which has been found to the predominant species in the as-systhesized catalysts [31]. According to previous research [32], metallic Pt is believed to the active sites for ethanol oxidation. The binding energies at 495.40 and 486.82 eV are assigned to Sn 3d_3/2 and Sn 3d_5/2 doublets of Sn(Ⅳ) and no peaks in other chemical states were found, indicating that tin species mainly exist as oxide (SnO_x). This could be attributed to the synthesis precedure and shows a good agreement with the XRD results. Furthermore, we compared the binding energies of Pt 4f peaks for different catalysts and found that the values of BE decrease in the following sequence: Pt/C (71.68 eV) > Pt/Graphene (71.53 eV) > Pt/SnO_x/Graphene (71.31 eV). It is noteworthy that the BE for Pt/SnO_x/Graphene shifts negatively by 0.37 eV with respect the Pt/C catalysts, which can be ascribed to the electron transfer from graphene and SnO_x to Pt, demonstrating electronic modification of support and metal oxide. Our previous work has confirmed that graphene has stronger electron donation ability with respect to XC-72 carbon black [29]. Lewera et al. [33, 34] studied the interaction between Pt and metal oxide using XPS and attribute this phenomenon (negative shift) to the strong metal-support interaction (SMSI) as well as the formation of Pt-M (from metal oxide) alloy, which can also be accounted for our case.

We have synthesized Pt nanocatalysts with the same morphology but different electronic structure. Next, electrochemical measurements were performed to investigate the effect of Pt electronic structure changes on ethanol oxidation. Firstly, the steady-state cyclic voltammetry (CV) results of Pt/SnO_x/Graphene, Pt/Graphene and Pt/XC-72 working electrodes were depicted in Fig. 3(a). As presented in Fig. 3(a), all the CV curves show similar electrochemical features, obvious hydrogen adsorption/desorption reaction (H_upd) peaks over the range of 0.05–0.3 V (H region) can be observed, where the peaks with various intensities are associated to different facets of polycrystalline Pt [35, 36]. In the double layer region (0.3–0.75 V), profiles exhibit much capacitive current (both presented in Fig. 3(a) and Fig. S3) for Pt/Graphene and Pt/SnO_x/Graphene than that of Pt/C, which can be ascribed to high specific surface area of graphene [37]. The electrochemical surface area (ECSA) per unit mass can be determined by integrating the electrical charge (with the subtraction of the background) in hydrogen adsorption/desorption region with an assumption that the charge accumulated in polycrystalline Pt for monolayer hydrogen adsorption is 210 μC/cm² [28, 38, 39]. The obtained ECSA (Table 2) in this work for Pt/SnO_x/Graphene, Pt/Graphene and Pt/C are 53, 57 and 55 m²/g, respectively, demonstrating a similar Pt exposed active actives for all the catalysts.

Fig. 3(b) shows the electrocatalytical acitivity for ethanol oxidation on Pt/SnO_x/Graphene, Pt/Graphene and Pt/XC-72 samples and all the electrochemical activity parameters for EOR are listed in Table 2. The forward anodic peaks reflect the ethanol oxidation capability of the electrocatalysts. The mass peak current densities for Pt/SnO_x/Graphene and Pt/Graphene are 0.79 and 0.51 A/mg-Pt, respectively, which is 2.82 and 1.82 times that of Pt/XC-72 counterpart (0.28 A/mg-Pt), indicating a superior EOR activity compared to the Pt/XC-72 counterpart [40]. We compared this activity with the literature in terms of mass peak current densities. As shown in Table S2, our Pt/SnO_x/Graphene exhibits the highest activity among the reported catalysts under the operation conditions. Moreover, the forward anodic peak potential for Pt/SnO_x/Graphene exhibits a relatively negative shift (0.89 V), about 30 to 40 mV lower than that of Pt/Graphene and Pt/C, which demonstrates a reduced overpotential and implys that the electro-oxidation process on heterostructure Pt/SnO_x is more readily realized than that on Pt [41]. The backward anodic peaks are related to the oxidation reaction of the intermediates on the surface of the catalysts [42], similar trends could also be found for the backward mass peak current densities (I_b) and peak potential (Fig. S4). Silva and his coworkers [24] constructed SnO_x@Pt/C core-shell electrocatalysts for EOR and believed that the enhanced ethanol oxidation activity was mainly associated with the electronic modification of Pt shell from SnO_x core, which changes the d-band structure and is favorable to the oxidation of intermediate species bound strongly to the active sites. Similarly, in our heterostructure Pt/SnO_x nanostructure, the addition of SnO_x tunes the electronic structure of Pt through electron donating effect, which weakens the interaction between CO intermediates and Pt active centers, and mitigates the poisoning effect of the catalysts.

Since the sluggish ethanol oxidation kinetics for Pt-based electrocatalysts are mainly attributed to the strongly adsorption of CO-like intermediate species, which reduces the amount of active sites, CO stripping experiment was then carried out to elucidate the mechanism for the ultrahigh ethanol oxidation activity on Pt/SnO_x heterostructure. Fig. 3(c) depicts the CO stripping voltammetry curves for the three as-prepared catalysts, where sharp oxidation peaks of CO can be observed. The onset potential of CO_ads oxidation removal is 0.84 V for Pt/XC-72 catalysts, while this value shift towards the negative potential about 40 mV and 60 mV for Pt/Graphene and Pt/SnO_x/Graphene, respectively. Similar trends can be found for the CO stripping peak potential, which is in the following order: Pt/SnO_x/Graphene (0.93V) < Pt/Graphene (0.94V) < Pt/C (0.95V). The negative shift of both onset potential and peak potential implys that CO could be oxidized more easily on Pt/SnO_x/Graphene. Yoo et al. and his coworkers [43] have confirmed that there exists much smaller CO adsorption rate on Pt/Graphene comparing to that Pt/XC-72. Therefore, the CO is more easier to remove from the surface of graphene-supported catalysts thus increasing the activity of the catalyst. Moreover, according to reports from Camacho et al. [21, 22], the electrocatalytic activity could be improved by the strong interaction between Pt center and metal of the oxide. Therefore, the enhancement of CO oxidation for the Pt/SnO_x/Graphene catalyst could be interpreted by that the electronic effect from both SnO_x and graphene reduces the electron-donating capability of Pt to the π antibonding orbitals of CO, which weakens the adsorption of CO on Pt active sites and thus significantly improves the oxidation rate of CO [44]. Furthermore, in the Pt/SnO_x heterostructure, the electronic donation effect of SnO_x may facilitates the dissociation of H₂O, produces more OH_ads on Pt and accelerates the oxidation of adjacent CO_ads on Pt active sites [45], which results in a significant enhanced activity towards ethanol oxidation.

Pt/SnO_x/Graphene also shows a good stability against EOR. To assess both the electrocatalytic activity and the durability of the catalysts studied in steady conditions, chronoamperometric measurements (i-t curves) were conducted in 0.5 mol/L H₂SO₄ and 1 mol/L ethanol electrolyte at a constant voltage of 0.5V vs RHE. As displayed in Fig. S5, all the profiles have a downward trend with time elapsing, showing a obvious decay of the activity of the catalysts. This phenomenon is closely related to the poisoning of active centers, which could be explained by that the poisoning carbonceous intermediates gradually occupy the active sites on the electrode surface and partially suppress the ethanol oxidation process. As can be seen, the mass-specific current densities on Pt/SnO_x/Graphene and Pt/Graphene electrodes decrease more slowly comparing to that of Pt/XC-72 electrode, demonstrating a superior stability towards EOR. The residual current densities on Pt/SnO_x/Graphene and Pt/Graphene electrode were found to 0.0225 and 0.0194 A/mg-Pt, respectively, which is 1.86 and 1.60 folder that of Pt/C electrode. The electron affinities of Pt surface are weaken due to electron donating effect from both graphene and tin oxide, which produces more oxygenated species (such as OH_ads from water splitting) and thus accelerate the oxidation and removal efficiency of the poisoning intermediates [42]. Moreover, the strong metal-support interaction (SMSI) between Pt and SnO_x is also of critical significance for the stability of the Pt/SnO_x/Graphene nanostructure [46].

In summary, ultrafine Pt NPs were prepared by a facile and green EG method, and the electronic structure were successfully tuned by adjacent metal oxide and carbon support. The best catalyst is the one with Pt/SnO_x heterostructure highly dispersed on graphene, which shows the highest activity and durability towards ethanol oxidation. The reason is the electron donating effect in Pt from both tin oxide and carbon support (graphene), which facilitates the removal of poisoning CO immediate spices on Pt nanocrystals. Therefore, tuning electronic structure could be used as a facile and effective strategy to improve the performance of electrocatalysts in fuel cells, and other electrochemical devices for clean energy, such as CO₂ reduction, and water splitting.

The authors are grateful to the financial support from the Key Research and Development Project of Tianjin (18ZXJMTG00180), and the National Nature Science Foundation of China (21433003).

The authors declare no competing interests.