High Active PdSn Binary Alloyed Catalysts Supported on B and N Codoped Graphene for Formic Acid Electro-Oxidation

A series of PdSn binary catalysts with varied molar ratios of Pd to Sn are synthesized on B and N dual-doped graphene supporting materials. The catalysts are characterized by X-ray diffraction (XRD) and Transmission electron microscopy (TEM). Formic acid electro-oxidation reaction is performed on these catalysts, and the results reveal that the optimal proportion of Pd:Sn is 3:1. X-ray photoelectron spectroscopy (XPS) measurements show that when compared with 3Pd1Sn/graphene, B and N co-doping into the graphene sheet can tune the electronic structure of graphene, favoring the formation of small-sized metallic nanoparticles with good dispersion. On the other hand, when compared with the monometallic counterparts, the incorporation of Sn can generate oxygenated species that help to remove the intermediates, exposing more active Pd sites. Moreover, the electrochemical tests illustrate that 3Pd1Sn/BN-G catalyst with a moderate amount of Sn exhibits the best catalytic activity and stability on formic acid electro-oxidation, owing to the synergistic effect of the Sn doping and the B, N co-doping graphene substrate.


Introduction
With the development of renewable energy resources, fuel cells as portable power sources have been enormously studied. Among the different varieties of fuel cells, direct formic acid fuel cells (DFAFCs), which possess low toxic, high energy conversion efficiency and ease of handling, have attracted considerable attention [1,2]. The development of the active anode catalysts for the oxidation of formic acid plays an important role in the DFAFCs' technologies. Nowadays, compared with Pt, Pd-based catalysts are considered as a good candidate for anode catalysts because of their high power density and anti-poisoning ability [3,4]. However, the high cost and the low stability of Pd-based catalysts limit their practical application.
In order to improve the electrocatalytic activity and reduce the usage of the Pd catalysts, one feasible strategy is to form Pd-based alloys by incorporating non-precious metals. PdCo, PdFe, PdCu, PdNiPdSn and PdBi binary electrocatalysts have shown higher catalytic activities and stabilities for the electro-oxidation of small molecules (e.g., alcohol and formic acid) as well as oxygen reduction reation (ORR), than commercial Pd/C catalysts [5][6][7][8][9][10]. For example, Du et al. presented Pd86Sn14 showing much enhanced current densities and durability toward ethanol oxidation reaction among

Structrural Analysis
The phase structures of the Pd-based catalysts were analyzed by XRD, and the results are given in Figure 1. The broad diffraction peak at about 2θ = 26 • is attributed to the C (002) reflection of the graphene or its derivatives. It is noted that the C (002) peak intensities of 4Pd1Sn/BN-G and Pd/BN-G samples are relatively low, which could result from the higher Pd contents on the carbon surface. The diffraction peaks at approximately 2θ = 39.5 • , 45.4 • , 66.4 • and 79.6 • are assigned to the typical Pd face-centered cubic (fcc) crystal structure patterns of the (111), (200), (220) and (311) planes, respectively [30]. It is obvious that the Pd diffraction peaks of all the PdSn/BN-G samples slightly shift to the low-angle region, compared with that of Pd/BN-G, reflecting that the Sn atoms are incorporated into Pd lattice and the alloyed Pd-Sn nanomaterials are formed [12,15,27]. The small peak around 41.5 • for xPdySn/BN-G may be ascribed to the partial oxidation of Sn [31]. The crystallite sized and the lattice parameter α can be estimated from the Pd (111) diffraction data using the Debye-Scherrer formula and Bragg's law, and the crystallite size thus obtained and other physicochemical parameters are summarized in Table 1. It can be clearly seen that the average crystallite size decreasesin the order:Pd/G < 3Pd1Sn/G < Pd/BN-G < 1Pd1Sn/BN-G < 2Pd1Sn/BN-G < 4Pd1Sn/BN-G < 3Pd1Sn/BN-G. Moreover, the lattice structure of Pd crystal is expanded by the incorporation of Sn to Pd. Moreover, the alloying degree (χ, mole fraction of Sn, at.%) can be determined by the following equation [32].
χ (1 − r Sn /r Pd ) = 1 − α alloy /α Pd (1) where the element radii r Sn and r Pd are 1.41 and 1.37 Å, respectively, and α alloy and α Pd are the lattice parameters of PdSn alloys and Pd. The atomic ratios of Pd to Sn determined by ICP analysis (Table 1) are similar to the nominal ones, implying that most of the metal ions in the reactants are reduced to the metals. The morphology and microstructure of the 3Pd1Sn/BN-G catalyst were investigated by TEM. As is shown in Figure 2a, most of the alloy particles homogenously distribute on the supporting materials except for some degree of aggregation. According to the statistical size distribution histograms in the inset of Figure 2a, the average particle size is about 5.61 nm. In contrast, the metal particles in 3Pd1Sn/G are heterogeneously decorated on the graphene with much more agglomerates, as displayed in Figure S1a,b (the Supporting Information), and the alloy particles have larger sizes, around 9.94 nm. This means that with the heteroatom doping of B and N into the graphene architecture, the dispersion of metal on the graphene support can be improved and the combination of metal nanoparticles with graphene may also be enhanced [33]. In addition, as is seen from Figure S1c-h, the average particle size in 1Pd1Sn/BN-G, 2Pd1Sn/BN-G and 4Pd1Sn/BN-G is 6.8 nm, 5.7 nm and 5.65 nm, respectively. The results are close to the XRD data. The HRTEM images in Figure 2b reveal that the lattice fringe with a crystal plane distance of 0.23 nm is attributed to the d-spacing of the PdSn (111) plane [34]. Further inspection by elemental mapping in Figure 2c-h indicates that the catalyst contains C, N, B, Pd and Sn as the main components, and these elements are evenly dispersed throughout the selected region.This shows that the catalytic PdSn alloy particles with smaller size are well dispersed and effectively anchored onto the supporting materials. The EDS resultsfrom TEM aresummaried in Table  S1 (the Supporting Information) and it is revealed that the doping amount of B and N in graphene is about 6.16% and 1.83% (wt%).
where the element radii rSn and rPd are 1.41 and 1.37 Å, respectively, and αalloy and αPd are the lattice parameters of PdSn alloys and Pd. The atomic ratios of Pd to Sn determined by ICP analysis (Table  1) are similar to the nominal ones, implying that most of the metal ions in the reactants are reduced to the metals.
The morphology and microstructure of the 3Pd1Sn/BN-G catalyst were investigated by TEM. As is shown in Figure 2a, most of the alloy particles homogenously distribute on the supporting materials except for some degree of aggregation. According to the statistical size distribution histograms in the inset of Figure 2a, the average particle size is about 5.61 nm. In contrast, the metal particles in 3Pd1Sn/G are heterogeneously decorated on the graphene with much more agglomerates, as displayed in Figure S1a,b (the Supporting Information), and the alloy particles have larger sizes, around 9.94 nm. This means that with the heteroatom doping of B and N into the graphene architecture, the dispersion of metal on the graphene support can be improved and the combination of metal nanoparticles with graphene may also be enhanced [33]. In addition, as is seen from Figure S1c-h, the average particle size in 1Pd1Sn/BN-G, 2Pd1Sn/BN-G and 4Pd1Sn/BN-G is 6.8 nm, 5.7 nm and 5.65 nm, respectively. The results are close to the XRD data. The HRTEM images in Figure 2b reveal that the lattice fringe with a crystal plane distance of 0.23 nm is attributed to the d-spacing of the PdSn (111) plane [34]. Further inspection by elemental mapping in Figure  2c-h indicates that the catalyst contains C, N, B, Pd and Sn as the main components, and these elements are evenly dispersed throughout the selected region.This shows that the catalytic PdSn alloy particles with smaller size are well dispersed and effectively anchored onto the supporting materials. The EDS resultsfrom TEM aresummaried in Table S1 (the Supporting Information) and it is revealed that the doping amount of B and N in graphene is about 6.16% and 1.83% (wt%).     Figure 3a shows the cyclic voltammetric curves of different catalysts in 0.5 M H2SO4 solution in the scan range between −0.2 V and 0.8 V vs. SCE at a scan rate of 50 mV s −1 . It can be seen that the representative reduction adsorption and oxidation desorption peaks of atomic hydrogen exist at the potential region of −0.2 to 0.1 V [35,36]. The integration of the hydrogen desorption areas can be used to evaluate the electrochemically active surface area (ECSA) of the Pd nanoparticles, regarding the number of electrochemically active sites per unit mass of catalyst [20,37]. As depicted in Figure 3b, the ECSA values follow the sequence: 3Pd1Sn/BN-G (136.5 m 2 g −1 ) > 4Pd1Sn/BN-G (99.1 m 2 g −1 ) > 2Pd1Sn/BN-G (70.3 m 2 g −1 ) > 1Pd1Sn/BN-G (50.9 m 2 g −1 ) >Pd/BN-G (21.1 m 2 g −1 ) > 3Pd1Sn/G (15.2 m 2 g −1 ) >Pd/G (14.2 m 2 g −1 ). 3Pd1Sn/BN-G catalyst exhibits the largest ECSA value, which is most likely due to the good dispersion of metal nanoparticles immobilized on supporting materials.   Figure 3a shows the cyclic voltammetric curves of different catalysts in 0.5 M H 2 SO 4 solution in the scan range between −0.2 V and 0.8 V vs. SCE at a scan rate of 50 mV s −1 . It can be seen that the representative reduction adsorption and oxidation desorption peaks of atomic hydrogen exist at the potential region of −0.2 to 0.1 V [35,36]. The integration of the hydrogen desorption areas can be used to evaluate the electrochemically active surface area (ECSA) of the Pd nanoparticles, regarding the number of electrochemically active sites per unit mass of catalyst [20,37]. As depicted in Figure 3b, the ECSA values follow the sequence:

Electrochemical Characterization
. 3Pd1Sn/BN-G catalyst exhibits the largest ECSA value, which is most likely due to the good dispersion of metal nanoparticles immobilized on supporting materials.
The electro-catalytic performances towards formic acid electro-oxidation for all the samples were tested by the CV at a scan rate of 50 mV s −1 in 1 M HCOOH with 0.5 M H 2 SO 4 , and the results are illustrated in Figure 4. It is generally accepted that the electrooxidation of formic acid follows a dual-pathway mechanism [8,38]. One is a dehydrogenation path which directly produces CO 2 by reaction (2), and the other is a dehydration path which makes CO 2 by multi-step reactions (3)-(5).
(I) Dehydrogenation path Pd + HCOOH →Pd + CO2 + 2H + + 2e − In Figure 4, a strong peak at around 0.2 V and a shoulder at about 0.5 V occur in the anodic scan. The strong anodic peak at ~0.2 V is assigned to the direct oxidation of HCOOH to CO2 (reaction 2), and the shoulder peak at ~0.5 V is ascribed to the indirect oxidation (reactions 3−5). In the cathodic scan, the broad oxidation peak is related to the re-oxidation of HCOOH [18,39]. Obviously, the current densities for the direct oxidation of all the samples are much higher than those of the indirect one. This suggests that these Pd-based catalysts mainly oxidize formic acid through the direct pathway, and only a small amount of formic acid is oxidized by the indirect pathway, which leads to COads accumulated on the surface of catalysts [8,40,41]. Generally, the peak current density in the reaction (2), and the other is a dehydration path which makes CO2 by multi-step reactions (3−5). In Figure 4, a strong peak at around 0.2 V and a shoulder at about 0.5 V occur in the anodic scan. The strong anodic peak at ~0.2 V is assigned to the direct oxidation of HCOOH to CO2 (reaction 2), and the shoulder peak at ~0.5 V is ascribed to the indirect oxidation (reactions 3−5). In the cathodic scan, the broad oxidation peak is related to the re-oxidation of HCOOH [18,39]. Obviously, the current densities for the direct oxidation of all the samples are much higher than those of the indirect one. This suggests that these Pd-based catalysts mainly oxidize formic acid through the direct pathway, and only a small amount of formic acid is oxidized by the indirect pathway, which leads to COads accumulated on the surface of catalysts [8,40,41]. Generally, the peak current density in the In Figure 4, a strong peak at around 0.2 V and a shoulder at about 0.5 V occur in the anodic scan. The strong anodic peak at~0.2 V is assigned to the direct oxidation of HCOOH to CO 2 (reaction (2)), and the shoulder peak at~0.5 V is ascribed to the indirect oxidation (reactions (3)- (5)). In the cathodic scan, the broad oxidation peak is related to the re-oxidation of HCOOH [18,39]. Obviously, the current densities for the direct oxidation of all the samples are much higher than those of the indirect one. This suggests that these Pd-based catalysts mainly oxidize formic acid through the direct pathway, and only a small amount of formic acid is oxidized by the indirect pathway, which leads to CO ads accumulated on the surface of catalysts [8,40,41]. Generally, the peak current density in the forward scan is used to measure the electro-activity of the catalysts. As shown in Figure 4a, by comparing with Pd/BN-G, all the PdSn/BN-G samples display better electrocatalytic activity, and the specific activity of these samples follows the order: 3Pd1Sn/BN-G (86.23 mA cm −2 ) > 4Pd1Sn/BN-G (73.65 mA cm −2 ) > 2Pd1Sn/ BN-G (64.41 mA cm −2 ) > 1Pd1Sn/BN-G (58.76 mA cm −2 ) > Pd/BN-G (54.14 mA cm −2 ). This indicates that the addition of Sn into the catalysts may improve the electrocatalytic performance toward the electrochemical oxidation of HCOOH. Furthermore, as presented in Figure 4b, the electrocatalytic current of 3Pd1Sn/BN-G is about 1.9 times as large as that of 3Pd1Sn/G (46.18 mA cm −2 ), while the electrocatalytic activity of Pd/BN-G is 1.3 times higher than that of Pd/G (42.20 mA cm −2 ). Based on the above results, it is clear that the B, N dual-doping graphene can promote the activity of the catalysts. It is worth noting that the 3Pd1Sn/BN-G catalyst has the largest current density, that is, it shows the best electrocatalytic performance on the electrochemical oxidation of HCOOH, resulting from the synergistic effect of Sn addition and B, N dual-doping grapheme.
The stability of the catalysts is an important factor to evaluate the application of the fuel cells. Chronoamperometry tests were used to estimate the stability of all the samples at 0.4 V for 1 h in a solution of 1 M HCOOH + 0.5 M H 2 SO 4 . As is shown in Figure 5, the current densities of all the catalysts rapidly decrease at the initial stage, and then they gradually decay and reach a pseudo-steady state. This phenomenon is quite similar to those of other Pd-based catalysts [38,42], suggesting the same poisoning mechanism of the intermediate species accumulated on the catalysts toward the electrochemical oxidation of HCOOH. The pseudo-steady current density of 3Pd1Sn/BN-G reaches 12.00 mA cm −2 , which is much larger than the counterpart catalysts 4Pd1Sn/BN-G (8.87 mA cm −2 ), 2Pd1Sn/BN-G (7.57 mA cm −2 ), 1Pd1Sn/BN-G (6.63 mA cm −2 ), Pd/BN-G (4.87 mA cm −2 ), 3Pd1Sn/G (4.01 mA cm −2 ) and Pd/G (2.49 mA cm −2 ). The electrocatalytic activity of 3Pd1Sn/BN-G at different time intervals (0 s, 500 s, 3600 s) during the stability test is illustrated in Figure S2; it can be seen that within and even after the stability test, the electrocatalytic activity is almost the same as that before the stability test. Figure S3 shows the TEM images of 3Pd1Sn/BN-G before and after the chronoamperometric test; it can be found that there is no significant change in the structural integrity of the 3Pd1Sn/BN-G catalyst and most of the alloy particles retain homogenously dispersed on the supporting materials except for some degree of aggregation after the stability test, which is responsible for its relatively high catalytic activity. By considering the cyclic voltammograms and stability results, it is obvious to find that 3Pd1Sn/BN-G has the highest electrocatalytic for fomic acid oxidation, which results from an appropriate incorporation amount of Sn in the alloy phase and the addition of B and N into the graphene as the supporting material.
performance toward the electrochemical oxidation of HCOOH. Furthermore, as presented in Figure  4b, the electrocatalytic current of 3Pd1Sn/BN-G is about 1.9 times as large as that of 3Pd1Sn/G (46.18 mA cm −2 ), while the electrocatalytic activity of Pd/BN-G is 1.3 times higher than that of Pd/G (42.20 mA cm −2 ). Based on the above results, it is clear that the B, N dual-doping graphene can promote the activity of the catalysts. It is worth noting that the 3Pd1Sn/BN-G catalyst has the largest current density, that is, it shows the best electrocatalytic performance on the electrochemical oxidation of HCOOH, resulting from the synergistic effect of Sn addition and B, N dual-doping grapheme.
The stability of the catalysts is an important factor to evaluate the application of the fuel cells. Chronoamperometry tests were used to estimate the stability of all the samples at 0.4 V for 1 h in a solution of 1 M HCOOH + 0.5 M H2SO4. As is shown in Figure 5, the current densities of all the catalysts rapidly decrease at the initial stage, and then they gradually decay and reach a pseudo-steady state. This phenomenon is quite similar to those of other Pd-based catalysts [38,42], suggesting the same poisoning mechanism of the intermediate species accumulated on the catalysts toward the electrochemical oxidation of HCOOH. The pseudo-steady current density of 3Pd1Sn/BN-G reaches 12.00 mA cm −2 , which is much larger than the counterpart catalysts 4Pd1Sn/BN-G (8.87 mA cm −2 ), 2Pd1Sn/BN-G (7.57 mA cm −2 ), 1Pd1Sn/BN-G (6.63 mA cm −2 ), Pd/BN-G (4.87 mA cm −2 ), 3Pd1Sn/G (4.01 mA cm −2 ) and Pd/G (2.49 mA cm −2 ). The electrocatalytic activity of 3Pd1Sn/BN-G at different time intervals (0 s, 500 s, 3600 s) during the stability test is illustrated in Figure S2; it can be seen that within and even after the stability test, the electrocatalytic activity is almost the same as that before the stability test. Figure S3 shows the TEM images of 3Pd1Sn/BN-G before and after the chronoamperometric test; it can be found that there is no significant change in the structural integrity of the 3Pd1Sn/BN-G catalyst and most of the alloy particles retain homogenously dispersed on the supporting materials except for some degree of aggregation after the stability test, which is responsible for its relatively high catalytic activity. By considering the cyclic voltammograms and stability results, it is obvious to find that 3Pd1Sn/BN-G has the highest electrocatalytic for fomic acid oxidation, which results from an appropriate incorporation amount of Sn in the alloy phase and the addition of B and N into the graphene as the supporting material.

XPS Analysis
In order to gain further insight into the promoting effect of 3Pd1Sn/BN-G catalyst toward formic acid oxidation, the XPS measurements were carried out to analyze the surface composition and the valent state of the elements in the catalysts. As displayed by the survey spectrum of XPS in Figure 6a, it is clear that C, B, N, Pd, Sn and O elements exist in 3Pd1Sn/BN-G catalyst. Percentages of the related fitting species are listed in Table 2. As depicted in Figure S4, N 1s spectrum can be divided into five peaks as B-N type, pyridinic N, pyrrolic N, graphitic N and oxidized N, and B 1 s pattern can be convoluted into three species, named as BC 3 (B-N), BC 2 O and BCO 2 [43].
As is shown in Figure 6b,c, the splitting patterns of the Pd 3d band of 3Pd1Sn/G and 3Pd1Sn/B-N-G contains three pairs of doublets. The intensive doublet (335.6, 340.8 eV) is related to Pd0, the doublet Catalysts 2020, 10, 751 7 of 12 at 336.8 and 342.1 eV may be assigned to Pd(II) in PdO, while another doublet at 338.2 and 343.6 eV corresponds to the Pd(IV) in PdO 2 [20,44]. The contents of Pd0, Pd(II) and Pd(IV) in 3Pd1Sn/G are 57%, 25% and 18%, respectively (Table 2). In contrast, the proportion of metallic Pd in 3Pd1Sn/B-N-G is increased to 61%, whereas, the proportions of Pd(II) and Pd(IV) are 26% and 13%, respectively. Additionally, in our previous work, compared with Pd/G, Pd/BN-G also possesses the increased metallic Pd [43]. Many investigations illustrated that N doping benefits the nucleation of metal particles [45,46] and the B atom with low electronegativity increases the electron density of Pd [33]. This indicates that B, N co-doping into the framework of graphene can combine the advantages of B and N to tune the electronic structure of graphene, thus favoring the formation and good dispersion of metallic nanoparticles. Figure 6a, it is clear that C, B, N, Pd, Sn and O elements exist in 3Pd1Sn/BN-G catalyst. Percentages of the related fitting species are listed in Table 2. As depicted in Figure S4, N 1s spectrum can be divided into five peaks as B-N type, pyridinic N, pyrrolic N, graphitic N and oxidized N, and B 1 s pattern can be convoluted into three species, named as BC3(B-N), BC2O and BCO2 [43].
As is shown in Figure 6b,c, the splitting patterns of the Pd 3d band of 3Pd1Sn/G and 3Pd1Sn/B-N-G contains three pairs of doublets. The intensive doublet (335.6, 340.8 eV) is related to Pd0, the doublet at 336.8 and 342.1 eV may be assigned to Pd(Ⅱ) in PdO, while another doublet at 338.2 and 343.6 eV corresponds to the Pd(Ⅳ) in PdO2 [20,44]. The contents of Pd0, Pd(Ⅱ) and Pd(Ⅳ) in 3Pd1Sn/G are 57%, 25% and 18%, respectively (Table 2). In contrast, the proportion of metallic Pd in 3Pd1Sn/B-N-G is increased to 61%, whereas, the proportions of Pd(Ⅱ) and Pd(Ⅳ) are 26% and 13%, respectively. Additionally, in our previous work, compared with Pd/G, Pd/BN-G also possesses the increased metallic Pd [43]. Many investigations illustrated that N doping benefits the nucleation of metal particles [45,46] and the B atom with low electronegativity increases the electron density of Pd [33]. This indicates that B, N co-doping into the framework of graphene can combine the advantages of B and N to tune the electronic structure of graphene, thus favoring the formation and good dispersion of metallic nanoparticles. Sn 3d signal can be deconvoluted into two symmetrical peaks, as illustrated in Figure 6d. The two peaks at 485.1 and 493.5 eV are ascribed to metallic Sn, and the other peaks at 486.9 and 495.3 eV belong to oxidized Sn species, mainly in the form of SnO2 [11]. The proportions of two components are 5% for Sn0 and 95% for Sn4+. So it is reasonable to believe that the Sn element in the surface layer Sn 3d signal can be deconvoluted into two symmetrical peaks, as illustrated in Figure 6d. The two peaks at 485.1 and 493.5 eV are ascribed to metallic Sn, and the other peaks at 486.9 and 495.3 eV belong to oxidized Sn species, mainly in the form of SnO 2 [11]. The proportions of two components are 5% for Sn0 and 95% for Sn4+. So it is reasonable to believe that the Sn element in the surface layer of the 3Pd1Sn/BN-G catalyst mainly exists in the form of SnO 2 . SnO 2 can provide OH species that aid the removal of CO-like intermediate poisonous species [47]. With the incorporation of Sn to Pd, the oxophilic character of Sn atoms are likely to be present at the surface of Pd metals to generate oxygenated species, which can help re-oxidize the intermediates, such as CO ads , thus preventing the accumulation of poisoning-species and leaving more active Pd sites available for the electrochemical oxidation of HCOOH. This is responsible for the improvement of the catalytic activity and stability of the PdSn catalysts. According to the XRD results, the alloyed Pd-Sn nanomaterials are formed in the xPdySn/BN-G catalysts, while in the XPS analysis Sn in the surface layer of these catalysts mainly exists in the form of SnO 2 . Combined with XRD and XPS data, we conclude that xPdySn/BN-G catalysts are homogeneous alloys with Pd and Sn, while the Sn atoms on the surface with most of the surface Sn oxidized into SnO 2 . Considering that Sn itself has no catalytic activity toward formic Catalysts 2020, 10, 751 8 of 12 acid oxidation [15], it not only acts as a co-catalyst, but also a sacrificial component in the alloy, so a moderate amount of Sn is a key factor affecting the catalytic performance of Pd. According to the above results, it is found that the the optimal ratio of Pd to Sn is 3:1, which is consistent with other functionalized carbon-supported PdSn catalysts [31,48]. Table 2. Fitting results of 3Pd1Sn/G and 3Pd1Sn/B-N-G from XPS in Figure 6.

Synthesis of BN-G Supporting Materials
BN-G was produced by a two-step hydrothermal method as described in our previous work [43]. Briefly, graphene oxide (GO) was firstly prepared by the Hummers' liquid phase oxidation method [49], using graphite powder (<20 µm, Sigmar-Aldrich, Shanghai, China) as a raw material, and B-G was synthesized using boric acid (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) as the B source. An amount of 200 mg GO was ultrasonically dispersed in 20 mL distilled water, and boric acid (4 wt%) (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China)was added. The as-obtained suspension was put into a 100 mL Teflon-lined stainless steel autoclave and maintained at 200 • C for 5 h, and then the intermediate product was centrifuged, washed and dried in a vacuum oven at 80 • C overnight. The intermediate product was placed in a tube furnace and heated at 400 • C for 4 h in Ar atmosphere, and then B-G sample was obtained. Then, 200 mg B-G and 50 mL of 25 wt% aqueous ammonia (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) were thoroughly mixed under ultrasonic vibration. The resulting slurry was transferred to an autoclave as the same condition as B-G, and then the product BN-G was prepared.

Synthesis of PdSn/BN-G Binary Catalysts
A series of xPdySn/BN-G binary alloyed catalysts were synthesized, with the x/y ratios varying from 1:1 to 4:1 and the total-metal loading of 20 wt% in each sample. A typical synthesis of 3Pd1Sn/BN-G catalyst was as follows. First, 80 mg BN-G was uniformly dispersed in 30 mL distilled water with ultrasonic stirring. Subsequently, 2.741 mL 0.05 M PdCl 2 (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) and 10.31 g SnCl 2 ·2H 2 O (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) were mixed with the above slurry and stirred for 30 min. The resultant suspension solution was transferred to a three-neck bottle, and potassium borohydride solution (KBH 4 , 316 mg in 120 mL water) (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) was added drop by drop, and this reaction was conducted at 60 • C for 6 h. Finally, the product was filtered and dried overnight. Other samples including 1Pd1Sn/BN-G, 2Pd1Sn/BN-G and 4Pd1Sn/BN-G were also synthesized by the same method. For comparison, 3Pd1Sn/G and Pd/BN-G were prepared by a similar process.

Materials Characterization
The crystalline structures of synthesized samples were measured by powder X-ray diffraction (XRD) analyses (Bruker D8, Germany). The morphologies of materials were investigated by transmission electron microscopy (TEM, FEI F20, USA). An Escalab 250Xi X-ray photoelectron spectrometer (XPS, USA) operating at 500 W and 15 kV with Mg Kα radiation was used to record XPS spectra, Catalysts 2020, 10, 751 9 of 12 where binding energies were referred to the adventitious C 1s peak at 284.6 eV. The XPS spectra were analyzed using XPS PEAK 4.1 software. Metal contents of all the samples were obtained by inductively coupled plasma optical emission spectrometer (ICP-OES, Agilent 725, USA).

Electrocatalytic Measurements
The electrocatalytic properties of the catalysts were conducted on a CHI760d electrochemical workstation (Shanghai, China) using a three electrode system, with the catalysts' modified glass carbon electrodes (4mm in diameter) as the working electrode, a Pt foil as the counter electrode and the saturated calomel electrode (SCE) as the reference electrode. All the electrochemical tests were carried out in a N 2 -saturated electrolyte at 25 • C. The working electrode was obtainedby the following procedure. First, 5 mg catalysts, 30 µL 5 wt% Nafion solution and 1.1 mL ethanol (analytical grade, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) were mixed together with ultrasonic stirring, and then 5 µL as-obtained slurry was spread onto the pre-polished glass carbon disk, where the loading amount of the catalysts was 0.175 mg cm −2 .

Conclusions
PdSnbinary alloyed catalysts have been successfully prepared on B, N co-doping graphene substrate by a chemical reduction procedure. The heteroatom doping of B and N into the graphene substrate modify the electronic structure of graphene, thus improving the deposition of PdSn alloys and enhancing the metal-support interaction. The doping of Sn in Pd may promote the removal of the accumulated CO-like residues, thus exposing more free active Pd sites for formic acid electro-oxidation. 3Pd1Sn/BN-G with an optimal proportion of Sn shows more active and stable electro-catalytic performance, benefiting from its small particle size and good dispersion. The above promising results may shed new light on the design of active Pd-based catalysts in the direct formic acid fuel cells.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/7/751/s1, Figure S1: (a) TEM image of 3Pd1Sn/ G and (b) its corresponding particle size distribution; (c) TEM image of 1Pd1Sn/ BN-G and (d) its corresponding particle size distribution; (e) TEM image of 2Pd1Sn/ BN-G and (f) its corresponding particle size distribution; (g) TEM image of 4Pd1Sn/ BN-G and (h) its corresponding particle size distribution. Figure S2: Cyclic voltammograms of 3Pd1Sn/BN-G in the solution with 1 M HCOOH + 0.5 M H 2 SO 4 solution during the stability test at different time intervals (0, 500, 3600s) , scan rate: 50 mV s −1 . Figure S3: TEM images of 3Pd1Sn/BN-G before (a) and after (b) the electrocatalytic stability test toward the electrochemical oxidation of HCOOH.

Conflicts of Interest:
The authors declare no conflict of interest.