Electrochemical Toluene Hydrogenation Using Binary Platinum-Based Alloy Nanoparticle-Loaded Carbon Catalysts

A couple of toluene (TL) and its hydrogenation product, methylcyclohexane (MCH), are promising high-density hydrogen carriers to store and transport large amounts of hydrogen. Electrochemical hydrogenation of TL to MCH can achieve energy savings compared with hydrogenation using molecular hydrogen generated separately, and development of highly active catalysts for electrochemical TL hydrogenation is indispensable. In this study, binary Pt3M (M = Rh, Au, Pd, Ir, Cu and Ni) alloy nanoparticle-loaded carbon catalysts were prepared by a colloidal method, and their activity for electrochemical TL hydrogenation was evaluated by linear sweep voltammetry. Each Pt3M electrode was initially activated by 100 cycles of potential sweep over a potential range of 0–1.2 or 0.8 V vs. reversible hydrogen electrode (RHE). For all activated Pt3M electrodes, the cathodic current density for electrochemical TL hydrogenation was observed above 0 V, that is the standard potential of hydrogen evolution reaction. Both specific activity, cathodic current density per electrochemical surface area, and mass activity, cathodic current density per mass of Pt3M, at 0 V for the Pt3Rh/C electrode were the highest, and about 8- and 1.2-times as high as those of the commercial Pt/C electrode, respectively, which could mainly be attributed to electronic modification of Pt by alloying with Rh. The Tafel slope for each activated Pt3M/C electrode exhibited the alloying of Pt with the second metals did not change the electrochemical TL hydrogenation mechanism.


Introduction
Most energy comes from fossil fuels, but the use of fossil fuels involves the generation of CO 2 . Hydrogen is a promising alternative energy source because it has much higher gravimetric energy density (120 MJ kg −1 ) than gasoline (44 MJ kg −1 ) [1,2] and its combustion product is only water. Hydrogen gas, however, has the disadvantage of low volumetric energy density (10.8 kJ L −1 under atmospheric conditions [2,3]), so there is an urgent need to develop hydrogen carriers that can store and transport large amounts of hydrogen.
In general, the hydrogenation of TL to MCH is carried out by adding molecular hydrogen to TL on an appropriate catalyst as follows.
C 6 H 5 CH 3 + 3H 2 → C 6 H 11 CH 3 (1) The molecular hydrogen is produced by chemical or electrochemical method. On the other hand, the hydrogenation reaction of TL to MCH is also possible using atomic hydrogen electrochemically generated by one-electron reduction of H + in water electrolysis. Hence, direct electrochemical TL hydrogenation to MCH coupled with water electrolysis is totally represented as follows [19][20][21][22][23][24][25].
The development of highly active catalysts for electrochemical TL hydrogenation is indispensable to suppress hydrogen evolution reaction (HER) as a side reaction. It has been reported that PtRu alloy nanoparticle-loaded carbon black (PtRu/C) catalysts showed high hydrogenation activity and faradaic efficiency (FE) for MCH production, which were higher than those of Pt/C [22,23,26,27]. We also have reported that the Pt 3 Ru alloy nanoparticle-loaded carbon black (Pt 3 Ru/C) catalyst exhibited the highest hydrogenation activity among Pt x Ru/C (x = 1, 3, 4) catalysts, and 95% FE for MCH production [24]. The Au/C, Pd/C, Ru/C, Rh/C, and Ir/C catalysts were investigated as single precious metal catalysts other than Pt/C for electrochemical TL hydrogenation [28]. The Ru/C catalyst achieved high FE for MCH formation even at low metal loading of 1 wt %, but required high overpotential for electrochemical TL hydrogenation. The mixture of 5 wt % Ru and 5 wt % Ir loaded on carbon was comparable to 50 wt % Pt supported on carbon in the hydrogenation due to a combination of fast electrochemical reduction of H + on Ir and fast TL hydrogenation on Ru via spillover of generated atomic hydrogen from Ir to Ru [28][29][30]. In this way, the synergistic effect of two elements playing different roles is important to enhance the activity for electrochemical TL hydrogenation. Meanwhile, as for binary Pt-based alloy catalysts, to our best knowledge, only Pt x Ru alloys have been available so far. Therefore, to further enhance the hydrogenation activity of the Pt catalyst, searching for a partner to replace ruthenium is essential. In this study, several Pt-based binary alloy, Pt 3 M (M = Rh, Au, Pd, Ir, Cu and Ni) nanoparticle-loaded carbon (Pt 3 M/C) catalysts are prepared by the colloidal method [24] and their electrochemical TL hydrogenation activity was evaluated.  Table 1. For each Pt 3 M/C catalyst, the evaluated metal loading was close to the theoretical one (50 wt %), suggesting that the reduction of Pt and M precursors was almost completed, and the produced metals were loaded on ketjen black.  Figure 1 shows X-ray diffraction (XRD) patterns of the Pt 3 M/C catalysts. All catalysts had a single diffraction peak around 40 • and 47 • , which were close to the (111) and (200) peaks of platinum (ICDD 00-004-0802) with face-centered cubic (fcc) crystal structure, respectively, suggesting that Pt and M in metallic phase for each catalyst were alloyed. For the catalysts other than Pt 3 Au/C, the (111) peak shifted towards higher angles compared to that of pure Pt, indicating that the lattice parameter of each catalyst is smaller than that of pure Pt, as shown in Table 1, because the radii of the second metals are smaller than that of Pt. In contrast, the Pt 3 Au/C had additional two peaks around 38.42 • and 44.74 • , which has also been reported previously [31]. This suggests that phase segregation occurred in this catalyst. These diffraction peaks shifted towards higher angles than the (111) and (200) peaks (38.18 • and 44.39 • ) for gold (ICDD 00-004-0784), suggesting the formation of a binary Au-rich alloy. As shown in Table 1, the 2θ value of the main (111) peak for Pt 3 Au/C was almost the same as that for pure Pt, suggesting the existence of the Pt nanoparticles.  shows Pt 4f and Rh 3d, Au 4f, Pd 3d, Ir 4f, Cu 2p or Ni 2p core level spectra for the Pt3M/C (M = Rh, Au, Pd, Ir, Cu and Ni) catalysts. For the Pt3Rh/C catalyst, Rh 3d spectrum exhibited doublets at 307.4 and 312.3 eV due to metallic Rh (Rh 0 ), which were higher in binding energy than pure Rh 0 (307.0 and 311.7 eV) [34]. For the Pt3Au/C catalyst, Au 4f spectrum exhibited doublets at 84.0 and 87.7 eV due to metallic Au (Au 0 ), which almost agree with those of pure Au 0 (84.0 and 87.7 eV) [35]. For the Pt3Pd/C catalyst, Pd  The mean crystallite size of each Pt 3 M nanoparticle loaded on ketjen black was estimated with Scherrer's equation (Equation (5)) for the (111) peak in each XRD pattern assuming a single Pt x M phase, and summarized in Table 1.

Results and Discussion
where L is the mean crystallite size (nm), λ is the wavelength of the X-ray (0.1541 nm), θ is the Bragg angle of the (111) peak, and B is the width (radian) of the (111) peak at half height [32,33]. The mean crystallite size was 3.0-4.8 nm, which was relatively close to each other. Figures S1 and S2 (Supplementary Materials) show transmission electron microscope (TEM) images and size distribution profiles of nanoparticles for Pt 3 M/C catalysts. The mean crystallite size of each catalyst estimated from its XRD pattern (Table 1) was in good agreement with that estimated from the corresponding TEM image. Figure 2 shows Pt 4f and Rh 3d, Au 4f, Pd 3d, Ir 4f, Cu 2p or Ni 2p core level spectra for the Pt 3 M/C (M = Rh, Au, Pd, Ir, Cu and Ni) catalysts. For the Pt 3 Rh/C catalyst, Rh 3d spectrum exhibited doublets at 307.4 and 312.3 eV due to metallic Rh (Rh 0 ), which were higher in binding energy than pure Rh 0 (307.0 and 311.7 eV) [34]. For the Pt 3 Au/C catalyst, Au 4f spectrum exhibited doublets at 84.0 and 87.7 eV due to metallic Au (Au 0 ), which almost agree with those of pure Au 0 (84.0 and 87.7 eV) [35]. For the Pt 3 Pd/C catalyst, Pd 3d spectrum exhibited doublets at 335.5 and 340.9 eV due to metallic Pd (Pd 0 ), which were higher in binding energy than pure Pd 0 (335.2 and 340.5 eV) [36]. For the Pt 3 Ir/C catalyst, Ir 4f spectrum exhibited doublets at 61.2 and 64.4 eV due to metallic Ir (Ir 0 ), which were higher in binding energy than pure Ir 0 (60.9 and 63.8 eV) [37]. Additionally, the Ir 4f spectrum exhibited doublets at 66.1 and 62.9 eV [38] because iridium was oxidized to cations with valences of 4 or more in the air. For the Pt 3 Cu/C catalyst, Cu 2p spectrum exhibited doublets at 932.2 and 952.1 eV due to metallic Cu (Cu 0 ), which were lower in binding energy than pure Cu 0 (932.5 and 952.3 eV) [39]. Additionally, the Cu 2p 3/2 peak at 934.6 eV and Cu 2p 1/2 peak at 954.6 eV due to CuO were observed with the satellite peaks at 941.5, 944.3 and 963.8 eV [39,40]. For the Pt 3 Ni/C catalyst, Ni 2p spectrum exhibited singlet at 852.7 eV due to metallic Ni (Ni 0 ), which was higher in binding energy than pure Ni 0 (852.4 eV) [41]. Also the Ni 2p spectrum exhibited doublets at 857.2 and 874.8 eV due to Ni(OH) 2 with the satellite peaks at 861.2 and 880.2 eV [42,43]. In all Pt 3 M/C catalysts, Pt 4f spectrum exhibited doublets due to metallic Pt (Pt 0 ), which were higher in binding energy than pure Pt 0 (71.1 and 74.4 eV) [44]. These results also support the alloying of Pt with the second metals.
The bulk composition and surface composition of the Pt 3 M/C catalysts were evaluated by energy dispersive X-ray (EDX) and X-ray photoelectron spectroscopy (XPS), respectively, and summarized in Table 2. The bulk composition of Pt 3 Ir/C, however, could not be evaluated because the Pt and Ir peaks in the EDX spectrum overlapped each other. The bulk composition for Pt 3 M/C catalysts other than Pt 3 Au/C was close to the theoretical one (Pt 75 at % and M 25 at %), and the Pt/M atomic ratio calculated from the bulk composition was 2.8-3.5. Thus, the composition of each catalyst is nominally represented as Pt 3 M hereafter. For the Pt 3 Au/C alloy, the surface composition was significantly different from the bulk composition probably because of the phase segregation and/or partial dissolution of the surface Au atoms into the Pt and Au-rich alloy particles. The bulk composition and surface composition of the Pt3M/C catalysts were evaluated by energy dispersive X-ray (EDX) and X-ray photoelectron spectroscopy (XPS), respectively, and summarized in Table 2. The bulk composition of Pt3Ir/C, however, could not be evaluated because the Pt and Ir peaks in the EDX spectrum overlapped each other. The bulk composition for Pt3M/C catalysts other than Pt3Au/C was close to the theoretical one (Pt 75 at % and M 25 at %), and the Pt/M atomic ratio calculated from the bulk composition was 2.8-3.5. Thus, the composition of each catalyst is nominally represented as Pt3M hereafter. For the Pt3Au/C alloy, the surface composition was significantly different from the bulk composition probably because of the phase segregation and/or partial dissolution of the surface Au atoms into the Pt and Au-rich alloy particles. able 2. Surface composition, bulk composition, metal loading mass and mean crystallite size for Pt3M/C (M = Rh, Au, Pd, , Cu and Ni) catalysts as prepared and after the linear sweep voltammetry measurement for electrochemical toluene (TL) ydrogenation.    Figure 3 shows cyclic voltammograms (CVs) of the Pt 3 M/C electrodes after initial activation to clean the electrode surface before evaluating the electrochemical TL hydrogenation activity. The initial activation was performed by 100 cycles of potential sweep over the potential range of 0.05-1.2 V vs. RHE for the Pt 3 Rh/C, Pt 3 Au/C, Pt 3 Pd/C and Pt 3 Ir/C electrodes, and 0.05-0.8 V vs. RHE for the Pt 3 Cu/C and Pt 3 Ni/C electrodes because Cu and Ni were easier to be oxidatively dissolve than the noble metals. In general, a polycrystalline Pt electrode has two couples of redox peaks due to hydrogen adsorption/desorption  [44,45]. Each Pt 3 M electrode also had these redox peaks, as shown in Figure 4, although they were not distinctly separated. The electric charge for the desorption of atomic hydrogen (Q H-des ) for a polycrystalline Pt is 210 µC cm −2 [45], indicating that an atomic hydrogen is adsorbed on a Pt atom. The oxidation peaks due to hydrogen desorption for pure Rh, Pd and Ir electrodes partly overlap those for the polycrystalline Pt electrode [46,47], but not for pure Au, Cu and Ni electrodes. If binary Pt-based alloys maintain the fcc structure and the atomic radii of Pt and foreign metals are similar to each other, the Q H-des value of 210 µC cm −2 is often used, supposing that each constituent (M) has a stoichiometric composition of M:H = 1:1 [48]. The Pt 3 Rh, Pt 3 Pd and Pt 3 Ir alloys have the fcc structure, as can be seen from Figure 1. So, the electrochemical surface areas (ECSAs) of these electrodes were calculated with the Q H-des values estimated from their CVs in Figure 3 as follows, supposing 210 µC cm −2 .

Electrochemical TL Hydrogenation Activity with Pt 3 M/C Electrodes
Cu and Ni were easier to be oxidatively dissolve than the noble metals. In general, a polycrystalline Pt electrode has two couples of redox peaks due to hydrogen adsorption/desorption between 0.05 and 0.4 V vs. RHE [44,45]. Each Pt3M electrode also had these redox peaks, as shown in Figure 4, although they were not distinctly separated. The electric charge for the desorption of atomic hydrogen (QH-des) for a polycrystalline Pt is 210 μC cm −2 [45], indicating that an atomic hydrogen is adsorbed on a Pt atom. The oxidation peaks due to hydrogen desorption for pure Rh, Pd and Ir electrodes partly overlap those for the polycrystalline Pt electrode [46,47], but not for pure Au, Cu and Ni electrodes. If binary Pt-based alloys maintain the fcc structure and the atomic radii of Pt and foreign metals are similar to each other, the QH-des value of 210 μC cm −2 is often used, supposing that each constituent (M) has a stoichiometric composition of M:H = 1:1 [48]. The Pt3Rh, Pt3Pd and Pt3Ir alloys have the fcc structure, as can be seen from Figure 1. So, the electrochemical surface areas (ECSAs) of these electrodes were calculated with the QH-des values estimated from their CVs in Figure 3 as follows, supposing 210 μC cm −2 .
The ECSAs of the Pt3Rh/C, Pt3Pd/C and Pt3Ir/C electrodes, which are the sum of ECSA for the Pt and M surfaces, were evaluated to be 532, 339 and 673 cm 2 , respectively. On the other hand, for the Pt3Au/C, Pt3Cu/C and Pt3Ni/C electrodes, the ECSA estimated from the QH-des value will be quite equal to that of the Pt constituent surface of each alloy. The ECSAs of the Pt3Au/C, Pt3Cu/C and Pt3Ni/C electrodes were 325, 500 and 455 cm 2 , respectively.   when neat TL was supplied into the cathode compartment of the electrochemical hydrogenation cell. The initial activation was performed by 100 cycles of potential sweep over the potential range of 0.05-1.2 V vs. RHE for the Pt3Rh/C, Pt3Au/C, Pt3Pd/C, Pt3Ir/C and Pt/C electrodes, and 0.05-0.8 V vs. RHE for the Pt3Cu/C and Pt3Ni/C electrodes. Figure  4a,b are represented by the geometric current density (jgeo), and current density per ECSA (jECSA) or specific activity (SA), respectively. In Figure 4a, the onset potential of jgeo (Eonset) for each LSV is summarized in Table 3. All Eonset values were more positive than 0 V vs. RHE or the standard potential for HER, suggesting that the cathodic current density of each LSV is attributed to the electrochemical TL hydrogenation [49].
The jgeo and jECSA values at 0 V vs. RHE (jgeo,0 and SA0) were evaluated from each LSV in Figure 4 to use as measures of electrochemical TL hydrogenation activity. The results are summarized in Table 3 and Figure 5. The SA0 was the highest for the Pt3Rh/C electrode, and decreased in the order of Pt3Rh/C > Pt3Au/C > Pt3Pd/C > Pt3Cu/C > Pt3Ir/C > Pt3Ni/C > Pt/C. This indicates that the alloying of Pt with the second metals is effective for improving electrochemical TL hydrogenation activity of Pt. Remarkably, for the Pt3Rh/C electrode, when the initial activation potential range was changed to 0.05-0.8 V vs. RHE, the jECSA or SA greatly increased, as shown in Figure 4b. The SA0 value was 1.60 mA cm −2 , which is about 2.4, 8.4 and 1.8-times as high as that for the Pt3Rh/C electrode with the initial activation between 0.05 and 1.2 V vs. RHE, the Pt/C electrode and the Pt3Ru/C electrode [24] probably due to the increased surface Rh content, respectively. It has been reported that electrochemical TL hydrogenation activity for pure Rh was lower than that for pure Pt [27,28]. Moreover, we have reported that the electrochemical TL hydrogenation activity for Rh-modified Pt/C catalysts, in which Rh did not form any alloys with Pt, was the maximum 2.7-times that for the commercial Pt/C catalyst due to the increase in the extent of the interface between Pt and Rh resulting from the modification of the Pt surface with Rh [49]. In contrast, the present Pt3Rh alloy was homogeneous, as expected from the XRD pattern (Figure 1), and the Rh content is about onethird of the Pt content, so the domains consisting of only Rh atoms will not have been formed in the alloy. Therefore, in the present study, the Rh constituent seems to contribute to the electronic modification of Pt atoms. In our previous study with Rh-modified Pt/C catalysts, the first step of electrochemical TL hydrogenation or one-electron reduction of H + to atomic hydrogen was anticipated to proceed on the Rh sites [49]. So, the surface Rh sites of the Pt3Rh alloy may also contribute to facilitating the first step.   The ECSAs of the Pt 3 Rh/C, Pt 3 Pd/C and Pt 3 Ir/C electrodes, which are the sum of ECSA for the Pt and M surfaces, were evaluated to be 532, 339 and 673 cm 2 , respectively. On the other hand, for the Pt 3 Au/C, Pt 3 Cu/C and Pt 3 Ni/C electrodes, the ECSA estimated from the Q H-des value will be quite equal to that of the Pt constituent surface of each alloy. The ECSAs of the Pt 3 Au/C, Pt 3 Cu/C and Pt 3 Ni/C electrodes were 325, 500 and 455 cm 2 , respectively. Figure 4 shows the linear sweep voltammograms (LSVs) of activated Pt 3 M/C (M = Rh, Au, Pd, Ir, Cu and Ni) and commercial Pt/C electrodes at a sweep rate of 50 mV s −1 when neat TL was supplied into the cathode compartment of the electrochemical hydrogenation cell. The initial activation was performed by 100 cycles of potential sweep over the potential range of 0.05-1.2 V vs. RHE for the Pt 3 Rh/C, Pt 3 Au/C, Pt 3 Pd/C, Pt 3 Ir/C and Pt/C electrodes, and 0.05-0.8 V vs. RHE for the Pt 3 Cu/C and Pt 3 Ni/C electrodes. Figure 4a,b are represented by the geometric current density (j geo ), and current density per ECSA (j ECSA ) or specific activity (SA), respectively. In Figure 4a, the onset potential of j geo (E onset ) for each LSV is summarized in Table 3. All E onset values were more positive than 0 V vs. RHE or the standard potential for HER, suggesting that the cathodic current density of each LSV is attributed to the electrochemical TL hydrogenation [49]. The j geo and j ECSA values at 0 V vs. RHE (j geo,0 and SA 0 ) were evaluated from each LSV in Figure 4 to use as measures of electrochemical TL hydrogenation activity. The results are summarized in Table 3 and Figure 5. The SA 0 was the highest for the Pt 3 Rh/C electrode, and decreased in the order of Pt 3 Rh/C > Pt 3 Au/C > Pt 3 Pd/C > Pt 3 Cu/C > Pt 3 Ir/C > Pt 3 Ni/C > Pt/C. This indicates that the alloying of Pt with the second metals is effective for improving electrochemical TL hydrogenation activity of Pt.  To investigate the stability of the Pt3M/C catalysts during the initial activation and the following LSV measurement for electrochemical TL hydrogenation, bulk composition and surface composition of each catalyst after the LSV measurement were evaluated by EDX and XPS ( Figure S3, Supplementary Materials), respectively, and summarized in Table 2. For the Pt3M/C electrodes other than Pt3Au, both surface and bulk contents of the second metals decreased after the LSV measurement, and the decrease in the surface content was larger than that in the bulk content because the second metals were easier to oxidatively dissolve than Pt. Presumably, the oxidative dissolution of the second metals proceeded during the initial activation, not during the LSV measurement. For the Pt3Au/C catalyst, the Au constituent was not oxidized under the initial activation conditions, so the surface composition hardly changed during the initial activation.
The mean crystallite size of each Pt3M alloy after the LSV measurement for electrochemical TL hydrogenation was estimated from the XRD pattern ( Figure S4, Supplementary Materials), and summarized in Table 2. In all cases, the mean crystallite size increased after the LSV measurement. There are several styles of particle growth: (1) Ostwald rip- Remarkably, for the Pt 3 Rh/C electrode, when the initial activation potential range was changed to 0.05-0.8 V vs. RHE, the j ECSA or SA greatly increased, as shown in Figure 4b. The SA 0 value was 1.60 mA cm −2 , which is about 2.4, 8.4 and 1.8-times as high as that for the Pt 3 Rh/C electrode with the initial activation between 0.05 and 1.2 V vs. RHE, the Pt/C electrode and the Pt 3 Ru/C electrode [24] probably due to the increased surface Rh content, respectively. It has been reported that electrochemical TL hydrogenation activity for pure Rh was lower than that for pure Pt [27,28]. Moreover, we have reported that the electrochemical TL hydrogenation activity for Rh-modified Pt/C catalysts, in which Rh did not form any alloys with Pt, was the maximum 2.7-times that for the commercial Pt/C catalyst due to the increase in the extent of the interface between Pt and Rh resulting from the modification of the Pt surface with Rh [49]. In contrast, the present Pt 3 Rh alloy was homogeneous, as expected from the XRD pattern (Figure 1), and the Rh content is about one-third of the Pt content, so the domains consisting of only Rh atoms will not have been formed in the alloy. Therefore, in the present study, the Rh constituent seems to contribute to the electronic modification of Pt atoms. In our previous study with Rh-modified Pt/C catalysts, the first step of electrochemical TL hydrogenation or one-electron reduction of H + to atomic hydrogen was anticipated to proceed on the Rh sites [49]. So, the surface Rh sites of the Pt 3 Rh alloy may also contribute to facilitating the first step.
To investigate the stability of the Pt 3 M/C catalysts during the initial activation and the following LSV measurement for electrochemical TL hydrogenation, bulk composition and surface composition of each catalyst after the LSV measurement were evaluated by EDX and XPS ( Figure S3, Supplementary Materials), respectively, and summarized in Table 2. For the Pt 3 M/C electrodes other than Pt 3 Au, both surface and bulk contents of the second metals decreased after the LSV measurement, and the decrease in the surface content was larger than that in the bulk content because the second metals were easier to oxidatively dissolve than Pt. Presumably, the oxidative dissolution of the second metals proceeded during the initial activation, not during the LSV measurement. For the Pt 3 Au/C catalyst, the Au constituent was not oxidized under the initial activation conditions, so the surface composition hardly changed during the initial activation.
The mean crystallite size of each Pt 3 M alloy after the LSV measurement for electrochemical TL hydrogenation was estimated from the XRD pattern ( Figure S4, Supplementary Materials), and summarized in Table 2. In all cases, the mean crystallite size increased after the LSV measurement. There are several styles of particle growth: (1) Ostwald ripening, (2) coalescence of metal nanoparticles via migration on the carbon support, and (3) metal nanoparticle agglomeration triggered by the corrosion of carbon support [50]. In the present study, the surface and bulk contents of the second metals in the Pt 3 M alloys were reduced by the oxidative dissolution of the second metals during the initial activation, so the particle growth seemed to be mainly due to the coalescence of metal nanoparticles via migration on the carbon support.
Mass activity (MA), cathodic current density per mass of Pt 3 M alloy, is also another measure of electrochemical TL hydrogenation activity. The higher MA leads to the smaller catalyst amount, so the higher MA is desirable from a practical point of view. To calculate MA, it is not suitable to use the initial loading mass of metals because the second metals other than Au can oxidatively dissolve during the initial activation. The metal loading mass after the LSV measurement (m a ) can be calculated using the metal loading mass and bulk composition of the as-prepared Pt 3 M/C catalyst and the bulk composition after the LSV measurement ( Table 2), assuming that only the second metal constituent in each Pt 3 M alloy dissolves [24]. The m a value of each Pt 3 M/C electrode is summarized in Table 3. In addition, the MA 0 and specific surface area (SSA) for each Pt 3 M/C catalyst were calculated using the ma value, and shown in Figure 5. As can be seen from Figure 5, the MA 0 value (107 mA mg −1 ) of the Pt 3 Rh/C electrode was the highest in the Pt 3 M/C electrodes, and higher than that (88 mA mg −1 ) of the commercial Pt/C electrode. The MA 0 value is a product of SA 0 and SSA. For SSA, the Pt/C electrode (456 cm 2 mg −1 ) was much larger than the Pt 3 Rh/C electrode (162 cm 2 mg −1 ), because the mean size (ca. 2.6 nm [49]) of Pt nanoparticles is smaller than that (4.4 nm) of Pt 3 Rh nanoparticles. For SA 0 , the Pt 3 Rh/C electrode was much larger than the Pt/C electrode. Overall, the increase in SA 0 outweighs the decrease in SSA, leading to the increase in MA 0 . Therefore, the electronic modification of Pt by alloying with Rh has a great influence on the high MA 0 for Pt 3 Rh/C electrode.
To discuss the reaction mechanism for the electrochemical TL hydrogenation on each Pt 3 M/C electrode, the Tafel slope was evaluated. Figure 6 shows the Tafel plots made from the LSVs in Figure 4, and the Tafel slopes are summarized in Table 3. Electrochemical TL hydrogenation is known to consist of the following elemental reactions [19,20].
C 6 H 5 CH 3 + 6H ads → C 6 H 11 CH 3 (8) Table 3, the Tafel slope of the Pt/C electrode was 39 mV dec , which was similar to that reported previously [49]. This strongly suggests that the rate-determining step of electrochemical TL hydrogenation on Pt/C is the second step (Equation (7)). Moreover, as can be seen from Table 3, the Tafel slope of each Pt3M/C electrode was similar to that of the Pt/C electrode, indicating that each Pt3M/C electrode has the same rate-determining step as the Pt/C electrode or the alloying of Pt with various second metals do not change the electrochemical TL hydrogenation mechanism.

Chemicals and Materials
Tetraoctylammonium bromide (N(Oct)4Br, 97.0%), tetrahydrofuran (THF, super dehydrated, stabilizer free, 99.5%), platinum chloride (PtCl2, 98.0%), rhodium chloride trihydrate (RhCl3•3H2O, 99.5%), iridium chloride (IrCl3, 97.0%), palladium chloride (PdCl2, The first step or one-electron reduction reaction of H + to atomic hydrogen (H ads ) is the same reaction as the Volmer reaction in HER, and the second step is the MCH formation reaction by adding the adsorbed H ads to TL. In the HER on a Pt electrode, if the Volmer reaction is the rate-determining step, the Tafel slope is 120 mV dec −1 , whereas if the Tafel reaction or the coupling reaction of two H ads is rate-determining, it is 30 mV dec −1 [51]. In Table 3, the Tafel slope of the Pt/C electrode was 39 mV dec −1 , which was similar to that reported previously [49]. This strongly suggests that the rate-determining step of electrochemical TL hydrogenation on Pt/C is the second step (Equation (7)). Moreover, as can be seen from Table 3, the Tafel slope of each Pt 3 M/C electrode was similar to that of the Pt/C electrode, indicating that each Pt 3 M/C electrode has the same rate-determining step as the Pt/C electrode or the alloying of Pt with various second metals do not change the electrochemical TL hydrogenation mechanism.

Characterizations of Pt 3 M/C Catalysts
To determine the metal loading of each catalyst, thermogravimetry (TG) was performed with the Thermo Plus TG8120 apparatus (Rigaku, Akishima, Japan) by heating the sample from room temperature to 800 • C in air at a rate of 1 K min −1 . Each sample was heated from room temperature to 800 • C in air at a rate of 1 K min −1 . The contents of Pt and M in at.% for each catalyst were evaluated by EDX analysis (VE-9800, Keyence, Osaka, Japan). The crystal structure of each catalyst was characterized by XRD measurements using the XRD-6100 diffraction instrument (Shimadzu, Kyoto, Japan) with Cu K α radiation (λ = 0.15405 nm, 50 kV, 30 mA). The particle size of each catalyst was characterized by TEM (JEM-2000FX, JEOL Ltd., Akishima, Japan). The size distribution profiles of the Pt 3 M nanoparticles for Pt 3 M/C were obtained by directly measuring the sizes of 300 random nanoparticles chosen from several TEM images. The electronic state of each catalyst was analyzed by performing XPS using the ESCA-3400 instrument (Shimazu, Kyoto, Japan) with Mg K α radiation (1253.6 eV, 10 kV, 20 mA). Binding energies of XPS spectra were calibrated by using Au 4f 7/2 peak of gold powder as 84.0 eV.

Preparation of Pt 3 M/C Electrodes and Membrane-Electrode Assemblies
Each catalyst ink was prepared by mixing the Pt 3 M/C powder, ultrapure water, 2-propanol and 5 wt % Nafion ® perfluorinated resin aqueous dispersion in a zirconia container (45 mL), and then ball-milling at 30 m s −2 for 2 h using a planetary ball-mill (P-7, Fritsch Japan Co., Ltd., Yokohama, Japan). The Nafion ® /Carbon support mass ratio in the catalyst ink was fixed to 0.8. The resultant catalyst ink was spread over the microporous layer side of the commercial carbon paper using a bar-coater (Matsuo Sangyo, K-coater), followed by drying it at 80 • C for 30 min to prepare a gas-diffusion electrode (GDE). The resultant Pt 3 M/C-loaded GDE is named Pt 3 M/C electrode hereafter.
To prepare a membrane-electrode assembly (MEA), a Nafion ® membrane (NRE212, 50 µm, Du Pont Co, Wilmington, DE, USA) was sandwiched between a Pt 3 M/C electrode as the cathode and a Pt/C electrode as the anode, and hot-pressed at 130 • C and 1.0 MPa for 3 min. The geometric surface areas of the cathode and anode were 5 cm 2 .

Electrochemical Hydrogen Pump TL Hydrogenation with Pt 3 M/C Electrodes
A commercial electrochemical cell (FC05-01SP, ElectroChem, Inc., Woburn, MA, USA) was used for electrochemical measurements. All electrochemical measurements were carried out at 50 • C. N 2 and H 2 gases of 0.1 MPa were fed into the cathode and anode compartments at a rate of 100 mL min −1 after being humidified at 70 • C, respectively. The anode was also used as the reference electrode (RHE). Each cathode was initially activated by 100 cycles of potential sweeps at 50 mV s −1 between 0.05 and 0.8 V vs. RHE or between 0.05 and 1.2 V vs. RHE with flowing the humidified N 2 gas. A CV was measured at 50 mV s −1 between 0.05 and 0.8 V vs. RHE to evaluate ECSA.
An electrochemical hydrogen pump TL hydrogenation system used in this study is shown in Figure 7. Neat TL was supplied to the cathode compartment at 5 mL min −1 .
LSVs were measured at 1 mV s −1 from an open circuit potential to −0.03 V vs. RHE to evaluate the activity of each catalyst for electrochemical TL hydrogenation. All potentials were corrected by IR drop.
were carried out at 50 °C. N2 and H2 gases of 0.1 MPa were fed into the cathode and anode compartments at a rate of 100 mL min −1 after being humidified at 70 °C, respectively. The anode was also used as the reference electrode (RHE). Each cathode was initially activated by 100 cycles of potential sweeps at 50 mV s −1 between 0.05 and 0.8 V vs. RHE or between 0.05 and 1.2 V vs. RHE with flowing the humidified N2 gas. A CV was measured at 50 mV s −1 between 0.05 and 0.8 V vs. RHE to evaluate ECSA.
An electrochemical hydrogen pump TL hydrogenation system used in this study is shown in Figure 7. Neat TL was supplied to the cathode compartment at 5 mL min −1 . LSVs were measured at 1 mV s −1 from an open circuit potential to −0.03 V vs. RHE to evaluate the activity of each catalyst for electrochemical TL hydrogenation. All potentials were corrected by IR drop.

Figure 7.
A schematic diagram of electrochemical hydrogen pump TL hydrogenation system.

Conclusions
In this study, the binary Pt3M (M = Rh, Au, Pd, Ir, Cu and Ni) alloy nanoparticleloaded carbon catalysts were prepared by colloidal method. The metal loadings and bulk compositions for the as-prepared Pt3M/C catalysts were close to the theoretical ones, suggesting the complete reduction of the Pt and M precursors. The XRD patterns of Pt3M/C catalysts other than Pt3Au/C were similar to of polycrystalline Pt with the fcc structure, and the lattice parameter in the former was smaller than that in the latter, suggesting that

Conclusions
In this study, the binary Pt 3 M (M = Rh, Au, Pd, Ir, Cu and Ni) alloy nanoparticleloaded carbon catalysts were prepared by colloidal method. The metal loadings and bulk compositions for the as-prepared Pt 3 M/C catalysts were close to the theoretical ones, suggesting the complete reduction of the Pt and M precursors. The XRD patterns of Pt 3 M/C catalysts other than Pt 3 Au/C were similar to of polycrystalline Pt with the fcc structure, and the lattice parameter in the former was smaller than that in the latter, suggesting that Pt and M in metallic phase for each catalyst were alloyed. The Pt 3 Au/C catalyst was segregated to Pt and binary Au-rich alloy phases. The mean crystallite size for the Pt 3 M/C catalysts was 3.0-4.8 nm, which was relatively close to each other. The crystallite size of each catalyst estimated from its XRD pattern was in good agreement with that estimated from the corresponding TEM image. In core-level spectrum of each constituent for the Pt 3 M/C catalysts, a peak shift was observed, suggesting the formation of the binary alloy.
When neat TL was supplied to the cathode compartment of the electrochemical hydrogenation cell, the E onset was more positive than 0 V vs. RHE or the standard potential of HER, suggesting that cathodic current density of each LSV was attributed to electrochemical TL hydrogenation. The SA 0 value was the highest for the Pt 3 Rh/C electrode, and decreased in the order of Pt 3 Rh/C > Pt 3 Au/C > Pt 3 Pd/C > Pt 3 Cu/C > Pt 3 Ir/C > Pt 3 Ni/C > Pt/C. This indicates that the alloying of Pt with the second metals is effective for improving electrochemical TL hydrogenation activity of Pt. Remarkably, for the Pt 3 Rh/C electrode, when the initial activation potential range was changed to 0.05-0.8 V vs. RHE, the SA 0 value was about 2.4, 8.4 and 1.8-times as high as that for the Pt 3 Rh/C electrode with the initial activation between 0.05 and 1.2 V vs. RHE, the Pt/C electrode and the previous Pt 3 Ru/C electrode, respectively due to electronic modification of Pt atoms with the Rh constituent.
After the LSV measurement, both surface and bulk contents of the second metal for the Pt 3 M/C catalysts other than Pt 3 Au decreased, and the decrease in surface content was larger than that in bulk content due to the oxidative dissolution of each constituent during the initial activation, whereas the mean crystallite size increased mainly due to the coalescence of metal nanoparticles via migration on the carbon support.
The MA 0 value of the Pt 3 Rh/C electrode was the highest in the Pt 3 M/C electrodes, and higher than that of the commercial Pt/C electrode. This is mainly attributed to the increase in SA 0 or the electronic modification of Pt by alloying with Rh. The Tafel slope of each activated Pt 3 M/C electrode was similar to that of the Pt/C electrode, indicating that each activated Pt 3 M/C electrode has the same rate-determining step as the Pt/C electrode or the alloying of Pt with various second metals do not change the electrochemical TL hydrogenation mechanism.

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