JES F OCUS I

Pt nanoparticle catalysts supported on carbon black coated with mixtures of Ru and Sn oxides (Pt/Ru-Sn oxide/C) have been screened for their activity and stoichiometry for ethanol oxidation. The compositions, loadings, and properties of the mixed Ru + Sn oxide support layers were varied by changing the concentration of base used during their deposition. Cyclic voltammetry at ambient temperature showed that Ru oxide alone provided a signiﬁcant increase in activity at low potentials over Pt/C, while a small amount of Sn provided a large additional beneﬁt. The most active catalyst had Ru and Sn loadings of 17% and 1.2% by mass, respectively. This was also the most active catalyst for ethanol oxidation at 80 ◦ C in proton exchange membrane electrolysis cells, where it provided four times more current than a commercial Pt/C catalyst at 0.25 V vs. a hydrogen evolving cathode. This is comparable to the performance of commercial PtRu/C, and the Pt/Ru-Sn oxide/C catalyst was more selective for the complete oxidation of ethanol to CO 2 . Thus, in direct ethanol fuel cells, Pt/Ru-Sn oxide/C catalysts would provide greatly increased current and power densities over Pt/C, and improved efﬁciencies over PtRu/C. ©

Although direct ethanol fuel cells (DEFC) are potentially an important future technology for many application, 1-4 they are currently impractical because of the low efficiencies of their anode catalysts and the production of incomplete oxidation products. 1,5 The development of catalysts for ethanol oxidation has focused on increasing activities to produce higher power densities and voltage efficiencies, but this has generally led to lower selectivity for the complete oxidation to carbon dioxide. [5][6][7][8] For example, PtSn based catalysts can provide attractive power densities, but the production of acetic acid and acetaldehyde as the main products results in low faradaic efficiencies and harmful emissions. 1,5 Early work on the development of electrocatalysts for ethanol oxidation focused on the use of PtRu, PtSn, and PtRuSn based materials, 9 while recent work has shifted toward PtRh and PtNi based systems. 7,[10][11][12] Alloying of Rh with Pt has been shown to promote cleavage of the C-C bond of ethanol, leading to increased selectivity for complete oxidation to CO 2 . 13,14 Activities have been increased by adding a third metal or an oxide phase, and control of the nanostructure. Sn, 15 Mo, 16 W, 16 Ru, 17 and Sb 18 oxides have been employed with Pt and/or PtRh nanoparticles.
The high conductivity and stability and RuO 2 makes it an attractive component for electrocatalytic systems, and its co-catalytic effect on the oxidation of CO and methanol is well known. [19][20][21] RuO 2 alone has been reported to have a small activating effect on Pt for ethanol oxidation, 17,22 and it can be used in combination with SnO 2 to prepare very active and stable ethanol oxidation catalysts. 23 Codeposition of SnO 2 with RuO 2 onto a carbon black support, followed by deposition of Pt nanoparticles, to produce a Pt/Ru-Sn oxide/C catalyst, has been shown to increase the low potential activity for ethanol oxidation relative to Pt black (higher voltage efficiency) and also increase the faradaic efficiency. 23 The purpose of the work reported here was to explore the effects of the synthesis conditions and composition of Pt/Ru-Sn oxide/C catalysts on their activity and stoichiometry for ethanol oxidation. The objective was to control the nanostructure of these materials in order to optimize the balance between activity at low potentials and selectivity for the complete oxidation of ethanol to CO 2 . Electrochemical measurements were made at ambient temperature in a conventional liquid electrolyte cell and in proton exchange membrane (PEM) electrolysis cells at 80 • C. Ultimately, this will allow optimization of the energy efficiency of DEFCs with Pt/Ru-Sn oxide/C anodes.
Preparation and characterization of the catalysts.-All of the catalysts were prepared by the reaction of KRuO 4 and SnCl 4 with carbon black in water or aqueous KOH, followed by deposition of Pt nanoparticles by reduction of H 2 PtCl 6 with NaBH 4 in the presence of sodium citrate. 23 Table I summarizes the conditions employed. In the catalyst descriptors, the targeted mass percent of the catalytic Pt nanoparticles is provided, followed by an identifier for the support (S x ) with a subscript related to the synthesis conditions. The support is taken to be the carbon + oxide composite, since these composites did not show significant activity for ethanol oxidation in the absence of Pt.
To deposit the mixed oxide, a mixture of KRuO 4 and SnCl 4 in aqueous KOH (or water) was added dropwise to a suspension of carbon black in water or KOH(aq). Following stirring for 24 h (or 30 min for S 110-30 ), the Ru-Sn oxide/C powder was collected by filtration, rinsed several times with water, and dried in air overnight. Pt nanoparticles were deposited onto an aqueous suspension of the Ru-Sn oxide/C support by adding H 2 PtCl 6 in 10 mL of water, stirring for 0.5 h, followed by dropwise addition of 50 mM trisodium citrate (12 mL), stirring for 1 h, dropwise addition of sodium borohydride (0.0902 g in 20 mL of water) and stirring for 3 h. The Pt/Ru-Sn oxide/C catalyst was collected by filtration, washed several times with water, and allowed to dry overnight. For comparison purposes, a Pt/Ru oxide/C catalyst was prepared in the same way as 25%Pt/S 110 , without SnCl 4 . Powder X-ray diffraction (XRD) measurements were made on a Rigaku Ultima IV using a Cu-Kα source (1.5418 Å). Transmission electron microscopy was carried out at the University of New Brunswick (The Microscopy and Microanalysis Facility) using a JEOL 2011 200 keV scanning transmission electron microscope.
Electrochemical measurements.-Cyclic voltammetry was carried out in a conventional three electrode glass cell with a Pine Instruments RDE4 Potentiostat. Glassy carbon working electrodes (0.071 cm 2 ) were coated with the catalysts as follows. The catalyst (2.0 mg) was dispersed in a mixture of water (120 μL), 1-propanol (30 μL) and Nafion solution (50 μL), and the resulting ink was sonicated for 1 h (Branson Ultrasonic Corporation Model 1510R-MTH, 70 W at 42 kHz). Then, 3.0 μL of ink was applied to a glassy carbon electrode that had been polished with 0.3 μm alumina. A calomel (SCE) reference electrode and Pt wire counter electrode were used. PEM cells.-Cells with nine separate 0.236 cm 2 anodes or a single 5 cm 2 anode, a single 5 cm 2 Pt black cathode, and a Nafion proton conducting membrane electrolyte (Nafion 117 for the 9-anode cell and Nafion 115 for the single anode cell) were used. 24 Each cell was based on a commercial (ElectroChem Inc.) PEM fuel cell. The anode current collectors in the nine-anode cell were graphite rods embedded in a Bakelite plate. Anodes were prepared by painting a catalyst ink onto carbon fiber paper (CFP; Toray TGP-H-090). For the 5 cm 2 anode, the CFP (wet proofed with 10 wt% PTFE; E-TEK, Inc.) was first coated with a carbon black layer prepared with an ink consisting of carbon black (10.2 mg) dispersed in a mixture of 1-propanol (100 μL) and Nafion solution (50 μL) by sonication for 3 h. The ink for the catalyst layer consisted of 41.3 mg of catalyst dispersed in 1-propanol (80 μL) and Nafion solution (200 μL). For the nine-anode cell, inks were prepared similarly, and painted onto CFP to give Pt loadings of 4 mg cm −2 and 20% Nafion by mass.
The applied potential was controlled with a MSTAT potentiostat from Arbin Instruments. The ethanol solution was supplied by a NE-300 New Era Pump Systems syringe pump. The cells were preconditioned at 0.7 V for one hour at the operating temperature. Polarization curves were then obtain from 0.9 V to 0.0 V in 50 mV steps. Each potential was held for three minutes, with the current recorded every second. The reported currents are averages over the final two minutes.
The nine-anode cell was operated in crossover mode with a 0.1 M ethanol solution pumped through the cathode flow field at 0.5 mL min −1 while the anode flow field was purged with 30 mL min −1 N 2 gas to prevent interference from oxygen. 24 The single-anode cell was operated in anode polarization mode, 25 with 0.1 M ethanol solution (0.2 or 0.5 mL min −1 ) at the anode and N 2 at the cathode. In both of these operating modes, the cathode approximates a dynamic hydrogen electrode, since the cathode reaction is H + + e − → 1 2 H 2 . 26 Product analysis.-Products and residual ethanol in the exhaust from the single-anode PEM cell were collected and analyzed as previously decribed. 25 Both the anode solution and the cathode gas were passed into a 50 mL vial cooled with a mixture of ice and dry ice. CO 2 in the N 2 stream was measured in real time with a commercial non-dispersive infrared CO 2 monitor (Telaire 7001). The concentrations of ethanol, acetaldehyde, and acetic acid in the solution collected in the cold trap were analyzed by 1 H-NMR spectroscopy (Bruker AVANCE III 300 MHz spectrometer), following mixing of 400 μL samples with 100 μL of D 2 O containing 32 mM fumaric acid as an internal standard. Chemical yields of CO 2 , acetic acid, and acetaldehyde are reported, based on their measured amounts (moles) and the amount of ethanol consumed.

Results and Discussion
Synthesis of the catalysts.-The reaction of KRuO 4 with carbon black to form a Ru oxide coating should be conducted under basic conditions to avoid precipitation of Ru oxide due to oxidation of water. A solution of KRuO 4 in 0.1 M KOH has been used in previous work, 27,28 and the previously reported method for preparation of Ru-Sn oxide/C employed a mixture of KRuO 4 and SnCl 4 in 0.1 M KOH. 23 However, a hydrous Sn(IV) oxide slowly precipitates from 0.1 M KOH due to hydrolysis. This would be expected to influence the composition, structure, homogeneity, and morphology of the resulting Ru-Sn oxide deposit, and thereby modulate the effect of the Ru-Sn oxide/C support on the catalytic activity of the Pt/Ru-Sn oxide/C catalyst. In order to investigate this, the KOH concentration of the precursor solutions was varied, as shown in Table I. The aim of these experiments was to optimize the composition, loading, and structure of the Ru-Sn oxide. It was thought that controlling the solubility of hydrous Sn oxide would significantly influence the amount of Sn 4+ that would be incorporated into the Ru oxide deposit, and that precipitation of hydrous SnO 2 could lead to a heterogeneous mixed oxide with discrete Sn oxide particles. The compositions of the catalysts, measured by ICP-OES, are listed in Table I. The catalyst name indicates the targeted Pt loading.
The supports (S x ) for the 25%Pt/S 110-30 , 25%Pt/S 110 and 70%Pt/S 110 catalysts were prepared following a previously reported procedure, 23 in which solutions of KRuO 4 and SnCl 4 in 0.1 M KOH were mixed and then added to a suspension of carbon black in water. The reaction time was 30 min for S 110-30 and 24 h for S 110 and all other supports. A black precipitate began to form when the KRuO 4 and SnCl 4 were mixed, and the resulting change in concentration of KOH (to 57 mM) when this mixture was added to the carbon suspension in water could cause further precipitation of Sn oxide during the reaction. In order to investigate whether this change in the KOH concentration produced a significant effect, the support for the 25%Pt/S 57mM catalyst was prepared with all of the precursors in 57 mM KOH. It can be seen from the Ru and Sn loadings reported in Table I that neither the increase in reaction time (relative to S 110-30 ) nor the dispersion of the carbon in water had a large influence on the Ru or Sn loading.
In order to increase the oxide loading, the amount of carbon was decreased by 40% for preparation of the low carbon (LC) support for 30%Pt/S LC . This resulted in a large increase in the Ru content of the catalyst and large increase in the Ru:Sn mass ratio (from 5   Table I, where very close to the target values, except for 25%Pt/S 110 and 25%Pt/S water which had higher measured Pt contents, presumably due to loss of oxide during Pt deposition. Apart from 70%Pt/S 110 and 30%Pt/S LC , which respectively had deliberately low and high Ru loadings, there were not large variations in the Ru mass%. Fig. 1 shows XRD of the 25%Pt/S 110-30 (a) and 25%Pt/S water (b) catalysts, and the S 110 support (c and d) without Pt. The diffraction peak at ca. 25 • in all of the spectra is due to the C(002) plane of the carbon black. The Pt peaks at 39.8 • , 46.6 • , 67.8 • , 81.8 • and 85.9 • for the Pt(111), Pt(200), Pt(220), Pt(311) and Pt(222) planes, respectively, are not significantly shifted from the values for pure Pt, indicating that there was not significant alloying with Ru and/or Sn. The mean Pt particle sizes were estimated from the width of the Pt(111) peak (Scherrer equation) to be ca. 3.5 nm for 25%Pt/S 110-30 and 4.3 nm for 25%Pt/S water . Mean particles sizes from XRD for the other catalysts are reported in Table I.

Characterization of the catalysts.-
Only the XRD pattern for 25%Pt/S water contained peaks other than for C and Pt, indicating that the Ru and Sn were present mainly as amorphous materials. Information about the nature of the Ru and Sn containing components was obtained by comparison of XRD patterns for S 110 before and after annealing at 200 • C for 24 h (curves d and c in Fig. 1). Prior to annealing, there was a broad peak at ca. 35.8 • that spans to the positions of the (101) peaks for pure SnO 2 and RuO 2 at 33.8 • and 35.0 • , respectively. This indicates that there were very small crystalline regions of these oxides, and/or a mixed oxide present. Annealing increased the height and decreased the width of this peak, indicating an increase in the size of the crystalline regions. The (101) peak shifted to 35.4 • for the annealed sample, indicating that it was predominantly due to RuO 2 . A new peak appeared at 54.4 • close to the position for the (211) peak for pure RuO 2 at 54.0 • . Since these peaks are not between the peaks for pure SnO 2 and RuO 2 , it appears that a crystalline mixed oxide was not formed in this case. 29 Since there are no peaks for SnO 2 , it must have been present in amorphous regions.
The XRD pattern for 25%Pt/S water in Fig. 1 does show small peaks in the Ru/Sn oxide regions, at ca. 34 • and 51.5 • , although they are very broad. These are close to the (101) and (211) peaks for SnO 2 at 33.8 • and 51.7 • , 29 respectively, which is consistent with the high Sn content of this catalyst (Table I) and the precipitation of hydrous Sn oxide at the neutral pH used for synthesis of the support. It can be seen in Fig. 1 that the predominantly SnO 2 peaks for 25%Pt/S water were at significantly lower angles than the predominantly RuO 2 peaks for S 110 .
A TEM image of the 30%Pt/S LC catalyst is shown in Fig. 2. It can be seen that the Pt particles (dark circles) were distributed reasonably uniformly over the support, while the morphology of the carbon black (agglomerates of ∼50 nm diameter spheres) was not changed significantly by the deposition of the oxides and Pt. This is consistent with the expectation that the Ru oxide would deposit uniformly as a thin layer over the carbon. 28 X-ray mapping by STEM (not shown) indicated that the Ru and Sn distributions mirrored the secondary electron image, but the resolution was insufficient to show whether the darker regions (10-20 nm) in Fig. 2 were due to thicker oxide deposits. The average diameter of the Pt particles from TEM was 5.0 ± 1.2 nm, which is similar to the average value of 6.4 ± 0.9 nm measured by XRD for two samples of 30%Pt/S LC prepared in the same way. with a Pt/Ru-Sn oxide/C catalyst in 1 M H 2 SO 4 (aq). Both voltammograms show broad redox waves at ca. 0.43 V that can be attributed primarily to the Ru(IV/III) couple of the oxide layer on the carbon. These are similar to the voltammograms reported for Ru oxide deposited in the same way on carbon nanotubes 27 and carbon fabric. 28 The main differences between the voltammogram of the catalyst and the support in Fig. 3 are in the −200 to +100 mV region where the catalyst shows well-defined peaks for hydrogen adsorption and desorption at the Pt nanoparticles. The charges passed during these processes are proportional to the electrochemically active area (0.21 mC cm −2 ), which allows the Pt utilization (percentage of the total area that is active) to be estimated. In order the measure this charge more accurately, the voltammogram of the support, after appropriate scaling, was subtracted from the voltammogram of the catalyst as shown in Fig. 3. The scaling factor was determined by trial and error to produce a difference "Pt" voltammogram (dotted line) that was most representative of Pt nanoparticles, focussing on the hydrogen and double-layer regions from −0.2 to +0.3 V. For this electrode, the electrochemically active area was 44 m 2 g Pt −1 and the Pt utilization was 45%. An electrode prepared with a similar loading of a commercial Pt/C catalyst also gave a Pt utilization of 45%. These are somewhat lower than state of the art values because of the high catalyst loadings (0.12 and 0.30 mg Pt cm −2 ) that were employed to enhance the ethanol oxidation currents. Fig. 4 compares voltammograms for ethanol oxidation at a Pt/Ru-Sn oxide/C catalyst, a Pt/Ru oxide/C catalyst, and a commercial Pt/C catalyst. These voltammograms have been normalized based on the mass of Pt, to account for differences in the Pt loading of each catalyst and the amount of catalyst applied to the electrode. On this basis, the peak currents at ca. 0.64 V on the forward scan are similar, and the differences at lower potentials are clearer. It can be seen that a Ru oxide layer on the carbon support (curve a) significantly decreased the onset potential for ethanol oxidation relative to the Pt/C catalyst (curve b) and increased the current greatly at potentials below 0.5 V. The peak at ca. 0.64 V was also broader than for Pt/C and shifted slightly to lower potential. These effects have been reported previously for a Pt/Ru oxide/glassy carbon electrode, and attributed to oxidation of the adsorbed CO intermediate through a bifunctional mechanism involving Ru-OH on the oxide support. 17 Incorporation of Sn into the Ru oxide layer (curve c in Fig. 4) resulted in a large additional enhancement of the ethanol oxidation current at low potentials (0.1 to 0.4 V), and this has been attributed to an enhanced electronic effect of the oxide on the Pt nanoparticles. 17 Electronic effects between metal nanoparticles and metal oxides have recently been reviewed, 30,31 and it has been shown that a SnO 2 support can increase the Pt 5d band vacancy and thereby promote ethanol oxi- dation under alkaline conditions. 32 The work reported here is focused on the use of mixed Ru-Sn oxide layers, rather than Ru oxide alone, since they provide higher electrochemical activities at low potentials. However, the results in Fig. 4 show that Pt/Ru oxide/C catalysts should also be considered for DEFCs.

Electrochemical measurements in a 3-electrode liquid
It should be noted that the voltammograms shown in Fig. 4 are all for the first cycle. In all cases, currents prior to the 0.64 V peak were lower on the 2 nd and subsequent forward scans. This has been previously documented and discussed for Pt nanoparticles on bare and oxide modified glassy carbon electrodes. 17 It has been attributed to the oxidation of pre-adsorbed species during the 1 st scan and instability of the oxide layers at high potentials. Since the effects of pre-adsorbed species should be similar for each catalyst, but stability is not, it is most instructive to compare the 1 st scans. Chronoamperometry at lower potentials has shown that the stability of the oxides is good over the potential range that would be encountered in a DEFC, and this was confirmed by the experiments in PEM cells reported below.
The reverse scans in the voltammograms in Fig. 4 are all quite similar (the higher peak at ca. 0.5 V for the Pt/Ru oxide/C catalyst is not significant based on the variability between electrodes prepared with the same catalyst). The peak is due to the oxidation of ethanol at the clean Pt surface that is formed when the oxide layer is reduced at ca. 0.5 V on the reverse scan (see the difference CV in Fig. 3). 33 The similarity of the reverse scan between the three catalysts therefore shows that the Pt nanoparticles are similar, and that the differences seen in the forward scan are due to differences in poisoning by adsorbed intermediates such as CO. Fig. 5 compares voltammograms for ethanol oxidation at Pt/Ru-Sn oxide/C catalysts with supports prepared under a variety of conditions. These have been normalized based on the electrochemically active area of the Pt. Only the first anodic scan is shown, which is most relevant to the use of these catalysts in DEFCs, where the anode potential should not exceed ca. 0.5 V vs. SCE. It can be seen that all of the Pt/Ru-Sn oxide/C catalysts provided much higher current densities than the commercial Pt/C catalyst from 0 to 0.4 V, and significantly higher currents at higher potentials. There were also large differences between the various Pt/Ru-Sn oxide/C catalysts in the 0 to 0.4 V region, with the 30%Pt/S LC catalyst providing the best performance up to 0.7 V. The lowest currents, at all potentials, were obtained with the 25%Pt/S 0.2M catalyst. The primary difference between 30%Pt/S LC and the other catalysts in Fig. 5 was the higher Ru loading, which is clearly beneficial. The primary difference between 25%Pt/S 0.2M and the other catalysts was the very low Sn content, which would be expected to results in low currents at low potentials based on the results in Fig. 4. It was therefore thought that increasing the Sn content by deposition of the oxide in the absence of base (S water ) would enhance activities at low potentials, but the result for 25%Pt/S water does not show a significant effect relative to 25%Pt/S 110-30 which had a similar Ru content and somewhat lower Pt loading. This, combined with the observation of SnO 2 peaks in the XRD of 25%Pt/S water , indicates that the Sn needs to be incorporated into the Ru oxide layer to be effective.

Ethanol electrolysis in a multi-anode PEM cell at 80 • C.-Al-
though cyclic voltammetry of ethanol in H 2 SO 4 (aq) at ambient temperature is useful for initial screening of catalysts, the results are not necessarily indicative of the performance that will be achieved in a DEFC. In order to investigate the correlation, selected Pt/Ru-Sn oxide/C catalysts were evaluated in a nine-anode PEM cell at 80 • C. The cell was operated as an electrolysis cell to avoid interference from oxygen, and in cross-over mode to compare activities under controlled mass transport conditions. 24 Thus, 0.1 M ethanol in water was supplied to the cathode and N 2 was passed through the anode flow field. Ethanol diffusing through the Nafion membrane was oxidized at the anode, while protons were reduced at the cathode to create a dynamic hydrogen pseudo-reference electrode (DHE). Evaluation of catalysts in this way allows the anode potential to be controlled much more accurately than in a DEFC, and the mass transport control provided by diffusion of ethanol through the membrane allows the number of electrons transferred (n av ) to be estimated. 24 Fig . 6 shows polarization curves for ethanol oxidation at 25%Pt/S 110 , 70%Pt/S 110 and 70% Pt/C anodes in a nine-anode PEM electrolysis cell. Three electrodes prepared with each catalyst were used, in order to statistically assess the differences between the activities of the catalysts. It can be seen that the onset of ethanol oxidation occurred at a significantly lower potential (ca. 0.15 V) for both Pt/Ru-Sn oxide/C catalysts relative to the commercial Pt/C catalyst (ca. 0.3 V), while the Pt/C catalyst provided higher currents at high potentials. The currents at the two Pt/Ru-Sn oxide/C anodes were not statistically different (t test at 95% confidence) at any potential, but the differences relative to the Pt/C anode were significant at 0.35 V, 0.4 V, and potentials above 0.5 V.
The lower onset of ethanol oxidation at the Pt/Ru-Sn oxide/C catalysts and higher currents at up to ca. 0.45 V are consistent with the voltammetric results in Figs. 4 and 5. This demonstrates that voltammetric screening provides a useful indication of fuel cell performance for these catalysts. It also shows that the catalysts have significant durability (>3 h of operation) under DEFC conditions. The lower currents at the Pt/Ru-Sn oxide/C anodes at potentials above 0.5 V indicate that the stoichiometry (n av ) was lower than at the Pt/C an- odes. The current for Pt/C in this region is limited by diffusion of ethanol though the cathode and membrane, and so is proportional to n av . 24 The decrease in this mass transport limited current with increasing potential has been shown to be due to a decreasing n av . 24,25 A similar trend is seen for the Pt/Ru-Sn oxide/C anodes.
Polarization curves for several other Pt/Ru-Sn oxide/C catalysts are shown in Fig. 7. The 30%Pt/S LC catalyst stands out here as the best, particularly over the 0.2 to 0.4 V potential range that is most relevant to DEFCs. Over this range it is statistically superior (95% confidence t test) to all of the other catalyst that were tested. This is consistent with the voltammetric results in Fig. 5, although the superiority is more pronounced in the PEM cell. The average current at the 30%Pt/S LC anodes was statistically higher than at the Pt/C anodes up to 0.45 V, but statistically lower at potentials above 0.5 V. This again indicates that n av was lower than at the Pt/C anodes.
Ethanol electrolysis and product analysis in a PEM cell at 80 • C.-Full evaluation of catalysts for ethanol oxidation requires measurement of the product distribution, in order to determine their faradaic efficiency and the amounts of acetaldehyde and acetic acid by-products. To do this accurately, the area of the anode needs to be relatively large. Fig. 8 shows polarization curves for ethanol electrolysis at 30%Pt/S LC and Pt/C anodes in a 5 cm 2 PEM cell, together with data for a commercial 75% PtRu/C catalyst (HiSPEC 12100 from Alfa  Aesar) from previous work. 25 In contrast to a previous report, in which a Pt/Ru-Sn oxide/C catalyst was evaluated with product analysis in a DEFC, 23 the same cell was operated in electrolysis mode here. This allows the anode potential to be accurately controlled and improves the accuracy of the product distribution. 25 It can be seen from Fig. 8 that the current at the 30%Pt/S LC anode was significantly higher than for the Pt/C anode at potentials below 0.5 V and that currents were similar a higher potentials. The enhancement in low potential performance for 30%Pt/S LC is similar to that previously reported for a 30%Pt/S 110-30 anode in a DEFC. 23 The performance of the 30%Pt/S LC anode was comparable to that of PtRu/C at low potentials and superior at potentials above 0.35 V. Table II reports the results of product analysis experiments at 0.45 and 0.50 V, together with data for a PtRu/C anode 25 that was obtained under the same conditions. The cell was operated at constant potential with monitoring the CO 2 in the combined anode and cathode exhausts, and collection of liquids in a cold trap. The average current during product collection is reported together with the chemical yields of each product. As observed in the polarization curves (Fig. 8), the current at 0.45 V in these longer timescale (ca. 50 min at each potential) experiments was higher at 30%Pt/S LC than at Pt/C, while the current at 0.5 V was higher at Pt/C. The sustained high currents observed for the 30%Pt/S LC catalyst in these experiments indicate that it has high stability at 80 • C under DEFC conditions (>15 h of operation over a 10 day period).
The product distributions show that the yield of CO 2 was lower at the 30%Pt/S LC anode than for Pt/C, while the acetic acid yield was much higher. The 30%Pt/S LC anode produced much less acetaldehyde and so would decrease harmful emissions from a DEFC. 4 From an efficiency perspective, the low CO 2 yields for the 30%Pt/S LC catalyst are partially offset by the higher acetic acid to acetaldehyde ratio. Stoichiometries (n av ) and faradaic efficiencies (ε f = n av /12) calculated from the product distributions 25 are included in Table II. It can be seen that n av , and consequently ε f , for the 30%Pt/S LC catalyst were both intermediate between the values for Pt/C and PtRu/C. Thus the increased electrochemical performance of 30%Pt/S LC over Pt/C comes at a lower trade-off in faradaic efficiency than for PtRu/C.
Overall, the 30%Pt/S LC catalyst represents a significant advance in the development of catalysts for DEFCs, with an increase in efficiency over a commercial high performance PtRu/C catalyst, and increase in low potential performance over Pt/C. In addition, there is much scope for improvement by further optimizing the composition and synthesis method for the Ru-Sn oxide/C, and the Pt deposition method. In particular, alloying of the Pt nanoparticles with Rh 13 and/or Ni 11 would be expected to increase the efficiency for breaking the C-C bond of ethanol.

Conclusions
The compositions, loadings, and properties of mixed Ru + Sn oxide layers on carbon black, deposited from solutions of KRuO 4 and SnCl 4 , can be varied by changing the concentration of KOH added. This consequently changes the activities for ethanol oxidation of Pt/Ru-Sn oxide/C catalysts prepared using these supports. Similar differences in electrocatalytic activities have been observed in both a liquid electrolyte cell at ambient temperature and polymer electrolyte electrolysis cells at 80 • C. Cyclic voltammetry showed that the increased low potential activities of Pt/Ru-Sn oxide/C over Pt/C were due to the presence of both Ru and Sn in the oxide support layer, with the highest activity corresponding to the highest Ru loading and a Ru:Sn atomic ratio of 12. The superiority of this catalyst was more pronounced in a PEM cell at 80 • C, where the onset potential for ethanol oxidation was >100 mV lower than at Pt/C and the current at 0.25 V vs. DHE was four time higher. These improvements in activity come at the cost of decreased selectivity for the complete oxidation of ethanol to CO 2 , although the best Pt/Ru-Sn oxide/C catalyst gave much higher CO 2 yields than PtRu/C. Overall, it can be concluded that Pt/Ru-Sn oxide/C catalysts are potentially better than both Pt/C and PtRu/C for use in direct ethanol fuel cells and ethanol electrolysis cells.