WO₃ decorated carbon nanotube supported PtSn nanoparticles with enhanced activity towards electrochemical oxidation of ethylene glycol in direct alcohol fuel cells

1 Introduction

There have been significant developments in the science and technology of low-temperature fuel cells during recent years, but there are still a number of fundamental problems to be resolved. Further progress is required, not only in fuel cell design, but also with regard to mechanistic aspects, the choice of electrode materials and their utilization, as well as the choice of fuel.

In recent decades, ethylene glycol (HOCH₂CH₂OH) has attracted considerable attention as a possible alternative fuel for low-temperature fuel cell applications, due to its easy transportation, low vapor pressure, low cost, convenience for fuel storage, lower flammability, and lower permeation through the polymer electrolyte membrane compared to methanol or ethanol (Serov and Kwak, 2010; An and Chen, 2016). Electrooxidation of ethylene glycol to CO₂ involves a complex mechanism, not only because the reaction produces 10 electrons per molecule, but due to the difficulty of breaking the C—C bond. In reality, it yields a number of partial oxidation products, including glycolaldehyde, glyoxal, glycolic acid, glyoxylic acid, and oxalic acid (Serov and Kwak, 2010; An and Chen, 2016; Dailey et al., 1998; Wang et al., 2009). It is believed that cleavage of the C—C bond is one of the factors that determines a fuel cell’s efficiency and its electrical energy yield. Besides electrochemical methods (such as cyclic voltammetry, linear sweep voltammetry, and chronoamperometry) spectroscopic measurements, such as Fourier transform infrared spectroscopy (FTIR) and differential electrochemical mass spectroscopy (DEMS), as well as combination methods such as in situ FTIR and online DEMS, have led to a better understanding of the mechanism of the electrooxidation of ethylene glycol (Dailey et al., 1998; Leung and Weaver, 1990; Chojak-Halseid et al., 2010; Wang et al., 2006a,b; Fan et al., 2004; da Silva et al., 2016; de Lima et al., 2003; Kadirgan et al., 1990; Demarconnay et al., 2007; Wang et al., 2011; Bayer et al., 2010).

Because of its stability and activity in contact with acid electrolytes, platinum is one of the most active catalytic metals for the oxidation of organic molecules, including the electrooxidation of alcohols at low and moderate temperatures. During the oxidation of all organic molecules, CO or CO-type (CHO) species are usually formed; they are strongly adsorbed on the surface of the Pt catalyst, thus blocking active sites and poisoning the catalytic material. Therefore, activation of the Pt catalyst by one or two additional metals, such as Ru, Sn, Mo, W, Rh, or their oxides, is often employed (Fashedemi et al., 2015; Chen et al., 2015; Lamy et al., 2002; Jusys et al., 2002; Li et al., 2013; Mann et al., 2006; Colmati et al., 2006; Li and Pickup, 2006; Neto et al., 2007; Miecznikowski and Kulesza, 2011; Miecznikowski, 2012). These metals are usually introduced as the alloying component, which leads to a significant enhancement of the electrooxidation of small organic molecules. This behavior could be related to the fact that the surface of the additional metals delivers a high population of OH species, which strongly participate in the removal of the intermediate CO species at much lower potentials than on bare Pt (Liang and Zhao, 2012). Moreover, the incorporation of extra transition metals may also lead to them participating in electronic modification by the “ligand effect”, resulting in a weaker binding energy of CO on the platinum. Consequently, the overpotential of the anodic process decreases (Kitchin et al., 2004).

The use of a suitable supporting material enables a high dispersion of metal nanoparticles onto its surface, which is particularly important with expensive noble metal catalysts, such as Pt and Pd. The appropriate supporting materials should be characterized by the following features: high surface area, corrosion resistance, and a stable surface to prevent the nanoparticles’ agglomeration. The substances most commonly used as supporting materials for various catalysts for fuel cell electrodes are carbon-based materials, as well as conductive oxides such as TiO₂, WO₃, ZrO₂, CeO₂, IrO₂ and SnO₂ (Miecznikowski and Kulesza, 2011; Miecznikowski, 2012; Shao et al., 2009; Baglio et al., 2013; Maiyalagan and Khan, 2009; Song et al., 2009; Anjos et al., 2006; Ou et al., 2011; Hepel et al., 2006; Kulesza et al., 2013; Baglio et al., 2014; Zignani et al., 2016; Sebastián et al., 2015). Carbon black (Vulcan XC-72) has been commonly used as a conventional supporting material for the Pt-based catalysts for fuel cell applications, due to its high surface area and conductivity, along with its low cost (Antolini, 2009). One drawback of the Vulcan support is that it is not very tolerant of fuel cell conditions, and is electrochemically oxidized to its surface oxides. In consequence, the oxidation process leads to the formation of large particles, resulting in a decrease in the fuel cell’s performance (Kangasniemi et al., 2004). To avoid these problems, carbon nanotubes (CNTs) have been introduced to replace carbon black (Vulcan) as supporting materials in electrocatalysis, due to their increased tolerance for carbon corrosion, and their excellent mechanical, electronic, and surface properties (Litster and McLean, 2004; Stevens and Dahn, 2003). Several previous papers have demonstrated an improved performance in the electrooxidation of small organic compounds and redox reactions when using carbon nanotubes in fuel cells, together with Pt-based nanoparticles, as supported electrode materials (Cui et al., 2011; Wang et al., 2011; Tang et al., 2007). Sieben et al. reported an improved performance in ethanol and ethylene glycol oxidation using Pt and PtSn nanoparticles supported on oxidized CNTs in sulfuric acid (Sieben and Duarte, 2011). Furthermore, recent studies have shown that commonly-used metal oxides, with CNTs and carbon black, are beneficial in improving the catalytic activity of supported Pd or Pt-based nanoparticles, or non-platinum catalysts (Ma et al., 2014; Miecznikowski et al., 2011). The presence of transition metal oxides in the immediate vicinity of Pt or Pt-based nanoparticles species leads to the generation of hydroxide groups at low potentials. These groups induce oxidation of passivating CO adsorbates on platinum sites; they can potentially break C—H bonds, as well as possibly weakening C—C bonds (Miecznikowski and Kulesza, 2011; Miecznikowski, 2012; Shao et al., 2009; Maiyalagan and Khan, 2009; Song et al., 2009; Anjos et al., 2006; Ou et al., 2011; Hepel et al., 2006; Kulesza et al., 2013). In particular, tungsten oxide (WO₃) has been applied as a support material or co-catalyst for fuel cells (Miecznikowski and Kulesza, 2011; Miecznikowski, 2012; Kulesza et al., 2013; Zeng et al., 2013; Trogadas and Ramani, 2008). It was reported that Pt supported on WO₃ demonstrates an excellent tolerance to CO, even at low potentials, such as ca. 0.1 V (vs. RHE) (Micoud et al., 2010).

Moreover, a positive effect was demonstrated for methanol electrooxidation in which WO₃ was introduced to platinum-based catalysts (Maiyalagan and Viswanathan, 2008; Tseung and Chen, 1997). Furthermore, PtRu catalysts supported on WO₃ displayed high activity towards the electrooxidation of methanol and ethanol (Cui et al., 2008; Jayaraman et al., 2005; Zhang et al., 2006; Zurowski et al., 2010; Barczuk et al., 2010; Lewera et al., 2011). Recently, an improved performance for formic acid oxidation utilizing WO₃ nanorods on Pd nanoparticles was also reported, which may be due to an increase in the catalytically active surface area, as well as metal–oxide interactions (Rutkowska et al., 2013). In our previous study, we reported that the catalytic activity of PtSn/C nanoparticles towards electrooxidation of ethanol was significantly enhanced through interfacial modification with an ultra-thin layer of tungsten oxide or a polyoxometalate (Zurowski et al., 2010; Barczuk et al., 2010; Lewera et al., 2011).

The purpose of this work was to investigate the fabrication and characterization of a catalyst for the electrooxidation of ethylene glycol in an acidic medium, based on WO₃ nanoparticles loaded onto multi-walled carbon nanotubes (MWCNT) to give a hybrid material, used as a support for PtSn nanoparticles, with the goal of obtaining an active, high surface area component of an electrocatalyst. The activating role of tungsten oxide may be due to the presence of OH groups on the oxide surface (Yang et al., 2004), which should facilitate oxidation of the poisoning CO intermediates (Tseung and Chen, 1997), and an increase in the electrochemically active surface area. The activating role may also be due to the existence of a hydrogen spillover effect at the interface between platinum and WO₃ (Kulesza and Faulkner, 1989; Shim et al., 2001). Our electrochemical diagnostic experiments involved cyclic voltammetric and chronoamperometric measurements. In situ FTIR spectroscopy was used to detect the formation of oxidation intermediates or products during the ethylene glycol oxidation. The morphology of the modified electrode surface was analyzed by using transmission electron microscopy (TEM), scanning electron microscopy (SEM-EDX), and X-ray powder diffraction (XRD).

2 Experimental section

All starting chemicals were of analytical grade quality and were used as received without further purification. Sodium tungstate (Na₂WO₄), a cationic exchange resin (Dowex 50 WX2-200), multi-walled carbon nanotubes (MWCNTs), SnCl₂·2H₂O, and Nafion (5 wt% solution in lower aliphatic alcohols) were supplied by Sigma-Aldrich. Hexachloroplatinic acid (H₂PtCl₆·6H₂O) was supplied by Alfa Aesar. Ethylene glycol, from Sigma-Aldrich, was used without further purification. Sulfuric acid (96%) and hydrochloric acid were from POCH, Poland. Aqueous solutions were prepared using double-distilled and subsequently de-ionized (Millipore Milli-Q) water. Argon was used to de-aerate the solutions and to keep an oxygen-free atmosphere over the solution during the measurements. The glass cell was cleaned in concentrated sulfuric acid for at least two hours before each experiment, and rinsed thoroughly with high purity water (resistance >18 MΩ cm).

For the preparation of multi-walled carbon nanotubes (MCNTs), modified tungsten oxide nanostructure films, and supported PtSn (PtSn/WO₃-MWCNT, and PtSn/MWCNT) materials, we used the following methods. The hybrid material (WO₃-MWCNT) was produced as follows. The precursor was a solution containing freshly prepared tungstic acid (made by the elution of 0.5 mol dm⁻³ Na₂WO₄ through a column filled with a proton exchange resin, Dowex 50 WX2-200) (Miecznikowski et al., 2011; Santato et al., 2001). A known quantity of MWCNTs (5 g) was dispersed in a 5 cm³ solution of tungstic acid, and the formation of nanostructured WO₃ films on the MWCNTs occurred through the sol–gel aggregation aging process. The resulting suspension was stirred for 24 h, and then was centrifuged and dried at 100 °C for 30 min in an oxygen atmosphere. The hybrid material obtained (WO₃-MWCNT) was used to support fabricated PtSn nanoparticles.

The procedure for MWCNT-modified WO₃-supported PtSn nanoparticles (PtSn/WO₃-MWCNT) and MWCNT supported PtSn were analogous to that described previously by Rojas (Garcia-Rodriguez et al., 2010). A mixture of SnCl₂ and H₂PtCl₆(6H₂O) was dissolved in de-ionized water and 3 mol dm⁻³ HCl, so as to give a Pt:Sn (atomic ratio) = 3:1. Then, an appropriate quantity of the WO₃-modified MWCNTs or MWCNT were added to the solution to obtain electrocatalysts with a metal loading of 20 wt%, and the solution was stirred for about 1 h. The reaction dispersion was then washed with water by rotating in a rotary evaporator, until the recovered water had reached a pH of approximately 6. Finally, the black dispersion was dried at 200 °C for 2 h, initially under an air atmosphere, and then under a reducing atmosphere (under flowing H₂/Ar, volumetric ratio 1:99). In that process, the nanostructured tungsten oxide film was first deposited on the surface of the MWCNTs, and secondly the PtSn nanoparticles were formed on the obtained hybrid material support, on which they interacted with the tungsten oxide, by the impregnation-reduction method. The metal loading of Pt₃Sn and the Pt-to-Sn ratio in a given catalyst system were evaluated using X-ray fluorescence (XRF); the resulting loading was approximately 20%, and the ratio was approximately 3-to-1 (Pt to Sn). The loading was fairly similar to that of the commercially available Vulcan-supported (Vulcan XC-72) 20% Pt₃Sn nanoparticles, which were used as reference PtSn nanoparticles.

To produce a suspension of WO₃-MWCNT supported PtSn nanoparticles (PtSn/WO₃-MWCNT) or MWCNT supported PtSn, approximately 5 mg of the catalyst was dispersed in 1 ml of water and the resulting suspension was magnetically stirred until a uniform suspension was formed. The catalyst film was fabricated through the dropping of approximately 5 μl of the suspension, using a micropipette, onto the surface of a glassy carbon disk electrode, and then letting it dry at room temperature. Before each fabrication of a modified electrode, the suspensions were treated in an ultrasonic bath for at least 10 min. After drying the catalyst layers, a drop (2 µl) of 0.2% Nafion solution was added onto the catalyst and allowed to dry. Nafion acts as the binding agent. The platinum loading of the catalysts was approximately 100 μg cm⁻². Prior to being used in electrooxidation processes, the modified electrodes were conditioned through the application of full oxidation/reduction cycles, at a scan rate, v, of 50 mV s⁻¹ between 0.0 V and 0.8 V in 0.5 mol dm⁻³ H₂SO₄, until steady-state currents were obtained.

All electrochemical measurements were performed in a three-electrode, single-compartment cell using a CH Instruments 750 A workstation. The base working electrode was made from glassy carbon with a surface area of 0.071 cm². A carbon rod was utilized as the counter electrode. All potentials in the present work were registered versus an Hg/Hg₂SO₄ saturated K₂SO₄ reference electrode, and later expressed versus the reversible hydrogen electrode (RHE). Current densities were calculated with respect to the electrochemical active surface area (S_A) of the PtSn catalysts. The electrochemical active surface area (S_A) of various PtSn catalysts were calculated according to the CO stripping analysis because the hydrogen adsorption-desorption peaks characteristic of the Pt active center overlap with the formation of hydrogen tungsten bronze and the reduction of WO₃. Furthermore, the S_A was calculated assuming a monolayer of adsorbed CO on the PtSn catalysts and measuring the charge required to oxidize this monolayer. The latter value was calculated by integrating CO stripping peaks and assuming that the coulombic charge was 420 μC cm⁻². The S_A values for the catalysts were 48, 47 and 45 m² g⁻¹ for bare PtSn/Vulcan, PtSn/MWCNT and PtSn/WO₃-MWCNT, respectively. It has found that there is no too much difference in the active area of PtSn in the presence and absence of WO₃. All experiments were performed in a thermostated cell at a temperature of 25 ± 0.5 °C.

The systems were systematically characterized to obtain information on particle morphology, composition, and crystal structure by various techniques, such as Scanning Electron Microscopy (SEM), High Resolution Transmission Electron Microscopy (HR-TEM), Energy Dispersion Spectroscopy (EDS), and X-ray diffraction (XRD). The SEM experiments were performed using a Carl Zeiss Merlin instrument with EDS analysis (Bruker Quantax 400, SEM-EDS). The HR-TEM measurements were performed using an instrument with an accelerating voltage of 200 keV. Samples for TEM measurements were prepared by depositing drops of diluted colloidal solutions of nanoparticles onto 400-mesh copper grids that supported a Formvar film (Agar Scientific); they were dried under ambient laboratory conditions (temperature, 22 ± 1 °C) for 24 h prior to TEM analysis. The crystallographic phase analysis was carried out by XRD using a Bruker D8 Discover system operated with a Cu X-ray tube (1.5406 Å) and Vantec (linear) detector (k = 1.5406 Å). The lattice parameter value and particle size were obtained from the position and the full-width at half-maximum (FWHM) of the (2 2 0) peak.

The in situ FTIR measurements were performed using a Nicolet iS10 spectrometer (ThermoScientific), equipped with a liquid nitrogen cooled MCT and an ATR VeeMAX III (PIKE Technology) accessory. The spectral resolution was 4 cm⁻¹ for each spectrum. The intensities of the spectra are presented in absorbance units defined as the reflectance ratio R/Ro, where R and Ro represent reflected IR intensities corresponding to the sample and reference single beam spectrum, respectively. The IR spectra were normalized to a spectrum recorded at 0.05 V in pure supporting electrolyte. A thin layer of catalyst was prepared by depositing ink onto a polished glassy carbon electrode with a diameter of 5 mm. A Ge hemisphere was used as the IR window, and the working electrode was located against the window, creating a thin solution layer with a thickness of a few micrometers. The incident angle of the IR radiation passing through the Ge window was 45°. Nitrogen was used to purge the electrolyte, while dry air was used to purge the spectrometer and chamber, reducing the spectral interference from CO₂ and water vapor.

For the preliminary single fuel cell measurement, a carbon paper supported anode electrode was prepared by brushing the appropriate ink composed of PtSn/WO₃-MWCNT or commercial PtSn/Vulcan catalysts with a loading of 2 mg cm⁻². The cathode consisted of a commercial 20% Pt/Vulcan catalyst, with a loading 2 mg cm⁻². The membrane-electrode assembly (MEA) was fabricated by hot-pressing the cathode and anode electrodes on either sides of a Nafion 117 membrane at 120 °C, under a pressure of 50 bar, for 120 s. The flow rate of 1 mol dm⁻³ ethylene glycol solution at the anodic compartment was set at 2.5 ml min⁻¹, and in parallel the oxygen gas flow rate at cathodic compartment was 200 ml min⁻¹. The fuel cell measurement was performed at 80 °C, under steady state conditions.

3 Results and discussion

The X-ray diffraction profiles of WO₃–MWCNT supported PtSn nanoparticles, PtSn/C, WO₃-MWCNT, and MWCNT are displayed in Fig. 1. The XRD patterns for PtSn/WO₃-MWCNT and PtSn/C exhibited diffraction peaks in the range 2θ = 24–26°, which were attributed to carbon-supported material with a (0 0 2) reflection plane for PtSn/C nanoparticles; the WO₃ reflections also arose in this (Hsu et al., 2000; Wang et al., 2007; Zignani et al., 2012). In the case of both PtSn/WO₃-MWCNT and PtSn/C (Fig. 1), diffraction peaks at the corresponding diffraction angles (39.9, 46.4, 67.7, and 81.6) occurred, which were attributed to the (1 1 1), (2 0 0), (2 2 0), and (3 1 1) planes, characteristic of the respective crystalline faces of Pt. A slight shift of the peak position at 40° towards lower degree values was observed for the PtSn/WO₃-MWCNT nanoparticles compared to pristine Pt/C (JCPDS# 04-0802) reflecting the influence of alloying (dos Anjos et al., 2008; Tanaka et al., 2005). While for the PtSn/WO₃-MWCNT nanoparticles (Fig. 1b), the other diffraction peaks appeared at 2θ = 34, 48, 55, and 61° (JCPDS no. 72-677); these are associated with the (1 0 1), (2 1 1), (2 2 0), and (3 0 1) reflections respectively, thus validating the formation of tungsten oxide (Choi et al., 2002). Moreover, the formation of peaks ascribed to the SnO₂ phase was not observed with either catalyst, since for PtSn/WO₃-MWCNT the signals from the SnO₂ and WO₃ peaks overlapped in the XRD profile, and most likely the SnO₂ signals were weaker than those of WO₃. No further effort was made to distinguish them and obtain quantitative structural information. To calculate the average particle size according to the Scherrer equation, the (2 2 0) reflections of Pt were used. The average Pt particle size obtained from XRD measurements for PtSn/WO₃-MWCNT and PtSn/MWCNT was of the order of 4–5 nm for both catalysts.

Close

Fig. 1 X-ray diffraction patterns for the MWCNT (a), WO3-MWCNT (b), PtSn/WO₃-MWCNT (c), PtSn/Vulcan (d) electrocatalysts.

Export to PPT

Fig. 2 displays representative TEM images and the size distribution of the PtSn nanoparticles on MWCNT decorated with WO₃. PtSn nanoparticles dispersed quite well on WO₃-decorated MWCNT, as can be seen from the TEM images (Fig. 2a), and had a relatively narrow distribution. The shapes of the Pt-based alloy nanoparticles were spherical or quasi-spherical. TEM characterizations, which reflected the analyses of several different portions of the catalysts, showed that the prepared PtSn nanoparticles had an average particle size of 5 nm (Fig. 2b). The values of the average particle sizes obtained using TEM analysis were in good agreement with those calculated above from the XRD results. High Resolution TEM images (Fig. 2c) were obtained to compare the lattice fringes of the PtSn/C and PtSn/WO₃-MWCNT.

Close

Fig. 2 Typical a) TEM and b) histogram of the nanoparticle size distribution and c) HR-TEM images of PtSn/WO₃-MWCNT nanoparticle.

Export to PPT

Initial electrochemical experiments were performed to characterize the PtSn/WO₃-MWCNT, PtSn/MWCNT and PtSn/C nanoparticles deposited on glassy carbon electrodes in the deaerated 0.5 mol dm⁻³ sulfuric acid supporting electrolyte. Typical results are presented in Fig. 3. The shape of the cyclic voltammetric curves for the PtSn/WO₃-MWCNT (Fig. 3a) were essentially different in the hydrogen adsorption and desorption region (between 0.0 V and 0.4 V), in comparison to PtSn/MWCNT and PtSn/C nanoparticles, because of the dependence on the surface composition. At higher potentials than 0.7 V, a platinum hydroxide/oxide redox transition was barely visible at the surface of the Pt-based nanoparticles, which is a characteristic of nanoparticles comprising Pt (Ribeiro et al., 2007). When the oxide is reduced in the backwards scan, Pt sites on the electrode surface are again available for the oxidation of fuel (e.g. ethylene glycol), and the current increases. In the range between the hydrogen adsorption/desorption and the oxide region, only the double layer capacitance current is present. The double layer capacitance of PtSn/WO₃-MWCNT and PtSn/MWCNT are smaller than that of PtSn/C, due to the presence of the carbon nanotubes support, reflecting the improved conductivity of supported hybrid materials. Furthermore, the voltammetric responses with carbon supported PtSn nanoparticles on the surface evidently show two well-defined adsorption/desorption peaks in the potential range (0.0 V–0.4 V vs. RHE), due to the oxidation and reduction of chemisorbed hydrogen on the intermetallic alloy phase. Performing the voltammetric experiment at the electrode surface that contains PtSn/WO₃-MWCNT results in one peak in this potential range. Such behavior is likely to be due to the overlap of the peak for the reduction of WO₃ with that for hydrogen adsorption/desorption on Pt-based catalysts. Comparison with the cyclic voltammograms of the electroactivity of WO₃-MWCNT in 0.5 mol dm⁻³ H₂SO₄ solution in the potential range between 0.0 and 0.9 V showed one peak at a potential lower than 0.2 V, which originated from the reversible reduction of tungsten oxide to hydrogen tungsten bronzes (Fig. 3c). It is apparent from Fig. 3 that there are no explicit distinguishing contributions from the redox process of tungsten oxide in the above-mentioned hydrogen adsorption/desorption peaks from the PtSn nanoparticles.

Close

Fig. 3 Cyclic voltammetric responses of a) PtSn/Vulcan (solid curve), b) PtSn/WO₃-MWCNT (dot curve), c) WO₃-MWCNT. Electrolyte: 0.5 mol dm⁻³ H₂SO₄. Scan rate: 10 mV s⁻¹.

Export to PPT

The electrocatalytic activities of the synthesized PtSn/WO₃-MWCNT nanoparticles and the commercial PtSn/C catalyst towards the oxidation of ethylene glycol were evaluated by cyclic voltammetry. Fig. 4 curve a shows representative cyclic voltammograms of PtSn/WO₃-MWCNT nanoparticles deposited on a glassy carbon electrode in a 0.5 mol dm⁻³ H₂SO₄ solution with 0.5 mol dm⁻³ ethylene glycol, at a scan rate of 10 mV s⁻¹. For comparison, the voltammetry of commercial PtSn/C nanoparticles is also shown (Fig. 4 curve b). The shape of the cyclic voltammetric curves of ethylene glycol oxidation represents the typical electro-oxidation process of alcohols, with two well-defined anodic peaks in the forward and backward scans in the investigated potential range. In the forward sweep, the ethylene glycol oxidation current density for PtSn/WO₃-MWCNT catalysts increased at 0.3 V, in order to reach the maximum at 0.82 V, which is almost located at the same potential seen with the commercial PtSn/C nanoparticles, as displayed in Fig. 4 curve b. During the reverse sweep, two peaks developed at 0.78 V and 0.55 V, which can be associated with the oxidative decomposition of by-products, including CO, glycol aldehyde, glycolate, glyoxylate, oxalate, and formate (Serov and Kwak, 2010; An and Chen, 2016; Dailey et al., 1998; Wang et al., 2009; Leung and Weaver, 1990; Chojak-Halseid et al., 2010; Wang et al., 2006a,b; Fan et al., 2004; da Silva et al., 2016; de Lima et al., 2003; Kadirgan et al., 1990; Demarconnay et al., 2007; Wang et al., 2011; Bayer et al., 2010).

Close

Fig. 4 Cyclic voltammetric responses for oxidation of 0.5 mol dm⁻³ ethylene glycol at a) PtSn/WO₃-MWCNT (dot line, b) PtSn/MWCNT (dash line) and c) PtSn/Vulcan (solid line). Electrolyte: 0.5 mol dm⁻³ H₂SO₄. Scan rate: 10 mV s⁻¹.

Export to PPT

In order to exclude the impact of the supported material on ethylene glycol oxidation, the cyclic voltammograms of the electroactivity of WO₃-MWCNT in 0.5 mol dm⁻³ H₂SO₄ solution, with and without ethylene glycol, in the potential range between 0.0 and 0.9 V were also recorded (not shown here). Those measurements showed that the glassy carbon electrode modified by WO₃-MWCNT does not exhibit any notable electrocatalytic behavior for the oxidation of ethylene glycol. As shown in Fig. 4, the peak current densities of ethylene glycol oxidation were greater than those when the reference catalyst (bare PtSn/C) was used. Furthermore, the ethylene glycol oxidation forward (anodic process) to backward (anodic peak current density) peak ratios were calculated for both catalysts, in order to evaluate the catalysts’ resistance to poisoning from ethylene glycol and its incompletely oxidized carbonaceous species. The ratios of the forward to the backward peak currents densities at 0.82 V, 0.81 V and 0.78 V for PtSn/C, PtSn/MWCNT and PtSn/WO₃-MWCNT were 1.0, 1.1 and 1.3, respectively. Our catalyst had a higher forward/backward peak current ratio, indicating that this catalyst was the more active toward ethylene glycol oxidation, and also that the catalytic centers were more tolerant of the adsorption of intermediate organic species. On the other hand, the lower forward-to-backward ratio for PtSn/C showed that this catalyst had a lower efficacy for the removal of adsorbed CO. These values were indicative of an enhancement by WO₃-MWCNT of the electrocatalytic activity of PtSn nanoparticles for the electrooxidation of ethylene glycol. Moreover, the onset potential of ethylene glycol oxidation at the electrodes modified with nanoparticles was determined. It was shown by the voltammogram that the onset potential of ethylene glycol starts at 0.34 V, 0.32 V and 0.29 V, for PtSn/C, PtSn/MWCNT and PtSn/WO₃-MWCNT nanoparticles, respectively. The onset potential of ethylene glycol was evaluated by a comparison of the anodic currents in the cyclic voltammetry in 0.5 mol dm⁻³ H₂SO₄, with and without ethylene glycol, at a scan rate of 10 mV s⁻¹. This potential was taken as the intersection of the background voltammetric current with an extrapolated line that was a linearization of the lower portion of the peak for the oxidation of ethylene glycol. Compared to the oxidation at PtSn/C, PtSn/MWCNT and PtSn/WO₃-MWCNT nanoparticles, the presence of WO₃-MWCNT caused a negative shift of the onset potential for ethylene glycol oxidation of ca. 50 mV. The low onset potential with the PtSn/WO₃-MWCNT nanoparticles was due to the facile formation of oxygenated species on the supporting material (WO₃-MWCNT) at a relatively lower potential in comparison to PtSn nanoparticles. According to the bifunctional mechanism, these oxygenated species assist the removal of CO_ad at lower potentials.

The evaluation of the stability and the electrocatalytic activity of the WO₃-MWCNT-supported PtSn nanoparticles as an electrochemical oxidation catalyst was performed through chronoamperometry measurements in a 0.5 mol dm⁻³ solution of ethylene glycol in 0.5 mol dm⁻³ H₂SO₄. The test was carried out at applied potentials of 0.3 V and 0.4 V in an unstirred solution. The lower value of the applied potential was selected as a value close to the onset potential for the oxidation of ethylene glycol, as determined previously in the voltammetry experiments, whereas 0.4 V is a value in the range that is suited for the potential application to fuel cells. On the other hand, from a practical point of view, direct alcohol fuel cells require catalytic systems that can become active at potentials that are as low as possible. The current-time curves are shown in Fig. 5.

Close

Fig. 5 Current – time responses at 0.3 V (a) and 0.4 V (b) for the oxidation of 0.5 mol dm⁻³ ethylene glycol at PtSn/WO₃-MWCNT (dot curve) and PtSn/Vulcan (solid curve). Electrolyte: 0.5 mol dm⁻³ H₂SO₄.

Export to PPT

Here, the current density was calculated using the geometric area of the glassy carbon electrode. The data in Fig. 5 confirmed the results of the cyclic voltammetric measurements; namely, that the electrochemical ethylene glycol oxidation on PtSn/WO₃-MWCNTnanoparticles had a greater catalytic activity than that on commercial PtSn/C nanoparticles. In the case of PtSn/WO₃-MWCNT nanoparticles on the electrode surface, the current density at an applied potential of 0.4 V was almost two times higher, in comparison to commercial PtSn nanoparticles at the near steady-state point of 3000 s. The gradual decrease of current density with time that was recorded at 0.3 V for carbon-supported PtSn nanoparticles may result from the poisoning of active sites on the Pt by various adsorbed organic intermediates. The incorporation of WO₃ on the surface of the MWCNTs may have one of the effects: firstly, it can keep the Pt site clean for the chemisorption of ethylene glycol by the formation and oxidation of hydrogen tungsten bronzes; alternatively, it can promote the formation of oxidation products, such as CO₂, that have a minor tendency to adsorb the active species.

To confirm the stability of our catalytic system, ICP-OES measurements were performed for an electrolyte solution (0.5 mol dm⁻³ H₂SO₄) before and after chronoamperometric (1 h) experiments (at 0.3 and 0.4 V) for both electrocatalysts. No leaching of platinum, tin, or tungsten metals was detected from either catalyst at room temperature in the given solutions.

The stability of the proposed catalysts was also investigated by cyclic voltammetry tests during the oxidation of 0.5 mol dm⁻³ ethylene glycol in 0.5 mol dm⁻³ H₂SO₄ in the potential range between 0.0 V and 0.9 V for a period of 2 h (scan rate 50 mV s⁻¹). In the case of the PtSn/WO₃-MWCNT electrode, the catalytic peak current density decreased by not more than 5% from the initial value after 200 cycles. When the same experiment was carried out for pristine PtSn/C nanoparticles, the decrease was 10%. Hence, the used of WO₃-modified MWCNT as supporting materials for PtSn nanoparticles did not make a significant difference to the stability of the peak current density, but the increase in the current density for the oxidation of ethylene glycol is an important outcome from the presence of hybrid supported materials (WO₃-MWCNT).

The electrooxidation of ethylene glycol yielded five incompletely oxidized intermediate species (glycol aldehyde, glycolic acid, glyoxal, glyoxylic acid, and oxalic acid) as well as CO₂ (Serov and Kwak, 2010; An and Chen, 2016; Dailey et al., 1998; Wang et al., 2009; Leung and Weaver, 1990; Chojak-Halseid et al., 2010; Wang et al., 2006a,b; Fan et al., 2004; da Silva et al., 2016; de Lima et al., 2003; Kadirgan et al., 1990; Demarconnay et al., 2007; Wang et al., 2011; Bayer et al., 2010). The typical cyclic voltammetric responses of these side-products in 0.5 mol dm⁻³ H₂SO₄ were examined with an electrode containing PtSn/WO₃-MWCNT nanoparticles and the commercial PtSn/C catalyst (Fig. 6). Glycolic acid was especially unreactive at all these electrodes in the investigated potential range (Fig. 6a). Such behavior suggests that it is likely to be one of the major products of ethylene glycol oxidation, even when the PtSn/WO₃-MWCNT nanoparticles are used in an acidic medium. The data for glycolic acid oxidation are consistent with the observation by Wang et al. (Wang et al., 2009) that the low reactivity of glycolic acid is due to its preferential adsorption on the Pt sides, where the reactive hydroxide group points away from the catalytically active Pt species. Moreover, it cannot be excluded that the presence of the OH group and the carboxyl group in the neighborhood may lead to oligomerization through ester formation that may play a role in the adsorption and oxidation process, by surface blocking and at least partly be responsible for the very low electrocatalytic activity of this molecule at potentials below 0.9 V. The shape of the cyclic voltammetric curve of oxalic acid (Fig. 6b) was slightly different to that for glycolic acid above 0.6 V. An anodic current that is not observed with glycolic acid as the test species was increased. When glyoxal (Fig. 6c) was used as the probe species, the current-voltage curve for PtSn/WO₃-MWCNT nanoparticles exhibited two anodic processes centered at about 0.66 V and 0.75 V, and an onset potential at 0.5 V. The process at 0.66 V was attributed to the oxidation of CO to CO₂, and the peak at 0.75 V resulted from the oxidation of glyoxal to intermediate species, perhaps glyoxylic acid and oxalic acid (Dailey et al., 1998). Also, voltammetry of glyoxylic acid at the PtSn/WO₃-MWCNT catalyst was performed (Fig. 6d). The shape of the anodic cyclic voltammetric curve was generally similar to that seen with glyoxal. The main features were two sharp peaks that appeared in the potential range of 0.5 V–1.0 V, and an onset potential of 0.45 V. The peak at the lower potential, centered at 0.8 V, was probably due to the oxidation of CO to CO₂. The peak at 0.95 V was probably due to the oxidation of glyoxylic acid to oxalic acid and CO₂. Noticeably, some oxidative decomposition of glycol aldehyde (Fig. 6e), glyoxal, and glyoxylic acid took place on PtSn/WO₃-MWCNT nanoparticles at a lower potential (at 0.45 V) in comparison to the reference PtSn/C nanoparticles. We need to note that none of the catalysts utilized in this study promoted the oxidation of glycolic acid in the investigated potential range. The same cyclic voltammetry measurements were also performed for PtSn/MWCNT nanoparticles; the analogous behavior was observed.

Close

Fig. 6 Cyclic voltammetric responses of oxidation of 0.5 mol dm⁻³ glycolic acid (a), oxalic acid (b), glyoxal (c), glyoxylic acid (d), glycol aldehyde (e), and respectively on PtSn/WO₃-MWCNT (dot curve) and PtSn/Vulcan (solid curve). Electrolyte: 0.5 mol dm⁻³ H₂SO₄. Scan rate: 10 mV s⁻¹.

Export to PPT

To detect the formation of oxidation intermediates or products during the ethylene glycol oxidation, in situ FTIR spectroscopy, in parallel with electrochemical measurements, was carried out on PtSn/WO₃-MWCNT nanoparticles and PtSn/C. These results are presented in Fig. 7. The reference spectrum was obtained in 0.5 mol dm⁻³ H₂SO₄ at 0.05 V. As can be seen in this figure, the band at 2345 cm⁻¹ corresponded to the asymmetric CO₂ stretching mode formation during full ethylene glycol oxidation. The inset in Fig. 7 shows the integrated band intensity, corresponding to the formation of CO₂ (2345 cm⁻¹) and carboxylic acids (1740 cm⁻¹) as a function of the potential applied during the electrooxidation of ethylene glycol, which was extracted from the spectra displayed in Fig. 7. It can be observed for the formation of CO₂ that in the initial potential region (0.0–0.3 V), the integrated band intensity located at 2345 cm⁻¹ remains at a constant small value, but in the second potential region (>0.4 V) it starts to increase with the potential. The integrated band intensity located at 1740 cm⁻¹ for carboxylic acid production starts to increase at potential higher than 0.5 V with these PtSn/WO₃-MWCNT and bare PtSn/C nanoparticles. The data indicate that, the formation of carboxylic acids seem to begin later than CO₂ and started earlier than on Pt (Arán-Ais et al., 2014).

Close

Fig. 7 IR spectra recorded during oxidation of ethylene glycol on PtSn/WO₃-MWCNT (a) and PtSn/C (b). The reference potential is 0.05 V. Inset plot shows the integrated band intensity of CO₂ and carboxylic acids as a function of the potential.

Export to PPT

The data indicate that, at least to some extent, ethylene glycol is liable to be completely oxidized on the PtSn surface at a potential higher than 0.4 V. The observed bands at 2050 cm⁻¹ and 1840 cm⁻¹ were assigned to linearly adsorbed CO (Leung and Weaver, 1988; Schnaidt et al., 2011). An additional band at 1650 cm⁻¹ was observed, which was due to the HOH- bending vibration of molecular water. The two bands around 1740 cm⁻¹ and 1250 cm⁻¹ corresponded to the C⚌O and C—O stretching modes of a carboxylic group. This functional group could be assigned to either glycolic acid or oxalic acid, but it was not possible to identify the appropriate acid. Another two bands detected at about 1210 and 1100 cm⁻¹ were related to the adsorption of bisulfate (Schnaidt et al., 2011). The in situ FTIR results are shown in Fig. 7; these data confirmed that the electrochemical oxidation of ethylene glycol on PtSn/WO₃-MWCNT led to the formation of CO₂ below 0.45 V, as a consequence of C—C bond cleavage in ethylene glycol. Moreover, the characteristic peaks that originated from the aldehyde and acid groups suggested partial incomplete oxidation. Further research is necessary to develop catalytic systems capable of more effective C—C bond cleavage in ethylene glycol.

Furthermore, a single ethylene glycol fuel cell performance of PtSn/WO₃-MWCNT catalyst and pristine PtSn/Vulcan nanoparticles were compared. Fig. 8 exhibits the power density curves and suitable polarization curves for these PtSn/WO₃-MWCNT and bare PtSn/C nanoparticles. The obtained result of fuel cell measurements showed that employing PtSn/WO₃-MWCNT as the anode catalyst reached the open-circuit voltages (OCV) of 0.66 V, whereas the anode composed of bare PtSn/Vulcan achieved lower value around 0.61 V at the operating temperature. Moreover, the single fuel cell with the PtSn/WO₃-MWCNT anode characterized substantially better performance than the pristine PtSn/Vulcan, with a current density of around 79.8 mA cm⁻² and an output power density of 20.5 mW cm⁻². A comparison of the fuel cell performance with the literature is not easy because of the diversity of operating conditions (pressure, temperature, and concentration of fuel) and the various materials used for both the cathode and anode. Despite the many different factors involved, the obtained power density found in this work is located in the midrange of the results reported in the literature. The lowest maximum power density found was achieved for a PtRu anode (10 mW cm⁻²), (Chetty and Scott, 2007) and the highest one was recorded for PtRu on NP-PCM (Livshits and Peled, 2006). These data are in good agreement with the above results that were obtained by diagnostic measurements (e.g. cyclic voltammetry, and chronoamperometry), and confirm the efficacy of the WO₃-MWCNT in enhancing the catalytic activity of PtSn. Further research is necessary to focus on the qualitative and quantitative analysis of product distribution (e.g. DEMS), in order to complete our understanding concerning the mechanisms of ethylene glycol electrooxidation.

Close

Fig. 8 Power density and polarization curves of the MEA with bare PtSn/Vulcan (dot line and circle) and PtSn/WO₃-MWCNT (solid line and black squares) anodes. Catalyst loading in all electrodes: 2 mg cm⁻². Cathode: pure oxygen flow rate 200 ml min⁻¹. Anode: 1 mol dm⁻³ ethylene glycol in 0.5 mol dm⁻³ H₂SO₄ concentration aqueous solutions, 2.5 ml min⁻¹.

Export to PPT

4 Conclusions

The utility of WO₃ nanoparticles loaded onto multi-walled carbon nanotubes (MWCNT) as a hybrid material for the support of PtSn nanoparticles has been demonstrated by the oxidation of ethylene glycol, with the goal of obtaining an active, high surface area component of an electrocatalyst. The oxidation of ethylene glycol with synthesized PtSn/WO₃-MWCNT and PtSn/MWCNT nanoparticles and commercial PtSn/C has been compared. The results of cyclic voltammetry showed that PtSn/WO₃-MWCNT enhanced the catalytic activity, in that higher current densities and lower onset potentials relative to the reference nanoparticles were observed. The efficacy of the WO₃-MWCNT is likely to be related to better conductivity. Moreover, the presence of tungstate, which has a high affinity for OH groups, contributes to a negative shift in potential for the oxidation relative to that on Pt, thereby participating in the removal of poisoning species (e.g. CO) from the platinum surface. Chronoamperometry at 0.3 and 0.4 V confirmed the result of the voltammetry. The reactivity of the intermediate species at electrode surfaces that contained PtSn/WO₃-MWCNT was also tested by cyclic voltammetry. The data illustrated that glycol aldehyde, glyoxal, and glyoxylic acid are electrochemically reactive by-products formed in the course of ethylene glycol oxidation. In situ FTIR spectroscopy was used to probe the reaction to determine the formation of oxidation intermediates or products during the ethylene glycol oxidation. In summary, the supported nanoparticles PtSn/WO₃-MWCNT are a promising catalyst for ethylene glycol oxidation in acidic electrolytes