Effect of acid functionalised carbon supports for Pd–Ni–Sn catalyst on ethanol oxidation reaction

Abstract

The sodium borohydride reduction method was employed to prepare a ternary catalyst with an atomic ratio of 20%Pd10%Ni10%Sn impregnated on four carbon supports, Vulcan XC-72R carbon black (CB), functionalised carbon black (CB_sn), multi-walled carbon nanotubes (MWCNTs) and functionalised MWCNTs (MWCNT_sn), to improve the catalytic activity for direct ethanol fuel cells (DEFCs). The acid functionalisation was performed using H₂SO₄ and HNO₃. The addition of Ni and Sn metals helped to increase catalytic activity as well as catalyst stability in the ethanol oxidation reaction (EOR). The XRD and XPS results confirmed the presence of Ni(OH)₂ and SnO₂ compounds. The crystalline size was in the range of 6.4 to 7.4 nm for all samples. TEM images of the catalyst on MWCNT_sn confirmed that the catalyst particles were uniformly distributed and dispersed with a particle size in the range of 6 nm. This result was consistent with the highest number of active sites investigated by CO chemisorption. This was due to the highly crystalline structure of MWCNT_sn. Moreover, the catalyst on MWCNTs exhibited the highest current density (291 mA cm⁻²) for EOR, great catalyst stability, the highest electrochemical surface area (77.155 m² g⁻¹ Pd) and excellent CO tolerance.

---

1 Introduction

Nowadays, the world’s demand for energy consumption is dramatically increasing, while the major source of energy coming from fossil fuels is limited and depleting at a fast rate. The use of fossil fuels also raises serious environmental concerns; hence the development of alternative and renewable energy technologies is needed. Among others, fuel cell technology has the potential to contribute to a sustainable and emission-free transport and energy system. A fuel cell produces electricity via electrochemical reaction but without combustion as in conventional combustion processes. The proton exchange membrane fuel cell (PEMFC) is one of the most widely used types of fuel cells being developed for stationary and portable applications, since it offers several benefits, for example, low operating temperature and pressure, low catalyst loading, and high durability.¹ There are various types of fuels being used in PEMFCs, such as hydrogen gas, formic acid, methanol, ethanol, etc. Commercially viable fuel cell systems use hydrogen or direct methanol as a fuel depending on the application. The conventional hydrogen fuel cell is usually used to deliver high power density, whereas the direct methanol fuel cell (DMFC) is more suitable for low power applications.² However, ethanol appears to be the preferred candidate fuel over methanol as the energy density and specific energy of ethanol are much higher.^3,4 Moreover, ethanol can be produced at low cost and in large quantities by fermentation of domestic biomass. For these reasons, a direct ethanol fuel cell (DEFC) has become a promising renewable and clean energy device with less issues with catalyst poisoning and fuel crossover when compared to a DMFC.⁵

The kinetic rate of ethanol oxidation reaction in alkaline DEFCs is higher than that of DEFCs in acid media.⁶ Thus, non-noble metal catalysts such as nickel and silver can be utilized along with low-cost noble metals such as palladium (Pd) and ruthenium (Ru) as anode catalysts.⁴ However, electrode poisoning often associated with these anode catalysts is a critical issue in a DEFC. The catalyst instability is caused by the adsorption of CO intermediate generated during the ethanol oxidation reaction (EOR). A considerable amount of work has been carried out to find alternative alloys of non-noble and noble metals which yield high EOR activity and durability.^7–9 High selectivity of the anode catalyst is also important as complete oxidation of ethanol to CO₂ is desired to produce the most electrons from one molecule of ethanol.¹⁰

Pd, one of the most abundant noble metals in nature and cheaper than platinum (Pt) by 50%, has been recognized as the most efficient catalyst for EOR in alkaline media.^11–14 Thus, the metal and its alloys serve as substitutes for Pt in catalytic reactions. However, its catalytic stability and activity need to be improved. This can be achieved by introducing binary or ternary catalysts such as Au, Ni, Ru, Sn, Ir, Bi, etc.^15–28 It has been reported that Ni, Au, or Ru could be used to increase the activity and the stability of Pd while Sn, Ir, or Bi have been used to facilitate the desorption of CO-intermediate during the EOR.²⁹

The type of catalyst support is also a key factor to enhance the catalytic activity.^17,30 In general, carbon black is the most widely used catalyst support in DEFCs.^12,31–33 The average size of the primary particles obtained from pyrolysis or incomplete combustion of carbon compounds is normally in the range of 10–50 nm.³⁴ The BET surface area (S_BET) is in the range of 10–1500 m² g⁻¹, depending on the method to produce the carbon black. There are many types of carbon black commercially available, such as acetylene black, Vulcan XC-72, and Ketjenblack. Apart from carbon black supports, varieties of carbon nanotubes are also exploited, including single-walled (SWCNTs) and multi-walled carbon nanotubes (MWCNTs). The unique structure and particular morphologies of nanotubes make them more attractive as catalyst supports for fuel cells. They have high surface area, corrosion resistance, good electro-conductivity, high stability, meso-porosity, and high degree of crytallinity.^35–38 As a result, they have exhibited excellent performance when used as catalyst supports. Treatment with HNO₃ and H₂SO₄ was carried out following Osorio et al. procedure.³⁹ They treated carbon nanotubes in different acids and found that the carbon nanotubes treated by nitric–sulfuric acid and nitric–sulfuric–hydrochloric acid showed good support characteristics.

Based on the preceding, Pd-based catalysts with Ni and Sn co-catalysts were developed in this present work to increase the catalyst efficiency for DEFC for the first time using the sodium borohydride reduction method and also to reduce the cost of the catalysts. Moreover, the developed catalysts were prepared on functionalised and non-functionalised carbon based supports. Ternary catalysts with an atomic ratio of 20%Pd10%Ni10%Sn were initially synthesized on different types of carbon supports: (i) Vulcan XC-72R carbon black, (ii) MWCNTs, (iii) functionalised Vulcan XC-72R carbon black, and (iv) functionalised MWCNTs. The catalyst ratio was studied first to select the most suitable ratio. Physical properties of these carbon supports were examined using temperature program reduction (TPR), chemisorption analysis, Raman spectroscopy, Fourier transform infrared spectroscopy (FT-IR), thermogravimetric analysis (TGA) and scanning electron microscopy-energy dispersion spectroscopy (SEM-EDX). In order to access their suitability for use in DEFC, 20%Pd10%Ni10%Sn deposited on different supports were investigated using X-ray diffraction (XRD), transmission electron microscopy (TEM), SEM-EDX, X-ray photoelectron spectroscopy (XPS), and CO-pulse and TPR chemisorption analysis. Additionally, several modes of voltammetry were performed to explore their catalytic activity, CO tolerance, durability, and electrochemical surface area.

2 Materials and methods

2.1 Materials

The metal precursors for the anode electro-catalysts were prepared from palladium chloride (PdCl₂), nickel chloride (NiCl₂), and tin chloride (SnCl₂), which were purchased from Merck. Sodium borohydride (NaBH₄) was purchased from HiMedia. Carbon black (Vulcan XC-72R) was supplied by Cabot Corporation. Ethanol was purchased from RCI, and potassium hydroxide (KOH) was purchased from Ajex. HNO₃ and H₂SO₄ were purchased from J. T. Backer and Merck, respectively.

2.2 Functionalisation of carbon supports

Firstly, 1 g of carbon support (either Vulcan XC-72R or MWCNTs) was mixed with 50 ml of a mixed concentrated H₂SO₄ : HNO₃ (3 : 1) solution in a closed vessel under sonication, at room temperature for 2 h. After that, the reaction was left undisturbed for another 15 h. The suspension was washed with deionised water until the effluent had a pH of 4–5. Finally, the treated support was dried in a convection oven at 100 °C overnight.

2.3 Physical characterisation of carbon supports

Characterisation of surface area and characterisation of pore size distribution of different carbon supports were carried out by TPR analysis (BELCAT-B). The sample was initially degassed at 300 °C for 2 h under vacuum prior to the N₂ adsorption measurement. Raman spectroscopy (Renishaw Raman Spectroscopy System 2000) was used to investigate the resulting chemical structure and the defects of the carbon supports after functionalisation. Raman spectra were obtained at ambient temperature with a green laser wavelength of 532 nm, radiated from an argon laser with a power of 2 mW. In addition, FT-IR (Thermo Nicolet 6700) was used to diagnose the functional groups obtained from the functionalisation. Each sample was mixed with KBr and was then analysed at wave numbers of 400 to 4000 cm⁻¹. In order to analyse the quantity of functional groups on the supports, TGA (Mettler Toledo TGA/SDTA 851^e) was performed at scanning temperatures from 70 to 850 °C, at a ramp rate of 5 °C min⁻¹, and under air flow conditions. Brauner, Emmet and Teller (BET) surface area was also investigated using an Autosorp-I to find specific surface areas of the prepared supports. The samples were degassed at 300 °C for 2 h under vacuum before the nitrogen adsorption measurement. Finally, SEM-EDX (FEI Quanta 400 & EDAX) was employed to assess the surface morphology and elemental compositions of supports.

2.4 Catalyst preparation

The 20%Pd10%Ni10%Sn by weight was prepared from the metal precursors – PdCl₂, NiCl₂ and SnCl₂ – on carbon supports. Firstly, the metal precursors were added to deionised water and vigorously stirred. One gram of carbon support was then suspended in the precursor solution for 5 min using a sonicator. After that, an excess amount of NaBH₄ solution was added under vigorous stirring for 6 h. The mixture was then centrifuged and washed with deionised water several times. The ternary catalyst impregnated on the carbon support was finally dried in a convection oven at 70 °C for 20 h. This method was applied to all types of prepared catalysts.¹³

2.5 Physical and chemical characterisation of catalyst impregnated on carbon support

Firstly, XRD (Philips X’pert) was used to investigate the crystallization and phase of the catalysts. The XRD patterns were obtained using a Cu K_α source operating at 45 kV and 30 mA with a scan rate of 0.04° s⁻¹. Additionally, TEM (JEOL 2010) was used to investigate the particle size and the dispersion of catalyst particles using a LaB₆ filament at 200 kV. SEM and EDX were also employed to find the morphology of the catalyst surface and metal loading ratio, respectively. The form of the resulting ternary catalyst was determined using XPS (AXIS Ultra DLD) operating in the lens-defined mode swept 3 times, at a pass energy of 20 eV, a maximum X-ray source power of 150 W, and an energy step size of 100 meV. A gas sorption chemisorption analyser was employed to perform TPR and CO pulsing analysis. The starting and ending temperatures of hydrogen reduction were firstly defined by TPR analysis using 5% H₂ in argon media at a controlled flow rate of 50 ml s⁻¹ and at a wide range of temperatures (from 0 to 650 °C). Next, the CO-pulse method was used to provide the number of active sites and metal dispersion. The CO-pulse steps include the following: (i) preheating at 400 °C for 25 min, (ii) oxidation with 99.999% O₂ for 15 min, (iii) purging with He for 15 min, (iv) reduction with pure H₂ gas for 15 min, (v) purging with He for 15 min, (vi) cooling to 50 °C, and (vii) injection of 0.89 ml of 10% CO–He until the change in adsorption peaks was below 2%.

2.6 Electrochemical characterisation of catalyst impregnated on carbon support

The electrochemical properties of 20%Pd10%Ni10%Sn impregnated on different carbon supports were investigated using a potentiostat and galvanostat (Autolab PGSTAT100) to perform the following measurements: catalytic activity, durability, CO tolerance, and electrochemical surface area. All the measurements employed a three-electrode system operating at room temperature. For the characterisation of catalytic activity, cyclic voltammetry (CV) was carried out in a mixture of 1 M ethanol and 1 M sodium hydroxide aqueous solution. The working electrodes (WEs) were composed of 20%Pd10%Ni10%Sn impregnated on different carbon supports coated on a gas diffusion layer (Toray Carbon Paper, TGP-H-60) with an active surface area of 0.28 cm² (Pd metal loading 0.0680 mg cm⁻²). In the process of WE fabrication, the catalyst ink was prepared from 5 mg of catalyst powder mixed with 1 ml of 5% wt Nafion® 117 solution and 1 ml of 5 M ethanol solution. The ink was sonicated for 1 h to provide a homogeneous suspension. The resulting catalyst ink was deposited on the Toray carbon paper and subsequently dried at 60 °C for 10 min. Ag/AgCl and Pt rods were used as a reference electrode (RE) and a counter electrode (CE), respectively. The potential across the WE and the CE was scanned from −0.75 to 0.25 V.

The durability of the prepared catalyst was tested in a mixture of 1 M KOH and 1 M ethanol solution using chronoamperometry (CA). A potential of −0.3 V was constantly applied to the three-electrode system for 1000 seconds, with an active surface area the same as that in the activity test. The electrochemical surface area was analysed from a cyclic voltammogram recorded from −1.2 to 0.5 V at a scan rate of 100 mV s⁻¹, in 1 M KOH solution purging with nitrogen gas with an active surface area of 0.40 cm² (Pd metal loading 0.0952 mg cm⁻²). The method for measuring CO tolerance was rather similar to that of electrochemical surface area. Initially, the CO adsorption on the ternary catalyst system was performed in 1 M KOH solution using CA under CO feeding conditions with an active surface area the same as that in the ESA test. After that, the electrolyte solution was purged with nitrogen gas to remove all the remaining dissolved CO gas. Finally the potential was scanned from −1.2 to 0.5 V (at 100 mV s⁻¹) to remove the pre-adsorbed CO molecules from the catalyst surface.

3 Results and discussion

The acid-treated carbon supports and the ternary catalysts were characterised by several methods to assess their suitability for use in DEFCs. The characterisation of carbon supports are detailed in the first subsection. The physical, chemical, and electrochemical characterisations of the catalyst are later explained in more detail in Section 3.2.

3.1 Functionalisation of carbon-based supports

Acid functionalisation is known to be a process of removing impurities and inserting acid functional groups on the surface of carbon supports. The hydroxyl groups (–OH) of carbon supports were replaced herein by carboxylic groups (–COOH).^40,41 Fig. S1† shows the SEM images of CB and MWCNTs before and after functionalisation. The modified surface can be observed to be a rougher surface when compared to the non-functionalised one. Additionally, the particle sizes of both CB_sn and MWCNT_sn were found to be decreased after the acid functionalisation. This may be due to residues or impurities being removed from the surface supports.

TGA results are shown in Fig. S2.† It can be seen that both functionalised supports started to decompose at a lower temperature when compared to the non-functionalised ones (i.e., untreated samples). This indicated that the existence of acid functional groups in the supports caused a decrease in the thermal stability of the supports. The weight loss of these functionalised supports was located at approximately 230 °C, a temperature associated with the decomposition of carboxylic groups.³⁹ It was also found that the decomposition temperature of MWCNT_sn was higher than that of CB_sn. The decomposition temperature of carbon-based supports was in the range of 400–700 °C. The residues obtained at the end of TGA measurement indicated the impurity level of the samples.³⁹ After the thermal decomposition, both functionalised supports contained less residue than that contained on the non-functionalised ones. This can imply that the acid functionalisation helped to remove some impurities despite causing the reduction of thermal stability. The weight percents of acid functional groups were 5.85% and 5.42% for CB_sn and MWCNT_sn, respectively.

Raman spectroscopy was used to verify the existence of –COOH functional groups from the change of bond structure among the functionalised carbon supports. Fig. S3† shows Raman spectra of all carbon supports. The D band at 1350 cm⁻¹ appears as a shoulder to the G band (located at around 1580–1620 cm⁻¹). It was suggested that the D band originated from a hybridised vibration associated with graphene edges, which usually indicates the presence of disorder in the graphene structure. The high intensity of the D band in carbon materials refers to a higher order of defects or disorder in the graphene structure. The high intensity of the G band indicated less amorphous carbon and highly crystalline structures of the untreated MWCNTs. The G′ band (the overtone of D band at 2D position) corresponds to the stacking order of the graphitic walls and the crystallinity of the graphitic planes.^42–44 The ratios of D band and G band intensity (I_D/I_G) of all supports are given in Table S1.† The I_D/I_G ratio of MWCNT_sn was higher than that of MWCNTs as a result of covalent functionalization of the MWCNT sidewall being increased after acid treatment.⁴⁵ However, the I_D/I_G ratio of CB_sn was slightly lower than that of CB, indicating that the degree of disorder of the microcrystalline structure of CB was not as badly effected as that of MWCNTs by acid treatment.³³

FTIR was used to confirm the functional groups attached to the original structures. The stretching and bending vibrations of atoms were used to determine the functional groups in the structures. As explained previously, the acid treatment enabled the purification and/or functionalisation of carbon supports to occur. The process was related to the oxidation of the double bond in the graphene wall yielding sulfonated groups (SO₃), carboxylic groups (COOH), and hydroxyl groups (OH) on the carbon surface.^46,47 Fig. 1 shows FTIR spectra of all the carbon supports. It can be seen that the characteristic peaks of CB_sn and MWCNT_sn were located at the same positions, but the peak intensity of CB_sn was weaker than that of MWCNT_sn. The fingerprint region below 1400 cm⁻¹ could hardly be used to identify any compounds because of its complexity. The peaks at around 3400 cm⁻¹ and at around 1640 cm⁻¹ refer to the H-bond of hydroxyl groups and the H-bond of carboxyl groups (C=O), respectively. These H-bonds were conjugated with C=C bonds (i.e., at approximately 1600 cm⁻¹) in the graphene wall of MWCNT_sn.⁴⁸ In addition, the peaks at approximately 2900 cm⁻¹ and 620 cm⁻¹ corresponding to the C–H bond were also obtained. It was believed to undergo an oxidation reaction due to the presence of C–O bonds at approximately 1475 cm⁻¹. Moreover, a peak at around 1100 cm⁻¹ can be observed, indicating the formation of sulfonated groups and corresponding to C–O vibration.⁴⁹

Fig. 1 The FT-IR spectra of CB compared with CB_sn (top) and MWCNTs compared with MWCNT_sn (bottom) revealing the effect of acid functionalisation.

The specific surface areas of supports strongly affected the activity of the catalyst. The multilayer adsorption model or BET method was used to analyse the surface area of the supports. According to IUPAC isotherms, all the support samples showed type-II adsorption isotherms, a normal form of isotherms shown by a finely divided non-porous adsorbent. This represents an unrestricted monolayer–multilayer adsorption.⁵⁰ The conversion of the isotherm graph to a BET graph resulted in a linear relationship between 1/W[(P/P₀) − 1] and P/P₀, as shown in Fig. S4.†

Fig. S4† reveals an increase in the slopes of the functionalised supports when compared to the non-functionalised ones. This means that the surface area of the functionalised supports was much higher than that of the non-functionalised supports. The values of BET surface area (S_BET) were calculated as given in Table S2.† The S_BET values of the CB and MWCNTs were 427.4 and 221 m² g⁻¹, respectively which agree very well with the values reported in the literature.^51,52 The S_BET of the pore diameters for all samples were below 250 Å or 25 nm, indicating the formation of mesopores. Interestingly, the S_BET values of MWCNT_sn and CB_sn decreased by 42% and 62%, respectively, which was also observed by Paulina et al. (2012). The BET surface area of the two functionalised carbon supports decreased due to the increasing functional group content. This can be related to the blocking of the inter-bundle galleries and intra-bundle interstitial channels by the functional groups.⁵³

3.2 Properties of electrocatalysts on different carbon supports

3.2.1 Physical properties. 20%Pd10%Ni10%Sn was impregnated on four different carbon supports: CB, CB_sn, MWCNTs, and MWCNT_sn. The selected catalyst composition showed the maximum current density for EOR as reported in our previous study.⁵⁴ The physical properties of 20%Pd10%Ni10%Sn deposited on different carbon supports were analysed using XRD, SEM-EDX, TEM, CO-pulse chemisorptions, and XPS. Then, the electrochemical properties were characterised using a potentiostat/galvanostat to explore the activity, durability, CO tolerance and electrochemical surface area (ESA) of those samples.

XRD was used to identify the forms of the metal catalysts as demonstrated in Fig. 2. All catalysts revealed similar peak positions, indicating the identical forms of metal catalysts deposited on different supports. The first peak at around 25–30° was the (0 0 2) plane of carbon structure in CB or MWCNTs. The characteristic peaks of Pd included the peaks at 39.901°, 46.209°, 67.618°, 81.364°, and 86.342°, which referred to (1 1 1), (2 0 0), (2 2 0), (3 1 1), and (2 2 2) planes, respectively. All peaks of Pd represented the face-centred cubic (FCC) structure. The peaks at 33.193° and 59.303° refer to the (1 0 0) and (1 1 0) planes of Ni(OH)₂, respectively, whereas the peaks at 26.511°, 33.769°, 51.612°, and 65.752° refer to the (1 1 0), (1 0 1), (2 1 1), and (3 0 1) planes of SnO₂, respectively. The SnO₂ peaks were difficult to observe as three peaks of SnO₂ were overlapped with the peaks of amorphous carbon (25–30°), Ni(OH)₂ (33.193°), and Pd (67.618°).

Fig. 2 XRD patterns of all prepared catalysts at 2θ = 20–90° ( carbon, Ni(OH)₂, SnO₂, Pd).

The peak of Pd (2 2 0) at 67.618° was used to calculate spacing, lattice parameter, and crystalline size in all samples. The lattice parameter is the physical dimension of unit cells in a crystal lattice, and d-spacing is the spacing between the planes. The results are given in Table 1. It was found that the crystalline size of catalysts was in the range of 6.4–7.4 nm, and similar values of d-spacing and lattice parameter were obtained.

Table 1 d-Spacing, lattice parameter, and crystalline size calculated from the Pd peak at (2 2 0) plane

Composition	2θ	d-Spacing (nm)	Lattice parameter (nm)	Crystalline size (nm)
20%Pd10%Ni10%Sn/CB	67.845	0.13802	0.39038	6.8
20%Pd10%Ni10%Sn/CBsn	67.911	0.13791	0.39007	6.6
20%Pd10%Ni10%Sn/MWCNT	67.916	0.13790	0.39004	6.4
20%Pd10%Ni10%Sn/MWCNTsn	67.813	0.13808	0.39054	7.4

---

SEM images of all samples are shown in Fig. 3. The metal particles were dispersed uniformly on the surface of the carbon supports. The actual compositions of the as-prepared catalyst samples were examined using SEM-EDX with 5 sampling points per sample. It was found that the actual compositions were quite similar to the desired compositions as shown in Table 2.

Fig. 3 SEM images (×30 000) of 20%Pd10%Ni10%Sn catalyst on (a) CB, (b) CB_sn, (c) MWCNTs, and (d) MWCNT_sn.

Table 2 Elemental compositions of 20%Pd10%Ni10%Sn catalyst on different carbon supports

Catalyst/carbon support	Desired composition (% wt)			Actual composition using EDX (% wt)
	Pd	Ni	Sn	Pd	Ni	Sn
20%Pd10%Ni10%Sn/CB	50	25	25	49.21 ± 2.92	23.58 ± 5.45	27.22 ± 2.58
20%Pd10%Ni10%Sn/CBsn	50	25	25	52.11 ± 4.69	16.45 ± 2.00	31.44 ± 3.34
20%Pd10%Ni10%Sn/MWCNT	50	25	25	53.78 ± 3.99	16.81 ± 2.38	29.41 ± 2.09
20%Pd10%Ni10%Sn/MWCNTsn	50	25	25	45.37 ± 6.47	29.25 ± 6.33	25.37 ± 1.83

---

TEM images of 20%Pd10%Ni10%Sn on various supports (see Fig. 4) show that the metal catalysts were well dispersed on all the carbon supports. The histograms of catalyst particle size distribution on various supports (see Fig. 5) reveal a narrow average particle size of 6.0–6.7 nm. The catalyst particle sizes greater than 14 nm shown in Fig. 5 may come from overlapping of the supports which was not averaged.

Fig. 4 TEM images at (×50 000) of 20%Pd10%Ni10%Sn catalyst on (a) CB, (b) CB_sn, (c) MWCNTs, and (d) MWCNT_sn.

Fig. 5 Particle size distribution of 20%Pd10%Ni10%Sn catalyst on (a) CB, (b) CB_sn, (c) MWCNTs, and (d) MWCNT_sn.

The metal dispersion (D_m) and the number of active sites (N_s) of 20%Pd10%Ni10%Sn on various supports assessed by chemisorption are given in Table 3. It was found that the catalyst on both MWCNT_sn and CB_sn exhibited high levels of D_m and N_s implying that well-dispersed and distributed metal particles were achieved with the use of acid functionalised carbon supports. The D_m result was consistent with the catalyst distribution results observed by TEM.

Table 3 Metal dispersion (D_m) and number of active sites (N_s) of 20%Pd10%Ni10%Sn on different carbon supports

Catalyst	Dm (%)	Ns (site)
20%Pd10%Ni10%Sn/CB	1.4474	1.64 × 10^19
20%Pd10%Ni10%Sn/CBsn	2.1705	2.46 × 10^19
20%Pd10%Ni10%Sn/MWCNT	0.9251	1.05 × 10^19
20%Pd10%Ni0%Sn/MWCNTsn	2.8822	3.26 × 10^19

---

XPS was carried out to evaluate the oxidation number of the metal catalysts. Fig. 6a shows the XPS survey scan of 20%Pd10%Ni10%Sn catalyst on CB_sn and MWCNT_sn supports. The survey scan contained the atomic peaks of carbon (C 1s), oxygen (O 1s), nickel (Ni 2p), tin (Sn 3d), and palladium (Pd 3d). The XPS area ratios of chemical state of Pd, Ni and Sn species and their oxidation states are detailed in Table 4. The XPS spectrum of the Pd 3d peak consisted of two dominant peaks and three sub-peaks, corresponding to Pd metal (Pd⁰) and PdO (Pd²⁺) forms, respectively. The XPS spectrum of the Ni 2p peak was composed of two dominant peaks and four sub-peaks associated with Ni, NiO, Ni(OH)₂, and NiOOH with the oxidation states of Ni⁰, Ni²⁺, Ni²⁺, and Ni³⁺, respectively. The XPS spectrum of Sn 3d included two main peaks and three sub-peaks associated with Sn–SnO and SnO₂ with the oxidation states of Sn⁰–Sn²⁺ and Sn⁴⁺, respectively. In both catalyst samples, the highest ratios of Ni(OH)₂ and SnO₂, which are known to promote the EOR and CO oxidation, respectively, were obtained.^11,13,17,18 Fig. 6b–d show XPS spectra of Pd 3d, Ni 2p, and Sn 3d, respectively, of catalyst on MWCNT_sn (top) and MWCNT (bottom). The carbon support showed the interaction between metal loading by the shift of metal peak. The XPS results show a positive shift of 0.1 eV for Pd 3d and a negative shift 0.4 eV for Sn 3d on catalyst on MWCNT_sn.^55,56 In the case of Ni on MWCNT, the peak contained much noise and was of a low intensity making it hard to identify the position of peak for the peak shift.

Fig. 6 (a) Survey XPS spectra of 20%Pd10%Ni10%Sn on CB_sn and MWCNT_sn; (b) XPS spectra of Pd 3d of catalyst on MWCNT_sn and MWCNT; (c) XPS spectra of Ni 2p of catalyst on MWCNT_sn and MWCNT; (d) XPS spectra of Sn 3d of catalyst on MWCNT_sn and MWCNT.

Table 4 XPS area ratios of chemical state of Pd, Ni, and Sn species

Form of metal	XPS area ratio (%)
	Metal on CBsn	Metal on MWCNTsn
Pd(0)	87.5	12.5
PdO(2+)	87.5	12.5
Ni(0)	9.0	5.7
NiO(2+)	15.3	9.2
Ni(OH)2(2+)	60.0	51.8
NiOOH(3+)	15.7	33.3
Sn–SnO(0,2+)	4.5	95.5
SnO2(4+)	3.3	96.7

---

3.2.2 Activity, stability, CO tolerance, and electrochemical surface area (ESA). The cyclic voltammograms of the catalyst samples for EOR were recorded in a mixture of 1 M ethanol and 1 M KOH aqueous solution and are shown in Fig. 7. The EOR can be observed on the anodic peak which was located between −0.2 and 0.1 V. It can be seen that the onset potentials were relatively similar for all samples. Table 5 shows the corresponding anodic peak potentials and current densities for EOR. The 20%Pd10%Ni10%Sn/MWCNT_sn exhibited the maximum current density or activity (291 mA cm⁻²) and mass activity (4288 mA mg_Pd⁻¹). It has been suggested that the catalytic activity towards EOR was more favourable when the catalyst exhibited a higher magnitude of anodic current density but a lower value of the onset potential.⁵⁷ Compared to other works, it was found that the catalyst activity Pd₂Ni₃/C prepared by Shen et al. showed lower current density (about 80 mA cm⁻²) and the Pd/C catalyst prepared by Yi et al. also showed a current density about 50% lower than that obtained with the catalysts prepared in this work.^13,57 The highest mass activity of the catalyst on MWCNT_sn (about 2–3 times that of others) suggested that Pd supported on MWCNT_sn has a more effective utility value.⁹ The results also confirmed that the 20%Pd10%Ni10%Sn deposited on functionalised supports generated higher current density than those on non-functionalised supports. This may be due to the high crystallinity and strong covalent bond of the support, leading to a good dispersion and a nanoparticle size of metal.^58,59 In terms of catalyst activity, MWCNT_sn was found to be a promising catalyst support for EOR in DEFC.

Fig. 7 Voltammograms of 20%Pd10%Ni10%Sn deposited on different catalyst supports recorded in a mixture of 1 EOH and 1 M KOH at a scan rate of 50 mA s⁻¹.

Table 5 The data extracted from cyclic voltammograms (Fig. 10) of 20%Pd10%Ni10%Sn deposited on different catalyst supports recorded in a mixture of 1MeOH and 1 M KOH

Catalyst on	Cyclic voltammetry of ethanol oxidation				CO-stripping		ESA
	Onset potential (V)	Anodic peak voltage (V)	Anodic peak current density or activity (mA cm^−2)	Mass activity (mA mgPd^−1)	CO stripping onset potential (V)	CO stripping peak potential (V)	Area of desorbed OH^− peak (mC cm^−2)	ESA (m^2 g^−1 Pd)
CB	−0.414	−0.077	146	2151	−0.478	−0.162	22.856	59.255
CBsn	−0.419	−0.141	177	2608	−0.480	−0.168	23.791	61.680
MWCNT	−0.472	−0.182	107	1576	−0.415	−0.174	19.711	52.615
MWCNTsn	−0.396	0.096	291	4288	−0.368	−0.179	29.760	77.155

---

The stability of 20%Pd10%Ni10%Sn catalyst deposited on different carbon supports was determined by chronoamperometry. The test method reported by Yaojuan et al. and Changwei et al. was followed.^60,61 The stability was found unchanged after 1000 s. Therefore, a stability test of the catalyst was carried out and the results are demonstrated in Fig. 8. The percentages of current drop of MWCNT_sn, MWCNT, CB, and CB_sn were 2.58, 1.15, 0.76, and 0%, respectively. Thus, 20%Pd10%Ni10%Sn/MWCNT_sn exhibited the greatest catalyst stability. The reason for this was because the covalent bonds of carboxylic functional groups attached to the MWCNT_sn surface enabled a good interaction between metals and support⁵⁸ and the high degree of crystallinity of the MWCNT_sn support resulted in a good electrical connection at a molecular level.⁶²

Fig. 8 Chronoamperometric curves of 20%Pd10%Ni10%Sn deposited on different catalyst supports recorded in a mixture of 1 M ethanol and 1 M KOH solution at an applied potential of −0.3 V.

A CO-stripping test was also used to determine the catalytic ability in the presence of CO poisoning in the system. It is also a useful method for determination of electrochemical surface area, especially for Pd-based catalysts. CO-stripping voltammograms of 20%Pd10%Sn10%Ni on various supports are demonstrated in Fig. 9; the inset shows the onset potential and the peak potential for CO oxidation. The peak for CO oxidation occurred at around −0.1 to −0.2 V with the onset potential in a range of −0.368 to −0.480 V. The metal promoters (Sn and Ni) facilitated the removal of CO poison on the Pd surface and refreshed the reaction sites for ethanol oxidation. As regards to the onset potential of the CO oxidation peak, a lower value indicates well facilitated CO removal on the Pd surface for the route of ethanol oxidation. The catalyst on MWCNT and MWCNT_sn supports showed lower onset potentials than the catalyst on CB and CB_sn supports. This might be due to the better effect of Ni and Sn oxides on the catalyst surface of MWCNT and MWCNT_sn, or it could be due to the effect of Pd–Ni and Pd–Sn alloys on CO removal reaction.⁶³ The reaction pathways to remove CO poisoning of metal oxide (MtO_x: SnO_x, NiO_x) are shown in eqn (1)–(3), with the metal oxide forms revealed by XPS and XRD results. Moreover, the catalyst on functionalised support exhibited a lower onset potential (the lowest onset potential was the catalyst on MWCNT_sn) than the non-functionalised one, as shown in Fig. 9. The onset and peak potentials for the CO-stripping process extracted from Fig. 9 are detailed in Table 5. The peak potential was in the range of −0.162 to −0.179 V. The catalyst on MWCNT and MWCNT_sn had lower onset potentials than those of the catalyst on CB and CB_sn, indicating the higher efficiency for CO tolerance obtained when using MWCNT supports.

2Pd + CH₃CH₂OH ⇒ 2Pd–CO_ads + 6H⁺ + 6e⁻ (1)

MtO_x + H₂O ⇒ MtO_x–OH_ads + H⁺ + e⁻ (2)

Pd–CO_ads + MtO_x–OH_ads ⇒ Pd + MtO_x + CO₂ + H⁺ + e⁻ (3)

where MtO_x is a metal oxide such as SnO_x, NiO_x.

Fig. 9 CO stripping voltammograms of 20%Pd10%Sn10%Ni on various carbon supports and the corresponding forward scan of CO stripping voltammograms (inset).

Furthermore, another anodic peak which is associated with Ni metal and, in particular, the oxidation of Ni(OH)₂ to NiOOH was found at approximately 0.4–0.5 V. As regards another cathodic peak, it was found that the reduction reaction of NiOOH to Ni(OH)₂ was at about 0.4–0.2 V, as shown in eqn (4) and (5). These peaks are associated with the increase of activity of the catalyst. Ni metal on the support catalyst can work similarly to Pd (on catalyst activity) and reduce CO-poisoning in the same manner as Sn.

Ni(OH)₂ + OH⁻ ↔ NiOOH + H₂O + e⁻ (4)

NiOOH + H₂O + e⁻ ↔ Ni(OH)₂ + OH⁻ (5)

ESAs of the 20%Pd10%Ni10%Sn on various supports were determined by hydroxyl (OH⁻) adsorption in 1 M KOH solution. Fig. 10 shows the cyclic voltammograms of these catalyst samples. OH⁻ desorption was used to calculate the ESA values (given in Table 5). The anodic peak appeared in the forward scan, and the cathodic peak showed in the backward scan. In the forward scan, the oxidation peak was at −0.65 V, associated with the formation of the OH⁻ group adsorption (OH_ads) on the Pd surface which was then converted to Pd–OH_ads. After that, the backward scan caused the electrodesorption of the pre-adsorbed OH_ads from the Pd surface. From Table 5 it can be seen that the use of functionalised supports exhibited higher ESAs than those obtained with the non-functionalised supports. MWCNT_sn showed the highest ESA which may be because the metal was well attached to and dispersed in this support. This agreed well with the TEM results. Moreover, the result was consistent with the number of active sites measured from the CO chemisorption test. Among the samples, MWCNT_sn was found to be the most suitable support for the deposition of 20%Pd10%Ni10%Sn as a ternary catalyst in DEFC, as it had excellent catalytic activity and stability, great performance for CO tolerance, and the highest ESA.

Fig. 10 Cyclic voltammograms of all samples for OH⁻ adsorption and desorption in 1 M KOH solution.

4 Conclusions

The 20%Pd10%Ni10%Sn tri-metallic electrocatalyst was synthesised on four different carbon supports (i.e., CB, CB_sn, MWCNT, and MWCNT_sn) to investigate the most suitable support for DFEC electrodes. The catalyst samples were produced using a NaBH₄ reduction method, which is a simple and a versatile technique. Then, their physical and electrochemical properties were characterised using several techniques. The results obtained by TGA diagram, Raman spectra, and FTIR spectra verified the existence of acid functional groups on both functionalised supports: CB_sn and MWCNT_sn. However, TGA and BET results suggested that the thermal stability and S_BET were decreased after acid functionalisation. The actual catalyst compositions in all samples were investigated by EDX, and the results agreed very well with the desired composition of 20%Pd10%Ni10%Sn. TEM micrographs demonstrated the narrow size distribution and good dispersion of catalyst particles on the MWCNT_sn support with an average particle size of 6.08–6.65 nm. The acid functionalisation was found via characterisation by CO chemisorption to increase the number of active sites. Moreover, the crystalline forms of Pd, Ni(OH)₂, and SnO₂ were confirmed by the XRD and XPS results. For the electrical properties of 20%Pd10%Ni10%Sn deposited on different supports, it was found that the 20%Pd10%Ni10%Sn/MWCNT_sn exhibited outstanding electrical properties, including a maximum current density of 291 mA cm⁻² for EOR, high catalyst stability and ESA, and excellent CO tolerance. This may be due to high crystallinity and strongly covalent bonds of MWCNT_sn and the metals, leading to a good dispersion of nanoparticles of metal.

Acknowledgements

This work was financially supported by the Kasetsart University Research and Development Institute (KURDI) and Center for Nanocale Materals Design for Green Nanotechnology, Kastsart Universty. Thanks are also to National Center of Excellence for Petroleum, Petrochemical, and Advanced Materials and National Metal and Materials Technology Center (P1550433) Thailand.

References

1. C.-T. Hsieh, Y.-Y. Liu, W.-Y. Chen and Y.-H. Hsieh, Int. J. Hydrogen Energy, 2011, 36, 15766–15774.
2. H. David, T. Reginald, B. Steve and J. Amanda, Hydrogen for fuel cell electric vehicles, http://www.renewableenergyfocus.com/blog/2013/9/19/hydrogen-for-fuel-cell-electric-vehicles/1006.aspx, accessed September 12, 2014.
3. Y. H. Chu and Y. G. Shul, Int. J. Hydrogen Energy, 2010, 35, 11261–11270.
4. E. Antolini and E. R. Gonzalez, J. Power Sources, 2010, 195, 3431–3450.
5. K. S. Roelofs, T. Hirth and T. Schiestel, Mater. Sci. Eng., 2011, 176, 727–735.
6. Y. Li and Y. He, RSC Adv., 2014, 4, 16879–16884.
7. N. Wongyao, A. Therdthianwong and S. Therdthianwong, Energy Convers. Manage., 2011, 52, 2676–2681.
8. S. Meenakshi, P. Sridhar and S. Pitchumani, RSC Adv., 2014, 4, 44386–44393.
9. R. Wang, J. Yang, K. Shi, B. Wang, L. Wang, G. Tian, B. Bateer, C. Tian, P. Shen and H. Fu, RSC Adv., 2013, 3, 4771–4777.
10. L. An, T. S. Zhao, R. Chen and Q. X. Wu, J. Power Sources, 2011, 196, 6219–6222.
11. T. Ramulifho, K. I. Ozoemena, R. M. Modibedi, C. J. Jafta and M. K. Mathe, Electrochim. Acta, 2012, 59, 310–320.
12. H. Liu, C. Song, L. Zhang, J. Zhang, H. Wang and D. P. Wilkinson, J. Power Sources, 2006, 155, 95–110.
13. S. Y. Shen, T. S. Zhao, J. B. Xu and Y. S. Li, J. Power Sources, 2010, 195, 1001–1006.
14. P. K. Shen and C. Xu, Electrochem. Commun., 2006, 8, 184–188.
15. I. Merino-Jimenez, M. J. Janik, C. Ponce de Leon and F. C. Walsh, J. Power Sources, 2014, 269, 498–508.
16. X. Wang, Y. Tang, Y. Gao and T. Lu, J. Power Sources, 2008, 175, 784–788.
17. R. M. Modibedi, T. Masombuka and M. K. Mathe, Int. J. Hydrogen Energy, 2011, 36, 4664–4672.
18. C. Jin, X. Sun, Z. Chen and R. Dong, Mater. Chem. Phys., 2012, 135, 433–437.
19. M. Nie, H. Tang, Z. Wei, S. P. Jiang and P. K. Shen, Electrochem. Commun., 2007, 9, 2375–2379.
20. J. B. Xu, T. S. Zhao, S. Y. Shen and Y. S. Li, Int. J. Hydrogen Energy, 2010, 35, 6490–6500.
21. J. Cai, Y. Huang and Y. Guo, Electrochim. Acta, 2013, 99, 22–29.
22. A. O. Neto, M. M. Tusi, N. S. de Oliveira Polanco, S. G. da Silva, M. Coelho dos Santos and E. V. Spinacé, Int. J. Hydrogen Energy, 2011, 36, 10522–10526.
23. Z. Qi, H. Geng, X. Wang, C. Zhao, H. Ji, C. Zhang, J. Xu and Z. Zhang, J. Power Sources, 2011, 196, 5823–5828.
24. Z. Liu, X. Zhang and L. Hong, Electrochem. Commun., 2009, 11, 925–928.
25. M. R. Tarasevich, V. N. Titova, A. A. Yavich, N. V. Petrova and V. A. Bogdanovskaya, Russ. J. Phys. Chem., 2009, 83, 1850–1854.
26. J. C. Castro, R. M. Antoniassi, R. R. Dias, M. Linardi, E. V. Spinacé and A. O. Neto, Ionics, 2012, 18, 781–786.
27. H. An, H. Cui, D. Zhou, D. Tao, B. Li, J. Zhai and Q. Li, Electrochim. Acta, 2013, 92, 176–182.
28. H. An, L. Pan, H. Cui, B. Li, D. Zhou, J. Zhai and Q. Li, Electrochim. Acta, 2013, 102, 79–87.
29. J. Tayal, B. Rawat and S. Basu, Int. J. Hydrogen Energy, 2011, 36, 14884–14897.
30. V. Bambagioni, C. Bianchini, A. Marchionni, J. Filippi, F. Vizza, J. Teddy, P. Serp and M. Zhiani, J. Power Sources, 2009, 190, 241–251.
31. J. M. Jacob, P. G. Corradini, E. Antolini, N. A. Santos and J. Perez, Appl. Catal., B, 2015, 165, 176–184.
32. S. Meenakshi, K. G. Nishanth, P. Sridhar and S. Pitchumani, Electrochim. Acta, 2014, 135, 52–59.
33. X. Xue, J. Ge, T. Tian, C. Liu, W. Xing and T. Lu, J. Power Sources, 2007, 172, 560–569.
34. M. Uchida, Y. Aoyama, M. Tanabe, N. Yanagihara, N. Eda and A. Ohta, J. Electrochem. Soc., 1995, 142, 2572–2576.
35. D.-J. Guo and H.-L. Li, J. Power Sources, 2006, 160, 44–49.
36. H. S. Nalwa, Nanostructured Materials and Nanotechnology (Concise edn.), Academic Press, San Diego, 2002.
37. S. Tang, G. Sun, J. Qi, S. Sun, J. Guo, Q. Xin and G. M. Haarberg, Chin. J. Catal., 2010, 31, 12–17.
38. N. Karousis, N. Tagmatarchis and D. Tasis, Chem. Rev., 2010, 110, 5366–5397.
39. A. G. Osorio, I. C. L. Silveira, V. L. Bueno and C. P. Bergmann, Appl. Surf. Sci., 2008, 255, 2485–2489.
40. A. M. Zaiton, A. M. S. Nur, A. B. Nor and S. Shafinaz, Journal of Fundamental Sciences, 2010, 6, 51–55.
41. Y. Rike, O. S. Holia, S. Yukie, I. Tadahisa and J. I. Azuma, Open Mater. Sci. J., 2011, 5, 242–247.
42. Z.-M. Dang, L. Wang and L.-P. Zhang, J. Nanomater., 2006, 2006, 1–5.
43. M. A. Pimenta, G. Dresselhaus, M. S. Dresselhaus, L. G. Cancado, A. Jorio and R. Saito, Phys. Chem. Chem. Phys., 2007, 9, 1276–1291.
44. G. E. Romanos, V. Likodimos, R. R. N. Marques, T. A. Steriotis, S. K. Papageorgiou, J. L. Faria, J. L. Figueiredo, A. M. T. Silva and P. Falaras, J. Phys. Chem. C, 2011, 115, 8534–8546.
45. A. E. Daugaard, K. Jankova, J. M. R. Marín, J. Bøgelund and S. Hvilsted, Eur. Polym. J., 2012, 48, 743–750.
46. H. Hu, B. Zhao, M. E. Itkis and R. C. Haddon, J. Phys. Chem. B, 2003, 107, 13838–13842.
47. M. W. Marshall, S. Popa-Nita and J. G. Shapter, Carbon, 2006, 44, 1137–1141.
48. Y.-T. Shieh, G.-L. Liu, H.-H. Wu and C.-C. Lee, Carbon, 2007, 45, 1880–1890.
49. A. G. Osorio, I. C. L. Silveira, V. L. Bueno and C. P. Bergmann, Appl. Surf. Sci., 2008, 255, 2485–2489.
50. K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol and T. Siemieniewska, in Handbook of Heterogeneous Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, 2008,  DOI:10.1002/9783527610044.hetcat0065.
51. H. Y. Quan, X. H. Peng, B. Shuo, Z. W. Mao and M. C. Hui, Chem. Phys. Lett., 2001, 345, 18–24.
52. S. Inoue, N. Ichikuni, T. Suzuki, T. Uematsu and K. Kaneko, Phys. Chem., 1998, 102, 4689–4692.
53. N. Jha, P. Ramesh, E. Bekyarova, X. Tian, F. Wang, M. E. Itkis and R. C. Haddon, Sci. Rep., 2013, 3, 2257.
54. S. Jongsomjit, K. Sombatmankhong, S. Mori and P. Prapainainar, PACCON2014 (Pure and Applied Chemistry International Conference 2014) conference proceeding, Khon Kaen, Thailand, January 8–10, 2014.
55. D. He, C. Zeng, C. Xu, N. Cheng, H. Li, S. Mu and M. Pan, Langmuir, 2011, 27, 5582–5588.
56. D. He, Y. Jiang, H. Lv, M. Pan and S. Mu, Appl. Catal., B, 2013, 132–133, 379–388.
57. F. N. Qingfeng Yi and Lizhi Sun, Fuel, 2011, 90, 2617–2623.
58. Z.-L. Liu, R. Huang, Y.-J. Deng, D.-H. Chen, L. Huang, Y.-R. Cai, Q. Wang, S.-P. Chen and S.-G. Sun, Electrochim. Acta, 2013, 112, 919–926.
59. G. Wu and B.-Q. Xu, J. Power Sources, 2007, 174, 148–158.
60. C. Xu, L. Cheng, P. Shen and Y. Liu, Electrochem. Commun., 2007, 9, 997–1001.
61. Y. Hu, P. Wu, Y. Yin, H. Zhang and C. Cai, Appl. Catal., B, 2012, 111–112, 208–217.
62. A. Santasalo-Aarnio, M. Borghei, I. V. Anoshkin, A. G. Nasibulin, E. I. Kauppinen, V. Ruiz and T. Kallio, Int. J. Hydrogen Energy, 2012, 37, 3415–3424.
63. Y.-C. Wei, C.-W. Liu, W.-D. Kang, C.-M. Lai, L.-D. Tsai and K.-W. Wang, J. Electroanal. Chem., 2011, 660, 64–70.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra07508d

---