Gold nanocrystal arrays as electrocatalysts for the oxidation of methanol and ethanol

1 Introduction

The clean conversion of chemical fuels into electricity with high efficiency can be realized by fuel cell technology, i.e. the reaction of gaseous oxygen with hydrogen [1]. Hydrogen can be delivered for the reaction from either methanol or ethanol through direct electrochemical oxidation. However, the main hindrances for the commercialization of direct methanol fuel cells are the poor kinetics of the methanol oxidation reaction (MOR) [2]. Higher energy densities can be achieved with direct ethanol fuel cells, which furthermore have the advantage of ethanol being less toxic than methanol [3]. The interest in the ethanol oxidation reaction (EOR) is also connected with the worldwide availability of ethanol as a hydrogen feedstock.

Oxidation reactions of such small molecules at the anode require electrocatalysts that decrease significantly the activation energy, providing a high current density. The development of electrocatalysts, which are active and stable under operating conditions, is therefore the key for future fuel cell technology [4]. The most studied catalysts in acidic media are platinum-based materials because of their high electrocatalytic activity [5]. However, platinum-based electrocatalysts are known to suffer significant deactivation throughout alcohol oxidation [6]. This is due to the formation of strongly adsorbed intermediate species such as carbon monoxide [7]. This severe catalyst poisoning, i.e. an irreversible loss of the catalytically active surface area, leads to the decrease in catalytic performance.

In addition to monometallic platinum catalysts, most research efforts for MOR and EOR in acidic media are focused on bimetallic platinum- and/or palladium-based alloys [8–13]. In contrast to acidic environments, alkaline electrolyte media offer higher stability of the catalysts and in many cases higher activities [14]. For instance, gold is known to exhibit higher current densities than catalysts containing platinum and palladium for MOR and EOR in alkaline electrolytes [15].

Kwon et al. [16] elucidated the origin of the electrocatalytic activity of gold for alcohol oxidation reactions in alkaline media. Their studies have demonstrated that the main source of the high electrocatalytic activity is the base catalysis from the solution accompanied with a second gold-catalyzed reaction step. In combination with the superior resistance of gold toward surface poisoning by reaction intermediates, this results in higher electrocatalytic oxidation activity with respect to platinum- and/or palladium-based materials [17–19].

The general difficulty in the comparison of catalysts regarding their electrochemical activities is the significant dependency on the electrode preparation method. In addition to single-crystal [20], thin-film [21, 22], and polycrystalline [23] electrodes, most electrocatalysts contain a physical mixture of catalytically active nanocrystals (NCs), conductive carbon support, and binding agent [24–27]. This composite is deposited on another conductive carbonaceous substrate. This type of preparation makes the agglomeration of NCs to larger entities inevitable and decreases simultaneously the catalytically active surface area. Furthermore, the mass activity, i.e. current per catalyst loading, is an important criterion when determining realistic materials costs for fuel cells.

We have fabricated two-dimensional arrays [28] of monodisperse Au NCs with a mean particle size of 8 nm by self-assembly and investigated their electrocatalytic activity for the MOR and EOR. We have calculated the mass activities for 1.0 m alcohol and different base concentrations at a scan rate of 50 mV s⁻¹ during cyclic voltammetry.

2 Results and discussion

The synthesized Au NCs are capped with oleylamine, and their experimental powder X-ray diffraction patterns match those calculated for the gold face-centered cubic (fcc) structure (Fig. 1). Under certain film growth conditions (see Experimental section), the self-assembly of such Au NCs yields a two-dimensional NC array over an area of several square centimeters. The NC array can be mounted onto hydrophobic substrates, such as high purity vitreous carbon (HPVC) discs, as demonstrated by scanning electron microscopy (SEM) images. HPVC discs are an ideal electrode support for the deposition of NC arrays because they provide high planarity for NC deposition and electrical conductivity. The SEM image in Fig. 2 shows the coated side of the HPVC, which is exposed to the electrolyte in the electrochemical MOR and EOR. The inset of Fig. 2 demonstrates a complementary transmission electron microscopy (TEM) image of the Au NCs at a higher magnification. The NCs are of spherical shape with a particle size of 8.0±1.0 nm (based on 200 counted entities). The investigation of the catalytic effect of Au NC arrays for MOR was performed for a constant methanol concentration of 1.0 m and KOH concentrations of 0.1 and 1.0 m. To draw conclusions on the catalytic effect of the Au NC arrays for the electrochemical MOR, two additional types of electrochemical experiments were conducted. First, methanol was oxidized electrochemically on a working electrode consisting solely of HPVC without Au NCs. This measurement served as the baseline and ensured that carbon served only as the conducting support for Au NCs. The second type of experiment was cyclic voltammetry of the Au NC array in 0.1 m KOH in absence of the alcohol.

Fig. 1:

Experimental powder X-ray diffraction patterns of synthesized Au NCs and data calculated for fcc gold. The inset contains an image of a gold NC film.

Fig. 2:

SEM image of a self-assembled Au NC array after the transfer on the supporting carbon electrode. The inset shows a TEM image of the Au NCs in a higher magnification (scale bar, 10 nm).

Without methanol, the anodic currents are related to hydroxide adsorption and at higher potentials to the oxidation of gold [29, 30]. The surface electrochemistry of Au NCs is highly determined by its hydroxide coverage, and the cathodic currents in the reverse potential sweep are a consequence of the reduction of gold oxide to gold [31]. In the presence of methanol, the electrocatalytic effect of the Au NC array is visible through the appearance of an oxidation wave between 1.0 and 1.3 V (Fig. 3). In the presence of methanol, this oxidation peak is clearly distinct from the HPVC baseline and the characteristics of the methanol-blank Au NC system. The decrease in electrocatalytic activity, and thus the current density, at higher potentials is related to the formation of gold oxide, which inhibits the MOR [32].

Fig. 3:

Cyclic voltammograms of 1.0 m MOR at 50 mV s–¹ for 0.1 m (solid black line) and 1.0 m (solid blue line) KOH. The baseline (dotted green line) represents the oxidation of methanol on carbon support (0.1 m KOH) in the absence of the Au electrocatalyst. The electrochemical behavior of Au NCs in 0.1 m KOH in the absence of methanol is drawn as a red solid line.

The reverse sweep in the presence of methanol is characterized by the absence of an oxidation peak, resulting in a total negative current. In other words, the surface reduction of gold oxide is faster than the alcohol oxidation. This is due to the low concentration of methanol at the interface between the Au NCs and the electrolyte [33]. The electrocatalyst is reactivated in the reverse sweep accompanied by a simultaneous reduction of gold oxide back to elemental gold [34]. During the MOR, the oxidation peaks reach maximum current densities of 49 and 74 μA cm⁻² for 0.1 and 1.0 m KOH electrolyte solutions, respectively. It is known that for gold, the activity in oxidative processes varies in a highly nonlinear fashion with pH [35]. The higher activity of Au NCs at higher pH values is linked to an increased activation of the gold surfaces [36]. The adsorption of oxygen species, such as an OH⁻ anion, on the gold surface has a promoting role to the electro-oxidation of methanol or other alcohols [37].

To calculate the mass activity, i.e. current per mass of gold, quantitative analyses by means of atomic absorption spectroscopy (AAS) have been conducted on the Au NC arrays. The mass of gold in the Au NC arrays on HVPC support was determined to be 1.61±0.02 μg (based on four analyzed NC arrays). The calculated mass activities, expressed in the units A g⁻¹, are displayed on the right y scale of Fig. 3. At a scan rate of 50 mV s⁻¹ and with 1.0 m KOH, the oxidation peak of the electrochemical MOR on Au NC arrays gives a mass activity of 48.3 A g⁻¹.

In addition to the work on MOR, we have also investigated the performance of self-assembled Au NC array electrodes in the EOR [32]. During alcohol electro-oxidation in alkaline media, the electrocatalytic activity strongly depends on the pK_a value of the corresponding alcohol [16]. The initial step in the reaction is thought to be a base-catalyzed deprotonation of the alcohol into the catalytically active alkoxide [38]. A consecutive deprotonation step relies on the interaction of the alkoxide with the Au NCs [39]. Although one may expect that MOR would provide a higher current than EOR as a consequence of the lower pK_a value of methanol, the second deprotonation step is easier for ethanol because of its weaker C-H bonds [40]. The results of the alkaline EOR on the Au NC array at a scan rate of 50 mV s⁻¹ therefore show significantly higher current densities and mass activities in comparison with the MOR under identical conditions (Fig. 4). The higher catalytic activity of Au NCs for the second deprotonation step, in comparison with platinum-based catalysts under alkaline conditions, originates from the excellent resistance toward surface poisoning by reaction intermediates [41].

Fig. 4:

Cyclic voltammograms of 1.0 m EOR at 50 mV s–¹ for 0.1, 0.5, and 1.0 m KOH. The baseline (0.1 m KOH, absence of Au NCs) and the electrochemical behavior of Au NCs (0.1 m KOH, absence of ethanol) are drawn as a dotted green line and solid red line, respectively.

Contrary to the MOR, a new oxidation peak occurs at 1.05 V in the reverse sweep of the EOR (Fig. 4). This is due to a higher concentration of ethanol at the working electrode. The strength of interaction between gold and saturated aliphatic alcohol molecules during electrochemical oxidation is known to be modulated by the chain length [42]. The highest mass activity of 830 A g⁻¹ at the position of the oxidation peak for the EOR (1.0 m ethanol) was obtained for 1.0 m KOH and at a scan rate of 50 mV s⁻¹. This mass activity of EOR on Au NC arrays is approximately 17 times larger than that reached under identical conditions for MOR.

3 Conclusion

We have fabricated two-dimensional arrays of self-assembled monodisperse Au NCs and investigated their activity for the MOR and EOR at different pH values and constant alcohol concentrations and scan rates. Having monodisperse Au NC in the form of self-assembled arrays as an electrocatalyst enables to obtain high mass activities for MOR and EOR in alkaline media because of the suppression of NC agglomeration. For EOR, the mass activity of the Au NC arrays in alkaline media (1.0 m KOH, 1.0 m alcohol and at a scan rate of 50 mV s⁻¹) is with 830 A g⁻¹, approximately 17 times higher than for MOR under identical conditions.

4 Experimental section

Anhydrous gold(III) acetate (99%) was purchased from Alfa Aesar. Oleylamine (90%), 1,2-hexadecanediol (90%), and 1-octadecene (90%) were purchased from Aldrich. Methanol (99.95%) and ethanol (99.95%) were purchased from Geyer Chemsolute GmbH. Potassium hydroxide (99.99%) was obtained from Aldrich.

For the synthesis of Au NCs, 20 ml of 1-octadecene was loaded in a three-neck flask and degassed at 130°C for 30 min under nitrogen. After cooling down to room temperature, the flask was briefly opened to add 1.316 ml (4 mmol) oleylamine, 1.00 g (4 mmol) of 1,2-hexadecanediol, and 299.2 mg (0.8 mmol) of gold(III) acetate. The temperature was increased to 200°C under nitrogen and kept for 2 h. The dark-red solution was cooled to room temperature, and 35 ml ethanol was added to precipitate the NCs, followed by centrifugation at 12 000 rpm for 20 min. The Au NCs were redispersed with hexane and washed once with ethanol. The NCs can be redispersed in nonpolar solvents, such as hexane and toluene.

Self-assembled Au NC arrays were fabricated using glass containers, which were filled with deionized water. In a typical procedure, 2–3 ml of a 500-μg ml⁻¹ solution containing Au NCs in hexane and toluene was carefully deposited on the water surface using a micropipette. The self-assembly of Au NCs into NC arrays was achieved by the evaporation of hexane and toluene from the water surface. The resulting thin film is strongly hydrophobic and can be easily mounted on the surface of the HPVC electrode. The organic ligands were removed by O₂ plasma treatment (Zepto, Diener Electronic) at a power of 100 W for 120 s. The amount of Au NCs on the surface of the working electrode for electrochemical experiments was determined by AAS. The electrochemical surface area was calculated by cyclic voltammetry analysis of gold oxide reduction in 0.5 m H₂SO₄ at a scan rate of 100 mV s⁻¹ (Fig. 5) [43]. The current densities for all measurements were obtained by dividing the current by the electrochemically active surface area of 1.06 cm².

Fig. 5:

Cyclic voltammogram (0.5 m H₂SO₄ at a scan rate of 100 mV s⁻¹) for the determination of the electrochemically active surface area.

Powder X-ray diffraction (XRD) patterns were recorded in transmission mode with CuK_α1 radiation on a STOE STADI-P diffractometer equipped with an image plate detector. The morphology and self-assembly of Au NCs was analyzed by a Libra 200 FE TEM (Zeiss). SEM images of the working electrode covered with Au NC arrays were acquired on a Leo Supra 35VP SMT (Zeiss). Electrochemical measurements were performed on a Bio-Logic SP-50 potentiostat with the EC-Lab^® software package. Electrochemical measurements were conducted in a closed three-electrode cell. Potentials were recorded versus 1 m KCl Ag/AgCl reference electrode (Metrohm AG) and converted versus the RHE. Platinum wire was used as a counterelectrode. For the preparation of the working electrode, planar HPVC discs (Plano GmbH, surface roughness <50 nm) were used as substrate for Au NC arrays. The voltammetric characterization of the Au NC arrays was performed under alkaline conditions at room temperature. Before each electrochemical test, the electrodes and the electrochemical cells were carefully cleaned to eliminate impurities. The electrolytes were prepared from water purified with a Milli-Q system (18.3 MΩ·cm) and deoxygenated by bubbling with high-purity nitrogen, 99.998%, for 30 min before each cyclic voltammetry experiment.