Au/ZnO as catalyst for methanol synthesis: The role of oxygen vacancies

Abstract

Gold catalysts supported on zinc oxide with Au loadings of 1, 2, and 3 wt% were prepared by the colloidal deposition method and applied in methanol synthesis in CO₂-free (CO + H₂) and in CO₂-containing (CO + CO₂ + H₂) synthesis gas. The characterization by transmission electron microscopy and X-ray diffraction before and after the catalytic high-pressure tests demonstrated a very narrow and uniform Au particle size distribution and a high stability against sintering. Reactive frontal chromatography (RFC) experiments with N₂O were performed aiming at the titration of oxygen vacancies. With increasing Au loading, the amount of consumed N₂O increased in good correlation with the number of Au perimeter atoms present in the Au/ZnO catalysts suggesting an enhanced formation of oxygen vacancies at the Au/ZnO interface. In both synthesis gas mixtures the presence of the Au particles led to an increased activity compared with pure ZnO. All Au/ZnO samples exhibited higher catalytic activity in the absence of CO₂, as had been observed for pure ZnO with similar apparent activation energy. It is concluded that oxygen vacancies in ZnO are also the active sites in methanol synthesis over Au/ZnO, and that the presence of the Au particles enhances the number of exposed oxygen vacancies in ZnO, presumably located at the interface region.

Introduction

Although bulk gold is essentially chemically inert, gold can be a catalyst for many reactions, such as CO oxidation, selective oxidations of hydrocarbons, alcohols and aldehydes, hydrogenation reactions or the water gas shift reaction (WGSR, Eq. (1)), when it is applied as nanoscaled supported particles [1]. It has been shown that gold can also be an active catalyst in methanol synthesis (Eqs. (2), (3)) [2], [3], [4], [5]. Catalysts such as Au/TiO₂, Au/Al₂O₃, Au/Fe₂O₃, and Au/ZnO were investigated [3], [4], and all of these systems were found to exhibit catalytic activity for the WGSR and for the hydrogenation of carbon oxides to methanol and methane [3], [4], [6], [7].

$$\text{CO}+{\text{H}}_2\text{O}\rightleftharpoons {\text{CO}}_2+{\text{H}}_2\text{;}\Delta H_R=-42\text{kJ/mol}$$

$$\text{CO}+2{\text{H}}_2\rightleftharpoons {\text{CH}}_3\text{OH;}\Delta H_R=-98\text{kJ/mol}$$

$${\text{CO}}_2+3{\text{H}}_2\rightleftharpoons {\text{CH}}_3\text{OH}+{\text{H}}_2\text{O;}\Delta H_R=-58\text{kJ/mol}$$

Compared to the industrial Cu/ZnO/Al₂O₃ catalyst, which has been extensively characterized [8], [9], these Au catalysts are still barely understood. ZnO as support material for Au was shown to be essential to obtain a high selectivity towards methanol and to minimize methane formation [3], [4], [10]. Nevertheless, the role of ZnO has not been elucidated in detail yet [10].

Generally, the catalytic activity of supported Au particles depends on three key features: the size of the Au particles, the type of the support, and the structure of the interface, suggesting that the catalytic performance is significantly influenced by the interaction between the Au particles and the support [1]. It was claimed that the active sites are located at the interface between the Au particles and the support at the perimeter line of the particles [3], [4], [5], [7], [10], [11]. This concept is also believed to be valid for the performance of Au catalysts both in methanol synthesis [3], [4] and in the WGSR [7]. Smaller Au particles are thus desired to increase the total length of the perimeter at constant Au loading [3], [4], [5], [7], [10]. The reason for the interface being the active part of the catalyst is still under debate. For Cu/ZnO Spencer and coworkers [12], [13] suggested hydrogen dissociation to take place on ZnO followed by reverse spillover to Cu. King and Nix [14] found experimentally that the presence of the metal enhanced the reducibility of ZnO in the system Cu/ZnO. This may also be true for the presence of nanoscaled metallic Au. Frost [15] proposed already in 1988 that the presence of small clusters of any of the Ib group metals (Cu, Ag, Au) perturbs the oxide defect equilibria in semiconducting oxides by metal/oxide junctions enhancing the surface concentration of oxygen vacancies. It is discussed in Refs. [3], [4], whether these oxygen vacancies in the support act as active sites for methanol synthesis in oxide-supported Au catalysts.

It has been proposed that on pure ZnO oxygen vacancies on the polar oxygen-terminated ZnO $(000\overline{1})$ surface are the active sites for methanol synthesis. Reaction pathways starting from CO [16], [17] and from CO₂ [18], [19] were investigated. Experimentally, a higher activity in methanol synthesis on ZnO was observed, when CO₂ was absent [16], [17]. In a combined kinetic and theoretical study, an energetically preferable reaction pathway at the oxygen vacancy site was identified starting from CO instead of CO₂ [16]. It was proposed that in presence of CO₂ a change in the reaction mechanism occurred. From CO₂, a very stable formate species can be formed in the defect site, which can be hydrogenated to methanol. Due to its high stability, this rate is supposed to be much lower than methanol formation from CO [16]. Experimental evidence for the fast hydrogenation of CO over oxygen vacancies was obtained by investigating highly oxygen-deficient ZnO samples with electron paramagnetic resonance (EPR) spectroscopy. The samples exhibited a remarkably high catalytic activity in CO/H₂, which scaled with the integrated EPR signal intensity and which was quenched in the presence of CO₂ [17].

A facilitated formation of oxygen vacancies at the Au/ZnO interface, according to the enhanced reducibility of the oxide as suggested by Frost [15], would result in a larger amount of highly active sites in methanol synthesis. A better catalytic performance compared to pure ZnO should then be observed, especially in CO₂-free synthesis gas. Yet, if oxygen vacancies in ZnO are also the active sites in Au/ZnO, the catalytic behavior should still resemble that of pure ZnO [16], [17], [20].

As the particle size of Au is a crucial parameter for the catalytic performance, the colloidal deposition route was applied to obtain samples with constant high quality [21]. A method to derive the number of surface oxygen vacancies in Au/ZnO would be desirable, but is not yet available. In Cu catalysts, the reactive frontal chromatography with N₂O (N₂O RFC) is used to titrate the exposed Cu surface [22]. For Cu/ZnO, it was proposed that a quantification of surface oxygen vacancies can also be performed by N₂O RFC [14]. This method was therefore applied to the Au/ZnO catalysts. It was shown recently that microcalorimetry provides a method for a quantification of the amount of removable lattice oxygen: by applying reduction–oxidation steps using subsequent dosing sequences with CO and O₂, the amount of removable lattice oxygen in different Au/oxide samples was determined including Au/ZnO [23]. Comparing the amount of removable lattice oxygen to N₂O RFC results, the potential of this method for a quantification of the oxygen vacancies is evaluated. Furthermore, the comparison of the N₂O RFC results with the estimated values of Au surface and perimeter atoms should help to identify the location of the removable oxygen atoms.