Selective oxidation of ethanol on modified supported Au catalysts: from fundamental understanding to improved performance

Abstract

The transition from crude oil as primary feedstock for the chemical industry towards renewable sources is of paramount importance. Bioethanol is one of the most promising candidates for new raw materials, with an annual production over 100 billion liters. Selective oxidation of ethanol opens the pathway to acetaldehyde, ethyl acetate and acetic acid. In this thesis, the selective oxidation of ethanol on (bimetallic) gold catalysts was investigated. First, a fundamental insight is given into the mechanism of supported gold catalysts. Different support materials were tested, TiO2 both in the rutile and anatase modification (and mixtures), ZnO and Al2O3, whereas the rutile-supported Au catalyst exhibited the highest activity with high selectivity towards acetaldehyde (> 97%) between 200 and 300C. The main by-product was ethyl-acetate, a result of the coupling of acetaldehyde with ethoxy species. Total oxidation to CO2 was negligible. Bimetallic catalysts, however, have attracted lots of interest as they offer a wide range of possibilities to tune catalytic properties by the choice of the second metal component. For the preparation of bimetallic Au catalysts, different promoter metals were screened (Ag, Ru, Pt), and silver was proven the best suited showing a clear promoting effect on reaction rates. As a non-oxide reference carbon-supported Au and AuAg were used. Thus, kinetic studies performed on Au/rutile, AuAg/rutile as well as the carbon-supported references gave insight into the reaction mechanism and were compared to computational DFT results obtained by J. E. de Vrieze in M. Saeys group, Ghent University, Belgium. With good agreement between the Au/C and theoretical results, different reaction steps (proton transfer of ethanol to atomic oxygen, surface hydroxyl species and molecular oxygen and at higher temperatures ethoxy -H elimination) were identified as rate limiting rather than one dominant route. STEM-HAADF measurements showed comparable particle size distributions for all catalysts with mean particles sizes between 2.7 and 3.5 nm. STEM-EDX revealed bimetallic nature of the AuAg particles on both carbon and rutile after the pretreatment. A range of operando spectroscopy techniques was employed, DRIFTS (Diffuse Reflectance Infrared Fourier Transform Spectroscopy), NAP-XPS (Near Ambient Pressure-XPS) and XAS (X-Ray Absorption Spectroscopy). For both Au/rutile and AuAg/rutile operando DRIFTS revealed ethoxy species as the dominant intermediate on the surface as well as bands of surface acetate that are dependent on conversion. XAS at the Au LIII-edge reveals that the gold remains metallic under all reaction conditions on both Au/rutile and AuAg/rutile, confirming results from the NAP-XPS and computational results that predict the ability of Au0 to activate molecular oxygen. XAS measurements at the Ag K-edge revealed a significant proportion of silver, approximately 50% for AuAg/rutile and ca. 40% for Ag/rutile, to be oxidized under reaction conditions, possibly located within the rutile support and too far away from the surface to be accessible for NAP-XPS: The latter showed a massive depletion from the surface. This observation contradicts the generally accepted hypothesis of Ag2O segregation on the nanoparticle surface. For the AuAg/C, an alloying and no Ag depletion from the surface is hypothesized, explaining the different kinetic results. Thus, under reaction conditions the nanoparticle surface is strongly enriched in Au. NAP-XPS further revealed an increased amount of Ti3+ species under ethanol dehydrogenation conditions, without oxygen, as compared to ethanol oxidation, thus possibly explaining different reaction orders on carbon-supported catalysts. The influence of water was investigated and differences between the gas-phase and the aqueous phase, the latter of which was investigated by partners at the ETH Zurich, were discussed. Finally, the influence of water in the reaction feed was investigated and compared to data from aqueous-phase reaction, the latter performed by S. Mostrou-Moser in the van Bokhoven group. A reversed trend of active catalysts was found, with the catalyst activity ordered as follows in the liquid-phase (where the main product is acetic acid): AuPt > Au > AuRu > AuAg (with Au being the more interesting catalyst as it shows higher selectivity). Surprisingly, the AuAg/rutile, favored in the gas-phase reaction, is the least active one in the liquid phase (gas phase: AuAg > Au > AuRu > AuPt). Gas-phase experiments with water in the feed revealed that water has an activating effect on Au/rutile and overall just minor effects on Ag/rutile. For the bimetallic AuAg, a mixed effect of Au and Ag was shown. Interestingly, the promotional effect was dependent on the presence of rutile and could not be observed for Au/C for the investigated temperatures. Thus, it could be proved that the results of the liquid-phase could not be reproduced in the gas-phase by water introduction, showing that the different behavior must be attributed to the liquid-phase and a different reaction mechanism must be assumed. Also, for the gas-phase, NAP-XPS involving a water feed could not show any water-induced electronical changes and DRIFTS did not reveal any different species on the catalyst surface leading to the conclusion that the effect is a merely kinetic one. In conclusion, the rutile supported AuAg catalyst is a very promising system for ethanol oxidation to acetaldehyde. In this thesis, the common believe that Ag atoms dispersed in the Au surface are responsible for the increased catalyst performance has been disproved. The Ag atoms migrate into the rutile support and in doing so, cause a promotion effect. The specific mechanism behind this effect, however, remains strongly debated and will require further investigation.