Ceria (CeO) has recently been found to be a promising catalyst in the selective hydrogenation of alkynes to alkenes. This reaction occurs primarily on highly dispersed metal catalysts, but rarely on oxide surfaces. The origin of the outstanding activity and selectivity observed on CeO remains unclear. In this work, we show that one key aspect of the hydrogenation reaction-the interaction of hydrogen with the oxide-depends strongly on the presence of O vacancies within CeO. Through infrared reflection absorption spectroscopy on well-ordered CeO(111) thin films and density functional theory (DFT) calculations, we show that the preferred heterolytic dissociation of molecular hydrogen on CeO(111) requires H pressures in the mbar regime. Hydrogen depth profiling with nuclear reaction analysis indicates that H species stay on the surface of stoichiometric CeO(111) films, whereas H incorporates as a volatile species into the volume of partially reduced CeO(111) thin films (x ∼ 1.8-1.9). Complementary DFT calculations demonstrate that oxygen vacancies facilitate H incorporation below the surface and that they are the key to the stabilization of hydridic H species in the volume of reduced ceria.

An atomic-level understanding of dioxygen activation on metal oxides remains one of the major challenges in heterogeneous catalysis. By performing a thorough surface-science study of all three low-index single-crystal surfaces of ceria, probably the most important redox catalysts, we provide a direct spectroscopic characterization of reactive dioxygen species at defect sites on the reduced ceria (110) and (100) surfaces. Surprisingly, neither of these superoxo and peroxo species was found on ceria (111), the thermodynamically most stable surface of this oxide. Applying density functional theory, we could relate these apparently inconsistent findings to a sub-surface diffusion of O vacancies on (111) substrates, but not on the less-closely packed surfaces. These observations resolve a long standing debate concerning the location of O vacancies on ceria surfaces and the activation of O on ceria powders.

Density functional theory is used for periodic models of monomeric vanadia species deposited on the CeO2(111) surface to study dissociative adsorption of methanol and its subsequent dehydrogenation to formaldehyde. Dispersion-corrected PBE+U calculations are performed and compared with HSE and B3LYP results. Dissociative adsorption of methanol at different sites on VO2·CeO2(111) is highly exothermic with adsorption energies of 1.8 to 1.9 eV (HSE+D). Two relevant pathways for desorption of formaldehyde are found with intrinsic barriers for the redox step of 1.0 and 1.4 eV (HSE+D). The calculated desorption temperatures (370 and 495 K) explain the peaks observed in temperature-programmed desorption experiments. Different sites of the supported catalyst system are involved in the two pathways: (i) methanol can chemisorb on the CeO2 surface filling a so-called pseudovacancy and the H atom is transferred to an V-O-Ce interphase bond or (ii) CH3OH may chemisorb at the V-O-Ce interphase bond and form a V-OCH3 species from which H is transferred to the ceria surface, providing evidence for true cooperativity. In both cases, ceria is directly involved in the redox process, as two electrons are accommodated in Ce f states forming two Ce(3+) ions whereas vanadium remains fully oxidized (V(5+)).

We study the dissociative adsorption and oxidative dehydrogenation of methanol at the pristine and O-defective ceria (111) surfaces to understand the role of surface oxygen vacancies. The accuracy of two density functional theory based approaches (PBE+U and the HSE hybrid functional) is assessed on available experimental data. In addition, the impact of dispersion correction and zero-point vibrational energy contributions is discussed. Calculated vibrational frequencies are compared with experimental IR spectra. Using dispersion-corrected PBE+U, we obtain comparably large intrinsic barriers for the oxidation step at the pristine (104 kJ/mol) and defective (119 kJ/mol) ceria surfaces. Compared to HSE+D, these barriers are underestimated by 20 kJ/mol. Adsorption energies for the pristine surface agree well between PBE+U+D (−88 kJ/mol) and HSE+D (−86 kJ/mol). However, adsorption energies for the defective surface vary by 25 kJ/mol (PBE+U+D: −231 kJ/mol; HSE+D: −206 kJ/mol). Nonetheless, adsorption into surface oxygen defects is thermodynamically highly favored. As a result, oxygen vacancies are preferred active sites for methanol oxidation in temperature-programmed desorption experiments.

Graphene-based single-atom catalysts are promising alternatives to platinum-based catalysts for fuel cell applications. Different transition metals have been screened using electronic structure methods by estimating onset potentials from the most endergonic elementary reaction step. We calculate onset potentials for the oxygen reduction reaction on metal atoms embedded in Nsubstituted graphene di-vacancies by virtue of first-principlesinformed microkinetic analysis. We find that for more oxophilic metals (Cr, Fe, Mn, and Ru), purely thermodynamic models systematically underestimate onset potentials. Furthermore, the oxophilic metals (Cr, Fe, Mn, and Ru) are oxidized under reaction conditions, leading to an increase in activity compared to their reduced state. Importantly, coadsorbed O m H n species actively participate in the reaction, which requires a dynamic treatment of spectator species. These findings highlight the limitations of thermodynamic analyses for electrocatalytic processes, which commonly assume the same oxidation state for each metal, and show that deviations between computational and experimental onset potentials cannot be solely attributed to the shortcomings of the electronic structure methods.

We study 14 transition metals on pristine and N-doped graphene using density functional theory. For double vacancies, nitrogen doping increases the binding strength of harder transition metals to the support and reduces their oxygen affinity. Inversely, the oxygen affinity of softer metals increases. Since O2 binding energies are correlated with the CO oxidation barrier in a volcano-like trend, doping also affects the activity of the single-atom catalyst. Among these systems, Fe atoms embedded in N-doped graphene are the most active CO oxidation catalysts. These insights can be used to guide the synthesis of highly active oxidation catalysts from nonprecious metals.

Catalytic performance is known to be influenced by several factors, with the catalyst surface oxidation state being the most prominent of all. However, in the great majority of industrial heterogeneous catalytic reactions the oxidation state of the active sites is a subject of intense debate. Preferential oxidation of carbon monoxide in hydrogen-rich mixtures (COPrOx) is an important heterogeneous catalytic reaction involved in the purification of hydrogen. In the search for an economically affordable catalyst for this reaction, cobalt appears as one of the most promising materials. Therefore, it is essential to distinguish if a certain surface state of cobalt exists that favors the reaction more than the others. Here we directly compare the activity of CoO and Co3O4 phases by combining ex situ and operando analytical methods with first-principles modeling. CoO is identified as the most favored cobalt surface oxidation state for the COPrOx process, providing a critical element for the rational design of efficient and durable COPrOx catalysts.

Ceria nanoparticles may expose (100) and (111) facets depending on the preparation method. Motivated by that, we study the reactions of methanol with the CeO 2 (100) surface using dispersion-corrected PBE+U subject to periodic boundary conditions and compare with results for the CeO 2 (111) surface. At (100) facets, the oxidative dehydrogenation of methanol to formaldehyde occurs with an intrinsic barrier of only 0.94 eV (91 kJ/mol), which is significantly lower than the corresponding barrier of 1.23 eV (119 kJ/mol) at the (111) surface. The lower barrier maps to lower formaldehyde desorption temperatures as observed in temperature-programmed desorption spectroscopy. The higher activity is in accordance with predicted lower oxygen defect formation energy in the (100) surface. Thus, defect formation energies are valid reactivity descriptors for oxidation reactions following a Mars-van Krevelen mechanism. At both surfaces, formaldehyde either desorbs or forms a bridging dioxymethylene intermediate, which can be further oxidized into carbon oxides. However, formaldehyde desorption from the (100) surface is significantly more endothermic. Thus, methanol oxidation at the (100) surface is predicted to yield both formaldehyde and carbon oxides, whereas at the (111) surface, formaldehyde can easily desorb. Therefore, the desorption step appears to be the key to the selectivity of methanol oxidation to formaldehyde.