This talk is driven by the vision of computational design of cluster-based nanocatalysts. The discovery of the extraordinary activity in catalysis exhibited by small metal-oxide clusters has stimulated considerable research interest. However, in heterogeneous catalysis, materials property changes under operational environment (e.g. temperature (T) and pressure (p) in an atmosphere of reactive molecules). Therefore, a solid theoretical understanding at a realistic (T, p) is essential in order to address the underlying phenomena. In this talk, I shall first introduce a robust methodological approach that integrates various levels of theories combined into one multi-scale simulation to address this problem[1]. I shall show one application of this methodology in addressing (T, p) dependence of the composition, structure, and stability of metal oxide clusters in a reactive atmosphere at thermodynamic equilibrium using a model system that is relevant for many practical applications: free metal (Mg) clusters in an oxygen atmosphere[2].

More recently, I have extended this development in designing clusters with desired properties. The novelty of this implementation is that it goes beyond the interpretation of experimental observations and addresses the challenging “inverse problem” of computationally designing clusters with target properties. The methodology is applied and thoroughly benchmarked on (TiO₂)_n clusters [n=2, 3,...., 10, 15, 20][3]. All the results are duly validated using the highest level of theories currently achievable within Density Functional Theory (DFT).

References:

[1] S. Bhattacharya, S. Levchenko, L. Ghiringhelli, M. Scheffler, New J. Phys. 16, 123016 (2014).

[2] S. Bhattacharya, S. Levchenko, L. Ghiringhelli, M. Scheffler, Phys. Rev. Lett. 111, 135501 (2013).

[3] S. Bhattacharya, B. Sonin, C. Jumonville, L. Ghiringhelli, N. Marom, Phys. Rev. B 91, 241115 (R), (2015).

Dry reforming is a process wherein CH₄ and CO₂ react to give synthesis gas and/or liquid fuels. Dry reforming is normally done under high temperature and pressure conditions, with a Ni catalyst, however it has recently been discovered that if a plasma is also present near the catalyst, then it is possible to get this reaction to go under modest conditions close to room temperature and atmospheric pressure. The role of the plasma in this process is poorly understood. In this talk I will describe several electronic structure studies that my group is doing which are designed to describe the processes involved in plasma enhanced dry reforming, including both the role of the plasma, and the gas-surface chemistry that occurs in the presence of plasma species. The plasma is known to fragment the reacting gases, especially CH₄, so we will study the interaction of methane fragments with various Ni surfaces, to show how this enhances chemisorption, surface dissociation, and subsequent reaction with species already on the surface. A highlight of this work involves the reaction of subsurface hydrogen with adsorbed CO₂ to give CO, water and other products. We have also studied the influence and ions and electrons on surface chemistry.

We have studied the structure and reactivity of individual Pt atoms supported on oxidized Cu using scanning tunneling microscopy (STM) and temperature programmed reaction (TPR). We first elucidated the structure of the support, a previously observed but unsolved Cu₂O overlayer on Cu, by density functional theory (DFT) and comparison of simulated STM images with experimental STM. We discovered that the oxide surface is inert towards CO oxidation, while the single Pt atom decorated surface is efficient for low-temperature CO oxidation. This system is also capable of water activation. Our combination of TPR and STM studies suggest that single Pt atoms supported on the oxide are indeed catalytically active and may be a good catalyst for the water-gas shift reaction.

The energetics of the reactions of water with metal oxide surfaces are of tremendous interest for catalysis and electrocatalysis, yet the energy for the dissociative adsorption of water was only previously known on one well-defined oxide surface, Fe₃O₄(111).[1] Here we report the first calorimetric measurement of the heat of reaction for the dissociative adsorption of water on NiO(111)-2x2 as a function of coverage, showing that the heat of dissociative adsorption decreases with coverage from 177 kJ/mol to 119 kJ/mol in the first 0.25 ML of coverage. These measurements provide an important benchmark for validating computational estimates of adsorption energies of molecular fragments on correlated metal-oxide such as NiO and for oxide surface chemistry in general, which is more challenging in this respect than for metal or wide-gap semiconductor surfaces. We also present DFT calculations of the energetics of this reaction, and compare it to the calorimetric results.

[1] P. Dementyev, K.-H. Dostert, F. Ivars-Barceló, C.P. O'Brien, F. Mirabella, S. Schauermann, X. Li, J. Paier, J. Sauer, H.-J. Freund, Water Interaction with Iron Oxides, Angewandte Chemie International Edition, 54 (2015) 13942-13946.

In spite of its technological relevance in the energy conversion and storage, our knowledge about the microscopic structure of electrochemical electrode-electrolyte interfaces and electrical double layers is still rather limited. The theoretical description of these interfaces from first principles is hampered by three facts. i) In electrochemistry, structures and properties of the electrode-electrolyte interfaces are governed by the electrode potential which adds considerable complexity to the theoretical treatment since charged surfaces have to be considered. ii) The theoretical treatment of processes at solid-liquid interfaces includes a proper description of the liquid which requires to determine free energies instead of just total energies. This means that computationally expensive statistical averages have to be performed. iii) Electronic structure methods based on density functional theory (DFT) combine numerical efficiency with a satisfactory accuracy. However, there are severe shortcomings of the DFT description of liquids, in particular water, using current functionals.

Despite these obstacles, there has already significant progress been made in the first-principles modeling of electrochemical electrode-electrolyte interfaces. In this contribution, I will in particular focus on how the electrochemical environment can be appropriately taken into account using numerically efficient schemes. In the presence of an aqueous electrolyte, metal electrodes are in general covered by either cations or anions. Based on the concept of the computational hydrogen electrode, the equilibrium coverage of Pt(111) with hydrogen (1,2) and halides (3) as a function of the electrode potential has been derived showing that halide and hydrogen adsorption is competitive, in agreement with experimental findings. The presence of the aqueous electrolyte has been taken into account modeling water layers either implicitly through a polarizable medium (2) or explicitly in ab initio molecular dynamics runs (3). To obtain a proper description of the water-water and the water-metal interaction, it turns out that the consideration of dispersion corrections is essential (4). The importance of the electrochemical environment in electrocatalytic processes will be demonstrated using the methanol electrooxidation on Pt(111) (5) as an example.

References

(1) S. Sakong, M. Naderian, K. Mathew, R. G. Hennig, and A. Gross, J. Chem. Phys. 142, 234107 (2015).

(2) T. Roman and A. Gross, Catal. Today. 202, 183 (2013).

(3) F. Gossenberger, T. Roman, and A. Gross, Surf. Sci. 631, 17 (2015).

(4) K. Konigold and A. Gross, J. Comput. Chem. 33, 695(2012).

(5) S. Sakong and A. Gross, submitted.

With the recent explosion in computational catalysis and related microkinetic modeling, the need for a fast yet accurate way to predict equilibrium and rate constants for surface reactions has become more important. In such calculations, adsorbates are usually treated within either the 2D lattice gas or 2D ideal gas approximation to estimate their partition functions and entropies. Here we present a fast new method to estimate the partition functions and entropies of adsorbates that is much more accurate than those approximations, and recognizes the true oscillating nature of the adsorbate’s potential energy for motions parallel to the surface. As with previous approaches, it uses the harmonic oscillator (HO) approximation for most of the modes of motion of the adsorbate. However, it uses hindered translator and hindered rotor models for the three adsorbate modes associated with motions parallel to the surface, and evaluates these using an approach based on a method that has proven accurate in modeling the internal hindered rotations of gas molecules. The translational and rotational contributions to the entropy of a hindered translator / hindered rotor calculated with this new method are, in general, very closely approximated (to within <0.25R error per mode) by the corresponding harmonic oscillator (i.e., 2D lattice gas) entropy when kT is less than the barrier. When kT exceeds the barrier, the hindered translator / hindered rotor model is closely approximated (to within 0.1 R) by the entropy of an ideal 2D gas. The harmonic oscillator / lattice gas model severely overestimates the entropy when kT greatly exceeds the barrier. The cutoff between the temperature ranges of applicability of these simple two approximations is very sharp but with our combined hindered rotor/hindered translator approach the whole temperature range is covered with the same approach.

Heterogeneous catalysis by singly-dispersed metal atoms on non-metallic surfaces offers great potential for maximizing the efficiency of metal atoms, and optimizing their activity and selectivity. Herein, we present results from our ab-initio density functional theory (DFT) calculations for methanol partial oxidation (MPO), an industrially important reaction for the production of H₂, on Pd₁/ZnO(101¯0). To begin with we find that the Pd atom prefers to adsorb at the oxygen vacancy site i.e. the anion vacancy is responsible for stabilizing singly-dispersed Pd atom on ZnO(101¯0). We discuss the adsorption characteristics of a set of gas molecules (CH3OH, O₂, CO, CO₂, H₂O, H₂), and the potential energy profile including activation barriers for the reaction processes associated with MPO on Pd₁/ZnO(101¯0), and compare them with those on Pd16Zn16 nanoparticle and pristine ZnO(101¯0). We find that the singly dispersed Pd sites offer a high activity towards the formation of CO₂ and H₂ over that of CO and H₂O. We trace this reactivity to the electronic structure of the single Pd site as modified by its local environment which in turn facilitates a strong binding of CO to the Pd site, thereby increasing the CO desorption barrier and stabilizing O₂ on ZnO(101¯0), which is essential for further oxidation steps. With activation energy barriers and pre-exponential factors calculated from DFT, for a large set of reaction intermediates, we perform kinetic Monte Carlo simulations to determine the turn over frequencies and rate limiting steps in the formation of CO₂ and H₂ on Pd1/ZnO(101¯0), under ambient conditions.

Work supported in part by DOE grant DE-FG02-07ER15842.