Reports: DNI552279-DNI5: Rapid Prediction of Sulfide and Oxide Catalyst Surface Structures Using Insight from Density Functional Theory

Lars C. Grabow, PhD, University of Houston

Introduction. We use density functional theory (DFT) calculations to study equilibrium surface terminations of related metal sulfide and oxide surface under hydrotreating conditions. Hydrotreating is a technology used in petroleum refineries to remove sulfur from the refined products and is the largest application of catalysis by the amount of feed processed.1 To describe surface reactions on metal sulfide catalysts (e.g. MoS2) during hydrodesulfurization an accurate representation of the active site model under reaction conditions is needed.2 Similarly, metal oxides are often used as catalysts or support material for downstream processes in the petrochemical industry, and their surface structure also depends on reaction conditions. Our work is based on the central hypothesis that metal sulfide and oxide surfaces exhibit similar behavior under hydrotreating conditions and the knowledge from one system can be translated to the other. Such correlation enables faster computational discoveries of suitable materials for metal sulfide and oxide catalysts.

Current Progress. The challenge to translate surface termination knowledge from a metal sulfide to the corresponding metal oxide is illustrated in Fig. 1. The comparison of the thermodynamically stable MoS2(-1010) and MoO3(010) phases with varying pressures of H2S/H2O and H2 clearly indicates significant differences, which are to a large part caused by the structural differences between the two crystal lattices.

Description: Macintosh HD:Users:grabow:Documents:UH:Proposals:2011:11 ACS-PRF DNI:Report 2014:MoO3 and MoS2 Phase Diagram.png

Figure 1 Surface Phase Diagram for MoS2(-1010) and MoO3(010) under hydrotreating conditions.

To make the problem more tractable we have first focused on a set of materials with the same crystal structure. We choose to investigate metal oxides in the rutile crystal phase, including the very popular catalyst components TiO2 and RuO2. Figure 2 shows a systematic enumeration of the possible RuO2(110) surface terminations obtained by adding H atoms (vertical) or creating O vacancies (horizontal).

Figure 2 Grid representation of surface terminations obtained by adding H (vertical) or creating O vacancies (horizontal) to the stoichiometric RuO2(110) surface (bottom left).

We calculated the surface free energies of the phases given in Fig. 2 for RuO2, also for the rutile (110) surfaces of PdO2, RhO2, PtO2, Ru/TiO2, IrO2, VO2, and TiO2. We found that the changes in surface free energy on across all surfaces can be described by just the O vacancy formation energy only. This result is summarized in Fig. 3.

Figure 3 Linear scaling relations for rutile surface free energies as a function of oxygen vacancy formation energy.

The linear scaling behavior for surface free energies shown in Fig. 3 can be used to rapidly predict phase diagrams for rutile oxides from the vacancy formation energy only.

Figure 4 Comparison between the RuO2(110) phase diagram from (a) DFT calculations, and (b) using linear scaling relations (b).

Last year, we reported that also the vacancy formation kinetics can be related to the vacancy formation energy, which allowed us to predict a kinetic phase diagram using kinetic Monte Carlo simulations. The phase diagram in Fig. 5 is given for the simple reaction H2 + RuO2 ¨ RuO2-VO + H2O, but includes 8 elementary reaction steps and their respective activation energy barriers. While the graphs in Fig. 4 and 5 look similar, they should not be expected to be identical: the key difference is that Fig. 4 includes only thermodynamics, whereas Fig. 5 also accounts for reaction kinetics. Further, in Fig. 5 lateral interactions between surface species was neglected, which is not the case in Fig. 4.

 

Figure 5 Kinetic phase diagram for RuO2(110) obtained from kinetic Monte Carlo simulations.

Impact on Career and Participating Students.  This ACS-PRF supported project has significantly contributed to multiple successful outcomes in my research group. Most notably, we used the insights we obtained from studying rutile metal-oxide surface structures for the development of a comprehensive Au/TiO2 catalyst model. In a NSF/DOE sponsored project in collaboration with the groups of Drs. Chandler and Pursell at Trinity University, we have then studied CO oxidation on this catalyst model and were able to determine the roles of water, support hydroxyls, and the metal/support interface. The work resulted in a novel reaction mechanism involving proton transfer from support water for oxygen activation and was published in Science.3 Another career highlight was the receipt of the DOE Early Career Award in 2014, which was a direct outcome of our ACS-PRF supported research. I received two invitations to present our work at the ACS Spring and Southwest Regional Meeting, and contributed talks to the ACS Fall and AIChE Annual Meeting. The participating graduate students had opportunities to disseminate their research in one oral presentation (ACS Spring 2014, B. Baek) and four poster presentations. S. Kasiraju won the Best Poster Award for his work comparing sulfides and oxides under hydrotreating conditions at the International Summer School on Electronic Structure Theory and Materials Design offered at DTU, Denmark. Most recently, UH undergraduate student Connor Fernandez received the Provost Undergraduate Research Scholarship (PURS) to continue our work on rutile metal oxides and their ability to cleave C-O bonds.

Summary and Future Work. We have demonstrated that the surface free energy of rutile oxides can be estimated from the oxygen vacancy formation energy only. Surface phase diagrams based on linear scaling, ab-initio thermodynamics, and sophisticated kinetic Monte Carlo methods are in good agreement, suggesting that time-consuming DFT calculations can be minimized. A detailed comparison to MoS2-type metal sulfides is currently underway. In the future we plan to extend the approach to more complex sulfides and oxides and have begun characterizing the MoO3 surface, which presents a bigger challenge because of its three distinguishable oxygen atoms.

References

1.  R. Prins, "Hydrotreating" in Handbook of Heterogeneous Catalysis (Wiley-VCH Verlag GmbH & Co. KGaA: 2008, pp. 2695–2718).

2.  P. G. Moses, L. C. Grabow, E. M. Fernandez, B. Hinnemann, H. Tops¿e, K. G. Knudsen and J. K. N¿rskov, "Trends in Hydrodesulfurization Catalysis Based on Realistic Surface Models", Catal. Letters 144 (2014) 1425–1432.

3.  J. Saavedra, H. A. Doan, C. J. Pursell, L. C. Grabow and B. D. Chandler, "The critical role of water at the gold-titania interface in catalytic CO oxidation", Science 345 (2014) 1599–1602.