Theoretical study of structure and catalytic properties of heterogeneous catalysts.

摘要

Two embedded cluster models, the surface charge representation of the electrostatic embedding potential (SCREEP) method and the full quantum embedded cluster (FQEC) method, have been presented and applied to systems of heterogeneous catalysts, particularly Zeolites and the TiO2 surface. In the SCREEP method, the active region is treated by a quantum calculation and the Madelung potential from the extended lattice is accounted for by the surface charges enclosing the quantum region. In a more rigorous approach, the FQEC method, the interactions in crystal structure of infinite lattices are modeled by an approximate periodic ab initio calculation and the interactions in the active region are corrected by a more accurate ab initio calculation. The accuracies of these two approaches have been evaluated on the system of H-Chabazite, known to be very sensitive to its environment, to access the importance of the crystal effects on electronic and catalytic properties. Three other systems, i.e., CO and NO adsorptions on Cu-ZSM-5, Pt/H-ZSM-5, and the water dissociation on TiO2 (110) surface have been examined by the SCREEP method. The FQEC method has been revealed to provide accurate structures and energetics consistent with periodic quantum calculations but requires modest computational resources, especially in the case that the crystal effects are crucial. Despite the lack of crystal polarization, the SCREEP embedding method can model the chemistry and adsorption mode of those systems with satisfactory accuracy while lesser computational resources are required. Both cluster size and lattice effects, especially the Madelung potential, have been indicated to be very crucial for modeling crystalline systems. Several issues have been elucidated successfully, in particular the electronic states of the catalytically active species and the metal-support interactions, which are critical to the catalysis. The understanding of these systems at the molecular level has been achieved by using the embedded cluster models without several problems encountered in conventional periodic calculations, such as the symmetry constraints, and the computational burden required. Thus the embedded cluster models presented in this dissertation are promising models for zeolite and metal oxide studies.