Influence of Crystal Facet and Phase of Titanium Dioxide on Ostwald Ripening of Supported Pt Nanoparticles from First-Principles Kinetics

Metal oxide plays an important role on stability and catalytic performance of supported metal nanoparticles, but mechanistic understanding of structure sensitivity and optimization of the oxide supports remains elusive in heterogeneous catalysis. Taking Ostwald ripening of platinum nanoparticles supported on titanium dioxide (TiO2) as an example, we reveal here a great structure sensitivity of oxide facets and crystal phases on sintering of supported metal nanoparticles through first-principles kinetic simulation. Total activation energies of the Pt ripening on various pristine TiO2 surfaces of both anatase and rutile phases are calculated by density functional theory, and Ostwald ripening under isothermal condition and temperature programmed condition are simulated numerically. Calculated total activation energies are found inversely proportional to the corresponding oxide surface energies, and vary considerably from 1.76 to 3.56 eV. The Pt ripening rate on the pristine TiO2 surfaces follows the order of r(001) ≈ a(001) ≫ a(100) ≈ r(101) > r(100) > a(101) ≈ r(110). For TiO2 support exposing different facets, not only their intrinsic ripening rate but also their relative surface area determines the overall ripening kinetics and formation of transit bimodal particle size distribution. For pristine anatase TiO2 exposing a(001) and a(101) facets, ripening starts on a(101) facets only after ripening on a(001) facets finishes due to their order of magnitude difference in ripening rate, resulting a step-wise increase of average particle size with ripening time. For pristine rutile TiO2 exposing r(101) and r(110) facets, ripening could proceed simultaneously on both facets due to their modest difference in ripening rate, and the average particle size increases monotonically with ripening time. Compared to rutile TiO2, anatase TiO2 supports are less resistant to the metal nanoparticles ripening since a(001) facets with high ripening rate is likely to be exposed. The present work is compared to available experiments and the theoretical framework established could be expanded to various metal and oxide systems