The science and technology of catalysis is more important today than at any other time in our history due to the grand energy and environment challenges we are facing. With the explosively growth of computation power nowadays, computer simulation can play an increasingly important role in the design of new catalysts, avoiding the costly trail-and-error attempts and facilitating the development cycle. The goal to inverse design of new materials with desired catalytic property was once far off, but now achievable. The major focus of this dissertation is to find the general rules that govern the catalytic performance of a nanoparticle as the function of its structure. Three types of multi-metallic nanoparticles have been investigated in this dissertation, core-shell, random alloy and alloy-core@shell. Significant structural rearrangement was found on Au@Pt and Pd@Pt nanoparticle, which is responsible for a dramatic improvement in catalytic performance. Nonlin- ear binding trends were found and modeled for random alloy nanoparticles, providing a prescription for tuning catalytic activity through alloying. Studies of ORR on Pd/Au random alloy NP and hydrogenation reaction on Rh/Ag random alloy NP revealed that binding on individual ensemble should be in- vestigated when large disparity of adsorbate affinity is presented between two alloying elements. In the alloy-core@shell system, I demostrated a general linear correlations between the adsorbate binding energy to the shell of an alloy-core@shell nanoparticle and the composition of the core. This relation- ship allows for interpolation of the properties of single-core@shell particles and an approach for tuning the catalytic activity of the particle. A series of promising catalysts were then predicted for ORR, HER and CO oxidation. As a first attempt to bridge the material gap, bimetallic nano clus- ter supported on CeO₂(111) was investigated for CO oxidation. A strong support-metal interaction induces a preferential segregation of the more reac- tive element to the NC-CeO₂ perimeter, generating an interface with the Au component. (Au-Cu)/CeO₂ was found to be optimal for catalyzing CO oxida- tion via a bifunctional mechanism. O₂ preferentially binds to the Cu-rich sites whereas CO binds to the Au-rich sites. A method called distributed replica dynamics (DRD) is proposed at last to utilize enormous distributed computing resources for molecular dynamics simulations of rare-event in chemical reac- tions. High efficiency can be achieved with an appropriate choice of N [subscript rep] and t [subscript rep] for long-time MD simulation. / text

京都大学 / 新制・課程博士 / 博士(エネルギー科学) / 甲第25398号 / エネ博第477号 / 新制||エネ||89(附属図書館) / 京都大学大学院エネルギー科学研究科エネルギー基礎科学専攻 / (主査)教授 片平 正人, 准教授 中田 栄司, 教授 菅瀬 謙治 / 学位規則第4条第1項該当 / Doctor of Energy Science / Kyoto University / DFAM