(101c) Physical Descriptors That Control Metal-Support Interactions Identified with DFT and Statistical Learning | AIChE

(101c) Physical Descriptors That Control Metal-Support Interactions Identified with DFT and Statistical Learning

Heterogeneous catalysts featuring transition metal nanoparticles supported on oxide surfaces play an essential role in energy, environment, and chemical technologies. Synergistic interactions between the supported metal and the oxide surface can alter catalytic behavior and therefore must be understood at a fundamental level to tune overall catalytic activity, selectivity, and stability. Such interactions are complex, often resulting from the simultaneous action of multiple independent phenomena. It is imperative that we identify fundamental descriptors related to properties of the metal, the support, and adsorbates from the reaction environment that control metal-support interactions. Identification of such descriptors enables one to screen metal-support pairs for interaction properties that are well-suited for the reaction environment in which the catalyst will operate. To address this need, we employ density functional theory (DFT) in concert with screening approaches derived from statistical learning (SL) to identify generalized physical descriptors of the metal-support interaction. DFT is used to generate adsorption energy and charge distribution data for a wide range of metal-support pairs in the presence of various common adsorbates. This data then serves as the training set for SL screening tools that scan descriptor feature space for properties that significantly impact the overall metal-support interaction. Our results demonstrate that the reducibility of the oxide support, the oxophilicity of the supported metal, and the metal-metal bond strength between the adsorbed metal and the parent metal of the oxide support all play key roles governing metal-support interactions. Insight gained from this methodology leads to concrete principles informing the rational design of oxide-supported metal catalysts, as it provides predictive models for identifying useful modifications of the support via the introduction of defects, dopants, and promotors that will alter the extent and nature of the metal-support interaction.

Topics