| Summary: | Transition metal carbides and nitrides are promising catalytic supports because of their well-defined and intrinsically tunable structures. Such supports span a vast space of materials (~103) and exhibit tunable dimensionality (2D MXenes or 3D structures), surface chemistry (unary or binary combinations of p-block elements as surface terminations), and bulk compositions. For these materials to fulfil their promise as next-generation catalyst supports, we first require a systematic understanding of how modifications in dimensionality, surface chemistry, and composition, influences the thermal stability, reactivity, and selectivity of the metal catalysts that they support. Density Functional Theory methods combined with machine learning are well-suited to establish this systematic understanding, because these methods can efficiently parse through the vast space of different carbide and nitride materials.
Using machine learning models and experiments, the work function of a given 2D MXene is shown to vary by >0.5 eV, by tailoring the surface chemistry of the MXene. Feature importance analyses of the machine learning models provide qualitative insights into the elemental properties that influence this work function. This tunability in the work function indicates that tailoring the surface chemistry of MXenes can alter the extent of charge transfer or the polarization of adsorbed metal films. Building on this central hypothesis, the influence of polarization of epitaxial metal films induced by carbide/nitride supports is investigated on the adhesion energies and sintering temperatures of these metal films. A linear model to estimate adhesion energies of arbitrarily thick metal films that are supported on carbides/nitrides is presented. This model predicts adhesion energies of metal films with errors of 0.01-0.05 eV/Å2 relative to the DFT training set. Model-predicted adhesion energies are then used to assess the stability of single atom catalysts on carbides/nitrides and determine the sintering temperatures of 5000+ supported metal films. My analysis reveals that sintering temperatures of metals can be maximized by tuning the relative composition of two p-block elements (e.g., C* and F*) that terminate the support.
Upon considering metrics of thermal stability, the trends in reactivity descriptors are assessed on supported metal carbides and nitrides. Using site-specific scaling relationships and CO* as a probe adsorbate, adsorption trends are systematized at active sites spanning different nuclearity (adatoms, corner sites, edge sites, closed packed sites) and different proximities to the metal/support interface. The slopes of the scaling relationships indicate that the structure sensitivity of CO* adsorption at these interfacial metal sites is influenced by the support. Negatively charged interfacial gold active sites exhibit structure insensitive CO* adsorption with adatoms and closed-packed active sites yielding similar adsorption strengths. Conversely, positively charged interfacial gold active sites possess strongly structure sensitive CO* adsorption. This relationship between structure-sensitivity and interfacial charge transfer is observed across different carbide, nitride, sulphide, and oxide supports, and is explicated using acid-base interactions.
After systematizing the stability and reactivity trends on supported metal carbides/nitrides; the influence of tuning the extent of polarization induced by the support is examined. The two-electron reduction of CO2 to CO, HCOOH, and H2 is selected as a probe reaction. Linear scaling relations and machine learning methods are employed to systematize adsorption energies of various reaction intermediates. Using the adsorption energies of CO* and OH* as descriptors, the thermodynamic driving force is classified towards different CO2 electro-reduction products. My findings indicate that the polarization of metal sites that is induced by the support, can fundamentally alter selectivity. For example, while gold sites located at least three metal layers away from the interface favour the formation of CO, interfacial gold sites exhibit a Cu-like reactivity, thereby favouring the formation of C1+ products. More broadly, this principle of polarizing interfacial metal sites can be used to widen the space of materials yielding C1+ products, to beyond traditional Cu-based catalysts. Taken together, this thesis advances first principles-based strategies for evaluating the stability, reactivity, and selectivity of metal catalysts that are supported on 2D and 3D carbides and nitrides.
|
|---|