Model-aided catalyst prediction through descriptor-based hybrid semi-empirical approach
MetadataShow full item record
Model-aided catalyst prediction through descriptor-based hybrid semi-empirical approach Microkinetic model aided catalyst prediction is an advanced method for developing improved and novel catalysts without intensive empirical testing in heterogeneous catalysis. The model offers the possibility to probe reaction mechanisms and predict the activity and selectivity of catalysts. Density functional theory (DFT) calculation is currently the most prevalent approach to accurately estimate input parameters (surface energetics) of microkinetic model. However, DFT calculations are normally complex, time-consuming and computationally expensive. As such, a descriptor-based hybrid semi-empirical approach (UBI-QEP + BEP) is advocated here to rapidly acquire energetics for transition metal surface reactions. In the present thesis, computational catalyst prediction is performed based on following strategy: Descriptors → Adsorption energies → Activation energies → Activity. An improved UBI-QEP method effectively associates adsorbates binding energies to atomic binding energies (descriptors), with which estimated adsorption heats satisfactorily fit DFT and experimental values. BEP relationships correlate activation energies with reaction heats, which are obtained towards C-H, C-O, C-C, O-H and C-O-H bond activation from DFT data. Microkinetic modeling serves as a measure to relate energetics to activity, and model refinement is conducted via adopting DFT estimated energetics of rate-determining steps. A preliminary microkinetic modeling of steam methane reforming (SMR) demonstrates that the strategy displays reasonable accuracy with respect to DFT computations as well as experiments, but reduces radically six orders of magnitude in the computational expenses. Subsequently, the strategy is utilized in SMR and Fischer-Tropsch synthesis (FTS) to investigate the size-dependent and metal-dependent activity as well as mechanism. Size-dependent activity of SMR is researched on Rh, Ni, Pt and Pd catalysts by using microkinetic modeling combined with a truncated octahedron model, which illustrates the higher activity on the smaller sized metal particles. Substantially higher activity of M(211) than M(111) and M(100) accompanied by decreased M(211) surface fraction results in reduced metal activity as the particle size increases. Towards metal-dependent activity, the activity trend is Rh > Ni > Pd ~ Pt for SMR at both small and large particle size. Pd and Pt are energetically unfavorable; Ni activity is restricted by surface blocking of C*/CH* on Ni(100) and O* on Ni(211) surface. This work offers a feasible approach to gain insight into the size-dependent and metal-dependent activity as well as mechanism in heterogeneous nanocatalysis. The metal-dependent mechanism of FTS demonstrates that CO insertion is the main mechanism for FT reaction. CO activation occurs primarily via H-assisted CO dissociation. The major C1/ C2 chain growth take place through CH + CO/ CH3C + CO coupling. Metal-dependent activity indicates that Cu, Pt and Pd are energetically unfavorable and perform quite low activity towards light olefin formation. Fe is the energetically most preferred catalyst, but extremely low vacancy coverage leads to low activity. Co, Ni, Ru and Rh are active due to both appropriate free energy and vacancy coverage, among which Co exhibits the highest activity. Ni3Cu and Ni3Zn are suggested as potential bimetallic catalysts to aid rational catalyst design for FTO (Fischer-Tropsch to Olefins) through screening of large amounts of A3B type bimetallic catalysts, which show both high activity and economic advantage. It is illustrated that the proposed strategy is a powerful method to discriminate reaction mechanisms and predict catalyst activity, and may extend to other reaction systems in future.