Committee Members Information

• Ph.D. Advisor: Professor Aamir Farooq
• External Examiner: Professor Joris Thybaut from Ghent University
• Committee Chair: Professor Yoji Kobayashi
• 4th Committee Member:  Javier Martinez

Abstract:  

The emission of volatile organic compounds (VOCs), particularly methane and ethylene, from industrial and automotive sources poses significant environmental and health risks. Catalytic oxidation is a highly effective strategy for VOC abatement, but its widespread application is hindered by the high cost and susceptibility to poisoning of traditional noble-metal catalysts by common exhaust components like water and CO. Transition metal oxides offer a cost-effective alternative, though they often exhibit lower activity and stability, especially under realistic conditions involving water vapor and CO. This thesis addresses these challenges by designing and developing robust, noble-metal-free catalysts through a multi-faceted approach that integrates advanced synthesis strategies with a data-driven design framework.

Three key strategies were employed. First, a nano-confinement "med-synthesis" approach was developed to encapsulate manganese oxide nanoparticles within the mesoporous channels of SBA-15. This method resulted in highly dispersed (~7.3 nm) and thermally stable nanoparticles that demonstrated significantly enhanced catalytic activity and durability for ethylene oxidation, achieving 90% conversion (T₉₀) at 250 °C, a 50 °C improvement over its wet impregnation prepared counterpart. Second, Co₃O₄/LaCoO₃ heterostructures were fabricated via a one-step hydrothermal synthesis to create catalysts with robust poison resistance for methane oxidation. The optimized La_0.25Co_0.75 catalyst exhibited exceptional stability in wet (10% H₂O) and CO-rich feeds, a performance attributed to synergistic interfacial effects that enhanced lattice oxygen mobility and surface hydrophobicity.

 Finally, a data-driven framework was established to accelerate the discovery of complex quintinary (La-Co-Fe-Mn-Ni) multi-element oxide catalysts. By integrating Bayesian neural networks with a genetic algorithm, this approach successfully navigated the vast compositional space to identify two distinct, high-performance catalysts: La_0.06Co_0.33Fe_0.02Mn_0.07Ni_0.52 for dry methane combustion and, using transfer learning, La_0.1Co_0.16Fe_0.09Mn_0.38Ni_0.27 for the simultaneous oxidation of methane and ethylene in a wet, mixed-gas feed. A pivotal contribution of this work is the extension of the machine learning models to accurately predict not only catalytic performance but also fundamental material properties, including full XRD patterns and BET surface areas, directly from precursor compositions.

Collectively, this research provides novel synthesis methodologies and establishes a new paradigm for the rational, accelerated design of next-generation VOC abatement catalysts. The findings demonstrate a clear pathway toward developing highly active, durable, and cost-effective solutions for environmental catalysis under industrially relevant conditions.