{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_53412"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_53412","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"INTEGRATING MACHINE LEARNING AND DENSITY FUNCTIONAL THEORY TO UNRAVEL AND OPTIMIZE CATALYTIC REACTION MECHANISMS","abstract":"Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","abstract_html":"Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","abstract_has_math":false,"creators":["Rodriguez, Austin R."],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Mendoza-Cortes, Jose JM","Lira, Carl CL","Ross, Arun AR","Ferrier, Robert RF","Calabrese Barton, Scott SCB"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T18:59:53Z","subjects":["Chemical engineering","Computational chemistry","Computational physics","Chemistry--Computer simulation","Physics--Computer simulation"],"languages":["eng"],"rights":["Attribution 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53412","local:Rodriguez_grad.msu_0128D_22471"],"render_values":[{"text":"etd:53412","href":null,"code":true},{"text":"local:Rodriguez_grad.msu_0128D_22471","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/5cw3-te49","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/5cw3-te49","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_53412","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Mendoza-Cortes, Jose JM","Lira, Carl CL","Ross, Arun AR","Ferrier, Robert RF","Calabrese Barton, Scott SCB"]},{"key":"dc:creator","label":"Author","values":["Rodriguez, Austin R."]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Chemical engineering","Computational chemistry","Computational physics","Chemistry--Computer simulation","Physics--Computer simulation"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53412","local:Rodriguez_grad.msu_0128D_22471","https://doi.org/doi:10.25335/5cw3-te49"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","Catalytic reaction mechanisms are governed by features of the potential energy surface that are difficult to observe directly: reactive intermediates, transition states, competing selectivity branches, and the molecular environments that stabilize them. This dissertation develops an integrated computational strategy for using density functional theory (DFT), curvature-aware machine-learning interatomic potentials (MLIPs), and auditable agentic workflows to unravel these mechanisms and provide the mechanistic information needed to optimize catalysts toward preferred pathways.DFT provides the mechanistic reference for this work. Studies of aluminum-based epoxide polymerization, PDI-supported CO2 reduction, and CO2/epoxide copolymerization show how electronic-structure calculations connect catalyst structure to intermediates, rate- or selectivity-determining transition states, and competing reaction branches. These applications demonstrate the value of DFT for mechanistic interpretation while also exposing the cost of exhaustive transition-state and pathway screening.To address this scaling problem, this dissertation develops curvature-aware MLIPs for reactive potential energy surfaces. Hessian-trained models improve reactive extrapolation beyond conventional energy and force training, particularly near transition states, intrinsic-reaction-coordinate structures, and off-equilibrium geometries. Projected Hessian Learning replaces full Hessian supervision with Hessian-vector-product labels, retaining much of the benefit of curvature supervision while avoiding explicit construction of dense Hessian matrices and reducing training cost relative to full-Hessian learning.The same integration requires reliable computational infrastructure. Bounded agentic workflows are developed to generate chemically meaningful structures, preserve fragment metadata, pre-optimize with lower-cost methods, prepare and audit Gaussian16 calculations, classify failures, and record retry decisions. This provenance layer makes DFT and MLIP data generation more reproducible and supports closed-loop catalyst discovery.Together, these contributions support a platform in which DFT validates critical mechanistic steps, curvature-aware MLIPs accelerate exploration of reactive potential energy surfaces, and bounded agentic systems prioritize the calculations that most improve catalyst design. The aim is not to optimize a mechanism independently of the chemistry that governs it, but to identify how catalyst structure can lower barriers, suppress unproductive branches, or promote the pathway that gives the desired product. The dissertation concludes by proposing this integration for CO2/epoxide copolymerization, where projected-Hessian MLIPs could enable fast transition-state search and soft-mode enhanced sampling within an auditable catalyst-discovery loop.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (114 p.) : ill."]},{"key":"dc:title","label":"Title","values":["INTEGRATING MACHINE LEARNING AND DENSITY FUNCTIONAL THEORY TO UNRAVEL AND OPTIMIZE CATALYTIC REACTION MECHANISMS"]}]}],"canonical_facts":{"dc:contributor":["Mendoza-Cortes, Jose JM","Lira, Carl CL","Ross, Arun AR","Ferrier, Robert RF","Calabrese Barton, Scott SCB"],"dc:creator":["Rodriguez, Austin R."],"dc:date":["2026"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Chemical Engineering - Doctor of Philosophy, 2026","Catalytic reaction mechanisms are governed by features of the potential energy surface that are difficult to observe directly: reactive intermediates, transition states, competing selectivity branches, and the molecular environments that stabilize them. This dissertation develops an integrated computational strategy for using density functional theory (DFT), curvature-aware machine-learning interatomic potentials (MLIPs), and auditable agentic workflows to unravel these mechanisms and provide the mechanistic information needed to optimize catalysts toward preferred pathways.DFT provides the mechanistic reference for this work. Studies of aluminum-based epoxide polymerization, PDI-supported CO2 reduction, and CO2/epoxide copolymerization show how electronic-structure calculations connect catalyst structure to intermediates, rate- or selectivity-determining transition states, and competing reaction branches. These applications demonstrate the value of DFT for mechanistic interpretation while also exposing the cost of exhaustive transition-state and pathway screening.To address this scaling problem, this dissertation develops curvature-aware MLIPs for reactive potential energy surfaces. Hessian-trained models improve reactive extrapolation beyond conventional energy and force training, particularly near transition states, intrinsic-reaction-coordinate structures, and off-equilibrium geometries. Projected Hessian Learning replaces full Hessian supervision with Hessian-vector-product labels, retaining much of the benefit of curvature supervision while avoiding explicit construction of dense Hessian matrices and reducing training cost relative to full-Hessian learning.The same integration requires reliable computational infrastructure. Bounded agentic workflows are developed to generate chemically meaningful structures, preserve fragment metadata, pre-optimize with lower-cost methods, prepare and audit Gaussian16 calculations, classify failures, and record retry decisions. This provenance layer makes DFT and MLIP data generation more reproducible and supports closed-loop catalyst discovery.Together, these contributions support a platform in which DFT validates critical mechanistic steps, curvature-aware MLIPs accelerate exploration of reactive potential energy surfaces, and bounded agentic systems prioritize the calculations that most improve catalyst design. The aim is not to optimize a mechanism independently of the chemistry that governs it, but to identify how catalyst structure can lower barriers, suppress unproductive branches, or promote the pathway that gives the desired product. The dissertation concludes by proposing this integration for CO2/epoxide copolymerization, where projected-Hessian MLIPs could enable fast transition-state search and soft-mode enhanced sampling within an auditable catalyst-discovery loop.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (114 p.) : ill."],"dc:identifier":["etd:53412","local:Rodriguez_grad.msu_0128D_22471","https://doi.org/doi:10.25335/5cw3-te49"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution 4.0 International"],"dc:subject":["Chemical engineering","Computational chemistry","Computational physics","Chemistry--Computer simulation","Physics--Computer simulation"],"dc:title":["INTEGRATING MACHINE LEARNING AND DENSITY FUNCTIONAL THEORY TO UNRAVEL AND OPTIMIZE CATALYTIC REACTION MECHANISMS"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:53Z"}