Unveiling Surface Reaction Networks in Biomass-Derived Compound Upgrading
Year: 2026
Type: preprint
Abstract: Biomass-derived platform molecules often undergo complex surface reactions in which productive routes compete with catalyst deactivation, limiting catalyst selectivity and stability. Identifying the molecular origins of these competing processes remains challenging, as conventional computational approaches generally rely on chemical intuition and hypothesis-driven rules, while strongly adsorbed intermediates and poisoning species remain difficult to characterize experimentally. Here, we combine enhanced sampling reaction discovery techniques with machine learning interatomic potentials, experimental catalytic tests, and in situ DRIFTS to unravel the surface reaction network controlling methyl levulinate conversion on zirconia. Our simulations reveal a highly interconnected network of elementary transformations in which surface Lewis acidic Zr sites and basic oxides cooperatively promote carbonyl activation and α-deprotonation, giving rise to competing pathways toward cyclization and catalyst poisoning. This mechanistic picture is supported by experimental evidence, which reveal a progressive decrease in angelica-lactone formation and the accumulation of persistent carboxylate-like surface species. Our results show that the same cooperative acid base surface motif which activates the substrate and promotes productive cyclization, also induces irreversible surface trapping leading to catalyst poisoning. By revealing this intrinsic competition at the molecular level, our work provides a clue on how to steer zirconia reactivity toward selective biomass conversion while suppressing the formation of the persistent surface species responsible for deactivation. More broadly, our work establishes a general strategy to tackle complex reaction networks in heterogeneous catalysis.
Source: ChemRxiv
Institutions: Italian Institute of Technology, University of Bologna
Language: English
Subfield: Catalysis
Field: Chemical Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7208777492