Work

From Empirical Tuning to Predictive Design: Decoupling Dynamic Bottlenecks in Photocatalytic H 2 O 2 Production via Z‐Score Importance Analysis

Year: 2026
Type: article
Abstract: Deconvoluting the intertwined dynamic bottlenecks of photogenerated charge and reaction substrates remains a fundamental bottleneck in heterogeneous photocatalysis due to the lack of reasonable and effective tools. This study presents a quantitative, interpretable framework to deconvolve the entangled factors limiting solar‐driven photocatalysis. Using Au(I)‐based covalent metal‐organic framework (CMOF‐Au) isomers with controlled stacking (AA, AB, and ABC) as a model platform for hydrogen peroxide photosynthesis, we applied a Z‑score and LOFO (Leave‐One‐Feature‐Out) importance analysis. This approach revealed that the dominant rate‐determining factor shifts with stacking: bulk charge transfer governs performance in AA stacking, whereas surface adsorption limits AB and ABC stacking. The method successfully quantifies the relative contribution weights of key functions (charge dynamics, interfacial transfer, surface adsorption, and mass transport) and demonstrates high predictive power across diverse photocatalysts, including covalent organic frameworks, porous aromatic frameworks, C 3 N 4 , and overall water‐splitting systems. By uncovering counterintuitive mechanisms, such as Simpson's paradox, our work establishes a generalizable strategy for decoupling multivariate factors, moving photocatalyst design from empirical tuning toward a predictive, function‐aware paradigm.
Language: English
Field: Energy
Open Access status: bronze
OpenAlex ID: W7236021975
FWCI:
Cites: 67
Cited by:
Related to: 10