Jul 2026· Journal of the American Chemical Society· Vol 148 31, pp.
33934-33944
· 0 citations· 72 references
Medicine
TL;DR
A mechanism-driven approach to alleviate data dependence and develop a multisource transfer learning (MS-TL) framework that leverages the knowledge embedded in abundant adsorption data sets while accurately capturing local structural dependence, enabling a deep fusion of multidimensional thermodynamic knowledge while preserving local structural information.
Abstract
In complex heterogeneous systems, data-driven catalyst discovery is severely hindered by the scarcity of kinetic data and the breakdown of traditional linear scaling relationships caused by the diverse local coordination environments. Herein, we formulate a mechanism-driven approach to alleviate data dependence and develop a multisource transfer learning (MS-TL) framework that leverages the knowledge embedded in abundant adsorption data sets while accurately capturing local structural dependence. Taking methane C-H activation as a representative case, this framework extracts key thermodynamic descriptors corresponding to the initial, transition, and final states as source domains, enabling a deep fusion of multidimensional thermodynamic knowledge while preserving local structural information. Using this framework, we achieved universal predictions of barriers across various facets and compositions in complex alloys. Subsequent data-driven analysis recovers the classical Sabatier principle beyond the limits of linear scaling, revealing a multidimensional volcano-shaped trend that delineates the optimal catalytic window. Furthermore, we propose a temperature-barrier composite kinetic descriptor that quantitatively bridges microscopic theoretical calculations with macroscopic experimental methane oxidation rates, establishing a new data-driven paradigm for rational catalyst design under realistic operating conditions.
Dual-atom catalysts (DACs) provide a powerful platform for oxygen electrocatalysis, yet rational design remains limited by the lack of transferable mechanistic principles. Machine learning (ML) has the potential to address this gap, yet its role in mechanistic discovery remains largely underexplored despite its wide use in catalyst screening. Here, using extended phthalocyanines (M1M2-ePc), we establish an integrated DFT-ML-experiment framework that maps catalytic performance onto an interpretable electronic landscape. Screening 81 DFT-computed and 360 ML-predicted metal pairs identifies FeM-ePc as a promising bifunctional catalyst family. Notably, SHapley Additive exPlanations (SHAP) analysis highlights the importance of electronic background and the key role of the secondary metal in regulating catalytic activity. First-principles calculations further uncover a cooperative dual-descriptor mechanism, in which d-band center and charge transfer jointly govern bifunctional activity. Combining LASSO with SISSO yields compact analytical formulas that quantitatively reproduce ηORR and ηOER, providing interpretable descriptors for DACs. Guided by these findings, FeCo-ePc-L with atomically dispersed Fe-Co sites was synthesized to experimentally examine the ML-guided prediction. This work highlights the utility of interpretable ML for mechanistic discovery in DACs by revealing role-asymmetric electronic cooperation between paired metal centers.
Shao-Bo Jia, Lu Yang, Chou Wu et al.· Advances in Materials· 0 citations
A data-driven framework combining explainable machine learning (ML) with large-scale virtual library generation with large-scale virtual library generation is presented, establishing a practical route from experimental data to actionable catalyst designs.
Xuefeng Li, Haoke Qiu, Hanwen Pei et al.· Journal of Physical Chemistr...· 0 citations
Porous catalytic materials, including metal-organic frameworks (MOFs), covalent organic frameworks (COFs), zeolites, and porous carbons, provide structurally defined microenvironments for controlling reactivity and are increasingly being investigated in electrocatalysis. In electrochemical systems, potential-dependent adsorption energetics, electric double-layer structure, solvent effects, and mass transport within confined pores introduce additional layers of complexity beyond conventional heterogeneous catalysis. Decoding structure-reactivity relationships under such conditions therefore requires representation strategies that are explicitly aligned with reaction-relevant states. This review summarizes recent data-driven strategies used to interrogate porous catalysts, organized around three themes: (i) chemically informed, descriptor-based models that connect local structure to activity/selectivity/stability; (ii) graph-based representations that encode connectivity and topology to learn reactivity-relevant motifs; and (iii) multimodal and transferable learning approaches that integrate structural, spectroscopic and energetic information across material classes. Representative examples across MOFs/COFs, zeolites and porous carbons are discussed, with emphasis on studies that pair modelling with mechanistic reasoning and targeted experiments. Key bottlenecks remain, including the scarcity of reaction-resolved electrochemical datasets, limited treatment of dynamic restructuring under bias, and mismatches between computational descriptors and experimentally measurable observables. We conclude by outlining priorities for reaction-relevant descriptor design and model-experiment feedback loops to accelerate porous electrocatalyst development for sustainable chemical transformation.
Hao Wang, Zhiming Feng, Jie Yang et al.· Chemical Communications· 0 citations
We introduce an active-learning framework that closes the loop between high-throughput EIS measurements and structure-aware composition descriptors to discover superionic candidates under realistic processing constraints. Starting from a small seed set, Gaussian-process and tree-based models propose batched experiments that maximize information gain on conductivity and activation energy while enforcing uncertainty-aware Kramers–Kronig quality gates. Descriptor families integrate interpretable features: ionic radius mismatch, framework softness, site connectivity from simple graph-derived motifs, and processing proxies (grain size from Scherrer, porosity, interphase penalty terms). We demonstrate rapid convergence to high-conductivity regions in multi-component chalcogenide and halide spaces using the automated multi-site EIS workflow described separately. Across three material spaces, the approach reduces experiments ~3× versus grid sampling while yielding candidates with improved conductivity at moderate temperatures and stable impedance upon cycling. We release a lightweight, reproducible stack (metadata schema, analysis notebooks, and synthetic datasets) to encourage community benchmarking without proprietary infrastructure. The result is a pragmatic path to self-driving electrolyte discovery that prioritizes experimental tractability and interpretability—features that matter for industrial translation and cross-lab reproducibility.
Keywords:
active learning; Bayesian optimization; EIS QC; interpretable descriptors; high-throughput screening; solid electrolytes
We report an end-to-end computational-experimental workflow for the discovery of metal-organic frameworks (MOFs), demonstrated by the computational design and synthesis of two novel Zn-based frameworks, UCHI-1 and UCHI-2, exhibiting enhanced methane uptake and selectivity at low pressure under ambient conditions (298 K, 1 bar). The workflow enables the rational selection and experimental realization of metal-organic frameworks combining data mining, machine-learning driven adsorption prediction, and structure generation, with experimental synthesis and validation within a closed-loop discovery pipeline. Analysis of existing and newly generated MOFs reveals the structure-property relationships governing low-pressure methane adsorption, identifying an optimal pore size and shape, framework densities, linker functionalities, and framework topologies that maximize dispersive C-H/π and van der Waals interactions. Beyond the specific materials identified herein, the results establish this workflow as a scalable and extensible platform for accelerated MOF discovery, with clear routes toward further optimization and automation while demonstrating practical applicability beyond purely theoretical exploration of hypothetical materials.
Andrea Darù, Jianheng Ling, Xiaoliang Wang et al.· Journal of the American Chem...· 0 citations
Emerging opportunities in physics-informed machine learning, graph neural networks, generative artificial intelligence, active learning, and autonomous closed-loop DFT-ML-MKM workflows are discussed as promising directions for accelerating the discovery of next-generation electrocatalysts with enhanced activity, selectivity, and long-term stability.
Swetarekha Ram, Shalini Tomar, S. Bhattacharjee· Chemical Communications· 0 citations