Using Machine Learning to Forecast the Conductive Substrate-Supported Heteroatom-Doped Metal Compound Electrocatalysts for Hydrogen Evolution Reaction DOI
Nana Zhou, Yaling Zhao,

Qingzhang Lv

et al.

The Journal of Physical Chemistry C, Journal Year: 2024, Volume and Issue: 128(41), P. 17274 - 17281

Published: Oct. 8, 2024

The heteroatom-doped metallic compounds supported on conductive substrates are excellent catalysts for the hydrogen evolution reaction (HER) thanks to their tunable properties, e.g., and nonmetallic compositions, especially bimetallic active centers synergistic effect, as well morphology interaction between substrate. Only optimal combination these adjustable properties other external factors could endow remarkable HER catalytic activity of catalysts. Therefore, in this study, machine learning (ML) database based plenty from publicly available data was conducted train three different ML models, various features including electrolyte type, catalyst morphology, compositions (metallic nonmetallic) ratios, additive, substrate were analyzed figure out impacts overpotential (OP) values determine outstanding According feature importance Spearman coefficient analysis, metal elements ratio determined be Pt, Mo 0.5, heteroatoms nitrogen, sulfur, nickel foam. Finally, model predicts that foam nickel-supported composed Pt Mo2S3 codoped with nitrogen sulfur (N, S-doped Pt@Mo2S3) exhibits admirable performance alkaline electrolytes a pretty low OP value 33 mV. database-guided provides an alternative rapid screening prediction electrocatalysts.

Language: Английский

Interpretable physics-informed machine learning approaches to accelerate electrocatalyst development DOI Open Access
Hao Wu, Mingxuan Chen, Hao Cheng

et al.

Journal of Materials Informatics, Journal Year: 2025, Volume and Issue: 5(2)

Published: Feb. 26, 2025

Identifying exceptional electrocatalysts from the vast materials space remains a formidable challenge. Machine learning (ML) has emerged as powerful tool to address this challenge, offering high efficiency while maintaining good accuracy in predictions. From perspective, we provide brief overview of recent advancements ML for electrocatalyst discoveries. We emphasize applications physics-informed (PIML) models and explainable artificial intelligence (XAI) development, through which valuable physical chemical insights can be distilled. Additionally, delve into challenges faced by PIML approaches, explore future directions, discuss potential breakthroughs that could revolutionize field development.

Language: Английский

Citations

0
Establishing Quantitative Structure–Activity Relationships for the Degradation of Aromatic Organics by UV–H2O2 Using Machine Learning DOI

Zhongli Lu,

Jiming Liu, Xuqian Zhang

et al.

Industrial & Engineering Chemistry Research, Journal Year: 2025, Volume and Issue: unknown

Published: March 26, 2025

The degradation of aromatic organic compounds in aquatic environments is critical due to their persistence and toxicity. This study establishes a machine learning (ML)-driven quantitative structure–activity relationship model predict the pseudo-first-order reaction rate constants (K) for UV–H2O2 organics. A data set comprising 134 experimental observations 30 was constructed, integrating conditions, quantum chemical parameters, physicochemical properties. Among six ML algorithms evaluated, gradient boosting decision tree emerged as optimal model, with feature importance analysis identifying H2O2 concentration, topological polar surface area, q(C)min dominant factors. Theoretical calculations supported by linking higher reactivity o,p'-dicofol lower energy gaps elevated electrophilic susceptibility. Additionally, establishment interpretable expressions not only provides transparency clarity predictions but also aids economic analysis, which highlighted that mildly acidic pH low UV light intensity, along suitable concentrations, are cost-effective conditions process. work bridges chemistry elucidate mechanisms, offering rapid resource-efficient tool optimizing advanced oxidation processes.

Language: Английский

Citations

0
Data-Driven Insights into Porphyrin Geometry: Interpretable AI for Non-Planarity and Aromaticity Analyses DOI Creative Commons

Shachar Fite,

Zeev Gross

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: April 20, 2025

Porphyrins are involved in numerous and very different chemical biological processes, due to the sensitivity of their application-relevant properties subtle structural changes. Applying modern machine learning methodology is appealing for discovering structure-activity relationships that can be used design tailor-made porphyrins specific purposes. For achieving this goal, a high-quality set consisting 425 metal was established via curation 7590 porphyrin structures from Cambridge crystallographic database. Using data-driven techniques analyzing nonplanarity "structural aromaticity" allowed validation common knowledge field as well discovery new relations. Aromaticity found influenced differently by distinct nonplanar distortions. Nonplanarity more sensitive macrocycle substitutions than or axial ligand effects, while ruffled distortions dominated size properties. These findings offer insights into structure-property porphyrins, providing foundation targeted synthesis tune aromaticity nonplanarity. Despite data limitations, work demonstrates value uncovering complex trends.

Language: Английский

Citations

0
Design and Application of Electrocatalyst Based on Machine Learning DOI Creative Commons

Yulan Gu,

Hailong Zhang, Zhen Xu

et al.

Interdisciplinary materials, Journal Year: 2025, Volume and Issue: unknown

Published: May 15, 2025

ABSTRACT Data‐driven artificial intelligence provides strong technical support for addressing global energy and environmental issues. The powerful data processing analysis capabilities of machine learning (ML) can quickly predict electrocatalytic performance, improving the efficiency catalyst design time‐consuming inefficient nature traditional design. By integrating ML with theoretical calculations experiments, catalytic reaction processes be precisely regulated. This not only accelerates discovery new catalysts but also drives development more efficient environmentally friendly sustainable technologies. In this article, we discuss approaches to discovering novel driven by ML, focusing on activity prediction, barrier optimization, innovative materials. We systematically application in field electrocatalysis explore future prospects domain. provide a comprehensive in‐depth its potential development.

Language: Английский

Citations

0
Unraveling the effect of single atom catalysts on the charging behavior of nonaqueous Mg–CO2 batteries: a combined density functional theory and machine learning approach DOI
Rafiuzzaman Pritom, Rahul Jayan, Md Mahbubul Islam

et al.

Journal of Materials Chemistry A, Journal Year: 2023, Volume and Issue: 12(4), P. 2335 - 2348

Published: Dec. 18, 2023

The role of single atom catalysts in improving the charging phenomenon nonaqueous Mg–CO 2 batteries to realize improved performance.

Language: Английский

Citations

8
A review of advancements in theoretical simulation of oxygen reduction reaction and oxygen evolution reaction single-atom catalysts DOI
Ninggui Ma, Yu Xiong, Yuhang Wang

et al.

Materials Today Sustainability, Journal Year: 2024, Volume and Issue: 27, P. 100876 - 100876

Published: June 8, 2024

Language: Английский

Citations

3
Electroreduction of nitrate to ammonia on graphyne-based single-atom catalysts by combined density functional theory and machine learning study DOI

Yushan Pang,

Zongpeng Ding,

Aling Ma

et al.

Separation and Purification Technology, Journal Year: 2024, Volume and Issue: 354, P. 129422 - 129422

Published: Aug. 30, 2024

Language: Английский

Citations

3
Design Principle of Carbon-Supported Single-Atom Catalysts – Interplay between d-Orbital Periodicity and Local Hybridization DOI

Zhengda He,

Jingyang Wang, Bin Ouyang

et al.

Chemistry of Materials, Journal Year: 2024, Volume and Issue: 36(3), P. 1405 - 1412

Published: Jan. 31, 2024

Carbon-based single-atom catalysts (SACs) have been widely investigated as a potential alternative for noble-metal-based the hydrogen evolution reaction (HER) and oxygen reduction (ORR). The rational design of such requires not only physical intuitions but also practical descriptors that can be directly applied in experiments. In this work, we establish theoretical framework based on comprehensive data set SACs compromising 28 metals, 5 types local environments, adsorption calculations 4 adsorbates (e.g., H/O/OH/OOH). We disentangle complex trend H/OH an interplay between d-orbital periodicity hybridization, allowing estimation catalytic performance solely basis number valence electrons. By utilizing framework, identified several promising catalyst candidates overlooked strategies.

Language: Английский

Citations

1
Active learning accelerated exploration of single-atom local environments in multimetallic systems for oxygen electrocatalysis DOI Creative Commons
Hoje Chun,

Jaclyn Lunger,

Jeung Ku Kang

et al.

Published: March 14, 2024

Single-atom catalysts (SACs) exhibit high activity for a wide range of sluggish reactions and allow performance tunability at atomic-level through the selection central metals, ligand environments, secondary metal sites. However, design space with varying structures compositions significantly hinders fast accurate identification desired multimetallic SACs. In this work, we demonstrate self-driving computational strategy exploring binary metallic sites combinations 3d transition metals different resulting in over 30,000 single atom electrochemical catalysis oxygen reduction evolution (ORR/OER). This approach is based on density functional theory (DFT) calculations binding energies atomic descriptors as target properties utilizes an equivariant graph neural network (GNN) surrogate model predicting DFT labels directly from structure. The chemical environments learned by GNN lead to capturing composition-structure-property relationships ORR/OER selectivity. Active learning facilitates investigation search balancing exploration unseen exploitation active ones. predictions promising Co-Fe, Co-Co, Co-Zn pairs are consistent state-of-the-art results experimental measurements reported literature. GNN-based analysis multiple surface catalytic reaction can be extended broader class multi-element entropic materials systems.

Language: Английский

Citations

1
A generalized model for estimating adsorption energies of single atoms on doped carbon materials DOI
Maria G. Minotaki, Julian Geiger, Andrea Ruiz‐Ferrando

et al.

Journal of Materials Chemistry A, Journal Year: 2024, Volume and Issue: 12(18), P. 11049 - 11061

Published: Jan. 1, 2024

Single metal atoms on doped carbons constitute a new class of extremely appealing materials, as they present the best utilization for catalysis.

Language: Английский

Citations

1