Electronic Structure Modulation in Dopant‐Controlled Single‐Atom Graphene Catalysts for Efficient Hydrogen Evolution: A Machine Learning and First‐Principles Study
ENERGY & ENVIRONMENTAL MATERIALS · 19.05.2026
DOI: 10.1002/eem2.70421 ↗Kurz gefasst
Fachartikel in der Zeitschrift ENERGY & ENVIRONMENTAL MATERIALS, online erschienen am 19.05.2026. Verfasst von Sajjad Ali und 3 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Single‐atom catalysts (SACs) have emerged as a promising class of materials for achieving high catalytic efficiency and maximum atom utilization. However, their design requires atomic‐level understanding of metal–support interactions. Herein, we combine first‐principles density functional theory (DFT) and machine learning (ML) to study 26 transition metals (3 d –5 d ) on pristine, boron (B)‐ and nitrogen (N)‐doped graphene (G) in axial and pyridinic sites (104 single‐atom configurations). Heteroatom doping significantly improves the metal binding; N‐doped graphene leads to the highest adsorption energies (E ads ), lowest d‐band centers, and shortest bond lengths to the support, with an inverse relationship between adsorption energy, bond length, charge transfer, and d‐band position across all systems. Mo, Ti, Hf, Zr, Os, and Nb (early/mid‐TM) form the most stable TM/G single‐atom sites, while Zn, Ag, and Au are weakly adsorbed due to filled d orbitals. Furthermore, the adsorption of CO 2 , CO, H 2 O, H, H 2 , O 2 reveal dopant‐dependent‐reactivity. Hydrogen adsorption free‐energy calculations reveal Pd/B‐G(axial), Y/B‐G(pyridinic), Zn/N‐G(pyridinic), and V/N‐G(axial) as near‐thermoneutral systems at the hydrogen evolution reaction (HER) valcono peak, where N‐G preferentially stabilizes hydrogen‐adsorption and while B‐G allows for balanced turnover in adsorption–desorption. A follow‐up descriptor analysis and machine‐learned gradient‐boosted regression (GBR) model based on DFT features shows R 2 = 0.72 on cross‐validation and R 2 = 0.64 on leave‐one‐metal‐out (LOMO) validation. SHapley Additive exPlanations (SHAP) analysis provides a descriptor hierarchy: Bader‐charge (Q) > period > elevation (h) > d‐band center (ε d ), where the metal‐to‐support charge transfer is the most influential descriptor for ΔG H* , and moderate charge transfer (Q ≈ 0.3–0.8 |e|) on B‐G is the most active. Overall, these results provide a mechanistic picture and a design principle for graphene‐based single‐atom catalysts for HER.
Zitieren
APA
Ali, S., Ismail, P. M., Humayun, M., & Bououdina, M. (2026). Electronic Structure Modulation in Dopant‐Controlled Single‐Atom Graphene Catalysts for Efficient Hydrogen Evolution: A Machine Learning and First‐Principles Study. ENERGY & ENVIRONMENTAL MATERIALS. https://doi.org/10.1002/eem2.70421
Vancouver
Ali S, Ismail PM, Humayun M, Bououdina M. Electronic Structure Modulation in Dopant‐Controlled Single‐Atom Graphene Catalysts for Efficient Hydrogen Evolution: A Machine Learning and First‐Principles Study. ENERGY & ENVIRONMENTAL MATERIALS. 2026. doi:10.1002/eem2.70421
BibTeX
@article{Ali2026,
author = {Ali, Sajjad and Ismail, Pir Muhammad and Humayun, Muhammad and Bououdina, Mohamed},
title = {Electronic Structure Modulation in Dopant‐Controlled Single‐Atom Graphene Catalysts for Efficient Hydrogen Evolution: A Machine Learning and First‐Principles Study},
journal = {ENERGY & ENVIRONMENTAL MATERIALS},
year = {2026},
doi = {10.1002/eem2.70421}
}
EUVIMED