Molecular fingerprint and machine learning enhance high-performance MOFs for mustard gas removal

• DOI: 10.1016/j.isci.2024.110042
• Publication Date: 2024-05-21T02:21:08Z
• Authors: Jing Ni; Jinfeng Li; Shuhua Li; He Zheng; Zhongyuan Ming; Li Li; Heguo Li; Shouxin Zhang; Yue Zhao; Hong Liang; Zhiwei Qiao
• Keywords: Chemistry; Science; Natural sciences; Q; 0210 nano-technology; Materials science; Article; 0104 chemical sciences

Highlights•Three key descriptors were found to influence the HD adsorption loading of MOFs•Ten high-performance MOFs identified for removal mustard gas from air•Three crucial substructures screened out design new materials•Random Forest and MACCS molecular fingerprinting can obtain a good predicted abilitySummaryChemical warfare agents (CWAs), epitomized by notoriously used (HD), represent class exceptionally toxic chemicals whose airborne is paramount battlefield safety. This study integrates high-throughput computational screening (HTCS) with advanced machine learning (ML) techniques investigate efficacy metal-organic frameworks (MOFs) in adsorbing capturing trace amounts present air. Our approach commenced comprehensive univariate analysis, scrutinizing impact six distinct on efficiency MOFs. analysis elucidated pronounced correlation between MOF density Henry coefficient effective capture HD. Then, four ML algorithms employed train predict performance The Random (RF) algorithm demonstrates strong model generalization, achieving best prediction result 98.3%. In novel exploratory stride, we incorporated 166-bit (MF) identify critical functional groups within adsorbents. From top 100 analyzed, 22 optimal identified. Leveraging these insights, designed three innovative substructures, grounded groups, enhance efficiency. this work, combination MF could provide direction efficient outcomes offer substantial potential revolutionize domain CWA capture. represents significant stride toward developing practical solutions that both environmental protection security.Graphical abstract