[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-126766-en":3,"doc-seo-126766-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},126766,962084928432,"Emma Wilson","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","Automated transition metal catalysts discovery and optimisation with AI and Machine Learning - Review","Automated exploration of chemical space using AI and machine learning enables the discovery and optimisation of catalysts, extending data-driven strategies proven in organic and pharmaceutical contexts. The review addresses why transition-metal catalyst design is harder, due to coordination geometry, spin states, stability and selectivity, and the higher cost of evaluating excited and transition-state properties. It outlines workflows combining efficient descriptor generation—from molecular mechanics to DFT—with ML-guided screening across complex reaction dimensions such as solvent, temperature and additives, while emphasizing the need for accurate, low-cost modelling to generate training data.","ChemCatChem  \nReview  \n[doi.org/10.1002/cctc.202301475](doi.org/10.1002/cctc.202301475)  \n[www.chemcatchem.org](www.chemcatchem.org)  \nAutomated transition metal catalysts discovery and optimisation with AI and Machine Learning  \nSamuel Mace,[a] Yingjian Xu,*[b] and Bao N. Nguyen*[a]  \nDedicated to Prof. John M. Brown (FRS) on his 84th birthday  \nSignificant progress has been made in recent years in the use of AI and Machine Learning (ML) for catalyst discovery and optimisation. The effectiveness of ML and data science techniques was demonstrated in predicting and optimisingenantioselectivity and regioselectivity in catalytic reactions through optimisation of the ligands, counterions and reaction conditions. Direct discovery of new catalysts/reactions is more difficult and requires efficient exploration of transition metal  \nchemical space. A range of computational techniques for descriptor generation, ranging from molecular mechanics to DFT methods, have been successfully demonstrated, often in conjunction with ML to reduce computational cost associated with TS calculations. Complex aspects of catalytic reactions, such as solvent, temperature, etc., have also been successfully incorporated into the ML optimisation and discovery workflow.  \nIntroduction  \nAutomated chemical space exploration with the help of AI/ Machine Learning (ML) is a highly important methodology in modern chemical discovery. Progresses in this area with organic compounds have resulted in the first AI discovered Active Pharmaceutical Ingredient (API) entering Phase II trials. [1,2] The same benefits can be extended to catalyst discovery through chemical space exploration of organometallic compounds. However, this is significantly more challenging due to the additional constraints, e.g. coordination geometry, and complexity, e.g. spin state, catalyst stability and selectivity, etc. Evaluating the desired function of catalysts for in silico screening is also more computationally demanding compared to API discovery, due to the need to calculate and/or estimate properties of excited states and transition states. In homogeneous catalysis, additional dimensions such as solvent, temperature and additives can have a significant impact on reaction outcome and need to be included in the evaluation methodology. Synthetic catalytic reactions often involves chemo- and stereoselectivity, competing side reactions, and multiple possi-  \n[a] S. Mace, B. N. Nguyen  \nInstitute of Process Research & Development  \nSchool of Chemistry, University of Leeds  \nWoodhouse Lane, Leeds, LS2 9JT, United Kingdom  \n[E-mail: b.nguyen@leeds.ac.uk](E-mail: b.nguyen@leeds.ac.uk)  \n[b] Y. Xu  \nGoldenKeys High-tech Materials Co., Ltd.  \nBuilding 3, Guizhou Industrial Investment Technology Industrial Park Gui’an New District, Guizhou Province, 550008, China  \n[E-mail: goldenkeys9996@thegoldenkeys.com.cn](E-mail: goldenkeys9996@thegoldenkeys.com.cn)  \n © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH. This isan open access article under the terms of the Creative Commons Attribution Non-Commercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.  \nble mechanistic pathways,[3–9] depending on the substrate and catalyst (Figure 1) . [10–13]  \nThese complex and demanding challenges led to the need for of AI/ML models which can predict catalytic activity. This  \nFigure 1. An example mechanism of the Ullmann-Goldberg coupling reaction, with a reaction condition dependent deactivation pathway and multiple possible mechanisms for the rate determining step (RDS) . [17]  \nChemCatChem  \nReview  \n[doi.org/10.1002/cctc.202301475](doi.org/10.1002/cctc.202301475)  \n18673899,  \n0,  \napproach can be particularly powerful for complex and difficult substrates, which tend to occur in high value chemical synthesis. Unfortunately, experimental data on catalytic activity in this area is scarce, with ","cbCaifDsKHpgZzO5","https://ap.wps.com/l/cbCaifDsKHpgZzO5","pdf",13847523,1,13,"English","en",105,"# Introduction\n## Automated chemical space exploration\n## Challenges in transition-metal catalyst evaluation\n# Automated exploration of ligand space\n## Chemical space as a cornerstone\n## ML-guided screening workflow\n# Computational methods for data generation\n## Descriptor generation (from mechanics to DFT)\n## Using ML to reduce transition-state cost\n# Dealing with complex catalytic aspects\n## Selectivity, reaction conditions, competing pathways","[{\"question\":\"Why is AI/ML-based catalyst discovery harder for transition metals than for many organic targets?\",\"answer\":\"Transition-metal systems add constraints such as coordination geometry and spin states, and require evaluating properties like catalyst stability and selectivity. In silico screening also becomes more computationally demanding because excited states and transition states must be calculated or estimated.\"},{\"question\":\"How does the review describe typical ML workflows for catalyst discovery and optimisation?\",\"answer\":\"A workflow begins with experimental or computational data for a small set of ligands/catalysts, then trains a machine learning model to predict catalytic properties. The model is used to extrapolate performance to a much larger, generated set of candidates.\"},{\"question\":\"What role do descriptor generation and computational chemistry play in enabling ML-guided screening?\",\"answer\":\"Descriptor generation is used to represent catalysts and reactions, ranging from molecular mechanics to DFT methods. These computational tools are often combined with ML to reduce the cost associated with transition-state calculations and to support low-cost training data generation.\"}]","Automated transition metal catalysts discovery and optimisation with AI and Machine Learning - Review | PDF",1785934669,33,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"automated-transition-metal-catalysts-discovery-and-optimisation-with-ai-and-machine-learning-review","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/automated-transition-metal-catalysts-discovery-and-optimisation-with-ai-and-machine-learning-review/126766/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is AI/ML-based catalyst discovery harder for transition metals than for many organic targets?","Question",{"text":75,"@type":76},"Transition-metal systems add constraints such as coordination geometry and spin states, and require evaluating properties like catalyst stability and selectivity. In silico screening also becomes more computationally demanding because excited states and transition states must be calculated or estimated.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the review describe typical ML workflows for catalyst discovery and optimisation?",{"text":80,"@type":76},"A workflow begins with experimental or computational data for a small set of ligands/catalysts, then trains a machine learning model to predict catalytic properties. The model is used to extrapolate performance to a much larger, generated set of candidates.",{"name":82,"@type":73,"acceptedAnswer":83},"What role do descriptor generation and computational chemistry play in enabling ML-guided screening?",{"text":84,"@type":76},"Descriptor generation is used to represent catalysts and reactions, ranging from molecular mechanics to DFT methods. These computational tools are often combined with ML to reduce the cost associated with transition-state calculations and to support low-cost training data generation.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]