Algorithm-Driven Robotic Discovery of Polyoxometalate-Scaffolding Metal–Organic Frameworks

• DOI: 10.1021/jacs.4c09553
• Publication Date: 2024-10-09T13:05:56Z
• Authors: Donglin He; Yibin Jiang; Melanie Guillén-S...; Zack Geary; Lucia Vizcaíno-Anaya; Daniel Salley; Maria Del Carmen ...; De-Liang Long; Leroy Cronin
• Keywords: Polyoxometalate

The experimental exploration of the chemical space crystalline materials, especially metal-organic frameworks (MOFs), requires multiparameter control a large set reactions, which is unavoidably time-consuming and labor-intensive when performed manually. To accelerate rate material discovery while maintaining high reproducibility, we developed machine learning algorithm integrated with robotic synthesis platform for closed-loop polyoxometalate-scaffolding (POMOFs). eXtreme Gradient Boosting (XGBoost) model was optimized by using updating data obtained from uncertainty feedback experiments multiclass classification extension based on POMOF their constitution. digital signatures POMOFs were represented universal description language (χDL) to precisely record synthetic steps enhance reproducibility. Nine novel including one mixed ligands derived individual through imidization reaction POM amine derivatives various aldehydes have been discovered good repeatability. In addition, maps plotted XGBoost models whose F1 scores are above 0.8. Furthermore, electrochemical properties synthesized indicate superior electron transfer compared molecular POMs direct effect ratio Zn, type used, topology structures in modulating abilities.