Integrating High Throughput Experimentation (HTE) and Machine Learning (ML) in Catalytic Olefin Polymerization: Worth the Effort ?

The integration of High Throughput Experimentation (HTE)[1-2] and Machine Learning (ML)[3-5] is considered the best practice in important areas of chemical R&D, notably pharma. In others instead, like organometallic chemistry (OMC), that is not (yet) the case, hinting that out-of-the-box ML-guided workflows face significantly steeper challenges than in adjacent fields.[6] An inherent drawback of the HTE/ML approach to OMC is that the available experimental databases are comparatively small, while Quantitative Structure-Properties Relations (QSPR) are very complex and difficult to determine. Here, we address the question whether the benefits of implementing and exploiting an ML-aided state-of-art HTE workflow for a specific case of OMC, namely catalytic olefin polymerization, are worth the high Capex, Opex and technical complexity of the endeavor.