Machine Learning\nModels for Predicting Polymer Solubility\nin Solvents across Concentrations and Temperatures

Artificial intelligence\nand machine learning have become\nessential\ntools in predicting material properties to aid in the accelerated\ndesign of new materials. Polymer solubility, critical for new formulations\nand solution processing, is one such property. However, current models\nare limited by inadequate experimental data sets that cannot capture\nthe complexity and detail for many features contributing to polymer\nsolubility. Here, we provide a data set for polymer solution behavior\nbased on Crystal16 turbidity measurements that includes high quality\npercent transmission data for polymer solutions for a variety of polymers,\nsolvents, concentrations and temperatures. We use this data set to\ntrain a model that predicts the experimental transmission data at\nmany temperatures and multiple concentrations. From this, we are able\nto classify the polymer/solvent pairs into three solubility categories\nproviding a level of granularity to predictions beyond prior binary\nclassification models considering only solvent/nonsolvent classes.\nThe inclusion of multiple concentrations, temperatures and partially\nsoluble data expands solubility prediction capability beyond prior\nwork into predictions more attractive for use by formulators and process\ndesigners working with industrial polymer solutions.