A Bayesian Approach to Predict Solubility Parameters

Abstract

<jats:p>Solubility is a ubiquitous phenomenon in many aspects of science. While solubility can be determined by considering the cohesive forces in a liquid via Hansen solubility parameters (HSP), the prediction is often done using Quantitative structure-property relationship models due to its low computational cost. Herein, we report an interpretable and versatile probabilistic approach (gpHSP) to determining HSP. Our model is based on Gaussian processes (GP), a Bayesian machine learning approach, which also provides uncertainty bounds to prediction. gpHSP achieves its flexibility by leveraging a variety of input data, such as SMILES strings, COSMOtherm simulations, and quantum chemistry calculations. gpHSP is built on experimentally determined HSP: a general solvents set aggregated from literature, and a polymer set, in-house characterized. In both sets, we obtained a high degree of agreement, surpassing well-established machine learning methods. We demonstrate the general applicability of gpHSP to miscibility of organic semiconductors, drug compounds and in general solvents, which can be further extended to other domains. gpHSP is a fast and accurate toolbox, which could be applied to molecular design for solution processing technologies.&lt;br&gt;</jats:p>

Machine Learning Orchestrating the Materials Discovery and Performance Optimization of Redox Flow Battery

“…For example, it successfully identifies materials with specific characteristics, such as high redox potential and optimal solubility in electrolytes. [103] This can be achieved while minimizing the computational burden, requiring only a minimal number of DFT calculations. This highlights the potential of Bayesian optimization to accelerate material discovery while reducing reliance on computationally expensive evaluations.…”