Active Learning Exploration of Thermally Conductive Polymers Under Strain

Finding amorphous polymers with higher thermal conductivity (TC) is technologically important, as they are ubiquitous in a wide range of applications where heat transfer is crucial. While TC is generally low in in amorphous polymers, it can be enhanced by subjecting them to strain since it facilitates the alignment of polymer chains. However, using the conventional Edisonian approach, the discovery of polymers that may have high TC after strain can be time-consuming and lack guarantee of success. In recent years, machine learning approaches have shown promise in the prediction of polymer properties with high accuracy and efficiency. In this work, we employ an active learning scheme to speed up the discovery of amorphous polymers with high TC under strain. Polymers under strain were simulated using molecular dynamics (MD), and their TCs were calculated using the non-equilibrium MD method. A Gaussian Process Regression (GPR) model is then trained using these data. The GPR is used to screen the PoLyInfo database, and the predicted mean TC and uncertainty are used towards an acquisition function to recommend new polymers. The TC results of these selected polymers are then labeled using MD simulations and the obtained data are incorporated into the training set, initiating a new iteration of the active learning cycle. Over a few cycles, we identified more strained polymers with significantly higher TC than the original dataset, and the results offer valuable insights into the structural characteristics that contribute to enhanced energy conduction from a physical perspective.