(420f) Molecular Weight Distribution Modulation and Mechanical Property Prediction of Linear Polymers: Combining Molecule Simulation and Machine Learning

The molecular weight distribution (MWD) of linear polymers plays a critical role in determining their mechanical properties, which are essential for designing materials that exhibit high tensile strength and excellent rheological property and processability¹. However, conventional experimental methods for studying the specific effects of MWD on mechanical properties can be time-consuming and labor-intensive. Inspired by structure–property framework with organosilica synthesis,² we propose an integrated design that combines kinetic Monte Carlo (KMC), molecular dynamics (MD) simulation, and machine learning (ML) to efficiently modulate MWD and uncover the intricate relationship between MWD and the resulting mechanical properties. First, we employ KMC simulations to generate various linear polymer with different MWD by adding initiators and monomers in a semi-batch manner. The simulated KMC results are validated by comparing them to published experimental data.³ Next, the MD simulations are performed based on the KMC results to investigate the effects of different structural parameters on the mechanical properties, such as interaction, bond energy, average chain length, and polydispersity index (PDI). The results show that strong interactions and bond energies enhance the Young's modulus of the system due to more significant bond orientation and chain entanglement. The impact of chain length and PDI on mechanical properties are also investigated, and the results show that their effects on mechanical properties are non-monotonous. Furthermore, we build machine learning models based on molecular dynamics simulation results to link the structural parameters to the mechanical properties. These models predict the mechanical properties of the polymer, such as tensile strength, Young's modulus, and elongation at break. Our results demonstrate that the proposed machine learning models exhibit high predictive accuracy and robustness. This study showcases the potential of integrating molecular simulations and machine learning techniques for efficiently modulating MWD and predicting the mechanical properties of linear polymers. The proposed method can significantly expedite the development of new polymeric materials with customized mechanical properties, paving the way for their use in a diverse range of fine and everyday chemical applications.

References:

(1) Gentekos, D. T.; Sifri, R. J.; Fors, B. P. Controlling Polymer Properties through the Shape of the Molecular-Weight Distribution. Nat Rev Mater 2019, 4 (12), 761–774.

(2) De Keer, L.; Kilic, K. I.; Van Steenberge, P. H. M.; Daelemans, L.; Kodura, D.; Frisch, H.; De Clerck, K.; Reyniers, M.-F.; Barner-Kowollik, C.; Dauskardt, R. H.; D’hooge, D. R. Computational Prediction of the Molecular Configuration of Three-Dimensional Network Polymers. Nat. Mater. 2021, 20 (10), 1422–1430.

(3) Gentekos, D. T.; Dupuis, L. N.; Fors, B. P. Beyond Dispersity: Deterministic Control of Polymer Molecular Weight Distribution. J. Am. Chem. Soc. 2016, 138 (6), 1848–1851.