Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 9: Crystallization, Nucleation and Self-Assembly II

CPP 9.4: Vortrag

Montag, 18. März 2024, 15:45–16:00, H 2032

Nucleation patterns of polymer crystals analyzed by machine learning models — •Atmika Bhardwaj^{1, 2}, Jens-Uwe Sommer^{1, 2}, and Marco Werner¹ — ¹Institut Theorie der Polymere, Leibniz-Institut für Polymerforschung Dresden e.V., Hohe Strasse 6, 01069 Dresden, Germany — ²Institute für Theoretische Physik, Technische Universität Dresden, Zellescher Weg 17, 01062 Dresden, Germany

We employ unsupervised machine learning algorithms to investigate conformation patterns during polymer crystallization from undercooled melts using simulation data obtained from coarse grained molecular dynamics simulations. Our focus is on data-driven techniques that establish decision boundaries to detect the crystalline state of individual monomers without any prior knowledge, as opposed to classical methods that position these boundaries based on the analysis of stem length distributions [C. Luo and J.-U. Sommer, Macromolecules 44 (2011), 1523]. We utilize self-supervised auto-encoders to compress local conformation fingerprints of individual monomers. We show that the compressed fingerprints are organized in a lateral map that can be associated with the degrees of conformation and orientation order in the monomer’s environment. The Gaussian mixture model allows to distinguish the crystalline from the amorphous phase by identifying a dense region within that map that reflects the reduced degrees of freedom. The high specificity of the method allows us to uncover the intricate temporal patterns related to crystalline order, even before any clear indications of the transition became evident in thermodynamic properties, such as specific volume.

Keywords: Polymer crystallisation; Machine learning; Nucleation patterns