Spontaneous Wrinkle Collapse in Anisotropic Condensed Matter Predicted by Deep Learning.

Area of Science:

• Condensed Matter Physics
• Materials Science
• Computational Physics

Background:

• Nematic liquid crystals exhibit complex orientational order and topological defects.
• Predicting these configurations typically requires computationally intensive simulations.
• Controlling topological textures is crucial for advanced optical applications.

Purpose of the Study:

• To develop a fast and accurate deep learning framework for predicting nematic liquid crystal configurations.
• To validate the model's predictions against experimental observations.
• To establish a generalizable data-driven surrogate for nematic systems.

Main Methods:

• A 3D U-Net deep learning model was trained on data from a finite element Landau-de Gennes solver.
• Simulated director fields were compared with experimental data from photoaligned wrinkle substrates.
• Polarized optical microscopy (POM) images were used for validation.

Main Results:

• The 3D U-Net model accurately predicts global orientational order and local defect structures.
• Predictions are generated in milliseconds, significantly faster than conventional simulations.
• The model successfully reproduces complex defect behaviors, including collapse and splitting.
• Experimental validation confirmed the model's reliability and fidelity across diverse boundary conditions.

Conclusions:

• The deep learning framework provides a robust, data-driven surrogate for simulating nematic liquid crystals.
• This approach bridges computational theory and experimental validation.
• It offers a pathway for designing and controlling topological textures in materials for photonics and optics.