Deep Learning Analysis of Localized Interlayer Stacking Displacement and Dynamics in Bilayer Phosphorene

Area of Science:

• Materials Science
• Condensed Matter Physics
• Nanotechnology

Background:

• Interlayer displacement is key for tuning properties in layered crystals.
• Transmission electron microscopy (TEM) analyzes crystal structures but struggles with precise interlayer displacement measurement.
• Conventional TEM analysis limits recognition of varying unit-cell patterns and structure factors.

Purpose of the Study:

• To introduce a deep learning-based analysis for atomic resolution TEM images.
• To enable accurate unit-cell pattern recognition and precise interlayer displacement determination in bilayer phosphorene.
• To demonstrate the model's capability in analyzing large-scale in situ TEM data.

Main Methods:

• Development of a deep learning model for analyzing atomic resolution TEM images.
• Application of the model to identify unit-cell patterns and stacking displacement in bilayer phosphorene.
• Processing of in situ TEM data to capture dynamic interlayer displacement.

Main Results:

• The deep learning model accurately determines stacking displacement in bilayer phosphorene with 3.3% error.
• Achieved spatial resolution approaches the individual unit-cell level.
• Successfully captured time-dependent interlayer displacement dynamics during edge reconstruction in large in situ TEM datasets.

Conclusions:

• Deep learning provides a powerful approach for precise interlayer displacement analysis in TEM.
• The developed model offers high accuracy and resolution for studying layered materials.
• This method has significant potential for analyzing complex, dynamic phenomena in large microscopy datasets.