Reconstructing the exit wave of 2D materials in high-resolution transmission electron microscopy using machine...

Area of Science:

• Materials Science
• Computational Science
• Physics

Background:

• High-resolution transmission electron microscopy (HRTEM) requires accurate exit wave function reconstruction for image interpretation.
• Conventional methods for exit wave reconstruction can be complex and time-consuming.

Purpose of the Study:

• To develop and validate a novel method for exit wave reconstruction using convolutional neural networks (CNNs).
• To assess the fidelity of CNN-based reconstruction compared to traditional techniques.
• To demonstrate the applicability of this method across a wide range of 2D materials.

Main Methods:

• Utilized a fully convolutional neural network based on the U-Net architecture.
• Trained the network on simulated exit waves and HRTEM images of graphene-supported molybdenum disulphide (MoS2).
• Applied the trained network to experimentally obtained HRTEM images.

Main Results:

• Achieved exit wave reconstruction fidelity comparable to conventional methods.
• Successfully reconstructed atomically resolved structures of MoS2 nanoparticles and graphene support.
• Demonstrated successful training for reconstructing exit waves of 3400 different 2D materials from a database.

Conclusions:

• CNNs offer a powerful and efficient approach for exit wave reconstruction in HRTEM.
• This AI-driven method enables accurate structural analysis of various nanomaterials and 2D materials.
• The approach is scalable and applicable to a vast library of materials for advanced characterization.