Compressive multi-beam scanning transmission electron microscopy

Area of Science:

• Materials Science
• Microscopy
• Image Processing

Background:

• Scanning transmission electron microscopy (STEM) is crucial for materials analysis.
• Conventional STEM imaging can be time-consuming, limiting throughput.
• Super-resolution techniques are needed to enhance detail and speed.

Purpose of the Study:

• To develop a faster STEM imaging method using multi-beam probes and compressive sensing.
• To achieve high-fidelity image reconstruction from down-sampled data.
• To explore potential acceleration of analytical scanning methods.

Main Methods:

• Utilized a custom condenser aperture to generate a six-beam STEM probe.
• Employed defocus to tune beam shape and distribution.
• Applied compressive sensing framework with Adam optimization and total variation normalization for image reconstruction.

Main Results:

• Successfully reconstructed high-fidelity STEM images from down-sampled data.
• Demonstrated that the reconstructed images closely reproduce original sample structures.
• Showcased the effectiveness of multi-beam sparse sampling and computational reconstruction.

Conclusions:

• The proposed multi-beam STEM approach significantly accelerates imaging acquisition.
• This method enables super-resolution reconstruction, preserving sample details.
• Offers a promising pathway for advancing analytical scanning microscopy techniques.