Method for retrieval of the three-dimensional object potential by inversion of dynamical electron scattering

Area of Science:

• Electron microscopy
• Computational imaging
• Materials science

Background:

• Dynamical electron scattering is crucial for high-resolution imaging.
• Current methods often simplify complex physical interactions.
• Accurate 3D object potential retrieval is essential for nanoscale characterization.

Purpose of the Study:

• To develop a novel, nonheuristic method for inverting dynamical electron scattering.
• To integrate advanced physical effects into the imaging reconstruction process.
• To enable accurate 3D object potential retrieval using artificial neural networks.

Main Methods:

• Recasting the multislice algorithm as an artificial neural network (ANN).
• Iterative retrieval of the 3D object potential.
• Incorporating CCD modulation transfer function, coherence effects, and inelastic scattering.
• Utilizing prior knowledge of atomic potential shape, sparseness, and positivity.

Main Results:

• Successful demonstration on simulated bright-field images.
• Accurate reconstruction of 3D object potentials.
• Nonheuristic treatment of complex scattering phenomena.

Conclusions:

• ANN-based inversion of dynamical scattering offers a powerful approach for 3D imaging.
• This method enhances the accuracy and scope of electron microscopy.
• It paves the way for more detailed nanoscale material analysis.