Lattice Multislice Algorithm for Fast Simulation of Scanning Transmission Electron Microscopy Images.

Area of Science:

• Materials Science
• Computational Physics
• Microscopy

Background:

• Numerical simulations are crucial for interpreting Scanning Transmission Electron Microscopy (STEM) images.
• Existing methods like PRISM offer speedups but have limitations in memory consumption and numerical accuracy.

Purpose of the Study:

• To introduce a novel numerical approach for simulating STEM images.
• To improve computational efficiency and reduce errors in STEM image simulations.

Main Methods:

• Developed the Lattice Multislice Algorithm, approximating electron wave propagation in a localized real-space basis.
• Utilized the linearity of the Schrödinger equation, similar to the PRISM algorithm.
• Focused on functions localized in real space rather than Fourier space for approximation.

Main Results:

• Achieved computational speedup comparable to the PRISM algorithm.
• Demonstrated significantly lower memory consumption compared to PRISM.
• Reduced numerical errors by avoiding interference from virtual probe wave copies.
• Enabled faster recomputations for localized specimen changes.

Conclusions:

• The Lattice Multislice Algorithm offers an efficient and accurate alternative for STEM image simulations.
• This method presents advantages in memory usage, numerical stability, and adaptability for localized changes.
• The approach holds promise for advancing materials characterization through improved computational modeling.