Homepage

mmc2021 Abstract Database

Three-dimensional electron crystallography (3DED) and the particle-crystal transition and implications for structure solution

Three-dimensional electron crystallography (3DED) and the particle-crystal transition and implications for structure solution

Session

Stream 2: EMAG - Automated Control, Advanced Data Processing

Authors

Gearóid Mangan (1), David Landers (1), Eoin Walsh (1), Andrew Stewart (1)

Affiliations

1. University of Limerick

Keywords

3DED, structure solution, particle-crystal transition, simulation, machine learning, low signal to noise, electron fluence

Abstract text

In the last decade, 3D electron diffraction (3DED)[1] has emerged as a new and highly successful method for ab initio structure solution of Nanosized crystals. With the resolving power of modern Transmission Electron Microscopes (TEM), it is possible to image a single atom. The ability to resolve atomic resolution detail in an image poses a question of electron crystallography which would not be a consideration for X-ray or Neutron crystallography. How many repeating unit cells does it take to be considered a crystal? When is a collection of atoms a particle, and when is it a crystal? Where is the boundary between these two states of matter, is the boundary between them sharp or ill-defined? The International Union of Crystallography definition of a crystal states, “A material is a crystal if it has essentially a sharp diffraction pattern.” [2] We will present a framework to assist in refining the definition of a crystal in the context of a 3DED experiment, determining the minimum requirements to define the particle-crystal transition and propose where the experimental limits of a crystal and, by extension, electron crystallography. 

The approach taken has been to build crystal models with increasing numbers of unit cells and calculate their diffraction patterns via multislice software.  Analysis of the calculated diffraction patterns provides insight into where the particle-crystal transition occurs and the nature of the transition. 

Experimentally observing such tiny crystals requires careful consideration of the electron beam fluence, as too high an electron beam flux could melt the crystal under observation. Most multislice software does not provide the capability to control the electron budget within a simulation; therefore, we have developed this capability via a Monte Carlo approach taking into account the detector characteristics and electron budget of the simulated image.  Enabling the simulation of low electron beam current images for a more realistic understanding of the experimental data will help the experimentalist understand signal to noise under limited electron budget experiments, which is a critical factor for beam sensitive materials. 

Operating in low beam fluence conditions can be a challenging problem. Therefore, we have developed machine learning algorithms to detect objects within the field of view for low signal to noise and noisy data, which will also be presented. 

References

1.  Gemmi, M., et al., 3D Electron Diffraction: The Nanocrystallography Revolution. ACS Central Science, 2019. 5(8): p. 1315-1329.

2.  Report of the Executive Committee for 1991. Acta Crystallographica Section A, 1992. 48(6): p. 922-946.