Understanding Local Crystallography in Solar Cell Absorbers with Scanning Electron Diffraction.

Area of Science:

• Materials Science
• Nanotechnology
• Crystallography

Background:

• Controlling grain structure and local crystallography is crucial for high power conversion efficiency in thin film photovoltaic devices.
• Structural defects, grain boundaries, and unwanted phases in photovoltaic materials can arise from compositional inhomogeneities or synthesis parameters.
• Studying crystallographic properties in complex photovoltaic systems with multiple phases or numerous grains is challenging.

Purpose of the Study:

• To demonstrate the application of 4D Scanning Transmission Electron Microscopy (4D-STEM) for nanoscale characterization of photovoltaic materials.
• To showcase the use of unsupervised machine learning for analyzing large 4D-STEM datasets.
• To provide a framework for understanding and engineering crystallographic properties in complex photovoltaic systems.

Main Methods:

• Utilized 4D Scanning Transmission Electron Microscopy (4D-STEM) on cross-sections of Cu(In,Ga)S2, halide perovskite, and Sb2Se3 thin films.
• Applied unsupervised machine learning techniques, including dimensionality reduction and hierarchical clustering, to analyze 4D-STEM data.
• Developed an analytical framework adhering to FAIR principles, using open-source software for data sharing.

Main Results:

• 4D-STEM successfully unraveled nanoscale crystallographic and microstructural properties of diverse photovoltaic materials.
• Machine learning algorithms effectively extracted key information from complex and large 4D-STEM datasets.
• The study demonstrated the capability to analyze materials with multiple phases, complex stoichiometry, electron beam sensitivity, and high grain densities.

Conclusions:

• 4D-STEM is a powerful technique for comprehensive crystallographic analysis in challenging photovoltaic materials.
• Unsupervised machine learning is essential for extracting actionable insights from large 4D-STEM datasets.
• The developed analytical framework supports data sharing and reproducible research in photovoltaic materials science.