Multicrystalline Informatics Applied to Multicrystalline Silicon for Unraveling the Microscopic Root Cause of...

Area of Science:

• Materials Science
• Crystallography
• Computational Materials Science

Background:

• Multicrystalline silicon (MCS) wafers are crucial in solar cells, but their performance is limited by defects like dislocation clusters.
• Understanding the formation mechanisms of these defects is essential for improving material quality and device efficiency.

Purpose of the Study:

• To investigate the formation mechanism of dislocation clusters in multicrystalline silicon using advanced computational and data science methods.
• To develop a realistic 3D model of dislocation cluster generation and analyze stress distributions within the material.

Main Methods:

• Utilized machine learning for image analysis (dislocation cluster extraction, grain segmentation, crystal orientation prediction) on silicon wafers.
• Performed finite element stress analysis coupled with crystal growth simulations on a 3D model.
• Conducted multiscale analysis of grain boundaries (GBs) and ab initio calculations.

Main Results:

• A realistic 3D model accurately predicted dislocation cluster generation points.
• Complex stress distributions were identified, with clusters forming along high-shear-stress slip planes and bending grain boundaries.
• A novel dislocation generation mechanism involving nanofacet formation at GBs was discovered, not explained by existing models.

Conclusions:

• The study reveals a unique dislocation generation mechanism in multicrystalline silicon driven by shear stress concentration and GB bending.
• This research introduces multicrystalline informatics, integrating diverse data science approaches for multi-scale analysis.
• The findings advance materials science by elucidating complex phenomena in multicrystalline materials.