About this Abstract |
| Meeting |
MS&T25: Materials Science & Technology
|
| Symposium
|
Advances in Dielectric Materials and Electronic Devices
|
| Presentation Title |
Machine Learning-Assisted Identification of Crystal Symmetries in Ferroelectric Perovskites Using X-Ray Diffraction Patterns |
| Author(s) |
Luiz Fernando Cotica, Hugo N. Machado, Gustavo Sanguino Dias, Ivair Aparecido Santos, Ruyan Guo, Amar Bhalla |
| On-Site Speaker (Planned) |
Luiz Fernando Cotica |
| Abstract Scope |
X-ray diffraction (XRD) is crucial for characterizing ferroelectric materials with perovskite structures, providing insights into crystallographic phases and structural distortions. Accurate interpretation of XRD patterns is essential for identifying symmetry changes and atomic displacements that influence ferroelectric behavior. Concurrently, machine learning (ML) techniques are increasingly employed in materials science for efficient data analysis and property prediction. This study integrates XRD analysis with ML algorithms to classify the crystal symmetries of ferroelectric perovskite materials. We developed a database of 11,000 powder XRD patterns derived from crystallographic information files (CIFs) using a Python simulation tool. The diffraction peaks served as input features, while the crystal symmetries were the classification targets. We evaluated ten ML algorithms, achieving an accuracy of 70% for triclinic systems and 62% for monoclinic systems. Ongoing efforts will focus on enhancing accuracy and expanding the dataset to improve model robustness across additional symmetry classes. |