About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Algorithms Development in Materials Science and Engineering
|
| Presentation Title |
H-11: Data-Driven Visualization of Molybdenum Deposition in Hybrid Semiconductors Using Machine Learning |
| Author(s) |
Aline Hernandez-Garcia, Rafael Torres-Escobar, Osvaldo de Melo, Leon Hamui |
| On-Site Speaker (Planned) |
Aline Hernandez-Garcia |
| Abstract Scope |
This work presents a machine learning-based methodology to enhance the visualization of molybdenum (Mo) concentration in hybrid semiconductor films deposited on silicon substrates. SEM-EDS data were cleaned and analyzed with over 1,000 sampling points per region. The area under the curve (AUC) of Mo-specific spectral peaks was calculated, and background signals from the substrate were subtracted to reduce noise. Spatial interpolation was performed using a nearest-neighbor assignment technique, allowing for the reconstruction of concentration maps with improved clarity. The workflow combined Python for data processing and ImageJ for image segmentation and preprocessing. The resulting visualizations showed an improvement of up to 80% in the spatial resolution of Mo distribution, enabling a more accurate assessment of deposition quality. This approach supports the correlation between compositional uniformity and semiconductor performance, offering a valuable tool for materials optimization in data-intensive experimental workflows. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Computational Materials Science & Engineering, Machine Learning |