Multi-Modal Machine Learning for PVD Mo Thin Films

Developed a multi-modal machine learning model to predict the microstructure of molybdenum thin films based on physical vapor deposition (PVD) process parameters. Integrated data from X-ray diffractograms, electron microscopy, and resistivity measurements to learn a joint latent representation of film structure. Used neural networks to map process conditions to structural outcomes, enabling accelerated process optimization in semiconductor materials.

The approach can be then used for inverse modeling of materials