Presentation Information
[8a-N102-1]Real-Time Surface Temperature Distribution Measurement of Silicon Wafers Using Deep Learning-Assisted Optical Interferometric Contactless Thermometry (OICT)
〇Jiawen Yu1, Harumu Ono1, Hiroaki Hanafusa1, Seiichiro Higashi1 (1.Hiroshima Univ.)
Keywords:
deep learning,machine Learning,real-time temperature measurement
Real-time monitoring of wafer temperature distribution is essential for advanced semiconductor manufacturing processes. In this study, a deep-learning-based framework was developed to estimate the surface temperature distribution of a silicon wafer from a limited number of temperature measurements obtained by Optical-Interference Contactless Thermometry (OICT). Training data were generated using three-dimensional heat diffusion simulations under various heating conditions, and a structured Deep Neural Network (DNN) was trained to reconstruct the wafer temperature distribution from multi-point temperature data. The proposed model achieved an average prediction error of approximately 0.4 K and a maximum prediction error below 2.5 K, while requiring less than 0.1 s for a single prediction. These results demonstrate the potential of combining multi-point OICT measurements with deep learning for real-time and accurate monitoring of wafer temperature distributions in semiconductor manufacturing processes.