서강대학교 전자공학과

Professor Kang Suk-ju’s Research Team Publishes Paper in Prestigious International Journal IEEE

Transactions on Instrumentation and Measurement

▲ (From left) Professor Kang Suk-Ju, master’s students Hwang Ye-eun and Song Min-seo from the Department of Electronic Engineering

Professor Kang Suk-ju’s research team from the Department of Electronic Engineering (including master’s students Hwang Ye-eun and Song Min-seo) has published a paper in the prestigious international journal IEEE Transactions on Instrumentation and Measurement (TIM) after conducting joint research with the CSE team at the Semiconductor Research Institute of Samsung Electronics.

In the paper, titled “SO-Diffusion: Diffusion-based Depth Estimation from SEM Images and OCD Spectra,” the research team introduces a new model for predicting semiconductor structures using semiconductor images captured by a scanning electron microscope (SEM) and optical critical dimension (OCD) spectra. Notably, the team developed a CNN-based spectrum encoder (SEFO) to effectively preprocess the OCD spectra and applied it to a diffusion-based network with SEM images, improving the accuracy of semiconductor structure prediction. As a result, the proposed model using SEM images and OCD spectra significantly outperformed existing models in predicting semiconductor depth.

Hwang Ye-eun, a master’s student who participated in the research, said, “It is a great honor to have my paper published in IEEE TIM during my master’s course.” She added, “Thanks to the support from Samsung Electronics’ CSE team and the guidance of my professor, we were able to achieve excellent results. As semiconductor research is advancing rapidly, I hope this study serves as a foundation for various follow-up studies.”

The proposed algorithm was developed in response to the industry’s recent trend of increasing miniaturization and complexity of semiconductor structures. By providing more precise measurement data when analyzing SEM semiconductor images, this research is expected to contribute to the effective reconstruction of 3D semiconductor structures.

▲ Overall structure of the SO-Diffusion network proposed in the paper

▶ Paper Title: SO-Diffusion: Diffusion-based Depth Estimation from SEM Images and OCD Spectra

▶ Journal: IEEE Transactions on Instrumentation and Measurement

▶ Authors: Hwang Ye-eun (First author, Sogang University), Song Min-seo (Sogang University), Ma Ah-mi (Samsung Electronics CSE), Kim Gyu-hwan (Samsung Electronics CSE), Jang Gyu-baek (Samsung Electronics CSE), Jung Jae-hoon (Samsung Electronics CSE), and Kang Suk-ju (Corresponding author, Sogang University)