SK hynix and Gauss Labs presents 2 papers on AI semiconductor metrology at SPIE AL 2024
Date: 09/03/2024
SK hynix and Gauss Labs have recently participated in the SPIE AL1 2024 conference, a prestigious international event focusing on advanced lithography and patterning technologies. At the conference, they presented two papers highlighting cutting-edge advancements in AI-based metrology, a critical aspect of semiconductor manufacturing.
Figure: Gauss Labs CEO Mike Kim(center) poses with his colleagues
Enhanced Prediction Accuracy: Gauss Labs introduced the "aggregated AOM3" algorithm, enhancing the prediction accuracy of its AI-based virtual metrology solution, Panoptes VM. This algorithm aggregates data from processing machines and chambers with similar patterns, addressing data scarcity issues and boosting prediction accuracy.
Improved Process Variability: SK hynix utilized Panoptes VM for virtual measurements on over 50 million wafers, achieving a significant 29% improvement in process variability. This technology enables rapid virtual measurements, contributing to enhanced semiconductor yield and productivity.
Universal Denoiser for Image Enhancement: Gauss Labs presented a "universal denoiser" algorithm designed to remove random variations (noise) from Critical Dimension Scanning Electron Microscope (CD-SEM) images. By employing AI, this denoiser improves measurement accuracy and reduces image acquisition time by up to ¼, thereby enhancing metrology equipment productivity by 42%.
These innovations address key challenges in semiconductor manufacturing, including data scarcity, process variability, and image noise.
Mike Kim, CEO of Gauss Labs, said that his company is working on research and development for applications of industrial AI in real-world semiconductor manufacturing fabs. “We will continue to launch innovative AI-based solutions to revolutionize manufacturing.”
