Comparison of Leak Sensors for Detection Sensitivity in a Vacuum Chamber: Self-Plasma Optical Emission Spectroscopy and Residual Gas Analyzer

Since leak in the semiconductor fabrication process is one of the potential threats causing device degradation, it is significant to evaluate the leak detection performance of current leak sensors for advancing leak detection technology. In this study, we evaluated the leak detection performance of widely used leak sensors: self-plasma optical emission spectroscopy (SPOES) and residual gas analyzer (RGA). Specifically, we examined two SPOES systems: 1) SPOES-i, which employs an inductively coupled plasma source, and 2) SPOES-c, which employs a capacitively coupled plasma source. Leak is injected into a vacuum chamber with an Ar background gas, using a mass flow controller (MFC) with concentrations ranging from 2.0 to 20 ppm. Before evaluation, we found the optimal operating conditions for each sensor in detail. We then evaluated their leak detection performance, limit of detection (LoD), and sensitivity at chamber pressures of 6.7 Pa (50 mTorr), 13.3 Pa (100 mTorr), and 26.7 Pa (200 mTorr). In terms of the LoD, the experimental results demonstrated that, at 6.7 Pa, the SPOES-c detects the leak concentrations as low as 2.0 parts per million (ppm), while the SPOES-i and the RGA detected between 2.4 and 6.0 ppm. At higher pressures (>6.7 Pa), all three sensors detected at least 2.0 ppm. This enhancement of the detection limit results from the increase in the absolute leak amount with higher chamber pressure at the same concentration. In terms of sensitivity, the SPOES-i exhibited the highest value, followed by SPOES-c, and then the RGA. It might result from the fact that the sensitivity is proportional to the density of electrons of each sensor, the higher the electron density, the larger the sensitivity. Detailed discussions on LoD and sensitivity are provided. These findings will contribute to the future development of in situ leak detection for sub-ppm level leak detection.