Silicon wafer surface temperature measurement using light-pipe radiation thermometers in rapid thermal processing systems

Light-pipe radiation thermometers (LPRTs) are widely used to monitor temperature during thermal processes, particularly semiconductor wafer Rapid Thermal Processing (RTP). According to the International Technology Roadmap for Semiconductors 2004 (ITRS), temperatures for semiconductor wafer processing should be measurable to within an uncertainty of ± 1.5 ºC at 1,000 ºC with temperature calibration traceable to ITS (international temperature standard)-90. To achieve this uncertainty, there are several issues associated with LPRTs to be resolved. There are three basic effects which could affect LPRT’s temperature measurement are studied in this paper. They are the “drawdown effect”, the “shadow effect” and the “environment effect”. viii The “draw-down effect” is the temperature depression on the target surface caused by the heat loss from the LPRT probe. It is examined both experimentally and numerically in the temperature range of 500 ºC to 900 ºC in this paper. The “shadow-effect” is the radiosity distortion on the target surface caused by the presence of the LPRT probe. In this paper the shadow-effect has been studied through experiments and computer simulations. A Monte Carlo model has been created to simulate the signal transport from the measurement surface (shadowed object) through the light-pipe. Both experiment and computer simulation results show that the distance from the target surface to the light-pipe tip plays an important role in LPRT’s temperature measurement. The relation between an effective emissivity for the system and the light-pipe-source separation distance has been investigated and compared to the experiment results. Significant differences between the apparent measured temperature and the actual object temperature are found if the shadow effect is not compensated for. The “environment effect” is the noise signals from surroundings other than target surface get into the LPRT and influences the temperature indicated by the LPRT sensor. In this paper, I concentrate on the non specular reflection caused by the roughness of the sidewall of the LPRT’s probe. A Monte Carlo computer simulation was conducted and compared to experiment study [Puttitwong, 2006] to quantitively investigate the effect of surface imperfections on the light-pipe probe on differences in calibration.