Lightpipe Radiation Thermometers Research Articles

Efficient Signal Transport Model for Remote Thermometry in Full-Scale Thermal Processing Systems

The rapid thermal processing of semiconductor devices is very temperature sensitive and requires precise temperature measurement. Light pipe radiation thermometers are widely used for temperature control during manufacturing by industry, and there is concern about errors associated with light pipe measurements. Modeling in simplified systems has helped in understanding the signal transport process in light pipes and errors associated with measurements in the past. Considering the small sensor area compared to the size of the semiconductor wafer and the remaining system components, modeling of the complete system has not been done due to the computational demand. A reverse Monte Carlo model can be used efficiently to model the signal transported to the photodetector in conjunction with a thermal model of the system to better characterize the system. The proposed method is demonstrated in a full-scale instrumented system with a light pipe thermometer, and the results are compared against previously published measurements from the system.

Monte Carlo Modeling of a Light-Pipe Radiation Thermometer

Light-pipe radiation thermometers (LPRTs) are widely used to monitor temperature during thermal processing of materials, particularly semiconductor wafer rapid thermal processing. According to the International Technology Roadmap for Semiconductors 2004, temperatures for semiconductor wafer processing should be measurable to within an uncertainty of plusmn1.5 degC at 1000 degC with temperature calibration traceable to International Temperature Standard-90. To achieve this accuracy level, the radiation signal transport process inside the light-pipe probe has to be fully understood. Few studies have been conducted to model the radiation transfer of LPRTs. In this paper, a Monte Carlo model has been created to simulate the signal transport from the measurement surface through the light-pipe probe. The model predicts the acceptance angle or aperture of the light-pipe. We also investigated the effect of nonspecular reflections caused by the sidewall surface roughness of the light-pipe probe using this model

Errors Associated With Light-Pipe Radiation Thermometer Temperature Measurements

Accurate measurement of surface temperature distribution is of concern in the semiconductor industries, particularly in rapid thermal processing. The International Technology Roadmap for Semiconductors 2004 has established requirements of uncertainties of plusmn1.5 degC at a temperature of 1000 degC, with temperature calibration traceable to ITS-90 (International Temperature Scale-1990). Light-pipe radiation thermometers (LPRTs) are becoming increasingly important as an industrial tool for temperature measurement, especially in the semiconductor industry. However, achieving improved accuracy will be difficult without fully understanding several issues associated with LPRTs. In this paper, the "drawdown effect," the "shadow effect," and the "environmental effect" are investigated. The drawdown effect is caused by the physical mass of the light-pipe probe acting as a radiation heat sink for the measured object. The shadow effect is caused by distortion of the wafer radiosity distribution due to the presence of the light-pipe probe. The environmental effect is caused by nonspecular reflection due to surface imperfection or impurity of the light-pipe material. Monte Carlo simulation was conducted to better understand the underlying physics, and the simulation results are compared to the experiment results. The latter two effects were found to be very important in the present study

A Summary of Lightpipe Radiation Thermometry Research at NIST.

During the last 10 years, research in light-pipe radiation thermometry has significantly reduced the uncertainties for temperature measurements in semiconductor processing. The National Institute of Standards and Technology (NIST) has improved the calibration of lightpipe radiation thermometers (LPRTs), the characterization procedures for LPRTs, the in situ calibration of LPRTs using thin-film thermocouple (TFTC) test wafers, and the application of model-based corrections to improve LPRT spectral radiance temperatures. Collaboration with industry on implementing techniques and ideas established at NIST has led to improvements in temperature measurements in semiconductor processing. LPRTs have been successfully calibrated at NIST for rapid thermal processing (RTP) applications using a sodium heat-pipe blackbody between 700 °C and 900 °C with an uncertainty of about 0.3 °C (k = 1) traceable to the International Temperature Scale of 1990. Employing appropriate effective emissivity models, LPRTs have been used to determine the wafer temperature in the NIST RTP Test Bed with an uncertainty of 3.5 °C. Using a TFTC wafer for calibration, the LPRT can measure the wafer temperature in the NIST RTP Test Bed with an uncertainty of 2.3 °C. Collaborations with industry in characterizing and calibrating LPRTs will be summarized, and future directions for LPRT research will be discussed.

Investigation of Lightpipe Volumetric Radiation Effects in RTP Thermometry

A major obstacle to the widespread implementation of rapid thermal processing (RTP) is the challenge of wafer temperature measurement. Frequently, lightpipe radiation thermometers are used to measure wafer temperatures in RTP reactors. While the lightpipe distorts the wafer temperature profile less than temperature measurement techniques which require physical contact, the presence of the lightpipe influences the wafer temperature profile. This paper presents the results of a theoretical study exploring that influence for an idealized RTP reactor in which the wafer is treated as a nonconducting, opaque, constant-heat-flux surface imaged by the lightpipe. The coupled radiation/conduction transport in the lightpipe measurement enclosure is solved numerically. Radiation transfer in the system is modeled with varying levels of rigor, ranging from a simple volumetrically nonparticipating treatment to a full spectral solution of the radiative transfer equation. The results reveal a rather significant effect of the lightpipe on the wafer temperature, which depends on the separation between the lightpipe tip and the wafer. The study illustrates clearly the need to model the lightpipe as a volumetrically participating, semitransparent medium, and further, the importance of accounting for spectral variation of the lightpipe properties in the prediction of the radiative transfer. Finally, two primary mechanisms are identified by which the lightpipe affects the wafer temperature distribution.