Eighth Annual SCR / ARPA CIM - IC Workshop

Over the last decade, universities have been undertaking more industrially oriented research in addition to their traditional role of fundamental science. This new opportunity offers a new range of exciting research problems and gives students an exposure to real-world issues, but it can require a change in the way that university researchers (and their industrial sponsors) approach problems. At Princeton, we have an SRC-funded research project on the measurement of the temperature of silicon wafers in non-equilibrium processing environments, such as in a rapid thermal processor or a lamp-heated cluster tool, that has illustrated several of these issues. The measurement of the wafer temperature is critical for control of virtually any manufacturing process, but existing techniques for temperature measurement are subject to several practical and fundamental limitations. As an alternative, we have investigated the potential for extracting the temperature of a silicon wafer through the in-situ measurement of its infrared optical absorption, which is a strong function of temperature. The apparatus has been implemented in a RTP reactor (Figure 1). Our early work showed that one could indeed measure this change in absorption and, hence, could measure temperature from ~500 to 800°C (Figure 2). The next phase of the work was to establish the ultimate ranges for practical application in terms of temperature, wavelength, resolution , etc. This was done in large part by developing a model of the temperature dependence of optical absorption in silicon. Figure 1. Schematic diagram for the measurement of silicon wafer temperature in an RTP-CVD reactor by the in-situ measurement of transmission at 1.3 and 1.5 µm. Because of the complexity of physical processes such as electron-phonon coupling, this was clearly a task well suited for university work. The resulting model was sufficient to reveal the ranges of application, but it also revealed the possible presence of new second-order physical effects. Rather than pursue further study of these Physical effects, as might Figure 2. Experimental and modeled have been done tradi-transmission vs. temperature (normalized tionally, a decision was to room temperature values) at 1.3 and made to move the project 1.5 µm vs. temperature. forward. By discussing the implications of the model with industrial colleagues, it was established that metal-coated wafers were the most promising application. The next tasks involved the development of a double-pass version of the technique for this application. While these results demonstrated the feasibility of the approach, they also showed what …