OES-based sensing for plasma processing in IC manufacturing

Commercial semiconductor technology is expected to reach the 0.18;In this work we are developing a methodology for estimating etching performance using real-time sensors. Specifically, the sensor readings collected throughout plasma etching processes are used in combination with statistical techniques to model etch rate, within-wafer uniformity, aspect-ratio dependent etching (ARDE), and critical dimension (CD) reduction. Most of this effort has been devoted to developing reliable models that relate process performances to signals acquired from optical emission spectroscopy (OES), a sensor providing spatially resolved, in-situ, real-time readings without disturbing the plasma or interfering with the process. Our study is based on the OES sensor systems installed on various commercial plasma etchers including dielectric etchers (AMAT 5300 Centura), polysilicon etchers (Lam 4400 and 9400), and metal etchers (Lam 9600).;Another aspect of this work is to model the spatial variation of critical dimension (CD) using the spatial and temporal information provided by the OES sensor. Both physical and statistical approaches are employed to extract the spatial and temporal variation from the original OES sensor readings for each wafer at selected wavelengths. An additional goal of this work is to demonstrate a novel technique for monitoring the spatial CD variation and detecting abnormal wafers using OES signals. This is accomplished by a statistic describing the temporal variation from the real-time OES signals collected from three spatially resolved beams, and can be employed to indicate the nature of the CD spatial non-uniformity.;Conventional endpoint detection techniques often provide temporally-resolved information of the etch process, but without spatial resolution. With the emerging of the 300mm wafer production, the potential utility of a spatially-resolved endpoint detector is expected to increase greatly. In this project we propose a novel spatially resolved sensor to detect the endpoint and monitor spatial uniformity for a plasma etching process. A scanning spatially-resolved optical emission spectroscopy (SROES) system was built and installed in the Berkeley Microfabrication Laboratory. This sensor system consists of a stepper motor, controller, and a monochrometer, which provides an in-situ real-time monitoring of the etching endpoint spatially. The most interesting feature of collected plasma spatial images is their temporal dependence. This information can be then used to detect the spatially-resolved endpoints and monitor the etch uniformity. This concept also extends to a scanning SROES system with a full spectral-range spectrometer. The results suggest that full color plasma emission spatial profiles are useful for characterizing and monitoring the plasma conditions. It is our expectation that these signals will play a useful role in designing the next generation equipment controllers.