This Small Business Innovation Research Phase II project will develop a multiple-applications, low-cost, real-time process monitoring and control tool for micro-electro-mechanical system (MEMS) deep-etch fabrication. Deep-etch processes are used to manufacture high aspect ratio structures up to several hundred microns thick, and promise to deliver new devices with increased performance and functionality at lower cost. A major difficulty in deep-etch technology is the control of the etch depth, which is currently measured post-etch using ex-situ destructive scanning electron microscopy. This is extremely inefficient, and is a major hurdle to be surmounted before extensive production takes place. During Phase I, an FTIR-based sensor was designed, constructed and installed on top of an etcher chamber. Etch depth and photoresist thickness measurements were obtained, for the first time ever, in-situ and in real-time on several MEMS structures. An excellent correlation between the FTIR measurements and SEM measurements was found. During Phase II, analysis models will be developed and implemented to measure the widest possible range of MEMS structures. These models will extract multiple parameters on any type of patterns, and will allow the use of the sensor for various applications, including deep trenches in silicon or SOI (silicon on insulator) wafers, membranes, thick photoresist, and mainstream silicon applications such as DRAM (Dynamic Random Access Memory) trenches. Hardware will be optimized for spot size, measurement spot range, compactness and, very importantly for the cost-sensitive MEMS industry, for cost. The result of this project will be the development of a metrology tool with capabilities currently unavailable, and which are in high and increasing need. The specific anticipated results of the use of the proposed metrology are: (1) to reduce cost through the reduction of destructive measurements and the improvement in process control,(2) to increase the reproducibility of the MEMS structures through better process control (run to run accuracy is currently ~3 % and is expected to be lowered by the use of the sensor to <0.5 %), (3) to provide useful feed-back for process development, thus reducing development time. These results will have a great impact on the deep-etch MEMS market, as they willhelp future MEMS applications to mature and come to market at a faster pace through cheaper characterization and improved process control. In addition, this first-of-a-kind real-time wafer-state monitoring and control technology will lead to applications within mainstream semiconductor processing such as DRAM.
