ABSTRACT - McAvoy - CTS-9616085 At present most chemical plants are operated without the benefit of real time composition measurements. The reasons include cost, reliability, and robustness of composition sensors. Silicon based microsensors, hold promise for having a major impact on the operation of chemical processes. Each microsensor occupies approximately 400 square microns of surface area, so it is feasible to place several hundred sensors on a single, small chip. Tin dioxide (SnO2) gas microsensors are being developed at the National Institute of Standards and Technology (NIST), with whom the PI's are collaborating. These sensors have several unique features including their small size, approximately 200 micrometers in diameter, their fabrication by a conventional complementary metal oxide semiconductor (CMOS) foundry, and the ability to have their operating temperature changed rapidly, in a cyclic pattern, to optimize their sensitivity to a particular gas species. There are a number of problems that remain to be solved before these sensors can be used to solve process problems. Some of these problems include drift and noise effects, lack of sensitivity to some chemicals, and material problems. The SnO2 microsensors are nonspecific and they can respond to a number of chemical species. The research project involves further investigations of SnO2 microsensors. In using microsensors to broad problems need further study: (1) One can use the sensors to determine whether or not a chemical species is present, which is a problem in classification; and (2) one can use the sensors to determine which chemicals are present as well as their specific concentrations, which is a problem in quantitative analysis. The short range goal of the PI's research is to apply signal processing and dynamic modeling techniques to these microsensors to advance their use in the classification problem. Many applications of microsensors involve classification, e.g. hazardous leak, odor, and fire detection . The signal processing and modeling methods planned have the potential to be expanded to the quantitative analysis problem---thus, the long range goal of the research is to help advance the use of microsensor arrays for solving the quantitative analysis problem.
