Date of Award

Summer 2007

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)


Electrical and Computer Engineering

First Advisor

Josse, Fabien

Second Advisor

Schneider, Susan

Third Advisor

Yaz, Edwin


It has been established that a polymer-coated SH-SAW sensor on 36° rotated Y-cut LiTa03 is a very sensitive platform for direct liquid-phase chemical sensing. In this work, two partially selective coatings, poly (epichlorohydrin) (PECH) and polyurethane (PU) are investigated for the detection of organophosphate pesticides (phosmet and parathion) in aqueous solutions. Organophosphates (OPs) are chemical compounds that are produced by reacting alcohols with phosphoric acid and are widely used in agriculture, homes, and veterinary medicine to control pests, as well as in chemical warfare. The results show that PECH and PU-coated SH-SAW devices can be used to detect phosmet and parathion with high sensitivity. For the same coating thickness, the sensor responses for phosmet and parathion varied greatly even though they have similar functional groups and molecular properties (mass and size). In particular, the response time for parathion is almost five times as long as phosmet in spite of the fact that parathion is only 3.86% smaller than phosmet; the faster sorption process for phosmet may be due to the double benzene ring in its chemical structure. Moreover, the higher sensitivity of the films towards parathion suggests that parathion induces stronger interaction with the polymers, leading to greater viscoelastic contribution to the frequency response. For the present non-optimized experiments, a detection limit in the ppb range can be achieved. In order to meet the need for sensitive, selective, and rapid chemical detection systems, novel signal processing techniques are employed for on-line analysis of the sensor signal during the detection process. To achieve the above, the sensor response is first modeled in terms of all relevant contributions (mass loading and viscoelastic change) using a state-space approach. Transient information, often unique to a given analyte/coating pair or to a class of analytes, is then extracted using estimation theory. This additional sensor information is then utilized to improve analyte recognition and to explain observed sensor responses. Furthermore, a derivative technique, in the form of Savitzky-Golay algorithm is used in conjunction with an extended Kalman filter to predict detection well before steady-state is reached. As a result, detection decision can be made over a relatively shorter time, thus partially satisfying one of the design requirements for rapid chemical sensors.



Restricted Access Item

Having trouble?