Modeling the Transient Response of Microcantilever Sensors and Analyte Classification Using Estimation Theory

State-space models for the response of polymer-coated microcantilever chemical sensors operating in the static mode are developed. The models are developed from the physics governing the transduction mechanisms from chemical presence to microcantilever bending. The models incorporate well-known effects of the coating including plasticization and stress relaxation. Furthermore, each model contains parameters that are dependent on the analyte/coating pair. Estimation theory, in the form of the extended Kalman filter or a gradient descent technique, is used to extract the transient information I from the sensor response signatures in the form of system parameters. Once extracted, it is shown that this information can be used to improve the analyte identification process, as the transieJ?t response information is specific to a given analyte/coating pair. The results indicate that with this added information the selectivity of a given sensor array can be drastically improved. Also, the sensor array response, when processed using this technique, is less affected by noise due to aging and non-repeatability. Finally, analyte identification and classification can be performed well before steady-state (equilibrium) is reached, thus increasing the speed of detection without loss of accuracy.