Sensor signal processing for gravimetric chemical sensors based on a state-space model

One of the major drawbacks of gravimetric (mass-sensitive) chemical sensors with selective coatings is the slow dynamic response due to diffusion processes during analyte molecule absorption. In general, one must put up with a trade-off between sensitivity and response time because both quantities increase with the coating thickness. This issue can be resolved by decoupling the evaluation time from the dynamic behavior of the sensor. To this end, we developed a linear state-space model based on the generally accepted physical model of diffusion and absorption. Then we applied signal processing strategies such as Kalman filtering and Wiener deconvolution to obtain stochastically optimal estimates for the sensor signal features of interest. By this approach, which is novel for the class of sensors under discussion to the best of our knowledge, a change in the ambient analyte gas concentration can be determined quantitatively well before the sensor has reached the steady-state. We discuss the features of this approach in detail and demonstrate that the measurement time is almost independent of system time constants. Hence, one can use sensitive (slow) sensors and still guarantee fast effective response times. Finally, we also present a straightforward extension to the even more general nonlinear problem.