Development of optimization procedures for application-specific chemical sensing

Abstract

We present methods for selecting optimized operating conditions from tunable sensor systems that produce multi-dimensional data streams. To demonstrate our approach in a case study, a chemical sensing dataset was collected using a microsensor array with temperature-modulated elements. The top electrodes of the array elements were coated with SnO2, In2O3, and CuO sensing films that were formed on the individual microhotplate platforms by annealing microcapillary deposited metal-hydroxide sol–gel films. Chemical sensing data was collected while interfacial interactions were influenced via pulsed temperature programming. During the collection of the chemical sensing database an analyte exposure schedule consisting of a dry air background and μmol/mol concentrations of three reducing analytes was cycled as the temperature program operating the microsensor array was systematically varied. The data was processed using linear discriminant analysis and principle component analysis to quantify analyte selectivity. The methods discussed in this manuscript produced sensing materials-specific temperature programs optimized for the studied analytes which contained a factor of 2.6 fewer temperature pulses than the original inspection temperature programs. These shortened temperature programs increased sample frequency, preserved analyte concentration information, and improved target cross-selectivity.