The measurement of organic vapors with surface-acoustic-wave microsensor arrays.

Abstract

Arrays of SAW sensors with different, partially-selective coatings exhibited response patterns that were analyzed to identify and quantify individual vapors and the components of simple vapor mixtures. In analyzing binary mixtures responses, successful identification rates of over 85 percent were obtained. Three semi-empirical physicochemical models of sensor responses were developed based on the vapor boiling point, vapor/coating solubility parameters, and vapor/coating solvation parameters in conjunction with linear solvation energy relationships (LSER). All three models provided useful information but the latter provided greatest accuracy with all modeled values falling within 25 percent of the corresponding experimental values. These models are useful in estimating the limit of detection and the degree of selectivity that an array provides for a given individual vapor. A limited investigation of humidity effects revealed that water vapor has no appreciable effect on sensor responses for most coating-vapor pairs. An examination of empirical limits of detection revealed a range of 1 $\mu$g/L (for the low-volatility polar vapor aniline on a polar cyano substituted coating) to 5620 $\mu$g/L (for the highly volatile non-polar vapor ethyl ether on the same polar cyano substituted coating). Liquid-crystal coatings were investigated to test if their anisotropic structure would aid in distinguishing vapors based on subtle differences in shape or size. For three of the four isomeric vapor pairs investigated, a significantly higher response was observed for the more rod-like isomer on the sensors with liquid-crystal coatings relative to sensors with isotropic coatings. In some cases the liquid crystal coatings yielded a sensitivity that was 25 percent greater than isotropic coatings for the rod-like isomer relative to the non-rod-like isomer. A method of choosing optimal coatings for inclusion in a four sensor array was developed that allows an assessment of the array performance. Application of this coating selection method in combination with LSER-modeled sensor responses showed that, in lieu of experimental calibration data, one can use modeled responses to determine an optimal coating set and estimate correct identification rates for a given set of vapors.