Obtaining Chemical Selectivity from a Single, Nonselective Sensing Film: Two-Stage Adaptive Estimation Scheme with...

A new approach is reported to detect and quantify the members of a group of small-aromatic-molecule target analytes: benzene, toluene, ethylbenzene, and xylenes (BTEX), dissolved in water, in the presence of interferents, using only the data collected from a single polymer-coated SH-SAW (shear horizontal surface acoustic wave) device and a two-stage adaptive estimation scheme. This technique is composed of exponentially weighted recursive least-squares estimation (EW-RLSE) and a bank of Kalman filters (BKFs) and does not require any prior knowledge of the initial concentration range of the target analytes. The proposed approach utilizes the transient sensor response to sorption and/or desorption of the analytes as well as the error range associated with the response time constants to provide more information about the analyte-specific interactions with the polymer film. The approach assumes that the sensor response to contaminated groundwater is a linear combination of the responses to the single target analytes, the interferents that interact with the selected polymer sensor coatings, and measurement noise. The proposed technique was tested using actual sensor responses to contaminated groundwater samples containing multiple BTEX compounds with concentrations ranging from 10 to 2000 parts per billion, as well as common interferents including ethanol, 1,2,4-trimethylbenzene, naphthalene, n-heptane, and MTBE (methyl tert-butyl ether). Estimated concentration values, accurate to ±10% for benzene/toluene and ±15% for ethylbenzene/xylenes, are obtained in near-real time. The utilization of sorption and/or desorption data enables detection and quantification of BTEX compounds with improved accuracy, high tolerance to measurement noise, and improved chemical selectivity.