Environmental Sensors and Subsystems

Print or save page as PDF
Multi-Transducer Arrays Using Nanoparticle Interface Layers for Vapor Discrimination
Lindsay K. Wright, Kee Scholten, and Edward T. Zellers
PCA plots for the (a) 4-sensor TSMR array, (b) 4-sensor CR array, and (c) the best performing sensor array: an MT array composed of 3 TSMR sensors and 1 CR sensor, for the 5 vapors tested. Data points are plotted for Monte-Carlo generated synthetic responses with ε =1% induced error, while ellipses represent the 95% confidence interval for data points distributed with ε =5% error.
PCA plots for the (a) 4-sensor TSMR array, (b) 4-sensor CR array, and (c) the best performing sensor array: an MT array composed of 3 TSMR sensors and 1 CR sensor, for the 5 vapors tested. Data points are plotted for Monte-Carlo generated synthetic responses with ε =1% induced error, while ellipses represent the 95% confidence interval for data points distributed with ε =5% error.

This project explores the development of multi-transducer (MT) arrays as detectors for microscale gas chromatographs (µGC). To date, most sensor arrays are composed of a set of single transducers (ST) coated with different sorptive interface layers to impart selectivity. However, such devices have a response diversity limited by the range of vapor-film interactions, which is often insufficient for accurately differentiating between vapors, especially in mixtures. MT arrays should offer greater diversity by virtue of probing multiple aspects of the vapor-interface interactions. We are studying combinations of two transducers: chemiresistors (CR), which respond to volumetric swelling and changes in the dielectric properties of the interface film, and thickness shear mode resonators (TSMR), which respond to changes in the mass of the interface. Thin films of thiolate-monolayer protected gold nanoparticles (MPN) with different thiolate functionalities have been selected as chemically selective interfaces for both transducers. In recent work we examined the responses of CRs and TSMRs coated with four different MPN films to a set of 5 vapors (toluene, TOL; n-propanol, POH; nitromethane, NME; 2-butanone, MEK; n-octane, OCT) using extended disjoint principal components regression (EDPCR) analysis. The results showed that an optimally chosen MT array of 3 or 4 sensors had a higher recognition rate than any ST array, when considering both single vapors and binary mixtures. This work was funded by the Department of Homeland Security, Science and Technology Directorate.

Updated 04/05/2012