Adaptation and Evaluation of a Personal Electronic Nose for Selective Multivapor Analysis

Abstract

The evaluation of a commercial, belt-mountable “electronic nose” modified for the rapid recognition and quantification of individual solvent vapors and simple vapor mixtures at low ppm concentrations is described. Marketed under the name VaporLab® this direct-reading instrument was designed for qualitative determinations of the presence or absence of selected individual vapors and was adapted in this study for quantitative determinations of vapors and vapor-mixture components. Vapor samples are concentrated on a small adsorbent bed and then thermally desorbed for analysis by an array of four polymer-coated surface acoustic wave sensors. Tests were performed with 13 organic solvent vapors individually and in selected binary, ternary, and quaternary mixtures at concentrations ranging from 0.1 to 12 times the respective American Conference of Governmental Industrial Hygienists' (ACGIH®) threshold limit value (TLV®). Pattern recognition analyses yielded a library of response patterns to which subsequent actual and virtual (i.e., Monte-Carlo simulated) samples were compared to assess performance. Limits of detection < 0.025 × TLV are achieved (based on the most sensitive sensors) for 0.25 L of preconcentrated air samples collected over a 2-min period. Individual vapors from different functional group classes can be recognized, quantified, and discriminated from other vapors with little error, and discrimination of the components of binary mixtures is possible where the component vapor response patterns are sufficiently different. Within-class individual vapor and binary-mixture discriminations are more difficult and most ternary and higher-order mixtures could not be analyzed with acceptable accuracy. Changes in ambient humidity have no effect on responses and changes in temperature lead to well-behaved and compensable changes in responses. Tests of fluctuating concentrations demonstrate the capability for accurately tracking short-term variations in exposure. Overall, results suggest that this instrument could serve effectively as a personal exposure monitor in previously characterized occupational environments with proper revisions in design.

Keywords:

• gas detector
• electronic nose
• personal monitor
• sensor array
• surface acoustic wave sensor
• vapor mixture

ACKNOWLEDGMENTS

The authors wish to acknowledge Hank Wohltjen and Brent Busey of Microsensor Systems, Inc., for providing the instrumentation tested in this study and for valuable technical support; Kevin Vetelino of Sawtek, Inc., for coating the sensors in the array; and Jeongim Park and William A. Groves for revisions of the EDPCR routines and helpful comments on the manuscript.

Funding for this research was provided in part by a National Institute for Occupational Safety and Health (NIOSH) Education and Research Center Pilot Research Grant and also by NIOSH Grant R01-OH03692.

Notes

^A Sensitivity values are in Hz/ppm.

^B LOD values are in ppm.

^A See text for explanation of the boundary conditions placed on each analysis.

^B Average of the absolute quantification error values for tests with correct recognition outome.

^C Rates obtained with the modeled error/noise terms for the OV-275-coated sensor reduced by half.

^A Rates obtained with the modeled error/noise terms for the OV-275-coated sensor reduced by half.