Quantitative trace analysis of benzene using an array of plasma-treated metal-decorated carbon nanotubes and fuzzy adaptive resonant theory techniques.

The functionalization of carbon nanotube sidewalls with metal nanoparticles is exploited here to improve the sensitivity and selectivity of gas sensors operated at room temperature. An array of sensors using oxygen plasma treated multiwalled carbon nanotubes (bare and decorated with Pt, Pd or Rh nanoparticles) is shown to selectively detect traces of benzene (i.e., 100 ppb) in the presence of carbon monoxide, hydrogen sulfide or nitrogen dioxide at different humidity levels. Employing a quantitative fuzzy adaptive resonant theory (ART) network whose inputs are the responses of the sensor array, it is possible to accurately estimate benzene concentration in a changing background. The quantitative fuzzy ART is especially suited for compensating the nonlinear effects in sensor response caused by changes in ambient humidity, which explains why this method clearly outperforms partial least squares calibration models at estimating benzene concentration. These results open the way to design new affordable, wearable, sensitive and selective detectors aimed at the personal protection of workers subject to occupational exposure to benzene, toluene, ethyl benzene and xylenes.