Identification and quantification of individual volatile organic compounds in a binary mixture by SAW multisensor array and pattern recognition analysis

Abstract

We have developed a surface acoustic wave (SAW) multisensor array with five acoustic sensing elements configured as two-port resonator 433.92 MHz oscillators and a reference SAW element to recognize different individual components and determine their concentrations in a binary mixture of volatile organic compounds (VOCs) such as methanol and acetone, in the ranges 15-130 and 50-250 ppm, respectively. The SAW sensors have been specifically coated by various sensing thin films such as arachidic acid, carbowax, behenic acid, triethanolamine or acrylated polysiloxane, operating at room temperature. By using the relative frequency change as the output signal of the SAW multisensor array with an artificial neural network (ANN), a recognition system has been realized for the identification and quantification of tested VOCs. The features of the SAW multisensor array exposed to a binary component organic mixture of methanol and acetone have been extracted from the output signals of five SAW sensors by pattern recognition (PARC) techniques, such as principal component analysis (PCA). An organic vapour pattern classifier has been implemented by using a multilayer neural network with a backpropagation learning algorithm. The normalized responses of a reduced set of SAW sensors or selected principal components scores have been used as inputs for a feed-forward multilayer perceptron (MLP), resulting in a 70% correct recognition rate with the normalized responses of the four SAW sensors and in an enhanced 80% correct recognition rate with the first two principal components of the original data consisting of the normalized responses of the four SAW sensors. The prediction of the individual vapour concentrations has been tackled with PCA for features extraction and by using the first two principal components scores as inputs to a feed-forward MLP consisting of a gating network, which decides which of three specific subnets should be used to determine the output concentration: the first subnet for methanol only, the second subnet for acetone only and the third subnet for methanol and acetone in the binary mixture. Good 0.941 and 0.932 correlation coefficients for the predicted versus real concentrations of methanol and acetone, respectively, as individual components in a binary mixture have been obtained. The experimental results demonstrated that the proposed binary organic vapour mixture classifier is effective in the identification of the tested VOCs of methanol and acetone. Also, the combination of PCA and ANN techniques provides a rapid and accurate quantification method for the individual components' concentration in the tested binary mixture of methanol and acetone.