Multi-channel piezoelectric crystal gas sensor with principal component analysis for organic solvent pollutants from polymer plants

A multi-channel piezoelectric (PZ) quartz crystal gas sensor coated with various organic adsorbents and a home-made microcomputer interface for data processing have been fabricated for the detection of various organic vapors such as methanol, formic acid, propionaldehyde, carbon disulfide and N,N-dimethyl formamide which are generally used as solvents in the polymer industry. The home-made computer interface includes a digital Altera and an Intel-8255 data processing system. The Principal Component Analysis (PCA) method was employed to analyze the frequency response signals from each channel with different coating materials and was applied to select the appropriate coating materials for these organic vapors. A dataset for the multi-channel sensor with 29 PZ quartz crystals for these organic vapors was collected. Six representative coating materials such as C60-polyphenyl acetylene (C60-PPA), polyethylene glycol, Nafion, triphenyl phosphine, cryptand-22 and polyvinyl pyrrolidone were selected after comparing the correlation between the 29 coating materials and the first four principal component (PC) factors of the PCA analysis. The 5 organic vapors can be effectively distinguished clearly by the six-channel piezoelectric crystal sensor and which can be obviously found from the PCA Scores Map. Furthermore, different organic vapors tend to have distinguishing profile discrimination maps with the six channel PZ crystal sensor. The profile discrimination maps can be applied as fingerprints for these organic molecules. The six-channel PZ crystal sensor has quite good detection limits for these organic vapors. Furthermore, the multivariate linear regression (MLR) analysis was also employed to compute the concentrations of methanol and carbon disulfide in the mixture of organic vapors.