Simultaneous enzymatic spectrophotometric determination of ethanol and methanol by use of artificial neural networks for calibration

Binary mixtures of ethanol-methanol were resolved by use of an enzymatic sp ectrophotometric method using artificial neural network (ANN) methodology f or multivariate calibration. The chemical system involves two coupled react ions, viz the oxidation of the primary alcohols to the corresponding aldehy des in the presence of alcohol oxidase and the oxidation of p-phenylenediam ine to Bandrowski's base by hydrogen peroxide, catalysed by the previously formed aldehydes. The high complexity of the system studied entails the use of this non-linear calibration methodology, which provides significantly i mproved results relative to a multi-variate bilinear calibration, principal component regression (PCR), which was used for comparison. The optimized A NN allows the quantitation of both mixture components in ethanol to methano l mole ratios from 20 : 1 to 400 : 1, with relative standard errors of pred iction in the region of 5% for both analytes. (C) 1999 Elsevier Science B.V . All rights reserved.