Quantification of binary mixtures of the freones R22 and R134a by surface plasmon resonance

For the recycling process of freones a high purity of the analytes has to be guaranteed. A fast and inexpensive online monitoring is desirable. In this study the quantification of binary mixtures of the freones R22 and R134a by surface plasmon resonance (SPR) was studied using a self-built Kretschmann set-up. A microporous polycarbonate was used as sensitive layer. Due to a size dependent sorption and desorption kinetics of the two freones a quantitative characterization of binary mixtures is possible by the use of only one sensitive layer. 441 different binary mixtures for calibration and 400 test mixtures were measured over time. The measured signals were subdivided into points of time and used as input variables for the data analysis by neural networks. An optimization of the number and type of these points of time resulted in a small relative RMSE of prediction of 1.68% for R22 and 2.33% for R134a for the test mixtures.