Application of modern analytical techniques and chemometric methods to the chemical characterisation of South African wines : determination of non-volatiles

[摘要] ENGLISH ABSTRACT:The present study deals in the first instance with the improvement of current analyticaltechniques for the analysis of the non-volatile content of wines. An improved samplepreparation method, using solid phase extraction (SPE), was initially developed for theanalysis of organic acids, sugars and phenolic compounds. Consequently, modemanalytical methodologies were assessed to obtain optimal techniques for the separation ofvarious non-volatile compounds. A capillary electrophoresis (CE) method, demonstrablymore reliable than currently used high performance liquid chromatography (HPLC) and CEmethods, is proposed for the analysis of organic acids. HPLC with refractive index (RI) orevaporative light scattering detection (ELSD) proved more suitable than CE for the analysisof sugars in dry wines. Liquid-chromatography-mass spectroscopy (LC-MS) offeredsuperior sensitivity and resolution compared to the relatively new technique of CE-MS forthe analysis of wine phenolics. LC-MS was further applied for the efficient and sensitiveanalysis of non-coloured phenolics and anthocyanins in wine. Negative- and positiveelectrospray ionisation, respectively, were used in conjunction with an ion-trap massanalyzer, for the identification of 34 phenolics and 31 anthocyanins in red wine samples.Complementary CE and LC methods were developed to allow the identification of artificialdyes in red wines, added illegally to improve their colour. Also, the application of stir barsorptive extraction (SBSE) with liquid desorption and micellar electrokineticchromatography (MEKC) for the analysis of bitter acids in beer is reported.In the second part of the thesis, the analytical results obtained for South African red andwhite wines were evaluated. Following comparison of the results with literature reports,several pattern recognition techniques were employed. A classification function obtained bylinear discriminant analysis (LDA) was used to classify both red and white wines accordingto variety, based on their chemical composition. This classification is achieved independentof the factors of wine age or geographical origin, making it useful for authenticityevaluation.

[发布机构] Stellenbosch University