Analysis of N-dimensional data arrays from fluorescence spectroscopy of an intermediary sugar product

Abstract

Unwanted formation of colour takes place during the production of crystalline sugar. The degree of colouration depends partly on the necessary processing conditions, e.g. heating and pH, and partly on the initial composition and condition of the sugar beets used as raw material. Reducing sugars are formed during the process. These are reactive compounds forming a variety of coloured complexes and strong precursors to further formation of colour and many of these compounds contain fluorophores. In the present work it is discussed if spectrofluorometric screening of intermediary sugar products prior to the final heating stages combined with a multi-way chemometric approach can provide information that significantly reflects the condition of the process and the beets. The model used is the N-way PCA (Principal Component Analysis) which is an exploratory model, not necessitating explicit modelling of single parameters nor any assumptions towards parameter interaction. By use of a 4-way PCA of order (3,2,3,3) satisfactory classification of 47 thick juice samples belonging to 5 factories has been obtained from a spectrofluorometric screening method. Also, a temporal trend has been found to evolve during the time of production. The investigation substantiates the use of modern models from data analysis for extracting significant information from large and complex data sets.