Evaluation of detection and sample clean-up techniques for on- and off-line fermentation monitoring systems

Two different analytical systems for monitoring of fermentation processes were evaluated for the analysis of three carbohydrates: glucose, xylose and galactose. Samples taken during a controlled fermentation of the three sugars to ethanol were treated and analysed with both systems for the carbohydrate content. The first was an off-line system, based on manual sampling and clean-up, and column liquid chromatography (LC) in combination with refractive index (RI) detection and the other was an on-line set-up, based on microdialysis sampling, LC and an amperometric carbohydrate biosensor as the detection device. The manual clean-up consisted of centrifugation, filtration of the supernatant and dilution. First, the different clean-up procedures were compared, and the other parts of the system were kept constant, viz. the LC separation and the RI detector. Evaluation of the analysis response using the two clean-up procedures yielded in 12 or 13 cases higher values for the microdialysis probe than for the manual procedure. Two different statistical methods, one based on linear regression and another, a paired t-test, found the difference between the two clean-up procedures to be statistically significant. Detailed investigation of the manual clean-up procedure using analysis of variance (ANOVA) revealed that for identical samples treated by this method, a significant difference in-between the samples after treatment was found. Next, the two detection techniqueswere compared, using the manual clean-up procedure in combination with LC. It was found that the sugar content in 15 out of 16 samples was higher for the RI detector than for the biosensor and also in this case both the linear regression method and the paired t-test showed the difference to be statistically significant. Finally, studies on the performance of microdialysis probe membranes after use in a fermentation medium are presented.