On-line bioprocess monitoring with a multi-wavelength fluorescence sensor using multivariate calibration

Cultivations of Pseudomonas fluorescens were monitored with a multi-wavelength on-line fluorescence sensor. The multi-wavelength fluorometer used excitation light from 270 to 550 nm with 20 nm steps and measured fluorescence emission from 310 to 590 nm. The fluorescence, on-line exhaust gas measurements and off-line analysis of nitrate, succinate, optical density and protein were compared chemometrically by multivariate calibration, i.e. computing partial least square (PLS) regression models. Based on the multivariate regression models, it was possible to determine CO2 and O-2 composition in the exhaust gas (the correlation coefficients, R-2 between the predicted values by the PLS model and the measured values was 0.97 for CO2 and 0.97 for O-2, respectively). Also to make quantitative determinations of succinate (R-2 = 0.97), protein (R-2 = 0.94), optical density (R-2 = 1.0) and nitrate (R-2 = 0.98) in the medium based on the fluorescence spectra. Only a limited data set was available but the results indicated that the sensor could indirectly determine non-fluorescent compounds, i.e. nitrate and succinate, which probably is due to the stoichiometric relationship between fluorescent cellular components and non-fluorescent compounds. Consequently multi-wavelength fluorescence is an interesting technique for a wide range of applications. (C) 2001 Elsevier Science B.V. All rights reserved.