Résumé
It is seen as a promising source of biomass for crucial applications such as energy production,1 source of highly valuable molecules. Optimizing the monitoring of the mass cultivation process is challenging2 as it requires high frequency biochemical and physical characterization of the microalgal cells throughout the cultivation process. Visible (VIS) and near infrared (NIR) spectroscopy have the potential to characterize dense microalgal cultures status. This measurements are fast and can be performed with minimum or no sample preparation. Applying spectroscopic methods to dense algal cultures can be troublesome due to multiple scattering effects resulting from high cell density (106− 109cells/mL). In this study, we attempt to model the optical properties of dense algal sample in which multi scattering occurs. Our work aims at providing a simulation method that can be used to interpret the spectral properties of a dense algal culture, and to provide information on the biochemical and physical characteristics of the algal cells during growth. For this purpose, we implemented a method based on Kerker's solution of the extended Mie theory3 coupled with the numerical resolution of the radiative transport theory4 was implemented. The algal cells are thus modeled as multilayered spheres composed of different organic materials and dissolved photopigments5. Two different monospecific batch cultures of Isochrisys galbana and Phaeodactylum tricornutum were tested. The biochemical compositionof cell, the number, mean size and mean dry weight and the total transmittance over the [400-750 nm] spectral range were measured at three different growth (beginning, exponential and stationary) stages of each of the two cultures. The measured transmittance spectra were used in the simulation method in order to predict cell mean size and density on the one hand, and the pigment quantity and composition on the other for the two species at each growth phase. Thus the mean equivalent spherical diameter was retrieved for both strains and at each sampling date with a relative error below 10%. Cell density was also successfully predicted for I. galbana for each growth phase studied, with a relative error below 7%, which suggest that cell dynamics could be successfully monitored during cell population growth. However, more discrepancies are observed for P. tricornutum, for which the cell density is predicted with a higher relative error, up to 27% at day 34. This higher relative error is due to the non-spherical shape of the P. tricornutum cells. Predictions of the relative proportions of each pigments are closely match HLPC measurements, with relative errors below 9%. Therefore changes in the carotenoids/Chl ratio may also be successfully predicted with low relative errors (3-7%). For P. tricornutum, the predicted values show generally more discrepancies with the measurements than for I. galbana. This is likely explained also by the shape effect of these microalgae.