Online multiparameter sensor for microalgae cultivation: prediction of nitrate, biomass and pigment concentrations

Abstract

Industrial microalgae production is widely recognized as a sustainable and promising method for generating nutraceuticals, biofertilizers, food ingredients, and feed. However, the lack of automated process control in current industrial facilities is a significant limitation, primarily due to the absence of sensors capable of measuring the growth medium's chemical composition and biomass. In this study, a novel method for simultaneously measuring several critical process parameters, including nitrate, various pigments, and biomass is presented. The approach is based on optical transmission and machine learning-based chemometrics and is reagent-free and non-invasive, making it ideal for development into an automated online device. Measurements of optical spectra and chemical concentrations were conducted for two microalgae species, Arthrospira platensis and Tetradesmus obliquus, across different levels of biomass and nitrate. Through Partial Least Squares analysis of the data, biomass concentrations can be inferred in a range up to 8 g.L−1 (root mean square error (RMSE) of 0.55 g.L−1 and R2 0.75 for A. platensis, and RMSE of 0.52 g.L−1 and R2 0.88 for T. obliquus), using the same model for the different species (RMSE of 0.72 g.L−1 and R2 0.74). Nitrate concentrations can be estimated in a range between 0 and 800 mg.L−1 (RMSE of 41.70 mg.L−1, R2 of 0.82) in the presence of biomass and without the need to discriminate between the microalgae tested. For T. obliquus, total chlorophyl (from 2 to 18 mg.g−1 biomass, RMSE 1.24 mg.g−1 and R2 of 0.71) and carotenoids content (from 1 to 12 mg.g−1 biomass, RMSE 1.12 mg.g−1 and R2 of 0.59) can be estimated directly during the cultivation.