Complex biogenic mixtures such as biologically derived hydrolysates (e.g. yeastolate), cell culture media, and bioreactor broths are widely encountered in biotechnology. They are generally very complex, and thus comprehensive analysis of these types of mixtures poses a very significant challenge due to the potentially vast array of different molecular species present and their widely varying concentrations. For example, the complex cell culture media used in the pharmaceutical manufacturing of therapeutic proteins can be generically characterized as consisting of low (0.5 to 5%) concentration of dissolved solids in water, where the number of chemical species varies from 5-10, up to many hundreds. These mixtures need to be rapidly identified and classified, e.g. identifying raw materials or blended formulations for quality control, and have the compositions of specific components measured accurately, e.g. a specific amino acid in a cell culture feed medium. Finally there is also a drive to try and correlate media variance with global process parameters such as final product yield .
Traditional separation-based methods (often coupled with mass spectrometry) that rely on sample fractionation are expensive, time-consuming, and often require considerable analyst/operator expertise. In contrast, spectroscopy-based methods such as fluorescence, Raman, and NIR spectroscopy can offer qualitative characterization and quantitative analysis of these mixtures without many of these drawbacks. These non-contact, non-destructive, rapid, and relatively inexpensive methods can deliver some (but not all) of the information required. Here we show that multidimensional fluorescence spectroscopy (excitation emission matrix, EEM) and chemometrics offers significant benefits for rapid and efficient analysis of the cell culture media [1,2], biogenic raw materials , and bioreactor broths  encountered in industrial biotechnology. Specifically, the EEM method can be used to quantify some amino acids in media, analyze filtration and sterilization issues, and correlate media variance with process yield.