The Choice of Suitable Online Analytical Techniques and Data Processing for Monitoring of Bioprocesses - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 272

further developed by Roychoudhury et al. [ 142 ], who used a multiplexed

calibration technique.
As with MIR, NIR predictive models have also been applied to control systems
in order to allow fed-batch cultures to react in ''real time''. As early as 1994
Vaccari et al. proposed the use of NIR to control the glucose feed in the production
of lactic acid by Lactobacillus casei [ 143 ]. Many others have developed control
strategies for various yeast and microbial cultures [ 144 , 145 ].

Raman Applications

The reported use of Raman spectroscopy for monitoring bioprocesses in situ and in
real time is limited, and this is most likely due to the need for low-frequency lasers
to avoid fluorescence, which can have heating effects due to the long exposure
times necessary for such lasers. Most reported studies describe use of Raman
spectroscopy to monitor yeast cultures. One of the earliest applications of in situ
Raman spectroscopy was monitoring the production of ethanol in yeast fermen-
tations [ 146 ]. In this study the concentrations of fructose and glucose were also
measured. Shaw et al. used a dispersive Raman instrument to monitor the change
in substrate and metabolite concentrations as well as product formation in yeast
fermentation and found it necessary to include a by-pass filter to remove cells as
they were causing interference to the photon scattering process [ 147 ]. The pro-
duction of carotenoids in Phaffia rhodozyma cultures has also been monitored by
dispersive Raman spectroscopy [ 148 ]. Bacterial cultures with monitoring of glu-
cose, acetate, formate, lactate and phenylalanine by in situ measurements have also
being reported [ 149 ]. In a more recent study, Raman spectra were collected in situ
in a mammalian cell bioreactor. As well as monitoring substrates and metabolites,
the spectra were correlated to total cell density and viable cell density, showing
that it may be possible for Raman spectroscopy to distinguish between live and
dead cells [ 119 ]. While these studies all demonstrate the potential of Raman
spectroscopy as a monitoring tool, it has yet to be proved capable of control in
industrial bioprocesses.
Although separate techniques, both MIR and NIR have similar applications in
bioprocessing; both have been used for monitoring and control purposes. Raman
spectroscopy has been used to monitor bioprocesses, but to a lesser degree than the
other vibrational spectroscopies. The manner in which these techniques are
exploited is similar. In all cases, multivariate chemometric models are developed
based on synthetic, semi-synthetic or actual samples from a cell culture. Typically,
these models are then validated and applied to a culture online. These techniques
all have their benefits and limitations, but to date NIR has been the subject of more
investigation and as a result is more developed in terms of applications in bio-
processing. However, the potential of MIR and Raman should not be underesti-
mated or overshadowed.