Biomedical Engineering Reference
In-Depth Information
in chemical manufacturing. Here, limitations include a poor remote sensing
capability in particular regarding solids manufacturing processes such as dry
mixing, granulation and tabletting, high water absorption and very poor light
penetration in solid samples. Thus, Raman spectroscopy presents very in-
teresting opportunities in pharmaceutical manufacturing processing with its
unique features. Further development is, however, needed in some areas for
Raman spectroscopy to take the final step and be a widespread tool in phar-
maceutical manufacturing.
The influence of particle size has been discussed by, e.g. De Beer et al.
[70]. The authors developed a polynomial model for amount of salicylic acid
in vaseline ointments and pointed out the necessity of controlling the particle
size for quantitative assessment. This is related to the more general problem
of quantitative assessment in turbid media in which the optical path length
changes with elastic light scattering probability. This effect was studied in
greater detail by Abrahamsson et al. for NIR spectroscopy who showed the
benefit of simultaneous measurement of light absorption and scattering for
solid samples of highly varying morphology [71]. This issue is less of a prob-
lem for Raman spectroscopy of solid formulations such as tablets or capsules,
for which several authors have shown satisfactory calibrations of content API
as discussed above, e.g. [16]. One reason why it appeared to be dicult for
crystallisation monitoring [38, 70] may be that in these examples a single
Raman scatterer was monitored whose intensity was measured on an abso-
lute scale rather than evaluated against a calibration based on spectral shape
including several components.
Fluorescence from excipients is another factor that limits the applicability
of Raman spectroscopy. Excipients such as microcrystalline cellulose (MCC)
and lactose exhibit very strong fluorescence upon excitation at 785 nm and
shorter. This has the effect of filling up the dynamic range of the detector and
also introducing the inherent noise associated with the fluorescence increase.
This limits the sensitivity of detecting much weaker Raman bands. In order
to deal with the fluorescence problem, bleaching by prolonged laser irradia-
tion has been demonstrated [57]. By irradiating a sample for a few minutes
the fluorescence background can be substantially lowered. This is, however,
not a very attractive solution for process applications where the acceptable
sampling time is very short, typically on the order of a sample per second.
An alternative approach may be to go to longer excitation wavelengths for
which fluorescence excitation eciency decreases significantly. FT-Raman in-
struments, for instance, employ Nd:YAG lasers that emit at 1064 nm and
result in very low fluorescence but these are not very well suited for process
applications. Another possibility is to use dispersive systems at a fairly longer
excitation wavelength in the range of 800-900 nm. For excitation at 830 nm
much of the Raman emission will still fall within the sensitivity of a silicon-
based CCD detector, which has a cut-off at 1050-1100 nm. For excitation
wavelengths above 900-1000 nm, another option is dispersive spectrometers
employing InGaAs detector elements for better sensitivity in the NIR region.
Search WWH ::
Custom Search