Discovery and Formulation - Emerging Raman Applications and Techniques in Biomedical and Pharmaceutical Fields - page 226

However, these limits of detection must be treated with caution. Where
single particles ( > 1-2 m in size) of a polymorph exist, a Raman spectrum can

be obtained provided they can be located. For example, a common pharma-
ceutical excipient used to lubricate formulation blends is magnesium stearate.
This is typically present at ca. 0.5-1.0% w/w, though formulations with a few
%w/w content exist where the excipient is also used to control API dissolution
rate. Magnesium stearate has a strong and characteristic Raman spectrum but
is rarely detected in Raman chemical images. Although the excipient is present
at nominally detectable levels, it is not observed because being a lubricant it
is spread throughout the sample and as a result at any sample location it is
below its Raman LOD (unless it fails to lubricate the sample and is present
as a lump).
With such low detection levels being possible there is always a question
of how much data should be collected to ensure all present materials are de-
tected in a mapping experiment. Experimental optimisation and statistical
approaches have been reported [63], but Sasic and Whitlock have specifically
addressed this question for systems where the detection of API in formula-
tions at concentrations of less than 1% is required [64]. Using probability and
statistical theory a low-level component should be detectable making 5

μ

1 /p
measurements, where p represents the concentration of the targeted compo-
nent ( p = 1 for a pure component). Thus for API concentrations of 0.5 and
0.2% w/w, 1000 and 2500 measurements are required, respectively, to ensure
detection at the 95% confidence limit. Of course, multiple experiments are re-
quired as insurance against failure. Probability theory indicates that between
20 and 60 experiments are required to ensure success, but in practice this num-
ber is significantly less as demonstrated by the experimental data produced to
test the theoretical hypothesis. Another interesting feature of this work was
that the mapping step size was > 200

×

m to ensure maximum coverage of the
tablet surface. This stage shift is considerably larger than typical API particle
sizes. Here the Raman microscope system was used as a convenient way of
collecting many spectra automatically in the search for low-level components.
However, any subsequently generated chemical image would be meaningless
in terms of component domain sizes.
An emerging application for Raman chemical images is in the characteri-
sation of inhaled products. The size of the API particles is critical for inhaled
delivery. A particle > 10

μ

m particles are essen-
tial for reaching the deep lung, while a cough reflex will expel particles < 1

μ m will not enter the lung; 2-5 μ

m.
As a consequence particle size analysis of ingoing API and formulations is a
key activity in the development and production of inhalation products. The
use of impactors (e.g. Anderson cascade and new generation impactor (NGI))
provides a method for separating dispensed formulation into specific aerody-
namic ranges that are linked to deposition profiles. These fractions can then
be assessed visually by light and scanning electron microscopy methods while
quantitative analysis is performed by HPLC. Laser diffraction methods are
the primary tools used for dynamic particle size analysis. One drawback of

μ