of this text. Unknowns can be identified by matching their spectra with stored library spectra, by a
process of pattern recognition. Quantitative analysis at a single wavelength is not normally possible
using the Beer–Lambert law (p. 357) as almost all NIR spectra are recorded in diffuse reflectance rather
than transmittance mode. Diffuse reflectance is a complex phenomenon that is highly dependent on the
physical and chemical characteristics of the sample and does not have a simple relation with sample
concentration. For calibration purposes, it is necessary to record the reflectance spectra of a series of
standards of known composition over a particular wavelength (wavenumber) range. Reflected
intensities measured at multiple wavelengths selected by the software are used in conjunction with
concentrations to generate sets of equations. Empirical coefficients evaluated from these equations are
then used to calculate the concentrations of specified sample components after recording reflected
intensities from the samples at the same wavelengths. In effect, a mathematical model of the reflectance
spectrum of each standard is computed. These models are then combined and refined by a learning
process so as to generate the best set of calibration data to quantify samples run under the same
conditions. Detection limits are of the order of 1%, although levels down to around 0.1% have been
reported.
Applications of Infrared Spectrometry
The unique appearance of an infrared spectrum has resulted in the extensive use of infrared
spectrometry to characterize such materials as natural products, polymers, detergents, lubricants, fats
and resins. It is of particular value to the petroleum and polymer industries, to drug manufacturers and
to producers of organic chemicals. Quantitative applications include the quality control of additives in
fuel and lubricant blends and to assess the extent of chemical changes in various products due to ageing
and use. Non-dispersive infrared analysers are used to monitor gas streams in industrial processes and
atmospheric pollution. The instruments are generally portable and robust, consisting only of a radiation
source, reference and sample cells and a detector filled with the gas which is to be monitored.
The original applications of NIR were in the food and agricultural industries where the routine
determination of the moisture content of foodstuffs, the protein content of grain and the fat content of
edible oils and meats at the 1% level and above are typical examples. The range of industries now using
the technique is much wider and includes pharmaceutical, polymer, adhesives and textile companies.
The first in particular are employing NIR spectrometry for the quality control of raw materials and
intermediates and to check on actives and excipients in formulated products. Figure 9.26(b)
demonstrates that even subtle differences between the NIR spectra of enantiomers can be detected.
Instrumentation incorporating fibre optics are utilized for remote sensing in process streams, and
identity checks on products in glass vials, plastic