Unless otherwise stated this page contains Version 1.0 content (Read more about versions)
3.8.5 Infrared and Raman spectrophotometry
The absorption of infrared (IR) radiation (at wavelengths above ~ 1 μm) distorts a dipole in the molecule, inducing molecular vibrations with energy shifts an order of magnitude smaller than in electronic absorption spectra (see section 3.8.7). Simple diatomic molecules have no dipole, and IR stretching vibrations are only possible with assymetrical molecules. However, there may be energy absorbed for other types of vibration, such as bending and rocking motions, which give rise to IR region vibrations even with symmetrical bond systems.
Measurements, classically expressed as wavelengths of absorption maxima, may be converted to frequency using the relationship: v = c/λ, where c = 300 Mm/s; with λ in μm, the frequency in THz (1012 Hz) = 300/ λ. For example, a band at 4 μm has frequency 75 THz. In analytical practice, frequency is expressed in reciprocal wavelength (as cm−1), called ‘wavenumbers’, calculated as 10 000 divided by wavelength (in μm). Thus, 4 μm 2500 cm−1. The typical working range for IR spectra is 3500–650 but may optionally be extended up to 4000, or down to 200, cm−1.
Raman emission arises from a quite different process. While the majority of monochromatic exciting radiation in the ‘visible’ range passing through a transparent solution of substance undergoes elastic general (‘Rayleigh’) scattering, a small proportion experiences inelastic collisions with polarisable molecules which may re-emit energy characteristic of a series of vibrational modes. If the polarisability of the molecule changes with the vibrational mode, Raman spectra are observed on the longer wavelength side of the exciting line. The frequency differences from the exciting line are characteristic of the molecule and often identical with infrared frequencies for that substance (Bentley et al., 1974). Because the intensities of the Raman lines are barely 0.01% of incident radiation, the technique has been little more than a research tool until the availability of laser pulsed excitation. Raman spectra are complementary to IR spectra and may provide structural information where no IR absorption takes place. For a discussion of modern Raman spectroscopy, see Baranska and Halina (1987).
Infrared dispersion instruments use a Nernst (rare earth oxide) filament as a source. In earlier instruments, the radiation was dispersed by a cam-driven halide prism, resulting in a linear wavelength scale. In more modern instruments, a series of gratings and filters provide better discrimination and result in a linear wavenumber (frequency) scale. The change in energy, usually measured by a.c. amplified thermocouples, is expressed as the ratio of transmitted to incident radiation and displayed as a series of peaks corresponding to vibration bands. The fundamental limitation of dispersive instruments is that the monochromator allows only about 2% of the incident energy to reach the detector, even at low resolution.
Interferometric instruments employ a beam splitter and moving mirror. The interferometry pattern generated either reinforces or cancels, with absorption at specific wavelengths. Computer processing analyses the interferogram employing software with a Fourier Transform (FT) algorithm. The FT-IR detector may be a doped germanium bolometer or photon detector cooled in liquid nitrogen. The advantages of FT-IR are that it provides a series of quick scans (10 per sec)—so that it can be used ‘on the fly’, e.g. for transient samples or chromatography eluates; it captures all frequencies at once and mathematical averaging improves the signal/noise ratio, thereby giving better resolution and/or facilitating examination of much smaller samples; and it generates much lower stray light (transmittance < 0.02%)—which helps quantitative options. Moreover, computer assistance, as well as providing the necessary Fourier analysis for averaging successive scans, also facilitates matching a current specimen with a subsequently run spectrum of a comparison substance; retrieval of one or more matching spectra from a library database; and subtraction of individual reference spectra relating to components in a mixture of substances.
Possible sources of calibration errors in IR spectrometry include the instrument optico-mechanical system being out of adjustment; or the chart paper being stretched, misprinted or not in registration with wavelength standards. A wide range of error in wavelength/wavenumber scale has been tolerated in older dispersive instruments, e.g. 5000 ± 50 w/n, but with FT-IR instruments 5000 ± 5 and 2000 ± 2 are expected. It is advisable to calibrate regularly: preferably every day that the instrument is used.
The British Pharmacopoeia 1993 [Appendix IIA] lays down a series of operational criteria for use of prism and grating instruments in BP tests. For resolution, there are two measurements of % Transmittance (%T) of a 0.05 mm polystyrene film: (i) %T difference between max. @ ~2851 and min. @ ~2870 cm−1 must be better than either 6% for prism instruments or 18% for grating equipment; and (ii) the difference in %T between max. @ ~1583 and min. @ ~1589 must respectively exceed 6 or 12%. Calibration of wavenumber is checked by reference to a polystyrene film at 14 specified wavenumbers; the reference values range from 3027.1 ± 0.3 to 698.9 ± 0.5 cm−1.
Several options are available. In liquid film presentation, the sample is sandwiched between suitable parallel (halide) plates separated with a gasket (0.1 to 0.01 mm) or is transferred as a mobile liquid into an IR-transparent microcell of suitable pathlength. A solution of a sample may be introduced into a cell of suitable pathlength, selecting a concentration for optimal spectral absorption which is neither too dense nor too transparent: a typical range is 1 to 10% concentration for a path length of 0.5 to 0.1 mm. Despite the presence of the same solvent in the reference beam, which gives approximate compensation for that solvent in the solution, it is generally desirable to choose a solvent that does not absorb at key wavenumbers (e.g. carbon disulphide or carbon tetrachloride—although both have safety problems). (See table of spurious peaks.) The Nujol mull is the easiest sample presentation to prepare: 5–10 mg substance are triturated in an agate mortar with one drop of liquid paraffin, mixed to a cream, and spread between suitable (halide) plates, taking special care to exclude bubbles. For the very widely used halide disc presentation, 1–2 mg of test substance is triturated with approximately 300 mg dry KBr or KCl, ground thoroughly in a mini-mortar, spread uniformly in a die, compressed under vacuum at approximately 8000 atm, and the disk mounted in a lightpath holder. With trace samples, it is useful to use a microdisk. Alternatively, a chloroformic solution is injected via a Hamilton syringe onto KBr pellets, the solvent evaporated and the residue ground, compressed, etc. as before.
Problems may arise with the appearance or clarity of the spectrum. In KBr disks moisture causes a rising baseline and broad OH bands around 3500 cm−1. This interference can be avoided by use of KCl which is less hygroscopic. Inadequate grinding may cause the curious repetitive pattern of Christiansen symmetrical bands, whereas excessive grinding may result in a change of polymorphic form, or even cause anion transfer, e.g. partially convert a base hydrochloride into the corresponding hydrobromide salt. This generally slight risk is avoided by use of KCl if a base-HCl salt is being examined. In Nujol mulls the methyl and methylene bands confuse interpretation around 2800 to 2950 and 1350 to 1470 (see table of spurious peaks). If this is a serious handicap to spectral examination yet a mull presentation is still desirable, then it is possible to use a perfluoroalkane instead of Nujol. In solution there may be incomplete solvent compensation, e.g. a flat baseline or rectangular steps in a solvent absorption region; or evaporation of solvent, which affects quantitation; and sometimes solvation which will modify solute frequencies in solution. Interfering bands may also arise from silicones, plasticisers, salts and plasticsware—see below, and Szymanski (1971) for more detail.
IR Reference Standards may be of three kinds. Authentic Chemical Reference Substances (CRS) permit parallel testing with the substance being examined. A KBr (or KCl for a base-HCl) disk of the CRS and a similar disk of the test sample are prepared, and their IR spectra run from 4000 to 625 cm−1. If unexpected wavenumber differences are observed, then the two substances should be recrystallised in an identical manner and their spectra re-run.
Peak matching relies on inter-laboratory collaborative assessment establishing for a finite data collection the optimum number of strong peaks, e.g. Ingle and Mathieson (1976) used ten named peaks for better than 95% confidence in identification between laboratories. Subsequent experience with medicinal substances concluded that comparison with reference spectra is safer than relying on literature values alone.
For Reference Spectra published by an official agency (e.g. by the British, and International, Pharmacopoeias), specific procedures are recommended for the sample presentation, conditions of calibration, and the spectral range to run. The wavenumber registration of the obtained spectrum may be checked with polystyrene reference peaks preprinted on the Reference Spectrum (usually 2851,1601 and 1028 cm−1) and the concordance of the obtained and reference spectra should be carefully compared as to position, relative intensity and shape of the corresponding absorption maxima. The disadvantages of using Reference Spectra are lack of access to a currently approved authentic specimen in the ‘correct’ crystalline form; inability directly to compare differences arising from use of prism with grating instruments; quality differences between dispersive and FT-IR spectra; and reliance on authenticity of a local specimen for addition of in-house spectra to a data station (see below). The advantages are clear savings over the cost of creation of CRS substances when these would only be required for an IR test; the originating agency saves the need for collaborative authentication of reference specimens; there is no storage and distribution cost; and it avoids distributing any CRS which is a substance scheduled in misuse legislation or is extremely expensive to procure.
The major historic data collection (Sadtler)
provides spectra of well over 60 000 substances, including around 4000
Pharmaceuticals, drugs of abuse, steroids and poisons. Other large collections
include the CRC Handbook (1989) and multivolume series by Lang (1973–80).
Most modern FT-IR instruments have an associated data station which digitalises
the spectrum obtained with the test substance and automatically seeks to match
this with a computer-stored library of spectra. The collection purchased with
the data station can be supplemented by adding in-house run spectra of
substances of special interest. The comparison can be displayed on a split
screen and optionally subtracted to examine the degree of difference.
Valuable diagnostic information is obtained from the extent and manner by which the absorption frequency is affected by the structural environment of the bond system—and by the physical state of the molecule (its crystal habit or solution). Many absorption bands can be assigned to particular chemical bonds in a functional group and within a specific structural environment. However, many other bands, especially in the so-called ‘fingerprint’ region, have NOT been unequivocally assigned and are empirically recorded as part of the ‘fingerprint’ of that molecule. Direct assignment of individual functional groups may help to determine the structure of impurities, metabolites, reaction products, newly extracted natural products, and other material of unknown or inexact origin. Band assignment is not usually necessary for the identification of a known substance, for which there is direct comparison of an IR spectrum of the test substance with that of a CRS, or by comparing with a reference spectrum. The CHART of Characteristic Infrared Frequencies provides a summary of commonly required frequency correlations. Much more extensive functional group tabulations will be found in Bellamy (1975) and most textbooks incorporate correlation tables derived or developed from the original table by Colthup (1950). For a detailed treatment of IR frequencies below 700 cm−1, see Bentley, Smithson and Rozek.
These require measurement of suitable peak intensity in solution which is proportional to concentration. It is the limitation of readability and reproducibility (usually within ± 5%) that generally restricts IR spectra to semi-quantitative use. Examples include determining the proportion in mixtures of two polymorphs with separate fingerprint frequencies; distinguishing and determining the approximate proportion of cis and trans isomers of a substance; and assessing the extent (in atom%) to which 2D has replaced a labile 1H atom (e.g. following treatment with NaOD or DCl). It is also possible to follow a reaction sequence, such as the course of an oxidation or reduction of a functional group and the appearance of the reaction products, by examining the IR spectra of successive chromatographic or other sequentially extracted specimens.
This site is hosted and maintained by the National Physical Laboratory