Good Reasons to Use ATR Libraries

1. Introduction

Transmission spectroscopy was the only method in general use until the early 1960s. The first applications of attenuated total reflection (ATR) or internal reflection spectroscopy were reported independently by Fahrenfort and Harrick. Especially macromolecular materials (rubbers, fibers, fabrics, coatings, laminates etc), which often failed to yield useful transmittance spectra, became accessible for infrared spectroscopic investigations by the ATR technique. Meanwhile, the ATR technique has been proven to be an extremely useful technique for material analysis and has become more widely used for all but gaseous samples.

For ATR, light is introduced into a suitable prism at an angle exceeding the critical angle for internal reflection producing an evanescent wave at the reflecting surface. A sample is brought in contact with or in close proximity to the reflecting surface. From the interaction of the evanescent wave with the sample, a spectrum can be recorded with little or no sample preparation. If there is a sufficient solid sample, then either single or multiple reflections ATR can be used.

The ATR technique is non-destructive and can be used for solids, liquids, and powders. All of these features make the ATR technique an ideal tool for fast routine analysis.

For thin films, the ATR spectra are usually identical to transmission spectra. For thick films, the absorption bands are more intense at longer wavelengths. As the angle of incidence approaches the critical angle, the bands tend to broaden for lower wave numbers and the minima are displaced to lower wave numbers.

2. Impacts of the ATR method on Library Search

Identification of unknown materials is typically achieved by searching the spectrum of the material in large datasets of spectra (libraries of reference spectra). However, most of these reference spectra are recorded as transmittance spectra and stored as absorbance spectra in reference libraries. Therefore, these libraries are not directly applicable on ATR spectra, because the spectra obtained by the ATR technique may be quite different from the transmittance spectra.

There are three major reasons why the identification of unknown ATR spectra usually leads to unsatisfactory results.

2.1 Intensity differences because of penetration depth

Fig. 1: Infrared "pathlength" for transmission and ATR spectra.

Fig. 1: Infrared "pathlength" for transmission and ATR spectra.

This effect is mainly due to the fact that the infrared beam is internally reflected within a crystal, and the sample is placed directly onto the crystal. However, as the penetration depth of the IR radiation strongly depends on the frequency the band intensities differ between the two methods. Because of these fundamental differences between both techniques, the transmittance and ATR spectra of a material can be considerably different.

As mentioned above, the most significant difference between transmission and ATR is related to the interaction of the infrared beam with the sample. This interaction is quantified by the infrared "path length," and the difference between the two techniques is illustrated in Figure 1.

In transmission, the path length depends only on the sample thickness. In ATR, however, the path length depends on the infrared "penetration depth," which is a function of (among other things) the wave number of radiation. For an ATR measurement, lower wavenumber radiation interacts with a sample more strongly than higher wavenumber radiation.

Figure 2 shows the KBr, the corrected ATR and the original ATR spectrum of Vitamin B12. The absorption peaks in the uncorrected ATR spectrum are more intense at lower wave numbers, relative to those at the higher wave numbers. In addition the peak position of the bands at high wave numbers are slightly lower than in the KBr spectrum.

A common way to approach the pathlength issue is to correct the ATR spectrum. This so-called ATR correction simply scales the ATR absorbance values by a penetration depth function, effectively compensating the wavenumber dependence.

full-screen Fig. 2: KBr, corrected ATR and original ATR spectrum of Vitamin B12.

Fig. 2: KBr, corrected ATR and original ATR spectrum of Vitamin B12.

However, even after such a correction the search results in a library created from transmittance spectra are far away from being satisfactory. Commonly used search algorithms are based on intensity comparisons of sample and library reference spectrum. Thus, a match with a high hit quality can only be expected if intensity and band positions are very similar, which is not possible for the comparison of ATR spectra with transmittance (or absorbance) spectra or vice versa.

2.2. Dispersion

A further significant difference between ATR and transmittance spectra is caused by "dispersion" on peak shapes and locations in ATR data. Dispersion is the change in the sample’s refractive index as a function of wave number. As it turns out, the penetration depth is also a function of the sample refractive index, so when the refractive index changes, the penetration depth changes as well. As consequence considerable changes in the peak shape and position can be observed. This has severe implications for identifying substances affected by this dispersion effect.

2.3. Artifacts in Transmittance Spectra

Libraries of transmittance spectra are of little assistance for the identification work if these libraries contain artifact peaks because of the sampling technique used. It is common to analyze solid samples by suspending them in a Nujol (hydrocarbon oil) matrix. Nujol itself exhibits an infrared spectrum, constituting a spectral interference. The Nujol peaks can lead to erroneous interpretations of the data if they are not accounted for. However, excluding the spectral ranges where Nujol peaks appear from spectrum search automatically excludes most of the characteristic peaks of methylene sequences and methyl groups, which are for many compounds the most characteristic bands.

3. Results

The influence of the above three major objections against using transmittance libraries for ATR spectra is illustrated in Table 1.

As basis for this table three different samples (Vitamin B12, Polyvinyl Alcohol (PVA) and Polystyrene (PS)) have been selected and searched against a set of 21,140 KBr transmittance spectra (4,000 cm^-1 to 400 cm^-1, 4 cm^-1), and 10,000 ATR spectra*, (4,000 cm^-1 to 400 cm^-1, 4 cm^-1), [ATR-FTIR Spectral Library- ICATRall, S.T. Japan-Europe GmbH]. All three compounds were measured as KBr and ATR spectra and the data was included in the library. The search algorithms “SQDifference 1. Derivative” and “Pearson” as available in theIRSolution software were applied. These two algorithms usually give the best results in case of band shape and intensity problems.

The table lists the normalized hit qualities together with the rank in the hit list. The maximum number of hits was limited to 30.

Whereas in the case of PS and Vitamin B12 only one spectrum of each preparation is available in the library, there are 3 different PVA spectra of each preparation.

As expected in all searches of the three ATR spectra in the full collection (KBr and ATR) the first hit was always an ATR spectrum and showed a high hit. However, no KBr spectrum was found among the first 30 hits. The same is valid for a search with a KBr spectrum. The first hit always was a KBr spectrum and no ATR spectrum was listed among the first 30 hits.

The more interesting result however is the search of the corrected ATR spectrum (column ATRC).

When applying the Pearson algorithm on PS the KBr PS spectrum was found at rank 1, for the derivative algorithm however the ATR spectrum at rank 1 with a not really acceptable hit quality. The KBr PS spectrum was at position 12.

For PVA the situation is worse: Pearson algorithm rated rank 3 and derivative algorithm rank 1, but with a very bad hit quality. Even more striking is the fact: the other 2 PVA spectra of each preparation were not among the first 30 hits.

The Vitamin B12 case is even more unreliable: The KBr spectrum achieved rank 2 for the Pearson algorithm, but no hit for the derivative algorithm.

Table 1: Results of spectrum search in ATR and KBr libraries. [ATR-FTIR and FTIR Spectral Libraries, S.T. Japan-Europe GmbH].

4. Conclusion

The best search results are obtained when searching ATR spectra in ATR libraries and transmittance spectra in transmittance libraries. Searching ATR spectra in transmittance libraries without any ATR correction does not make any sense at all. Searching corrected ATR spectra in transmittance libraries in certain cases may give a reliable result, but these cases are pretty rare.

Clearly, the best approach for identifying samples scanned by ATR-technique is to search ATR databases.

5. Acknowledgement

The searches were performed with IRSolution from Shimadzu in the spectral databases from S.T. Japan-Europe.