A Simple guide to complex world of overtone and combination bands: Theoretical simulation and interpretation of NIR spectra – summary of the workshop at NIR-2021 Beijing Conference

It is probably most important to mention that the background for some of the well-known concepts, such as the harmonic approximation of molecular oscillations, normal modes, or the intensity of the overtones, are most often not explained thoroughly in the textbooks. For instance, the famous ‘selection rule’ for infrared spectroscopy, stating that only the transitions between the adjacent vibrational levels (Δn = ±1) is ‘permitted’ in the harmonic oscillator cannot be explained without invoking the behaviour of the dipole moment function (which is, in this case, approximated by a linear function). In a similar manner, the example of a diatomic molecule, beyond which the textbooks seldom extend, is entirely unsuitable to explain the appearance of the combination bands. These features, among many others (e.g. the normal vs. internal coordinates describing vibrational motion), have been clarified in detail during the workshop.

It is, unfortunately, not possible to provide in-brief a good basis for the listed characteristics of vibrational spectroscopy. However, an interested reader should find comprehensive introductions to these essential topics in our recently published textbook chapters^2,3 and review articles.^4–6

Recent progress in theoretical NIR spectroscopy largely improved our understanding of NIR spectra, e.g. it is currently possible to accurately simulate NIR absorption lineshape of various molecules and assign in full detail the measured bands to the corresponding vibrational modes.^2,4,5 The example demonstrating the complex structure of overtones and combination bands, in this case in the NIR spectrum of thymol, is shown in Figure 1. Thymol is an interesting molecule, a simple terpene compound that is very popular in medicinal plants and it is responsible for the antioxidant potential of the plant and the medicinal products derived from it.⁷ It is, therefore, a molecule of a quite simple structure but still very relevant to one of the typical applications of NIR spectroscopy – phytopharmaceutical analysis.⁸ Despite the structural simplicity, the NIR spectrum of thymol isolated in a fairly inert solvent (carbon tetrachloride), which largely eliminates any interfering matrix effects, still presents a considerable complexity with multiple overlapping peaks.

The opportunity to accurately reconstruct NIR lineshape brings a new perspective on NIR spectroscopy, which was thoroughly discussed during this workshop.⁴ The workshop aimed to debunk several myths as well. For example, it is often said that NIR spectroscopy is the spectroscopy of overtones, but in fact, it is the spectroscopy of combination bands – and the example of thymol recapitulated here should illustrate it very well. That example also demonstrated how much different the concept of ‘band assignments’ in NIR spectroscopy should be approached, in contrast to the fairly straightforward one routinely used in mid-IR or Raman spectroscopy. In contrast to pinpointing the peaks visible in IR spectrum, we observe ‘diffused’ contributions that reflect the convoluted (i.e. overlapping) nature of the absorption regions in NIR spectra, exactly as demonstrated in the case of thymol (Figure 1).

The molecules with more complex structure, e.g. nucleic acid bases (i.e. nucleobases: adenine, guanine, cytosine, thymine),⁹ short-, medium- and long-chain fatty acids,^10–13 or polymers,¹⁴ reveal understandably more convoluted NIR spectra. These cases were discussed in detail in the workshop, while here I provide the example of the NIR spectrum of PVC polymer for illustrative purposes – in Figure 2. As one should notice, the numerous individual simulated bands are ‘hidden’ beneath the observable NIR lineshape (Figure 2). The following part of the workshop aimed to demonstrate how and when the structural fingerprint appears and can be identified in NIR spectra, with examples of e.g. six-membered aromatic ring giving rise to a clearly identifiable absorption feature clearly visible, for instance, in the spectra of polymers.¹⁴

Finally, the majority of the third video workshop was devoted to demonstrating how the ability to accurately predict NIR spectra can be used for the benefit of the analytical applications of NIR spectroscopy. The most basic examples, obviously, showed that a detailed interpretation of the spectra of analysed constituents becomes available. The benefit of performing a more conscious analysis, with the understanding of the ‘signal’ that is sought for in the analysed spectra should not be underestimated.¹⁵ However, even more direct applications become viable with this knowledge. The interpretation of the meaningful variables determined in the calibration models should be mentioned first.^16–19 What follows, is the ability to elucidate the contribution of the chemical information to the performance profile of different spectrometers and the instrumental difference in general. This becomes particularly essential for miniaturized sensors, given their often very limited (i.e. narrow) operational spectral region.²⁰ This information cannot be easily deducted from the typically emphasized purely technical parameters of a given instrument.^21,22 This, in turn, offers the ability to perform ‘smart design’ of the analysis by selecting the most suitable sensor without prior necessity to perform any standard measurements. Finally, a light is shed on the differences in analytical performance of competing vibrational techniques, e.g. mid-IR vs NIR spectroscopy, in performing exactly same analysis. Here, the example may be served by the case of melamine content in milk powder quantification.²³ The analysis of the meaningful variables showed that entirely different vibrational modes of melamine matter for the regression models describing melamine content for mid-IR and NIR spectra.²³ Such knowledge gives an even broader perspective on the applicability of different techniques to a certain analytical scenario and opens pathway for a smart design of future analyses.