page     1     2     3     printable version

Fundamentals and pharmaceutical
applications of near-infrared spectroscopy

by I. Antal, Á. Z. Dávid

The near-infrared region
The electromagnetic (EM) spectrum (Figure 1.) covers the range from corpuscular gamma rays to radio radiation. Near infrared (NIR), as defined by the International Union of Pure and Applied Chemistry (IUPAC) [2] is the radiation range of the electromagnetic spectrum, extending from 780 – 2 500 nm (12 800 – 4 000 cm-1). The absorption in the NIR region arises from overtones and combinations of the fundamental mid-infrared bands. Thus in a wider sense, NIR specroscopy (NIRS) deals including wavelengths between 700 - 3000 nm, near the red of the visible spectrum and near the beginning of fundamental infrared stretches of organic compounds.

Figure 1: The electromagnetic spectrum

History of NIR spectroscopy
According to the famous experiment of Isaac Newton in 1665, the white light composes of all the visible colors in the electromagnetic spectrum, and the prism merely separates them. This breakthrough in optics excited the imagination of scientists to develop modern spectroscopy.
In 1880 William Herschel published his experiment on the heating effect of sunlight projecting a rainbow on to a bench by the aid of a prism (Figure 2.) [1]. The temperature increased as the thermometers were moved from violet to red and reached a maximum beyond the red end of the visible spectrum [2]. Herschel concluded that the sunlight contains more than just the colors, he referred the “invisible light” as “radiant heat” and the “thermometrical spectrum”. Herschel’s discovery was a milestone to the rest of the electromagnetic spectrum .

Figure 2: Heating effect of light discovered by Herschel

In 1821 Thomas Johann Seebeck discovered that when any conductor (such as a metal) is subjected to a thermal gradient, it will generate a voltage. This is now known as the thermoelectric effect, which opened up the possibility to compose thermocouple detectors. Eight years later, in 1829, Nipece and Daguerre invents the photographic plate, which turns out to be sensitive to NIR radiation. It was left to Ampere, in 1835, to demonstrate that NIR had the same optical characteristics as visible light and conclude that they were the same phenomenon, by using thermocouple detectors. Later on in the century Abney and Festing recorded the spectra of organic liquids in the 1 – 1,2 mm range in 1881 and recognized atomic grouping and the importance of hydrogen bonds in the NIR region
Inspired by their work, W.W. Coblentz constructed a spectrometer which was highly sensitive to both vibration and thermal disturbances. Coblentz had to leave the room in which the instrument was set up, after each step in the rotation of the prism. It took one whole day to record a single spectrum, but he ultimately recorded the spectra of several hundred compounds in the 1 – 15 µm wavelength region. Using his instrument he was the first to verify Planck's Law, and was also the first to show that different atomic and molecular groupings absorb characteristic wavelength “fingerprints” in the infrared region [3].
Following, in the first half of the twentieth century, the spectral database have been broadly extended by many contributors of NIR and IR spectroscopy. Though while infrared spectroscopy had moved away from being a scientific curiosity, it was used very little, since suitable spectrometers did not exist. Over half a century has passed, before infrared spectroscopy became a tool for routine chemical analysis, and almost two-third of a century was required for NIR measurements making their debut in everyday laboratory work.
Possibly the first quantitative NIR measurement was done by F.E. Fowle in 1912, by determining atmospheric moisture in the Mount Wilson observatory.
In the 1930s lead sulphide (PbS) was studied as a compound semiconductor for heat–sensing and the Second World War stimulated its development as an infrared detector. Twenty years later, in the 1950s, it turned out that it is very sensitive for the 1 – 2.5 µm wavelength region, and thus it became a commercially available NIR detector.
In 1931 Kubelka and Munk published their theory on diffuse light scattering which still serves as a background for measurements of solid materials [4].
Analytical use of NIRS have developed more slowly than applications in the visible and infrared range.
The lack of accurate calibration techniques and computational hardware and methods were a setback of NIR spectroscopy. Beside the construction of low cost NIR instruments with high signal-to-noise ratio, advances in mathematical and statistical analysis were necessary for industrial use. In the 1960s, Karl Norris from the U.S. Department of Agriculture did pioneer work in the field of NIR analysis and introduced it into practice for agricultural and food products [5]. Development of chemometrics and appropriate calibration models were a prerequisite for pharmaceutical applications from the 1980s [6,7].

Principles of NIR spectroscopy
NIR spectra are primarily the consequences of overtones and combinations of the many fundamental absorption bands of the mid and far infrared regions. The overtones are anharmonic, which makes NIR spectra complex and overlapping. Due to energy considerations, most of the overtones found in the NIR spectrum arise from the X–H stretching modes (O–H, C–H, S–H and N–H) [8]. Being quantum mechanically forbidden transitions, the overtones represent a 10 to 1000 times weaker band, than the fundamental mid-IR vibrational bands.
The absorption in the infrared region is a result of molecular vibrational and rotational states. The background of vibrational spectroscopy is the concept that atom-to-atom bonds within molecules vibrate with frequencies that may be described by the laws of physics and are, therefore, subject to calculation. When the material coming into contact with the given radiation, absorbs energy, it will be excited to a higher energy level, thus the difference in the energetic state of the material may be described by quantum mechanical calculations. Assuming, that the band energies arising within a molecule from the vibration of a diatomic harmonic oscillator, and obey Hooke’s Law, the lowest or fundamental frequencies may be roughly calculated by the following equation:

(1)

where v is the vibrational frequency, k is the classic force constant and µ is the reduced mass of the two atoms. This theory works well in case of the fundamental frequencies of diatomic molecules, and gives comparable results for calculation of the average value for two-atom stretch of polyatomic molecules. However in real molecules, the electron sucking-, or donating effect of nearby atoms can significantly influence the bond strength and length, thus the fundamental frequency of X-H bonds as well. By using quantum physical calculations the fundamental transition states of a molecule can be calculated:

(2)

where E is the energy level of the different (v1, v2, v3, …) vibrational states and hv is the quantum term. Since fundamental transitions between the ground and excited energetic states are only allowed by selection rules, whenever the molecule reaches a vibrational state above the fundamental, it is called overtone. Transition to a state, where the fundamental quantum levels are 1 in both direction, gives combination bands. By calculation, overtones and combinations are not allowed states, but do appear as weak bands due to anharmonicity or Fermi resonance.
In practice, the so called ideal harmonic oscillator has limits. This can be best explained with the so called spring model, where atoms of a molecule are imagined as being attached to each other with springs. In this way the oscillation of the molecules can be among bonds: stretching, or it can be among bond angles: vibration. These different types of oscillations are graphically shown in Figure 4.

Figure 4: Stretching and vibration model of molecules.

However in an ideal case, as the atoms close in on each other, real compression forces are fighting against the spring, while on the other end, if stretching is too much, the spring looses its capacity to return to its coil form. This means that in real molecules the electron clouds between the two bonding atoms mean a resistance against the nuclei closing in too much, while on the other hand, if the energy with which the molecule will stretch reaches the amount of the dissociation energy, the
bond will break up. To reach a state, where the bond breaks is always easier than to compress the atoms of the molecule. The barrier for decreasing distances among two neighboring nuclei increases at a rapid rate, while at the other direction, stretching slowly approaches zero, which is shown in Figure 5. As it is apparent from the graphic representation, energy levels are not equal in an anharmonic oscillator. In practice, anharmonicity of the first overtone of a fundamental band occurring at 3500 nm would be at 3500/2 plus a small shift to longer wavelengths, placing the signal between 1785 to 1925 nm.

Figure 5: The energy diagram of molecule’s vibrational model showing an (a) ideal diatomic or (b) anharmonic diatomic oscillator.

page     1     2     3     next page