Near-infrared spectroscopy

Visible near-infrared spectroscopy is a well-established analytical method for examining food materials. The attraction of this technology is that it is low cost, robust, and measurements can be made very fast and usually without damaging or even touching the sample. Vis-NIR is used in a wide range of factory and processing environments, for making measurements such as moisture in breakfast cereals and biscuits, fats in cheese and chocolate, and protein in grains and spices. Vis-NIR reflectance and transmittance measurements provide information on the molecular composition of the foods. Molecular bonds can be imagined as tiny springs, bending, rocking or stretching on a microscopic scale.

When a sample is illuminated with infrared radiation, the vibrating bonds absorb photons of specific frequencies which depend on the mode of vibration and crucially, the chemical composition of the bond. By examining the radiation after it has interacted with the sample, we can see what frequencies have been absorbed which in turn tells us about the nature and the amount of different molecular bonds that are present in the sample. Here, a Vis-NIR portable spectrometer is being used to authenticate olive oils. The bottle is illuminated with radiation delivered by an optical fibre from the infrared source.

Some of the radiation reflected back after passing through the sample is captured by a further fibre and returned to the spectrometer, where the detector records the spectrum and passes it onto a computer. Here is the spectrum from the olive oil sample. The software has been mathematically trained to recognise extra virgin olive oils, using a pattern recognition method. Samples that are a close match to the reference database are reported as authentic. This method can easily tell the difference between olive oil and other edible oil types, for example, hazelnut, sunflower, or rapeseed oil. The software has been configured to report these as suspicious. We can see from the reflectance spectra that they are a poor match with the olive oils.

This approach is fast, cheap, and non-destructive for screening oils to check their identity or to check that a product is within specification with regard say to its colour. The technological advantage of working with this wavelength range is that components needed to construct a spectrometer are low cost. Ordinary light bulbs can be used as broadband sources of visible near infrared radiation. Spectrometers can be built with no moving parts, using diffraction gratings. Detectors are similar to those used in cameras and mobile phones. And optical fibres can be used to achieve all kinds of different optical geometries and remote sampling possibilities. Vis-NIR does, however, have limitations.

Oil types that have very similar reflectance spectra may be hard to distinguish from one another, especially if there is variability within each type. Remember that edible oils are natural products which are affected by factors like growing conditions and production processes. Here we have some samples of extra virgin olive oil that have been deliberately adulterated with sunflower oil. One of these contains quite a high amount of adulterant 40% weight for weight, and the pattern recognition flags this up as suspicious. However, the other contains only 15% of the adulterant, and this sample incorrectly passes our test for authenticity.

This is because the difference between the adulterated and authentic samples is not large enough to be detected above the background of measurement noise on product variation in the reference database. To lower the detection limit, we need to look for more specific compositional markers of individual oil types. And in the next video, we’ll see that we can find these by moving to the longer wavelengths that are used in mid-infrared spectroscopy.