How can NIR spectroscopy be used to analyze beverages?

Jay R. Powell, PhD, Analytical Answers Inc.
July 2, 2016

 

Spectroscopy is a branch of the physical sciences which studies and applies the interactions between electromagnetic radiation and matter. This field is further divided based on the wavelength range, such as X-ray spectroscopy, visible spectroscopy, infrared spectroscopy, and microwave spectroscopy, to name just a few. Hamamatsu offers instruments and components covering both the visible range, i.e., light we can "see" (300-700 nm), and the near-infrared range (700-2500 nm). This last near-infrared, or NIR, range has found numerous applications in a wide range of different chemical, material, environmental, and other analysis.

 

In the visible range, light can be absorbed by atoms and molecules through electronic transitions. Absorption of light of the right wavelength or energy by an atom or molecule leads to a decrease in the amount of light transmitted. In the condensed phase (liquids and solids), visible light absorption occurs from complex conjugated double bonds and aromatic rings in organic molecules, although many aqueous transition metal ions, through complexion with associated water molecules, can also form strongly-light-absorbing species. This absorption of light through electronic transitions produces the broad range of colors we perceive from solids and liquids.

 

In the NIR range, the energy carried by the NIR light is too low to excite the electronic transitions noted above, and thus the intensely colored species observed in the visible range have less of an influence in the NIR. In the NIR range, light absorption occurs through vibrational transitions, where absorption of light of the right wavelength or energy increases the vibrational mode between two atoms bonded together in a molecule. As NIR absorption does not require conjugated bonds or solvent complexes as in the visible, it is more sensitive to both the structure of the molecule and how that molecule interacts with nearby molecules. It is this sensitivity to the molecular structure and molecular environment which gives NIR spectroscopy its power.

 

The most common solvent used, in fact often referred to as the "universal solvent," is water. Water, through interaction with O-H stretching and H-O-H "scissoring" vibrational modes, is a weak NIR absorber. Figure 1 shows the NIR absorbance profile of water, covering approximately 800 to almost 1400 nm in wavelength, as measured in a 1 cm (10 mm) thick quartz cuvette. The Y axis shows the absorbance value, commonly abbreviated "A," and which is the negative base 10 log of the percent of the light transmitted at that wavelength (A = -log10(%T/100), where %T, a.k.a. %Transmittance, is the ratio of the light measured by the instrument with and without the sample in place, or %T = I/I0). Absorbance values are often preferred in such applications, as a simple relationship between absorbance, concentration, and sample thickness (or pathlength) is given by:

 

Equation 1

 

Aλ=aλbc

where Aλ = measured Absorbance at a given wavelength λ, c = concentration, b = sample thickness or "pathlength," and aλ = absorptivity of the species of interest at the given wavelength λ. Assuming that a constant pathlength cell is used, such as the common 1 cm cuvette described above, then the pathlength term can be ignored and the measured absorbance can be directly related to the concentration of the species of interest.

 

In Figure 1, we note three areas where water shows stronger absorbances: a broad band around 970 nanometers (nm), a stronger broad band around 1200 nm, and a very strong absorbance starting around 1300 nm and increasing to the end of the range near 1400 nm. These absorbance areas arise from the interactions between the O-H and H-O-H scissoring vibrations of the water molecules, and the broadness of the bands is due to the extended interactions, or "hydrogen bonding" between adjacent water molecules. These hydrogen bonding interactions can be used in characterizing aqueous systems.

Figure 1. NIR absorbance profile of water

In Figure 2, three plots are presented showing deionized (DI) water, a diet beverage, and a regular beverage sweetened with a relatively high loading (ca 10% w/w) of one or more sugars. Note that in contrast to Figure 1, the strong absorbances in the DI water at 970, 1200, and 1300+ are no longer observed here, as these spectra were collected ratioed to DI water as a background reference. Using DI water as the background reference, if there is no difference in the interactions between water molecules between a sample and the DI water reference, little to no change will be observed in these regions. Adding more components to the (mostly) water beverages causes larger and larger changes in these absorbance bands, as seen by comparing the traces from the regular (sugar sweetened) beverage and diet (no sugars). Thus, one simple application of NIR spectroscopy using Hamamatsu instrumentation is to differentiate between sugar-sweetened and artificially sweetened beverages.

Figure 2. NIR absorbance profiles of water (blue curve), diet beverage (purple curve), and beverage with sugar(s) (red curve)

Instrumentation components for a beverage analyzer

For NIR absorbance measurements, there are four major components required. An overview of these components is presented in Figure 3.

Figure 3. Components of Hamamatsu beverage analyzer

First, a strong and stable source of NIR light is required. A strong or "bright" source is desired, in order to minimize the impact of light transfer and sampling losses. A stable source, that is a source whose output does not measurably change over time, is necessary as such changes in source output from one measurement to the next will be indistinguishable from the differences we are trying to measure in our samples. Here, we have used the Hamamatsu L7893 D2 / Quartz Halogen UV-Vis-NIR light source. This switch-selectable light source allows selection of the D2 lamp for UV-Vis sampling, or the Quartz Halogen NIR source, which is used here.

 

Next, a method to transfer the light from the source to the sample is needed. Here, creativity in optical coupling can be used, often based on optical tables, mirrors, and focusing lenses to move the light from the source to the sample, and from the sample onwards. One easy method to do this in the NIR range is to use dedicated NIR fiber optics, which are based on fine silica fibers formulated to minimize light scattering and loss from other foreign species, such as hydroxyl (-O-H) groups. In addition, as a reproducible sample thickness or pathlength is required, a standard 1 cm (10 mm) pathlength quartz cuvette is used, which is held in a dedicated cuvette holder with fiber optic connections. Here, we have used a pair of ThorLabs TP01195070 optical fibers with industry-standard SMA connectors to take light from the Hamamatsu L7893 source (also with industry standard SMA connectors) to a ThorLabs CVH100 cuvette holder with a pair of SMA connectors (one for light going in, one for light going out), with our samples and water reference solutions held in Pike Technologies 162-0223 glass cuvettes with covers. This is shown in more detail in Figure 4. These cuvettes are the most common visible to NIR range sampling accessory, are available from a wide variety of sources, are easy to clean, and are relatively inexpensive.

Figure 4. Quartz cuvettes, holder, and fiber light guides

Finally, the NIR light, after passing through the sample, has to be measured. A dispersive spectrometer collects the light, and separates or disperses it into the individual wavelengths. Here, we have used a Hamamatsu C11118GA dispersive spectrometer, with a thermoelectrically cooled integrated array detector, covering the range of 850 nm out to 2,500 nm (or 2.5 μm, which is the edge of the mid-infrared region of 2.5 μm to 25 μm). Here, our spectrometer and detector range is well beyond the range needed for our beverage analyzer. However, substitution of other source, light transfer, and sampling components allows easy reconfiguration of this spectrometer for other applications.

Controlling the beverage analyzer spectrometer

While the selected hardware components offer a very powerful and capable combination, these components on their own will not supply us with the differentiation between the artificial and sugar sweetened beverages we wish to demonstrate. Here, additional communications, control, and data acquisition of the C11118GA system is provided through a Hamamatsu software interface package. This interface package allows easy definition, access, control, and acquisition of the C11118GA spectrometer through common development tools, such as Microsoft Visual Studio’s Visual Basic (VB). Figure 5 shows an example of using these interfacing tools in VB, where the program calls up the function to acquire the spectral data from the spectrometer.

Figure 5. Example of "GetSpectrum" data acquisition from Hamamatsu's spectrometer interface

VB then provides the tools to easily lay out windows, screens, menus, and function buttons as part of a dedicated graphical user interface (GUI) for our beverage analyzer (Figure 6).

Figure 6. Visual Basic layout for beverage analyzer

In modern programming environments, program control occurs through a combination of objects, properties, events, and methods. For example, Figure 7 shows code written for the "btnScan_Click," an event which is "fired" when the user clicks on the "Scan" button seen in Figure 6. Here, Figure 7 shows that only a small amount of code is needed to first define variables to hold information (the Dim statements), reset the display, and then get a spectrum from the C11118GA "GetSpectrum" method. The remainder of the code shown then ratios the collected spectrum to the DI water background, and calculates the absorbance spectrum from the ratio spectrum.

Figure 7. Calling Hamamatsu's data acquisition in Visual Basic

Once the spectral data has been acquired, the programmer can then manipulate the data as necessary to extract the desired chemical information from the data array. For example, Figure 2 shows the NIR spectra of sugar-sweetened and artificially sweetened beverages (along with water), and Figure 8 shows the programmer’s "cross-check" on the data acquired, where the data has been exported into an Excel format and plotted to verify the program calculation steps are operating correctly. Note that in Figure 8, the peaks located at the X-axis values of 15 and 40 correspond to the peaks shown in Figure 2 at 970 and 1200 nanometers. Thus, the acquired spectrum can now be measured by selecting important peaks (based on pixel number or wavelength), baselines determined, band areas calculated, and converted to meaningful information for presentation to the operator.

Figure 8. Excel cross-check on captured spectra

Once Hamamatsu’s interface package has been defined in a VB project, the developer can then concentrate on program design, flow, data manipulation, and the user interface. For our beverage analyzer, we first want to prompt the operator to collect a blank distilled or deionized (DI) water sample, which will be used as the background reference. This is shown in Figure 9, where all the other dialog box components shown in Figure 6 are automatically hidden until the DI water background is collected.

Figure 9. Initial screen, prompting operator to collect DI water background spectrum

Once the DI water background is collected, the operator can then choose if they believe the beverage is sweetened with sugar(s) or artificial sweetener(s), shown in Figure 10.

Figure 10. Prompt to choose sweetener and start scan

The spectrum of the beverage is then collected and measured based on the selected pixels or wavelengths as shown in Figure 8 and Figure 2. If the measurements show large peaks at these locations, the program can then display a message indicating sugar(s) present (Figure 11), or alternately, if the measurements show minor or no peaks in those locations, the program can display a message indicating no sugar(s) present (Figure 12).

Figure 11. Operator chose sugar(s) and the beverage analyzer also finds sugar(s)

Figure 12. Operator chose sugar(s) but the beverage analyzer detected no sugar(s)

Although the creation of this "beverage analyzer" was not intended to be a commercial product, it does show the relative ease in using a Hamamatsu spectrometer system and interface package. Here, in conjunction with common optical components and simple programming tools, relatively sophisticated UV-Vis-NIR analyzers can be easily designed and implemented.