Spectroscopy is not just spectroscopy—or is it?
When talking with our project partners and customers, the topic of near-infrared (NIR) spectroscopy is often automatically associated with FT-NIR spectroscopy. So, why isn’t it just called NIR? What is the difference between IR and NIR? Some of you might even wonder: “Can I replace an old IR analyzer with NIR hardware?” And additionally: “Why should I replace the IR with a NIR process analyzer?”
This two-part series aims to explain the differences between these techniques and dispel some myths.
Click below to jump directly to a section:
The NIR wavelength range has a long history. As early as the 1880s, organic components were investigated in the NIR range and the strong –OH band relating to the presence of water was discovered as a very important piece of information. Shortly after, measurement of oils from the agricultural industry and investigations into various polymers followed. Some of the first industrial applications of dispersive NIR spectrometers were in the food and agricultural industries. In such applications, parameters including moisture, protein content, and fat content were analyzed quantitatively.
On the other hand, some strong advantages came from using the infrared (IR) wavelength range—high structural sensitivity and specificity—making it possible to obtain precise fingerprints for structural identification.
The hardware for NIR and IR analysis was fundamentally different. At that time, even though the evaluation of NIR spectra seemed to be too difficult and ambiguous due to the broad overlapping peaks, there was one major advantage: robust and cheap materials could be used for NIRS (e.g., PbS detectors, tungsten lamps, and simple glass materials for the optics). Since the NIR-bands were broad and overlapped, users were limited to only the essential information and therefore did not need higher resolution, so simple dispersive gratings (monochromator gratings) were sufficient.
For IR, Fourier transform infrared spectrometers (FT-IR) were used which operated based on Michelson interferometers. This was necessary to obtain the spectral resolution needed for structural interpretation (e.g., distinguishing the isomers 1-propanol and 2-propanol at about 2700 nm for the first time). These spectrometers were introduced to the market in the 1960s. Due to the high costs for interferometers, special optics, and lasers, they were mainly used for research purposes.
Significant progress has been made in the field of spectroscopy due to the development of more powerful computers in the 1980s and 1990s in combination with chemometrics. While IR spectroscopy, due to its high data point density, was still far behind in the application of computer-based chemometric methods, NIR spectrometers were already able to benefit from fast evaluation methods.
Chemometric tools combined with the hardware benefits of NIR technology led many manufacturers to transfer their existing FT-IR measurement technology to the NIR range. And other companies? They just used and improved their already existing measurement technology to achieve perfect synergy between spectrometer and chemometrics.
Now that some of the background has been laid out, it’s time to answer some myths about NIR, FT-NIR, and FT-IR spectroscopy.
This is a persistent myth that you could easily miss unless you look closely. What exactly does the «FT» mean and why doesn’t everyone use it to describe NIR spectroscopy?
When using a FT-NIR spectrometer, first an interferogram is generated—not a spectrum in that sense. The conversion of the interferogram into a spectrum is done by applying a mathematical operation, the Fourier transform (FT). This transforms the path-dependent information (e.g., relative mirror position of two mirrors in the spectrometer) into a frequency-dependent function. This means FT-NIR is nothing more than the methodology of generating the spectrum in the NIR wavelength range.
As before, it is NIR spectroscopy that provides the same information as dispersive NIR spectroscopy or diode array spectroscopy. FT-NIR uses the interferences produced by an interferometer to extract single wavelengths from white light (halogen lamp), while dispersive spectrometers use gratings. Gratings are produced by very modern lithographic techniques and offer the highest precision (of wavelength accuracy).
||Mathematical calculation (Fourier transformation) from the phase shift of two incident light beams (interferogram)||Diffraction or dispersion, movement by a digital encoder|
|Resolution depends on…||Setting the maximum offset of the moving mirror||Number of lines of the monochromator grid, slit width, encoder quality|
|Moving elements||Yes (motor of the interference mirror)
||Yes (motor of the grid)|
12500 cm-1–4000 cm-1
(expandable to 400 nm)
|Noise||Depending on the resolution, higher than dispersive NIR with comparable setup||Depending on the resolution, lower than FT-NIR with comparable setup|
|Method transferability (i.e., to other spectrometers)||Yes||Yes (due to the Metrohm calibration concept)|
|Consumables||Laser source, halogen lamp, and desiccant||Halogen lamp|
Compared to a dispersive analyzer, a FT-NIR spectrometer uses a laser to control the position of the interferometer mirror. This laser must be changed periodically, however this task is generally not done by the end-user themselves compared to halogen lamp, which is easily replaceable.
Looking a little bit deeper into the details of Table 1, it is clear that dispersive spectroscopic instruments are more suitable for industrial process applications. Why? A low acquisition time is critical for real-time measurements with the least time loss. Compared to a FT-NIR instrument, the acquisition time is lower for a dispersive analyzer (leading to faster results) at the same resolution.
If you have ever wondered why you might need to calculate a Fourier transform, you will know after the next myth is answered.
Myth 2: Method transfer is only possible with FT-NIR spectrometers and not with dispersive spectrometers
Where does this myth originate?
Visualize the internal structure of an interferometer. A He-Ne laser is used as the reference measurement for the precise determination of the mirror position and thus also obtains an exact spectrum with high wavelength reproducibility at the same spatial coordinate by the Fourier transformation.
What is different about the dispersive spectrum?
In this case, the spectrum is not calculated in a mathematically complex way but is recorded directly via the dispersing element (the monochromator) on the detector. Here, a high-resolution state of the art grating and an accurate digital encoder that is precisely matched with the detector play an important role. NIRS XDS Process Analyzers from Metrohm Process Analytics (Figure 2) use wavelength standards to achieve both high wavelength precision and reproducibility and to ensure the transferability of the developed method.
||Precision <0.015 nm||Valid|
|975.880 nm||0.0012 nm||Yes|
|1221.342 nm||0.0005 nm||Yes|
Figure 3. Results of the wavelength reproducibility test of a NIRS XDS Process Analyzer during a performance test. The precision meets the very narrowly defined testing specifications.
By using a wavelength and reference standard built into the spectrometer, additional diagnostics can be carried out at routine intervals (either as part of maintenance or automated within regular process operation) to check the wavelength accuracy and precision.
Due to the standardization concept with reference standards and wavelength standards, methods can be transferred without much effort even when changing or adjusting accessories (e.g., longer optical fibers, partially changed probes).
To summarize: a good standardization concept with NIST certified reference and wavelength standards as well as internally installed standards allow robust method transfers to other spectrometers and excellent long-term stability in the production process.
Myth 3: Many applications cannot be measured with dispersive NIRS, but require well-resolved FT-NIR spectroscopy
The Michelson interferometer and the monochromator grating were both developed in the 1800s. Both of these technologies have been used industrially since the advancement of computer technology and utilize the same light sources, detectors, optical fibers, and probes.
Monochromator gratings now consist of, e.g., a holographic concave diffraction grating with an optimized image plane to avoid aberrations and stray light. Holographic gratings are created by etching interference lines via laser into a photoresist layer. An advantage of this is very high spectral resolution, which together with a detailed adjustable encoder (and other components of the monochromator), provides very good resolution with the NIR spectrometer. For example, the NIRS XDS Process Analyzer (Figure 2) has a real resolution of 8.75 nm.
In comparison, higher resolutions can be achieved with interferometers, but this can also decrease the signal-to-noise (S/N) ratio. Usually, resolutions of approximately 8 cm-1 or 16 cm-1 are used, which corresponds to 10–25 nm at 2500 nm.
||Dispersive NIR (Metrohm)|
|Wavelength precision (nm)||~0.01||~0.005|
|Wavelength accuracy (nm)||~0.05–0.2||~0.05|
Figure 4. Noise spectra recorded with a Metrohm dispersive NIRS analyzer and a typical FT-NIR spectrometer.
Higher resolutions are usually not required for the majority of applications since harmonics/overtones and combination bands of pure substances in the NIR wavelength range have a broad bandwidth. The absorption peak with the smallest bandwidth currently known in the NIR region is talc at slightly more than 10 nm.
Very similar overlapping information (e.g., –OH bands or –COOH bands) are separated by using chemometric methods and are evaluated individually and specifically.
Another example that shows how powerful dispersive NIR spectroscopy is compared to FT techniques and IR spectroscopy can be seen in the separation of xylene isomers in a mixture of several aromatics/hydrocarbons (Figure 5).
Figure 5 shows that the three xylene isomers can be clearly distinguished spectroscopically. Applying chemometrics elaborates the information even more and ultimately all six components can be determined individually and quantitatively. In a production process, the real-time reaction monitoring can be performed for all six components (Figure 6).
The application shown in Figures 5 and 6 has often been implemented with IR photometers in the past. We have now shown that the application can not only be transferred to the NIR wavelength range, but that even minimal structural differences between the functional groups of molecules can be detected by dispersive NIR spectroscopy.
To summarize: dispersive spectrometers have very good spectral resolution—in some cases better than FT-NIR spectrometers—and can even distinguish between different isomers in complex mixtures, or between very similar components like –OH and –COOH functional groups.
In the first part of this series, we went more into detail about the practical differences between FT-NIR and dispersive NIR spectroscopy. Three myths were discussed: that NIR spectroscopy always means FT-NIR—False, method transfer is only possible with FT-NIR spectrometers and not with dispersive spectrometers—False, and many applications cannot be measured with dispersive NIRS, but require well-resolved FT-NIR spectroscopy—False.
Some myths should no longer be kept alive because they are not facts!
We have also compared FT-NIR spectrometers to dispersive spectrometers when used in a process environment. Some critical points to remember: the dispersive analyzer is less sensitive to vibrations; less maintenance is necessary, and the grating is post-dispersive and therefore less prone to pollution due to lower ambient light.
To prove those arguments, Part 2 will show that, contrary to some expectations, it is possible to replace IR measurement techniques in industrial processes with easy-to-implement NIR measurement techniques. We will dispel some more myths and go into detail of an extremely low water content measurement in a process with the support of our primary analysis methods.