— Back to Blogs & Publications FT-NIR Spectroscopy Guide for Food and Feed Manufacturers

What Is FT-NIR Spectroscopy? A Plain-Language Guide for Food and Feed Manufacturers

A plain-language introduction to Fourier-transform near-infrared (FT-NIR) spectroscopy for food and feed: how a Michelson interferometer and a Fourier transform produce a spectrum, what the near-infrared bands reveal about water, protein, fat and carbohydrate, and why every result rests on a calibration against reference methods — based on the spectroscopy literature and the ISO guidelines.

Near-infrared light as an analytical tool

Near-infrared (NIR) spectroscopy is widely used for compositional analysis in the food and feed industry. Pasquini defines the NIR region as 750 to 2,500 nm (13,300 to 4,000 cm−1) and describes the technique as fast, non-destructive and non-invasive, with minimal sample preparation and nearly universal application to molecules containing C–H, N–H, S–H or O–H bonds [1]. A basic introduction to the technique is given in What Is NIR Spectroscopy?; this guide concentrates on the Fourier-transform (FT) variant and on what its spectra can and cannot tell a food or feed producer.

How an FT-NIR spectrometer records a spectrum

Benchtop laboratory NIR spectrometers have converged on a fairly uniform design in which a Michelson or a polarization interferometer is used and the spectrum is obtained by the Fourier-transform principle [2]. In a Michelson interferometer the beam is divided by a beamsplitter between a fixed mirror and a moving mirror; when the two beams recombine, their interference depends on the optical path difference set by the mirror position, and the theory of how this produces an interferogram, and how finite mirror travel limits resolution, is treated in detail in the standard text by Griffiths and de Haseth [3].

The key point for a non-specialist is that the whole broadband beam reaches the detector at once. As Beć et al. put it, the detector registers an interferogram, a signal in the time domain, and the spectrum — the signal in the frequency domain — is reconstructed from it by a Fourier transform [2]. In other words, the instrument does not measure one wavelength after another; it measures a mixture of all of them and recovers the individual intensities mathematically [1][2]. Beć et al. name the gain in optical throughput and a precisely controlled wavelength axis as the primary benefits of this approach [2], and Pasquini attributes to FT instruments the best wavelength precision and accuracy among NIR designs and a throughput (Jacquinot) gain from the absence of entrance and exit slits [1].

These benefits come with trade-offs. Pasquini notes that sub-nanometre resolution is obtained at the cost of scan speed and that an interferometer with a moving mirror is not as robust as designs without moving parts [1]; Beć et al. show that when the Michelson interferometer is miniaturised it is no longer far superior to other solutions [2]. FT-NIR is therefore one of several valid instrument architectures rather than a universally better one; the comparison with grating and diode-array instruments is covered in NIR vs FT-NIR.

What the bands say about water, protein, fat and carbohydrate

NIR absorptions are overtones and combinations of fundamental molecular vibrations, mainly of C–H, O–H and N–H groups; they are broad, superimposed and weak, between about 10 and 1,000 times less intense than the corresponding fundamental absorptions [1]. The weak absorption goes together with the high penetration of the probing beam that Pasquini lists among the technique's characteristics, but the overlap means that individual constituents rarely appear as isolated peaks [1][2]. The table lists typical band positions for the main food constituents.

Approx. wavelength (nm) Vibration Typical constituent Source
1440 and 1930 O–H (water) Moisture [4]
1210 C–H second overtone Carbohydrates, lipids [2]
1435, 1500, 1565 N–H first overtone Proteins [2]
1480 O–H first overtone Carbohydrates [2]
1610–1725 C–H first overtone Carbohydrates, lipids [2]
2050, 2155, 2295 N–H and amide combinations Proteins [2]
2255 O–H stretch + deformation Carbohydrates [2]

Water dominates: Büning-Pfaue describes two broad water peaks near 1440 and 1930 nm in nearly every NIR spectrum and shows that hydrogen bonding and sample temperature affect the reliability of NIR moisture results [4]. Beć et al. stress that such tables are approximate and that identifying specific markers in NIR spectra is much less straightforward than in mid-infrared or Raman spectroscopy [2]. This is why NIR results are not read off a single peak.

Why calibration against reference methods is needed

Because the bands overlap, NIR relies on a multivariate model to quantify a property or a concentration in complex samples, and Pasquini stresses that the technique is and will always be heavily dependent on good reference methods, with the success of a model greatly dependent on the quality of the reference values in the training set [1]. The international guidelines formalise this. ISO 12099 covers the determination of constituents such as moisture, fat, protein, starch and crude fibre in animal feeding stuffs, cereals and milled cereal products [5]; ISO 21543 | IDF 201 does the same for milk and milk products in liquid, semi-solid or solid form, measured in transmittance, diffuse reflectance or transflectance [6]. In practice three points matter:

  • The reference method sets the ceiling. An NIR prediction cannot be more accurate than the laboratory values it was calibrated on, so reference analyses should follow the official procedures [1][5].
  • Matrix and presentation matter. Sample form and measurement mode determine the instrument set-up [6], and for moisture the effects of temperature and hydrogen bonding have to be represented in the calibration set [4].
  • Chemometrics must be used with care. Pasquini's 2018 review describes NIR as a mature technique with renewed interest driven by instrumentation and chemometrics, but also addresses the misuse of its mathematical tools [7]; independent validation guards against over-fitted models.

How USTECH approaches FT-NIR

USTECH uses the FT-NIR principle in the ProLine2550 process analyzer for blenders, reactors and pipelines; for grains, feed ingredients and forage at the bench or at-line it offers the MasterLine17EC NIR analyzer. Calibrations are developed and validated against reference analyses in caliX, following the approach of ISO 12099 and ISO 21543 | IDF 201 [5][6]. Which parameters an FT-NIR instrument can measure reliably in a given product is established in a feasibility study with your own samples and reference analyses.

References

  • [1] Pasquini, C. (2003). Near infrared spectroscopy: fundamentals, practical aspects and analytical applications. Journal of the Brazilian Chemical Society, 14(2), 198–219. doi:10.1590/S0103-50532003000200006
  • [2] Beć, K.B., Grabska, J., Huck, C.W. (2022). Miniaturized NIR Spectroscopy in Food Analysis and Quality Control: Promises, Challenges, and Perspectives. Foods, 11(10), 1465. doi:10.3390/foods11101465
  • [3] Griffiths, P.R., de Haseth, J.A. (2007). Fourier Transform Infrared Spectrometry, 2nd ed. Wiley-Interscience, Hoboken. doi:10.1002/047010631X
  • [4] Büning-Pfaue, H. (2003). Analysis of water in food by near infrared spectroscopy. Food Chemistry, 82(1), 107–115. doi:10.1016/S0308-8146(02)00583-6
  • [5] ISO 12099:2017. Animal feeding stuffs, cereals and milled cereal products — Guidelines for the application of near infrared spectrometry. ISO, Geneva. iso.org/standard/67352.html
  • [6] ISO 21543 | IDF 201:2020. Milk and milk products — Guidelines for the application of near infrared spectrometry. ISO, Geneva. iso.org/standard/77606.html
  • [7] Pasquini, C. (2018). Near infrared spectroscopy: A mature analytical technique with new perspectives – A review. Analytica Chimica Acta, 1026, 8–36. doi:10.1016/j.aca.2018.04.004
USTECH Innovations Logo

Discuss this application for your process

Our application team can walk through the method, share the supporting literature and discuss how it applies to your process.

Contact Us

Read Next

NIR vs FT-NIR: What's the Difference
GUIDE

NIR vs FT-NIR: What’s the Difference

How scanning-grating, diode-array and Fourier-transform NIR analyzers produce a spectrum, what the throughput, multiplex and wavelength-axis advantages really buy, and which trade-offs decide an inline application.

Read Guide
Inline vs. At-Line vs. Off-Line Analysis
GUIDE

Inline vs. At-Line vs. Off-Line Analysis

How the FDA PAT guidance defines at-line, on-line and in-line measurement, what each location trades away in delay, sampling error and validation effort, and how to choose the arrangement a production line actually needs.

Read Guide