Overtones in NIR Spectroscopy: The Physics Behind the Peaks

Why near-infrared spectroscopy works at all — and why that's a more interesting question than it first appears.

Near-infrared spectroscopy occupies a peculiar position in analytical science. It is simultaneously one of the most widely deployed techniques in industry — used daily in grain elevators, pharmaceutical production lines, and dairy processing plants — and one of the least intuitively understood. Ask most users why their NIR instrument can predict protein content in wheat or moisture in a tablet blend, and they will likely point to a calibration model built by someone else, running on chemometric software whose inner workings remain opaque.

At the heart of NIR spectroscopy — and at the root of much of its complexity — is a concept borrowed from physics: the overtone. Understanding what overtones are, why they arise, and what they reveal about molecular structure is the key to understanding both the power and the peculiarity of the near-infrared region.

The Basics: Molecular Vibrations and the Infrared

To understand overtones, we must first revisit how molecules interact with infrared radiation in the first place.

Molecules are not static assemblies. Their bonds stretch, bend, twist, and rock continuously, with frequencies determined by the masses of the atoms involved and the stiffness of the bonds connecting them. When infrared light of exactly the right frequency encounters a molecule, it can be absorbed — its energy transferred to the molecule, exciting one of these vibrational modes to a higher energy state.

The classical model of a chemical bond treats it as a harmonic oscillator: a spring connecting two masses, obeying Hooke's Law. In this idealised picture, the vibrational energy levels are perfectly evenly spaced, and a molecule can only absorb radiation at a single, fundamental frequency. This gives rise to the familiar fundamental absorption bands of mid-infrared spectroscopy — the C–H stretch near 3000 cm⁻¹, the carbonyl stretch near 1700 cm⁻¹, the O–H stretch near 3300 cm⁻¹, and so on.

In mid-IR spectroscopy, these fundamentals dominate. They are strong, well-resolved, and structurally diagnostic. The mid-IR region is rightly called the "fingerprint" of chemical structure.

But the harmonic oscillator model is an approximation — and it is in that approximation's failure that NIR spectroscopy finds its reason to exist.

Anharmonicity: Where Overtones Come From

Real chemical bonds do not behave as perfect springs. As two atoms are stretched further apart, the restoring force weakens — the bond is heading toward dissociation, not simply returning to equilibrium. Conversely, pushing the atoms too close together generates strong repulsive forces that stiffen rapidly. The true energy potential of a bond is asymmetric, and it is captured more accurately by the Morse potential than by a simple parabola.

This asymmetry is called anharmonicity, and it has two profound consequences.

First, the vibrational energy levels are no longer evenly spaced. They become progressively closer together as energy increases, until the molecule dissociates.

Second, and more importantly for NIR spectroscopy: the "selection rule" that governs which transitions are allowed is relaxed. In a perfectly harmonic oscillator, only transitions between adjacent energy levels are allowed — the molecule can only jump from v = 0 to v = 1, where v is the vibrational quantum number. This is the fundamental transition.

In an anharmonic oscillator, transitions to v = 2, v = 3, v = 4, and beyond become weakly allowed. These are overtones. The first overtone corresponds to the v = 0 → v = 2 transition, the second overtone to v = 0 → v = 3, and so on. Each successive overtone absorbs radiation at approximately (but not exactly) twice, three times, or four times the frequency of the fundamental, and each is substantially weaker than the one before — typically by one to two orders of magnitude per step.

It is precisely these overtones, along with combination bands (transitions where two different vibrational modes are simultaneously excited), that populate the near-infrared region of the electromagnetic spectrum, roughly 800–2500 nm (or 12500–4000 cm⁻¹).

Which Bonds Contribute to the NIR?

Not all molecular vibrations produce useful NIR absorptions. The strength of an overtone depends on the degree of anharmonicity of the bond involved, and anharmonicity is especially pronounced for bonds involving light atoms — specifically hydrogen.

This is why the NIR region is dominated by overtones and combination bands of bonds containing hydrogen:

O–H stretching overtones — particularly prominent in the 1400 nm and 1900 nm regions. Water absorbs strongly here, which is simultaneously one of NIR spectroscopy's greatest challenges (water ubiquity in biological and food samples) and one of its most useful features (moisture is often exactly what needs to be measured).

N–H stretching overtones — appearing in similar regions to O–H, with the first overtone around 1500 nm. Critical for protein and amino acid analysis.

C–H stretching overtones — the first overtone near 1700 nm, the second overtone near 1200 nm, combination bands in the 2000–2500 nm region. These are the workhorses of NIR analysis in food, agriculture, and petrochemicals, where hydrocarbons and organic molecules dominate.

S–H and P–H — present but weaker and less commonly exploited.

Heavier bonds — C–C, C=O, C–N — are far more harmonic and produce much weaker overtones that are largely buried in the NIR, though their combination bands can contribute, particularly in the longer-wavelength NIR (sometimes called the short-wave infrared, SWIR).

This hydrogen-centric view of NIR spectroscopy explains why it is so effective at analysing organic molecules and aqueous systems, and why it provides little direct information about inorganic materials that lack X–H bonds.

The Structure of the NIR Spectrum

The NIR region is conventionally divided into sub-regions, each populated by different classes of overtones and combination bands:

800–1100 nm (near-NIR or VNIR boundary region) Second and third overtones of C–H, N–H, and O–H stretches. Absorptions here are very weak, which is actually advantageous for samples with high water content or for diffuse reflectance measurements where pathlengths are variable.

1100–1400 nm Second overtones of C–H stretching, combination bands of C–H. Important for quantification of hydrocarbons, oils, and fats.

1400–1500 nm First overtone of O–H stretching (water, alcohols, carbohydrates) and N–H stretching (proteins, amines). This is a region of strong water absorption.

1600–1800 nm First overtone of C–H stretching. One of the richest regions for organic analysis — lipid, carbohydrate, and protein content can all be interrogated here.

1900–2000 nm Combination bands of O–H stretching and bending. Strong water absorption feature used extensively in moisture analysis.

2000–2500 nm Combination bands of C–H, N–H, and C=O. This region contains some of the most structurally informative NIR absorptions, though it requires instruments capable of measuring into the SWIR.

Why NIR Spectra Are So Broad and Overlapping

One immediate consequence of overtone spectroscopy is that NIR spectra look very different from mid-IR spectra. Where mid-IR spectra show sharp, well-resolved peaks that can often be assigned to specific functional groups by visual inspection, NIR spectra are characterised by broad, overlapping, poorly resolved features.

Several factors contribute to this:

Broad intrinsic linewidths. Overtone transitions are inherently broader than fundamentals. The anharmonic energy levels that enable them also mean that each transition is smeared over a range of frequencies, especially in condensed-phase samples where intermolecular interactions (hydrogen bonding above all) broaden and shift peaks further.

Overlap of multiple contributions. The NIR spectrum of even a simple organic molecule contains many overlapping overtone and combination bands from different vibrational modes. In a complex mixture like food, pharmaceutical formulation, or biological tissue, hundreds of such features from different molecules layer on top of each other.

Sensitivity to physical state. Overtone positions and intensities are sensitive to sample temperature, particle size, surface properties, and density in ways that fundamental absorptions in transmission mid-IR are not. This is both a complication and an opportunity — NIR can encode information about physical as well as chemical properties.

The result is that NIR spectra are essentially uninterpretable by visual inspection alone. This is precisely why chemometrics — multivariate calibration and pattern recognition — is indispensable in NIR spectroscopy. A PLS model does not need to know which peak corresponds to which bond. It learns empirically which combinations of spectral variables correlate with the property of interest across a representative set of samples.

Combination Bands: Overtones' Close Relatives

Alongside true overtones, NIR spectra contain combination bands — absorptions that arise when two (or occasionally more) different vibrational modes are simultaneously excited by a single photon. For example, a C–H stretching mode and a C–H bending mode might combine, producing an absorption at approximately the sum of their fundamental frequencies.

Combination bands are governed by similarly relaxed selection rules as overtones, and they also owe their existence to anharmonicity (specifically, to coupling between different vibrational modes — sometimes called Fermi resonance or mechanical coupling). They populate the NIR just as overtones do, and in practice the two are often treated together under the general label of "NIR absorptions," since distinguishing them requires detailed normal-mode analysis.

The interplay of overtones and combination bands gives the NIR spectrum its rich, if congested, information content. Different molecules have subtly different patterns of these absorptions, and it is this molecular fingerprint — however broad and overlapping — that calibration models learn to recognise.

Practical Implications: What Overtone Physics Means for NIR Analysis

Understanding the overtone origin of NIR spectroscopy has several practical consequences for how the technique is applied.

Longer pathlengths are possible. Because overtone absorptions are so much weaker than fundamentals — by factors of 10 to 1000 depending on the order — samples can be measured in much larger quantities without the detector being saturated. This makes NIR ideal for bulk measurements: bales of grain, intact tablets, whole fruit, large powder samples. Mid-IR, by contrast, typically requires thin films or attenuated total reflectance accessories to reduce the effective pathlength.

Non-destructive measurement is routine. The relative transparency of most solid organic materials in the NIR, combined with the availability of diffuse reflectance sampling, allows NIR spectrometers to measure samples without any preparation. Grain can be poured into a cup, tablets can be measured on a conveyor belt, intact fruit can be assessed without cutting.

Water is both problem and opportunity. The strong O–H overtone absorptions mean that water dominates NIR spectra of aqueous samples. This must be accounted for in calibration, but it also means that moisture content — a commercially critical quality parameter in foods, pharmaceuticals, and many industrial products — can be measured rapidly and non-destructively with excellent accuracy.

Calibration is essential and sample-dependent. The overlap and breadth of NIR features means that a calibration model built for one product matrix may transfer poorly to another, even if the analyte is nominally the same. This reflects not just different chemistry, but different overtone environments: the O–H overtone of water in a bread dough behaves differently from the O–H overtone of water in a fresh milk sample, because hydrogen bonding, temperature, and matrix interactions all shift and broaden the bands differently.

Overtones and the Limits of Interpretation

One of the enduring frustrations of NIR spectroscopy is the difficulty of direct spectral interpretation. A skilled mid-IR spectroscopist can often identify functional groups and even specific compounds by inspection. NIR spectra resist this kind of intuitive reading.

Some progress has been made. Second-derivative preprocessing sharpens the broad NIR features and can reveal shoulder peaks invisible in the raw spectrum. Loadings plots from PLS or PCA models can point to spectral regions of greatest analytical importance. Two-dimensional correlation spectroscopy (2D-COS) can help untangle overlapping bands. And density functional theory (DFT) calculations, increasingly practical for complex molecules, can predict overtone and combination band positions to guide interpretation.

But NIR spectroscopy remains predominantly empirical — driven by calibration datasets and validated by prediction performance, not by first-principles spectral assignment. This is a feature as much as it is a limitation: it allows NIR to be deployed without deep spectroscopic expertise, and it means that properties with no obvious spectroscopic origin (texture, hardness, geographic origin) can sometimes be predicted simply because they correlate with chemical features that NIR does capture.

Overtones are the reason NIR spectroscopy exists as a discipline. They arise from the anharmonic reality of chemical bonds — the fact that atoms connected by bonds do not vibrate like ideal springs, and that this imperfection allows transitions beyond the fundamental to occur. These transitions, weak and overlapping, populate the near-infrared region with a dense forest of absorptions that encode, however indirectly, the chemical composition and physical state of whatever is being measured.

The challenge — and the intellectual richness — of NIR spectroscopy lies in extracting that information. The broad, featureless-looking NIR spectrum is not poor-quality mid-IR. It is a different kind of signal entirely, one that requires different tools to interpret and rewards a different set of intuitions.

For the scientist or analyst willing to understand not just how to run a calibration but why the technique works, overtone physics offers a genuinely illuminating starting point. The messiness of the NIR spectrum is not a defect. It is the physics of real molecular bonds, encoded in light.