GUIDE

How to interpret an infrared spectrum (FTIR): principles and method

Updated on 10/25/2026 · espectroFTIR

Interpreting an infrared spectrum is reasoning from evidence: each band constrains compatible structures, and the pattern, together with sample and technique context, supports a hypothesis. This guide brings together the physical principles, acquisition problems and an interpretation method.

In brief
  • One band alone identifies nothing: the combined evidence matters.
  • Check the data (technique, Y-axis, noise, saturation, CO₂ and water vapor) before assigning any band.
  • Follow the method: observe, assign, seek corroboration and assess consistency.
  • An expected but undetected band is evidence, but “not detected” does not prove the group is absent.

1. Three clues from each band

Lesson 1: What is spectroscopy?Lesson 2: What is light?Molecular vibrations

Each band offers three clues. Its position indicates which bond vibrates, depending on bond strength and atomic masses. Its intensity shows how much the vibration changes molecular polarity and how much material is in the light path; this is why C=O, O–H, N–H and C–O dominate the spectrum. Its width reflects the environment: a narrow band suggests a homogeneous environment, while a broad band can reflect hydrogen bonding or multiple environments.

Explore further: equations behind the three clues

In the harmonic model, ν̃ = (1/2πc)√(k/μ), where k is the force constant and μ the reduced mass. Intensity is proportional to (∂μ/∂Q)², the square of the dipole-moment change with vibration, and to the amount of material in the optical path (A = εcl).

2. Data quality before interpretation

Before assigning bands, check whether the spectrum permits reliable interpretation:

3. The spectral regions and the information each provides

Region (cm⁻¹)Physical originTypical bands
4000–2500X–H stretching: small reduced mass and modes decoupled from the molecular skeletonFree O–H ≈ 3600; associated O–H 3550–3200 (broad); N–H 3500–3300; aromatic or alkene C–H > 3000; aliphatic C–H 2960–2850; aldehyde 2820 and 2720 (Fermi resonance)
2500–2000triple and cumulated bonds: high force constantsC≡N 2260–2210; C≡C 2260–2100; azides, isocyanates and isothiocyanates 2300–2000 (CO₂ at 2349 is a common artifact)
2000–1500double bonds and N–H bendingC=O 1850–1630 (position helps distinguish classes); C=C 1680–1620; aromatic ring ≈ 1600, 1580 and 1500, with weak overtones between 2000 and 1650 whose pattern depends on substitution; amide II 1570–1510
1500–650fingerprint: strongly coupled C–C, C–O and C–N stretching and bending modesδ CH₂/CH₃ 1470–1370; C–O and C–N 1300–1000; out-of-plane aromatic C–H 900–690; C–Cl 800–600

Boundaries are conventional and vary between authors and between this guide and the operational report windows; no region has rigid boundaries.

4. An interpretation method based on evidence

  1. Observation. List dominant bands by position, relative intensity, width and shape, without assigning them yet.
  2. Provisional assignment. For each band, list compatible groups, including alternatives: a broad band above 3100 cm⁻¹ may be alcohol, acid or water O–H, or N–H.
  3. Corroborating evidence. Each group implies other bands: ester (C=O and two C–O bands), acid (broad O–H, C=O and a band near 930 cm⁻¹), amide (amide I, amide II and N–H). An assignment without corroboration is a weak hypothesis.
  4. Negative evidence. Expected but undetected bands reduce support for a hypothesis under the conditions and within the region evaluated; they do not exclude it when overlap, low concentration or saturation is present.
  5. Overall consistency. The hypothesis should explain all relevant bands; unexplained bands indicate another component (mixture, additive, filler or water) or an artifact.
  6. External verification. Compare with references measured using the same technique and, for stronger conclusions, with an orthogonal technique.

Guiding questions for a first pass: is there a broad band above 3100 cm⁻¹ (O–H or N–H)? Is there C–H above and/or below 3000 cm⁻¹? Is there a carbonyl between 1850 and 1630 cm⁻¹ (see the carbonyl guide)? Is there C=C or an aromatic ring? What occurs between 1300 and 1000 cm⁻¹ (C–O, C–N; broad, intense bands without C–H suggest inorganic materials such as silicates, sulfates and phosphates)?

5. Fermi resonance, overtones and combinations

Not every band arises from one fundamental vibration. Overtones and combinations are weak but can gain intensity through interaction with a nearby fundamental: Fermi resonance. Classic examples include the aldehyde C–H doublet (2820 and 2720 cm⁻¹) and the weak aromatic pattern between 2000 and 1650 cm⁻¹. Recognizing these bands avoids spurious assignments and strengthens the hypothesis.

Explore further: the origin of Fermi resonance

Not every band is a fundamental. Overtones (≈ 2ν) and combinations (ν₁ + ν₂) are weak but can gain intensity through interaction with a fundamental of the same symmetry and nearby energy (Fermi resonance). Examples include the aldehyde C–H doublet at 2820 and 2720 cm⁻¹ (interaction of C–H stretching with a bending overtone at ≈ 1390 cm⁻¹), the weak aromatic pattern at 2000–1650 cm⁻¹, and amine-salt combination envelopes (see the hydrochloride guide). Ignoring these bands leads to spurious assignments; recognizing them strengthens the hypothesis.

6. When a band is absent

When a band does not appear, the appropriate statement is that it was not detected under the conditions and within the region evaluated. Absence of a carbonyl band reduces support for acid, ester or ketone hypotheses but is not sufficient to exclude them: low concentration, overlap and spectral quality also affect detection.

7. From functional group to substance: limits of the conclusion

The spectrum alone describes compatible functional groups and compound classes. Comparison with a reference measured using the same technique strengthens the hypothesis but does not by itself confirm identity. Always record technique, resolution, scan count and instrument. The analyst is responsible for the conclusion, supported when necessary by an orthogonal technique (NMR, mass spectrometry or diffraction).

Explore further: why numerical correlation does not prove identity

Similarity indicators such as correlation measure shape similarity. Distinct but closely related substances may have highly correlated spectra, while polymorphs, hydrates and mixtures can differ from the reference. Mixtures, additives and the matrix change the spectrum.

8. Following a reaction or a change through the spectrum

The spectrum also serves to follow a transformation (synthesis, curing, hydrolysis, oxidation, degradation). Instead of identifying the sample, watch what disappears, what appears and what shifts from one measurement to the next.

Quick summary
  • Disappears: the band of a reagent group (for example, N=C=O at ≈ 2270 cm⁻¹ in an isocyanate).
  • Appears: the band of a product group (for example, ester C=O at ≈ 1740 cm⁻¹).
  • Shifts: the same band changes position when the environment changes (conjugation, hydrogen bonding, salt formation), without the group disappearing.
  • Stays the same: an internal band that does not take part in the reaction (for example, the aromatic ring at ≈ 1600 cm⁻¹) serves as a reference to compare the others.
TransformationDisappears (cm⁻¹)Appears (cm⁻¹)
Esterification (acid + alcohol)Broad acid O–H (3300–2500) and acid C=O (≈ 1710)Ester C=O (≈ 1740) and C–O at 1300–1150
Oxidation of an alcohol to a ketoneAlcohol O–H (≈ 3350, broad) and C–O (≈ 1050)Ketone C=O (≈ 1715)
Neutralization of a carboxylic acid (salt)Acid C=O (≈ 1710) and broad O–HCarboxylate: 1610–1550 and 1440–1380
Nitrile hydrolysisC≡N (≈ 2250)Amide or acid C=O, with O–H or N–H
Polyurethane curingN=C=O (≈ 2270)N–H (≈ 3330), urethane C=O (≈ 1730–1700) and amide II (≈ 1530)
Epoxy curingepoxy ring (≈ 915)Broad O–H (≈ 3400)
Acrylate or vinyl polymerizationC=C (≈ 1640) and =C–H above 3000Stronger aliphatic C–H of the chain (2960–2850)

A simple method:

  1. Measure all samples with the same technique, the same accessory and the same conditions (in ATR, pressure and contact change the intensity).
  2. Choose a reference band that does not take part in the reaction and divide the band that changes by it.
  3. Follow this ratio (height or area) over time. The curve is relative and semi-quantitative, not an absolute conversion.
  4. Overlay the spectra and check that the changes go in the same direction. Residual solvent, water and CO₂ also change the spectrum between measurements.
  5. When the result is critical, confirm it with an orthogonal technique.

It is a supporting spectral screening: it does not prove the conversion or the identity of the product.

References

  1. Silverstein, R. M.; Webster, F. X.; Kiemle, D. J.; Bryce, D. L. Spectrometric Identification of Organic Compounds. 8th ed. Hoboken: Wiley, 2014.
  2. Pavia, D. L.; Lampman, G. M.; Kriz, G. S.; Vyvyan, J. R. Introduction to Spectroscopy. 5th ed. Stamford: Cengage, 2015.
  3. Smith, B. C. Infrared Spectral Interpretation: A Systematic Approach. Boca Raton: CRC Press, 1999.
  4. Socrates, G. Infrared and Raman Characteristic Group Frequencies: Tables and Charts. 3rd ed. Chichester: Wiley, 2001.
  5. Griffiths, P. R.; de Haseth, J. A. Fourier Transform Infrared Spectrometry. 2nd ed. Hoboken: Wiley-Interscience, 2007.
Have a spectrum ready? Upload the file (or an image of the spectrum plot) and view the free preview, with the full curve and a sample of bands, on the page and in PDF. Eligible new accounts receive 1 welcome credit to try a full report in PDF. Word and supplementary files are included in analyses using paid credits.

Analyze my spectrumFrequency tableOther guides

Frequently asked questions

How many regions does an infrared spectrum have?

The most common division uses four conventional regions: 4000–2500 (O–H, N–H, C–H), 2500–2000 (triple and cumulated bonds), 2000–1500 (carbonyl and C=C), and 1500–650 (fingerprint). Boundaries vary among authors.

Can a substance be identified from its spectrum alone?

The spectrum shows compatible functional groups and compound classes but does not by itself confirm identity. Comparison with a reference measured using the same technique helps assess similarity; the analyst is responsible for the final conclusion.

What is the difference between transmittance and absorbance?

In transmittance (%T), bands appear as valleys; in absorbance, as peaks. Absorbance A = −log₁₀ T is proportional to concentration under ideal conditions, according to the Beer–Lambert law.

Why do polar-bond bands dominate the spectrum?

Because intensity is proportional to the squared change in dipole moment with vibration, (∂μ/∂Q)². Bonds such as C=O, O–H and N–H cause large dipole changes; nonpolar or symmetric bonds change it less and give weak bands.

How do I follow a reaction by FTIR?

Measure each sample under the same conditions and watch what disappears (reagent), what appears (product) and what shifts. Compare the ratio between the changing band and a reference band that does not take part in the reaction. The result is relative and semi-quantitative.

Ranges are approximate and vary with technique, physical state, matrix and instrument. This guide describes functional groups and compatible classes; it does not identify substances.