Infrared (IR) spectroscopy identifies functional groups by the frequencies at which molecular bonds absorb infrared light and vibrate — stretch, bend, or twist. A bond absorbs only if the vibration changes the molecule’s dipole moment, and it absorbs at a characteristic wavenumber that depends mainly on the two atoms’ masses and the bond’s stiffness (its force constant). Because C=O always absorbs in roughly the same region regardless of what molecule it sits in, a spectrum can be read as a list of functional groups present, well before any full structure is assigned. This page gives the correlation table for doing that, plus a worked band-by-band walkthrough and the sampling and instrument choices that determine whether the spectrum you get is actually readable.
How to read an IR spectrum
An IR spectrum plots wavenumber (cm-1) on the x-axis, conventionally decreasing from left to right (4000 cm-1 down to 600-400 cm-1), against either %Transmittance or Absorbance on the y-axis. On a %Transmittance plot, absorption bands point downward — a bond that absorbs strongly shows as a deep dip toward 0%T, while a region with no absorption sits near 100%T. Absorbance plots invert this (peaks point up) and are generally preferred for quantitative work because absorbance is linear with concentration (Beer-Lambert), whereas transmittance is not.
The spectrum splits into two working regions:
- Group frequency region, roughly 4000-1500 cm-1. Bands here are reasonably transferable between molecules — an O-H stretch looks like an O-H stretch whether it’s in ethanol or in a steroid. This is the region the correlation table below is built from, and it’s where you do functional-group identification.
- Fingerprint region, roughly 1500-600 cm-1. Bands here arise from complex, coupled vibrations of the whole carbon skeleton and are highly specific to the exact molecule. Two different compounds with identical functional groups will still usually have distinguishable fingerprint regions. This region is not generally assigned bond-by-bond; it’s used for whole-spectrum comparison against a reference library or a known standard.
Three properties of a band carry information beyond its position: intensity (strong/medium/weak, driven by how much the dipole moment changes — C=O and C-O change it a lot and absorb strongly; C=C and C-H change it little and absorb weakly), shape (a sharp, well-defined band usually means a “free,” non-hydrogen-bonded group; a broad band usually means hydrogen bonding is spreading the vibration over a range of local environments), and multiplicity (a doublet where you expect a singlet, for example, often indicates Fermi resonance or two non-equivalent bonds of the same type, such as the asymmetric and symmetric N-H stretches of a primary amine).
IR functional group frequency table
Ranges below reflect the values commonly reported in organic-chemistry and analytical-chemistry references (e.g., Pavia/Lampman/Kriz-style correlation tables and the Sigma-Aldrich/Coblentz Society IR charts). Use them as ranges, not fixed numbers — the exact position within a range shifts with conjugation, ring strain, hydrogen bonding, and neighboring electron-withdrawing or -donating groups.
| Bond / group | Range (cm-1) | Intensity | Shape / notes |
|---|---|---|---|
| O-H, alcohol/phenol (free) | 3580-3650 | Variable, sharp | Only seen as a sharp band in dilute non-polar solution; in neat/concentrated samples it’s swamped by the H-bonded band below |
| O-H, alcohol/phenol (H-bonded) | 3200-3550 | Strong | Broad, rounded — the default appearance for neat liquids and solids |
| O-H, carboxylic acid | 2500-3300 | Weak-medium | Very broad, often described as a “sawtooth” that overlaps and roughens the C-H stretches beneath it — this breadth alone is diagnostic of a -COOH |
| N-H, primary amine | 3300-3500 | Medium | Two bands (asymmetric + symmetric stretch); distinguishes primary from secondary amines |
| N-H, secondary amine / amide (A-band) | 3300-3400 | Medium | Single band |
| Alkyne C-H (≡C-H) | ~3300 | Strong, sharp | Narrow — distinguishes a terminal alkyne from vinyl/aromatic C-H just below it |
| Aromatic / vinyl C-H (=C-H) | 3000-3100 | Medium | Just above 3000 cm-1 — the classic cutoff used to separate sp2 from sp3 C-H |
| Alkane C-H (sp3) | 2850-2980 | Medium-strong | Just below 3000 cm-1; usually 2-4 overlapping bands (asym/sym CH3 and CH2) |
| Aldehyde C-H | 2700-2900 | Weak | Characteristic Fermi-resonance doublet near 2720 and 2850 cm-1, on the low-wavenumber shoulder of the alkane C-H envelope |
| Alkyne C≡C | 2100-2260 | Weak-medium | Absent or very weak if the alkyne is internal and symmetrically substituted (no dipole change) |
| Nitrile C≡N | 2210-2260 | Medium | Sharp; sits in a generally “quiet” region of the spectrum, so even a weak band here is easy to spot |
| Acid chloride C=O | 1790-1815 | Strong | Highest carbonyl frequency of the common acyl derivatives |
| Anhydride C=O | 1740-1830 | Strong | Two bands (symmetric + asymmetric C=O stretch), roughly 60-70 cm-1 apart |
| Ester C=O | 1735-1750 | Strong | Higher than ketone/aldehyde due to the adjacent C-O |
| Aldehyde C=O | 1720-1740 | Strong | Paired with the 2700-2900 cm-1 C-H doublet, which is what distinguishes it from a ketone |
| Ketone C=O | 1705-1725 | Strong | Conjugation with a ring or double bond lowers this by 20-40 cm-1 |
| Carboxylic acid C=O | 1700-1725 | Strong | Often broadened/lowered a little by dimerization via H-bonding |
| Amide C=O (Amide I) | 1630-1700 | Strong | Lowest carbonyl frequency of the group, because of amide resonance delocalizing the C=O bond order |
| C=C, alkene | 1620-1680 | Weak-medium | Weak or absent for symmetrically substituted (e.g., trans-disubstituted, tetrasubstituted) alkenes |
| C=N, imine/oxime | 1640-1690 | Medium | Overlaps the alkene and amide region; context (presence of N-H, absence of C=O) helps distinguish it |
| N-H bend (amide II / amine scissoring) | 1550-1650 | Medium | Secondary marker for N-H-containing groups, used alongside the N-H stretch above 3300 cm-1 |
| Aromatic ring C=C | 1450-1600 | Variable | Typically 2-4 bands, commonly near 1600, 1580, 1500 and 1450 cm-1; intensity depends on ring substitution |
| Nitro group, asymmetric (N=O) | 1500-1550 | Strong | Paired with the symmetric stretch below — the two-band pattern is itself diagnostic of -NO2 |
| Nitro group, symmetric (N=O) | 1345-1385 | Strong | See above |
| C-O, alcohols/esters/ethers | 1000-1300 | Strong | Position within the range depends on whether the C is primary, secondary, tertiary, or aromatic/vinylic |
| C-N, amines | 1020-1250 | Medium | Weaker and less diagnostic on its own than the N-H stretch/bend pair |
| C-F | 1000-1400 | Strong | Multiple C-F bonds (e.g., -CF3) shift and multiply this band |
| S=O, sulfoxide | 1030-1070 | Strong | |
| S=O, sulfone | 1120-1160 and 1300-1350 | Strong | Two bands, analogous to the nitro pair |
| =C-H out-of-plane bend, alkene | 650-1000 | Strong | Fingerprint-region band whose exact position indicates cis/trans and substitution pattern |
| Aromatic C-H out-of-plane bend | 675-900 | Strong | Pattern (number and position of bands) indicates the ring’s substitution pattern (mono-, ortho-, meta-, para-) |
| C-Cl | 600-800 | Strong |
Worked example: reading an aspirin (acetylsalicylic acid) spectrum
Illustrative walkthrough: the values below are the typical, widely-published band positions for acetylsalicylic acid, used here only to demonstrate the method of going from a spectrum to a set of functional-group assignments — not a specific proprietary run. Aspirin is a convenient teaching molecule because it carries three distinct, well-separated functional groups on one small structure: a carboxylic acid, an ester, and an aromatic ring.
- ~2500-3300 cm-1, broad, weak-medium: the carboxylic acid O-H, overlapping and roughening the aromatic/aliphatic C-H stretches in the same window. Its unusual breadth (rather than a discrete sharp O-H) is the first clue that a -COOH, not a simple alcohol, is present.
- ~1750 cm-1, strong, sharp: the ester C=O stretch (the acetoxy group, -O-C(=O)-CH3).
- ~1690 cm-1, strong: the carboxylic acid C=O stretch. Two separate strong carbonyl bands this close together, at these two positions, is itself consistent with “one ester carbonyl plus one acid carbonyl” rather than two esters or two acids.
- ~1600, 1580, 1490 cm-1, medium, several bands: aromatic ring C=C stretches, confirming a benzene ring.
- ~1180-1300 cm-1, strong: C-O stretches associated with the ester linkage.
- ~700-750 cm-1, strong: aromatic C-H out-of-plane bending, consistent with the ring’s ortho-disubstitution pattern.
Put together, that band list — broad acid O-H, two carbonyls at two different characteristic positions, aromatic C=C, and ester C-O — reconstructs the molecule’s key functional groups (aromatic carboxylic acid + aromatic ester) directly from the correlation table above, without needing the fingerprint region at all. The fingerprint region (600-1500 cm-1, excluding the assigned C-O and C-H bands) would still be used to confirm identity against a reference spectrum, since it’s unique to acetylsalicylic acid specifically.
Sampling techniques compared
The functional groups present don’t change with sampling method, but band shape, resolution, and how much sample prep you need do. Choosing the wrong technique for a given sample is a common source of a spectrum that looks “wrong” when the chemistry is fine.
| Technique | Sample type | Prep required | Trade-offs |
|---|---|---|---|
| ATR (attenuated total reflectance) | Solids, liquids, pastes, powders — minimal restriction | Place directly on the crystal, apply pressure; little to no prep | Fastest and most common routine method; band intensities and slightly the band shapes differ from transmission spectra (ATR correction algorithms adjust for this) and penetration depth is only ~0.5-2 µm, so poor crystal contact on a rough or hard solid weakens the whole spectrum |
| Transmission (KBr pellet) | Solids | Grind sample with dry KBr, press into a pellet | Gives spectra closest to literature reference values; KBr is hygroscopic, so residual moisture introduces a broad O-H band that can mask or be mistaken for the analyte’s own O-H |
| Nujol mull | Solids | Grind sample into a mineral-oil paste between salt plates | Avoids KBr’s moisture problem, but the mineral oil itself contributes strong C-H bands (~2850-2950, ~1460, ~1375 cm-1) that obscure the sample’s own C-H region |
| Transmission (liquid cell) | Liquids | Thin film or fixed-pathlength cell between salt windows | Good for neat liquids; pathlength must be matched to concentration or strong bands saturate (go off-scale) |
| DRIFTS (diffuse reflectance) | Powders, rough solids, surfaces | Diluted in a non-absorbing matrix (e.g., KBr powder) or measured neat | Useful for samples that can’t be pressed into a pellet or pass light easily; requires a Kubelka-Munk conversion for quantitative comparison to transmission spectra |
Instrument parameters that affect what you see
- Resolution. Routine FTIR work is typically run at 4 cm-1 resolution, which is adequate to resolve the correlation-table bands above. Closely spaced bands (e.g., resolving a doublet, or gas-phase rotational fine structure) call for finer resolution (2 or 1 cm-1), which increases scan time and reduces signal-to-noise per scan for a fixed total acquisition time.
- Number of scans (co-addition). FTIR instruments co-add repeated scans to improve signal-to-noise, since noise averages toward zero while the real signal reinforces. More scans mean a cleaner spectrum at the cost of longer acquisition time; weak or dilute samples generally need more scans than a neat, strongly-absorbing one.
- Background/reference scan. Every sample spectrum is ratioed against a background scan (empty ATR crystal, empty cell, or blank pellet) taken under the same conditions, to cancel out the instrument’s own optical response and atmospheric absorption. A stale or mismatched background is one of the most common causes of a spectrum with an odd sloping baseline or spurious bands.
- Purge/atmosphere control. Because water vapor and CO2 both absorb in the mid-IR, many instruments are purged with dry air or nitrogen, or sealed, to suppress their bands. Without adequate purging, atmospheric interference (see the troubleshooting table below) is layered on top of the sample spectrum.
Troubleshooting table
| Symptom | Likely cause | Fix |
|---|---|---|
| Sharp, jagged spikes clustered around 3550-3900 and 1300-1900 cm-1 | Atmospheric water vapor rotational-vibrational lines, from inadequate purge or a stale background scan | Re-run the background immediately before the sample; purge with dry air/N2; keep the sample chamber closed between scans |
| Sharp doublet-like feature near 2350 cm-1 | Atmospheric CO2 asymmetric stretch, incompletely subtracted by the background | Same fix as above — fresh background scan taken close in time to the sample scan, adequate purge |
| Broad band around 3200-3550 cm-1 in a sample that shouldn’t contain O-H | Moisture in hygroscopic KBr, or adventitious water on the ATR crystal/sample | Dry KBr before pressing (oven or desiccator); switch to ATR for a moisture-sensitive solid; clean and dry the ATR crystal between samples |
| Peaks flattened at the top or “clipped” near 0%T | Sample too concentrated or pathlength too long — the detector is saturated, not resolving true peak height | Dilute the sample, use a thinner film, reduce pellet loading, or reduce ATR contact pressure for a soft solid |
| Overall spectrum weak, noisy, poor signal-to-noise | Sample too dilute, too thin, or poor contact with the ATR crystal | Increase concentration/pathlength, increase ATR pressure, or increase the number of co-added scans |
| Sloping or curved baseline rather than a flat 100%T background | Light scattering (particle size too large in a mull or pellet), or uneven ATR crystal contact | Grind solids more finely and evenly; re-seat the ATR anvil; apply a baseline correction only after confirming it isn’t masking a real broad band (e.g., acid O-H) |
| Unexpected extra bands that don’t match any expected functional group | Residual solvent, mineral oil (Nujol) contribution, or ATR crystal contamination carried over from a previous sample | Check against known solvent/Nujol reference bands; clean the ATR crystal thoroughly (including with a solvent wipe) before the next sample; run a blank |
| Carbonyl band at an unexpected position (e.g., lower than the ketone range) | Conjugation with a ring or adjacent double bond, or intramolecular/intermolecular hydrogen bonding, both of which lower the C=O frequency | Not necessarily an error — check the structure for conjugation or H-bond donors/acceptors near the carbonyl before assuming misassignment |
Frequently asked questions
What is the difference between wavenumber and wavelength in IR spectroscopy?
Wavenumber (cm-1) is the reciprocal of wavelength and is directly proportional to vibrational energy and frequency, which is why IR spectra are conventionally reported in wavenumber rather than wavelength (in µm) — it lets you read energy/frequency relationships straight off the x-axis.
What is the fingerprint region and why isn’t it assigned band-by-band?
The fingerprint region (roughly 1500-600 cm-1) arises from coupled vibrations across the whole molecular skeleton rather than one isolated bond, so individual bands in this region don’t map cleanly to individual functional groups the way the O-H, C=O, or C-H stretches above 1500 cm-1 do. It’s used for whole-pattern comparison against a reference or literature spectrum rather than for peak-by-peak assignment.
Why is my O-H peak broad but my C=O peak sharp?
O-H groups hydrogen-bond to neighboring molecules (or, in carboxylic acids, dimerize), and hydrogen bonding spreads the vibration across a range of slightly different local environments and bond strengths, broadening the band. C=O has no comparable hydrogen-bonding network in most samples, so it stays sharp.
Can I use IR spectroscopy alone to fully determine a structure?
No — IR tells you which functional groups are present and, via the fingerprint region, whether a spectrum matches a known reference, but it doesn’t establish the full connectivity of a novel structure on its own. It’s routinely used alongside NMR and mass spectrometry, which supply the connectivity and molecular-weight information IR doesn’t.
Why does my spectrum show a peak around 1715 cm-1 even though my compound has no obvious carbonyl?
Check for adventitious contamination first (residual solvent, a plasticizer, or oxidation product), then check the structure again for a carbonyl-containing group that’s easy to overlook (an ester, lactone, lactam, or carbamate) before concluding it’s an artifact.
Related CASRAI guides
IR spectroscopy is one of several instrumental identification and characterization techniques covered on CASRAI; the guides below cover adjacent methods and the calibration/prep steps that feed into a clean spectrum.
- Spectrophotometer Calibration: Wavelength and Photometric Accuracy Checks — the UV-Vis analogue of instrument qualification, for labs running both techniques
- Thermogravimetric Analysis (TGA): Running a Scan and Interpreting Mass-Loss Steps — often run alongside FTIR for thermal decomposition products
- Atomic Absorption Spectroscopy: Flame vs Graphite Furnace, Lamp Selection and Detection Limits — elemental analysis to complement IR’s functional-group information
- HPLC: Columns, Mobile Phases, and a Peak-Problem Troubleshooting Table — for separating a mixture before IR characterization of individual components
- Gas Chromatography: Columns, Carrier Gases and Detectors Explained
- LC-MS Explained: How Liquid Chromatography and Mass Spectrometry Are Coupled — the molecular-weight/connectivity complement to IR’s functional-group data
- Thin-Layer Chromatography: How to Run a Plate and Calculate Rf
- Analytical Balance Calibration and Proper Weighing Technique — for accurately weighing out KBr or sample for a pellet
- Agarose Gel Electrophoresis Protocol Basics: What It Is and How It Works
