Group Frequencies and Functional Group Identification

A carbonyl stretch near 1715 cm⁻¹ is one of the most reliable signals in organic chemistry. It tells a chemist, almost instantly, that a C=O bond is present. This single absorption can distinguish a ketone from an ester or an amide. It does so before a single spectral line is fully assigned to a normal mode. This reliability is not an accident. Certain bonds vibrate at frequencies that depend mostly on local mass and bond stiffness. They depend much less on the rest of the molecule. Chemists exploit this fact daily, using infrared spectra to identify functional groups within minutes. This complements the normal-mode picture developed for small polyatomic molecules. This article explains why group frequencies exist physically. It shows how to read the characteristic-group region above 1500 cm⁻¹. It also covers the fingerprint region below that value. You will see the harmonic-oscillator argument behind group frequency transferability. Two fully worked spectral identification problems follow. The article closes by previewing how symmetry, not just frequency, further restricts which vibrations appear in a spectrum at all.

Key takeaways

Group frequencies arise because light, stiff atoms (like H) vibrate largely decoupled from the heavy molecular frame.
The diatomic-oscillator formula predicts real trends: bond order and reduced mass set the frequency scale.
Above ~1500 cm⁻¹ absorptions are characteristic (O–H, N–H, C≡N, C=O); below it lies the fingerprint region.
The fingerprint region (1500–400 cm⁻¹) is unique to each molecule, like a barcode, and confirms identity by comparison.
Functional group identification uses both region types: characteristic peaks name the group, fingerprint peaks confirm the exact compound.
Key idea: A vibration behaves as a nearly isolated local oscillator when it involves a light atom or a very stiff bond, making its frequency a transferable signature of a functional group rather than a property of the whole molecule.

The core insight

Group frequencies exist because the vibrational frequency of a bond, approximated as a harmonic oscillator, depends primarily on the force constant and reduced mass of the atoms directly involved. When those atoms are much lighter, or the bond much stiffer, than the rest of the molecular framework, the vibration stays localized and nearly decoupled from other normal modes, so its frequency is transferable from molecule to molecule.

The physical picture

A normal mode, in the rigorous sense, is a collective motion of every atom in a molecule. Its frequency and displacement pattern come from diagonalizing the full mass-weighted force-constant matrix. In principle, every vibrational frequency depends on the entire molecular structure. Yet in practice, chemists routinely say "that peak at 3300 cm⁻¹ is an N–H stretch" without solving the full secular equation. This works because of a mechanical decoupling effect. Consider a C–H bond attached to a large carbon skeleton. The hydrogen atom is roughly 12 times lighter than carbon. When it vibrates, the heavy carbon frame barely moves — momentum conservation forces the light atom to do almost all the work. The vibration is therefore dominated by the local C–H stretching coordinate, with only a small admixture of skeletal motion. The same argument applies to O–H, N–H, and to unusually stiff bonds such as C≡N or C≡C, where a large force constant raises the frequency well above the dense cluster of skeletal modes, again isolating it energetically and mechanically from the rest of the vibrational manifold. This localization has a direct spectroscopic consequence. Any molecule containing an isolated N–H bond, whether it is ammonia, aniline, or a peptide, absorbs somewhere in a narrow window near 3300–3500 cm⁻¹. The precise frequency shifts slightly with hydrogen bonding, conjugation, or electronegativity of neighboring atoms, but the group is always recognizable. Below about 1500 cm⁻¹, this localization argument breaks down: bending and skeletal stretching modes have comparable frequencies and force constants, so they mix strongly. Motions of C–C, C–N, and C–O single bonds all fall in the same crowded window, and their frequencies genuinely depend on the whole molecular connectivity. This is the fingerprint region — highly diagnostic for confirming an exact compound by comparison to a reference spectrum, but not simply interpretable bond-by-bond.

Figure

Characteristic-group region versus fingerprint region across the mid-IR

Figure

Why light atoms localize a vibration: displacement amplitude versus mass

The core relations

\tilde{\nu} = \frac{1}{2\pi c}\sqrt{\frac{k}{\mu}}

where $\tilde{\nu}$ is the vibrational wavenumber, $k$ the bond force constant, $\mu$ the reduced mass, $c$ the speed of light.

\mu = \frac{m_1 m_2}{m_1+m_2}

where $m_1, m_2$ are the masses of the two atoms forming the (approximately local) oscillator.

\Delta \tilde{\nu} \approx \frac{1}{2\pi c}\left(\frac{1}{2\sqrt{k\mu}}\right)\Delta k

where a change in bond order or electronic environment shifts $k$ , giving a small, predictable frequency shift within a group's characteristic window.

\tilde{\nu}_\text{fingerprint} = f(\text{entire connectivity}),\quad \tilde{\nu} < 1500\ \text{cm}^{-1}

where fingerprint-region frequencies are genuinely global properties, not assignable to one bond.

Derivation AWhy light-atom bonds decouple from the molecular frame

Goal. Show that a diatomic-oscillator formula, applied to just two atoms of a polyatomic molecule, is a good approximation whenever one atom is much lighter than the rest of the structure it is attached to.

Step 1. Set up the local coordinate. Treat the X–H bond (X = C, N, O) as a two-body oscillator, ignoring for a moment that H is also weakly coupled to the rest of the molecule through X. The classical equation of motion for the stretching coordinate $q = r - r_e$ is

\mu\ddot q = -kq

giving harmonic solutions at angular frequency $\omega=\sqrt{k/\mu}$ .

Step 2. Bring in the rest of the molecule. In the full normal-mode treatment, H is also connected to a heavy "effective" mass $M$ representing the rest of the skeleton, with some coupling force constant $k'$ describing how the X–H stretch drags on the frame. A more complete two-mass model along the bond axis gives a reduced mass

\mu_{\text{eff}} = \frac{m_H M}{m_H + M}

where $M$ is of order tens to hundreds of atomic mass units for a typical organic fragment.

Step 3. Take the heavy-frame limit. Because $M \gg m_H$ , expand

\mu_{\text{eff}} = m_H\left(\frac{M}{m_H+M}\right) = m_H\left(1-\frac{m_H}{M}+\cdots\right) \approx m_H

so the effective reduced mass converges to the light-atom mass itself, essentially independent of what $M$ actually is. A CH stretch in methane, in a long alkane chain, or in a steroid all see nearly the same $\mu_\text{eff} \approx 1$ u corrected for the C–H splitting, because the heavy end of the "spring" barely recoils.

Step 4. Compare force constants across families. Typical single-bond force constants to hydrogen are $k \approx 500$ N m⁻¹; C=O double bonds have $k\approx 1200$ –1300 N m⁻¹; C≡N triple bonds have $k\approx 1800$ N m⁻¹. Because

\tilde\nu \propto \sqrt{k/\mu}

a bond that is either very light (small $\mu$ ) or very stiff (large $k$ ) is pushed to a wavenumber well outside the dense cluster of skeletal single-bond modes (typically $k\approx 400$ –600 N m⁻¹ among C, N, O framework atoms, all of comparable mass). Such modes are both mechanically isolated (little amplitude in the frame) and energetically isolated (far in frequency from other modes they could mix with by near-degeneracy), so mixing is weak and the frequency stays characteristic of the local X–H or X≡Y unit alone.

Step 5. Conclude the transferability principle. Since $\mu_\text{eff}$ and $k$ for a given bond type barely change between molecules, the predicted $\tilde\nu$ barely changes either, apart from small perturbations from neighboring electron-withdrawing or conjugating groups. This is the physical basis of the empirical group-frequency correlation tables.

\boxed{\tilde{\nu}_{X\text{–}H\ \text{or}\ X\equiv Y} \approx \frac{1}{2\pi c}\sqrt{\frac{k_\text{local}}{\mu_\text{local}}}\ \text{independent of the rest of the molecule, to good approximation}}

Worked examples

Example 1 — Predicting and checking the C=O stretch of acetone. Model the carbonyl as a simple diatomic C=O oscillator with force constant $k = 1200$ N m⁻¹ (a typical literature value for a ketone C=O). Reduced mass, using $m_C = 12.000$ u and $m_O = 15.999$ u, with $1\ \text{u} = 1.6605\times10^{-27}$ kg:

\mu = \frac{(12.000)(15.999)}{12.000+15.999}\ \text{u} = \frac{191.99}{27.999}\ \text{u} = 6.857\ \text{u}

\mu = 6.857 \times 1.6605\times10^{-27}\ \text{kg} = 1.1388\times10^{-26}\ \text{kg}

Angular frequency:

\omega = \sqrt{\frac{k}{\mu}} = \sqrt{\frac{1200}{1.1388\times10^{-26}}} = \sqrt{1.0538\times10^{29}} = 3.246\times10^{14}\ \text{rad s}^{-1}

Convert to wavenumber using $\tilde\nu = \omega/(2\pi c)$ , $c = 2.998\times10^{10}$ cm s⁻¹:

\tilde\nu = \frac{3.246\times10^{14}}{2\pi(2.998\times10^{10})} = \frac{3.246\times10^{14}}{1.8836\times10^{11}} = 1723\ \text{cm}^{-1}

Sanity check: acetone's experimental C=O stretch is observed at 1715 cm⁻¹. Our simple two-body harmonic estimate (1723 cm⁻¹) is within about 0.5%, well inside the expected error from ignoring anharmonicity and coupling to adjacent C–C stretches. This confirms both the force-constant value chosen and the diatomic approximation itself — exactly the kind of consistency chemists use to assign an unknown peak near 1715 cm⁻¹ confidently to a ketone carbonyl, distinguishing it from an ester (typically 1735–1750 cm⁻¹, higher $k$ from inductive withdrawal by the ester oxygen) or an amide (typically 1630–1700 cm⁻¹, lower $k$ from resonance delocalization into the C–N bond).

Example 2 — Identifying an unknown compound, $C_4H_8O_2$ , from its IR spectrum. An unknown liquid with molecular formula $C_4H_8O_2$ (one degree of unsaturation, since a saturated $C_4H_{10}O_2$ would have 10 H) shows a mid-IR spectrum with these key bands:

Strong sharp peak at 1740 cm⁻¹
Strong peak at 1240 cm⁻¹
No absorption between 3200–3550 cm⁻¹
No absorption near 2250 cm⁻¹ or 2100 cm⁻¹
Weak C–H stretches near 2980 and 2900 cm⁻¹
Complex fingerprint pattern between 1000–1300 cm⁻¹

Step 1 — Use the characteristic-group region to rule things out. The absence of any O–H or N–H absorption (3200–3550 cm⁻¹) rules out alcohols, carboxylic acids, and amines/amides. The absence of a band near 2250 cm⁻¹ rules out a nitrile; absence near 2100 cm⁻¹ rules out an alkyne or cumulated diene. So the one degree of unsaturation is not a C≡N, C≡C, or an isolated N–H/O–H feature.

Step 2 — Identify the carbonyl. A strong, sharp band at 1740 cm⁻¹ is squarely in the ester C=O range (1735–1750 cm⁻¹), higher than a simple ketone (≈1715 cm⁻¹, as in Example 1) because the adjacent ester oxygen's electron-withdrawing inductive effect raises the C=O force constant. This single peak accounts for the one degree of unsaturation and matches the O-rich formula ( $C_4H_8O_2$ , two oxygens, consistent with a –C(=O)–O– ester linkage).

Step 3 — Confirm with the companion C–O stretch. Esters show a second strong, diagnostic band from asymmetric C–O–C stretching, typically 1200–1250 cm⁻¹; the observed 1240 cm⁻¹ band matches this perfectly, reinforcing the ester assignment (a lone carbonyl at 1740 cm⁻¹ without this companion band would instead suggest, e.g., a strained lactone or an unusual ketone).

Step 4 — Use the fingerprint region only for final confirmation, not primary assignment. The complex absorption pattern between 1000–1300 cm⁻¹ is consistent with C–C and C–O skeletal vibrations of an alkyl ester chain but is not, by itself, interpretable bond-by-bond; it would be compared against a reference spectrum (e.g., ethyl acetate or methyl propanoate) to confirm the exact isomer, since $C_4H_8O_2$ is consistent with several ester constitutions (ethyl acetate, methyl propanoate).

Conclusion. The compound is an aliphatic ester, most likely ethyl acetate or methyl propanoate; characteristic-region data (1740 and 1240 cm⁻¹, absence of O–H/N–H/C≡N) identify the functional group unambiguously, while fingerprint comparison against reference spectra would pin down the precise isomer.

Limiting cases & common traps

Situation	What actually happens
X–H stretch (X = C, N, O)	Highly localized, transferable, appears 2800–3600 cm⁻¹ regardless of the rest of the molecule
C=O, C≡N, C≡C stretches	Stiff-bond isolation; appear 1650–2260 cm⁻¹, largely insensitive to remote substituents
Region below 1500 cm⁻¹	Genuinely collective skeletal modes; frequencies depend on whole connectivity — use for compound-level "barcode" matching, not group assignment
Hydrogen bonding	Broadens and red-shifts O–H/N–H stretches significantly (e.g., free O–H ≈ 3600 cm⁻¹ vs. H-bonded, broad, 3200–3400 cm⁻¹)
Conjugation with a carbonyl	Lowers C=O frequency by 20–40 cm⁻¹ (resonance reduces double-bond character, lowering $k$ )

The chemist behind it

William Weber Coblentz spent the early twentieth century painstakingly recording infrared spectra of hundreds of pure organic compounds, at a time when infrared detectors were crude thermopiles and every spectrum took hours to record. His published catalogs, compiled through relentless, repetitive measurement, first revealed that certain wavelength regions correlated consistently with specific chemical groups across unrelated compounds. This empirical foundation, established well before quantum mechanical vibrational theory matured, made group-frequency correlation tables possible and turned infrared spectroscopy into a practical tool for structural chemistry.

Origins · 1 minWilliam W. Coblentz and how this was discovered

William W. Coblentz

1873–1962 · American physicist whose systematic infrared surveys of organic compounds established the empirical basis for group-frequency correlation tables

Ask the tutor: Want to work through how conjugation and ring strain shift a carbonyl stretch away from its "textbook" value, bond by bond?

Why doesn't every vibration behave like an isolated bond, if normal modes are just linear combinations of bond stretches?

Normal modes are indeed linear combinations of all internal coordinates, in principle. But the mixing between two internal coordinates is large only when their local force constants and reduced masses are similar, putting them near the same natural frequency. Light-atom stretches (X–H) and very stiff multiple bonds (C=O, C≡N) sit far in frequency from the dense cluster of C–C, C–N, and C–O single-bond skeletal modes. That frequency separation, plus the mechanical decoupling from a heavy frame, keeps mixing small, so these particular modes stay localized while lower-frequency skeletal modes mix freely.

How much does hydrogen bonding actually shift an O–H stretch, and why?

A free (gas-phase or dilute-solution) O–H stretch appears near 3580–3650 cm⁻¹. In a hydrogen-bonded environment, such as a neat alcohol or a carboxylic acid dimer, the same stretch broadens dramatically and shifts down to roughly 3200–3400 cm⁻¹, sometimes lower. Hydrogen bonding weakens the O–H bond by donating electron density into the bonding orbital from the acceptor, lowering the effective force constant $k$ . Since $\tilde\nu \propto \sqrt{k}$ , even a modest reduction in $k$ produces a substantial, easily observed red shift and peak broadening from a distribution of hydrogen-bond geometries.

Can two completely different molecules have identical fingerprint regions?

In practice, no — the fingerprint region is sensitive to overall connectivity, ring size, branching, and substitution pattern, so it acts as a near-unique identifier, much like a barcode. Two different constitutional isomers, such as ethyl acetate and methyl propanoate, may share nearly identical characteristic-group peaks (both are esters) but show distinguishable fingerprint patterns. This is precisely why spectral libraries and reference-spectrum matching remain standard practice for confirming a specific compound's identity, beyond simply naming its functional groups.

How does this topic connect to the symmetry-based selection rules covered next?

Group frequencies tell you where to look for a functional group based on local mass and force constant. Whether a given normal mode of that group actually appears with strong intensity in an infrared spectrum, versus in a Raman spectrum, is a separate question governed by molecular symmetry. For centrosymmetric molecules, the Rule of Mutual Exclusion states that no vibration can be both IR-active and Raman-active. This selection-rule framework builds directly on the normal-mode and group-frequency ideas developed here.