Infrared Spectroscopy: Molecular Vibrations and the Harmonic Oscillator

Modern FTIR spectrometers (Fourier-transform infrared) use a Michelson interferometer and a moving mirror rather than a slow scanning prism, collecting an entire spectrum's worth of interference data in under a second and Fourier-transforming it into the familiar absorbance-versus-wavenumber plot. This speed makes IR the fastest structural technique available to a synthetic chemist: a reaction can be monitored in real time by tracking a single functional group's disappearance (an alcohol O–H band shrinking as it is oxidized to a ketone C=O band, for instance), and modern reaction-monitoring probes use ATR-FTIR (attenuated total reflectance) to record spectra directly inside a reaction flask without sampling. In pharmaceutical quality control, IR spectra serve as fingerprints for confirming the identity and polymorphic form of active pharmaceutical ingredients, since even different crystal packings of the same molecule (polymorphs) can shift hydrogen-bonding-sensitive bands measurably. In materials science, IR (and its complementary sibling, Raman spectroscopy, governed by the reversed selection rule derived above based on polarizability rather than dipole moment) characterizes polymers, catalysts, and semiconductor surfaces. In atmospheric and climate science, exactly the same physics — vibrational transitions with nonzero $\partial\mu/\partial x$ — explains why CO₂, water vapor, and methane are potent greenhouse gases (their bending and asymmetric-stretch modes absorb infrared radiation emitted from Earth's surface) while the far more abundant N₂ and O₂, being homonuclear diatomics with permanently zero dipole moment at every bond length, are completely IR-transparent and contribute nothing directly to the greenhouse effect. In biochemistry, IR spectroscopy of amide I and amide II bands (largely C=O stretch and N–H bend of the peptide backbone) reports directly on protein secondary structure, distinguishing α-helix from β-sheet content without needing crystallography. This same vibrational language — normal modes, dipole derivatives, force constants — underlies the interpretation of the specific correlation charts built in IR Functional Group Frequencies and the quantitative mass and bond-order trends explored in IR Hooke's Law and Frequency Trends.

Ask the tutor: Why do overtone bands (v=0→v=2) appear at all in real IR spectra if the harmonic oscillator selection rule strictly forbids them?

Start with the simplest possible molecule: a diatomic, say H–Cl. Classically, imagine the two nuclei connected by a spring with force constant $k$ (units N/m), the stiffness of the bond. If you stretch this spring by a displacement $x$ from its equilibrium bond length, Hooke's law says the restoring force pulling the atoms back together is

This is exactly the same law governing a mass on a spring in first-year mechanics — bond stretching is a mechanical oscillation problem in disguise. The molecule's electrons redistribute so fast (electronic motion is ~1000× faster than nuclear vibration, the Born–Oppenheimer approximation) that at every instant the electrons "see" a fixed internuclear distance and generate exactly the restoring force appropriate to that distance. This is why the harmonic spring model, despite treating atoms as classical balls, captures real chemistry so well. Two nuclei of masses $m_1$ and $m_2$ vibrating against each other don't move independently — the center of mass must stay fixed (no net translation), so the natural variable is the reduced mass,

A single heavy atom oscillating against an effectively infinite wall would need a different treatment; reduced mass converts the two-body oscillator into an equivalent one-body problem: a single particle of mass $\mu$ on a spring of constant $k$. Solving Newton's second law $\mu \ddot{x} = -kx$ classically gives simple harmonic motion at frequency

This is the classical vibrational frequency — but real bonds are quantum objects, and quantum mechanics changes the story in one crucial way: energy is quantized. A molecule cannot vibrate with an arbitrary amplitude; it can only occupy discrete vibrational energy levels. This quantization, worked out explicitly below, is what makes IR spectroscopy a spectroscopy at all — a bond absorbs a photon only when the photon's energy $h\nu_{\text{photon}}$ exactly bridges the gap between two allowed vibrational levels. Molecules with more than two atoms vibrate in more complex ways than a simple stretch. A nonlinear molecule with $N$ atoms has $3N$ total degrees of freedom (each atom moves in $x$, $y$, $z$); subtracting 3 translations of the whole molecule and 3 overall rotations leaves $3N-6$ normal modes of vibration (a linear molecule loses only 2 rotational degrees of freedom, since rotation about its own axis is not a real motion, giving $3N-5$ modes). Each normal mode is a specific, independent collective motion in which every atom oscillates sinusoidally at the same frequency and passes through equilibrium at the same instant — the molecular equivalent of the fundamental and overtone modes of a vibrating string.

The central chemical question is: which of these vibrations actually shows up in an IR spectrum? The answer is governed by a selection rule rooted in how light interacts with matter, developed rigorously below.

The quantum mechanical treatment of the harmonic oscillator replaces the classical potential energy of a spring,

directly into the time-independent Schrödinger equation for a particle of reduced mass $\mu$:

where:

• $\hbar = h/2\pi$, Planck's constant divided by $2\pi$
• $\mu$ is the reduced mass of the two vibrating atoms
• $x$ is the displacement from the equilibrium bond length
• $\psi(x)$ is the vibrational wavefunction, $E$ the vibrational energy

Solving this differential equation (carried out fully in the derivation below) yields quantized energies

where:

• $v$ is the vibrational quantum number
• $\nu = \frac{1}{2\pi}\sqrt{k/\mu}$ is the same classical oscillator frequency derived above
• the $+\tfrac12$ term is the zero-point energy — even in the ground state $v=0$, the bond never stops vibrating

Reading this equation: vibrational energy levels are evenly spaced ladder rungs, each separated by exactly $h\nu$, and no molecule at any temperature is ever perfectly motionless — a direct, testable consequence of the Heisenberg uncertainty principle applied to a bound particle. For IR spectroscopy specifically, chemists report frequencies not in Hz but in wavenumbers $\tilde\nu = \nu/c$ (units cm⁻¹), because wavenumber is directly proportional to photon energy via $E = hc\tilde\nu$ and gives convenient numbers (400–4000 cm⁻¹ for typical organic vibrations) without factors of $10^{13}$. The second essential relationship is the selection rule, which determines whether a transition between vibrational levels can absorb a photon at all:

where:

• $\mu_{\text{dipole}}$ is the molecular dipole moment (not to be confused with reduced mass, unfortunately sharing the same symbol by convention)
• $x$ is the vibrational displacement coordinate
• the derivative is evaluated at the equilibrium geometry

Reading this equation: a vibration is IR-active only if the dipole moment changes as the atoms move through that vibration. It is not enough for the molecule to have a permanent dipole (or even to lack one) — what matters is the rate of change of dipole with displacement along that specific normal coordinate.

Derivation 1 — the quantized vibrational energy ladder

Goal: show explicitly how solving the Schrödinger equation for a spring-like potential forces the vibrational energy to be quantized in equal steps of $h\nu$, starting from nothing but Hooke's law and the postulates of quantum mechanics.

Step 1 — set up the Hamiltonian. The total energy operator for a particle of reduced mass $\mu$ bound in a parabolic (harmonic) potential is

This comes directly from writing kinetic energy $\hat T = \hat p^2/2\mu$ with $\hat p = -i\hbar\, d/dx$ (the standard quantum momentum operator) and adding the classical Hooke's law potential energy $V(x) = \tfrac12 kx^2$. No new physics is introduced beyond quantizing the classical spring Hamiltonian $H = p^2/2\mu + \tfrac12 kx^2$.

Step 2 — nondimensionalize. Define $\omega = \sqrt{k/\mu}$ (angular frequency) and substitute the dimensionless variable $\xi = \sqrt{\mu\omega/\hbar}\, x$. This rescaling turns the Schrödinger equation

into the standard Hermite equation,

This is a well-studied equation in mathematical physics: it only has solutions that stay finite (do not blow up) as $\xi \to \pm\infty$ when $\lambda$ takes specific odd-integer values.

Step 3 — apply the boundedness (normalizability) condition. A wavefunction must be normalizable — $\int |\psi|^2\,dx$ must be finite — because $|\psi|^2$ represents a probability density that must integrate to 1. Solutions of the Hermite equation only satisfy this physical requirement when

Any other value of $\lambda$ produces a $\psi$ that diverges at large $|\xi|$ — an unphysical, non-normalizable wavefunction representing infinite probability. This is exactly analogous to how boundary conditions on a vibrating guitar string only permit standing waves of specific wavelengths; here, the "boundary condition at infinity" (wavefunction decay) plays the same quantizing role.

Step 4 — extract the energy levels. Substituting $\lambda = 2v+1$ back into the definition $\lambda = 2E/\hbar\omega$:

Converting angular frequency $\omega = 2\pi\nu$ back to ordinary frequency $\nu$:

This recovers exactly the quantized ladder quoted above, with rungs uniformly spaced by

Step 5 — connect to absorption. For a photon to be absorbed, its energy must exactly match a gap between occupied and unoccupied vibrational levels. At room temperature, thermal population is overwhelmingly in $v=0$ (by the Boltzmann distribution, since $h\nu \gg k_BT$ for typical bond vibrations — a C–H stretch at 3000 cm⁻¹ corresponds to an energy gap roughly twelve times $k_BT$ at 298 K). The dominant, allowed transition is therefore the fundamental, $v=0 \to v=1$, requiring photon energy

so the absorbed light frequency exactly equals the mechanical oscillator frequency $\nu = \frac{1}{2\pi}\sqrt{k/\mu}$ — the harmonic oscillator's quantum mechanics reproduces the classical resonance condition exactly, which is why the simple spring picture works so well for predicting where bands appear.

Derivation 2 — the dipole selection rule (why some vibrations are IR-silent)

Goal: derive, from time-dependent perturbation theory applied to light–matter interaction, the precise mathematical condition under which a vibrational transition can absorb a photon — explaining mechanistically why CO₂'s symmetric stretch is invisible in IR while its asymmetric stretch and bend are strong.

Step 1 — the light–matter interaction Hamiltonian. An oscillating electric field $\mathcal{E}(t) = \mathcal{E}_0\cos(2\pi\nu_{\text{light}} t)$ interacts with a molecule's electric dipole moment $\boldsymbol{\mu}$ through the perturbation

This is simply the classical energy of a dipole in a field, $-\boldsymbol{\mu}\cdot\boldsymbol{\mathcal{E}}$, promoted to an operator — light couples to matter through charge, and the dipole moment is the leading-order description of a molecule's asymmetric charge distribution.

Step 2 — expand the dipole moment along the vibrational coordinate. As the bond vibrates (coordinate $x$), the dipole moment $\mu(x)$ generally changes because the charge distribution reorganizes as bond length changes. Taylor-expand around the equilibrium geometry $x=0$:

where $\mu_0$ is the permanent (equilibrium) dipole moment and the derivative terms describe how the dipole changes as the bond stretches or bends.

Step 3 — compute the transition moment integral. Time-dependent perturbation theory (Fermi's Golden Rule) says the rate of transition between initial state $\psi_i$ (here $v=0$) and final state $\psi_f$ (here $v=1$) is proportional to the square of the transition dipole moment,

Substituting the Taylor expansion from Step 2:

Step 4 — evaluate each term using harmonic oscillator wavefunction symmetry. The first integral $\int\psi_1^*\psi_0\,dx$ is zero by the orthogonality of eigenstates ($v=0$ and $v=1$ are different energy eigenstates of the same Hermitian Hamiltonian, hence automatically orthogonal) — so the permanent dipole moment $\mu_0$ contributes nothing to the transition, regardless of how large it is. This already tells us that even a highly polar molecule can have IR-silent vibrations, and even a nonpolar molecule ($\mu_0 = 0$, like CO₂ or N₂) can have IR-active ones. The second integral, $\int \psi_1^{*}\, x\, \psi_0\, dx$, is a standard harmonic-oscillator matrix element that evaluates to a nonzero constant (proportional to $\sqrt{\hbar/2\mu\omega}$) precisely because $x$ is an odd function, $\psi_0$ is even, and $\psi_1$ is odd — making the full integrand ($\text{odd}\times\text{odd}\times\text{even} = \text{even}$) integrate to a nonzero value over all space. Higher terms (quadratic in $x$, etc.) vanish for the fundamental transition by similar symmetry arguments and are only responsible for weak overtone bands.

Step 5 — state the selection rule. Since the first (permanent-dipole) term vanishes and only the second term survives, the transition dipole moment for the fundamental IR transition reduces to

The transition probability is proportional to $|\mu_{fi}|^2$, so:

If the dipole derivative is zero — because the vibration is perfectly symmetric and any charge buildup on one side is exactly canceled on the other — the transition dipole moment vanishes and no photon can be absorbed, however large the static dipole moment $\mu_0$ might be.

Step 6 — apply to CO₂ explicitly. In the symmetric stretch, both C=O bonds lengthen and shorten in phase; by the molecule's center-of-symmetry, whatever dipole increment builds up on the left C=O is exactly canceled by the mirror-image change on the right C=O at every instant. Thus $(\partial\mu/\partial x)_0 = 0$ for this mode: it is IR-inactive (it is, in fact, Raman-active instead — a complementary vibrational technique with the opposite selection rule, based on polarizability rather than dipole moment). In the asymmetric stretch, one C=O shortens while the other lengthens, breaking the cancellation; a net oscillating dipole appears along the molecular axis, $(\partial\mu/\partial x)_0 \neq 0$, and this mode absorbs strongly at 2349 cm⁻¹. The bending mode also breaks the linear symmetry and is IR-active as well.

Example 1 (foundational): vibrational frequency of H–Cl from its force constant

HCl has an experimentally measured force constant $k = 480\ \text{N/m}$ (a typical value for a strong single bond). Find the fundamental vibrational frequency in both Hz and cm⁻¹, and the zero-point energy.

Reduced mass. Using atomic masses $m_H = 1.008\ \text{u}$, $m_{Cl} = 34.97\ \text{u}$ (dominant isotope ³⁵Cl), convert to kg using $1\ \text{u} = 1.6605\times10^{-27}\ \text{kg}$:

Frequency.

Convert to wavenumbers (dividing by speed of light $c = 3.00\times10^{10}\ \text{cm/s}$):

This lands almost exactly on the experimentally observed HCl fundamental at 2886 cm⁻¹ — a striking success for a model built entirely from Hooke's law and one quantum postulate.

Zero-point energy.

Converting to kJ/mol (multiply by Avogadro's number and convert J to kJ):

Even in its lowest possible energy state, an HCl molecule carries 17.2 kJ/mol of vibrational energy that can never be removed — a direct, numerically explicit demonstration of the zero-point energy predicted by Step 4 of the first derivation.

Limiting case check: as $k \to 0$ (an infinitely weak, floppy "bond"), $\nu \to 0$ and $E_0 \to 0$ — a bond with zero restoring force has no vibration and no zero-point energy, exactly as physical intuition demands. As $\mu \to \infty$ (arbitrarily heavy atoms), $\nu \to 0$ as well — heavy masses vibrate more sluggishly for the same spring stiffness, matching the everyday observation that a heavy pendulum bob swings more slowly than a light one on the same string.

Example 2 (applied): predicting IR activity in cis- and trans-1,2-dichloroethylene

Consider the two geometric isomers of 1,2-dichloroethylene, ClHC=CHCl. Predict whether the symmetric C–Cl stretching mode is IR-active in each isomer.

Cis isomer (both Cl atoms on the same side of the C=C double bond, $C_{2v}$ symmetry): the molecule has a net permanent dipole moment because the two C–Cl bond dipoles point in directions that partially reinforce rather than cancel. In the symmetric C–Cl stretch (both C–Cl bonds lengthening together), the dipole moment magnitude changes continuously as both bonds stretch in phase — there is no symmetry element that forces exact cancellation of the changing dipole contributions. Thus $(\partial\mu/\partial x)_0 \neq 0$: this mode is IR-active in cis-dichloroethylene.

Trans isomer (Cl atoms on opposite sides, $C_{2h}$ symmetry, with a center of inversion): here the molecule has zero permanent dipole moment because the two C–Cl bond dipoles point in exactly opposite directions and cancel. Consider the symmetric C–Cl stretch: by the inversion center, stretching the left C–Cl bond by $+\delta$ is mirrored by the right C–Cl bond also stretching by $+\delta$ in the inversion-related direction — the dipole increments from the two bonds remain equal and opposite at every point in the vibration, exactly analogous to the CO₂ symmetric stretch. Therefore $(\partial\mu/\partial x)_0 = 0$ for this mode by the same center-of-symmetry cancellation argument used in Derivation 2, Step 6: this mode is IR-inactive in trans-dichloroethylene (though it remains Raman-active). This exact IR/Raman complementarity — a consequence of a rule called the mutual exclusion principle, valid for any molecule with a center of inversion — lets chemists distinguish cis from trans isomers experimentally: comparing IR and Raman spectra of the two isomers reveals which vibrational bands appear in one technique but not the other, directly diagnosing the presence or absence of a center of symmetry, and hence the geometry.

Limiting case check: for a hypothetical isomer with the two C–Cl bonds at some intermediate dihedral angle (neither perfectly cis at 0° nor perfectly trans at 180°), there is no exact symmetry-forced cancellation at any angle other than the special $C_{2h}$ trans geometry — so as the dihedral angle is smoothly varied away from 180°, the symmetric-stretch IR intensity should smoothly grow from zero, confirming that the cis-isomer's activity is not a coincidence but the generic (symmetry-broken) case.

Example 3 (exam/competition level): estimating the C≡N stretching frequency and comparing to C≡C

A nitrile R–C≡N has a measured force constant of approximately $k = 1780\ \text{N/m}$ (triple bonds are much stiffer than single bonds — roughly triple the force constant of a comparable single bond, consistent with three bonding interactions instead of one). Estimate the fundamental stretching frequency in cm⁻¹, and compare with an alkyne C≡C stretch, $k \approx 1580\ \text{N/m}$, treating the terminal alkyne as a C–C oscillator against effectively the rest of the chain (approximate as isolated C≡C with both carbons as $m=12.00\ \text{u}$, i.e. treat as though vibrating against another carbon).

Nitrile C≡N. Reduced mass using $m_C = 12.00\ \text{u} = 1.9926\times10^{-26}\ \text{kg}$ and $m_N = 14.003\ \text{u} = 2.3253\times10^{-26}\ \text{kg}$:

This matches the well-known nitrile C≡N stretching region (2200–2260 cm⁻¹, real spectra run slightly higher because the true force constant and anharmonicity corrections push it up, but the order of magnitude and region are correctly reproduced by the bare harmonic estimate).

Alkyne C≡C (treating $\mu_{CC} = m_C/2 = 12.00\ \text{u}/2$, since both atoms are identical carbon):

Interpretation. Both triple-bond stretches fall in the same 2100–2260 cm⁻¹ window, well-separated from single (1000–1300 cm⁻¹) and double (1600–1750 cm⁻¹) bond stretches — directly reflecting the roughly linear increase of force constant with bond order predicted by Hooke's law, since $\tilde\nu \propto \sqrt{k}$ and $k$ scales upward with bond multiplicity. The small residual difference between C≡C (2113 cm⁻¹ estimated, 2100–2260 observed for internal/terminal alkynes) and C≡N (2160 cm⁻¹ estimated, 2200–2260 observed) arises mostly from the slightly higher reduced mass mismatch and slightly different intrinsic force constant of the more polar, more ionic-character C≡N bond — exactly the kind of fine structure explored quantitatively in IR Hooke's Law and Frequency Trends. Note also, importantly, that internal (disubstituted) alkynes R–C≡C–R′ often show very weak or absent C≡C stretches in the IR despite being present, because a nearly symmetric substitution pattern drives $(\partial\mu/\partial x)_0 \to 0$ — the selection rule derived above, not the force constant, governs whether the band is even observable, regardless of how accurately its predicted frequency matches theory.

The quantum harmonic oscillator itself emerges from the broader development of quantum mechanics by Schrödinger, Heisenberg, and Dirac in the 1920s, but the practical marriage of this theory to molecular structure determination is largely credited to the American physicist William Weber Coblentz, whose painstaking prism-spectrometer measurements of hundreds of organic and inorganic compounds (published 1905) first established that specific functional groups reliably absorb infrared radiation at characteristic, reproducible frequencies — the empirical foundation later explained rigorously by the quantum theory developed here. Coblentz worked with primitive, unstable rock-salt prism instruments in an era before electronic amplification, often recording spectra by hand from a thermopile detector, and his systematic catalog of vibrational frequencies for hundreds of pure compounds remains recognizable in modern correlation tables.