Put your hand on a warm mug and you feel heat. Tap the mug and you hear sound. It is one of the quiet marvels of condensed-matter physics that these two everyday sensations are, deep down, the same thing: the atoms of the solid, locked into their crystal lattice, are never sitting still. They are jiggling — springing back and forth against their neighbours in a vast, coupled dance. Sound is that jiggle organised into a travelling wave; heat is that jiggle scrambled into a chaos of overlapping waves. Understand the vibrations of the lattice and you have understood, in one stroke, how solids carry sound, store heat, and conduct (or refuse to conduct) thermal energy.

And here is the twist that makes it modern physics rather than nineteenth-century mechanics: the vibrations do not come in arbitrary amounts. Like light, they are quantised. A vibrational mode of frequency \omega can only gain or lose energy in discrete lumps of \hbar\omega. Each lump behaves so much like a particle — it carries energy, it carries a momentum-like quantity, it scatters off other lumps — that we give it a name and a life of its own: the phonon. This page builds the phonon from the ground up, starting with the simplest crystal imaginable: a single line of identical masses joined by springs. We lean throughout on the machinery of coupled oscillators and normal modes, and on the picture of a periodic crystal lattice.

The 1-D monatomic chain: masses and springs

Strip a crystal down to its barest cartoon. Take identical atoms of mass m, spaced a distance a apart along a line — a is the lattice constant — and join each one to its two neighbours by identical springs of stiffness K. Label the atoms by an integer n, and let u_n(t) be the small displacement of atom n from its resting site. Each atom feels a restoring pull from the spring on its left and the spring on its right, so Newton's second law reads

This is one equation for every atom — an infinite, coupled set. Solving it atom by atom would be hopeless. But the chain has a hidden symmetry we can exploit: it looks identical if we shift along by one lattice spacing. Whenever a system is periodic like this, its natural motions — its normal modes — are plane waves. So we guess a travelling-wave solution in which every atom does the same thing as its neighbour, just delayed in phase by an amount set by how far along the chain it sits:

Here q is the wavevector (how much the phase advances per unit length), na is the equilibrium position of atom n, and \omega is the angular frequency. The whole problem now collapses to finding which pairs (q, \omega) the chain actually supports.

The dispersion relation

Substitute the ansatz into Newton's law. Every atom carries the same factor e^{i(qna - \omega t)}; the neighbours differ only by u_{n\pm 1} = u_n\,e^{\pm i q a}. Dividing through by the common factor leaves an algebraic equation, no derivatives in sight:

Use the identity 1 - \cos qa = 2\sin^2(qa/2) to tidy the right-hand side:

Take the square root. The result — the relationship between frequency and wavevector — is the dispersion relation, the single most important curve in this whole subject:

That last point is the reason we only ever need q in the range -\pi/a < q \le \pi/a. This special interval is the first Brillouin zone — the fundamental "unit cell" of the wavevector world. Any q outside it is just a relabelling of one inside. Explore the curve below: drag the stiffness-to-mass slider and watch the whole dispersion scale up and down while its shape stays fixed.

Two limits worth knowing

Small q — long waves, and the speed of sound. When the wavelength is huge compared with the atomic spacing, qa is tiny and \sin(qa/2) \approx qa/2. The dispersion straightens into a line through the origin:

A frequency proportional to wavevector, \omega = v_s\,q, is precisely the signature of a non-dispersive sound wave travelling at a fixed speed. So the slope of the dispersion at the origin is the speed of sound in the chain:

This long-wavelength, low-frequency branch — where the atoms in each region move nearly together and the whole thing behaves like a continuous elastic rod — is called the acoustic branch, because it is literally the branch that carries sound.

Zone boundary — the shortest wave. At q = \pi/a we have \sin(qa/2) = \sin(\pi/2) = 1, so \omega = \omega_{\max} = 2\sqrt{K/m}. The wavelength here is \lambda = 2\pi/q = 2a — every atom moves exactly opposite to its two neighbours. This is the most violent, highest-frequency vibration the lattice can hold. And crucially the slope of the curve is flat there, so the group velocity d\omega/dq = 0: the wave does not travel at all, it is a standing oscillation. Energy piles up rather than flowing.

Seeing the modes

The chart shows the frequencies, but it helps to picture the motion. Reveal the figure step by step: first the chain at rest, then a gentle long-wavelength wave where a whole neighbourhood of atoms drifts together (that is the acoustic, sound-like limit), and finally the savage zone-boundary mode where each atom lunges the opposite way to its neighbours.

Two atoms per cell: acoustic and optical branches

Real crystals often have more than one kind of atom in the repeating unit — table salt alternates sodium and chloride, for instance. Model this as a diatomic chain: alternating masses m and M (with M > m), still joined by identical springs K. Now there are two atoms per unit cell, so the plane-wave ansatz needs two amplitudes, and solving the pair of coupled equations gives a quadratic in \omega^2 with two solutions for every q:

The two roots are the two branches of the dispersion, and they behave very differently:

• The lower branch \omega_- starts at zero at the zone centre and rises linearly — it is the acoustic branch again, in which the two atoms of a cell move in phase, drifting together like one heavy blob. This is the sound-carrying mode.
• The upper branch \omega_+ never touches zero; even at q = 0 it sits at a high frequency \sqrt{2K(1/m + 1/M)}. This is the optical branch, in which the light and heavy atoms of a cell move out of phase — against each other — while the cell's centre of mass barely stirs.

Between the top of the acoustic branch and the bottom of the optical branch lies a frequency gap: a band of frequencies the crystal simply cannot vibrate at. Slide the mass ratio below and watch the gap open and close.

From vibrations to phonons

So far everything has been classical springs. Now bring in quantum mechanics. Each normal mode of the lattice — each allowed (q, \omega) — is, mathematically, a simple harmonic oscillator. And the one thing everyone learns about the quantum harmonic oscillator is that its energy is not continuous: it comes in a ladder of equally spaced rungs,

The mode of frequency \omega can therefore only hold energy in whole multiples of \hbar\omega above its ground state. We call each of these quanta of vibrational energy a phonon — a quantised lattice vibration, the mechanical cousin of the photon. Exciting a mode from level n to n+1 is "creating one phonon"; the number n is simply how many phonons currently occupy that mode.

This is the pay-off. Once vibrations are quantised into phonons, a hot solid is simply a gas of phonons rattling around inside the crystal — being created, destroyed, and scattered. Heat flow becomes phonon flow; thermal resistance becomes phonon scattering. The whole thermal behaviour of an insulating solid is the statistical mechanics of this phonon gas.

Why this explains heat capacity

The nineteenth-century Dulong–Petit law said every solid should store 3k_B of energy per atom no matter the temperature, giving a constant heat capacity. It works at room temperature but fails badly as solids are cooled: measured heat capacities plunge towards zero near absolute zero. Classical physics had no explanation. Phonons do.

Because each mode's energy comes in lumps of \hbar\omega, a mode can only be "switched on" if the thermal energy k_BT is large enough to pay for at least one quantum. High-frequency modes with \hbar\omega \gg k_BT are effectively frozen out — thermal kicks are too feeble to excite them, so they store no energy. As you cool the crystal, more and more modes freeze, and the heat capacity slides toward zero. Two classic models capture this:

• Einstein model — pretend every mode has the same frequency \omega_E (a crude stand-in for the flat optical branch). Simple, and it correctly predicts the freeze-out, though it drops off too fast at very low T.
• Debye model — treat the low-frequency acoustic branch properly as sound waves with \omega = v_s q up to a cutoff. This nails the famous low-temperature result C \propto T^3 that experiments confirm beautifully.

Both are just the phonon gas doing thermal accounting. The moving atoms, the springs, the dispersion curve, the quantisation — it all comes together to explain a number you can measure with a calorimeter.

Worked examples

Example 1 — the maximum frequency. A monatomic chain has spring constant K = 50\ \text{N/m} and atomic mass m = 2\times 10^{-26}\ \text{kg}. Its highest phonon frequency sits at the zone boundary:

That is an infrared-scale frequency — exactly the ballpark where lattice modes and light meet.

Example 2 — the speed of sound. With the same numbers and a lattice constant a = 3\times 10^{-10}\ \text{m}, the long-wavelength sound speed is the slope of the dispersion at the origin:

A few kilometres per second — the right order of magnitude for sound in a stiff solid (steel is about 5\ \text{km/s}; our toy chain is stiffer and lighter, so faster).

Example 3 — counting the modes. A finite chain of N atoms, with periodic (wrap-around) boundary conditions, allows only N distinct wavevectors, evenly spaced by \Delta q = 2\pi/(Na) across the Brillouin zone. So the monatomic chain has exactly N normal modes — one per atom, matching the N degrees of freedom. A diatomic chain of N cells has 2N modes: N acoustic and N optical. The counting always works out to one mode per degree of freedom — a useful sanity check.