Phonons and collective modes

Let identical atoms of mass $m$ sit at equilibrium positions $ja$ in a one-dimensional periodic chain. Write $u_j$ for the small displacement from equilibrium. Nearest-neighbor harmonic springs contribute energy proportional to $(u_{j+1}-u_j)^2$. A uniform displacement costs no spring energy because all relative separations remain unchanged. Newton’s equation therefore contains a discrete second derivative. This starting model assumes a stable equilibrium, small displacements, identical masses, and interactions truncated to nearest-neighbor quadratic terms. It does not include bond breaking, thermal expansion, or arbitrary nonlinear motion.

Insert a plane-wave displacement with wavevector $q$ and frequency $\omega$. Translational symmetry makes these waves normal modes, so each allowed wavevector evolves independently in the harmonic approximation. The resulting acoustic branch approaches zero frequency at long wavelength: neighboring atoms move almost together, producing little restoring force. A finite chain has discrete allowed wavevectors, while an infinite-chain approximation replaces their sum by an integral. The exact zero mode in an unpinned chain is rigid translation of the center of mass; it is handled separately from an ordinary finite-frequency oscillator.

The normal-mode transformation diagonalizes the coupled harmonic Hamiltonian. Each nonzero-frequency real normal coordinate has the energy spectrum of a quantum harmonic oscillator. Its excitation number is the phonon occupation. Increasing that number adds one quantum of collective energy without adding an atom to the crystal. The zero-point term survives even at zero occupation. When using complex Fourier coordinates, the reality of atom displacements relates opposite wavevectors, so count independent degrees of freedom consistently rather than accidentally doubling the number of oscillators.

At small wavevector, the sine dispersion becomes approximately linear, with sound speed $c=a\sqrt{K/m}$. A wavepacket assembled from nearby modes travels with group velocity $d\omega/dq$, which falls to zero at the zone boundary in this particular chain. A diatomic unit cell introduces additional branches, including an optical branch in which the two sublattices can move oppositely. Optical here labels the branch structure and possible electromagnetic coupling; it does not mean that every mode must visibly emit light. Real three-dimensional crystals have several polarizations as well.

An isolated exactly harmonic crystal has no intrinsic phonon-phonon scattering: its normal occupations are constants of motion. Finite lifetimes appear when anharmonic couplings, defects, boundaries, electrons, or other degrees of freedom are included. A damped classical oscillator can model the decay of one measured coherent displacement over a suitable regime, but damping is not the same as assigning an exact phonon occupation. Even the distinction between well-defined quasiparticles and strongly broadened vibrations depends on comparing decay rates with oscillation frequencies and on the observable being measured.

The oscillator lab lets you inspect one classical unit-mass coordinate with initial displacement one and initial velocity zero. Change its natural frequency, then increase damping to compare oscillation and envelope decay. It does not contain a chain, wavevector-dependent dispersion, Bose occupation statistics, or quantum zero-point fluctuations. The honest bridge is that a harmonic normal coordinate obeys the same classical differential equation before quantization. To study heat capacity or thermal conductivity, one must additionally count modes, assign quantum populations, and, for transport, model how energy moves and relaxes.