Lattice Vibrations and Collective Modes

Atoms, electrons and the electromagnetic field can move together in a solid. A phonon describes a quantized lattice vibration; a plasmon is a collective charge oscillation; an exciton is a correlated electron–hole pair; and a polaron is a carrier dressed by its surroundings. These excitations can interact and mix, so their energies alone do not always reveal what is moving.

This chapter connects the microscopic motion to dispersions, thermodynamics and measured spectra. The first step is to identify the relevant coordinate or operator. The next is to calculate both its response and its coupling to the probe.

Helpful background. Lattices, Reciprocal Space, and Bloch Electrons supplies crystal and momentum labels. Band Theory and Electronic Structure supplies electron and hole states. Quasiparticles and Collective Modes explains poles, spectral weights and lifetimes.

From atomic motion to a spectrum

In the harmonic approximation, small displacements around a stationary crystal structure obey a coupled oscillator problem. A mass-weighted dynamical matrix has eigenvalues $\omega_\nu^2(\mathbf q)$ and eigenvectors that describe the relative atomic displacements. A three-dimensional crystal with $r$ atoms per primitive cell has $3r$ branches. Translational invariance supplies three acoustic branches approaching zero frequency at the zone center; other zero modes or instabilities require further interpretation.

Phonons develops that material eigenproblem, while Phonons as Many-Body Excitations explains its quantization. The mode spectrum determines harmonic vibrational thermodynamics, but it does not by itself determine how quickly a heat current relaxes. For that one needs interactions, disorder, boundaries and the relevant scattering processes.

A negative harmonic eigenvalue means a direction of negative energy curvature around the chosen structure. It is not a phonon with negative excitation energy. Finite-temperature anharmonicity can change the free-energy curvature, but stability requires considering the coupled coordinates and competing structures, not merely finding one positive fitted frequency.

Choose a physical question

Question	Article	Main calculation
Which atoms move, and at what frequencies?	Phonons	Force constants, polarization vectors and the dynamical matrix.
What can a simple vibrational heat-capacity model explain?	Debye–Einstein Models	Approximate densities of states with explicit mode counts.
Why do frequencies, widths and volume change with temperature?	Anharmonic Lattice Dynamics	Nonquadratic forces, free energies and phonon scattering.
How does lattice motion affect electrons and pairing?	Electron–Phonon Coupling	A material coupling vertex with different weights for spectra, transport and pairing.
How does a carrier’s surrounding distortion change its motion?	Polarons in Materials	Dressed energy, mass and spectral weight, with coherent or hopping motion.
How do density and dielectric surroundings set charge modes?	Plasmons in Materials	Coupled charge response and electromagnetic boundary conditions.
When is an electron–hole pair bound and optically visible?	Excitons in Materials	The screened pair problem, matched continuum and transition matrix elements.

Start with phonons before Debye–Einstein models or anharmonicity. Electron–phonon coupling combines that preparation with band states. Plasmons and excitons can be read separately when charge response or optical transitions are the immediate question. For collective spin motion, continue to Spin Waves and Magnons.

Energies, widths and intensities answer different questions

A normal-mode frequency comes from an eigenproblem. A measured peak additionally depends on how the probe couples to the mode, its intrinsic damping and instrumental resolution. An optical phonon need not be optically active; a neutral exciton can nevertheless couple strongly to light.

For a weakly damped resonance with a pole $\omega_p=\omega_0-i\gamma$ and time dependence $e^{-i\omega t}$, the oscillation amplitude decays as $e^{-\gamma t}$. Near an isolated pole, a Lorentzian has angular-frequency half width $\gamma$. Width conventions and experimental broadening must therefore be specified before extracting a lifetime. When a continuum dominates, one frequency and lifetime may no longer summarize the response well.

Neutron Scattering and X-Ray Scattering connect momentum-resolved correlations to scattering intensities. Raman and Optical Spectroscopy explains optical matrix elements and selection rules. First-order Raman scattering usually samples near-zero crystal momentum; it cannot automatically identify a soft mode at a distant wave vector.

Likewise, a polaron mass does not fix a mobility, a density of states does not fix thermal conductivity, and one electron–phonon coupling average cannot answer every transport and pairing question. Each observable weights the same microscopic motion differently.

Mixed modes and a common continuum

In a doped polar semiconductor, a longitudinal optical vibration can mix with a charge oscillation. As carrier density changes, the two modes may exchange ionic and electronic character. One must follow those eigenvectors and their spectral weights as well as their energies. A bulk longitudinal calculation uses the full complex dielectric response; a surface or film mode also depends on the electromagnetic boundary conditions. A zero of the real part of one dielectric entry alone is insufficient.

For an exciton, the comparison is instead with the unbound electron–hole continuum at the same total momentum and compatible quantum numbers. Its binding energy is the separation below that continuum. Comparing an optical peak with an unrelated Kohn–Sham gap does not determine the binding. Band Gaps gives explicit examples of the required momentum constraint.

Exercise: stable coordinates, unstable combination

A pair of dimensionless mass-weighted coordinates has harmonic energy

$$U=\frac{E_*}{2}\begin{pmatrix}x&y\end{pmatrix} \begin{pmatrix}1&0.9\\0.9&0.5\end{pmatrix} \begin{pmatrix}x\\y\end{pmatrix},\qquad E_*>0.$$

Both diagonal curvatures are positive. Is the stationary point $x=y=0$ locally stable? Explain why measuring or calculating one positive curvature is insufficient.

Solution

The determinant is $0.5-0.9^2=-0.31$, so one eigenvalue is negative. Explicitly,

$$\lambda_\pm=\frac{1.5\pm\sqrt{0.25+4(0.9)^2}}2 \simeq 1.6841,\;-0.1841.$$

For the negative eigenvalue, the first row gives $y/x=-(1-\lambda_-)/0.9\simeq-1.316$. Opposite displacements in this proportion lower the energy. Positive curvature along each coordinate axis therefore does not imply positive curvature in every combined direction. In a finite-temperature stability problem the analogous test uses the full coupled free-energy Hessian at a stationary structure; even local stability does not establish the globally preferred phase.

References

• Baroni, Stefano, Stefano de Gironcoli, Andrea Dal Corso, and Paolo Giannozzi. “Phonons and related crystal properties from density-functional perturbation theory.” Reviews of Modern Physics 73, 515–562 (2001). doi:10.1103/RevModPhys.73.515.
• Born, Max, and Kun Huang. Dynamical Theory of Crystal Lattices. Clarendon Press, 1954.
• Giustino, Feliciano. “Electron–phonon interactions from first principles.” Reviews of Modern Physics 89, 015003 (2017). doi:10.1103/RevModPhys.89.015003.
• Onida, Giovanni, Lucia Reining, and Angel Rubio. “Electronic excitations: density-functional versus many-body Green’s-function approaches.” Reviews of Modern Physics 74, 601–659 (2002). doi:10.1103/RevModPhys.74.601.