11 May 2026

Quantization of Elastic Waves and Phonons

Normal coordinates, canonical quantization of lattice vibrations, creation and annihilation operators, zero-point motion, and phonon crystal momentum.

msc semester-iv condensed-matter lattice-dynamics elastic-waves phonons canonical-quantization normal-modes zero-point-motion

In the harmonic approximation, a crystal is a system of coupled oscillators. Its normal modes are independent at the classical level, and each normal mode becomes a quantum harmonic oscillator after canonical quantization. A phonon is one quantum of such a normal mode. The mode label contains both a wavevector and a branch index; it does not identify a particular atom carrying the excitation.

The derivation below first diagonalizes the harmonic-crystal Hamiltonian and then quantizes its normal coordinates. The continuum elastic-wave result follows as the long-wavelength limit of the same construction.

Harmonic Hamiltonian of a crystal

Let $l$ label primitive cells and let $\kappa=1,\ldots,r$ label the atoms in a cell. The equilibrium position of atom $(l,\kappa)$ is

\[\mathbf R_l+\boldsymbol\tau_\kappa,\]

and its small displacement is $\mathbf u_{l\kappa}$. Expanding the potential energy about a stable equilibrium gives

\[H=\frac12\sum_{l\kappa}M_\kappa \dot{\mathbf u}_{l\kappa}^{\,2} +\frac12\sum_{l\kappa\alpha} \sum_{l'\kappa'\beta} u_{l\kappa\alpha} \Phi_{l\kappa\alpha,l'\kappa'\beta} u_{l'\kappa'\beta}.\]

Here

\[\Phi_{l\kappa\alpha,l'\kappa'\beta} =\left. \frac{\partial^2 U} {\partial u_{l\kappa\alpha}\partial u_{l'\kappa'\beta}} \right\rvert_{\mathrm{eq}}\]

is the force-constant matrix. There is no term linear in displacement because every atom is in mechanical equilibrium. The harmonic approximation neglects cubic and higher terms; consequently, normal modes neither decay nor exchange energy in this approximation.

Born–von Karman boundary conditions on $N$ primitive cells make the allowed wavevectors discrete. Translational invariance permits normal-mode solutions of the form

\[u_{l\kappa\alpha}(t) =\frac{1}{\sqrt{M_\kappa}} e_{\kappa\alpha}(\mathbf q,s) e^{i(\mathbf q\cdot\mathbf R_l-\omega t)}.\]

The polarization vectors obey the dynamical-matrix eigenvalue equation

\[\sum_{\kappa'\beta} D_{\kappa\alpha,\kappa'\beta}(\mathbf q) e_{\kappa'\beta}(\mathbf q,s) =\omega_{\mathbf q s}^{\,2} e_{\kappa\alpha}(\mathbf q,s),\]

where

\[D_{\kappa\alpha,\kappa'\beta}(\mathbf q) =\frac{1}{\sqrt{M_\kappa M_{\kappa'}}} \sum_{l'} \Phi_{0\kappa\alpha,l'\kappa'\beta} e^{i\mathbf q\cdot(\mathbf R_{l'}-\mathbf R_0)}.\]

For each $\mathbf q$, this is a Hermitian $3r\times3r$ matrix. Its eigenvectors may be chosen to satisfy

\[\sum_{\kappa\alpha} e_{\kappa\alpha}^{*}(\mathbf q,s) e_{\kappa\alpha}(\mathbf q,s')=\delta_{ss'}.\]

For real force constants one can choose

\[\mathbf e_\kappa(-\mathbf q,s) =\mathbf e_\kappa^{*}(\mathbf q,s), \qquad \omega_{-\mathbf q s}=\omega_{\mathbf q s}.\]

Normal coordinates

Expand an arbitrary real displacement field in the complete set of normal modes:

\[u_{l\kappa\alpha} =\frac{1}{\sqrt{NM_\kappa}} \sum_{\mathbf q s} e_{\kappa\alpha}(\mathbf q,s) Q_{\mathbf q s} e^{i\mathbf q\cdot\mathbf R_l}.\]

Reality requires

\[Q_{-\mathbf q s}=Q_{\mathbf q s}^{*}.\]

Define the normal-coordinate velocity

\[\Pi_{\mathbf q s}=\dot Q_{\mathbf q s}.\]

Classically, $\Pi_{-\mathbf q s}=\Pi_{\mathbf q s}^{*}$. After quantization the corresponding operator reality conditions are

\[Q_{\mathbf q s}^{\dagger}=Q_{-\mathbf q s}, \qquad \Pi_{\mathbf q s}^{\dagger}=\Pi_{-\mathbf q s}.\]

Using polarization orthonormality and the dynamical-matrix eigenvalue equation, the Hamiltonian becomes

\[\boxed{ H=\frac12\sum_{\mathbf q s} \left[ \Pi_{\mathbf q s}\Pi_{-\mathbf q s} +\omega_{\mathbf q s}^{\,2} Q_{\mathbf q s}Q_{-\mathbf q s} \right]. }\]

Thus the coupled atomic displacements have been transformed into independent normal oscillators. Because the sum contains both $\mathbf q$ and $-\mathbf q$, the momentum canonically conjugate to $Q_{\mathbf q s}$ is

\[P_{\mathbf q s} =\frac{\partial L}{\partial\dot Q_{\mathbf q s}} =\Pi_{-\mathbf q s}.\]

Therefore the canonical commutator may be written

\[[Q_{\mathbf q s},\Pi_{\mathbf q's'}] =i\hbar\,\delta_{\mathbf q,-\mathbf q'}\delta_{ss'}.\]

Creation and annihilation operators

Define bosonic operators by

\[Q_{\mathbf q s} =\sqrt{\frac{\hbar}{2\omega_{\mathbf q s}}} \left(a_{\mathbf q s}+a_{-\mathbf q s}^{\dagger}\right),\] \[\Pi_{\mathbf q s} =-i\sqrt{\frac{\hbar\omega_{\mathbf q s}}{2}} \left(a_{\mathbf q s}-a_{-\mathbf q s}^{\dagger}\right).\]

The canonical commutator is equivalent to

\[[a_{\mathbf q s},a_{\mathbf q's'}^{\dagger}] =\delta_{\mathbf q\mathbf q'}\delta_{ss'}, \qquad [a_{\mathbf q s},a_{\mathbf q's'}]=0.\]

Substitution into the normal-coordinate Hamiltonian gives

\[\boxed{ H=\sum_{\mathbf q s} \hbar\omega_{\mathbf q s} \left(a_{\mathbf q s}^{\dagger}a_{\mathbf q s}+\frac12\right). }\]

The number operator

\[\hat n_{\mathbf q s} =a_{\mathbf q s}^{\dagger}a_{\mathbf q s}\]

has eigenvalues $0,1,2,\ldots$. Acting on a number state,

\[a_{\mathbf q s}^{\dagger} \lvert n_{\mathbf q s}\rangle =\sqrt{n_{\mathbf q s}+1} \lvert n_{\mathbf q s}+1\rangle,\] \[a_{\mathbf q s} \lvert n_{\mathbf q s}\rangle =\sqrt{n_{\mathbf q s}} \lvert n_{\mathbf q s}-1\rangle.\]

Creation or destruction of one phonon changes the vibrational energy by $\hbar\omega_{\mathbf q s}$.

Classical normal mode mapped to a quantum harmonic oscillator ladder with phonon creation and annihilation transitions
Each classical normal coordinate becomes an independent quantum harmonic oscillator. Adjacent number states differ by one phonon and by the energy $\hbar\omega_{\mathbf q s}$; the ground state retains the zero-point energy $\hbar\omega_{\mathbf q s}/2$.

Quantized displacement and momentum fields

In terms of phonon operators, the atomic displacement is

\[\boxed{ u_{l\kappa\alpha} =\frac{1}{\sqrt{NM_\kappa}} \sum_{\mathbf q s} \sqrt{\frac{\hbar}{2\omega_{\mathbf q s}}} \left[ e_{\kappa\alpha}(\mathbf q,s) a_{\mathbf q s}e^{i\mathbf q\cdot\mathbf R_l} +e_{\kappa\alpha}^{*}(\mathbf q,s) a_{\mathbf q s}^{\dagger}e^{-i\mathbf q\cdot\mathbf R_l} \right]. }\]

The conjugate atomic momentum is

\[\boxed{ p_{l\kappa\alpha} =-i\sqrt{\frac{M_\kappa}{N}} \sum_{\mathbf q s} \sqrt{\frac{\hbar\omega_{\mathbf q s}}{2}} \left[ e_{\kappa\alpha}(\mathbf q,s) a_{\mathbf q s}e^{i\mathbf q\cdot\mathbf R_l} -e_{\kappa\alpha}^{*}(\mathbf q,s) a_{\mathbf q s}^{\dagger}e^{-i\mathbf q\cdot\mathbf R_l} \right]. }\]

These expressions satisfy

\[[u_{l\kappa\alpha},p_{l'\kappa'\beta}] =i\hbar\, \delta_{ll'}\delta_{\kappa\kappa'}\delta_{\alpha\beta}.\]

They are also the starting point for calculating neutron, x-ray, electron–phonon and infrared matrix elements. Every one-phonon amplitude contains the characteristic factor

\[\sqrt{\frac{\hbar}{2M_\kappa\omega_{\mathbf q s}}}.\]

Low-frequency modes therefore have larger displacement amplitudes than high-frequency modes of the same participating mass.

Quantization of a one-dimensional elastic wave

Consider a uniform elastic medium of length $L$, mass per unit length $\mu$, tension or one-dimensional elastic coefficient $T$, and transverse displacement $u(x,t)$. Its Hamiltonian is

\[H=\frac12\int_0^L \left[ \frac{\pi^2(x)}{\mu} +T\left(\frac{\partial u}{\partial x}\right)^2 \right]dx,\]

where $\pi(x)=\mu\dot u(x)$ and

\[[u(x),\pi(x')]=i\hbar\delta(x-x').\]

Periodic boundary conditions give $q=2\pi n/L$ and the elastic-wave dispersion

\[\omega_q=c\lvert q\rvert, \qquad c=\sqrt{\frac{T}{\mu}}.\]

The quantized field is

\[u(x)=\sum_q \sqrt{\frac{\hbar}{2\mu L\omega_q}} \left(a_qe^{iqx}+a_q^{\dagger}e^{-iqx}\right),\] \[\pi(x)=-i\sum_q \sqrt{\frac{\hbar\mu\omega_q}{2L}} \left(a_qe^{iqx}-a_q^{\dagger}e^{-iqx}\right),\]

and

\[H=\sum_q\hbar\omega_q \left(a_q^{\dagger}a_q+\frac12\right).\]

The $q=0$ mode is a rigid translation with $\omega=0$, not an internal harmonic vibration. It is removed by fixing the centre of mass or treated separately as free translational motion. A real crystal replaces the continuum spectrum by lattice branches and supplies a natural Brillouin-zone cutoff.

Zero-point motion and thermal occupation

Even the phonon vacuum has the energy

\[\boxed{E_0=\frac12\sum_{\mathbf q s} \hbar\omega_{\mathbf q s}.}\]

This zero-point energy cannot be removed when comparing structures whose phonon spectra differ. It contributes, for example, to isotope-dependent equilibrium volumes and structural stability. The mean-square displacement of an atom contains a zero-point term because

\[\langle n\lvert (a+a^{\dagger})^2 \rvert n\rangle=2n+1.\]

For a mode $(\mathbf q,s)$, its contribution is proportional to

\[\frac{\hbar}{2NM_\kappa\omega_{\mathbf q s}} \left(2n_{\mathbf q s}+1\right) \lvert\mathbf e_\kappa(\mathbf q,s)\rvert^2.\]

At temperature $T$, a harmonic phonon mode has the Bose–Einstein mean occupation

\[\boxed{ \bar n_{\mathbf q s} =\frac{1}{ \exp(\hbar\omega_{\mathbf q s}/k_{\mathrm B}T)-1}. }\]

Its mean energy is

\[\langle E_{\mathbf q s}\rangle =\hbar\omega_{\mathbf q s} \left(\bar n_{\mathbf q s}+\frac12\right).\]

Phonon wavevector, momentum and crystal momentum

The wavevector $\mathbf q$ is defined modulo a reciprocal-lattice vector $\mathbf G$ because

\[e^{i(\mathbf q+\mathbf G)\cdot\mathbf R_l} =e^{i\mathbf q\cdot\mathbf R_l}.\]

It is therefore sufficient to label phonons by $\mathbf q$ in the first Brillouin zone. A one-phonon state transforms under a primitive translation $\mathbf R$ with phase $e^{i\mathbf q\cdot\mathbf R}$, so it carries crystal momentum $\hbar\mathbf q$, defined modulo $\hbar\mathbf G$.

Crystal momentum is not generally the mechanical centre-of-mass momentum of the ions. The latter is

\[\mathbf P_{\mathrm{mech}} =\sum_{l\kappa}\mathbf p_{l\kappa},\]

for which the lattice sum selects the uniform $\mathbf q=0$ translation. A finite-$\mathbf q$ normal mode can have zero net mechanical momentum while possessing non-zero crystal momentum. In a phonon interaction, discrete translational symmetry gives

\[\sum_{\mathrm{initial}}\mathbf q -\sum_{\mathrm{final}}\mathbf q =\mathbf G.\]

Processes with $\mathbf G=0$ are called normal processes. Those with $\mathbf G\ne0$ are Umklapp processes; the lattice absorbs the difference in crystal momentum.

Worked example: quantized mode of a monatomic chain

For a nearest-neighbour monatomic chain,

\[\omega(q)=2\sqrt{\frac{C}{M}} \left\lvert\sin\frac{qa}{2}\right\rvert.\]

Take

\[C=60.0\,\mathrm{N\,m^{-1}},\qquad M=28.0u,\qquad a=0.500\,\mathrm{nm},\]

and $q=\pi/(2a)$. Then

\[\omega=\sqrt{\frac{2C}{M}} =5.080\times10^{13}\,\mathrm{rad\,s^{-1}},\] \[f=\frac{\omega}{2\pi}=8.086\,\mathrm{THz}, \qquad \hbar\omega=33.44\,\mathrm{meV}.\]

At $300\,\mathrm K$,

\[\bar n =\frac{1}{\exp(33.44/25.85)-1} =0.378.\]

For a periodic chain of $N=1000$ identical atoms, the displacement coefficient multiplying $a_q+a_{-q}^{\dagger}$ at any one atom is

\[u_{1\mathrm{ph}} =\sqrt{\frac{\hbar}{2NM\omega}} =1.49\times10^{-13}\,\mathrm m.\]

The corresponding mean-square contribution in a number state is $(2n+1)u_{1\mathrm{ph}}^2$. The factor $1/\sqrt N$ expresses the fact that a normalized extended mode is shared by all $N$ atoms.

Preparation questions

  1. Starting from the harmonic-crystal Hamiltonian, introduce mass-weighted normal coordinates and show that the Hamiltonian is a sum of independent oscillators.
  2. Derive the quantized displacement and conjugate-momentum operators of a crystal and verify their equal-time commutator.
  3. Explain the physical meanings of $a_{\mathbf q s}^{\dagger}$, $a_{\mathbf q s}$ and $a_{\mathbf q s}^{\dagger}a_{\mathbf q s}$.
  4. Quantize a one-dimensional elastic wave with periodic boundary conditions and discuss the treatment of its zero-frequency mode.
  5. Derive the zero-point and thermal contributions of one mode to the mean-square atomic displacement.
  6. Distinguish phonon crystal momentum from the true mechanical momentum of the ions. Explain normal and Umklapp processes.
  7. A monatomic chain has $C=40\,\mathrm{N\,m^{-1}}$, $M=32u$ and $a=0.40\,\mathrm{nm}$. Find the phonon energy and Bose occupation at $T=200\,\mathrm K$ for $q=\pi/(2a)$.

Maxima worksheet

© Rajesh Kumar, SKMU · Physics Lecture Notes · rajeshphy.github.io

Discussion

Share This Page