- Explain how phonon information is extracted from absorption and inelastic scattering experiments.
- Distinguish what infrared absorption, Raman scattering, Brillouin scattering, and neutron scattering can measure.
- Use energy and crystal-momentum conservation to understand which phonons are experimentally accessible.
- Derive the Brillouin frequency shift and interpret Stokes and anti-Stokes processes.
- Read the one-phonon neutron-scattering cross section and identify what sets peak position, intensity, and width.
13.1 Roadmap
- Start from the general kinematic logic of phonon measurements.
- Study optical probes: infrared absorption, Raman scattering, and Brillouin scattering.
- Move to neutron scattering as the direct probe of phonon dispersion throughout the Brillouin zone.
- Interpret the dynamical structure factor, Debye–Waller factor, and phonon linewidth.
- Bridge to the next lecture: from measured phonon spectra to lifetimes, anharmonicity, and thermal transport.
| Symbol | Meaning |
|---|---|
| \(\omega_i,\omega_f\) | angular frequency of the incident and scattered probe |
| \(\mathbf{k}_i,\mathbf{k}_f\) | wave vector of the incident and scattered probe |
| \(\mathbf{q}\) | phonon wave vector |
| \(\omega_\lambda(\mathbf{q})\) | phonon frequency in branch \(\lambda\) |
| \(\mathbf{G}\) | reciprocal-lattice vector |
| \(\Omega\) | energy transfer to the sample, defined by \(\hbar\Omega=E_i-E_f\) |
| \(\mathbf{K}\) | momentum transfer, defined by \(\mathbf{K}=\mathbf{k}_i-\mathbf{k}_f\) |
| \(n_B(\omega)\) | Bose occupation number, \(n_B(\omega)=1/(e^{\hbar\omega/k_B T}-1)\) |
| \(n\) | refractive index of the sample |
| \(\theta\) | scattering angle between incident and scattered light |
| \(c_\lambda\) | sound velocity for acoustic polarization \(\lambda\) |
| \(S(\mathbf{K},\Omega)\) | dynamical structure factor |
| \(W\) | Debye–Waller exponent |
| \(\Gamma_\lambda(\mathbf{q})\) | inverse lifetime of a phonon mode |
13.2 Direct Phonon Measurements
Lecture 8 established how phonons determine thermal properties such as the heat capacity. Today we turn to the complementary question: how are phonons measured directly?
The experimental logic is simple. A probe exchanges energy and momentum with the crystal. If the process creates or annihilates a single phonon, the measured changes in the probe identify the phonon energy and wave vector.
Here \(E_i\) and \(E_f\) are the initial and final energies of the probe, \(\mathbf{k}_i\) and \(\mathbf{k}_f\) are its initial and final wave vectors, \(\hbar\Omega\) is the energy transferred to the sample, and \(\mathbf{K}\) is the momentum transfer.
For a one-phonon process in a crystal,
Here \(\omega_\lambda(\mathbf{q})\) is the phonon frequency in branch \(\lambda\), \(\mathbf{q}\) is the phonon wave vector, and \(\mathbf{G}\) is a reciprocal-lattice vector. The plus sign corresponds to phonon creation and the minus sign to phonon annihilation with our convention for \(\Omega\).
These two relations are the backbone of the lecture. Different probes differ mainly in the ranges of energy and momentum they can access.
Why Probe Choice Matters
To map a phonon dispersion relation over an appreciable part of the Brillouin zone, the probe wavelength must be comparable to the lattice spacing, while the probe energy must also be comparable to phonon energies.
Typical phonon energies are on the order of a few to a few tens of meV. Visible photons have much larger energies but very small wave numbers on the scale of the Brillouin zone, so they access only phonons near \(\mathbf{q}\approx 0\). Thermal neutrons, by contrast, naturally have both energies and wave numbers in the relevant range.
13.3 Optical Methods
13.3.1 Infrared Absorption
The simplest optical process is direct photon absorption accompanied by phonon creation. Because infrared photon momenta are tiny on the scale of the Brillouin zone, the created phonon must satisfy \[ \mathbf{q}\approx 0. \]
Here \(\mathbf{q}\) is the phonon wave vector. The statement means that infrared absorption probes only zone-center modes.
In addition, direct coupling to the electric field requires a changing dipole moment. Long-wavelength acoustic phonons do not create such a dipole moment, so direct infrared absorption selects optical modes, and in the simplest picture specifically transverse optical modes because the electric field of the light is transverse to its propagation direction.
Only those zone-center optical modes that change the dipole moment are infrared active.
A useful sanity check is the hierarchy of scales:
- infrared wavelengths are much larger than the lattice constant;
- therefore photon momentum is negligible compared with reciprocal-lattice vectors;
- hence infrared spectroscopy is symmetry-selective but not wave-vector resolved.
If anharmonic interactions are present, the initially created optical phonon can decay into two phonons or scatter from an already occupied phonon state. Then absorption is no longer strictly confined to a delta-like line, and finite-width features or sidebands appear.
What Infrared Absorption Does and Does Not Give You
Infrared absorption is excellent for:
- identifying zone-center transverse optical modes;
- diagnosing symmetry and selection rules;
- locating infrared-active phonons in ionic and polar materials.
It is not a method for mapping the full dispersion relation \(\omega_\lambda(\mathbf{q})\) across the Brillouin zone.
13.3.2 Raman Scattering
Raman scattering is inelastic light scattering by optical phonons near the zone center. Because the photon momentum is again negligible on the lattice scale, the dominant one-phonon process involves \(\mathbf{q}\approx 0\) optical modes.
The energy balance is
Here \(\omega_i\) and \(\omega_f\) are the angular frequencies of the incident and scattered photons, and \(\omega_\lambda(\mathbf{q}\approx 0)\) is the zone-center optical-phonon frequency. The upper sign corresponds to phonon emission and the lower sign to phonon absorption.
This immediately produces two shifted lines:
- the Stokes line at lower photon frequency, from phonon creation;
- the anti-Stokes line at higher photon frequency, from phonon annihilation.
The thermal population of the phonon controls the intensities:
Here \(n_B(\omega)\) is the Bose occupation number, \(I_{\mathrm{AS}}\) is the anti-Stokes intensity, and \(I_{\mathrm{S}}\) is the Stokes intensity. The extra \(+1\) in the Stokes channel is the bosonic stimulated-emission factor.
A direct consequence is that the anti-Stokes line weakens strongly as temperature is lowered and vanishes at \(T=0\), while the Stokes line remains finite.
A short sanity check:
- anti-Stokes scattering needs a phonon to be present already;
- Stokes scattering can create a phonon even from the ground state;
- therefore anti-Stokes is the more temperature-sensitive branch.
Selection rules are again essential. A mode is Raman active if the optical vibration changes the polarizability of the basis. Thus Raman and infrared spectroscopy are complementary rather than redundant: a mode can be active in one channel, both, or neither, depending on symmetry.
Higher-order optical processes are also possible. In two-phonon Raman scattering, phonon pairs with wave vectors \(\mathbf{q}\) and \(-\mathbf{q}\) contribute, so the spectrum broadens into a continuum with strong weight where the phonon density of states is large.
13.3.3 Brillouin Scattering
Brillouin scattering is the optical analog of acoustic-phonon spectroscopy. Unlike Raman scattering, the relevant phonons are acoustic, and now the change in photon momentum must be retained.
Inside a medium of refractive index \(n\), the photon wave number is \(nk\), where \(k=\omega_i/c\). Simple geometry gives
Here \(q=|\mathbf{q}|\) is the magnitude of the acoustic-phonon wave vector, \(n\) is the refractive index, \(\theta\) is the scattering angle, \(\omega_i\) is the incident photon frequency, and \(c\) is the speed of light in vacuum.
For long-wavelength acoustic phonons, \[ \omega_\lambda(\mathbf{q})\approx c_\lambda q. \]
Here \(c_\lambda\) is the sound velocity for polarization \(\lambda\). Combining this with the previous equation yields the Brillouin shift
Here \(\Delta\omega_\lambda=\omega_f-\omega_i\) is the frequency shift of the scattered light. The sign distinguishes phonon creation and annihilation.
This formula means that the measured frequency shift directly gives the sound velocity.
Interpretation:
- Brillouin scattering probes only very small \(q\);
- therefore it measures the slope of the acoustic branches near the zone center;
- experimentally, this means access to elastic constants and sound velocities.
13.4 Neutron Scattering From a Thermally Vibrating Crystal
Optical methods are limited to \(\mathbf{q}\approx 0\). To map phonons throughout reciprocal space, the probe must have both a lattice-scale wavelength and meV-scale energy transfer. Thermal neutrons satisfy both requirements.
Sólyom rewrites the neutron energy as \[ \varepsilon \approx 2.1\times 10^{-3}\ [k(\mathring{\mathrm{A}}^{-1})]^2\ \mathrm{eV}. \]
Here \(\varepsilon\) is the neutron energy and \(k=|\mathbf{k}|\) is the neutron wave number in inverse ångströms. For \(k\sim 1,\mathrm{\mathring{\mathrm{A}}}^{-1}\), the energy is on the order of \(0.1,\mathrm{eV}\) and the wave vector is comparable to Brillouin-zone dimensions.
This is why thermal neutrons are such a natural probe of lattice dynamics.
Triple-Axis Spectroscopy
In inelastic neutron scattering, the incident beam is first monochromatized, then scattered by the sample, and finally energy-analyzed before detection. The standard instrument is the triple-axis spectrometer.
The key experimental capability is this: by adjusting the instrument geometry, one chooses a momentum transfer \(\mathbf{K}\) and then measures intensity as a function of energy transfer \(\Omega\), or vice versa.
13.4.1 The Dynamical Structure Factor and the One-Phonon Signal
For coherent scattering from a vibrating lattice, the measured doubly differential cross section is proportional to the dynamical structure factor:
Here \(d\Omega_{\mathrm{sc}}\) is the solid-angle element for the scattered beam, \(E_f\) is the final neutron energy, \(\mathbf{K}\) is the momentum transfer, and \(\Omega\) is the energy transfer to the sample.
The dynamical structure factor is the space-time Fourier transform of density correlations:
Here \(N\) is the number of primitive cells, \(\mathbf{R}_m\) and \(\mathbf{R}_n\) are equilibrium lattice vectors, and \(\mathbf{u}_m(t)\) is the displacement of atom \(m\) at time \(t\).
Expanding the exponentials to first order in the atomic displacements isolates the one-phonon contribution:
Here \(\mathbf{e}_\lambda(\mathbf{q})\) is the phonon polarization vector, \(n_B(\omega_\lambda)\) is the Bose occupation number, and \(e^{-2W}\) is the Debye–Waller factor.
This single formula contains the essential physics:
- the delta functions fix the phonon energy and wave vector;
- the Bose factors distinguish phonon creation from annihilation;
- the factor \(|\mathbf{K}\cdot\mathbf{e}_\lambda|^2\) gives a polarization selection rule;
- the Debye–Waller factor reduces intensity as thermal motion grows.
13.4.2 Elastic Scattering, Inelastic Peaks, and Bragg Diffraction
The zeroth-order term in the expansion of \(S(\mathbf{K},\Omega)\) gives \[ \Omega=0, \qquad \mathbf{K}=\mathbf{G}. \]
This is ordinary Bragg diffraction: no energy transfer, momentum transfer equal to a reciprocal-lattice vector.
The first-order terms give inelastic peaks shifted away from the Bragg condition by \(\pm\mathbf{q}\) and away from zero energy by \(\pm\omega_\lambda(\mathbf{q})\). This is the direct route to the phonon dispersion relation.
13.4.3 The Debye–Waller Factor
The Debye–Waller factor is introduced through \[ e^{-W}=e^{-\frac{1}{2}\langle(\mathbf{K}\cdot\mathbf{u})^2\rangle}. \]
Here \(\mathbf{u}\) is the thermal displacement of an atom, and \(\langle\cdots\rangle\) denotes the thermal average. The full scattering intensity is reduced by the factor \(e^{-2W}\).
In isotropic form, we can write
Here \(K=|\mathbf{K}|\) and \(\langle u^2\rangle\) is the mean-square atomic displacement.
The physical meaning is transparent: stronger thermal vibrations reduce the coherent scattering intensity, even though the Bragg peaks themselves remain sharp in the harmonic approximation.
Optional: Finite Lifetime and Phonon Linewidth
Anharmonicity makes phonons unstable against decay, so the ideal delta-function peak broadens. If the phonon amplitude decays as \(e^{-\Gamma_\lambda(\mathbf{q})|t|}\), then the spectral line becomes Lorentzian: \[ \int_{-\infty}^{\infty}dt\ e^{i\Omega t} e^{\pm i\omega_\lambda(\mathbf{q})t-\Gamma_\lambda(\mathbf{q})|t|} = \frac{2\Gamma_\lambda(\mathbf{q})} {\big(\Omega\pm\omega_\lambda(\mathbf{q})\big)^2+\Gamma_\lambda^2(\mathbf{q})}. \]
Here \(\Gamma_\lambda(\mathbf{q})\) is the inverse lifetime of the phonon mode. A smaller \(\Gamma_\lambda(\mathbf{q})\) means a sharper peak and a longer-lived phonon.
So neutron scattering does not only locate phonon branches. It also measures their damping.
13.5 Comparing the Methods
The four methods fit together naturally:
- infrared absorption: zone-center, dipole-active optical phonons;
- Raman scattering: zone-center, polarizability-active optical phonons;
- Brillouin scattering: very small-\(q\) acoustic phonons and sound velocities;
- neutron scattering: full phonon dispersion, branch polarization, temperature dependence, and linewidths.
The choice of method is therefore not arbitrary. It is dictated by kinematics and selection rules.
- Direct phonon measurements rely on simultaneous conservation of energy and crystal momentum.
- Optical probes are restricted to phonons very close to the Brillouin-zone center.
- Infrared absorption selects dipole-active transverse optical modes.
- Raman scattering measures zone-center optical phonons through Stokes and anti-Stokes shifts, with intensities governed by Bose occupation.
- Brillouin scattering measures long-wavelength acoustic phonons and gives sound velocities from the frequency shift.
- Thermal neutrons are ideal because their wavelengths and energies match lattice scales and phonon energies.
- In neutron scattering, the one-phonon signal is encoded in the dynamical structure factor and carries information about branch energy, polarization, and occupation.
- Finite phonon linewidths are direct experimental signatures of finite phonon lifetimes and anharmonicity.
13.6 Problem Set
Why Optical Methods Probe \(\mathbf{q}\approx 0\). Explain why both infrared absorption and one-phonon Raman scattering are restricted to phonons near the center of the Brillouin zone. Then state what additional selection rule distinguishes infrared-active from Raman-active modes.
Stokes Versus Anti-Stokes. Starting from \[ n_B(\omega)=\frac{1}{e^{\hbar\omega/k_B T}-1}, \qquad I_{\mathrm{AS}}\propto n_B(\omega), \qquad I_{\mathrm{S}}\propto n_B(\omega)+1, \] show that \[ \frac{I_{\mathrm{AS}}}{I_{\mathrm{S}}}=e^{-\hbar\omega/k_B T}. \] Interpret the result physically.
Brillouin Shift. Derive the Brillouin frequency shift for long-wavelength acoustic phonons, \[ \Delta\omega_\lambda = \pm 2n\frac{\omega_i}{c},c_\lambda\sin\frac{\theta}{2}, \] from the relations \[ q=2n\frac{\omega_i}{c}\sin\frac{\theta}{2}, \qquad \omega_\lambda(\mathbf{q})\approx c_\lambda q. \]
Reading the One-Phonon Neutron Cross Section. In the one-phonon neutron-scattering signal, what determines
- the phonon energy,
- the phonon wave vector,
- the difference between phonon creation and annihilation intensities, and
- the visibility of different polarizations?
Debye–Waller Factor and Linewidth. Explain the physical difference between the roles of the Debye–Waller factor \(e^{-2W}\) and the linewidth parameter \(\Gamma_\lambda(\mathbf{q})\) in neutron scattering.