Comparing Phonon Dispersion to Scattering Data

Hello! Welcome back to our study of lattice dynamics.

In our last lesson, we saw how the Debye model, by treating the solid as a continuous elastic medium, successfully explained the $T^3$ dependence of specific heat at low temperatures. However, we acknowledged that its assumptions—a linear dispersion and a sharp frequency cutoff—are simplifications. The true thermal properties of a crystal are dictated by its actual, often complex, phonon dispersion relations, $\omega (\mathbf{q})$.

This raises a critical question: How do we experimentally measure these dispersion curves? Today's lesson will answer exactly that. We will bridge the gap between our theoretical concept of phonons and the world of experimental condensed matter physics.

Our learning outcome is to relate phonon dispersion curves to experimental data from inelastic neutron or X-ray scattering. We will explore the fundamental principles of this technique, develop the necessary theoretical formalism, and learn how to interpret the resulting experimental data. This is a topic that sits at the nexus of theory, computation, and experiment, which I know is an area of great interest to you.

1. The Principle: Probing Phonons with Particles

The core idea behind measuring phonon dispersion is conceptually straightforward. We can probe the crystal with a beam of particles (like neutrons or photons) and analyze how their energy and momentum change after they scatter.

If a particle interacts with the crystal and either creates or absorbs a single phonon, the process must obey the conservation of energy and momentum. Given your background in field theory, you can think of this as a vertex in a Feynman diagram where the incoming particle, outgoing particle, and a phonon meet.

The conservation laws are:

1. Conservation of Energy: The energy lost or gained by the probe particle must equal the energy of the phonon, $\hslash \omega$.
   $E_i-E_f=\hslash \omega$
   Where $E_i$ and $E_f$ are the initial and final energies of the probe particle. If $\hslash \omega >0$, a phonon was created (an energy loss for the probe). If $\hslash \omega <0$, a phonon was absorbed (an energy gain for the probe).

2. Conservation of Crystal Momentum: The change in the probe's momentum, $\hslash \mathbf{Q}=\hslash ({\mathbf{k}}_i-{\mathbf{k}}_f)$, must equal the phonon's crystal momentum, $\hslash \mathbf{q}$, up to a reciprocal lattice vector $\mathbf{G}$.
   ${\mathbf{k}}_i-{\mathbf{k}}_f=\mathbf{Q}=\mathbf{q}+\mathbf{G}$
   Here, ${\mathbf{k}}_i$ and ${\mathbf{k}}_f$ are the initial and final wavevectors of the probe. The vector $\mathbf{Q}$ is called the scattering vector or momentum transfer. Since phonon momentum $\mathbf{q}$ is defined within the first Brillouin zone (BZ), any larger momentum transfer is equivalent to a $\mathbf{q}$ in the first BZ plus a reciprocal lattice vector $\mathbf{G}$.

By measuring the energy change $E_i-E_f$ and momentum transfer $\mathbf{Q}$ of the scattered particles, we can determine the energy $\hslash \omega$ and crystal momentum $\mathbf{q}$ of the phonon involved in the interaction. By repeating this for many scattering events, we can map out points on the $\omega (\mathbf{q})$ dispersion curves.

2. Choosing the Right Probe: Neutrons, Photons, or Electrons?

To be an effective probe for phonons, a particle must have a wavelength comparable to the interatomic spacing and an energy comparable to phonon energies. Let's see why thermal neutrons and X-rays are the tools of choice.

The following video provides an excellent overview of the scattering principle and evaluates the suitability of different probe particles.

As the video explains:

• Phonon Momentum: The Brillouin zone has a size of order $2\pi /a$, where $a$ is the lattice constant (\sim$$\mathrm{Å}). This means we need a probe with momentum $k\sim 1{\text{ A˚}}^{-1}\approx 10^{10}{\text{ m}}^{-1}$.
• Phonon Energy: Typical phonon energies are in the range of 1-100 meV.

The analysis shows:

• Visible Light: Has far too little momentum ($k\sim 10^7{\text{ m}}^{-1}$) to access most of the Brillouin zone. Light scattering techniques like Raman and Brillouin scattering are powerful but are restricted to probing phonons very near the BZ center ($\mathbf{q}\approx 0$).
• Electrons: Can have the right momentum, but their strong Coulomb interaction with the solid leads to multiple scattering events and a short penetration depth, making them difficult to use for probing bulk phonons.
• Thermal Neutrons: A thermal neutron ($T\sim 300$ K) has an energy $E\approx k_{B}T\approx 25$ meV and a de Broglie wavelength $\lambda \sim 1.8$ Å. This is a perfect match! Their energy is comparable to phonon energies, and their wavelength is comparable to the lattice spacing, giving them the right momentum to probe the entire Brillouin zone.
• X-rays: While conventional X-rays have very high energy ($\sim 10$ keV), modern synchrotron sources can produce highly monochromatic X-ray beams with extremely fine energy resolution. Inelastic X-ray Scattering (IXS) is a complementary technique to Inelastic Neutron Scattering (INS), especially for small samples.

3. The Formal Theory: From Cross-Sections to Phonons

The simple conservation laws tell us when scattering can happen. But to understand experimental data, we need to know the probability of scattering. This is described by the double-differential scattering cross-section, $\frac{d^2\sigma }{d\Omega d{E}^{\prime }}$, which represents the fraction of particles scattered into a solid angle $d\Omega$ with a final energy in the range $[E^{\prime },E^{\prime }+d{E}^{\prime }]$.

This quantity is what is actually measured in an experiment. A rigorous quantum mechanical derivation, which you can explore in the following resource, shows that this cross-section is directly proportional to a quantity called the dynamic structure factor, $S(\mathbf{q},\omega )$.

To summarize the key result from the reading:
$\frac{d^2\sigma }{d\Omega d{E}^{\prime }}\propto S(\mathbf{Q},\omega )$
The dynamic structure factor $S(\mathbf{Q},\omega )$ is the space and time Fourier transform of the density-density correlation function, $G(\mathbf{r},t)=⟨\rho (\mathbf{r},t)\rho (0,0)⟩$. It contains all the information about the collective excitations (like phonons) in the material at a specific momentum transfer $\hslash \mathbf{Q}$ and energy transfer $\hslash \omega$.

Connecting $S(\mathbf{Q},\omega )$ to Phonon Dispersions

So, how do we get from this general function $S(\mathbf{Q},\omega )$ to the sharp dispersion curves $\omega (\mathbf{q})$? The connection is made by considering scattering that involves only a single phonon. This is called the one-phonon approximation.

A detailed derivation shows that the one-phonon dynamic structure factor, $S_1(\mathbf{Q},\omega )$, takes a special form.

The crucial result from this reading is that the one-phonon scattering intensity contains Dirac delta functions in energy:
$S_1(\mathbf{Q},\omega )\propto {\sum }_{\nu ,\mathbf{q}}|F_{\mathbf{q}\nu }(\mathbf{Q}){|}^2\frac{1}{{\omega }_{\mathbf{q}\nu }}\left[\underset{\text{Phonon Creation}}{\underbrace{(n({\omega }_{\mathbf{q}\nu })+1)\delta (\omega -{\omega }_{\mathbf{q}\nu })}}+\underset{\text{Phonon Absorption}}{\underbrace{n({\omega }_{\mathbf{q}\nu })\delta (\omega +{\omega }_{\mathbf{q}\nu })}}\right]\delta (\mathbf{Q}-\mathbf{q}-\mathbf{G})$

Let's break this down:

• $\delta (\omega -{\omega }_{\mathbf{q}\nu })$ and $\delta (\omega +{\omega }_{\mathbf{q}\nu })$: These delta functions are the most important part. They state that scattering intensity is non-zero only when the energy transfer $\hslash \omega$ exactly matches the energy of a phonon mode $\hslash {\omega }_{\mathbf{q}\nu }$. This is why we see sharp peaks in the experimental spectrum that trace out the dispersion curves.
• $(n(\omega )+1)$: This is the Bose-Einstein factor for creating a phonon (Stokes scattering). It is non-zero even at T=0 K, which is why we can always measure phonon dispersions by creating phonons.
• $n(\omega )$: This is the Bose-Einstein factor for absorbing a phonon (anti-Stokes scattering). This process is only possible if phonons are already thermally excited in the crystal. This leads to an asymmetry in the measured spectrum, as predicted by the principle of detailed balance.
• $|F_{\mathbf{q}\nu }(\mathbf{Q}){|}^2$: The dynamical structure factor. This term involves a dot product $\mathbf{Q}\cdot {\mathbf{e}}_{\mathbf{q}\nu }$, where ${\mathbf{e}}_{\mathbf{q}\nu }$ is the polarization vector of the phonon. This is a powerful selection rule: if the momentum transfer $\mathbf{Q}$ is perpendicular to the atomic motion $\mathbf{e}$, that phonon mode will be "invisible" in the experiment at that $\mathbf{Q}$. This allows experimentalists to distinguish between longitudinal (LA/LO) and transverse (TA/TO) modes by choosing the measurement $\mathbf{Q}$ vector appropriately.

4. Interpreting Experimental Data

Now that we have the theory, let's see what the experimental results look like. The workhorse instrument for INS is the triple-axis spectrometer (TAS). In a typical experiment, the instrument is set to measure at a fixed momentum transfer $\mathbf{Q}$ while scanning the energy transfer $\hslash \omega$. Every time a peak in intensity is found, a point $(\mathbf{q},{\omega }_{\mathbf{q}\nu })$ is added to the dispersion plot.

Let's examine some classic and modern examples.

These plots show how rich the experimental data is:

• Monatomic crystals (Ne, K): Only show 3 acoustic branches (1 LA, 2 TA).
• Diatomic crystal (NaCl): Show 3 acoustic and 3 optic branches (LO, TO). Notice the energy gap between the acoustic and optic bands. Also, note the splitting between the LO and TO modes even as $\mathbf{q}\to 0$. This is the Lyddane-Sachs-Teller splitting in polar crystals, caused by the long-range electric fields associated with the LO phonon.

Modern spectrometers can collect data over a wide range of $\mathbf{Q}$ and $\omega$ simultaneously, producing intensity maps of the full $S(\mathbf{Q},\omega )$. These are often compared directly with theoretical calculations, for instance from Density Functional Theory (DFT).

These images beautifully illustrate the modern approach in condensed matter physics: a close dialogue between cutting-edge experiments and sophisticated computations to unravel the complex properties of materials.

Conclusion

In this lesson, we have established the crucial link between the theoretical concept of phonons and their experimental measurement.

Key Takeaways:

• Inelastic Neutron/X-ray Scattering is the primary tool for measuring phonon dispersion curves $\omega (\mathbf{q})$.
• The technique relies on the conservation of energy and crystal momentum between the probe particle and the created/absorbed phonon.
• The measured quantity is the double-differential cross-section, which is proportional to the dynamic structure factor $S(\mathbf{Q},\omega )$.
• In the one-phonon approximation, $S(\mathbf{Q},\omega )$ contains delta functions that make the scattering intensity sharply peaked at the phonon frequencies, allowing the dispersion to be mapped out.
• Selection rules based on the dot product $\mathbf{Q}\cdot \mathbf{e}$ allow experimentalists to distinguish between longitudinal and transverse modes.
• Modern experiments provide rich datasets of $S(\mathbf{Q},\omega )$ that are often interpreted in close comparison with theoretical calculations.

Preview of the Next Module

We have now concluded our module on pure lattice dynamics. We've built a picture from the classical chain of atoms to the quantum mechanical phonon, and now to its experimental measurement.

However, a crystal is not just a vibrating lattice; it's a lattice of ions filled with electrons. The next crucial step is to understand how these two components talk to each other. In our next module, "Electron-Phonon Interactions," we will begin to explore this fascinating and fundamentally important topic. Our first lesson will focus on deriving the electron-phonon interaction Hamiltonian, the mathematical foundation for understanding phenomena ranging from electrical resistance to conventional superconductivity.