Spin Waves and Magnons

A spin wave is a collective oscillation of an ordered magnetic texture; a magnon is a quantum of a normal mode obtained by quantizing that oscillation. The distinction matters. A phase-coherent spin wave has a transverse expectation value that precesses in space and time, whereas a one-magnon number state need not have a classical transverse mean.

In a material, the phrase “magnon spectrum” usually means more than a list of energies. Each branch carries a polarization, sublattice phase pattern, probe-dependent spectral weight, and linewidth. A convincing model must explain those quantities together and must remain consistent with the static magnetic structure.

The central material question is whether one model explains both the observed mode energies and their visibility. The Magnons article develops spin-wave quantization and interactions; here the emphasis is on magnetic structures, scattering matrix elements, measured linewidths, and extracting couplings from material data.

Required background. Exchange Interactions supplies material couplings and sign conventions, while Magnons supplies Holstein–Primakoff and bosonic Bogoliubov quantization.

Helpful background. Ferromagnetism and Antiferromagnetism provide the ordered branches and sublattice conventions used in examples.

Magnon Energies, Polarizations, and Spectral Weight

For each reduced wavevector $\mathbf q$ and branch $\nu$, the measured spectrum asks for four quantities:

$$\left\{ \hbar\omega_\nu(\mathbf q), \quad \mathbf v_\nu(\mathbf q), \quad \mathcal I_\nu(\mathbf Q), \quad \Gamma_\nu(\mathbf q) \right\}.$$

Here:

• $\hbar\omega_\nu$ is the mode energy;
• $\mathbf v_\nu$ is its polarization and sublattice eigenvector;
• $\mathcal I_\nu$ is its intensity in a specified probe geometry;
• $\Gamma_\nu$ is an intrinsic linewidth, requiring interactions, disorder, or coupling to other degrees of freedom beyond the closed harmonic model.

The crystal momentum seen in a scattering experiment is commonly written

$$\mathbf Q = \boldsymbol\tau_m+\mathbf q,$$

where $\boldsymbol\tau_m$ is a reciprocal vector of the magnetic lattice. If the magnetic unit cell is larger than the chemical unit cell, the magnetic Brillouin zone is smaller and the apparent branches are folded. Calling two folded images “two different magnons” without checking their eigenvectors and periodicity is a common source of confusion.

The number of positive-energy harmonic branches is generically the number $n_m$ of fixed-length spins retained in the magnetic primitive cell. That branch count is not the Goldstone-mode count. Broken continuous symmetries constrain only the low-energy modes; additional optical branches can remain gapped.

Start from the Ordered Structure

Spin-wave theory is an expansion about a reference state, not a machine that discovers the reference state automatically. Before introducing bosons, establish:

1. the propagation vector or vectors of the order;
2. the magnetic primitive cell;
3. the moment direction and magnitude on every inequivalent site;
4. the domains and twins represented in the sample;
5. whether the moments are well described by local spins at the energies of interest.

Elastic diffraction, magnetometry, local probes, and symmetry analysis constrain different aspects of the structure. A propagation vector alone does not determine the moment direction. A bulk moment alone can miss compensated order. A crystallographic structure alone does not identify the magnetic domain population.

Let $\mathbf n_{i0}$ be the unit vector along the reference spin or pseudospin expectation on site $i$. Infer this direction from the measured magnetic moment using the site’s moment tensor; they need not be parallel. The first nontrivial consistency test is stationarity:

$$\mathbf n_{i0} \times \mathbf h_i^{\mathrm{eff}} = \mathbf 0, \qquad \mathbf h_i^{\mathrm{eff}} := - \frac{\partial E_{\mathrm{cl}}} {\partial \mathbf S_i}.$$

If this condition fails, the harmonic expansion contains terms linear in the fluctuation coordinates. Diagonalizing the remaining quadratic matrix then produces a spectrum about a state that is already trying to rotate. Small linear terms should not be dismissed as numerical noise until the structure, couplings, and local axes have been checked.

A Material Spin Hamiltonian

A useful starting point for localized moments is

$$\hat H ={} \frac12 \sum_{ij} \hat{\mathbf S}_i^{\mathsf T} \mathsf J_{ij} \hat{\mathbf S}_j + \sum_i \hat{\mathbf S}_i^{\mathsf T} \mathsf K_i \hat{\mathbf S}_i + \mu_B \sum_i \mathbf B^{\mathsf T} \mathsf g_i \hat{\mathbf S}_i + \hat H_{\mathrm{dip}} + \hat H_{\mathrm{higher}}.$$

The spin operators are dimensionless, so their eigenvalues are labeled by $S_i$. The Zeeman term inherits the convention of Magnetic Moments: $\hat{\boldsymbol\mu}_i=-\mu_B\mathsf g_i\hat{\mathbf S}_i$ and $\hat H_Z=-\mathbf B\cdot\hat{\boldsymbol\mu}_i$. Thus a physical electronic spin with positive scalar $g$ points opposite to its field-aligned magnetic moment. A fitted pseudospin tensor must specify its basis and this sign convention. Take $\mathsf K_i$ real symmetric and $\mathsf J_{ji}=\mathsf J_{ij}^{\mathsf T}$ for this bilinear Hermitian model. A symmetric single-ion quadratic term for spin $1/2$ is only a constant; it cannot open a gap. Anisotropic exchange can still do so. The terms have the following roles:

Term	Typical content	Spectral consequence
$\mathsf J_{ij}$	isotropic exchange, symmetric anisotropic exchange, antisymmetric Dzyaloshinskii–Moriya exchange	bandwidth, ordering wavevector, nonreciprocity, branch mixing
$\mathsf K_i$	single-ion anisotropy and crystal-field projection	gaps, easy axes or planes, polarization splitting
$\mathsf g_i$	site-dependent magnetic moment coupling	field slopes, compensation dynamics, polarization
$\hat H_{\mathrm{dip}}$	long-range dipole–dipole interaction	shape dependence and nonanalytic low-$q$ behavior
$\hat H_{\mathrm{higher}}$	biquadratic, ring, multipolar, magnetoelastic, or other retained terms	additional shifts, hybridization, and interaction channels

There is no universal sign convention for $\mathsf J_{ij}$. A fitted number is incomplete unless the Hamiltonian, bond counting, spin normalization, and crystallographic bond are all stated.

This local-spin form is an effective model. It is strongest when low-energy magnetic weight is concentrated into identifiable moments and the mode energies lie below major charge or orbital excitations. In an itinerant magnet, transverse collective modes can still exist, but their dispersion and damping arise from poles of an interacting electronic susceptibility and can merge into a particle–hole continuum. A local-spin fit is then a parametrization to be tested, not a microscopic conclusion.

Bosonic Modes and Stability

Choose a local frame on each magnetic site whose local $z$ axis is $\mathbf n_{i0}$. After introducing one constrained boson per retained magnetic site and expanding to quadratic order, a periodic magnet has the generic Nambu form

$$\hat H_2 = E_0 + \frac12 \sum_{\mathbf q} \hat{\boldsymbol\Psi}_{\mathbf q}^{\dagger} \mathcal H_B(\mathbf q) \hat{\boldsymbol\Psi}_{\mathbf q},$$

with

$$\hat{\boldsymbol\Psi}_{\mathbf q} = \left( \hat a_{\mathbf q1}, \ldots, \hat a_{\mathbf qn_m}, \hat a_{-\mathbf q1}^{\dagger}, \ldots, \hat a_{-\mathbf qn_m}^{\dagger} \right)^{\mathsf T}.$$

$E_0$ includes the constant needed when rewriting the quadratic operator in Nambu form; it is not generally just the classical energy. The mode problem is bosonic. It is governed by the dynamical matrix

$$\mathcal D(\mathbf q) = \Sigma_3\mathcal H_B(\mathbf q), \qquad \Sigma_3 = \begin{pmatrix} \mathbb 1_{n_m}&0\\ 0&-\mathbb 1_{n_m} \end{pmatrix},$$

not by an ordinary unitary diagonalization of $\mathcal H_B$. The physical transformation preserves the bosonic metric:

$$T_{\mathbf q}^{\dagger} \Sigma_3 T_{\mathbf q} = \Sigma_3.$$

For a stable harmonic reference, the physical mode energies are real and nonnegative after zero modes required by exact symmetries are handled correctly. A complex frequency, a negative-norm positive-frequency eigenvector, or a negative curvature outside the intended symmetry zero modes is a diagnostic of instability, an incorrect reference state, or a faulty convention. The canonical Magnons article derives this construction and its ferromagnetic and antiferromagnetic limits.

Ordered States and Their Low-Energy Modes

For an isotropic ferromagnet with nonzero uniform spin density, the two broken transverse rotation generators form one canonical pair. This gives one quadratic, type-B Goldstone branch. Several spins in the cell can add optical branches without adding broken symmetries. In a short-range model at long wavelength,

$$E(\mathbf q)\simeq\Delta+D_{\alpha\beta}q_\alpha q_\beta.$$

The gap vanishes only when the relevant continuous symmetry is exact and no field explicitly pins the order. Anisotropy and dipolar forces can gap or reshape this form; dipolar effects need not be analytic at $\mathbf q=0$.

A collinear antiferromagnet with vanishing uniform spin density commonly has linear transverse modes near a magnetic zone center. Its propagation vector can be zero in the chemical zone when the opposing spins already belong to a multi-site basis. With anisotropy gaps, a continuum approximation is

$$E_\lambda(\mathbf q)\simeq \sqrt{\Delta_\lambda^2+\hbar^2 q_\alpha \mathsf V^{(\lambda)}_{\alpha\beta}q_\beta},$$

where $\mathsf V^{(\lambda)}$ has units of velocity squared. This notation keeps energy and frequency units distinct. The Antiferromagnetism article explains the magnetic basis and spin-flop limits.

An isotropic ferrimagnet with nonzero net spin density combines a quadratic acoustic mode with optical relative-sublattice motion. Near angular-momentum compensation, its effective gyrotropic term can become small and inertial dynamics can matter. Anisotropy, damping, and lattice coupling determine the actual crossover. See Ferrimagnetism for a specified alternating-spin benchmark and the distinction between the two compensation temperatures.

These symmetry statements do not determine all material branches. Yttrium iron garnet has many spins in its primitive cell and many optical modes. Noncollinear order requires site-dependent local frames and can allow cubic interactions and spontaneous decay. Folded representations must also be compared in a common magnetic zone: an enlarged plotting cell can create copies without adding independent physical degrees of freedom. Incommensurate order needs a symmetry-adapted rotating frame or a controlled rational approximation rather than an unspecified finite magnetic cell (Toth and Lake, 2015).

What Inelastic Neutrons Measure

Define a dynamical spin correlation function by

$$S^{\alpha\beta}(\mathbf Q,\omega) = \frac{1}{2\pi N} \int_{-\infty}^{\infty} dt\, e^{i\omega t} \left\langle \hat S_{\mathbf Q}^{\alpha}(t) \hat S_{-\mathbf Q}^{\beta}(0) \right\rangle,$$

where

$$\hat S_{\mathbf Q}^{\alpha} = \sum_j e^{-i\mathbf Q\cdot\mathbf r_j} \hat S_j^\alpha.$$

For equivalent ions with the same isotropic moment factor and form factor, the unpolarized magnetic neutron cross section has the form

$$\frac{d^2\sigma}{d\Omega\,dE_f} = C \frac{k_f}{k_i} \lvert F(\mathbf Q)\rvert^2 \sum_{\alpha\beta} \left( \delta_{\alpha\beta} - \widehat Q_\alpha\widehat Q_\beta \right) S^{\alpha\beta}(\mathbf Q,\omega).$$

The constant $C$ includes the neutron and moment factors and the frequency-to-energy conversion appropriate to this definition, and $F(\mathbf Q)$ is the ion form factor. For inequivalent ions or anisotropic $g$ tensors, form factors and moment tensors belong inside the site-resolved scattering operator before squaring its matrix element; one common scalar $F$ outside the correlator is insufficient. The tensor

$$P_{\alpha\beta}(\widehat{\mathbf Q}) = \delta_{\alpha\beta} - \widehat Q_\alpha\widehat Q_\beta$$

projects out magnetic fluctuations parallel to $\mathbf Q$. Consequently, rotating the sample can make the same mode appear or disappear without changing its energy.

A one-magnon intensity also contains:

• the mode eigenvector and sublattice phase factors;
• the magnetic form factor of each ion;
• bosonic coherence factors;
• thermal occupation factors;
• domain and twin populations;
• instrumental resolution and background.

Plotting $\omega_\nu(\mathbf q)$ as a line over an intensity map tests only the peak positions. It does not test whether the model predicts the correct scattering operator matrix elements.

With the convention above, equilibrium detailed balance is

$$S^{\beta\alpha}(-\mathbf Q,-\omega) = e^{-\beta\hbar\omega} S^{\alpha\beta}(\mathbf Q,\omega).$$

For diagonal response in a reciprocal situation, the energy-gain side is therefore suppressed relative to the energy-loss side by $e^{-\beta\hbar\omega}$. This is both a physical population effect and a useful normalization check. Structure Factors and Time-Dependent Correlations develop these correlation conventions.

A mode can disappear by interference

Take two equivalent sites separated by $\mathbf d$, with equal local transition amplitudes and normalized transverse mode patterns $(1,\pm1)/\sqrt2$. Suppress the common form factor, thermal factor, and polarization projection. Their scattering amplitudes and intensities are

$$A_\pm(\mathbf Q)=\frac{1\pm e^{i\mathbf Q\cdot\mathbf d}}{\sqrt2}, \qquad I_\pm=1\pm\cos(\mathbf Q\cdot\mathbf d).$$

At phase zero, only the in-phase mode is visible; at phase $\pi$, only the out-of-phase mode is visible. If $d=a/2$ along a cell direction, shifting $Q$ by $2\pi/a$ exchanges these weights while leaving the reduced momentum unchanged. Both eigenmodes still exist in both zones. For a real antiferromagnet, replace these simplified amplitudes by the actual transverse Bogoliubov matrix elements.

Two-site interference changes intensity without changing an eigenmode’s energy. The middle panel plots $I_\pm=1\pm\cos\phi$ with $\phi=\mathbf Q\cdot\mathbf d$. Bottom: illustrative modes at $E_0$ and $2E_0$ exchange visibility between reciprocal zones whose phases differ by $\pi$. Form factors and polarization factors are held fixed in this example.

Other Spectroscopic Windows

No single probe measures a universal “magnon intensity.”

Neutron Scattering develops the instrument kinematics, polarization analysis, magnetic form factor, resolution convolution, normalization, and background controls behind the neutron row.

X-Ray Scattering develops the resonant intermediate-state, polarization, attenuation, and resolution controls behind the RIXS row.

Probe	Typical momentum window	What it is especially good at	Main interpretive caution
Inelastic neutron scattering	broad three-dimensional $\mathbf Q$ and energy coverage	bulk dispersions, polarizations, absolute spectral weight	sample volume, neutron absorption, form factors, resolution
Resonant inelastic x-ray scattering	finite momentum with element and edge selectivity	spin, orbital, charge, and lattice excitations in small samples and thin films	the resonant cross section is not generally identical to $S(\mathbf Q,\omega)$
Brillouin light scattering	very small wavevector and low energy	thin-film, surface, dipolar, and nonreciprocal modes	limited momentum range and optical selection rules
Raman scattering	near-zero photon momentum	symmetry-selected one- or multi-magnon channels	often probes composite operators rather than a single transverse spin
ESR, ferromagnetic resonance, and terahertz spectroscopy	near $\mathbf q=0$	gaps, field slopes, anisotropy, coherent mode dynamics	only modes with the required dipole matrix element are visible
Spin pumping and transport	device-dependent, often mode-integrated	driven populations, damping, propagation, interfaces	voltage or nonlocal signals are not direct spectral functions

Cross-probe agreement is powerful because different selection rules interrogate different pieces of the same eigenvectors. Cross-probe disagreement is not automatically a contradiction; first translate each measurement into its actual correlation function and geometry.

Fit the Forward Model

A robust analysis calculates the measured intensity from the Hamiltonian and experimental geometry:

$$I_{\mathrm{obs}}(\mathbf Q,\omega) = \int d\mathbf Q'\,d\omega'\, R( \mathbf Q-\mathbf Q', \omega-\omega' ) I_{\mathrm{model}}(\mathbf Q',\omega') + B(\mathbf Q,\omega).$$

Here $R$ is the instrumental resolution function and $B$ is a background model. Fitting unconvolved delta-function dispersions to broadened data can bias both energies and linewidths, especially near steep branches, crossings, or kinematic boundaries.

Energies and intensities constrain different things

Mode energies often constrain combinations of exchange parameters. Two different Hamiltonians can agree along a few high-symmetry cuts. Intensities add sensitivity to:

• the relative phases on inequivalent sites;
• eigenvector polarization;
• mode hybridization;
• magnetic form factors;
• domain populations.

This is why a fit to peak positions alone can be precise but not identifiable.

Use global constraints

Subject to the stated normalization and complete momentum and energy coverage, the total-moment sum rule has the schematic form

$$\frac{1}{N} \sum_{\mathbf Q} \int_{-\infty}^{\infty} d\omega \sum_\alpha S^{\alpha\alpha}(\mathbf Q,\omega) = S(S+1)$$

for one spin-$S$ degree of freedom per Bravais site, with the sum over the $N$ reciprocal grid points. Elastic as well as inelastic weight is included. For a multi-site cell, the corresponding completeness sum uses the trace over sublattice-resolved correlations; summing the scalar interference-weighted intensity over just one zone does not generally isolate all local moments. In a real material, covalency, multiple ions, orbital contributions, finite energy windows, and form-factor normalization must be included before using this as an absolute test. Partial spectral weight is not a violation of the sum rule if weight lies outside the measured window.

Other global checks include:

• symmetry-enforced degeneracies and periodicity;
• the correct number of magnetic domains;
• field slopes consistent with the stated $g$ tensors;
• gaps consistent with the symmetry broken explicitly by anisotropy or field;
• thermodynamic moments and ordering temperatures on compatible scales;
• stability of fitted parameters when additional momentum cuts are included.

Linewidths, Lifetimes, and Hybridization

The harmonic theory gives infinitely sharp poles. Interactions replace the bare pole by a dressed response. A schematic retarded Green function is

$$G_\nu^R(\mathbf q,\omega) = \frac{1}{ \hbar\omega - \varepsilon_{\nu0}(\mathbf q) - \Sigma_\nu^R(\mathbf q,\omega) }.$$

Near an isolated weakly damped pole, write

$$G^R\simeq\frac{Z}{\hbar\omega-E+i\kappa}, \qquad \Gamma_{\mathrm{FWHM}}=2\kappa.$$

The real-time amplitude decays as $e^{-\kappa t/\hbar}$, so its coherence time is

$$T_2=\frac{\hbar}{\kappa} =\frac{2\hbar}{\Gamma_{\mathrm{FWHM}}}.$$

Only for relaxation-limited damping without additional pure dephasing does the occupation decay twice as fast, giving $T_1=T_2/2=\hbar/\Gamma_{\mathrm{FWHM}}$. In a simple Markovian relaxation-plus-dephasing model,

$$\frac1{T_2}=\frac1{2T_1}+\frac1{T_\phi}.$$

A linewidth alone therefore does not generally determine $T_1$. This also assumes an intrinsic Lorentzian line: instrumental resolution, unresolved branches, and static frequency inhomogeneity must be modeled separately. The general pole and residue discussion is in Lifetime and Spectral Weight.

Magnons can broaden through:

• magnon–magnon scattering;
• magnon–phonon hybridization and decay;
• coupling to electron–hole or Stoner continua;
• disorder and domain boundaries;
• finite-temperature scattering;
• coupling to other crystal-field or orbital excitations.

An avoided crossing with exchange of spectral weight is evidence for hybridization, not merely two dispersions passing nearby. The coupled-mode eigenvectors should rotate through the crossing, and both energies and intensities should be fitted.

When the Magnon Description Fails

A sharp dispersing peak is strong evidence for a quasiparticle when its linewidth is small compared with its energy and nearby spectral scales. The converse is not automatic: a broad magnetic continuum can come from interacting magnons, fractional excitations, disorder, itinerant particle–hole pairs, or unresolved branches.

Reconsider a single-magnon interpretation when:

1. $\Gamma_{\mathrm{FWHM}}$ becomes comparable to the excitation energy;
2. the fitted pole loses most of its spectral residue;
3. a continuum carries substantial weight where harmonic theory predicts only one mode;
4. the assumed ordered state is absent or strongly fluctuating;
5. the local-moment sum rule and field response are incompatible with the fitted spins;
6. the mode enters a symmetry- and kinematically allowed decay continuum;
7. a one-dimensional or quantum-disordered system has elementary excitations other than magnons.

Energy and momentum conservation are necessary for decay, but not sufficient. The interaction vertex and all symmetry selection rules must also be nonzero. Conversely, a sharp mode inside a naive two-particle continuum can remain stable because the relevant coupling vanishes.

Worked Material Inference: Simple-Cubic Ferromagnet

Consider the explicitly defined model

$$\hat H = - J \sum_{\langle ij\rangle} \hat{\mathbf S}_i\cdot\hat{\mathbf S}_j + g\mu_B B \sum_i \hat S_i^z, \qquad J\gt0,$$

with one dimensionless spin $S$ per simple-cubic cell of lattice constant $a$, and $g>0$, $\mathbf B=B\hat{\mathbf z}$ with $B\geq0$. The field-polarized reference has $S_i^z=-S$, while the magnetic moment points along $+\hat{\mathbf z}$. To use an expansion about a local highest-weight state, rotate by $\pi$ about $x$: $\widetilde S^x=S^x$, $\widetilde S^y=-S^y$, $\widetilde S^z=-S^z$. This proper rotation preserves the spin commutators and exchange, and gives $H_Z=-g\mu_BB\sum_i\widetilde S_i^z$. A local magnon costs positive Zeeman energy. For this fully polarized Heisenberg model, the one-magnon energy is exact and agrees with linear spin-wave theory:

$$\varepsilon(\mathbf q) ={} g\mu_B B + 2JS \left[ 3 - \cos(q_xa) - \cos(q_ya) - \cos(q_za) \right].$$

Near $\mathbf q=\mathbf 0$,

$$\varepsilon(\mathbf q) \simeq \Delta + Dq^2, \qquad \Delta=g\mu_B B, \qquad D=JSa^2.$$

At the $X$ point $\mathbf q_X=(\pi/a,0,0)$,

$$\varepsilon_X = \Delta+4JS.$$

Therefore two directly interpretable estimates are

$$J = \frac{\varepsilon_X-\Delta}{4S}, \qquad D_{\mathrm{pred}} = \frac{\varepsilon_X-\Delta}{4}a^2.$$

The second relation is a falsifiable consistency check. If the measured low-$q$ stiffness disagrees with the value predicted from the zone boundary, the nearest-neighbor model is incomplete. Further-neighbor exchange, dipolar terms, anisotropy, itinerancy, or energy-dependent renormalization may be required. Adding parameters is justified only after identifying which additional measurements separate those possibilities.

Exercises

1. Extract exchange and test stiffness

For the simple-cubic model above, let

$$S=1, \qquad a=0.40\,\mathrm{nm}, \qquad \Delta=0.20\,\mathrm{meV}, \qquad \varepsilon_X=8.20\,\mathrm{meV}.$$

Find $J$ and the predicted stiffness $D$.

Solution

The exchange follows from

$$J = \frac{8.20-0.20}{4\times1}\,\mathrm{meV} = 2.00\,\mathrm{meV}.$$

Then

$$\begin{aligned} D &= JSa^2 \\ &= (2.00\,\mathrm{meV}) (1) (0.40\,\mathrm{nm})^2 \\ &= 0.32\,\mathrm{meV\,nm^2}. \end{aligned}$$

A measured low-$q$ stiffness inconsistent with this value would falsify the nearest-neighbor model even if it reproduced $\varepsilon_X$.

2. Neutron polarization extinction

Suppose a mode fluctuates entirely along laboratory $\hat{\mathbf z}$. Compare its unpolarized neutron intensity for $\mathbf Q\parallel\hat{\mathbf z}$ and $\mathbf Q\parallel\hat{\mathbf x}$, ignoring form factors and all other geometry.

Solution

The relevant factor is

$$P_{zz} = 1-\widehat Q_z^2.$$

For $\mathbf Q\parallel\hat{\mathbf z}$, $\widehat Q_z=1$ and $P_{zz}=0$, so the fluctuation is extinguished. For $\mathbf Q\parallel\hat{\mathbf x}$, $\widehat Q_z=0$ and $P_{zz}=1$, so it is fully visible within the stated simplifications.

The eigenmode exists in both geometries. Only its projection into the neutron cross section changes.

3. Detailed-balance check

At $T=100\,\mathrm K$, a reciprocal magnetic excitation has energy $\hbar\omega=10\,\mathrm{meV}$. What is the equilibrium ratio of the energy-gain intensity to the energy-loss intensity?

Use $k_B=0.08617\,\mathrm{meV/K}$.

Solution

Detailed balance gives

$$\frac{I(-\omega)}{I(+\omega)} = \exp\left( - \frac{\hbar\omega}{k_BT} \right).$$

Here

$$k_BT = 8.617\,\mathrm{meV},$$

so

$$\frac{I(-\omega)}{I(+\omega)} = e^{-10/8.617} \simeq 0.313.$$

A substantially different ratio calls for checking background subtraction, detector normalization, the sign convention for energy transfer, or nonequilibrium driving.

4. Linewidth and lifetime convention

An isolated magnon at $12\,\mathrm{meV}$ has a deconvolved Lorentzian energy FWHM of $1.2\,\mathrm{meV}$. Find its coherence time and its population lifetime if damping is relaxation limited, using $\hbar=0.6582\,\mathrm{meV\,ps}$ and calculate the ratio of energy to FWHM.

Solution

The intrinsic Lorentzian width gives

$$T_2=\frac{2\hbar}{\Gamma_{\mathrm{FWHM}}} =\frac{2(0.6582)}{1.2}\,\mathrm{ps} \simeq1.10\,\mathrm{ps}.$$

With no pure dephasing, $T_1=T_2/2\simeq0.55\,\mathrm{ps}$. A finite $T_\phi$ permits a longer population lifetime for the same measured width. The energy-to-FWHM ratio is $12/1.2=10$, which describes a resolved resonance but does not separate population decay from phase noise.

5. Diagnose a partial fit

A four-parameter local-spin model reproduces all observed peak energies along one high-symmetry line. It predicts a strong optical branch at a second reciprocal vector, but no peak is observed there. Give a controlled diagnostic sequence before discarding the branch.

Solution

First compute the full probe intensity, including the mode eigenvector, sublattice phase factors, polarization projector, ion-specific form factors, domain populations, thermal factor, and instrumental resolution. The branch may be extinguished in that geometry.

Second inspect symmetry-related reciprocal vectors and rotate the polarization geometry. A genuine mode with a selection-rule zero at one $\mathbf Q$ can become visible elsewhere.

Third check whether the predicted branch overlaps another excitation, background, or a region of poor resolution. Fit a resolution-convolved two-mode or coupled-mode model rather than a single peak.

Fourth test for damping or hybridization by looking for redistributed spectral weight, an avoided crossing, or a continuum. Only after these checks should the missing intensity be used to reject the Hamiltonian or the assumed magnetic structure.

References

• L. J. P. Ament, M. van Veenendaal, T. P. Devereaux, J. P. Hill, and J. van den Brink, “Resonant Inelastic X-Ray Scattering Studies of Elementary Excitations,” Reviews of Modern Physics 83, 705–767 (2011), doi:10.1103/RevModPhys.83.705.
• S. O. Bayrakci, T. Keller, K. Habicht, and B. Keimer, “Spin-Wave Lifetimes throughout the Brillouin Zone,” Science 312, 1926–1929 (2006), doi:10.1126/science.1127756.
• S. P. Bayrakci et al., “Lifetimes of Antiferromagnetic Magnons in Two and Three Dimensions: Experiment, Theory, and Numerics,” Physical Review Letters 111, 017204 (2013), doi:10.1103/PhysRevLett.111.017204.
• V. Cherepanov, I. Kolokolov, and V. L’vov, “The Saga of YIG: Spectra, Thermodynamics, Interaction and Relaxation of Magnons in a Complex Magnet,” Physics Reports 229, 81–144 (1993), doi:10.1016/0370-1573(93)90107-O.
• G. L. Squires, Introduction to the Theory of Thermal Neutron Scattering, 3rd ed. (Cambridge University Press, 2012), doi:10.1017/CBO9781139107808.
• S. Toth and B. Lake, “Linear Spin Wave Theory for Single-Q Incommensurate Magnetic Structures,” Journal of Physics: Condensed Matter 27, 166002 (2015), doi:10.1088/0953-8984/27/16/166002.
• M. E. Zhitomirsky and A. L. Chernyshev, “Colloquium: Spontaneous Magnon Decays,” Reviews of Modern Physics 85, 219–242 (2013), doi:10.1103/RevModPhys.85.219.