Waves in Magnetized Plasmas

Plasma Frequency and Dielectric Response left one dial at zero: the background magnetic field. Turn it on and the plasma stops being a mere frequency filter and becomes the most structured wave medium in classical physics. The reason is geometric: a static $\mathbf{B}_0$ makes every electron gyrate at the cyclotron frequency $\omega_c$ , so the medium acquires a built-in rotation sense. Push charges east and they drift north; the response current is no longer parallel to the applied field, the scalar permittivity becomes a tensor, and light crossing the plasma splits into modes that depend on the angle between its wavevector and $\mathbf{B}_0$ . The consequences read like a naturalist's field guide, and every specimen earns its living. Along the field the natural modes are right- and left-circular — one of them, the whistler, rides the electron gyration and disperses a lightning stroke into a falling glissando early telephone operators could hear. The tiny speed difference between the two circles rotates the polarization of any linear wave — Faraday rotation — how astronomers measure magnetic fields they will never visit. Across the field, the ordinary mode ignores $\mathbf{B}_0$ entirely while the extraordinary mode acquires new cutoffs and an upper-hybrid resonance; the split between their reflection ceilings is how an ionosonde reads the Earth's magnetic field out of a radio echo. This article derives the cold magnetized dielectric tensor once, honestly, and then spends it three ways: the parallel-propagation pair with the whistler and Faraday rotation worked to real numbers, the perpendicular pair with the ionosonde's O/X split, and the Appleton–Hartree relation that organized a century of ionospheric physics.

Key takeaways

A static $\mathbf{B}_0$ makes the plasma gyrotropic: the cold dielectric response becomes the Stix tensor with elements $S$ , $D$ , $P$ — the off-diagonal $D$ encodes the cyclotron rotation sense, and $P = 1 - \omega_p^2/\omega^2$ shows the unmagnetized physics surviving untouched along $\mathbf{B}_0$ .
Propagation along $\mathbf{B}_0$ : two circularly polarized modes with $n_R^2 = R$ and $n_L^2 = L$ . The R mode resonates at $\omega = \omega_c$ (it rotates with the electrons); below $\omega_c$ it becomes the whistler, with $v_g \propto \sqrt{\omega}$ — lightning impulses arrive as descending tones.
$n_R \neq n_L$ rotates linear polarization: Faraday rotation, $\Delta\psi = \lambda^2 \times \mathrm{RM}$ with the rotation measure $\mathrm{RM} \propto \int n_e B_\parallel\, dl$ — combined with the dispersion measure it yields the line-of-sight magnetic field of the interstellar medium.
Propagation across $\mathbf{B}_0$ : the O mode ( $n_O^2 = P$ ) behaves as if unmagnetized; the X mode ( $n_X^2 = RL/S$ ) has shifted cutoffs $\omega_R$ , $\omega_L$ and a resonance at the upper hybrid frequency $\omega_{UH}^2 = \omega_p^2 + \omega_c^2$ .
Cutoffs ( $n \to 0$ ) reflect; resonances ( $n \to \infty$ ) absorb — the two catastrophes that organize every dispersion curve, and the working principles of ionosondes and of electron-cyclotron heating (a 140 GHz gyrotron matches $\omega_c$ at 5.0 T).
The magnetoionic identities $\omega_R - \omega_L = \omega_c$ and $\omega_R\omega_L = \omega_p^2$ let one radio instrument measure both $n_e$ and $B_0$ from the split reflection ceilings of the ionogram.
KEY IDEA: a magnetized plasma is a tensor dielectric; each propagation direction carries two modes, and their cutoffs and resonances are set by three frequencies — plasma, cyclotron, and hybrid.

The core insight

In a magnetized plasma the dielectric response is a tensor; the wave equation becomes quadratic in $n^2$ , so each direction of propagation carries two modes whose cutoffs ( $n\to0$ ) and resonances ( $n\to\infty$ ) are set by the plasma, cyclotron, and hybrid frequencies.

Where this lives today

The rotation measure has gone cosmological in step with its sibling. Pulsar RM surveys — thousands of sightlines, each yielding $\langle B_\parallel\rangle$ from RM/DM — have mapped the Milky Way's spiral magnetic field, reversals and all; RM grids of extragalactic sources trace fields in galaxy clusters; and the RMs of localized fast radio bursts now probe magnetization between galaxies, with one famous repeater's RM swinging by tens of thousands of rad/m² as it orbits something violent. Meanwhile the whistler mode has a second career in the radiation belts: whistler-mode chorus — the same R-branch, self-excited by unstable electron populations — is the dominant accelerator and scatterer of relativistic "killer electrons," and forecasting it is a working problem of satellite protection. In fusion, this article is operational hardware. ITER's electron-cyclotron system will inject 20+ MW at 170 GHz, aimed by steerable mirrors to land exactly on chosen resonance surfaces — suppressing instabilities by heating a magnetic island a few centimeters wide, at a distance of meters, with nothing but the condition $\omega = \omega_c(\mathbf{r})$ . Reflectometry bounces microwaves off plasma cutoffs (the same reflection physics as the ionosonde, O and X mode both) to profile density without touching the plasma; interferometry and polarimetry across the machine measure $\int n_e\,d\ell$ and $\int n_e B_\parallel\,d\ell$ — the laboratory's own DM and RM, closing the loop between the tokamak hall and the radio telescope.

Origins · 1 minDouglas Rayner Hartree and how this was discovered

Douglas Rayner Hartree

1897–1958 · Cambridge and Manchester

Hartree is best known for the self-consistent field method of atomic physics, but his first major work was ionospheric. In 1931, extending Appleton's magnetoionic analysis, he derived the full dispersion relation for radio waves in a magnetized electron plasma — the Appleton–Hartree relation of this article — providing the theoretical backbone for the radio soundings that Appleton's experiments (Nobel Prize, 1947) had made possible. The formula's O/X mode structure explained the doubled traces on ionograms and turned them into quantitative instruments reading both electron density and the geomagnetic field. Hartree's later career anticipated computational physics itself: he built a differential analyzer from Meccano parts to solve his self-consistent field equations, ran wartime firing-table and radar-propagation calculations on it, and championed early electronic computers in Britain — a fitting trajectory for the man whose dispersion relation is still evaluated millions of times daily inside every ionospheric ray-tracing code.

History

1912

— Eccles proposes that ions in the upper atmosphere refract radio waves, seeding magnetoionic theory

1924

— Appleton and Barnett demonstrate the reflecting ionospheric layer; Appleton's magnetoionic work soon reveals the O/X mode splitting

1931

— Hartree completes the Appleton–Hartree dispersion relation, the working formula of ionospheric radio for the century since

1953

— Storey's analysis of lightning whistlers proves the plasmasphere exists — dense plasma stretching thousands of kilometers along field lines

1962

— Stix systematizes cold-plasma waves (the S, D, P tensor and the CMA atlas), the notation this article uses

The physical picture

Picture one electron in the field $\mathbf{B}_0$ , pushed by a wave's oscillating electric field. Left alone it gyrates at $\omega_c = eB_0/m_e$ — the cyclotron motion owned by the magnetostatics tier's Cyclotron Motion, running at the handy rate of 28 GHz per tesla. Now the wave arrives. If its field rotates in the same sense as the electron's gyration and at nearly the same rate, the push stays in phase orbit after orbit: the electron is driven resonantly and drinks energy without limit (in cold theory; kinetics eventually collects). If the field rotates the other way, the push and the gyration fall out of step twice per orbit and the coupling is mild. That single asymmetry — the medium has a handedness — is the whole subject. It is why the natural modes along $\mathbf{B}_0$ are circular rather than linear, why one of them blows up at $\omega_c$ and the other does not, and why a linear wave (an equal mix of both circles) cannot keep its orientation. The second organizing idea is the pair of catastrophes a refractive index can suffer. At a cutoff, $n \to 0$ : the wavelength stretches to infinity, the wave slows its energy transport to zero, reflects, and is evanescent beyond — the generalization of the plasma cutoff at $\omega_p$ , which is exactly the special case $B_0 = 0$ . At a resonance, $n \to \infty$ : the wavelength collapses, phase fronts pile up, and the wave's energy is handed to the particles — absorption, in reality via the kinetic effects the cold model only gestures at. Every dispersion curve in this article is a path between cutoffs and resonances, and every application — reflecting radio from the ionosphere, heating a tokamak, hearing a whistler — is an engineering of where those catastrophes sit.

Figure

The mode atlas of a cold magnetized plasma — along the field two circular modes with the whistler riding electron gyration, across it the ordinary and extraordinary pair

Why does the whistler descend in pitch? Because just below the electron's own rhythm the coupling is dramatic: the closer the wave's frequency sits to $\omega_c$ , the more efficiently it surfs the gyration, and — over the whistler band — the faster its energy travels. A lightning stroke launches all frequencies at once into the magnetosphere; the high notes outrun the low notes along the field line, and a receiver in the opposite hemisphere hears the impulse unrolled into a one-second glide from a few kilohertz downward. The dispersion is not a defect of the medium; it is the measurement — the glide's shape reads out the plasma density along a field line tens of thousands of kilometers long, remote sensing by ear.

Figure

A whistler spectrogram, computed for the L = 4 field line — the lightning impulse arrives unrolled into a falling tone with a nose at one quarter of the cyclotron frequency

One more feature of anisotropy deserves its own sentence: the wave normal $\mathbf{k}$ and the energy flow $\mathbf{S}$ need not point the same way. Whistler energy is guided along $\mathbf{B}_0$ even when the phase fronts tilt — the reason lightning whistlers follow geomagnetic field lines between hemispheres instead of spraying isotropically, and the same physics that makes anisotropic crystals walk a beam sideways in the optics track.

The dielectric tensor, derived

Derivation AThe cold-plasma dielectric tensor in Stix form

The goal is the full linear response of a cold magnetized plasma: the tensor $\boldsymbol{\varepsilon}$ that replaces the scalar $\varepsilon_r = 1 - \omega_p^2/\omega^2$ when $\mathbf{B}_0 = B_0\hat{\mathbf{z}}$ threads the medium. The strategy is the same as the unmagnetized case — solve the electron equation of motion in the wave field, sum the currents, read off the permittivity — with one new term, $q\mathbf{v}\times\mathbf{B}_0$ , which couples the two transverse velocity components and manufactures the off-diagonal response. Watch the denominators: $\omega^2 - \omega_c^2$ appears the moment the coupling is inverted, and every resonance in the article descends from it.

Step 1. Equation of motion, linearized. For electrons (charge $-e$ , the wave's own $\mathbf{B}$ negligible at first order, thermal motion switched off — "cold"):

-i\omega m_e \mathbf{v} = -e\left(\mathbf{E} + \mathbf{v}\times B_0\hat{\mathbf{z}}\right),

where: all perturbations oscillate as $e^{-i\omega t}$ , and the one structural novelty is the magnetic coupling term. In components, with $\omega_c = eB_0/m_e$ (a positive number here; the sign bookkeeping is carried explicitly): $-i\omega v_x = -(e/m_e)E_x - \omega_c v_y$ and $-i\omega v_y = -(e/m_e)E_y + \omega_c v_x$ , while $v_z$ decouples entirely — motion along $\mathbf{B}_0$ never feels the magnetic force.

Step 2. Invert the coupled pair. Solving the two-by-two system,

v_x = \frac{-e/m_e}{\omega^2 - \omega_c^2}\left(i\omega E_x + \omega_c E_y\right), \qquad v_y = \frac{-e/m_e}{\omega^2 - \omega_c^2}\left(-\omega_c E_x + i\omega E_y\right).

Why this is already the whole story: the response to $E_x$ has picked up a component along $\hat{\mathbf{y}}$ — drive the electron east and it also moves north, because the magnetic force turns its velocity. The denominator $\omega^2 - \omega_c^2$ announces the cyclotron resonance: drive at the gyration rate and the cold response diverges.

Step 3. Current, conductivity, permittivity. The electron current is $\mathbf{J} = -n_e e\mathbf{v}$ (ions contribute identical terms with their own $\omega_{ps}$ , $\omega_{cs}$ ; at the frequencies of this article their inertia benches them, and the FAQ says when they must return). Feeding $\mathbf{J} = \boldsymbol{\sigma}\mathbf{E}$ into $\boldsymbol{\varepsilon} = \mathbf{1} + i\boldsymbol{\sigma}/\varepsilon_0\omega$ gives the Stix tensor

\boldsymbol{\varepsilon} = \begin{pmatrix} S & -iD & 0 \\ iD & S & 0 \\ 0 & 0 & P \end{pmatrix},

S = 1 - \frac{\omega_p^2}{\omega^2 - \omega_c^2}, \qquad D = -\frac{\omega_c}{\omega}\,\frac{\omega_p^2}{\omega^2 - \omega_c^2}, \qquad P = 1 - \frac{\omega_p^2}{\omega^2},

where: $P$ (the $zz$ element) is exactly the unmagnetized dielectric function of Plasma Frequency and Dielectric Response — along $\mathbf{B}_0$ the magnetic force is blind and the old physics survives verbatim; $S$ ("sum") is the transverse diagonal response; and $D$ ("difference") is the gyrotropy itself, odd in $B_0$ , vanishing cleanly when the field is switched off. (Sign conventions for $D$ vary with the assumed charge and rotation bookkeeping across texts; the physical content — which circular mode resonates — is fixed by nature, and Step 4 makes it explicit.)

Step 4. Diagonalize the handedness. In the rotating basis $\hat{\mathbf{e}}_\pm = (\hat{\mathbf{x}} \pm i\hat{\mathbf{y}})/\sqrt{2}$ the transverse block becomes diagonal with eigenvalues $R = S + D$ and $L = S - D$ :

R = 1 - \frac{\omega_p^2}{\omega(\omega - \omega_c)}, \qquad L = 1 - \frac{\omega_p^2}{\omega(\omega + \omega_c)}.

where: $R$ carries the resonant denominator — it diverges at $\omega = \omega_c$ , because a right-circular field rotates with the electron gyration and pushes in phase forever — while $L$ stays finite at every positive frequency (its resonance belongs to the ions, at the far-lower ion cyclotron frequency). These two functions, plus $P$ , are the complete cast; every mode in the article is one of them wearing a geometric costume. Verified: setting $B_0 \to 0$ sends $\omega_c \to 0$ , so $D \to 0$ and $R, L, S \to 1 - \omega_p^2/\omega^2 = P$ — the tensor collapses to the isotropic scalar, recovering the unmagnetized article exactly; and the identities $S = \tfrac{1}{2}(R+L)$ , $D = \tfrac{1}{2}(R-L)$ check by direct addition of the displayed forms.

With the medium characterized, propagation is bookkeeping. A plane wave $\mathbf{E}\,e^{i(\mathbf{k}\cdot\mathbf{r} - \omega t)}$ in Maxwell's equations gives the tensor wave equation

\left(\mathbf{k}\mathbf{k} - k^2\mathbf{1} + \frac{\omega^2}{c^2}\boldsymbol{\varepsilon}\right)\cdot\mathbf{E} = 0,

where: setting the determinant to zero yields the dispersion relation, and because $\boldsymbol{\varepsilon}$ has three independent elements the relation is quadratic in $n^2 = c^2k^2/\omega^2$ — two modes per direction, the structural headline of the subject. The two clean geometries are worth deriving in full; the general angle is a formula (Appleton–Hartree, below) whose anatomy is best understood after both limits are in hand.

Along the field: whistlers and Faraday rotation

Go deeper · Derivation B · 3 minParallel propagation — the whistler glide and the Faraday rotation of the sky

Derivation BParallel propagation — the whistler glide and the Faraday rotation of the sky

The goal is the complete parallel-propagation story: the circular eigenmodes $n_R^2 = R$ , $n_L^2 = L$ , the whistler's falling tone extracted from $R$ with real magnetospheric numbers, and the Faraday rotation formula with its astrophysical payoff, the rotation measure. The strategy: put $\mathbf{k} = k\hat{\mathbf{z}}$ into the wave equation (the tensor is already diagonal in the rotating basis, so the modes drop out immediately), then interrogate the R branch below $\omega_c$ , then subtract the two branches' phase advances. Every number is computed for a real path.

Step 1. The eigenmodes. With $\mathbf{k} \parallel \mathbf{B}_0$ the wave equation block-diagonalizes in the $\hat{\mathbf{e}}_\pm$ basis:

n_R^2 = R = 1 - \frac{\omega_p^2}{\omega(\omega - \omega_c)}, \qquad n_L^2 = L = 1 - \frac{\omega_p^2}{\omega(\omega + \omega_c)},

where: the modes are circularly polarized by necessity, not choice — the medium's handedness makes the rotating vectors, not the Cartesian ones, the states that propagate without changing character. Cutoffs sit where $R = 0$ and $L = 0$ (named $\omega_R$ and $\omega_L$ , cashed in the next derivation); the R branch also owns the cyclotron resonance at $\omega = \omega_c$ .

Step 2. The whistler branch. For $\omega < \omega_c$ (and $\omega_p \gg \omega_c$ , the usual magnetospheric ordering), $R$ is positive and large: the mode propagates below the plasma frequency — the magnetic field has pried open a propagation window that the unmagnetized plasma slams shut — with

n^2 \approx \frac{\omega_p^2}{\omega(\omega_c - \omega)}.

Then $k = (\omega_p/c)\sqrt{\omega/(\omega_c-\omega)}$ , and differentiating,

v_g = \frac{d\omega}{dk} = \frac{2c\,\sqrt{\omega}\,(\omega_c - \omega)^{3/2}}{\omega_p\,\omega_c} \;\xrightarrow{\ \omega \ll \omega_c\ }\; \frac{2c\sqrt{\omega\,\omega_c}}{\omega_p}.

where: in the deep-whistler limit $v_g \propto \sqrt{\omega}$ — high notes travel faster — so a broadband impulse traversing a path of length $\ell$ arrives with the time signature $t(f) = \ell\omega_p/(2c\sqrt{2\pi f\,\omega_c})$ , i.e. $t\sqrt{f} = \text{const}$ : Eckersley's law, the observed shape of whistler spectrograms for half a century.

Step 3. Real numbers on a real field line. Take the $L = 4$ geomagnetic shell: equatorial field $B_0 = 3.1\times10^{-5}/4^3 = 4.8\times10^{-7}$ T, so $f_c = 13.6$ kHz; plasmaspheric density $n_e = 10^8\ \text{m}^{-3}$ , so $f_p = 90$ kHz; path length along the field line $\approx 25{,}000$ km. The full $v_g$ gives travel times $t(1\ \text{kHz}) = 1.14$ s, $t(3.4\ \text{kHz}) = 0.85$ s, $t(8\ \text{kHz}) = 1.37$ s — a one-second glide, exactly the duration whistler listeners report. And note what the full formula adds to the deep-limit law: $v_g$ is maximal at $\omega = \omega_c/4$ (differentiate $\sqrt{\omega}(\omega_c-\omega)^{3/2}$ ), here 3.4 kHz — frequencies on both sides of it arrive later, so a strong whistler traces a "nose" on the spectrogram. Nose whistlers hand ionospheric physicists both numbers at once: the nose frequency reads $B_0$ (hence which field line), the nose delay reads $\int \omega_p\,d\ell$ (hence the density along it). Why one hears electron density thousands of kilometers up: the glide time scales with $\omega_p$ , the only plasma property in it. Whistler dispersions measured from the ground mapped the plasmasphere — including the sharp density cliff of the plasmapause — years before spacecraft flew through it.

Step 4. Faraday rotation. A linearly polarized wave is the equal superposition of R and L. Over a path $d\ell$ each circular component advances phase as $n_{R,L}\,\omega\,d\ell/c$ , so the plane of polarization (the bisector of the two rotating vectors) turns by half their phase difference. In the high-frequency limit $\omega \gg \omega_p, \omega_c$ , expanding both indices and subtracting kills the leading terms:

n_L - n_R \approx \frac{\omega_p^2\,\omega_c}{\omega^3}, \qquad d\psi = \frac{\omega}{2c}(n_L - n_R)\,d\ell = \frac{e^3}{2\varepsilon_0 m_e^2 c\,\omega^2}\, n_e B_\parallel\, d\ell.

Integrating along the line of sight and writing $\omega = 2\pi c/\lambda$ :

\psi = \lambda^2\,\mathrm{RM}, \qquad \mathrm{RM} = \frac{e^3}{8\pi^2\varepsilon_0 m_e^2 c^3}\int n_e B_\parallel\, d\ell = 0.812\ \frac{\text{rad}}{\text{m}^2} \int \left(\frac{n_e}{\text{cm}^{-3}}\right)\left(\frac{B_\parallel}{\mu\text{G}}\right)\frac{d\ell}{\text{pc}},

where: the rotation measure RM is to magnetic fields what the dispersion measure is to density — a line integral read from a $\lambda^2$ fit to polarization angles across a radio band. Only $B_\parallel$ enters (reversals along the path cancel), and only the electrons matter. Divide the two integrals and the density weighting cancels: $\langle B_\parallel\rangle = 1.232\,\mathrm{RM}/\mathrm{DM}$ in microgauss. For the Crab pulsar ( $\mathrm{RM} = -42.3\ \text{rad/m}^2$ , $\mathrm{DM} = 56.77\ \text{pc cm}^{-3}$ ) this gives $\langle B_\parallel\rangle \approx 0.92\ \mu\text{G}$ pointed away from us — a measurement of the Galaxy's magnetic field, made with a protractor. At 1.4 GHz the Crab's polarization arrives rotated by $|\mathrm{RM}|\lambda^2 = 1.94$ rad; the $\lambda^2$ signature is what separates the plasma's rotation from the source's intrinsic angle. Verified: the astro coefficient recomputed from CODATA constants gives $0.8119\ \text{rad m}^{-2}$ per $(\text{cm}^{-3}\,\mu\text{G}\,\text{pc})$ and its reciprocal pair $1.232$ ; the whistler times above come from the full group velocity, not the deep-limit law, and reproduce Eckersley $t\sqrt{f} \approx 32\ \text{s Hz}^{1/2}$ on the low-frequency side where the limit applies.

Across the field: O, X, and the hybrid resonance

Go deeper · Derivation C · 2 minPerpendicular propagation — the ordinary and extraordinary modes and their catastrophes

Derivation CPerpendicular propagation — the ordinary and extraordinary modes and their catastrophes

The goal is the second clean geometry, $\mathbf{k} \perp \mathbf{B}_0$ : the two modes $n_O^2 = P$ and $n_X^2 = RL/S$ , the X mode's cutoff pair $\omega_R, \omega_L$ with the magnetoionic identities that make ionograms legible, and the upper-hybrid resonance. The strategy is to take $\mathbf{k} = k\hat{\mathbf{x}}$ in the tensor wave equation and watch it factor: one polarization never sees the magnetic field at all, and the other pays for its cross-field electric field with brand-new physics.

Step 1. The ordinary mode. Put the wave's electric field along $\mathbf{B}_0$ ( $E_z$ only). The response is governed by $P$ alone — electrons driven parallel to $\mathbf{B}_0$ feel no magnetic force — and the wave equation gives

n_O^2 = P = 1 - \frac{\omega_p^2}{\omega^2}.

where: this is verbatim the unmagnetized dispersion of Plasma Frequency and Dielectric Response: same cutoff at $\omega_p$ , same evanescence below it, magnetic field invisible. The name "ordinary" is earned. Its practical role is the control experiment: whatever the O mode measures is pure density.

Step 2. The extraordinary mode. Put the electric field across $\mathbf{B}_0$ (in the $xy$ plane). Now the gyrotropic coupling acts, the mode develops a longitudinal component ( $\mathbf{E}$ partly along $\mathbf{k}$ — it is not a purely transverse wave), and eliminating the components gives

n_X^2 = \frac{RL}{S}.

where: the structure announces everything: $n_X^2$ vanishes where $R = 0$ or $L = 0$ (two cutoffs) and diverges where $S = 0$ (a resonance). Solving $R = 0$ and $L = 0$ as quadratics in $\omega$ :

\omega_{R} = \tfrac{1}{2}\!\left(\omega_c + \sqrt{\omega_c^2 + 4\omega_p^2}\right), \qquad \omega_{L} = \tfrac{1}{2}\!\left(-\omega_c + \sqrt{\omega_c^2 + 4\omega_p^2}\right),

with the magnetoionic identities $\omega_R - \omega_L = \omega_c$ and $\omega_R\,\omega_L = \omega_p^2$ following by inspection — the sum and product of the cutoffs hand over the field and the density separately. The stop-band map: the X mode propagates for $\omega_L < \omega < \omega_{UH}$ and again above $\omega_R$ , and is evanescent elsewhere.

Step 3. The upper-hybrid resonance. Solve $S = 0$ :

\omega_{UH}^2 = \omega_p^2 + \omega_c^2.

where: the name is honest — the mode at this frequency is a hybrid oscillation, part plasma sloshing (electrostatic restoring force) and part cyclotron gyration (magnetic restoring force), the two stiffnesses adding in quadrature. Near $\omega_{UH}$ the X wave slows, its wavelength collapses, its electric field turns almost purely longitudinal, and cold theory hands off to kinetics: the wave is absorbed or converted to electrostatic modes. This is a feature — driving a plasma near a resonance is how you heat it on purpose.

Step 4. The instrument reading. An ionosonde looking up through the Earth's field sees both reflection ceilings: the O trace ends at $f_c^{O} = f_p^{\max}$ , the X trace at $f_c^{X} \approx f_p^{\max} + f_{ce}/2$ (the large- $\omega_p$ expansion of $\omega_R$ ). Two ceilings, two unknowns: density from the O ceiling, magnetic field from the split — computed to real digits in the worked example. The same logic at millimeter waves runs fusion diagnostics and heating: a 140 GHz gyrotron beam matches the electron cyclotron frequency at $B = 5.0$ T (or its second harmonic at 2.5 T), placing a localized power deposit exactly where $\omega = \omega_c(\mathbf{r})$ on the tokamak's field map — electron cyclotron resonance heating, the most surgically targeted heat source in fusion. Verified: $\omega_R\omega_L = \tfrac{1}{4}[(\sqrt{\omega_c^2+4\omega_p^2})^2 - \omega_c^2] = \omega_p^2$ and $\omega_R - \omega_L = \omega_c$ algebraically; for $f_p = 9.0$ MHz, $f_{ce} = 1.4$ MHz the exact cutoffs are $f_R = 9.727$ MHz and $f_L = 8.327$ MHz (identities check: difference 1.400, product $81.0 = f_p^2$ ), and $f_{UH} = 9.108$ MHz sits between them.

At arbitrary angle $\theta$ between $\mathbf{k}$ and $\mathbf{B}_0$ the determinant yields the closed form that organized twentieth-century radio science, the Appleton–Hartree relation (electrons only, collisionless):

n^2 = 1 - \frac{X}{1 - \dfrac{Y^2\sin^2\theta}{2(1-X)} \pm \sqrt{\dfrac{Y^4\sin^4\theta}{4(1-X)^2} + Y^2\cos^2\theta}},

where: $X = \omega_p^2/\omega^2$ and $Y = \omega_c/\omega$ package the two frequency ratios, and the $\pm$ selects the two magnetoionic modes. Its anatomy is exactly the two limits just derived: at $\theta = 0$ the square root collapses to $Y\cos\theta$ and the denominators become $1 \mp Y$ — the R and L modes; at $\theta = 90°$ the plus sign returns $n^2 = 1 - X$ (the O mode) and the minus sign rebuilds $RL/S$ (the X mode). Between the limits the formula interpolates the mode structure continuously — the quasi-longitudinal regime that governs real oblique ionospheric paths, where the O/X pair behave approximately like weakly split R/L modes. The full atlas of which mode propagates where, as $n_e$ and $B_0$ vary, is the CMA diagram of plasma physics — a two-parameter map of cutoff and resonance surfaces that this article's two clean geometries let you read; Stix owns the full chart.

Limiting cases

case	result	meaning
$B_0 \to 0$	$D \to 0$ , $R = L = S = P$	gyrotropy vanishes; the isotropic plasma of the previous article returns
$\omega \gg \omega_p, \omega_c$	$n^2 \to 1$	the plasma cannot respond; vacuum propagation (X-rays, high GHz)
$\omega \ll \omega_c$ , $\mathbf{k}\parallel\mathbf{B}_0$	$n^2 \approx \omega_p^2/\omega\omega_c$	whistler branch: $v_g \propto \sqrt{\omega}$ , falling tones, field-aligned ducting
$\theta = 0$ in Appleton–Hartree	$n^2 = 1 - X/(1 \mp Y)$	R/L circular pair; Faraday rotation from their split
$\theta = 90°$ in Appleton–Hartree	$n^2 = 1-X$ or $RL/S$	O mode (density only) and X mode (cutoff pair + hybrid resonance)
$\omega \to \omega_c$ (R mode)	$n^2 \to \infty$	electron cyclotron resonance: absorption, ECRH heating window
$\omega \to \omega_{ci}$ scales	ion terms revive	ion cyclotron waves, lower hybrid, and the MHD family (Alfvén Waves)

Common trapThe magnetized plasma is not a scalar-index medium with corrections — linear polarization is not a propagating state

The most common error in this subject is computing with one refractive index and adding the magnetic field as an afterthought. Along $\mathbf{B}_0$ there is no such thing as a linearly polarized eigenmode: a linear wave is a superposition of R and L, and the medium relentlessly re-phases the pair — that is Faraday rotation, and it accumulates without bound. Across $\mathbf{B}_0$ the two modes do not even share a cutoff: an X wave reflects where an O wave of the same frequency sails on, and the extraordinary wave is partly longitudinal — $\mathbf{E}$ with a component along $\mathbf{k}$ , unthinkable for E8's isotropic plane waves. Decompose into the true eigenmodes first (circular along the field, O/X across), propagate each with its own $n$ , recombine at the end. Skipping this does not approximate the answer; it misses cutoffs, resonances, and rotation entirely.

Worked examples

1 · Reading both numbers from an ionogram (confidence builder)

A vertical ionosonde at a site where the geomagnetic field is $B_0 = 5.0\times10^{-5}$ T records an O-mode trace ending at $f_c^{O} = 9.00$ MHz and an X-mode trace ending near $f_c^{X} = 9.73$ MHz. Extract the peak electron density and check the magnetic field against the site value.

Step 1. Density from the O ceiling. The O mode ignores $B_0$ , so its critical frequency is the plasma frequency itself:

n_e = \left(\frac{9.00\times10^6}{8.98}\right)^2 \approx \mathbf{1.0\times10^{12}\ m^{-3}},

using the engineer's coefficient from the previous article — the magnetized theory's O mode hands the unmagnetized formula back unchanged.

Step 2. Field from the split. The X ceiling is the $R = 0$ cutoff, $f_R = \tfrac{1}{2}(f_{ce} + \sqrt{f_{ce}^2 + 4f_p^2})$ , which for $f_p \gg f_{ce}$ is $\approx f_p + f_{ce}/2$ . The observed split is $9.73 - 9.00 = 0.73$ MHz, so

f_{ce} \approx 2 \times 0.73 = 1.46\ \text{MHz} \quad\Rightarrow\quad B_0 = \frac{2\pi m_e f_{ce}}{e} \approx \mathbf{5.2\times10^{-5}\ T},

within a few percent of the site's $5.0\times10^{-5}$ T (the exact cutoff formula closes the residual: with $f_{ce} = 1.40$ MHz it predicts $f_R = 9.727$ MHz, our "observed" value). One radio sweep, two geophysical numbers — this dual readout has run continuously at hundreds of stations since the 1930s.

2 · A whistler as a plasmaspheric densitometer (exam level)

A ground station records a lightning whistler that traversed the $L = 4$ field line (path $\ell \approx 25{,}000$ km, equatorial $f_{ce} = 13.6$ kHz). The 1.00 kHz component arrives $1.14$ s after the stroke. (a) Verify the deep-whistler travel time for an assumed plasmaspheric density $n_e = 1.0\times10^8\ \text{m}^{-3}$ . (b) Invert the measurement into a density. (c) Where is the nose?

Step 1. Forward check. With $n_e = 10^8\ \text{m}^{-3}$ , $f_p = 8.98\sqrt{10^8} = 89.8$ kHz. At 1.00 kHz ( $\omega/\omega_c = 0.074$ , comfortably deep), $v_g \approx 2c\sqrt{\omega\omega_c}/\omega_p$ with a $(1 - \omega/\omega_c)^{3/2} = 0.89$ correction from the full formula:

v_g = 0.89\times\frac{2c\sqrt{(2\pi\times10^3)(2\pi\times1.36\times10^4)}}{2\pi\times8.98\times10^4} = 2.19\times10^7\ \text{m/s}, \qquad t = \frac{2.5\times10^7}{2.19\times10^7} = \mathbf{1.14\ s}. \checkmark

Step 2. Invert. In the deep limit $t \propto \omega_p \propto \sqrt{n_e}$ at fixed path and $f$ , so the measured delay reads density directly:

n_e = 10^8 \left(\frac{t_{\text{obs}}}{1.14\ \text{s}}\right)^2\ \text{m}^{-3}.

A 10% timing measurement is a 20% density measurement of a region 25,000 km away, using lightning as the free transmitter. Storey's 1953 analysis of exactly such glides proved dense plasma extends thousands of kilometers up — the discovery of the plasmasphere.

Step 3. The nose. Group velocity peaks at $\omega = \omega_c/4$ , i.e. $f_{\text{nose}} = 13.6/4 = \mathbf{3.4\ kHz}$ , arriving first at $t = 0.85$ s; components at 1 kHz and 8 kHz both arrive over a quarter-second later (1.14 s and 1.37 s). Measuring $f_{\text{nose}}$ alone hands over $f_{ce}$ , hence the field line the whistler rode — no spacecraft required.

Going further

References. Jackson, Classical Electrodynamics, 3rd ed., Ch. 7 (waves in magnetized media, magnetoionic theory) · Griffiths, Introduction to Electrodynamics, 4th ed., Ch. 9 (EM waves, dispersion) · Zangwill, Modern Electrodynamics, Ch. 18 (waves in dispersive matter) · Stix, Waves in Plasmas, Ch. 1–2 (cold-plasma dielectric tensor, CMA diagram) · Chen, Introduction to Plasma Physics and Controlled Fusion, Ch. 4 (waves in plasmas).

Provenance. Dielectric tensor and Stix S,D,P from Stix, Waves in Plasmas Ch.1, and Chen Ch.4. Dispersion relations (R,L,O,X; Appleton-Hartree; Stix general-angle) cross-checked against Jackson 3rd ed. Ch.7. A verifier should: (1) re-derive the Stix tensor from the cold-fluid equation and confirm $R=S+D$ , $L=S-D$ ; (2) verify the magnetoionic identities $\omega_R\omega_L=\omega_p^2$ and $\omega_R-\omega_L=\omega_c$ ; (3) confirm $\omega_{UH}^2=\omega_{pe}^2+\omega_{ce}^2$ from $S=0$ ; (4) check the whistler scaling $k\propto\sqrt{\omega}$ hence $v_g\propto\sqrt{\omega}$ and the Eckersley law $t\sqrt f$ =const. All symbols (omega_p, omega_c, n=ck/omega) follow standard conventions; no numerical constants are quoted beyond standard definitions. Track addendum: all worked numbers here (whistler times on the $L=4$ line from the full group velocity, nose at $\omega_c/4$ , exact cutoffs $f_R = 9.727$ / $f_L = 8.327$ MHz, RM coefficient 0.812 and $\langle B_\parallel\rangle$ coefficient 1.232, Crab 0.92 μG, 28.0 GHz/T, 5.0 T at 140 GHz) were recomputed from CODATA constants for this article.

Prerequisites: Maxwell's Equations in Matter · Plane Electromagnetic Waves · Plasma Frequency and Dispersion · Lorentz Force and Cyclotron Motion. Leads to: Faraday Rotation · Synchrotron Radiation · Kinetic Plasma Theory and Landau Damping.

Where do the ions actually matter, and how bad is neglecting them here?

Quantitatively negligible for everything this article computes, by mass ratio: ion terms in the Stix sums carry $\omega_{pi}^2 = \omega_{pe}^2(m_e/m_i)$ and resonate at $\omega_{ci} = \omega_{ce}(m_e/m_i)$ — for protons, 1836 times smaller. Whistlers at kilohertz, ionograms at megahertz, Faraday rotation at gigahertz all sit far above every ion frequency. Ions return in two places: near $\omega_{ci}$ and below, where they carry ion cyclotron waves and the magnetosonic/Alfvén family (the Alfvén Waves article — the smooth low-frequency continuation of this page's fast branches), and at the lower hybrid resonance $\omega_{LH} \approx \sqrt{\omega_{ce}\omega_{ci}}$ (a few GHz in tokamaks), the geometric-mean frequency where electrons and ions share the oscillation and lower-hybrid current drive operates.

Is Faraday rotation in a plasma the same effect as Faraday's magneto-optical rotation in glass?

The same symmetry, different microscopics. Faraday's 1845 discovery — polarization rotating in glass along a magnetic field — is circular birefringence from bound electrons whose Larmor precession shifts the R and L resonance frequencies; the plasma version is circular birefringence from free electrons whose gyration splits $n_R$ from $n_L$ . Both descend from the same statement: a medium magnetized along the propagation direction cannot treat the two circular polarizations equally, because $\mathbf{B}_0$ defines a rotation sense. Both are non-reciprocal — reflect the wave back through the medium and the rotation doubles rather than undoing, which is why optical isolators can be built from magnetized media and why interstellar rotation cannot be dismissed as a round-trip artifact.

The whistler propagates below the plasma frequency. Doesn't that contradict the cutoff from the unmagnetized article?

It repeals it, legitimately — that cutoff was a theorem about isotropic plasma. The unmagnetized argument said: electrons respond faster than the wave, over-cancel it, no propagation below $\omega_p$ . The magnetic field breaks the argument by constraining the response: electrons trying to cancel a slowly rotating R-mode field cannot move freely across $\mathbf{B}_0$ — they are tethered to gyro-orbits — and their partial, phase-lagged response adds to rather than cancels the wave, driving $n^2$ large and positive. The whistler window ( $\omega < \omega_c$ , R mode, along the field) is thus a genuinely new passband pried open by the field. The general lesson: cutoffs are properties of a mode in a medium, not of a medium alone.

What physically happens to a wave at the upper hybrid resonance if nothing in cold theory can absorb it?

Cold theory delivers the wave to the scene of the crime and then leaves. As $S \to 0$ the X mode's wavelength collapses, its group velocity stalls, its field turns nearly longitudinal, and its energy density piles up — at which point the neglected physics takes over. Finite temperature converts the wave into short-wavelength electrostatic waves (electron Bernstein modes) that cold theory does not contain; collisions, however rare, get unlimited time to act on a stalled wave; and at high amplitude the field traps and heats particles directly. The honest statement: cold theory locates resonances exactly and predicts the pile-up, kinetics decides the corpse's fate. In practice the location is what engineering needs — ECRH aims power at a resonance surface, and the plasma reliably absorbs it there.

Why does the rotation measure use only the parallel component of B while synchrotron emission measures the perpendicular one?

Because they are different physical couplings, and the complementarity is a gift. Faraday rotation is circular birefringence: the R/L split exists only for the field component along $\mathbf{k}$ , so RM integrates $n_e B_\parallel$ — with sign, reversals canceling. Synchrotron emission comes from relativistic electrons gyrating around field lines: its intensity and polarization respond to $B_\perp$ , sign-blind. A radio map of a galaxy therefore carries two independent projections of the same field: synchrotron polarization angles give the sky-plane field orientation, RM gives the line-of-sight component and its direction. Stacking both across many sightlines is how the three-dimensional spiral structure of galactic magnetic fields — including the Milky Way's field reversals between arms — was reconstructed.

Ionograms show the X trace ending above the O trace, but the L cutoff is below the plasma frequency. Why doesn't a third trace appear there?

A wave launched from the ground climbs the ionosphere's rising density profile and reflects at the first surface where its $n^2$ hits zero; for the upgoing X wave near the layer peak the operative boundary is the $R = 0$ cutoff at $f_R \approx f_p + f_{ce}/2$ — above the O ceiling. The $L$ -cutoff branch is not fiction, though: it supports a "Z-mode" trace reachable through mode-coupling windows near vertical propagation, and Z-mode echoes are occasionally observed on high-latitude ionograms — a rare-but-real third trace whose explanation is exactly the extra cutoff you noticed.

How is the CMA diagram actually used, and do I need it at this level?

The Clemmow–Mullaly–Allis diagram is an atlas: axes $\omega_p^2/\omega^2$ (density) and $\omega_c/\omega$ (field), with every cutoff and resonance drawn as a boundary curve dividing the plane into regions where each mode exists. Its power is navigational — a wave crossing a nonuniform plasma traces a path across the diagram, and you read off where it reflects, resonates, or changes character without re-deriving anything. You already own its skeleton: the $P = 0$ , $R = 0$ , $L = 0$ cutoff curves and the $\omega_c$ , $\omega_{UH}$ resonance curves derived above are the CMA boundaries. For waves crossing widely varying density and field, reach for Stix's chart; for a homogeneous slab, the four named modes here are the whole toolkit.