Plasma Frequency and Dielectric Response

The dispersion measure has become cosmology. Fast radio bursts — millisecond flashes from other galaxies — arrive with DMs far beyond anything the Milky Way can supply, so each burst weighs the ionized matter along a gigaparsec sightline. The Macquart relation (2020) stacked localized FRBs against host-galaxy distances and found the "missing baryons" — the half of ordinary matter surveys could not see — sitting as diffuse intergalactic plasma, counted electron by electron with the $f^{-2}$ law derived above. Engineering, meanwhile, has learned to set $\omega_p$ by hand. Metamaterial wire lattices dilute the effective electron density to bring the plasma frequency down from ultraviolet to gigahertz — artificial "metals" with designer cutoffs. Ultrafast optics runs the trick in time: a femtosecond pulse ionizing a solid's surface raises $\omega_p$ above optical frequencies within one cycle — a "plasma mirror" that snaps from transparent to reflecting fast enough to carve attosecond structure out of light. And the space-weather enterprise — GPS plasma-delay corrections worth meters of position error, HF outage forecasts after solar flares — is at bottom the daily operational monitoring of one number, $n_e(h,t)$, through the formulas on this page.

Start with what a plasma is at rest: free electrons riding a sea of much heavier ions, so thoroughly mixed that any macroscopic volume is neutral. The mixing is enforced, not accidental: separating charge builds a field that fights the separation, and the field of even a tiny imbalance is enormous — the lesson of the whole electrostatics tier, now applied to a medium that can move. So move it. Take a slab of plasma and slide all its electrons a small distance $x$, leaving the ions fixed (thousands of times heavier; on electron timescales, furniture). Two charged sheets appear: exposed ions on one face, surplus electrons on the other — a parallel-plate capacitor built from the plasma's own charge, whose interior field pulls every displaced electron straight back. The electrons rush home, overshoot — nothing damps them — and the slab rings like a struck bell. The ringing rate is the plasma frequency, and the first derivation shows it does not care about size or shape, only density: $\omega_p \propto \sqrt{n_e}$. Denser plasma, stiffer spring, faster ring.

The second face of $\omega_p$ appears when the pushing is done by a passing light wave. A wave of frequency $\omega$ asks the electrons to oscillate at $\omega$; they answer with an amplitude set by inertia, and their re-radiated field either cancels the wave or merely dilutes it. The bookkeeping — done once, in the Absorption and Dispersion article's Lorentz-oscillator machine — says the contest is decided by $\omega$ versus $\omega_p$. Ask slowly ($\omega < \omega_p$) and the electrons respond more than enough to cancel the wave inside: the field dies within a skin depth $\sim c/\omega_p$, and everything reflects. Ask quickly ($\omega > \omega_p$) and inertia wins: the electrons barely move before the field reverses, cancellation fails, and the wave propagates — faster in phase than vacuum light, slower in energy.

One scale completes the cast. A plasma screens static intrusions over the Debye length $\lambda_D = \sqrt{\varepsilon_0 k_B T_e/n_e e^2}$ — the distance over which thermal jitter can hold charge apart against the restoring field. Quasineutrality, the slab picture's premise, is the statement that the system dwarfs $\lambda_D$ with many electrons per Debye sphere: the F-region has $\lambda_D \approx 2$ mm and $N_D \approx 4\times10^{4}$, so the smooth-fluid treatment is honest. Note the pairing: $\lambda_D$ is the plasma's ruler for statics, $\omega_p$ its metronome for dynamics, tied by $\lambda_D \, \omega_p = v_{\text{th}}$. Everything here is the cold, unmagnetized story — the two dials left at zero are switched on next door: a background magnetic field in Waves in Magnetized Plasmas (the response becomes a tensor), and temperature in Landau Damping (the oscillation gains a wavelength-dependent correction and a collisionless death).

The table of rings is worth a pause, because it spans sixteen orders of magnitude in density with one formula:

The slab mode is the plasma alone. Now send in a wave. The machinery is owned by Absorption and Dispersion: a medium of driven charges acquires $\tilde{\varepsilon}_r(\omega) = 1 - \omega_p^2/(\omega^2 + i\gamma\omega)$ when the springs are cut ($\omega_0 = 0$), and in the collisionless window $\omega \gg \gamma$ this is real:

where: $\gamma$ is the electron collision rate, negligible in the F-region ($\gamma/\omega \sim 10^{-4}$ at HF) and in interstellar space; the same Drude expression with $\gamma$ kept is the engine of Electromagnetic Waves in Conductors — copper at 60 Hz ($\gamma \gg \omega$) and the ionosphere at 10 MHz ($\gamma \ll \omega$) are one formula in two limits. The derivation below takes the collisionless form and asks the question that matters: what does a wave do in a medium whose permittivity can go negative?

Above roughly 60 km, solar EUV and X-rays ionize the thin air faster than recombination heals it, and Earth wears a permanent plasma blanket: the ionosphere, layered (D below 90 km, E near 100, F peaking near 250–350 km) because ionizing flux falls from above while air density falls from below. Its peak density, $n_e \sim 10^{12}\ \text{m}^{-3}$ by day, is thirteen orders below the neutral air's — yet it owns every radio wave below about 10 MHz, because ownership is decided by $f_p$, not mass fraction. The instrument that maps it is the ionosonde, radio's echo sounder: chirp upward from 1 to ~20 MHz, time the echoes. Each frequency reflects at the altitude where the local $f_p$ has climbed to meet it, so the echo-delay-versus-frequency trace — the ionogram — is an altitude profile of $\sqrt{n_e}$ read from the ground. The trace ends at the critical frequency $f_c = f_p^{\max}$: above it, vertical shots pierce the layer and never return. A century of ionograms is how the layers are known at all. Oblique incidence is more forgiving, and this is the entire economics of shortwave. A shallow ray needs only its vertical wavenumber killed, so reflection requires $f\cos\theta_i < f_p$ — the secant law $f_{\text{MUF}} = f_c \sec\theta_i$. Geometry caps $\sec\theta_i$ near 3–4 for realistic hops (the first worked example computes one), so the maximum usable frequency for long paths runs to roughly $3f_c \approx 25$–30 MHz by day. Between $f_c$ and the MUF lies the working shortwave spectrum; each bounce covers up to $\sim$3500 km, and a few hops cross any ocean — the physics that carried intercontinental traffic for decades. And the day–night drama is a tale of the two Drude limits. By night the D-layer (dense air, high $\gamma$) recombines away, leaving the high, collisionless F-layer mirror: AM broadcast at 0.5–1.6 MHz, far below $f_p$, skips beautifully a thousand kilometers. By day the D-layer regrows under the mirror; AM must cross it twice per hop with $\gamma$ comparable to $\omega$, the ohmic limit applies, and absorption (worst at low $f$) devours exactly the band the F-layer would have reflected. Same ceiling, new toll booth: AM dies to groundwave range at noon, and the skywave returns at dusk. Two more entries close the file. Re-entry blackout: a capsule's shock layer ionizes air to $n_e \sim 10^{19}\ \text{m}^{-3}$ — $f_p \approx 28$ GHz — a cutoff above every band aboard, hence the famous minutes of silence. Wood's window: in 1933 R. W. Wood found alkali metals turn transparent in the ultraviolet — sodium's conduction-electron density gives $\hbar\omega_p = 5.9$ eV, plasma wavelength 209 nm, and measured sodium indeed opens up near 210 nm. A shortwave trick and a shiny metal are the same inequality, $\omega \gtrless \omega_p$, evaluated twenty-two orders of magnitude apart in density.

Above cutoff the plasma is transparent but never innocent: $v_g = c\sqrt{1-\omega_p^2/\omega^2}$ depends on frequency, so a sharp broadband pulse crossing a plasma arrives smeared, high frequencies first. For the nearly-empty interstellar medium this is a minuscule effect per meter and a magnificent one per kiloparsec — and since 1968 it has been radio astronomy's tape measure.

1 · Reading the sky with an ionosonde (confidence builder)

A midday ionosonde measures a critical frequency $f_c = 9.0$ MHz for the F2 layer at a true height of 300 km. (a) What is the peak electron density? (b) A broadcaster wants a single-hop link to a receiver 2000 km away off this layer. Up to what frequency does the link work?

Step 1. Invert the engineer's formula. From $f_p = 8.98\sqrt{n_e}$,

A cubic millimeter of the layer holds a thousand electrons; that whisper of charge is the whole mirror.

Step 2. Set up the hop geometry. Single hop, reflection at the path midpoint: horizontal half-range 1000 km, mirror height 300 km (flat-Earth geometry is adequate at this range for a first pass). The ray meets the layer at incidence angle

Step 3. Apply the secant law. Reflection survives as long as the vertical piece of the wave is below cutoff, $f\cos\theta_i < f_c$:

Anything between 9 and 31 MHz that would pierce the layer straight up makes this 2000 km hop off the same layer — the whole shortwave allocation fits in the gap. (A circuit planner would shave this with spherical-Earth corrections and stay ~15% below the MUF for reliability; the physics of the margin is exactly this calculation.)

2 · The Crab pulsar's smeared clock (exam level)

The Crab pulsar has $\mathrm{DM} = 56.77\ \text{pc cm}^{-3}$. (a) How late is its pulse at 1.4 GHz, relative to infinite-frequency arrival? (b) A 400 MHz receiver has 8 MHz of bandwidth; how badly does the pulse smear across the band, against the 33 ms period? (c) Estimate the distance, taking $\langle n_e\rangle = 0.03\ \text{cm}^{-3}$.

Step 1. Absolute delay at 1.4 GHz. Straight from the boxed formula,

The radio pulse loses 120 ms to the X-ray pulse ($f^{-2}$ effectively zero) — nearly four full rotations of a 30 Hz pulsar. Uncorrected, radio and X-ray data would disagree about which rotation emitted a given pulse.

Step 2. Differential smear across the 400 MHz band. Evaluate the delay at both band edges:

The pulse drifts $\mathbf{57\ ms}$ across a mere 8 MHz of band — nearly two full pulse periods. Summed naively, the 33 ms pulse would blur away: a plasma-physics fact, not an instrument defect. The cure is dedispersion — split the band into channels, realign with the $f^{-2}$ law, then add — and getting it right is how the Crab's 33 ms clock was pinned down within months of the 1968 discovery.

Step 3. The column as a ruler. $\mathrm{DM} = \langle n_e\rangle d$ gives

in agreement with the Crab Nebula's independently known 2 kpc. Run in reverse — parallaxes calibrating $\langle n_e\rangle$, then DM ranging everything else — this is how the Galactic pulsar map gets its third dimension. A free-electron formula is doing astrometry.

References. Jackson, Classical Electrodynamics, 3rd ed., §7.5 (plasma dispersion) · Griffiths, 4th ed., Ch. 9 (dispersion; free-electron limit) · Chen, Introduction to Plasma Physics and Controlled Fusion, Ch. 1, 4 (oscillations, Debye shielding, waves) · Stix, Waves in Plasmas, Ch. 1 · Lorimer and Kramer, Handbook of Pulsar Astronomy (dispersion-measure practice).

Provenance. Authored fresh for the track: the source export carried this topic as a title-and-definition stub with its serial held, and the content was built to the track's standard from the canonical treatments above. The slab derivation follows Tonks–Langmuir (Chen Ch. 1); the dielectric limit is the $\omega_0 \to 0$, $\gamma \to 0$ reading of the Lorentz model owned by Absorption and Dispersion; the dispersion-measure convention is $\mathcal{K} = e^2/8\pi^2\varepsilon_0 m_e c = 4.149$ ms GHz² (pc cm⁻³)⁻¹, recomputed from CODATA constants, as were all worked numbers (8.98 coefficient, 10.8 eV copper, 209 nm sodium, 120 ms / 57 ms Crab delays, 31 MHz MUF, 28 GHz blackout).

Prerequisites: Absorption and Dispersion · Electromagnetic Waves in Conductors · Phase and Group Velocity · Maxwell's Equations in Matter. Leads to: Waves in Magnetized Plasmas · Landau Damping · Magnetohydrodynamics.