Electromagnetic Waves in Linear Dielectric Media

The linear-dielectric plane wave is the most economically consequential solution in this track: essentially all long-haul data on Earth rides it through fused-silica fiber at $n \approx 1.468$, with the residual frequency-dependence of that n (dispersion, article 0598's business) actively compensated so that petabits per second survive transoceanic distances. Lens design, anti-reflection coatings, and photolithography (where "immersion" machines exploit water's n = 1.44 at 193 nm to sharpen features) are all engineering in this article's variables — and the lithography case is a pointed reminder that n must be looked up at the working wavelength, water's UV index differing measurably from its visible one. The frontier plays with the parameters themselves. Metamaterials engineer ε and μ independently — impedance-matched slabs ($Z = Z_0$ despite $n \neq 1$) transmit without reflection, and negative-ε/negative-μ bands realize the reversed-phase-velocity media the polarization article's FAQ teased. Epsilon-near-zero materials stretch λ/n toward infinity, making phase effectively uniform across a chip. And at the highest intensities, the "linear" qualifier is the research program: nonlinear optics begins precisely where this article's constitutive assumption ends, one power of E at a time.

What does a light wave actually do inside glass? It does not slow down the way a runner enters sand — nothing drags on the photons between atoms, where there is only vacuum. Instead, the wave's $\mathbf{E}$ polarizes the medium as it passes: every molecule becomes an oscillating dipole (the E3 induced-dipole machinery, driven now at $10^{15}$ Hz), and those dipoles re-radiate coherently at the same frequency. The observed wave is the sum — incident plus every molecular echo — and the sum, worked out by superposition, is a wave of the same frequency whose phase crests advance more slowly: speed $c/n$. The index is not friction; it is organized re-radiation. (This picture, made quantitative, is the Ewald–Oseen theorem; this article takes the macroscopic shortcut through $\varepsilon$, which packages all the dipole physics the E3 chapter already audited — Clausius–Mossotti connects the two lenses, by name.) The macroscopic consequences are geometrically vivid. Frequency cannot change — the medium is driven at ω and linear systems respond at the driving frequency; equivalently, wave crests cannot pile up or vanish at the boundary, so the arrival rate equals the departure rate on both sides. With $f$ fixed and $v$ reduced, the wavelength must compress: $\lambda_{\text{med}} = v/f = \lambda_0/n$. Green light of 532 nm becomes a 400 nm-spaced comb inside water while remaining, to every eye and detector, green — color is frequency's property, not wavelength's.

The medium also retunes the wave's internal exchange rates. The Faraday lock becomes $B = E/v$: for the same electric amplitude, matter carries relatively more magnetic field — glass at $n = 1.5$ carries 50% more B per unit E than vacuum. The energy and flux bookkeeping inherit ε and μ, and the ratio $E/H$ — the wave impedance — drops from vacuum's 377 Ω to $Z_0/n$ (251 Ω in glass, 283 Ω in water). Impedance is the number to watch leaving this article: two media with equal $n$ but different $Z$ are optically distinguishable the moment a boundary appears, and the reflection articles run entirely on $Z$-mismatch (with $n$-mismatch as the non-magnetic shorthand). One boundary of validity, drawn honestly at the door: linear means the induced polarization is proportional to E — true for everyday intensities, false at the $10^{11}$ V/m crests of article 0585's femtosecond focus, where $\chi^{(2)}$ and $\chi^{(3)}$ physics (harmonic generation, self-focusing) switches on. Homogeneous postpones interfaces to the next article. And ε is frequency-dependent — the article's closing scandal — with genuine absorption arriving when ε goes complex (article 0595 owns that machinery, by name).

Begin where the materials chapters left Maxwell's equations. In a linear, homogeneous medium with no free charges or currents ($\rho_f = 0$, $\mathbf{J}_f = 0$ — an insulator away from its boundaries), the macroscopic equations close on E and B alone:

where: ε, μ — the medium's permittivity and permeability, packaging the bound-charge and bound-current response (the E3/E5 linear-media articles own the constitutive story; the E6 in-matter Maxwell article assembled these equations). Read the structure: (1) is exactly the vacuum system with the single substitution $\mu_0\varepsilon_0 \to \mu\varepsilon$ — the whole article follows from that observation, made once and cashed repeatedly.

Maxwell's $n = \sqrt{\varepsilon_r}$ faces the data. For nonpolar gases and many solids it triumphs: hydrogen, benzene, paraffin match to a percent or better. For water: $\varepsilon_r(\text{DC}) = 80$, predicting $n = 8.94$ — while every swimming pool shows 1.33. The resolution is not that Maxwell was wrong but that the question "what is water's ε?" is ill-posed until a frequency is named. The E3 polarization chapter built three mechanisms with three speed limits: orientational polarization (permanent dipoles physically rotating — water's specialty, worth ~76 of its 80) manages only ~$10^{10}$ Hz before rotational inertia and collisions defeat it (Debye relaxation, the E3 orientational article's business); ionic/atomic displacement follows to ~$10^{13}$ Hz; electronic cloud distortion follows to ~$10^{15}$–$10^{16}$ Hz. Light at $5.6\times10^{14}$ Hz outruns all but the electrons: the ε that light experiences is the electronic remnant, $\varepsilon_r(\text{optical}) = 1.78 = (1.333)^2$ — Maxwell's relation, vindicated at the right frequency.

Each drop-out step is not a clean cliff but a lossy shoulder: near a mechanism's cutoff, the response lags the drive, ε acquires an imaginary part, and the medium absorbs — which is exactly where the microwave oven parks itself (2.45 GHz, on the flank of water's Debye relaxation: slow enough to couple, mismatched enough to dissipate). The full machinery — complex ε(ω), the Lorentz oscillator, anomalous dispersion, why glass is transparent and UV-opaque — is the absorption-and-dispersion article's (0595, by name). What this article fixes permanently is the question discipline: n is shorthand for $n(\omega)$, and every use of "the" dielectric constant must name its frequency.

1 · A green laser enters the pool (confidence builder)

A 532 nm laser beam (vacuum wavelength) with in-water amplitude $E_0 = 500$ V/m propagates through water, $n = 1.333$. Find the frequency, in-water wavelength and speed, $B_0$, impedance, and intensity; audit the equal partition.

Step 1. Frequency — fixed by the source, unchanged by the medium: $f = c/\lambda_0 = 2.998\times10^8/5.32\times10^{-7} = \mathbf{5.64\times10^{14}\ Hz}$. Still green, everywhere.

Step 2. In-water kinematics: $v = c/n = \mathbf{2.25\times10^8\ m/s}$; $\lambda = \lambda_0/n = 532/1.333 = \mathbf{399\ nm}$ — a wavelength that would be violet in vacuum, yet the light is green: color rides frequency.

Step 3. The retuned lock: $B_0 = E_0/v = 500/2.25\times10^8 = \mathbf{2.22\ \mu T}$ — a third more magnetic field per volt-per-metre than vacuum would carry ($nE_0/c$).

Step 4. Impedance and intensity: $Z = Z_0/n = 376.7/1.333 = \mathbf{283\ \Omega}$; $I = E_0^2/2Z = 25\times10^4/565.2 = \mathbf{442\ W/m^2}$. Cross-check via $\tfrac12\varepsilon vE_0^2$ with $\varepsilon = 1.777\varepsilon_0$: 442 W/m² ✓. Equal-partition audit at a crest: $u_E = \tfrac12\varepsilon E_0^2 = 1.97\times10^{-6}$ J/m³ and $u_B = B_0^2/2\mu_0 = 1.97\times10^{-6}$ J/m³ — equal to four figures ✓.

2 · Prosecuting the water scandal (exam level)

(a) Predict water's refractive index from its handbook dielectric constant $\varepsilon_r = 80$; compare with the observed $n = 1.333$ and locate the error. (b) Using the E3 mechanism split (orientational ≈ 76, ionic ≈ 2, electronic ≈ 1.8 of the 80), predict n in the far-infrared plateau (~$10^{12}$ Hz, above Debye but below the ionic cutoff) and at optical frequencies. (c) Explain the microwave oven's choice of 2.45 GHz on this ladder.

Step 1. (a) The naive prediction: $n = \sqrt{80} = \mathbf{8.94}$ — wrong by a factor of 6.7. The error is not in Maxwell's relation but in the frequency label: 80 is measured with slow fields (kHz capacitor bridges), where all three polarization mechanisms contribute; light polls only the mechanisms fast enough to answer at $5.6\times10^{14}$ Hz.

Step 2. (b) Between Debye (~$2\times10^{10}$ Hz) and the ionic cutoff, the orientational 76 has dropped out, leaving $\varepsilon_r \approx 80 - 76 = 4$: predicted far-IR index $n \approx \sqrt{4} = \mathbf{2.0}$ (measured values in the THz window: ~2.1 — the ladder works). At optical frequencies only the electronic 1.78 survives: $n = \sqrt{1.78} = \mathbf{1.333}$ ✓ — Maxwell's relation, exact when interrogated at the wave's own frequency.

Step 3. (c) The oven's 2.45 GHz sits deliberately on the rising shoulder of the Debye drop: low enough that dipoles still couple strongly to the field, high enough that their rotation lags — and lag means the work done on the dipoles is not returned but dissipated (the complex-ε loss that 0595 formalizes). Too low a frequency: lossless following, no heating. Too high: no coupling at all (which is why the oven leaves dry glass and air cold, and why it would fail above ~100 GHz). The appliance is a Debye-relaxation exploit with a door. Answers: (a) 8.94 vs 1.333 — frequency mislabel; (b) ≈2.0 far-IR, 1.333 optical; (c) maximal loss on the relaxation shoulder.

References. Griffiths, Introduction to Electrodynamics, 4th ed., §9.3.1 · Jackson, Classical Electrodynamics, 3rd ed., §7.1–7.2 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §9.5 · Hecht, Optics, 5th ed., §3.5. Key self-consistency checks: the substitution $\mu_0\varepsilon_0 \to \mu\varepsilon$ threads through the curl-of-curl derivation to give $v = c/n$ and $n = \sqrt{\varepsilon_r\mu_r}$; $B_0 = nE_0/c$ follows from the Faraday constraint. Worked Example 1 numbers: $f = 5.64\times10^{14}$ Hz, $\lambda = 399$ nm, $v = 2.25\times10^8$ m/s, $B_0 = 2.22\ \mu$T, $Z = 283\ \Omega$, $I = 442$ W/m² (two routes agree), $u_E = u_B = 1.97\times10^{-6}$ J/m³. Water-ladder arithmetic: $\sqrt{80} = 8.94$; $80 - 76 = 4 \Rightarrow n \approx 2.0$ in the far-IR plateau; $\sqrt{1.777} = 1.333$ at optical frequencies.

Builds on: The Electromagnetic Wave Equation · Monochromatic Plane Waves in Vacuum · Complex Notation for Electromagnetic Waves.

Leads to: Reflection and Transmission at Normal Incidence · The Fresnel Equations · Brewster's Angle and Total Internal Reflection · Energy Density and the Poynting Vector.