Dielectric Polarization

An insulator has no charge free to cross the sample, so a field cannot drive a steady current through it. What a field can do is move charge a short way inside each molecule, and stop. Two mechanisms do this. Induced polarization shifts an atom's electron cloud against its nucleus; it operates in every material, hydrogen through polyethylene, and it vanishes the instant the field is removed. Orientational polarization acts only on molecules that are already dipoles — water, HCl, plastics with polar side groups — and biases their random tumbling so that slightly more point along E than against it. Both leave the same signature: each small volume acquires a net dipole moment while every molecule in it stays neutral. The sample as a whole stays neutral too. Nothing has been added or removed; charge has only been redistributed, and by remarkably small distances.

Model a neutral atom as a nucleus of charge +q sitting inside a uniform electron cloud of radius R. An applied field pulls the two apart until the internal Coulomb attraction balances it, which for this model happens at a dipole moment p = αE with α = 4πε₀R³ — the polarizability, fixed by the atom's volume. Take R = 1.0 × 10⁻¹⁰ m and E = 1.0 × 10⁶ V m⁻¹, about a third of the breakdown field of dry air. Then α = 1.11 × 10⁻⁴⁰ C m² V⁻¹ and p = 1.11 × 10⁻³⁴ C m; dividing by e gives a displacement of 6.9 × 10⁻¹⁶ m, seven millionths of the atomic radius. That tininess is why the response stays linear across any laboratory field: the restoring force never leaves its Hooke's-law region. It is also why α, and the part of χₑ it produces, barely depends on temperature — nothing here is competing with thermal energy.

A polar molecule already carries a dipole moment; water's is 6.2 × 10⁻³⁰ C m. A field applies the torque τ = p × E that would align it, thermal collisions keep knocking it out of line, and Boltzmann statistics settle the contest between the alignment energy pE and the thermal energy kT. While pE ≪ kT the mean component along the field is ⟨p_z⟩ = p²E/(3kT), linear in E and falling as 1/T: warm a polar dielectric and it responds less. For water at 300 K in E = 1.0 × 10⁶ V m⁻¹, pE = 6.2 × 10⁻²⁴ J against kT = 4.14 × 10⁻²¹ J, so ⟨cos θ⟩ = 5.0 × 10⁻⁴. Polarization is a statistical bias, not a parade. The same model predicts χₑ = np²/(3ε₀kT); with n = 3.35 × 10²⁸ m⁻³ for liquid water that returns χₑ ≈ 12, well short of the measured 79, because each molecule sits in its neighbours' fields as well as the applied one.

Add the molecular dipole moments in a small volume and divide by it: P = (Σ p)/V is the polarization, in C m⁻². Inside a uniformly polarized slab nothing accumulates, because each molecule's positive end sits against its neighbour's negative end — but at the surfaces the cancellation runs out and leaves bound sheets, positive where P points out of the material and negative where it points in. The general statements are σ_b = P⋅n̂ and ρ_b = −∇⋅P, so bound charge fills a volume only where P varies from place to place. Take a slab filling a capacitor gap whose plates hold σ_f = 8.85 × 10⁻⁶ C m⁻², giving E₀ = σ_f/ε₀ = 1.00 × 10⁶ V m⁻¹; with K = 4.0 the faces carry 6.64 × 10⁻⁶ C m⁻². These charges are bound in the strict sense — earth a face and nothing flows, because they are molecular ends, not carriers.

The two bound sheets face each other across the slab exactly as the capacitor plates do, so together they make a uniform field of magnitude P/ε₀ pointing against E₀. The field inside the slab is therefore E = E₀ − P/ε₀, smaller than what you applied. But P was set by that reduced field, not by E₀. Substitute the linear law P = ε₀χₑ E: E = E₀ − χₑ E, so E(1 + χₑ) = E₀ and E = E₀/(1 + χₑ). Define K = 1 + χₑ and the result is E = E₀/K. This is the point of the lesson. K is not an extra ingredient bolted on afterwards; it is the solution of a feedback loop, and χₑ measures how hard the material argues back. χₑ = 0 is vacuum. χₑ = 3.0 gives K = 4.0, a field cut to a quarter and bound faces of σ_f(1 − 1/K) = 0.75σ_f. Push χₑ → ∞ and E → 0 with σ_b → σ_f: that is a conductor.

Four assumptions hide inside P = ε₀χₑ E. Linear: χₑ independent of E. It survives because saturation needs pE comparable to kT, which for water means E ≈ kT/p = 6.7 × 10⁸ V m⁻¹, while dry air breaks down at 3 × 10⁶ V m⁻¹ and the toughest polymer films fail well below 10⁹ — a dielectric arcs long before its dipoles come close to lining up. Isotropic: P parallel to E, which fails in a crystal such as calcite, where χₑ is a tensor and P leans off-axis. Homogeneous: one χₑ throughout, which fails at interfaces and in composites. Static: χₑ independent of frequency. Molecular rotation in water takes of order 10⁻¹¹ s, so its K falls from 80 at low frequency to n² = 1.77 in visible light, where only the electron clouds keep up. Ferroelectrics abandon linearity outright, with K in the thousands and a hysteresis loop.