Dielectrics

What are the four Maxwell's equations (Heaviside form) and what does each describe physically?

div D = rho (Gauss: charges produce D-field divergence); div B = 0 (no magnetic monopoles); curl E = -dB/dt (Faraday+Lenz: changing B induces circulating E); curl H = j + dD/dt (Ampere+Maxwell: conduction current + displacement current produce circulating H).

Define the electric displacement field D in terms of E and P. Why is D needed at all?

D = eps0*E + P. When a dielectric polarizes, bound charge partially screens the applied field. D packages the free-charge-only response so Gauss's law (div D = rho_free) stays valid without explicitly tracking bound charge.

For a linear isotropic dielectric, write P and D in terms of susceptibility chi_e and relative permittivity eps_r.

P = eps0chi_eE; D = (1+chi_e)eps0E = eps_reps0E, so eps_r = 1 + chi_e.

Derive the radial Coulomb field of a point charge Q using Gauss's law.

Integrate div D = rho over a sphere of radius r surrounding the charge: closed-surface integral of D.dA = Q, so D(r)4pir^2 = Q, giving E(r) = Q/(4pieps0eps_r*r^2).

What two conditions define a dielectric material at the atomic/band-structure level?

(1) It is an insulator - essentially no free carriers, electrons are localized/bound to atoms. (2) It is polarizable - the bound charge distribution can shift under an applied field. The band gap Eg is so large that Eg >> kT keeps electrons out of the conduction band at room temperature.

Derive the capacitance of a parallel-plate capacitor filled with permittivity eps_r, and state the engineering implication of increasing eps_r.

Q = eps0eps_rEA, V = Ed, so C = Q/V = eps0eps_rA/d. Higher eps_r stores more charge at a fixed voltage and fixed geometry - the basis for using high-permittivity ceramics in compact capacitors - but the same material typically has a lower breakdown field, which limits how much of that theoretical capacity is usable.

Derive the stored energy density of a charged parallel-plate capacitor in terms of E and eps_r.

W = (1/2)QV = (1/2)CV^2. Substituting C = eps0eps_rA/d and V = Ed gives W/Volume = (1/2)eps_reps0E^2 - energy density scales with permittivity and with the square of the field.

A Li-ion battery reaches ~350 Wh/L volumetric energy density. If a capacitor dielectric has eps_r=10, what E-field would match this, and why is a capacitor impractical here?

350 Wh/L = 1.26e9 J/m^3. Solve (1/2)eps_reps0E^2 = W/V for E: E = sqrt(2(W/V)/(eps_r*eps0)) ~ 5.3e9 V/m = 5340 MV/m. This is far beyond real dielectric breakdown fields (~10-30 MV/m), so a capacitor would break down long before storing that much energy - which is why electrochemical batteries, not capacitors, dominate for energy-dense storage.

How does curl H = j + dD/dt unify conduction current and displacement current, and why was this historically important?

It shows that a time-varying E-field (dD/dt) generates a magnetic field just as a real conduction current j does, even with no charges physically moving. This 'displacement current' term was Maxwell's key addition needed to keep charge continuity self-consistent, and it is precisely what allows electromagnetic waves to propagate through free space.

Dielectrics span roughly 10 orders of magnitude in conductivity, out of ~26 orders spanned by all materials. What does this tell you about dielectric classification?

Dielectric behaviour (how tightly electrons are bound, defect/ionic conduction pathways) still varies enormously between near-perfect insulators and leaky/lossy dielectrics, but this range sits entirely below metallic conductivities - dielectrics are a broad but comparatively low-conductivity slice of the full spectrum.

Write the definition of complex permittivity eps* and explain what the real and imaginary parts represent physically.

eps* = eps' + i*eps''. eps' is the in-phase polarization response (energy storage in the field). eps'' is the out-of-phase response, representing energy dissipated as heat (loss) each cycle.

Derive eps'' for a medium with real Ohmic conductivity sigma, starting from Ampere's law with an oscillating field.

curl H = sigmaE + eps0eps_rdE/dt. For E = E0exp(-iomegat), dE/dt = -iomegaE, so curl H = (sigma - iomegaeps0eps_r)E = -iomega[eps0eps_r + isigma/omega]E. Comparing to curl H = -iomegaeps0eps_rE gives eps_r = eps_r + isigma/(omegaeps0), i.e. eps'' = sigma/(omega*eps0).

Sea water has sigma = 5 S/m. Calculate eps'' at f = 2 GHz.

eps'' = sigma/(omegaeps0) = 5/(2pi2e98.85e-12) ~= 45.

Why does eps'' from Ohmic conductivity decrease as frequency increases, even though sigma itself is constant?

eps'' = sigma/(omegaeps0). The displacement current (proportional to omegaeps0eps_r) grows with frequency while the conduction current (sigmaE) does not, so conduction losses become relatively less significant at higher frequencies.

Define the loss tangent tan(delta) and explain what a value near 1 means for wave propagation.

tan(delta) = eps''/eps'. A value near 1 means the field amplitude decays substantially within about one wavelength of travel - the medium is highly lossy, which is why, e.g., radio and radar signals cannot penetrate far into seawater.

Given the complex refractive index n* = n + ikappa and eps = (n)^2, show that eps' = n^2 - kappa^2 and eps'' = 2n*kappa.

(n + ikappa)^2 = n^2 - kappa^2 + i(2nkappa) = eps' + ieps''. Equating real and imaginary parts gives eps' = n^2 - kappa^2 and eps'' = 2n*kappa.

In the low-loss limit (kappa << n), show that tan(delta) ~= 2*kappa/n.

eps' ~= n^2 and eps'' = 2nkappa, so tan(delta) = eps''/eps' ~= 2nkappa/n^2 = 2*kappa/n.

A 1.5 micron optical fibre (n=1.5) has an absorption coefficient of 1 dB/km. Find the extinction coefficient kappa and the loss tangent.

Over 1 km, P_out/P_in = 10^(-0.1) = 0.79 = exp(-2kappak0z), k0 = 2pi/lambda. Solving gives kappa ~= 2.8e-11. Loss tangent tan(delta) = 2*kappa/n ~= 3.5e-11 - an extraordinarily low-loss material, which is exactly why silica fibre supports long-haul telecom links without repeaters every few metres.

How are complex conductivity sigma* and complex permittivity eps_r* related for a medium that behaves as both dielectric and Ohmic conductor?

They are equivalent descriptions of the same physics: sigma* = sigma - iomegaeps0eps_r (equivalently eps_r = eps_r + isigma/(omegaeps0)). You can always 'emulate' a lossy dielectric with an effective complex conductivity, or emulate an Ohmic conductor's displacement current with an effective complex permittivity.

Why is it engineering-useful to describe a real lossy material (salty water, tissue, food) with one complex permittivity rather than separate sigma and eps parameters?

A single eps*(omega) self-consistently captures both energy storage and dissipation at each frequency, so wave propagation, absorption and reflection can all be computed with the same complex-refractive-index machinery used for ordinary dielectrics - essential for designing microwave ovens, radar, and biomedical microwave sensing across a range of frequencies.

Starting from Maxwell's curl equations in a source-free, non-magnetic dielectric of permittivity eps_r, derive the 3D wave equation for E.

Take curl of curl E = -dB/dt: curl(curl E) = -mu0(d/dt)(curl H) = -mu0eps_reps0d^2E/dt^2. Using the identity curl(curl E) = grad(div E) - Laplacian(E), and div E = 0 (Gauss, no free charge), gives Laplacian(E) - mu0eps0eps_r*d^2E/dt^2 = 0.

From the wave equation Laplacian(E) = mu0eps0eps_r*d^2E/dt^2, derive the phase velocity and refractive index of light in a dielectric.

Comparing to the generic wave equation Laplacian(X) - (1/v^2)d^2X/dt^2 = 0 gives v = 1/sqrt(mu0eps0eps_r) = c/sqrt(eps_r) = c/n, so n = sqrt(eps_rmu_r), which reduces to n = sqrt(eps_r) for non-magnetic media.

What value did Maxwell compute for the speed of light purely from electrical and magnetic constants, and why was it historically significant?

c = 1/sqrt(mu0*eps0) ~= 3.107e8 m/s (a few percent high but close to the measured speed of light). The match convinced Maxwell that light itself is an electromagnetic wave, unifying optics with electricity and magnetism (1862, 1864).

Derive the intrinsic impedance Z of a plane wave in a linear, isotropic, non-magnetic dielectric, and state its value in free space.

From the transverse plane-wave solution, Z = Ex/Hy = sqrt(mu0mu_r/(eps0eps_r)) = Z0/n for non-magnetic media. In vacuum (n=1): Z0 = sqrt(mu0/eps0) ~= 376.7 ohms.

Show that the wave intensity (Poynting vector magnitude) for a plane wave can be written |P| = n*Ex^2/Z0.

|P| = ExHy = Ex^2/Z, and Z = Z0/n, so |P| = nEx^2/Z0 = eps_r*Ex^2/Z0 - intensity scales linearly with refractive index and with the square of the field amplitude.

How does introducing a complex refractive index n* = n + i*kappa modify the plane-wave solution, and what does kappa represent?

Ex(z,t) = Re[Aexp(i(nk0z - omegat + phi))] = exp(-kappak0*z) * Re[Aexp(i(nk0z - omega*t + phi))]. The imaginary part kappa (extinction coefficient) produces exponential decay of the field amplitude with propagation distance z - it is the absorption/loss of the medium.

Derive the absorption (1/e) length for intensity, l_1/e, in terms of vacuum wavelength lambda0 and extinction coefficient kappa.

Intensity is proportional to exp(-2kappak0z). Setting 2kappak0l_1/e = 1 gives l_1/e = 1/(2kappak0) = lambda0/(4pikappa).

A THz beam at f = 500 GHz reflects off a silicon wafer with return loss S11 = -5.2 dB. Determine the refractive index and real permittivity.

R = 10^(S11/10) = 0.54. Using R = ((n-1)/(n+1))^2, n = (1+sqrt(R))/(1-sqrt(R)) ~= 3.37, so eps' = n^2 ~= 11.35.

Derive the Fresnel field reflection coefficient r and transmission coefficient t at normal incidence between refractive indices n1 and n2.

Continuity of tangential E and H at the boundary gives A_forward + A_backward = A_transmitted and (A_forward - A_backward)/Z1 = A_transmitted/Z2. Solving (with Z = Z0/n) gives r = (n1-n2)/(n1+n2) and t = 1+r = 2*n1/(n1+n2).

Why are E and H (rather than D or B) the natural field variables for deriving electromagnetic boundary conditions?

Tangential E is continuous across an interface (from Faraday's law applied to a thin loop straddling the boundary), and tangential H is continuous when there is no surface current (from Ampere's law similarly). These are the quantities directly conserved at interfaces, making them the correct basis for matching wave amplitudes.

Show that reflectance R = r^2 and transmittance T = 4n1n2/(n1+n2)^2 satisfy energy conservation R + T = 1.

R + T = [(n1-n2)^2 + 4n1n2]/(n1+n2)^2 = (n1^2 - 2n1n2 + n2^2 + 4n1n2)/(n1+n2)^2 = (n1+n2)^2/(n1+n2)^2 = 1.

A return loss measurement gives -10 dB. What fraction of incident power is reflected?

R = 10^(-10/10) = 0.1, i.e. 10% of the power is reflected.

Derive the field reflection coefficient for air (n=1) reflecting off a dielectric of index n, and interpret the limits n->1 and n->infinity.

r = (n-1)/(n+1). As n->1, r->0 (no index mismatch, no reflection). As n->infinity, r->1 (perfect reflector) - this is why very high-index materials, and ultimately metals, behave like mirrors.

Light in air (n=1) strikes glass (eps_r = 2.25, n = 1.5) at normal incidence. Calculate reflectance R and transmittance T.

R = [(1-1.5)/(1+1.5)]^2 = (0.5/2.5)^2 = 0.04 (4%); T = 1 - R = 0.96 (96%).

Derive the condition on a quarter-wave anti-reflection coating (medium 2 between media 1 and 3) for zero net reflection.

With k2d = pi/2 (thickness d = lambda/4 in medium 2) and matching boundary conditions at both interfaces, zero reflection requires Z2 = sqrt(Z1Z3) - the coating's impedance must be the geometric mean of the impedances on either side, the basis of anti-reflective lens coatings.

Why must a general reflection problem include both a forward (incident) and backward (reflected) wave on the incidence side, but only a forward wave on the transmission side?

There are two independent boundary conditions (continuity of tangential E and tangential H) that must be satisfied simultaneously. On the incidence side you need two unknown amplitudes (incident and reflected) to have enough freedom to satisfy both equations; the transmission side only needs one outgoing wave since there is nothing beyond it to reflect back.

Write the Debye relaxation equation for the complex permittivity of a polar liquid and define each symbol.

eps(omega) = eps_inf + (eps_s - eps_inf)/(1 - iomega*tau), where eps_inf is the high-frequency (electronic) permittivity, eps_s is the static (low-frequency) permittivity, tau is the dipole relaxation time, and omega is angular frequency.

Derive the Debye equation starting from an exponentially relaxing polarization P(t) = P0*exp(-t/tau) after a field is switched off.

Fourier transforming P(t) = 0 for t

Split the Debye equation into real and imaginary parts and identify the frequency of peak loss.

eps'(omega) = eps_inf + (eps_s - eps_inf)/(1 + omega^2tau^2); eps''(omega) = (eps_s - eps_inf)omegatau/(1 + omega^2tau^2). The loss eps'' peaks when omega*tau = 1.

Why does a polar liquid like water show a much higher static permittivity eps_s than its optical-frequency permittivity eps_inf?

At low/DC frequency, permanent molecular dipoles have time to rotate and align with the field, contributing large orientational polarization and hence high eps_s. At optical frequencies (~10^15 Hz) the dipoles' rotational inertia and friction (tau ~ picoseconds) are far too slow to follow the field, so only fast electronic polarization contributes, giving a much lower eps_inf.

Why is the microwave-derived refractive index of water (n ~ sqrt(eps_s) ~ 9) so much larger than its optical refractive index (n ~ 1.3)?

The static/microwave permittivity includes the large, frequency-independent orientational (dipole reorientation) contribution, whereas the optical refractive index reflects only the much smaller electronic polarizability, since the dipoles cannot respond fast enough at optical frequencies to contribute.

Water has tau ~ 8 ps. Estimate the frequency of peak dielectric loss and explain why this makes water problematic for underwater radar or wireless signals.

Peak loss occurs at f = 1/(2pitau) ~= 1/(2pi8e-12) ~= 20 GHz, squarely in the microwave band used for radar and wireless links. Because tan(delta) approaches order unity there, EM waves decay within roughly one wavelength of entering water, which is why radar and mobile signals cannot function underwater.

Extend the Debye model to include ionic conductivity sigma. Write the full complex permittivity.

eps(omega) = eps_inf + (eps_s - eps_inf)/(1 - iomegatau) + isigma/(omegaeps0). The added isigma/(omega*eps0) term is Ohmic loss, which dominates eps'' especially at low frequency.

Stella Artois beer has sigma = 0.14 S/m, tau = 8 ps, eps_s = 80, eps_inf = 4. At 1 GHz (omega*tau << 1), estimate the total loss tangent and its two contributions.

At this low omegatau, eps' ~= eps_s = 80. The Debye relaxation loss tangent is about 0.05, and the separate Ohmic loss tangent (sigma/(omegaeps0*eps_s)) is about 0.03. The two mechanisms add to give a total loss tangent of about 0.08 - orientation loss and ionic conduction loss are physically distinct and combine linearly in eps''.

Explain the engineering logic behind choosing ~2.45 GHz for microwave ovens, referencing dielectric relaxation of water in food.

Water's Debye loss behaviour places strong absorption in the low-microwave range once bound water and dissolved ions in real food are accounted for. 2.45 GHz is chosen as a practical compromise in the ISM band: high enough for efficient dielectric heating (Pabs proportional to omegaeps''E^2), but with a penetration depth still large enough to heat food through its volume rather than just scorching the surface.

Derive the time-averaged power absorbed per unit volume by a dielectric in an oscillating field, in terms of eps'', E, and omega.

Pabs = (omegaeps0eps''/2)E0^2 for peak field amplitude E0 (equivalently omegaeps0eps''

Derive the Drude AC conductivity sigma(omega), starting from Newton's second law for a free electron with a friction (scattering) term.

m_effdv/dt = -(m_eff/tau)v - eE(omega)exp(-iomegat). Writing current density j = n_eev and solving the resulting linear ODE for a harmonic j(t) gives sigma(omega) = sigma0/(1 - iomegatau), where sigma0 = n_ee^2tau/m_eff is the DC (Ohm's law) conductivity.

What does the 'omega*tau product' mean physically in the Drude model, and what are its two limiting regimes?

omegatau is roughly the number of electron scattering (collision) events per cycle of the driving field. When omegatau << 1, electrons scatter many times per cycle and behave Ohmically (real conductivity dominant, sigma ~= sigma0). When omega*tau >> 1, electrons are effectively collisionless ('ballistic') over one cycle, conductivity becomes dominated by its imaginary part, and the metal behaves like a plasma with negative permittivity.

For copper at room temperature, estimate whether omega*tau << 1 or >> 1 at visible light (500 nm) versus microwave (2 GHz) frequencies.

Using typical copper parameters, tau ~ 2-4e-14 s. At visible light (omega ~ 4e15 rad/s), omegatau ~ 90-plus, so omegatau >> 1 (ballistic regime, imaginary conductivity dominates). At microwave (f = 2 GHz), omegatau ~ 3e-4, so omegatau << 1 (Ohmic regime, imaginary part negligible).

Derive the real part of a metal's permittivity in the high-frequency (omega*tau >> 1) limit, and show how it leads to the plasma frequency omega_p.

In the omegatau >> 1 limit, sigma(omega) ~= isigma0/(omegatau) = i(n_ee^2)/(m_effomega). Converting to permittivity form gives eps'(omega) ~= 1 - (omega_p/omega)^2, where omega_p^2 = n_ee^2/(eps0m_eff).

Explain physically why a metal reflects light for omega < omega_p but becomes transparent for omega > omega_p.

For omega < omega_p, eps'(omega) < 0, so the refractive index is purely imaginary and the wave becomes evanescent (exponentially decaying, non-propagating) inside the metal - it is almost entirely reflected. For omega > omega_p, eps'(omega) > 0 again, the wave can propagate, and the metal becomes transparent (assuming no interband absorption interferes).

Gold has n_e = 5.9e28 per cubic metre and m_eff = 1.1*m_e. Calculate its plasma wavelength, and explain why gold is still not transparent in the UV despite this.

omega_p = sqrt(n_ee^2/(eps0m_eff)) gives a plasma wavelength lambda_p = 2pic/omega_p ~= 144 nm (deep UV). Classically this predicts UV transparency, but real gold stays opaque there because of interband transitions - quantum electron transitions across filled/empty bands that the classical free-electron Drude model cannot capture.

Why does the ionosphere reflect AM radio waves (~100 kHz-1 MHz) but is transparent to FM radio (~100 MHz) and satellite links (~10 GHz)?

The ionosphere's free-electron plasma frequency is a few MHz. AM frequencies lie below omega_p, so eps' < 0 and the waves reflect, enabling long-range terrestrial AM broadcasting via ionospheric bounce. FM and satellite frequencies exceed omega_p, so eps' > 0 and the ionosphere is transparent, allowing satellite communication and line-of-sight FM to pass through.

What is the forced-pendulum analogy for explaining the sign of a metal's permittivity, and what does each frequency regime correspond to?

Below resonance (f << f_r) a driven pendulum moves in phase with the driving force - analogous to bound electrons in an atom at low frequency, giving positive permittivity. Above resonance (f >> f_r) the pendulum's inertia dominates and it moves out of phase with the driving force - analogous to free electrons in a metal at optical frequencies, giving negative permittivity.

How does correcting the Drude model for background electronic (core) polarizability change the permittivity expression, and why is the correction needed?

eps'(omega) = eps_inf - (omega_p/omega)^2 = eps_inf*(1 - (omega_p'/omega)^2), where eps_inf accounts for bound-electron/ionic-core polarizability that also contributes even at high frequency. Free conduction electrons are not the only source of polarization in a real metal, so omitting eps_inf would misestimate the plasma-edge frequency.

Silver has resistivity rho = 1.6e-8 ohmm and electron density n_e = 5.9e28 per cubic metre. Estimate tau and the omegatau product for red light (600 nm), and interpret the result.

Using sigma0 = 1/rho = n_ee^2tau/m_eff gives tau ~= 3.7e-14 s. For lambda = 600 nm, omega = 2pic/lambda ~= 3.14e15 rad/s, giving omegatau ~= 115. Since omegatau >> 1, conduction electrons are ballistic at optical frequencies - conductivity is dominated by its imaginary part, consistent with silver's strongly negative permittivity and high visible reflectivity.

Derive the skin depth delta for a plane wave penetrating a good conductor.

In the good-conductor limit (sigma >> omegaeps), the wave equation reduces to k^2 = imusigmaomega, giving k = sqrt(muomegasigma/2)(1+i). Substituting into Ex = E0exp(i(kz - omegat)) produces exponential decay exp(-z/delta) with delta = sqrt(2/(muomegasigma)).

Calculate the skin depth of aluminium (sigma = 4e7 S/m) at 2.4 GHz, and explain the engineering implication for EM shielding.

delta = sqrt(2/(mu0omegasigma)) = sqrt(2/(4pi1e-7 * 2pi2.4e9 * 4e7)) ~= 1.6 micrometres. Because the skin depth is this tiny, even a thin aluminium wall or foil (far thicker than 1.6 micrometres) fully blocks and reflects microwaves - the reason thin metal coatings work as effective EM shields at microwave frequencies, e.g. microwave oven walls.

Derive the surface impedance Z_m of a good conductor, and show it is tiny compared to free-space impedance Z0 = 376.7 ohms.

Z_m = Ex/Hy = (1-i)/(sigmadelta) = (1+i)sqrt(muomega/(2sigma)), so |Z_m| = sqrt(mu*omega/sigma). For aluminium at 2.4 GHz this gives |Z_m| ~= 0.022 ohms, about four orders of magnitude smaller than Z0 - so the air-metal reflectance R = [(Z2-Z1)/(Z2+Z1)]^2 is around 0.9998, meaning metals reflect over 99.9% of incident microwave power, making them excellent mirrors and waveguide walls.

Derive the characteristic impedance Z_coax and signal velocity of a coaxial cable from its per-unit-length capacitance and inductance, and evaluate for eps_r = 3, b/a = 4.

C' = 2pieps0eps_r/ln(b/a); L' = mu0ln(b/a)/(2pi). Z_coax = sqrt(L'/C') = [ln(b/a)/(2pi)] * sqrt(mu0/eps0)/sqrt(eps_r) = [ln(b/a)/(2pisqrt(eps_r))]Z0; signal velocity v = 1/sqrt(L'C') = c/sqrt(eps_r). For eps_r = 3, b/a = 4: Z_coax = ln(4)/(2pisqrt(3))*376.7 ~= 48 ohms, v ~= 0.58c - which is why standard coax uses plastic dielectrics (eps_r ~ 2-3) to hit the common 50 ohm standard, and why signals travel slower than the vacuum speed of light.